Wittgensteins Enduring Arguments

Wittgenstein’s Enduring Arguments

Fifty years after Wittgenstein’s death, his philosophy and the arguments it embodied remain vital and applicable. Wittgenstein’s Enduring Arguments illustrates the use of Wittgenstein’s thought for continuing philosophical debates, old and new. Featuring essays by leading international philosophers, the collection examines the key theme of representation in Wittgenstein's philosophy. Organised into three clear parts the book considers representation in mind, in language and in what cannot be represented – the absolute. The first part applies Wittgenstein to leading questions concerning qualia, the grammar of phenomenology and developmental psychology. The second part applies Wittgenstein to vexing knots in the philosophy of language like language and concept acquisition, the normativity of meaning and linguistic understanding. The final section addresses Wittgenstein’s unique philosophical approach to logic, self, religion and ethics. Each specially commissioned chapter demonstrates the successful application of Wittgenstein’s philosophy; collectively they express a confidence that Wittgenstein’s arguments and his philosophy will endure. Wittgenstein’s Enduring Arguments is essential reading for those seeking to examine and assess the philosopher's lasting contribution to modern thought. Edoardo Zamuner holds a Ph.D. from the University of Edinburgh where he worked on the philosophy of emotion. Currently he is a research associate at the Institute of Neuroesthetics of University College, London. With David Levy, he edited a scholarly edition of Wittgenstein's Lecture on Ethics (2007). D. K. Levy is a lecturer in philosophy at the University of Edinburgh. He received his Ph.D. from King’s College, University of London. He has published concerning Wittgenstein and moral philosophy.

Wittgenstein’s Enduring Arguments

Edited by Edoardo Zamuner and D. K. Levy

This edition published 2009 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Simultaneously published in the USA and Canada by Routledge 270 Madison Ave, New York, NY 10016 This edition published in the Taylor & Francis e-Library, 2008. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.”

Routledge is an imprint of the Taylor & Francis Group, an informa business © 2009 Edoardo Zamuner and D. K. Levy for selection and editorial matter; individual contributors for their contributions All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data Wittgenstein’s enduring arguments / edited by David K. Levy and Edoardo Zamuner. p. cm. Includes bibliographical references and index. ISBN 978-0-415-44296-1 (hardback : alk. paper) 1. Wittgenstein, Ludwig, 1889-1951. I. Levy, David K. II. Zamuner, Edoardo. B3376.W564w5893 2008 192–dc22 2008018365 ISBN 0-203-88257-1 Master e-book ISBN

ISBN13: 978-0-415-44296-1 (hbk) ISBN13: 978-0-203-88257-3 (ebk) ISBN10: 0-415-44296-6 (hbk) ISBN10: 0-203-882570-1 (ebk)

To the memory of Denis Paul and Michele Ranchetti

Contents

Notes on contributors Abbreviations for Wittgenstein’s and others’ works

ix xi

Introduction

1

EDOARDO ZAMUNER AND D. K. LEVY

1

To represent as so

4

CHARLES TRAVIS

PART I

Mind

31

2

33

Wittgenstein, qualia, and the autonomy of grammar JULIAN KIVERSTEIN

3

Understanding (other) minds: Wittgenstein’s phenomenological contribution

60

SØREN OVERGAARD AND DAN ZAHAVI

4

Wittgenstein, the secondary use of words, and child psychology

87

MICHEL TER HARK

PART II

Language

107

5

109

Concepts and primitive language-games EVA PICARDI

6

On the teaching and learning of words

135

MICHAEL LUNTLEY

7

Meaning, rules, and conventions HANS-JOHANN GLOCK

156

viii 8

Contents Wittgenstein’s critique of referential theories of meaning and the paradox of ostension: Philosophical Investigations §§26–48

179

DAVID G. STERN

PART III

The absolute 9

The architecture of meaning: Wittgenstein’s Tractatus and formal semantics

209

211

MARTIN STOKHOF

10 De-transcendentalizing religion

245

ANAT BILETZKI

11 Morality without agency

262

D.K. LEVY

Afterword

281

12 Wittgenstein passages

283

DENIS PAUL

Bibliography Index

293 310

Notes on contributors

Anat Biletzki has been teaching at the philosophy department at Tel Aviv University since 1979. Her publications include Talking Wolves: Thomas Hobbes on the Language of Politics and the Politics of Language (1997), (Over)Interpreting Wittgenstein (2003), and articles on Wittgenstein, Hobbes, analytic philosophy, political thought, digital culture, and human rights. Outside academia Biletzki is actively involved in the peace movement and in several human rights projects in Israel. Hans-Johann Glock is Professor of Philosophy at the University of Zurich and Visiting Professor at the University of Reading. He is the author of A Wittgenstein Dictionary (1996) and Quine and Davidson on Language, Thought and Reality (2003). He has edited The Rise of Analytic Philosophy (1997), Wittgenstein: a Critical Reader (2001), Strawson and Kant (2003) and co-edited Wittgenstein’s Philosophical Investigations, with Robert Arrington (1991). Michel ter Hark is Professor of History of Philosophy at the University of Groningen. He published a book on Wittgenstein’s philosophy of psychology (1990) as well as a number of articles. He also published a monograph on Karl Popper’s indebtedness to the psychologist Otto Selz (2004). Currently he works on Wittgenstein’s notion of secondary use in relation to developmental psychology and synaesthesia. Julian Kiverstein is a post-doctoral fellow with the European Science Foundation collaborative research project, “Consciousness in a Natural and Cultural Context.” He works on issues in philosophy of mind and phenomenology. D. K. Levy is a Lecturer in Philosophy at the University of Edinburgh. His research concerns understanding, especially moral understanding. Areas in which he has published include Wittgenstein, moral philosophy, and cognitive psychology. Michael Luntley is Professor of Philosophy at the University of Warwick. His most recent book is Wittgenstein: Meaning and Judgement (2003). In

x

Contributors addition to Wittgenstein, his interests include the philosophy of thought and language, and the philosophy of education, especially the concept of expertise.

Søren Overgaard is RCUK Academic Fellow and Lecturer in Philosophy at the University of Hull. He is the author of Husserl and Heidegger on Being in the World (2004) and Wittgenstein and Other Minds (2007). Denis Paul was educated in Oxford where he became interested in Wittgenstein’s work. He helped G. E. M. Anscombe in her translation of Wittgenstein’s work as well as making the majority of the published translation of On Certainty (1969). His lifelong interest in and work on Wittgenstein’s Nachlass culminated in Wittgenstein’s Progress 1929–1951 (2007), published posthumously by the Wittgenstein Archives in Bergen. He died in 2006. Eva Picardi is Professor of Philosophy at the University of Bologna. She is the author of La Chimica dei Concetti (1994), Le Teorie del Significato (1999) and of various papers in the history of logic and in the philosophy of language, with a special emphasis on the work of Frege, Quine, Davidson, and Dummett. Together with Annalisa Coliva she edited Wittgenstein Today (2004). David G. Stern is Professor of Philosophy at the University of Iowa. He is the author of Wittgenstein’s Philosophical Investigations: An Introduction (2004) and Wittgenstein on Mind and Language (1995); and co-editor of Wittgenstein Reads Weininger, with Béla Szabados (2004) and The Cambridge Companion to Wittgenstein, with Hans Sluga (1996). Martin Stokhof is Professor of Philosophy at the University of Amsterdam. Together with Jeroen Groenendijk he has published on topics in formal semantics and pragmatics. He is also the author of World and Life as One: Ethics and Ontology in Wittgenstein’s Early Thought (2002). Charles Travis is Professor of Philosophy at King’s College, London. He has written several books, including The Uses of Sense (1989) and Thought’s Footing (2006). He works on a set of problems about thought, representation, and experience, spanning parts of philosophy of perception, of psychology, of language, of logic, and epistemology. Dan Zahavi is Professor of Philosophy and Director of the Danish National Research Foundation’s Center for Subjectivity Research at the University of Copenhagen. He has authored eight books including Husserl und die transzendentale Intersubjektivität (1996), Self-awareness and Alterity (1999), Husserl’s Phenomenology (2003) and Subjectivity and Selfhood (2005). Edoardo Zamuner is a research associate at the Institute of Neuroesthetics of University College, London. With D.K. Levy, he edited a scholarly edition of Wittgenstein’s Lecture on Ethics (2007).

Abbreviations for Wittgenstein’s and others’ works

AWL BB BT CEIA CV GT LC LW I LW II LWL MWL N OC PESD PG PI PI II PO PR RC RFGB RFM RLF RPP I RPP II TLP

Wittgenstein’s Lectures: Cambridge, 1932–1935 Preliminary Studies for the “Philosophical Investigations.” Generally Known as The Blue and Brown Books Big Typescript (page references are to the manuscript page numbers given in WA 11 and the Blackwell scholars’ edition) Cause and Effect: Intuitive Awareness (page references to PO reprint) Culture and Value (2nd edition) Geheime Tagebücher, 1914–16 Lectures and Conversations on Aesthetics, Psychology and Religious Belief Last Writings on the Philosophy of Psychology, Vol. 1 Last Writings on the Philosophy of Psychology, Vol. 2 Wittgenstein’s Lectures: Cambridge, 1930–1932 G. E. Moore: “Wittgenstein’s Lectures in 1930–33” Notebooks 1914–1916 On Certainty Notes for Lectures on “Private Experience” and “Sense Data” (page references to numbering in PO edition) Philosophical Grammar Philosophical Investigations Philosophical Investigations, Part II (where page references are given, these are for pre-2001 and 2001 editions) Philosophical Occasions 1912–1951 Philosophical Remarks Remarks on Color Remarks on Frazer’s Golden Bough Remarks on the Foundations of Mathematics (3rd edition) “Some Remarks on Logical Form” Remarks on the Philosophy of Psychology, Vol. 1 Remarks on the Philosophy of Psychology, Vol. 2 Tractatus Logico-Philosophicus

xii

Abbreviations

WA 4 WA 11 WVC Z

Wiener Ausgabe Band 4. Bemerkungen zur Philosophie, Bemerkungen zur philosophischen Grammatik Wiener Ausgabe. Band 11. “The Big Typescript” F. Waismann: Ludwig Wittgenstein and the Vienna Circle. Conversations recorded by Friedrich Waismann Zettel

Introduction Edoardo Zamuner and D. K. Levy

The initial impact of Wittgenstein’s philosophy was tremendous. His first work, coming in the wake of Russell, Frege, and Husserl’s work on logic, attracted considerable attention in Cambridge and Vienna. With the Tractatus Logico-Philosophicus, it seemed that an entire research programme in philosophical logic had reached a terminus. His later work, though not published until after his death, also had a remarkable impact. For a time, after his death, it seemed as if philosophy, at least in the Anglo-American academic tradition had everywhere been affected by Wittgenstein’s work. It has been called an epochal change.1 An entire generation of philosophers, some of whom learned at his knee, went forth to use Wittgenstein’s philosophical method.2 Philosophy, it seemed, might never be the same. How things have changed, if the doomsayers are to be believed. A recent series of speakers at the Wittgenstein archive in Bergen decried the present place of Wittgenstein’s philosophy.3 Not only are mainstream academic philosophers not interested, but students are discouraged from advertising their interest in Wittgenstein if they are to progress.4 Wittgenstein, it might seem, was a flash in the pan, his impact more to do with personality than philosophical substance.5 Doubtless every philosopher’s work is subject to the vagaries of academic fashion. Rather than give arguments for the importance or endurance of Wittgenstein’s philosophy, this volume offers demonstrations of the application of Wittgenstein’s philosophy. This volume claims a place for Wittgenstein’s philosophy in the Western philosophical tradition. His philosophy is neither an epochal change nor a temporary distemper. We propose that Wittgenstein’s philosophy will have an enduring place in the Western tradition, because his philosophy addresses basic perplexities that give philosophy its impetus. For example, immediate perceptual experience seems the sole certainty, but the certainty is an illusion because it is no basis for logical identity and, a fortiori, unity. For another, the experiences one has seem peculiar to oneself, yet for them to be thoughts rather than illusions they must not be peculiar to oneself. Each of the chapters of this book is an illustration of the use of Wittgenstein’s arguments for continuing philosophical debates, old and new.

2

Edoardo Zamuner and D. K. Levy

It is well known that there is a difficulty demarcating a work of Wittgenstein’s since his published works are a tiny fraction of the writings he left, his Nachlass. By Wittgenstein’s philosophy, we mean the arguments he gave in his work. However, there is a similar difficulty in specifying Wittgenstein’s arguments owing to his unusual working methods and style of composition. By arguments, we mean the considerations he adduced and advanced for understanding our experience of the world in one way—systematic or not—rather than another. No doubt this makes the reception and interpretation of Wittgenstein’s philosophy different and difficult. There is however no more concrete a demonstration of Wittgenstein’s arguments than their application to specific philosophical questions. The chapters of this volume are such demonstrations. Their character suggests that the enduring application of Wittgenstein’s philosophy will inter alia be characterized by: 1 Drawing examples and descriptions from throughout the Nachlass; 2 Showing movements of thought from one conception of representing to another; 3 Or replacing one description with another, with all that may imply. Insofar as the previous reception of Wittgenstein’s philosophy was not characterized by this approach to his work, a new era of Wittgenstein scholarship may be emerging. Our volume opens with a focal piece by Charles Travis. In it he ties Wittgenstein’s work to Frege’s, the origin of the linguistic turn in recent philosophy. More importantly, he engages with perhaps the key theme in Wittgenstein’s philosophy: representation. In one aspect, representation is cognitive—an activity of the mind—and it is for this reason that the first part of the volume is organized around the mind. This part comprises contributions by Julian Kiverstein on qualia and the grammar of phenomenology; Dan Zahavi and Søren Overgaard on phenomenology and inter-subjectivity; and Michel ter Hark on secondary sense and its implications for developmental psychology. In another aspect, representation is linguistic, where representations are bearers of symbolic properties that are used in understanding. For this reason, the second part is organized around language and comprises contributions by Eva Picardi on the place of primitive language-games; Michael Luntley on language and concept acquisition; Hans-Johann Glock on the normativity of language; and David Stern on referential theories of meaning. A complementary aspect of representation essential to Wittgenstein’s philosophy is that there are matters that cannot be represented. These are the absolute, such as logic, self, religion, and ethics. The absolute is the orientation of our third part, which counts contributions by Martin Stokhof on formal semantics; Anat Biletzki on religion without transcendence; and D.K. Levy on ethics. The volume concludes with a bookend, by Denis Paul,

Introduction

3

to whom the volume is dedicated. His reflections are less to do with the specific content of Wittgenstein’s philosophy than with the passage to understanding possible from doing philosophy the way Wittgenstein did. The topics addressed in the individual chapters are various and, mostly, contemporary. Had this volume been composed even ten years earlier, the topics would have been different again—ten years earlier, again different. Such are the parochial currents of professional academic philosophy. That being so, if this volume is a demonstration of the successful application of Wittgenstein’s philosophy, then it is prima facie reason for confidence that Wittgenstein’s arguments and his philosophy will endure. The Pre-Socratics, unlike the Sophists, and like Plato, were interested in the nature of reality. Specifically, they sought what is common to all things and thereby what makes discourse possible. Theirs was a quest for understanding what makes experience and its objects intelligible—that which is the same and that which changes.6 Wittgenstein’s pursuit of philosophy was no different.7

Notes 1 Henry LeRoy Finch, Wittgenstein, The Spirit of Philosophy (Rockport, MA: Element, 1995), ch. 1. 2 For example: G. E. M. Anscombe, P. T. Geach, G. Ryle, D. Dennett, and J. Searle. 3 See Alois Pichler and Simo Säätelä, eds., Wittgenstein: The Philosopher and His Works, vol. 17, Working Papers from the Wittgenstein Archives at the University of Bergen (Bergen: WAB, 2005), pp. 74–76, 167–68, 236. 4 Ibid., p. 168. 5 Ernest Gellner, Words and Things: A Critical Account of Linguistic Philosophy and a Study in Ideology (London: Gollancz, 1959). 6 Cp. PI II, xii. 7 See “The Fundamental Problems of Philosophy,” in D. Z. Phillips, ed., Wittgenstein and the Possibility of Discourse, 2nd ed. (Oxford: Blackwell, 2006), Preface.

1

To represent as so Charles Travis

“Agreement … in judgement is part of communication.”1

Throughout, Wittgenstein had Frege in mind. We should too, to understand him. This is as true for Philosophical Investigations as for the Tractatus. In fact, the later work is, in an important way, closer to Frege than the first—even though the Philosophical Investigations makes a target of what seems a central Fregean idea. It directs Frege’s own ideas at that target, using something deeply right in Frege to undo a misreading of what, rightly read, are mere truisms. The Tractatus presents a view of what it is, essentially, to represent as so. The Philosophical Investigations, I will suggest, presents a different view, but on the same topic. Wittgenstein, of course, rejects essences, on some conceptions, in some employments: “You speak of language games [ … ] but never arrive at saying what the essence of a language game, and thus of language, is. What is in common to all these processes and which makes them all language, or parts of language. So you let yourself off that part of the investigation which used to give you the biggest headache, namely what to say concerning the general form of a proposition and of language.” And that is true.—Instead of specifying what is in common to all that which we call language, I say that all these phenomena have nothing in common in virtue of which we use the same word for all [ … ]. (PI §65) We see that what we call “proposition,” “language,” is not the formal unity I had imagined [ … ]. (PI §108) But he also says, “It is part of what communication is, not only that there is agreement in definitions, but also (strange as this may sound) agreement in judgements.”2 It is hard to do philosophy, even of a new sort, while espousing no commitment to what things—in this case, propositions, representing—are

To represent as so

5

as such. As we shall see, what I have just quoted is a view of this. The Philosophical Investigations opens with a fundamental point—a thesis— which paves the way for this.

1.1 The opening move The Philosophical Investigations makes its first point by deploying a notion, language game. The items it thus deploys to make its point are fictions, precisely what they are said to be; best thought of as identified by how they are to be played. It will help here to think of each as containing some specified set of moves, each move governed by specified rules, these jointly fixing how the game is to be played. A game is thus just what its rules make it. Those rules spell out a notion of correctness: a move is correct, on this notion, if it conforms to the rules which govern it. A language game may have a given point; something to be achieved in playing it. Such can sometimes give sense to the idea that there is something its rules ought to be. For present purposes, language games are (as Wittgenstein repeatedly insists) objects of comparison. We speak of them to serve philosophic ends.3 To think of them as fixed by their rules, apart from any playing of them is just to assign them a particular role in the comparisons to be made. One thing they may accomplish, thus conceived, is to make perspicuous ways for representing to connect to action. For this end, and for suitably chosen games, we may think of the rules governing a move as dividing into introduction and elimination rules. An introduction rule specifies under what circumstances a move would be made correctly. It may say, for example, that a player may say “Slab here” only if there is a slab at some specific place. The elimination rule specifies what must, or may, be done if the move in question was correctly made. The prescribed consequences of the move may just be more talking. But they may be actions of various non-linguistic kinds. A correct move “Slab!” in some game might, e.g., require bringing a slab to a certain point. In ways I will try to bring out, connections with action are central to a representation’s content being what it is (or, indeed, to its being anything at all). So, that a language game may connect words with action in a given way is an important feature of it, given the purposes it will serve in the Philosophical Investigations. I begin now on the basic point. One half of it is stated in §10: Now what do the words of this language designate?—What is supposed to show what they signify, if not the kind of use they have? And we have already described that. So we are asking for the expression “This word designates this” to be made a part of the description. In other words the description ought to take the form: “The word … designates … ” [ … ]. As said already, the fundamental point at the start of the Philosophical Investigations departs from Frege. But we see already how the point is made

6

Charles Travis

by taking very seriously another on which Wittgenstein and Frege agree. Frege made the point this way: What is distinctive about my conception of logic is that I begin by giving pride of place to the content of the word “true,” and then immediately go on to introduce a thought as that to which the question “Is it true?” is in principle applicable. So I do not begin with concepts and put them together to form a thought or judgement; I come by the parts of a thought by analysing the thought.4 A thought, for Frege, is, in this sense, the fundamental unit of analysis. In just this sense, a whole move in a language game is the fundamental unit of analysis when it comes to content. What does this mean? Suppose that, taking a thought (judgment, proposition) as fundamental, we ask the question whether such-and-such is an element of such-and-such thought—for example, whether Frege, or something naming him, is an element of a proposition that Frege smoked. If the proposition comes first, we can begin with when it would be correct, in that distinctive way in which a proposition, as such, is liable to be correct. For a given proposition, there is such a thing as things being as they are according to it. The distinctive sort of correctness, seen one way, just consists, for it, in things being as they are according to it. So thinking, we can say: for Frege, or something representing him, to be an element of a given proposition just is for him, or it, to make (or be correctly viewable as making) a particular identifiable contribution to when things would be as that proposition has them: whether things being as they are is their being that way turns on how Frege thus is. The proposition has an element naming Frege just where that element so works to make the condition on that proposition’s correctness what it is. As Frege notes, what the elements of a given proposition are is all relative to an analysis. There are two points here. First, there might be various alternative ways of carving up a whole proposition into parts. For to carve a proposition into parts is just to divide up its main task—being true (correct) just when it would be—into identifiable sub-tasks: the elements, on that carving, are just what perform, respectively, these sub-tasks. A thought that Sid grunts would be true just where Sid grunted. So it would be true just where Sid is a given way. A sub-task can thus be making its truth thus turn on Sid. It would be true just where someone was a grunter. So a sub-task can be to make its truth turn on whether someone or other grunts. These tasks, if performed jointly, make the whole true just where Sid grunts. To be an element of a proposition on this analysis just is to perform (or perhaps be) one of these sub-tasks. That a proposition’s task can be parceled out into subtasks in some given way does not preclude also parceling it out differently. Any such parceling out what the proposition does into sub-tasks would be a correct analysis of it, on which what performed some sub-task thus identified would be an element of the proposition.

To represent as so

7

If elements are what one thus carves out of a proposition, on some division of its job into sub-tasks, there is no intelligible question as to how the elements of a proposition (on an analysis) can form that unity which the proposition itself is. For there is nothing it could be for something to be an element other than to contribute to that unity. The element has no identity other than as that which performs such-and-such sub-task. The second point here is Frege’s context principle.5 An element of the thought that Sid grunts can make the truth of that whole turn on whether Sid is some way or other only in concert with some other element making the truth of that whole turn on whether someone is such-and-such way—say, a grunter. There is no such thing as making truth turn on how Sid is apart from any way it is so to turn. Words which made the truth of a whole turn on whether Sid was some way might be said to name Sid. Similarly, words making the truth of a whole turn on whether someone was a grunter might be said to name being a grunter. By analogy one might say the same for an element of a proposition, or of a thought. Where naming is of logical significance, to name X just is to perform some such role. Naming so conceived could be done only in the context of a whole in which elements, in naming what they did, jointly performed the task that whole did—here being true just when that whole would be. In that sense, naming cannot be done in isolation. This is the context principle. A move in a language game is the basic unit in the game in just the same way as a proposition (thought) is the basic unit in thinking. For something to be an element of a move is for it to contribute, in some definite way, to the conditions for the correctness of that whole move. All the above remarks apply to this. This is Wittgenstein’s point in §49. For an element of a move to name something just is for it to play a particular logical role in that whole move—to have a particular place in a particular parceling out of the task of the whole move into sub-tasks. If an element of a move names X, does that fact identify a unique role as the one it thus performs? That is the main question to be addressed. First, though, I will pause for a moment over a different conception of what it is to be an element.

1.2 Occam’s Razor This second conception is what Wittgenstein refers to in the Tractatus as “Occam’s Razor.”6 He stated it clearly in January 1930.7 It is there put thus. For something to be an element in a proposition is for that proposition itself to be an element in some specific system of propositions. For the proposition to contain that element is (inter alia) for there to be, within that system, a range of propositions which are the same in some determinate respect. The element thus shared is, or identifies, that respect. For there to be that element is for there to be that way for those propositions (the range) to be the same. For it to occur in that proposition is for that proposition to belong to that range, within this system.

8

Charles Travis

To be an element of a proposition, on Frege’s view, just is to contribute, in a way there is for an element to contribute, to the truth conditions of a whole. That way of contributing, Frege insists, is detachable from its occurrence in that whole. There could be a range of cases of an element contributing in that way. So far, this does not require there to be such a thing as the range (or set of them), or the system, which is that to which the whole thus analyzed belongs. For one thing, for Frege, the same proposition may be analyzed in many ways—and would fit in different ranges of propositions on each. Nor is it some Fregean quirk to allow for multiple analyses. If a proposition were, essentially, such-and-such construction out of such-andsuch building blocks, it would not have the pride of place Frege sees for it. One could give no more than lip-service to that idea of pride of place. Tractarian Wittgenstein cannot share this view. For, as he says in January 1930, “It is the essential feature of a proposition that it is a picture and has compositeness.”8 That is part of his conception, there, of what it is for a proposition to represent: it represents what its elements name as structured just as that proposition is by its elements. There is an identity of structure between that of what is so if the proposition is true and that proposition itself. Such a proposition must be built up in just one way, of just one set of parts. Different structurings of parts are ipso facto structurings of different propositions. Merely insisting that a proposition occupies a particular place in a particular system does not guarantee that this is so. A whole system of propositions might divide into sets of ranges, each of propositions the same in some respect, in different ways on different analyses of it. The whole system might admit of multiple analyses just as, for Frege, a proposition does. If it is the system that is to guarantee the uniqueness Tractarian Wittgenstein needs, something else must guarantee that the system decomposes uniquely into some one set of such ranges. We are left with this in common to Frege and Tractarian Wittgenstein: an element of a proposition performs a role which is one for an element to perform in a proposition; a role thus admitting of further exemplars. A role is essentially detachable from an occurrence of it. Let us now put this idea to new use. I have spoken, so far, of two ideas of an element, each of which leaves it open that the same proposition might be structured in different ways, out of different elements. What, aside from a particular structuring of particular elements, identifies a given proposition as the one it is? We began with one feature of a proposition: when things would be as it has them. We might cite that as an answer. But there is another (non-rival) view available. Frege remarks in “Der Gedanke”:9 to judge something is to expose oneself to the risk of error. (This happens in two steps. To judge of an environment is to run the risk of error; an environment—as he has just argued before making that point—is all there is, or could be, to judge about.) A proposition is the content of a judgment, detachable, Frege insists, from any particular judging of it.10 So a proposition is a way for a thinker to be exposed to risk of error.

To represent as so

9

The idea of an element as a respect in which some range of items are the same can now come into play. A proposition represents a particular, determinate, exposure there is to suffer to the risk of error. So it is a way in which a range of things would be the same. Which things? Each of us, at a time, occupies his own position of exposure to risk: each of us is exposed to risk of error, and exposed as he is. For each, there is his own distinctive way in which plans may go awry, actions may miss their mark, surprises may be in store. If today is Friday, I am in my own particular sort of trouble. (Proofs due by noon.) A proposition represents a determinate way for such distinctive individual exposures to be the same; so a range of cases of a thinker being exposed in the same way. The proposition identifies, and is identified by, that understanding of the same. (This is already to rule out the possibility of a private language). I think today is Thursday. I am not alone. Each of us who thinks so is thus exposed to a different specific form of trouble. Which of us, in risking what we in fact do, instances just this particular shape of stance toward things? What risk would one run in judging today Thursday? I leave that, momentarily, open. Language games are apt for capturing specifics of exposure to risk of error. A move in one identifies common elements in different thinkers differently than a proposition does. If a proposition is that the flat is dark, what risk does one run in supposing it? When would one be running just that risk? Well, you risk your hand just wherever your answer to the question what the thing for you to do is depends on whether things are as that proposition has it. Where is that? Consider the film-developing game. Two players go through a series of moves, at the end of which, if all is well, film has been developed. In the game there is a move, “The room is dark.” By its elimination rule, player 2 may remove film from its canister and place it in developer. When one removes film from a canister (for this purpose), there is a way in which darkness matters to whether one’s project will go awry. So, with this project in view, there is a way in which darkness matters to whether removing the film is the thing for one to do. If we were designing a game that was to have this envisioned end, then, that elimination rule settled, there is something the move’s introduction rule ought to be. Let us now suppose we have made the introduction rule accordingly. You are playing the game. The room is dark enough for you to bump into things unintentionally. Objects loom as shadows. May you make the move, “The room is dark,” correctly? Not according to the rule we would thus have made. Those conditions would spoil film. If, in the game, you make the move, “The room is dark,” you are in error just in case you breach the imagined introduction rule. But now that will be just where relying on the room’s being dark where it bears on the thing for you to do would lead to doing the wrong thing—what missed its mark— where that thing was what the elimination rule licensed, and the mark was

10

Charles Travis

what showed here what the introduction rule ought to be. In the case at hand, this is just where opening the canister would spoil film. What of our envisioned proposition that the room is dark? Again, in supposing it you are in error just where, in fact, things are not as they are according to it. But that introduction rule for it has not yet been linked to any elimination rules by the route just covered for the above move in the game. So would (supposing) that proposition be in error in the circumstances just envisioned—objects looming as shadows, barked shins, etc.? Nothing said so far about what proposition it is decides this.

1.3 Frege’s limitations Consider the proposition that Sid grunts. How might this decompose? Perhaps into a part which makes it about Sid, and one which makes it about being a grunter. Making a proposition about Sid could count as naming him. Similarly for naming being a grunter (a way for someone to be). To be an element of a proposition is to play a logical role. So to name Sid is to play some logical role. To be an element of a proposition is to make some determinate contribution to its condition for correctness (truth). It is to be part of a particular way of parceling out that condition into sub-parts. Is to name such-and-such to make some one such contribution? If I say that an element names such-and-such, have I thereby said what its contribution to its whole is? Here we come to an idea of Frege’s which, from late Wittgenstein’s point of view, is a bad one (on one straightforward reading). It damages the good idea of a proposition enjoying pride of place which Frege and Wittgenstein share. If we accept everything Frege says on this question, then what an element of a proposition names does determine what its contribution to its whole is. There is just one such contribution thus to make. For Frege, in the simple sort of case now on hand, logical roles fall into two categories: naming objects, and naming concepts. If, in the present case, there is an element which names Sid, then there must be another which names a concept. This would be the element proposed, above, as naming being a grunter. Naming Sid is identifying what it is that has to be some way or other (for the proposition to be true); naming that other element is identifying which way something or other must be. So far, so good. Now the fatal step. For Frege, a concept is a function, namely, one from objects to truth-values.11 If the words “Sid grunts” decompose into an element “Sid,” which names an object, and an element “grunts,” which names a function from objects to truth-values, then the whole, “Sid grunts,” names the value of that function for a certain argument, namely, Sid. Which is to say that it names a truth-value: true if Sid grunts, false if he does not. Which is to say that for parts jointly to play these roles is, ipso facto, for them to decide a unique and determinate truth condition for their whole. Mutatis mutandis for propositions, of which words “Sid

To represent as so

11

grunts” could be but one instance, or expression. Which is why Frege’s answer to the question must be yes. Not that taking concepts for functions is at all quirky. It is a natural reading of a truism: to apply a concept to (predicate something of) an object is to say (or think) the object to be thus and so; which it is (if it is) or is not (if not) purely in being as it is. Only the object’s being as it is makes that predication true or false. On the reading of this truism, a concept thus behaves as a function would: fix the object it applies to, and you fix the truth-value such application takes. To see this as misreading, one must first find another. To which task we now turn. In matters of decomposition (on the view late Wittgenstein and Frege share), a proposition and a move in a language game raise exactly the same questions. So we may now ask this. Suppose there is a move in a language game—say, “The room is dark”—which we have decomposed, somehow, into parts, and that, for each part we have identified something it in fact names (on that analysis). For example, on our analysis, “the room” names a certain room, “is dark” names (the property, or concept, or whatever, of) being dark. Have we fixed what the correctness (truth) condition for the move is? Do our parts, in jointly so functioning, uniquely, or univocally, fix some one such condition? Clearly not. For we can imagine two games. In the first, if I have made the move, “The room is dark,” correctly, then I am excused from carrying out certain orders. For example, I need not bring you your copy of Zettel, on grounds that I cannot see to find it. In the second, if I have made the move correctly, then you may remove the film from its canister (and rely on doing so safely). Let us suppose that each game is well constructed: the move is made correctly in it just in case there would, in fact, be those consequences (or at least such consequences would be to be expected). Then the correctness conditions for the move must differ from the one game to the other. For I may be, reasonably, unable to find your copy of Zettel (knowing it is on the shelf) where it is quite unsafe for you to remove the film. One might correctly say of either game that in it “is dark” names (speaks of) being dark. But the role it plays in naming this differs from the one game to the other. In the one game, but not the other, it contributes to a condition on being as said which is not satisfied if, where whether to remove film turns on whether the room is dark, removing film is not the thing to do. So if a move consists (on an analysis) of parts, for each of which there is a suchand-such it names, those facts about the move are compatible with any of indefinitely many mutually exclusive conditions on correctness (answers to the question when things would be as thus said). The point is general. That words name such-and-such determines no unique contribution which is that which such words make to conditions on the correctness of the whole they thus are part of. Such is just part of what naming is. It holds equally for naming in the context of a move in a language game, and naming in the context of a proposition. The fact of my speaking

12

Charles Travis

of being dark is compatible with my saying any of many things in, and by, doing so. There are many different things, each of which being dark may, sometimes correctly, be taken to be (or come to). Being dark admits of understandings. The same goes for being a grunter. On Wittgenstein’s new view of representation, these are to be taken as illustrating something intrinsic to the idea of a way for a thing to be; if a concept is identified by what it is of, then something intrinsic to the notion concept. Philosophical Investigations opens with this point. It is the first point the idea of a language game is used to make. It advances us far toward our emerging new picture of representing as so. We needed a new reading of a truism to see how it could be a misreading to make concepts into functions. We now have it. It is on an application of a concept to an object, on an occasion, that one says the object to be thus and so. The rest of the truism then holds. The concept as such admits of many applications, each excluding others. So it alone cannot assign an object, in being as it is, a truth-value.

1.4 Comparisons Words may name in the context of a proposition (or its expression). May they also name in the context of a language (such as English or French)? Not in the same sense. For a language asserts nothing, thus neither correctly nor incorrectly. Our fundamental point about naming makes this no mere nicety. The words “is dark,” in the context of a move in a particular game, contribute to a particular condition of correctness. That is what it is for them to name, in the sense in which they may name in such a context. What the English words “is dark” mean does not choose between different conditions of correctness such as the two just indicated. The English “The room is dark” is governed no more by one such condition than by any other. So there cannot be the proposition they express. They can be used to express various propositions, where the use in question fixes just how one may be in error in speaking of such-and-such as dark; some one such condition on correctness. But there is truth to tell in saying the English “is dark” to name (speak of) being dark. Here naming is not playing a role in some given thought. Aspects make the difference here. There is a switch in my kitchen which turns the dishwasher on and off. This does not make me apprehensive as to what the switch may be up to while I am away. If I did worry, I would be missing the aspect with which “turns on” occurs in that expression of truth. The point thus made is: to turn the dishwasher on, or off, flip that switch. Turning the dishwasher on and off is what the switch is for. Similarly (at least at first approximation) for the English “is dark.” English does not go around telling people things. I need not worry whether English can keep an Oxford secret. But if you want to tell someone something, and it is to your purpose to speak of something being dark, then, if you are speaking English,

To represent as so

13

uttering the words “is dark” will generally be a way of doing so. That is what they are for. Crucially, there is no particular thing you have to be in course of saying in order to use “is dark” to speak of being dark, just so long as speaking of being dark has a role in saying it. It does not matter whether you are speaking of a room being dark in a way that would license opening film canisters, or merely in a way that would excuse your failing to find Zettel on the shelf. English is indifferent to such things. It is, as one might say, plastic. Which is why it is so useful. To put things (so far) picturesquely, we might say: the words “is dark” would speak of what they do in English, in any move in any language game, provided only that, in that move, they speak of (name) being dark. What language games are played in some speaking “The room is dark” (in speaking English)? The notion playing a language game in speaking English is not yet defined. It is (normally) not as if, where I say, “The room is dark,” we have agreed to speak according to such-and-such rules. Nor does English provide the rules of some game for us then to play. But we can think of playing a game in another way. If, in given circumstances, I say of a given room, “The room is dark,” the rules of a game with that rule may be such that I would thus have spoken truly only if I would have moved correctly by those rules. Such rules identify some of what it would be for things to be as I said. We might speak of that move, so that game, as modeling my words. In the game, if “The room is dark” is said correctly, another player may remove film. A parallel condition can govern what someone said: things are as he said only if removing film (for developing) can be the thing to do, insofar as this turns on the present darkness of the room. (That there is no other way the film will be spoiled need not be in this bargain.) Such may be a condition on the correctness of what one says in calling the room dark. It contributes to an understanding of what it would be for the room to be as thus said. When I call the room dark, I may be said, in this sense, to be playing any game which thus captures some of what it would be for things to be as I thus say. By that measure, I might correctly be said to be playing indefinitely many different games. Where I speak of the room being dark, I must do so on some understanding of it so being. The language games I thus play, on our current understanding of playing one, may make explicit some of what that understanding is. Their elimination rules show some of how things would be if as I said: what is licensed or demanded by them is what one can, or must, do so far as that turns on things being as said. Their introduction rules must then be what these elimination rules require. If I said things to be thus and so, things would be that way, on the understanding on which I said so just where those introduction rules permitted saying so. Where I speak of a room as dark, there are things you may expect if I am right, among which ways you may, on the occasion, expect this to bear on the things for you to do. Licensing such expectations is a way of exposing

14

Charles Travis

oneself to risk of error. An identifiable such exposure may be a common element in a range of cases of thinkers taking things to be as they do. It may thus identify something to be thought or said—a proposition. For a given way for things to be—thus and so—there are many such exposures to risk one may incur in taking things to be that way, so many propositions. There are many ways of exposing oneself in taking a room to be dark. Supposing this point to hold for any things words might name (in speaking of sublunary affairs), and a concept to be what one names in speaking of a determinate way for things to be, a concept cannot be a function from objects to truth-values. The fact of having named one cannot by itself fix when what one thus said of something would be true of it. Logic is so built that, when it speaks to particular thoughts, it requires treating what plays the role of a concept (on an analysis of a thought) to be such a function. Which means that when we apply logic, that to which we do so must be, for our purpose, viewable in those terms. So far as matters to an application, it may be as if “is red,” as such, has some definite extension. Concepts are sometimes treatable as functions. That is another way an identification may be understood. Which suggests, perhaps, another way of reading Frege. There are as many games to play in calling a room dark as there are things one might understand by a room being dark. Who is “one” here? When I speak, there are those my words are for; for whom they should be usable as they ought to be. I cannot speak to people on understandings of being dark they cannot grasp, or could not recognize as understandings of being dark. On the other hand, for anything those my words are for would sometimes understand being dark to be, I can sometimes speak of being dark on that understanding of it. This remark fits together with another. We sometimes recognize someone as having spoken of a room as dark, while also recognizing certain things as to when things would be as he said. We are constrained here by the limits of our ability to understand being dark in different ways. Those limits are also the limits of our ability to take an audience as having taken being dark to be what was spoken of. Our limits here, and those of those for whom I speak, are fixed by an entirely parochial sense for when it would be being dark that was in question. To call such a sense parochial is to say that sharing it is not part of being a thinker as such. A being would not fail to be a thinker merely in lacking it. The sense in question is part of our equipment for engaging with the world; for all of which, what it helps pull off is engagement with the world. Where I did say “The room is dark,” what standard of correctness governed it? Only the parochial can answer this. I tell you the room is dark. You pry the lid from the canister. The light on the smoke detector blinks. No one will ever see that judge’s peccadilloes. Did I speak on an understanding of being dark on which the room was that? Is there such an understanding? If I spoke understandably, there are those equipped to understand me. There are things such fellows would be prepared to recognize—say, that there is an understanding on which the room was dark, and another on which it was

To represent as so

15

not; that, circumstances what they were, I ought to have been taken, or one had a right to take me, to be speaking in the second way. In which case, things were not as I said. There need not always be such outcomes. Those for whom I speak need not always see such things in one way rather than another. But sometimes there are. I call the room dark. The canister is opened. The tabloids have their field day. Is there an understanding of being dark on which, for all that, the room was not dark? That may be hard to say. But should I be so understood that things being as they were is things being as I said? About that there may be no doubt. We have arrived at the new picture of what representing is. What I said to be so in representing things as some given way is what it ought to be given how one ought to expect to be able (or need) to act on what I said as the introduction rules of a good game with a point are what they ought to be given the elimination rules they mesh with. Questions as to what it is I said—where I spoke of things being thus and so, on what understanding of their so being I in fact spoke—are thus answered by our common parochial sense for what ought to be expected of my speaking as I did in the circumstances in which I did—in the artificial terms here used to make the point, by our parochial sense for the games I was, or would be, playing in speaking as I did, on our present notion of what it would be for me to be playing a language game in speaking, say, English.

1.5 Truth A further contrast with Frege’s view may help focus the new picture. Frege’s conception of a truth bearer starts from this question: “Is a picture as bare visible, tangible, thing really true? And a stone, or leaf, not true?”12 The answer is, clearly, no. A picture, so conceived, might represent anything, or nothing, as so. So far, as Frege sees it, there is no intelligible question as to whether what it represented as so is so. There is no “what it represented as so.” And (a critical assumption), one cannot ask whether words represented as so what is so except by asking of that which they represent as so (that such-and-such) whether that is so. Suppose a painting does represent something as so—something, say, as to how the cathedral at Rouen looks. No question of truth arises, the idea is, until it is fixed just how the cathedral is thus supposed to look. Only with that fixed can we ask whether the cathedral is as represented. Suppose the painting is a Monet. In its image of a wall is a patch of blue paint. Does this represent it as so that the cathedral has a partly blue wall? Or that the wall was in shadow at the time it was painted? Monet’s ways aside, there is no saying. If not, the idea is, no (determinate) question of truth has arisen. A question of truth thus arises for Frege only where there is no unanswered one of this form: “Is the representation to be taken as representing this as so, or rather that?” For what raises a question of truth, there is no

16

Charles Travis

such open question as to what is so according to it; when things would be as it represents them. Frege thinks that beginning with visible, tangible representations we can arrive at non-visible, non-tangible, representations which meet this condition—and that any representation which did so would be neither visible nor tangible. This conviction is another version of the idea from which I have already signaled Wittgenstein as departing: the idea that a concept is a function from objects to truth-values. Suppose there is an identifiable way for things to be such that if things were represented as that way, there can be no further open question as to whether they were thus represented as this way, or rather that. Such a way for things to be would admit no understandings. The zero-place concept of it would thus map things being as they are into a truth-value. Where naming a (non-zero place) concept is a contribution to representing things as this way, the concept named could only behave as a function from objects to truth-values would. What raises such a question of truth is a representation on a different understanding of that term: something to represent as so. Such a thing (as Frege argues) is neither visible nor tangible. That the parochial decides which way words represented things is, for Frege, an anodyne idea. Whether “is dark” then named this concept or another may all depend on us. But, on this conception, there is no role for the parochial once a question of truth has been raised. From the question when someone should be counted as having spoken truth—as having said no more than what is so—we have separated out a question which does not refer to him, or his utterance, at all: for a given thing that might be said, when that thing would be true. On the Fregean idea, such a thing—something to represent as so—is not susceptible to understandings, as mere visible, tangible representers—words, pictures, etc.—are. It is an understanding for such things to bear. So there is no role for the parochial, or anything else, in choosing between understandings of it. A language game provides a different story. We can now abandon the idea that there are ways for things to be which brook no competing understandings; and the idea that the only questions of truth are whether things are one or another of these ways. For any way for things to be, there may always be questions as to how one may expect to act on that, admitting of different, and competing answers on different occasions for considering a question how to act. On the other hand, on the new picture, words are true if one can do all they license, must do all they require if correct. How things are according to the words is then fixed by those standards for their introduction which such standards for their elimination impart. If a painting represents part of a cathedral wall as blue, there is this question: Might it be true to represent the wall like that? The question may have a clear answer. Or, again, not. There are all sorts of things it could be for a wall to be blue there; all sorts of understandings of it so being. Would a crude graffito count? For each such thing, there are the language games in which it is what would so count. That there are such further understandings

To represent as so

17

of being blue need not matter to truth in the case at hand (given the way the wall in fact is), though it might. As to whether the painting is subject to any such further standards of correctness—whether it is the playing of any such further games—this is a matter for the parochial to rule on, just as it does where Frege concedes it room to operate. My calling the room dark raises a question of truth: Is it dark on the understanding on which I spoke of that? This question still leaves room for the parochial to work. There are things to be decided as to the understanding of that on which I did so speak. Now, the idea is, say what you like as to what that understanding is, and there is still the same sort of room for the parochial to work. There is always room for fresh questions of truth to arise. Find new circumstances of which my words might be true or not, and there are, correspondingly, new language games for me to have been playing or not; games which provide for such circumstances in any of many ways (or none). With that, there is more of the usual work for the parochial to do. The parochial, like rust, never sleeps. On this new picture, as opposed to Frege’s, it never cancels out absolutely. Invisible, intangible truth bearers thus lose the special status Frege accords them. The parochial takes on an ineliminable role in representing as so. Such a view must be seen as a picture of what representing is as such—though it is as yet unclear how it could be that. Which sets an agenda for much of what follows this opening move in the Philosophical Investigations. I now turn to that agenda, and one bit of its execution.

1.6 Thinking, meaning, understanding So far I have discussed the work of the first 27 paragraphs of the Philosophical Investigations. It sets an agenda for the rest—not the only agenda for it, but at least the agenda for a significant part of it: the rule following discussion.13 There are two issues on that agenda. Here is the first. I spoke of a picture of what belongs to representation as such. So far we have considered only linguistic representation: representation in, or by, vehicles. I utter some words, or make some signs by waving my hands, or use smoke signals, or whatever, to which one is to react in a certain way. The signs—the physical bearers of this way of being taken—make my representing perceivable, so that you can be aware of being represented to, and then, if things go well, take yourself to be represented to in a certain way. Where there is such a thing as how signs are to be taken, a parochial sensibility may decide that. Wittgenstein’s point could then just be that there are no bounds on the room for such decision—what Frege suggests there must be for there to be questions of truth at all. But representing with such a physical face to it—with bearers of content— is one case of two. There is representing to, or for, an audience—those who are to take the representing in one way or another. Then there is representing things to oneself as so. One can be the sole intended audience for representing

18

Charles Travis

of the first type: I leave myself a note (which, on finding, I may disbelieve). But there is a case where it is not that. Representing something as so may just be taking (holding) a certain view. I will call that autorepresentation. For me to think that my keys are in my pocket is, per se, for me to represent the world to myself as a certain way. For me to represent the world to myself as a certain way is for me to stand toward it in a certain way—in a certain posture. To write a note would not be so to stand. Representational vehicles have no clear role in my so standing (though some have postulated them by way of supposed empirical hypothesis). Nor, in any clear sense, is my representing something to myself as so something that is to be taken in one or another way, so that a parochial sensibility might decide how. Where there is something which is to be taken in a certain way, there is a role for the parochial to have: determining what way. But autorepresentation seems to involve no such thing. So it is as yet unclear what role the parochial could have there; so how the new picture of representing—which was to be of what representing is as such—has any application to this central case. The first item on the agenda is to see how it might. The second is to see why the new picture is the right one. This section takes up the first item only. The next deals with the second. By way of preview, if when things would be as said in saying them to be thus and so is fixed, somehow, by what we, as thinkers of the sort we are, are prepared to recognize, this can seem to threaten the objectivity of judgment; hence its very possibility, “subjective judgment” being an oxymoron. An answer to the second question should pave the way to seeing why such apparent threat is mere illusion. But I will not treat this last matter in full here. Philosophical Investigations §§84–87 is the first discussion of following rules. It discusses how the explanations we give one another can explain— can say no less than what to do, or how things are. I may explain to you, “To find the post office turn right at the second street,” or, if that does not suffice, “That alleyway is not a street.” I may tell you no less than what route to follow. The explanation, “Turn right at the second street,” of course admits of interpretations, as does every other. But my explanations, more generally, the things I say, are for thinkers of the sort we are. Such thinkers would take them in a certain way, or are prepared to as part of being the sort they are. What explanations which are for such thinkers in fact say is what they would thus be taken to. This is a core idea which I will appeal to in approaching the first agenda item. For want of space, I will omit anchoring in text, though I take there to be such. I will simply answer the question posed. How does the new picture apply to autorepresentation? I will focus on one case of it: someone meaning his words in a certain way—so as to say such-and-such. Here meaning is what someone does, not what “vrijgezel” does in meaning bachelor. It is non-factive, so unlike Pia’s frown meaning Sid is driving. I meant to post the letter. I meant to say “begleiten,” not “beledigen.” It aims at something, with all the usual accompaniments of that. If I mean to say X to be so, and for X to be so just is for Y to be so, I

To represent as so

19

need not have meant to say Y to be so. Which need not mean that I can mean to say something where there is no such thing to say. What can one mean the words one speaks to say? On one view, anything one can think. Thus, as Wittgenstein portrays the view “But it is just the queer thing about intention, about the mental process, that the existence of a custom, of a technique, is not necessary to it.”14 The counter idea here can be put: Ignorance is strength. A small child can jump off a tall building and try to fly—if, as may be with a small child, he does not know that he cannot do this. But only if that. I can jump off a tall building and flap my arms. But I cannot thus be trying to fly. For I know I cannot. You cannot try to do what you are fully convinced you cannot do. So I cannot, in the bakery, try to ask for bread by saying, “Would you like to go for a drink?” At least not in normal circumstances. For, as I know full well, I will not have asked for bread in doing that. English does not work that way. I am a competent speaker of English. So I know what I will say (if anything) in the bakery in uttering “Would you like to go for a drink?” So, contraposing “Ignorance is strength,” I cannot try, in so speaking, to ask for bread. So I cannot mean my words to be asking for bread. Insofar as I am competent—I know what would be said in speaking given words on a given occasion—I can only mean my words to say what they will, in fact, say. If I say “The room is dark” in circumstances in which that would be taken to mean it is safe to remove film, and I know this is how it would be taken, I cannot, consistent with that, mean my words to say that the room is dark on some other understanding of it so being. The new picture of representation now applies. What would I say, on a given occasion, in saying, “The room is dark”? Refer to the new picture. Now insofar as I am competent, that is what I meant my words to say. My readers and I are all competent in speaking English, but none of us perfectly so. Even where competent enough, we do not always see what we will say in uttering given words on a given occasion. We may always fail to appreciate fully the circumstances in which we speak. I may, say, misuse the word “careen” because I do not know, or properly grasp, its nautical origins. Or I may misuse it in a particular situation because though I do know those origins, my audience does not, so that I am aiming for a metaphor unrealizable under the circumstances. A boat tilting from side to side in high seas may be said to be careening (on what is now, anyway, one recognized use). I might think of a truck heaped high, creeping from the dust bowl toward California, as like such a ship. So I speak of it as careening down the road. The metaphor falls flat. I do not say what I mean to. Equally, I might threaten someone, “I’ll careen you,” and not succeed in saying what I mean to say—that I will keelhaul him—because that meaning of “careen” is currently unknown. There is that much room for divergence between what I mean my words to say and what they do say. None of which spoils the core in the idea that ignorance is strength. I thought “careen” meant keelhaul. In the context I

20

Charles Travis

used it, it did not. But we know what it would be (roughly) for a word to mean keelhaul. The word “keelhaul” means that, for example. Suppose “careen” had meant keelhaul (in that context). What would I then have said? In the above case, that is what I meant my words to say. Again, the new picture of representation stays in place. Suppose I drag you under my canoe. Would that be keelhauling on the understanding of this on which that is what I would have threatened my audience with? My perceptions as to what I would say in given words may intelligibly deviate from what is so. They may be, understandably, that I would say suchand-else there is to be said. Then, in choosing my words as I did, that is what I meant to say. I was able to suppose I would then speak, in those words, of things being such-and-such way. So I meant them to speak of that. I was able to suppose the circumstances to be thus and so. So I meant thus to say what one would in speaking of things as that way in such circumstances. When would things be as I thus meant to say they were? Well, what language games would I be playing were my suppositions so? Our new idea of what representation is now takes its grip. How I meant things were now falls within the scope of the parochial. Here, then, one illustration of how the new picture can be a picture of autorepresentation. Details differ from case to case. There is a further point. In saying that Sid thinks the room is dark, one might say of him that he thinks any of various things, depending on what, for purposes of so saying, one is to understand by a room being dark. But not: What Sid thinks in thinking the room dark (where he so thinks) depends on what one understands by a room being dark. (Nor even, I think, on what he understands by this.) Rather, in thinking the room dark, Sid thinks something so. What can be thought so is what, ceteris paribus, either is so or not. Which it is can then only depend on how things are. Equally, then, for whether things are as he thus thinks. So equally again for what he thus thinks. There is (roughly) no room here for whether things are as Sid thinks to depend on what one understands by a room being dark; equally little for what he thus thinks to so depend. So far what Sid thinks seems to follow the pattern of a Fregean truthbearer. Such a truth-bearer represents things as such-and-such way, where there are just no competing understandings of something being that way; so whether something is that way cannot depend on what you understand by something so being. Similarly, what Sid thinks in thinking what he does here is that things are a certain way; whether things are as he thus thinks cannot, it seems depend on what one understands by things being such-and-such way, for any substitution for that “such-and-such.” The phenomenon of thinking so seems to take this shape. Such may be one important source of the idea of a Fregean truth-bearer. But how can there be objects of thinking so—thoughts, or propositions—which thus admit of no competing understandings? We need to understand the terms of the line of thought, above, which can seem to suggest so.

To represent as so

21

Wittgenstein explores this topic in §§429–65. The key idea here is: “An order orders its own execution.” So it knows its execution before it is there?—But that was a grammatical proposition, and it says: If the order runs “Do such-and-such” then carrying out the order is called “doing such-and-such.” (PI §458) “If you said that Sid grunts, what you thus said would be true if Sid grunts.” Whether this states truth depends on what one understands saying that Sid grunts to be. But if we understand this antecedent as speaking of your having said such-and-such to be so, where that is (presumably) something so or not—if, that is, we understand it as we did Sid’s thinking something so, above—then the remark is a truism, unfolding part (though not all) of the concept of truth; a grammatical proposition in Wittgenstein’s sense. But why? In the consequent, the words “Sid grunts” speak of Sid being a way he (presumably) is or not. To do that, they must bear some particular understanding of being a grunter. If I made this remark, then I spoke, in that part of it, of Sid being a grunter, on some particular understanding of him so being. In the antecedent I speak of you speaking of Sid being a grunter. Again, if I thus speak of you saying things to be thus and so, then I thus speak of Sid being a grunter on some particular understanding of him so being. And now a grammatical point: it must be the same understanding both times, which guarantees the truth of the whole. Which is to say nothing as to just what understanding this might be. So far we have learned nothing as to just when you would be right. Suppose now that I say “If Sid thinks the room is dark, what he thinks is so just in case the room is dark.” If, in the antecedent, I speak of Sid thinking things to be thus and so, then all the above remarks apply. In the consequent I use the words “the room is dark” to speak of a room as dark on some particular understanding of its so being. In the antecedent I do the same. In the consequent I do it to say how things must be for Sid to be right. In the antecedent I do it to say what it is he will be right or wrong about. Again, necessarily, one understanding of a room being dark is in force throughout. Again we so far know nothing as to what it is that Sid thinks. “In thinking the room dark, Sid thinks something so; which is so, or not, independent of what you understand by a room being dark, or by things being any other given way.” That is so in just the way it is so that if I say the room to be dark, I thus say something so; which is so, or not, independent of what you understand by a room being dark. So things are as I said, or not, independent of any such understanding, just as things are as Sid thus thought independent of this. Neither truism suggests Fregean truth-bearers. Rather, the parochial retains the role assigned it, on our new idea of representing, equally, in the one case, in making determinate what it is I said, and when things would be as thus said, and, in the other, what it is Sid thinks,

22

Charles Travis

and when things would be as thought. That role once played, there is no need for playing it anew in fixing those understandings which, as we saw above, are not there to be fixed. (But if Sid thinks the room dark, could it not turn out that whether he thus thinks correctly depends on what you understand by being correct in so thinking? Of course.)

1.7 Particular and general For Frege there is something on which the parochial must be silent: when (in what particular cases) things would be thus and so. A way for things to be speaks for itself in all such matters. If such speaking is meant to silence the parochial, then for Wittgenstein the idea is incoherent. Here is the key point: We say “The order orders this—” and do it; but also “The order orders this: I am to … ”. We translate it at one time into a proposition, at another into a demonstration, and at another into action. (PI §459) Such translation—from something to be done to a case of doing it, from things being such-and-such way to a case of their so being—is blocked where the parochial is so silenced—for reasons Frege himself identifies. The rule following discussion points this out. Wittgenstein is here most deeply Fregean. There is, Frege notes, something intrinsically general about a thought: “A thought always contains something reaching out beyond the particular case so that this is presented to us as falling under something general.”15 Reaching beyond the particular case: a thought true of things as they are might still have been true without things being just as they are (in every respect). Nor need things have been just as they are in order for there to have been such a thought at all. If it is true that my cup is empty, it might still have been had it just been drained of different tea. Were it undrained, there would still be the thought that it is empty. The generality of thoughts is found in a notion of a way for things to be. Being empty is a way for a cup to be. Being fuller than is a way for an ordered pair of cups to be. My cup being empty is a way for things to be. My cup being as it is may be (or not) its being empty. As I will say, it may instance this (or not). Not everything about my cup matters to it, as it is, instancing being empty. What being empty is fixes what does matter and how; what is demanding for qualifying as an instance. There is an indefinitely extensible range of cases of what would instance being empty; equally, a range of different ways of being (doing) what a thing thus must. That my cup is now empty is not a way for things to be. It is the circumstance of things so being. In things being as they are, they are so. There is indefinite variety in the way things could be while so being (that my cup is empty). There is, again, a range of cases. So the generality of a way for things to be (as I will say, of the conceptual) is the generality of thoughts.

To represent as so

23

“A satisfies (falls under) the concept W” is here just a variant for “A’s being as it is is its being W.” A concept is, intrinsically, of being thus and so. A concept may be satisfied by single items, by order pairs, and so on. A zeroplace concept is a thought. There is a class of concepts (first-order, I will say) which are satisfied by what is not itself conceptual. The characteristic mark of the non-conceptual (the particular, as I will say) is that there is no range of cases of something’s being as it is being it. There is no satisfying the particular. It makes no demands on, so nothing could matter to, so doing. My cup is not a way for things to be. Nothing’s being as it is would be its my cup. There is, by contrast, such a thing as being my cup. This is a way for things to be. To be sure, the only thing that could be that way is my cup. But my cup may do that chipped, in Dublin, and so forth. There is a range of cases here. What satisfies a concept? In mathematics, one can say, “an object.” The number 2 satisfies the concept smallest prime. In sublunary affairs this will not do. My cup satisfies the concept empty. Now it does. It is about to cease. Tea is on the way. It is my cup’s being as it is now that is its being empty. This is what satisfies concepts. Equally for 0-place concepts. It is things being as they are which is its being so that my cup is empty. There is no range of cases of my cup being as it now is. To be that, a thing would need to be my cup, as it now is. There is just one case of that. So my cup’s being as it is, things being as they are, belong to the particular. My cup is the way it is. But that is not a way for a thing to be; only a way for my cup now to be. Terminology should not mislead us into misclassifying my cup’s being as it is with the conceptual. In defending his idea of the invisibility of truth-bearers Frege points to a crucial difference between particular and conceptual. My cup—a particular—is visible. That my cup is empty is not visible on the same understanding of visible. For, for my cup to be empty is for things being as they are to belong to a certain range of cases. To see that it is empty, one must recognize the membership. Neither such membership, nor the range thus joined, are visible things. One cannot see that to which precisely what would be cases of a cup being empty belong; nor are those other cases themselves in sight.16 Frege tells us that the particular’s relation to the general in belonging to ranges is the fundamental logical relation. In his words, “The fundamental logical relation is that of an object’s falling under a concept: all relations between concepts reduce to it.”17 For a way for things to be, or concept, to bear this relation to given particulars is for a generality to reach all the way to the particular case—Wittgenstein’s translation from order to action: I am to do such-and-such; I am to do this (doing it). A cup’s being empty is a cup being empty; this is a cup being empty. The rule-following discussion concerns this fundamental relation. What could it be for it to be instanced? The relation holds between my cup as it now is and its being empty, but not between my cup as it is about to be and

24

Charles Travis

its being empty. What is the difference? The obvious answer: my cup is now empty, but is about not to be. But that is just to say: the relation holds where it holds, and does not where it does not. It says nothing as to when it would hold. Frege speaks of a function, so a relation, as identified by a “law of correlation.”18 No such law has yet swum into view. So what is to be said as to when, or by satisfying what demand, the fundamental relation holds? We can, it seems, explain a given instance of its holding in terms of other instances of its holding—my cup is innocent of liquids. But there seems no other sort of thing to say. We understand relations within the conceptual in terms of the fundamental relation, falling under. We cannot understand it in the same way. But perhaps in no other way either. Which fits with Frege’s conception of the zero-place case. For things being as they are to fall under the zero-place concept of my cup’s (now) being empty is just for it to be true that my cup is empty. Truth, Frege tells us, is indefinable (equivalently, truth is part of the content of any judgment). To know when it would be true that my cup is empty is just to know when my cup would be empty—that is, when things being as they are would fall under the concept my cup being empty. But this last is just what we were asking after. To know what would fall under my cup being empty is to know what the correct transitions are from the conceptual, my cup being empty, to the particular cases of what are, or not, its instances. But what is it for given transitions to be the right ones? How can there be something to be known here? I can, it seems, only explain the instancing of the relation by my cup (as it is) and being empty by appeal to other instances of the relation. So perhaps asking for more than this is asking a bad question. Frege says something on these lines for questions as to why a law of logic holds. Such may admit of explanation within logic. But it admits of no extra-logical explanation.19 In logical matters, there is no such thing as things being other than as logic has them. For, insofar as there were, there would be room for explanation as to why things were not otherwise; such explanation being, necessarily, extralogical. But if there is no such thing as saying why logic is not otherwise, one can say why this is so. Logic (on Frege’s conception) unfolds the most general structure of judgment, or truth. What was not so structured would, ipso facto, not be judgment, nor truth. So there is no such thing as thought being otherwise than logic says; but all logic says is how thought must be. (I do not defend this story. What matters here is just its availability.) Now the thought would be: perhaps there is no saying why the fundamental relation is instanced as it is and not otherwise in the same way as there is no saying why logic is as it is and not otherwise. We need to work out what this means. But there is at the start a crucial disanalogy. Logic traces relations within the conceptual. It does not speak to, but rather presupposes, the fundamental relation. If, as Frege says, truth is part of every judgment, then knowing when it would be true that my cup is empty is just knowing when my cup would be empty. So when it would be is part of what

To represent as so

25

truth is. But logic is not concerned with that part. What we want to see, then, is whether those same ideas as to there being no such thing as otherwise, which find some application for that part of thought, or truth, which logic does unfold also apply for that part which it does not. Logic, conceived so as to fit Frege’s story, deals in a certain structure which is intrinsic to the conceptual; a certain way in which given bits of the conceptual relate to others. To repeat, relations between bits of the conceptual are understandable in terms of the fundamental relation. It is in terms of such understanding that we can understand what it is for such a relation to hold intrinsically. If Frege’s story fits, then this understanding of intrinsic will make sense of the relevant idea of there being no such thing as things being otherwise. But we do not have the same means at our disposal for understanding the fundamental relation itself; so nor for understanding what it could be for it to have a certain structure intrinsically. And that idea will have quite a lot of work to do here. Take any instance of the fundamental relation holding, or of it not, between some particular and some concept—say, between my cup as it now is and the concept empty. If the relation holds, then it is part of what being empty is that my cup’s being as it is is so to count. That is the correct translation from the fact of being empty being thus and so to its being this. Any other translation is a mistake. But there is no saying what it is for things to be that way. So that they are must be intrinsic to the structure of falling under: whatever did not have that feature simply could not be that relation. Similarly for every instance. How to understand that? The picture is now this. There is no saying why falling under has the instances it does (except why it has certain ones given that it has certain others). For it is intrinsic to it to have precisely those instances that it does. Since the ordered pair, <my cup as it now is, being empty>, instances falling under just in case my cup as it now is instances being empty, the same goes for every concept. It is equally intrinsic to being empty to have just the instances it does. All of this is part of the intrinsic structure of the relation of the conceptual to the particular. But we can remind ourselves that we do not yet understand what intrinsic could mean here. We understand its application within the conceptual because we understand the conceptual in terms of the fundamental relation. We cannot have that understanding of it applied to relations between the conceptual and something else. Which leaves us, so far, with no understanding at all. If my cup’s being as it is is its being empty, one can rightly say: it is part of what being empty is that this particular is to count as something being empty. What we want to understand is what truth we thus state. The wanted reading cannot plausibly be: anything which did not have just this instance would not be being empty. There are two reasons. The first is the point with which the Philosophical Investigations opens. The concept concept does not work like that. My cup is at the bottom of the fish tank, along with others. The others are all filled with fish food. Mine is not. Is mine empty? We

26

Charles Travis

would still be speaking of being empty if we spoke so as to count it as that way, but also if we spoke so as to count it as not that way (water being what it is). It is not intrinsic to being empty to be instanced by precisely what may so count. This point points in several directions. Here it makes vivid an understanding on which the conceptual does not relate intrinsically to the particulars which instance it—to this being as it happens to be. Whatever we count as its instances, a bit of the conceptual—a way for things to be—does not require just those for being the way for things to be it is. That would be a misreading of the correct idea that this just is what it is for something to be empty. The second point is obvious. My cup might have been full. Had it been, there would, and could, not have been that very particular, my cup’s being as it is, which in fact bears the falling under relation to being empty. Being empty would still have been the very thing it is. Ten minutes from now my cup will be as it then happens to be. Its being as it then is may be its being empty, or, again, may not. What way for a thing to be being empty is will be the same no matter what the outcome. So, again, there being such a thing as being empty, and its being what it is, does not turn on just which instances this turns out to have. That, again, is an understanding on which it is not intrinsic to being empty (or to any bit of the conceptual) to have just the instances it does. Our aim is to understand the transition from what being empty is—being thus and so—to the particular case—my cup, as it will be in ten minutes, being that way, or, again, not. The point is not that what being empty is does not decide whether my cup is that. But what is it for it to do this? We have found no help in the idea that it is part of the intrinsic structure of being empty—or, equally, of falling under—that that very particular, my cup’s being as it then will, is one of its instances, or again, as one must say, is not. It is hard to see how “intrinsic” could bear an understanding on which it would both be true, and advance understanding, to say so. Which is just to develop Frege’s idea. We can understand relations between bits of the conceptual in terms of the fundamental relation—one between particulars and the conceptual. We can thus give sense to the idea (whatever its merits) of such relations holding intrinsically. But we cannot understand the fundamental relation itself in such a way. So nor can we so understand what it would be for it to hold intrinsically between such-andsuch particular and such-and-such bit of the conceptual. Nor would it accord with our understanding of the conceptual—of concept, of way for things to be—to equate “holds intrinsically” with “just does.” In fact, it is simply obscure what “intrinsically” could mean here. One understands falling under in grasping, e.g., when things would count as my cup being empty; just as one grasps what truth is in grasping such things as how the world decides the correctness of a judgment that a cup is empty. Here the first rule-following discussion comes into play.20 I can explain (rightly or wrongly) where the post office is by pointing the way.

To represent as so

27

What I thus explain—how I say to go—is fixed by how such explanation would be taken by thinkers of the sort we are. As we saw in the previous section, with the parochial thus in operation, I can say, “The post office is closed” and mean it in a certain way; and my meaning can stop nowhere short of the particular case: instances of the fundamental relation. So if one can walk into the building, but the counters are all closed, things being just as they thus are can be their being as I meant, or, again, their not. The explanations we give, so the things we say, and mean, are, normally, for thinkers like us; which fact shapes how it is they represent things; so what the correct translations are from “he meant that/to do …” to “he meant this.” Thus does my meaning stop nowhere short of the facts.21 But the point now is: only with the parochial thus in operation is this so. When would a cup count as being empty? Exactly which instances of a cup being as it was would be it being that? When, that is, would it be the way I just asked about? Which way did I ask about? Now the above point applies. Operating as per the first rule-following discussion, the parochial fixes which way it is I spoke of, or meant. In this it stops nowhere short of fixing that way’s participation in the fundamental relation: which instances of things being thus are instances of their being that way. Here stopping nowhere short is stopping— nowhere—short. Whatever the parochial achieves by way of relations between the bit of the conceptual I spoke of and other bits of the conceptual, this will not achieve relations between the way I spoke of and bits of the particular—those things which instance it. Such is just what it is for the fundamental relation not to be reducible to relations between concepts as relations between concepts are (on Frege’s view) reducible to it. Nor is there, operation of the parochial aside, any other way for there to be facts as to how being such-and-such way translates into being, for example, thus. The parochial at work, it can be correct to say that it is intrinsic to what being empty is that this is a cup being empty: things could not be any other way in that respect. For it to count as true that the cup may well be for such instancing to count as thus mandated. Not that there is some other mandating than that which the parochial allows for. Nor does this exclude what is sometimes so: that it is intrinsic to what being empty is that this is being empty, only on a certain understanding of what being empty is. The parochial reaches all the way from the conceptual to the particular. With that idea properly in place there is no problem as to how rational relations, such as falling under, can have given instances. Our parochial equipment grants us understandings of what it is for the fundamental relation to hold. No others are to be had. The parochial is precisely that of which one cannot say, “There is no such thing as things being otherwise.” It is part of human thought; thus of human life, of our ways of going right or wrong in our projects in, or for, the world. So, it is something psychological; part of how things are, not how things must be. If the parochial forges transitions from the conceptual to the

28

Charles Travis

particular, as just sketched, then, with this in mind, it can seem that: “We are at most under a psychological, not a logical, compulsion. And now it looks quite as if we knew of two kinds of case.”22 Logic does not treat relations between the conceptual and the particular. If logical compulsion is what logic must place us under, then there is no logical compulsion to count my cup’s being as it is as its being empty. But, while one might turn my cup over to see whether it is empty, one would not do so to see whether its being as it is is its being empty. Nor would I pass out questionnaires on trams to see whether I am getting such transitions right. Except insofar as being empty admits of understandings, there are few or no open questions as to what would so count. If I say, truly, that my cup is empty, that is because it does so count. What plays a given role in our exposure to risk of error (as Frege put it) is shaped by what we would, or could, count as error; what might abash us. For all of which, you must count my cup as empty if you are to reach the goal truth—a dictate of the form Frege tells us it is logic’s role to give. If it seems as though there must be some other sense in which our hand ought to be forced here—as if we knew of some other kind of compulsion than that just sketched—then we need to recall the fundamental point of this discussion. We understand how there can be compulsion in transitions from one bit of the conceptual to another; an understanding available to us by reference to the fundamental relation. In the case of the fundamental relation itself we cannot understand compulsion in the same way. We do not have the same means available. It is this that casts suspicion on the idea that we do know of two kinds of case. If we were different sorts of thinkers we would speak to, and for, different sorts. That might change which ways for things to be we spoke of. We would at least speak on different understandings of things being as we said. What it would be for things to be the ways we thus spoke of, so when the world would bear the fundamental relation to the concepts we thus expressed (deployed as thus deployed) would change accordingly. Which does not mean that if we thought differently, my cup might be as it is without being empty. The conceptual that figures in our thought—the ways we think of things as being or not—is formed by that thought. It is, for all that, the conceptual, relating to the particular as is the conceptual’s prerogative. “For all that”: there is no other way of enjoying those prerogatives. Such is the moral of the rule-following discussion. In drawing out this line of thought in the Philosophical Investigations I have had two morals of my own in mind. First, to understand Wittgenstein it is always a good idea to think of Frege. Second, and correlatively, the Philosophical Investigations treats the very same problems that concerned Frege. It speaks to his concerns. These morals are for those—both admirers and detractors—who share the view that at some point, probably around the summer of 1930, Wittgenstein had an experience like Saul’s fall on the road, or Godard’s, then abandoning philosophy for some new form of self-help. The Philosophical Investigations offers new approaches to problems very much philosophy’s.

To represent as so

29

Notes 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

PI §242. Translations given are generally my own. PI §242. Cf. PI §§81, 130–31. Gottlob Frege, “Notes for Ludwig Darmstädter,” in Posthumous Writings, ed. Hans Hermes et al. (Oxford: Blackwell, 1979), p. 253. Cited by Wittgenstein in PI §49. TLP 3.328, 5.47321. Cf. Friedrich Waismann, Wittgenstein and the Vienna Circle, trans. Joachim Schulte and Brian McGuinness (Oxford: Basil Blackwell, 1979), p. 90. Ibid., p. 90. Gottlob Frege, “Der Gedanke: Eine Logische Untersuchung,” Beiträge zur Philosophie des deutschen Idealismus 2 (1918): 58–77. Gottlob Frege, “My Basic Logical Insights,” in Posthumous Writings, p. 51. Cf. Gottlob Frege, “Funktion und Begriff,” in Funktion, Begriff, Bedeutung: Fünf logische Studien, ed. Günther Patzig (Göttingen: Vandenhoeck & Ruprecht, 1962). Frege, “Der Gedanke: Eine logische Untersuchung,” p. 59. PI §§84–87, §§138–242. PI §205. Gottlob Frege, “17 Key Sentences on Logic,” in Posthumous Writings, p. 175. Cf. Gottlob Frege, “Logic,” in Posthumous Writings, pp. 137–38; Frege, “Der Gedanke: Eine logische Untersuchung,” p. 61. Gottlob Frege, “Comments on Sense and Meaning,” in Posthumous Writings, p. 118. Gottlob Frege, “Was ist eine Funktion?” in Funktion, Begriff, Bedeutung: Fünf logische Studien, ed. Günther Patzig (Göttingen: Vandenhoeck & Ruprecht, 1962). Gottlob Frege, The Basic Laws of Arithmetic: Exposition of the System, trans. and ed. Montgomery Furth (Berkeley: University of California Press, 1964), p. xviii. PI §§84–87. Cf. PI §95. PI §140.

Part I

Mind

2

Wittgenstein, qualia, and the autonomy of grammar1 Julian Kiverstein

Wittgenstein is commonly read as an ally of those who are skeptical about the existence of qualia.2 A skeptic about qualia either flatly denies the existence of such properties or he says that supposing such properties were to exist, we could know nothing about them.3 Wittgenstein might be thought a skeptic in the first sense when he argues against the possibility of widespread color inversion, such that we could not know that another person doesn’t see colors entirely differently from ourselves. Anyone who accepts the existence of qualia will typically also embrace the possibility of such a scenario. Indeed, thought experiments involving widespread spectrum inversion are standard ways of arguing for the existence of qualia.4 I will argue that Wittgenstein provides sound arguments against widespread spectrum inversion. However one such argument depends on a controversial assumption I will call “the effability requirement” which anyone inclined to accept the existence of qualia would simply reject. If Wittgenstein’s skepticism about qualia rested on such an assumption, it would carry little weight. I will outline an alternative Wittgensteinian argument that doesn’t rest on the effability requirement. However again we will see that such an argument fails to establish a more general skepticism about qualia. We must look elsewhere than to his discussion of spectrum inversion if we are to find compelling Wittgensteinian reasons for denying the existence of qualia. Skepticism about qualia in the second sense might be attributed to Wittgenstein on the basis of his attack on the possibility of a private language. It is natural to think of qualia as properties that are known only to a single subject—the individual that has the experience to which the qualia belong. Only I can know the qualia of my experience through introspection, for instance. Everyone else can know about the qualia of my experiences only by means of inference from my behavior, and it might be objected that my behavior isn’t a reliable indicator of what I experience. Hence only I can know the qualia of my experiences—my experiences have properties that are known only to me, and are in principle unknowable to others. I will call qualia, logically private properties for this reason. Now Wittgenstein’s private language argument has been read as an attack on the idea that our experiences have properties in virtue of which they are logically private.5

34

Julian Kiverstein

Wittgenstein is read as arguing that if there were a private component to experience, we would not be able to tell when this aspect of our experiences recurred.6 There would be no criterion of correctness entitling us to say that the item we are now identifying is the same item we encountered on an earlier occasion. But in the absence of such criteria of correctness we could not be said to know we are having an experience, and this is of course to endorse the second strand of skepticism about qualia. It is to say that even if qualia exist, we could know nothing about them. I will argue that, appearances notwithstanding, the private language argument doesn’t provide grounds for skepticism about qualia of the second kind. At best it establishes that there couldn’t be a private language—a language for talking about sensations and experience understood by a single person, the user of this language. This is to say that the language we use to communicate about our experiences must be a public language. However it doesn’t follow that what these elements of language refer to cannot be something private, an episode knowable in a first-person way only to the subject undergoing the episode, and in principle unknowable to others. We must therefore continue the search for an explanation of Wittgenstein’s skepticism about qualia. It is undeniably the case that in his mature writings Wittgenstein held a skeptical position about the existence of qualia, but it would be a mistake to suppose he always held such a position. In a neglected period of his writings, the so-called “middle period,” we find a philosopher wholly committed to the existence of qualia. Wittgenstein described his work of this period as “phenomenology.” His phenomenological project had as its aim the construction of a language for “the description of immediate sense perception without any hypothetical addition.”7 When we are describing immediate experience, he explains, we do not need to decide “about the presence or absence of an object.”8 Wittgenstein goes on to write that if we try to decide whether what we are seeing is a physical object, we make the mistake of erroneously applying “our physical mode of expression to sense data. ‘Objects,’ i.e., things, bodies in the space of a room—and ‘objects’ in one’s visual field; the shadow of a body on the wall as an object!”9 I take it that such an account of sense-perception which distinguishes physical objects from sense data commits the Wittgenstein of this period to a principle Howard Robinson has dubbed the “phenomenal principle”: If there sensibly appears to a subject to be something which possesses a particular sensible quality then there is something of which the subject is aware which does possess that sensible quality.10 Anyone who endorses the phenomenal principle must also accept a commitment to the existence of qualia. Take the case where there appears to be something possessing a particular sensible quality q, but where nothing exists that is q. According to the phenomenal principle one is nevertheless aware of something possessing q. What one is aware of in this case is a sense-datum

Wittgenstein, qualia, and the autonomy of grammar

35

and the sensible quality q is a quale, a property of this sense-datum. The middle-period Wittgenstein seems to have been committed to the phenomenal principle, but we have just seen how the phenomenal principle entails a commitment to qualia. Thus we can conclude that qualia occupied a central place in the middle-period Wittgenstein’s account of perceptual experience. Soon after Wittgenstein undertook the project of constructing a phenomenological language, he began to express doubts about the need for such a project. In the opening section of his Philosophical Remarks, he wrote of phenomenological language: “I no longer hold it to be necessary.” I will argue that a correct understanding of Wittgenstein’s skepticism about qualia lies not in his discussion of inverted spectrum thought experiments or even in the private language argument. Instead we must understand his skepticism about qualia in the context of the considerations that led him to abandon the project of constructing a phenomenological language. We will see how Wittgenstein gave up on the latter project when he abandoned a view of language as having a single logical structure disclosed through philosophical analysis. With this change of mind came a rejection of a theory of meaning for experience and sensation terms according to which these terms derive their significance from their referents. Instead Wittgenstein was to argue that these terms have the meaning they do because of the place they occupy in a language game. Just as our language games are not accountable to reality,11 so we can also say in the specific case of our sensation and experience terms that they are not accountable to experience for their meaning. It is this latter claim about the autonomy of grammar that motivates Wittgenstein’s skepticism about qualia.

2.1 Wittgenstein on the possibility of inverted spectra In his posthumously published “Notes for Lectures on Private Experience and Sense Data”12 written between 1934–36, Wittgenstein offers some muchdiscussed comments on the possibility of inverted color experience.13 Wittgenstein discusses two kinds of spectrum inversion, only one of which he regards as logically possible (i.e., conceivable). In the first scenario: Someone says “I can’t understand it, I see everything red blue today and vice versa.” We answer “it must look queer!” He says it does and, e.g., goes on to say how cold the glowing coal looks and how warm the clear (blue) sky. I think we should under these or similar circumstances be inclined to say that he saw red what we saw blue. And again we should say that we know that he means by the words “blue” and “red” what we do as he has always used them as we do. (PESD, p. 284) Ned Block argues that Wittgenstein is willing to allow the possibility of this kind of spectrum inversion because a subject would be able to communicate to

36

Julian Kiverstein

us how his experience had changed.14 The differences between the experiences of our imagined subject and normal color experience are tracked by what the subject says about his experience. These differences can be tracked because our inverted subject uses color terms just as we do. Hence the descriptions he gives of his experiences are ones we can understand. Call this first kind of spectrum inversion “detectable spectrum inversion” since the subject can make known to others how his experience has changed. There is a second kind of spectrum inversion, which I will call “widespread spectrum inversion,” that Wittgenstein holds to be logically impossible (i.e., inconceivable or incoherent): We said that there were cases in which we should say that the person sees green what I see red. Now the question suggests itself: if this can be so at all, why should it [not] always be the case? It seems, if once we have admitted that it can happen under peculiar circumstances, that it may always happen. But then it is clear that the very idea of seeing red loses its use if we can never know if the other does not see something utterly different. So what are we to do: Are we to say that this can only happen in a limited number of cases? (PESD, p. 316) The case Wittgenstein describes in this passage is not peculiar but is instead something that is widespread, and so commonplace. The subjects in question have not undergone a sudden change. Perhaps they were born with crossed neural wiring so that they have always enjoyed inverted color experiences— Block imagines such a scenario.15 They have grown up calling things red which look to them to be green and vice versa. It is this possibility Wittgenstein seems unwilling to admit. Wittgenstein’s worry seems to be that if spectrum inversion were widespread, the terms we use in reporting how we experience color would, as he puts it, “lose their use.” In the first kind of case we were at least agreed on the meaning of these terms, and this is what allowed us to understand what had happened to the color experiences of the inverted subject. Not so in the case of widespread color inversion. If spectrum inversion were widespread, there would be no agreement about how to use color terms in describing one’s experience. Some subjects will use the expression “looks red” when experiencing things that look green, others will use this expression when experiencing things that do indeed look red to them. Who is to say which of the groups is using this expression correctly? Whatever grounds we have for judging one use to be correct, these grounds would seem to count equally in favor of the other usage. Without any agreement on which usage counts as normal, talk of how things look with respect to color will be meaningless. Of course our imagined community could use some objective feature to secure agreement about how to use expressions such as “looks red.” They could agree to use this expression to refer to whatever experience a subject is

Wittgenstein, qualia, and the autonomy of grammar

37

having when seeing something red. The members of such a community would still lack a means of expressing what it is like for them to experience color. There would be differences in the ways a colored thing looked which would not show up in what subjects say about their experience. An inverted subject would say they were experiencing something that looks red when in fact they were seeing something that looks the way green things look to us. Let us call a position like Wittgenstein’s, which accepts the possibility of detectable spectrum inversion while denying widespread spectrum inversion, Restricted Spectrum Inversion. Wittgenstein’s defense of restricted spectrum inversion would seem to depend on an assumption I shall call the Effability Requirement. The Effability Requirement says: (ER) There are no differences in the phenomenal character of experience, which cannot be given linguistic expression. If we assume ER, a rejection of widespread spectrum inversion follows. We have just seen how, if widespread spectrum inversion held, there would be ways that colored things look which couldn’t be given linguistic expression. Inverted subjects couldn’t point to anything public to explain the different ways colored things look to them. Nor could they use any expression of the form “looks F” where “F” stands for some color term.16 We have just seen how an inverted subject could correctly use an expression of this form even though the way he experiences F-ness is the way we experience G-ness. We must conclude then that in a situation of widespread spectrum inversion there would be differences in phenomenal character that couldn’t be given linguistic expression. Unfortunately an argument against widespread spectrum inversion based on ER will fail to establish grounds for skepticism about qualia. Anyone accepting the existence of qualia will take the possibility of widespread spectrum inversion to be grounds for rejecting ER. To reject ER is to accept the existence of qualia, as has been persuasively argued by Block (in his forthcoming Wittgenstein and Qualia). If we are to motivate skepticism about qualia, we must either provide some additional defense of ER or look for a defense of restricted spectrum inversion independent of ER.

2.2 In defense of restricted spectrum inversion Immediately following his discussion of widespread spectrum inversion, Wittgenstein goes on to add: “we introduced the expression that A sees something else than B and we mustn’t forget that this had only use under the circumstances under which we introduced it.”17 I take Wittgenstein to be reminding us in this passage that we have a linguistic framework for comparing our color experiences with the experiences of others. It is by employing this framework that we can make sense of two individuals, A and B, experiencing one and the same property (e.g., redness) differently. In a

38

Julian Kiverstein

situation of widespread spectrum inversion, no such framework could get off the ground. The expressions such a community used to talk about the ways colored things look would cut across differences in the ways subjects experienced colors in such a way as to make these differences inexpressible. It follows that when considering such a situation we are using the expression “A experiences something else than B” outside of the circumstances or context in which it is normally used. Detectable spectrum inversion strikes Wittgenstein as being possible because in this situation we can draw on a framework for comparing experiences. When we think about a situation in which there is widespread spectrum inversion we are considering a situation in which no such framework could get off the ground. Subjects couldn’t express the way colored things look to them in such a situation. If they cannot express the ways in which colored things look to them, nor can they compare their respective experiences. If they have no means of comparing their experiences, how can one subject “A” make sense of the possibility that he might be experiencing something different from another subject “B”? Yet it is the possibility of A experiencing something different from B that we are trying to conceive of in entertaining the possibility of widespread spectrum inversion. So it looks like Wittgenstein’s defense of restricted spectrum inversion lies in considerations about what must be possible if we are to compare experiences—a possibility that the very intelligibility of any kind of spectrum inversion may rest on. A situation in which widespread spectrum inversion held would be a situation in which subjects could make no sense of comparing their color experiences with respect to their sameness or difference. The idea of widespread spectrum inversion presupposes the possibility of comparing color experiences. Yet we have seen that if widespread spectrum inversion obtained, there would be no framework for comparing experiences. Thus the idea of widespread spectrum inversion is self-undermining or incoherent. Compare the case of widespread spectrum inversion with what Wittgenstein says about pain experiences which are never displayed: “And remember that we admit that the other may have pain without showing it! So if this is conceivable, why not that he never shows that he has pain; and why not that everybody has pain constantly without showing it; or that even things have pain?!” What strikes us is that there seem to be a few useful applications of the other person’s having pain without showing it, and a vast number of useless applications, applications which look as though they were no applications at all. (PESD, p. 317) We can make sense of someone’s being in pain but not displaying it in certain circumstances. Perhaps the person is an athlete wholly absorbed in trying to win a competition so that she is not attentively aware of her pain experience. Wittgenstein points out however that there are also a “vast

Wittgenstein, qualia, and the autonomy of grammar

39

number of useless applications,” and I take it that widespread spectrum inversion and pain experiences that are never expressed count as paradigm examples of “useless applications.” Application of the concept of pain in a situation where pain is never displayed would be a “useless” application partly because the concept of pain would have no use in such a situation. We use the concept of pain, in part, to talk about behaviour expressive of pain. If pain were never displayed, our concept of pain would lack this use. Of course this is not the only use our concept of pain is put to—in addition, we use this concept to refer to that unpleasant state we are in when some part of our body hurts.18 Still a Wittgensteinian could argue that in the absence of displays of pain such applications of the concept of pain could not get off the ground. To have mastered the concept of pain is to know not just when it is appropriate to ascribe pain to oneself, but also when it is appropriate to ascribe pain to others. In a situation in which pain were never displayed, one would not know when it was appropriate to ascribe pain to others. To the extent that our concept of pain is used not just to talk about our own pain but also about the pain of others, our concept of pain would be without use. The concept of pain would be “useless” in a second sense in such a situation. There would be no need for such a concept, which is to say that the concept would serve no function or purpose. We need such a concept to talk about behaviour (our own and others) expressive of pain. We use this concept to explain to others what we are feeling when we are in pain. If no one ever expressed the pain they felt, there would be no need of such a concept. The same moral applies to the concepts we use to talk about the ways in which colored objects look to us. In a situation of widespread spectrum inversion these concepts would have no use in either of the senses just rehearsed. There would be no need to talk about the ways colored things look since there would be no practice of comparing experiences. Indeed if the above argument is correct, not only is it the case that these concepts would have no use, nor could they have a use. In a situation of widespread spectrum inversion there would be no agreement in how color words are used in contexts where subjects are reporting on how colors look to them. Without such agreement the framework would be missing required for making sense of our talk about color. Of course it might still be objected that a practice of comparing experiences could get off the ground in a situation of widespread spectrum inversion. An individual subject that has undergone a procedure for inverting his experience could use his memory to compare his post-inversion experience with his pre-inversion experience. He could arrive at judgments about similarity or difference in experience, relying only on his memory.19 In response it might be argued that memory cannot provide the kind of evidence needed to warrant judgments of this kind about similarity or difference in experience. We might insist that memory could never justify a subject’s claim to know that the colors he experiences now are the same or different from the colors he experienced in the past. However, if such an

40

Julian Kiverstein

argument can be made to work against the individual, it would also tell equally against a public practice of comparing experiences. I suggested above that a practice of comparing experiences can get off the ground because we are in agreement about what we mean by the terms we use to describe our experiences. However the skeptic could reasonably wonder how we know that our current use of terms is in agreement with our past usage of those terms. He could raise the very same objection it is suggested we level against the appeal to memory. On the other hand, if we suppose that there is some solution to this skeptical problem at the level of public practice, this solution could surely also be brought to bear by the individual.20 A more persuasive response would insist that memory can only do its work because there are certain practices already in place, in particular practices for reporting color experiences and comparing them. Individuals can use their memory to compare earlier and later experiences but this use of memory presupposes existing practices such as the practice of giving a name to one’s experience. Wittgenstein points out in Notes for Lectures on Private Experience and Sense Data: “‘To give a sensation a name’ means nothing unless I know already in what sort of game this name is to be used.”21 It is only because one is already a participant in a practice of naming one’s sensations and experiences that one can use memory to compare earlier and current sensations or experiences. I will have more to say in defense of such a line of argument below. However what has been said suffices for a Wittgensteinian defense of restricted spectrum inversion. Widespread spectrum inversion can be rejected because it is incompatible with our practice of comparing experiences, and without this practice we can make no sense of the idea of one subject experiencing something different from another subject. What is the place of the Effability Requirement (ER) (the claim that there can be no differences in phenomenal character that cannot be given linguistic expression) in this defense? I’ve argued that the practice of comparing experiences depends on subjects being able to express the ways in which they experience things. From this claim, it follows that we cannot make sense of differences in phenomenal character that do not show up in what a subject says and does. This is to make a claim about our concepts of phenomenal character: namely that there is a constitutive connection between mastery of these concepts and what a subject says and does. I take Wittgenstein to be committed to a semantic thesis to the effect that our sensation and experience concepts must have a public sense. I take this to be a semantic thesis, concerned with how it is possible for us talk meaningfully about our experience and their phenomenal character. It consists in part in the negative claim that we cannot talk meaningfully about our experiences using a private language—a language understood only by a single user. This is the moral of the so-called private language argument. In addition, Wittgenstein makes a positive claim to the effect that it is expressive behavior—facial expressions, gestures, and other bodily movements—which

Wittgenstein, qualia, and the autonomy of grammar

41

get our talk of experiences, sensations, and the like off the ground. In Philosophical Investigations, §244, for instance, Wittgenstein asks how it is possible for words to refer to sensations. He returns the following answer: Here is one possibility: words are connected with the primitive, the natural, expressions of the sensation and used in their place. A child has hurt himself and he cries; and then the adults talk to him and teach him exclamations and, later, sentences. They teach the child new pain-behavior. I take “natural expressions” in this quote to refer to the facial expressions, gestures, bodily movements, exclamations, etc. a child makes when she is undergoing a painful experience. Thus the hypothesis Wittgenstein is putting forward in this section suggests that we can use words to refer to sensations or experiences because we already have a range of expressive behaviors for communicating them. Sensation and experience terms derive their meaning from pre-established practices.22 In the absence of expressive behavior, no context could get established for talking about our sensations and experiences. (Recall the discussion above of the possibility of a situation in which pain experiences were never displayed). Without such a context our sensation and experience talk would be meaningless. Thus a reasonable case can be made for the claim that we cannot make sense of differences in phenomenal character except by reference to expressive behavior. However, ER makes an altogether stronger claim that there are no differences in phenomenal character that cannot be given linguistic expression. The case that has just been made for rejecting widespread spectrum inversion doesn’t seem to warrant this conclusion. A proponent of qualia might concede that the differences in phenomenal character we can make sense of (i.e., conceptualize), we can make sense of only by reference to expressive behavior. Still he will deny that all differences in phenomenal character are such that we can make sense of them. He will continue to insist that there are differences in phenomenal character, which aren’t expressible. One might wonder whether there can be differences in phenomenal character we cannot make sense of, but it is precisely such a possibility to which a proponent of the ineffable is committed. We must conclude that a Wittgensteinian defense of widespread spectrum inversion may be independent of ER. The search must continue for a Wittgensteinian defense of skepticism about qualia.

2.3 The distinction between a phenomenological language and ordinary language Perhaps a better place to look if we are to understand Wittgenstein’s skepticism about qualia is to a change that took place in his thinking in the socalled middle period. The Wittgenstein of this period was, for a time at least, very much committed to the existence of qualia. I will argue that the change

42

Julian Kiverstein

that took place in his thinking had as a consequence a rejection of qualia. If I am right, it is these often-neglected writings that hold the key to a Wittgensteinian defense of skepticism about qualia. Consider the distinction, which looms large in his writings from the early 1930s, between the ordinary language used to talk about what J. L. Austin called “medium sized dry goods” (henceforth “ordinary physical language”) and what Wittgenstein sometimes called a “primary language” or “a phenomenological language.” In Philosophical Remarks for instance he writes: The worst philosophical errors always arise when we try to apply our ordinary—physical—language in the area of the immediately given. If, for instance, you ask, “Does the box still exist when I am not looking at it?”, the only right answer would be “Of course, unless someone has taken it away or destroyed it”. Naturally a philosopher would be dissatisfied with this answer, but it would quite rightly reduce his way of formulating the question ad absurdum. All our forms of speech are taken from ordinary, physical language and cannot be used in epistemology or phenomenology without casting a distorting light on their objects. (PR §57) Wittgenstein is pointing out in this passage that skeptical problems concerning the existence of things unperceived are the result of confusing the language we use for talking about experience with ordinary language we use to talk about physical things. If we keep to ordinary physical language, the question the skeptic raises loses its force. We can give Wittgenstein’s disarming response that of course the box exists unless someone has removed or destroyed it. True, the box as-it-is-perceived-by-us doesn’t continue to exist unperceived but, as Wittgenstein notes, it is not this box whose existence the skeptic is questioning. The box the skeptic is interested in is the box we talk about using what Wittgenstein calls physical language. The skeptical problem is only pressing because we confuse these two languages. The argument Wittgenstein employs in this passage for disarming skepticism rests on a distinction between a language which is employed for the description of immediate experience (henceforth “phenomenological language”) and ordinary physical language. When we judge that something is the case we presumably do so using ordinary physical language. We judge that the box continues to exist unless it has been removed or destroyed. Wittgenstein, at least at this stage in his writings, believed in the possibility of constructing a phenomenological language distinct from ordinary physical language. Wittgenstein was however soon to abandon this distinction between phenomenological language and ordinary physical language. In a 1929 conversation with Moritz Schlick recorded by Friedrich Waismann, Wittgenstein is recorded as saying:

Wittgenstein, qualia, and the autonomy of grammar

43

I used to believe that there was the everyday language that we all usually spoke and a primary language that expressed what we really knew, namely phenomena. I also spoke of a first system and a second system. Now I wish to explain why I do not adhere to that conception any more. I think that essentially we have only one language, and that is our everyday language. We need not invent a new language or construct a new symbolism, but our everyday language already is the language, provided we rid it of the obscurities that lie hidden in it.23 We will see how Wittgenstein conceived of a phenomenological language as uncovering the logical structure of phenomena, something he took to be covered over by ordinary physical language. He abandoned the distinction between phenomenological and ordinary physical language when he ceased believing that language has a single logical structure disclosed through logical analysis. In the next section I will argue that his skepticism about qualia may be understood as a consequence of this development in his thought.

2.4 Phenomenology as grammar Wittgenstein introduced the distinction between phenomenological language and ordinary physical language as a response to a difficulty he had identified with the conception of philosophical analysis described in the Tractatus. In his early writings Wittgenstein took the task of philosophical analysis to be to lay bare the logical structure of our language. A complete analysis of a proposition will clearly display this proposition’s logical structure. It will reveal a set of elementary propositions connected by logical connectives out of which a complex proposition is formed. Each elementary proposition is made up of well-formed arrangements of names standing for individual objects.24 Elementary propositions do not stand in logical relations with one another: just as the states of affairs which elementary propositions describe are logically independent of each other (i.e., externally related), so too must elementary propositions be logically independent of one another.25 Wittgenstein was soon to realize that this latter thesis could not be defended given the existence of some obvious counterexamples. It was in an attempt to bring to light the logical relations between elementary propositions that he was led to introduce the idea of a phenomenological language. Consider for instance the proposition that a single object cannot be colored red and green at the same time. An object’s being red all over logically implies that it cannot also be green all over at one and the same time. This seems to suggest that, contrary to the claims of the Tractatus, elementary propositions are not logically independent of one another. This must be so if an elementary proposition such as can stand in logical relation of implication to a distinct elementary proposition such as <no green color patch here> where “here” designates the same space–time location. However, the logical relations between these two elementary

44

Julian Kiverstein

propositions are concealed by the subject-predicate form of ordinary language. As Wittgenstein was to note in his 1929 essay “Some Remarks on Logical Form,” the syntax of a sentence such as “This object is simultaneously red and green all over” fails to show how it is a contradiction and hence necessarily false.26 He writes: “It is, of course, a deficiency of our notation that it does not prevent the formation of such nonsensical constructions, and a perfect notation will have to exclude such structures by definite rules of syntax.”27 Wittgenstein conceived of a phenomenological language as a means of making perspicuous the logical relations between elementary propositions. He conceived of the construction of this language as a kind of analysis, which would reveal the logical structure of phenomena. Wittgenstein was thereby conceding that propositions do not have a single logical form, the familiar logical form of the propositions of ordinary language as he had supposed in the Tractatus. A phenomenological language will disclose: the forms of space and time and the whole manifold of spatial and temporal objects, as colours, sounds etc., etc., with, their gradations, continuous transitions, and combinations in various propositions, all of which we cannot seize by our ordinary means of expression.28 He distinguished ordinary physical language from a phenomenological language—a language for describing the phenomena of immediate experience— because he believed that the propositions of ordinary language were limited in what they could show. Studying the logical form of ordinary language sentences doesn’t show, for instance, the many different degrees of brightness we can see, or the variety of shades between red and blue. Nor does it show us the structure of the relationships between the colors we see. We need another sort of analysis to uncover relationships such as these, a task which he initially seems to have thought we could achieve through the construction of a phenomenological language. In the Philosophical Remarks, Wittgenstein introduces a color octahedron to represent the full range of relations in which colors can stand to one another. “The colour octahedron,” he writes, “is grammar, since it says that you can speak of a reddish blue but not of a reddish green, etc.”29 A color octahedron will clearly show which color combinations we can meaningfully say are possible and which we cannot meaningfully talk about. Such an apparatus succeeds where ordinary physical language fails, since it constitutes a notation for correctly representing the logical structure of experience. We now have a handle on why Wittgenstein came to distinguish phenomenological language from ordinary physical language. We saw at the end of the previous section how this distinction was one that Wittgenstein soon came to repudiate. In the opening section of the Philosophical Remarks for instance, Wittgenstein writes that he no longer has a phenomenological primary language as his goal—he says that he no longer views such a language

Wittgenstein, qualia, and the autonomy of grammar

45

as “necessary.”30 To give up the goal of a phenomenological language is to give up the task of constructing a notation capable of representing the structure of experience. So what led Wittgenstein to this change of mind? I will follow Stern’s suggestion that Wittgenstein abandoned his goal of a phenomenological language because he abandoned the idea of there being a “single unitary order that was concealed by our ordinary language [ … ] a general form of the proposition shown by the structure of language as a whole.”31 Stern describes how over the course of the 1930s Wittgenstein came instead to view language as “a family of interrelated games [ … ] a number of autonomous systems of representational conventions, each with its own logical structure.”32 Wittgenstein rejected the goal of a phenomenological language because he rejected the project he had pursued in the Tractatus, of constructing a form of representation which would make perspicuous what is only confusedly expressed in ordinary language. He came to the view that propositions do not have a single general form which logical analysis can uncover, a view we already find expressed in “Some Remarks on Logical Form,” where he introduced the idea of a phenomenological language. There is no single notation that can make perspicuous what is concealed by ordinary language. Instead there are a multitude of different methods of representation or notations each of which serves a different purpose. Once he abandoned this view of language (as having a single logical structure), he no longer needed to think of phenomenological language as having a life distinct from that of ordinary physical language: “Our language is completely in order, as long as we are clear about what it symbolizes.”33 We get clear about what ordinary language symbolizes by examining and comparing different methods of representation or notations so as to lead us to a recognition of “what is essential and what inessential in our language if it is to represent, a recognition of which parts of our language are wheels turning idly…”34 Such a project he goes on to say “amounts to the construction of a phenomenological language.”35 It took Wittgenstein a short time to fully recognize the implications of his rejection of the necessity of formulating a phenomenological language. If a phenomenological language is not distinct from ordinary language, the sentences we use to talk about our experience must get their meaning in the same way as any other sentence of ordinary language. Yet throughout 1929 he continues to write of propositions describing experience as deriving their significance from experience: The phenomenon is not a symptom of something else; rather, it is reality. The phenomenon is not a symptom of something else which then makes the sentence true or false; rather, it is itself that which verifies it [the sentence]. (MS 107, pp. 223–24)36

46

Julian Kiverstein

By the 1930s Wittgenstein rejected this account of meaning. He came to conceive of phenomenology as concerned with, to borrow Noë’s expression, “what it makes sense to say” about experience.37 The rules which determine what it makes sense to say are rules of grammar. Moreover it is rules of grammar which determine “what is essential” and what is not in our language. Grammar determines which parts of our language “are wheels turning idly.”38 The grammar of our talk about color experience determines, for instance, that it makes sense to speak of reddish-yellow but not of reddishgreen or that there can only be one color at one place at any given time: “Red and green don’t come together at the same location at the same time” does not mean that in actuality they are never together but, rather, that it is nonsense to say that they are at the same place at the same time, and therefore also nonsense to say that they are never at the same place at the same time. (BT §100, p. 477) Wittgenstein came to conceive of a word’s use as determining its meaning. In §19 of Big Typescript, for instance, Wittgenstein asks why the proposition “one can’t divide red” makes no sense.39 The answer he returns is that this sentence is without meaning because “it doesn’t belong to the particular game that, judging from its appearance, it seems to belong to.” Games are of course defined by the rules which determine, for instance, which moves within the game are legitimate, and which illegitimate. So Wittgenstein is saying that the rules of the game, of which our talk of colors is a part, determine that the sentence “one cannot divide red” is without meaning. It is these rules which fix what it makes sense to say about color experience. Consider once again the proposition “two primary colors cannot simultaneously occupy the same place at the same time” (call this the “color exclusion proposition”). Wittgenstein came to view the color-exclusion proposition as expressing a rule of grammar which part determines what it makes sense to say about color experience. As an expression of a rule of grammar this proposition cannot be justified by appeal to any empirical fact—to a fact about the intrinsic character of our color experience for instance. Wittgenstein wrote in his Philosophical Grammar: “Grammar is not accountable to any reality.”40 Any empirical statement we formulate relating to color will depend for its meaning on the very same principles of grammar we are trying to justify. If we try to justify our color grammar by appeal to the fact that “there really are four primary colours,” we are presupposing our ways of dividing up the world.41 This division of color space is of course what we are trying to justify. What is it then that accounts for our being warranted in asserting the proposition “two primary colors cannot be in the same place at the same time,” if it is not some empirical fact? To answer this question it will be helpful to turn briefly to a passage from the Investigations:

Wittgenstein, qualia, and the autonomy of grammar

47

So does it depend wholly on our grammar what will be called (logically) possible and what not—i.e. what grammar permits?—But surely that is arbitrary! Is it arbitrary?—It is not every sentence-like formation that we know how to do something with, not every technique has an employment [Verwendung] in our life; and when we are tempted in philosophy to count some quite useless thing as a proposition, that is often because we have not considered its application [Anwendung] sufficiently. (PI §520) What this quotation suggests is that the color-exclusion proposition is a statement of grammar because this proposition has an application in our lives. Those grammatical statements that lack such an application in our lives do not genuinely express propositions, and thus do not genuinely possess a meaning even though they may seem to. They are like the useless propositions of the philosopher we encountered above. (Recall the discussion of widespread spectrum inversion and of pain that is never displayed). Thus a proposition genuinely expresses a rule of grammar if the rule in question has a function, which is to say that the rule achieves some practical purpose in our lives. It is only because the color-exclusion proposition is useful—i.e., fulfils a practical purpose in our lives—that it counts as a rule of grammar. What practical purpose does the color-exclusion proposition serve? Why couldn’t its contrary (i.e., the statement that two primary colors can occupy the same place at the same time) serve us just as well? While it is true that we cannot, without the threat of circularity, appeal to any facts to justify our use of such a grammatical principle, we can however appeal to facts to explain a rule of grammar’s usefulness. A rule like the color-exclusion rule is useful because certain empirical regularities obtain. If different regularities obtained, different rules would be useful and the color-exclusion rule would be useless. In his Remarks on the Foundations of Mathematics, Wittgenstein writes: “Calculating would lose its point, if confusion supervened. Just as the use of the words ‘green’ and ‘blue’ would lose its point [ … ] the arithmetical proposition would not be false, but useless, if confusion supervened?”42 Our talk of green and blue is useful only because it doesn’t depend on confusion, but is instead a way of classifying colors in accordance with the regularities we find in experience. If only confusion prevailed when we made use of such a classification, the rules which govern the use of these color terms would be useless. We have the rules of grammar we do because they are useful. In a situation where only confusion prevailed we wouldn’t have the rules of grammar we do, we would employ some alternative rules.43 I have suggested that understanding why Wittgenstein abandoned the distinction between a phenomenological and an ordinary physical language will help us to understand his skepticism about qualia. We are finally in a position to draw some conclusions. First we can understand Wittgenstein’s skepticism about qualia as the result of his rejection of the distinction between phenomenological and

48

Julian Kiverstein

ordinary physical language. A proponent of qualia will say that when we talk about the qualitative character of our experiences, the terms we use have their meaning partly in virtue of the qualia belonging to our experiences. Consider once again, the color-exclusion proposition: a proponent of qualia will say that what makes this proposition true is the intrinsic character of our color experiences. We cannot experience two primary colors as simultaneously occupying the same place at the same time because an experience can only have a single intrinsic qualitative character at any given time. Just as an object cannot be in two different places at the same time so experiences cannot have two different intrinsic characters at the same time. In a similar fashion, it is the intrinsic character of our color experiences which explains why it is possible for us to experience reddish yellow, but impossible for us to experience reddish green. If we say that the meaning a proposition expresses is fixed by a proposition’s truth conditions, it follows that propositions which make reference to the qualitative character of our experiences have their meaning partly in virtue of qualia. It is the fact that an experience instantiates qualia which, according to proponents of qualia, makes a proposition concerning the qualitative character of an experience true. This sort of explanation of color exclusion is just what Wittgenstein came to reject once he abandoned the distinction between phenomenological and ordinary physical language. The color-exclusion proposition, and other similar examples, seemed to show that propositions describing immediate experience had a logical structure which was hidden by ordinary language. Wittgenstein came to realize that it was a mistake to think of language as having a single logical structure. With this realization he abandoned the view that phenomena have a particular logical structure in virtue of which certain ways of talking about experience are meaningful and other ways of talking are meaningless. He came to believe that we talk about experience as we do, because of the rules of grammar which fix what it makes sense to say about experience. What makes it the case that certain ways of talking are permitted and others are proscribed is ultimately whether these ways of talking serve a practical purpose in our lives. Wittgenstein’s brief flirtation with a phenomenological language taught him that we talk about color experience as we do not because of the facts of color experience. We cannot appeal to our color experiences having particular intrinsic, non-representational properties, as proponents of qualia would have us do, to explain why some ways of talking are meaningful and others meaningless. No putative factual claim such as this one can explain why we divide up color space as we do. The meaningfulness of any such factual claim derives from our employment of certain rules of grammar. It cannot explain our employment of these rules of grammar without lapsing into vicious circularity. We have seen how proponents of qualia explain the meaning of our experience and sensation talk by appeal to our acquaintance with qualia. Wittgenstein’s skepticism about qualia lies in his rejection of this account of meaning. A proponent of qualia is committed to saying that the terms we

Wittgenstein, qualia, and the autonomy of grammar

49

use to talk about the qualitative character of our experience have their meaning in virtue of qualia. Wittgenstein rejects this account of meaning. Hence he rejects the existence of qualia. It might be thought that an argument can be given for the existence of qualia quite independent of a commitment to any particular theory of meaning. However this doesn’t seem to be the case for some of the leading arguments given in support of qualia. Take Frank Jackson’s knowledge argument.44 When Mary, the expert on color vision, is released from her black-and-white room she gains some new propositional knowledge. A proponent of qualia will say that the propositions she learns make reference to color qualia. He will say that the propositions Mary could not know prior to her having color experiences are propositions which have their meaning in virtue of color qualia. Alternatively consider Ned Block’s inverted earth thought experiment, which is supposed to establish that a subject’s intentional states can change their content while the phenomenal aspects of their experience remain the same.45 I take it that the force of this argument is supposed to be that there is something about the meaning of our color terms which changes and something that stays the same. To explain what it is about the meaning of our color terms that stays the same we have to appeal to qualia—non-representational, intrinsic properties of experience. So it would seem that another of the major arguments for qualia depends on considerations to do with a theory of meaning for experience terms. According to the reading I have been advancing Wittgenstein is a skeptic about qualia because of his views about the grammar of our talk about sensations and experiences. He argues that such grammar isn’t explained by any factual claims, such as the claim that our experiences have particular qualia. Rather such factual claims are made meaningful by the rules of grammar that regulate our experience and sensation talk. What explains the existence of these rules is the practical purpose they serve in our lives. It is not only proponents of qualia who claim that our experience talk acquires its meaning from acquaintance with our experiences. This is also a commitment of those who posit the existence of phenomenal concepts in accounting for the semantics of our talk about sensations and experiences. These philosophers have appealed to the existence of phenomenal concepts in arguing for the conclusion that the so-called explanatory gap46 is a cognitive illusion.47 However if the account I have given of Wittgenstein’s skepticism about qualia is correct, it not only puts into question that account of the meaning of our experience and sensation terms given by proponents of qualia. It also challenges those who would defend materialism by appeal to phenomenal concepts. It is with this challenge that I will conclude.

2.5 Phenomenal concepts Proponents of phenomenal concepts typically defend what Chalmers has called “type B materialism.”48 Consider Jackson’s knowledge argument

50

Julian Kiverstein

involving Mary the color-scientist who, confined to a black-and-white room for her entire life, learns the complete science of color. Type B materialists accept that Mary learns something new when she experiences color for the first time, but they deny that she learns a new fact. Instead they say she acquires concepts that allow her to think about facts she already knew in a new way. Some type B materialists suggest we think of phenomenal concepts as recognitional concepts.49 Phenomenal concepts have a characteristic functional role—they allow us to recognize the differences between phenomenal qualities and state directly on the basis of introspection. Since what is recognized is a physical state, the deployment of phenomenal concepts does not require Mary to think about a new object or property distinct from the ones she already thought about in mastering color science. Her mastery of phenomenal concepts enables Mary to think about the same physical states she was thinking about before but in such a way that she can discriminate between them based on introspection. Perry argues that we should conceive of phenomenal concepts as functioning like indexical expressions.50 When Mary sees red for the first time she acquires concepts that allow her to run through the following chain of thoughts: (1) QR is what it is like to see red (where “QR” is the concept she knows others employ in thinking about what it is like to see red). (2) This is what it is like to see red (the thought Mary has when attending to her first ever experience of seeing red). (3) QR is this subjective character. So when Mary has an experience for the first time she acquires an ability to think thoughts of the form (2) and this ability allows her to learn identity statements of the form (3). Crucially (3) doesn’t describe a new fact Mary has learned. Instead it tells us something about Mary’s ways of thinking about redness. As a final example, consider Papineau’s quotational account of phenomenal concepts.51 According to Papineau, when Mary has a color experience for the first time she acquires abilities—the ability to imaginatively recreate and introspectively classify color experiences. These abilities enable Mary to think new thoughts, thoughts she couldn’t entertain prior to having her experiences of color. By imaginatively recreating her experience she can think for instance, “This is what people experience when they look at ripe tomatoes.” When Mary imaginatively recreates an experience, the concepts she employs in forming thoughts like this one literally co-opt an instantiation of the experience being thought about. On this view when we form a thought using a phenomenal concept we reproduce an example of the experience the thought is about. This exemplar is then included in the content of thought. When we deploy a phenomenal concept in an act of imagination, what this concept refers to are experiences that appropriately resemble the imaginatively recreated experience.52 When we use a phenomenal concept in forming an

Wittgenstein, qualia, and the autonomy of grammar

51

introspective thought, what the phenomenal concept refers to will be an act of perceptual classification. Introspective uses of phenomenal concepts include the experiences they refer to. What these different accounts of phenomenal concepts have in common is that they all posit concepts available only to persons who have undergone the experiences the concepts have as their referents. They deny that these concepts refer to qualia but they agree with proponents of qualia that to master a phenomenal concept, a subject must be acquainted with the experiences the concept designates. They just disagree about the nature of the experiences in question, with proponents of phenomenal concepts arguing that such experiences are either physical or functional in nature. We have seen how proponents of qualia claim that the terms we use to talk about subjective experience derive their meaning from what they designate, the experiences themselves and their qualia. The story is slightly more complicated in the case of proponents of phenomenal concepts who tend to tell different stories about (1) what fixes a concept’s meaning, and (2) the conditions a subject must satisfy if she is to possess a concept. They agree that a subject must be acquainted with an experience if she is to deploy a phenomenal concept in thinking about the experience in question. However they deny that it is acquaintance with an experience that fixes the meaning a concept expresses. Despite this complication, to which I will return in a moment, proponents of phenomenal concepts and qualia are ultimately agreed that the meaning a phenomenal concept expresses derives from the experience the concept has as its referent. It is this claim that we have seen Wittgenstein dispute. If my reading of Wittgenstein is correct, he would argue that we can refer to such items as sensations or experiences only because we make utterances and form thoughts according to publicly defined rules of grammar which determine what it makes sense to say in this domain. These publicly defined rules make it possible for us to compare experiences and so to identify and re-identify experiences. It is through mastery of a shared language that we can compare experiences and hence identify them. Papineau is the only proponent of phenomenal concepts to directly engage with Wittgenstein’s arguments, so far as I am aware. He writes: [C]ontent does not derive from normative rules, but rather from the kind of non-normative natural facts invoked by causal or teleosemantic theories of representation [ … ]. So, on my view, it is no deficiency in Mary’s concept that she is not sensitive to any normative principles tying its use to public criteria. It is enough that her concept has appropriate causal or teleosemantic credentials, since this in itself will ensure that her concept refers determinately, and that judgements made by using it have definite truth conditions.53 According to Papineau, a phenomenal concept derives its meaning from facts about the causes and biological function of uses that are made of those

52

Julian Kiverstein

concepts. Phenomenal concepts, he conjectures, may have helped our ancestors to better predict the behavior of others, to anticipate their own future experiences, to facilitate reflection on the epistemic standing of their own beliefs.54 For each phenomenal concept there will be something about the use that was made of the concept by our evolutionary ancestors which Papineau is gambling will enable us to fix its content. Thus there is no need to appeal to grammar to do the work of fixing what it makes sense to say about experience. The biological function of our phenomenal concepts will do this work. There is a dispute between Papineau and a Wittgensteinian about how the meaning of our phenomenal concepts is fixed. I want to finish by suggesting that it is an advantage of the Wittgensteinian account that it doesn’t separate the question of how the meaning of our concepts is fixed from the question of what conditions a user of concept must satisfy if he is to count as possessing a concept. Even if Papineau has a satisfactory story to tell about how phenomenal concepts get their meaning, a question on which I will remain neutral, he doesn’t have an adequate story to tell about what it takes to master a phenomenal concept. Arguably to understand a phenomenal concept one must know which experience it is the concept refers to—one must be able to identify the experiences of the type the concept denotes. Papineau describes two forms that such an understanding can take.55 The first account he suggests requires a subject to know that people acquire phenomenal concepts concerning their experiences upon having those experiences: [L]et us suppose that Mary has a companion, Jennifer, who similarly has never seen colours but knows all about colour vision in material terms. Jennifer isn’t shown the piece of paper that Mary sees but is told in material terms that it is red. Then Jennifer who knows that people acquire phenomenal concepts of experience once they have had those experiences, will be able to understand Jane’s “qual” as expressing just such a phenomenal concept, and indeed one which refers to the experience caused in her by seeing something red.56 Does Jennifer have an understanding of the phenomenal concept Mary employs in this case? She understands that the concept Mary has employed has as its referent, the experience caused in Mary upon seeing something red. It is not clear however that Jennifer knows which experience this is, because it is not clear that a description of this kind is a uniquely referring expression. From the possibility of spectrum inversion it follows that there is more than one experience Mary might be undergoing caused by her seeing something red. It follows that grasping this description isn’t sufficient to give Jennifer knowledge of which experience Mary is undergoing. Yet it is plausibly a condition on grasping a phenomenal concept that Jennifer be able to identify which experience the concept has as its referent.

Wittgenstein, qualia, and the autonomy of grammar

53

The second, more demanding, account Papineau describes requires Jennifer to deploy her own phenomenal concept in identifying the experience the concept denotes. The difference from the first case is that this time we allow that Jennifer can see the red piece of paper which Mary can see: Then she (Jennifer) too will acquire the phenomenal concept of seeing something red, and will thus be able to think about the experience referred to be “qual” in the same way in which Mary does, and not just materially.57 Is this really an improvement on the first account? Can Jennifer know which experience it is that Mary is referring to when she uses the expression “qual” just by deploying her own “qual” concept? In order for us to say that Jennifer understands Mary’s “qual” concept we must have grounds for saying that when she deploys her own “qual” concept she is applying the very same concept as Mary. We need to have grounds for saying that she is thinking about seeing something red in the same way as Mary. Papineau seems to assume that because both subjects use their “qual” concepts to think about experiences they are having in the presence of something red they must be using the same concept. However, again we can say that the two subjects may be having different experiences in the presence of something red—one may see something that looks red, the other may see something that looks green. If they are having different experiences, they will each be deploying different phenomenal concepts. Again we have no grounds for saying that Jennifer has understood what Mary means when she uses the word “qual” to express her phenomenal concept. To be fair to Papineau he does notice this difficulty, and he rejects the possibility of spectrum inversion on a posteriori grounds.58 Even if we suppose Papineau has good reasons for rejecting spectrum inversion, still my point stands that he has not given us an adequate account of what is involved in understanding a phenomenal concept. Of course it is probably the case that Papineau does not assign much importance to a theory of understanding. He does after all take concepts to be individuated by their biological function as we have already seen. It is a concept’s biological function—roughly the conditions under which it is used in such a way as to further an organism’s fitness—which fixes what the concept has as its referent. I will return to this point in my conclusion. A Wittgensteinian way around difficulties of this kind would be to invoke the metaphor of moves within language games. How does Jennifer know which experience it is that Mary is talking about? There are rules governing the use of the word “qual” that fix how this word must be used if it is to be used correctly. This rule determines which linguistic moves Mary can legitimately make with this word. The word “qual” is used correctly when it is used to refer to the experience Mary undergoes when she experiences something red. What rule must Jennifer follow if she is to grasp how the word “qual” must be used?

54

Julian Kiverstein

Here it will be useful to introduce a distinction Hintikka and Hintikka make between two kinds of language games, which they call “primary” and “secondary” respectively.59 In a primary language game a subject gives spontaneous expression to an experience through a gesture, expressive bodily movement or exclamation. Using a secondary language game we can talk about an experience separately from its behavioral manifestation as is the case when someone lies; simulates or play acts.60 However our doing so in an important sense presupposes the establishment of meaning brought about by the primary language game. Thus our use of the word “pain,” for instance, is given its primary meaning by spontaneous reactions to pain such as exclamations, bodily movements, facial expressions, etc. We can subsequently replace such spontaneous expressions of pain with the use of the word “pain.” When we do so, we replace one means of expressing pain with another. Someone can then pretend to be in pain thus breaking the connection between pain behavior and pain. However they can do so, only because the primary language game is already established. We understand which experience “pain” refers to then because of the constitutive connection between pain and expressions of pain, of which the use of the word “pain” is one instance. Understanding the word “pain” is thus knowing the place this word has in a language game—a game in which pain experiences are expressed by a variety of different behaviors. Can we make an analogous move for Mary’s use of the word “qual” to explain under what conditions Jennifer can be said to understand this term and so know which experience it refers to? “Qual” is of course different from “pain” in that it denotes a colorexperience—the experience Mary has when she sees red for the first time. What is the language game in this case which Mary’s use of this word depends on for its meaning? The following passage seems to address our question directly: How is he to know what colour he is to pick out when he hears “red”?— Quite simple: he is to take the colour whose image occurs to him when he hears the word.—But how is he to know which colour it is “whose image occurs to him”? Is a further criterion needed for that? (There is indeed such a procedure as choosing the colour which occurs to one when one hears the word “ … ”) “‘Red’ means the colour that occurs to me when I hear the word ‘red’”—would be a definition. Not an explanation of what it is to use a word as a sign. (PI §239) The language games we play with color are extremely complex but one facet of this game will be that our uses of color terms tend to go together with visual impressions of certain sorts. Mary’s use of “qual” is much like an exclamation of pain that occurs along with a pain sensation. It institutes a

Wittgenstein, qualia, and the autonomy of grammar

55

connection between the private mental episode known in a first person way only to Mary—the visual impression she is undergoing—and an expression in a language game. This relation between the visual impression and the use of the word is grammatical or conceptual: it fixes the meaning of the word. It is because of this connection between use of a color term and the occurrence of some visual impression that the visual impression as something private, known in a first person way only to its subject, can enter into a language game. Superficially this looks like the same story that proponents of phenomenal concepts tell about how we can think about our conscious experiences. On Papineau’s account, for instance, when Mary thinks about her first experience of redness she does so by including the token experience she is undergoing in her thought. It looks like Wittgenstein is proposing something similar in saying that “red” means the color that occurs to me when I hear the word “red.” Isn’t this equivalent to saying that the phenomenal concept “red” refers to the color that occurs to me when I token this concept? There is however a key difference. On Wittgenstein’s story the color term “red” gets its meaning from its place in a language game. Rules of grammar determine the moves that can be made within this language game that fix how this term is to be understood. It is because the visual impression of redness is correlated with a language game that we are able to make ourselves understood to one another in talking about our visual impressions. Proponents of phenomenal concepts do not have anything to play the same role as language games, to which they can appeal in explaining how we make sense of phenomenal concepts. This is why they fail to explain how Jennifer can know which experience it is that Mary picks out when she uses “qual.” If I am right, the Wittgensteinian appeal to language games escapes this difficulty. Now it might be objected that the Wittgensteinian hasn’t explained how Jennifer can know to which experience Mary is referring when she uses her “qual” concept, any more than the proponent of phenomenal concepts. All that has been said above is that “qual” refers to a visual impression correlated with a token use of this term. However it could be that each person has a different visual impression when they use this term. Wittgensteinians have a response to this objection: they can appeal to the same argument that was deployed against the possibility of widespread spectrum inversion. In order for us to talk meaningfully about color experiences there has to be some agreement about how the terms constitutive of such talk are to be used. It is this agreement that makes it possible for us to compare our color experiences. In the absence of this agreement, talk of how things look with respect to their color would lack meaning and the practice of comparing experiences would not be able to get off the ground. Yet the very idea of differences in experience, which this objection raises, rests on the possibility of our making sense of comparing experiences. So while it is possible for us to differ in the visual impressions we undergo when experiencing color,61 the language games we engage in talking about such experiences

56

Julian Kiverstein

allow us to side-step such possibilities of variation. They establish sufficient agreement in meaning for us to talk about and compare experiences. The key advantage the Wittgensteinian has over both the proponent of qualia and the proponent of phenomenal concepts is that he can explain how it is possible for a subject to refer to subjective experience. Both proponents of qualia and of phenomenal concepts lack a satisfactory answer to this question. I will conclude by making this failing explicit. The proponents of qualia take the meaning of our experience and sensation terms to be fixed by their referents but they fail to explain how it is possible for us to refer to our experiences and sensations. This can be brought out through reflection on the possibility of widespread spectrum inversion a commitment to qualia entails. Suppose that the meaning of our experience terms was given by their referents. Further suppose that our experiences have qualia such that I have no way of knowing that what I refer to when I talk about my experiences is not something different from what you refer to when you talk about your experience. If we concede the latter possibility, there could be no agreement in what we mean by our experience terms. If we have not accounted for such agreement, we have arguably not accounted for how it is possible to refer to color experiences. We will not have explained how it is possible for different individuals to refer to one and the same experience. A proponent of phenomenal concepts makes appeal to evolutionary function to explain what fixes the meaning of our experience and sensation terms, distinguishing the question of what fixes a concept’s meaning from the question of what conditions must be met if a subject is to possess a concept. However because they fail to return a satisfactory answer to this latter question, they fail to explain how it is possible for us to refer to our subjective experiences. They fail to provide an account of what it is to possess mastery of a phenomenal concept because they fail to explain how the user of the concept can know which experience it is a phenomenal concept designates. This is to fail to explain how it is possible for phenomenal concepts to refer. The Wittgensteinian doesn’t separate a theory of meaning for phenomenal concepts from a theory of understanding. Thus he has a satisfying story to tell about how it is possible for us to refer to our experiences and sensations, something that we have seen proponents of phenomenal concepts and proponents of qualia both fail to explain. Some may regard the autonomy of grammar thesis too high a price to pay. However my suspicion is that it is only once we recognize the autonomy of grammar that we will find a satisfying story to tell about how it is possible for us to talk about subjective experience and make ourselves understood.

Notes 1 This work was carried out as part of the European Science Foundation funded project CONTACT–Consciousness in Interaction.

Wittgenstein, qualia, and the autonomy of grammar

57

2 I shall use the term “qualia” to denote non-representational, intrinsic properties of experience, known through introspection. It is in virtue of an experience’s qualia that there is something it is like for a subject to undergo an experience. Occasionally I will use the term “phenomenal character” rather than “qualia” to refer to properties of experience in virtue of which there is something it is like for a subject to undergo an experience. I will take “phenomenal character” to be a philosophically less loaded term than “qualia.” Many philosophers deny that experiences have non-representational, intrinsic properties, and I take the later Wittgenstein to be amongst them, but nevertheless accept that experiences have phenomenal character. 3 Dennett—perhaps the leading exponent of scepticism about qualia—claims to be following Wittgenstein: “Several philosophers have seen what I am doing as a kind of redoing of Wittgenstein’s attack on the ‘objects’ of conscious experience. Indeed it is. [ … ] My debt to Wittgenstein is large and long standing.” Daniel C. Dennett, Consciousness Explained (London: Penguin, 1993), p. 462. Other philosophers who have taken Wittgenstein to be a skeptic about qualia include Ned Block, “Wittgenstein and Qualia,” in Philosophical Perspectives, ed. John Hawthorne, Philosophical Perspectives series (Oxford: Blackwell, forthcoming); Robert Kirk, Raw Feeling: A Philosophical Account of the Essence of Consciousness (Oxford: Clarendon Press, 1994), ch. 2; Howard Robinson, Perception (London: Routledge, 1994), ch. 4; Sydney Shoemaker, “The Inverted Spectrum,” Journal of Philosophy 79:7 (1982): 357–81; Robert Stalnaker, “Comparing Qualia across Persons,” Philosophical Topics 26 (2000): 385–405. 4 See, for instance, Ned Block, “Inverted Earth,” in Action Theory and Philosophy of Mind, ed. J. Tomberlin, Philosophical Perspectives series (Atascadero, CA: Ridgeview, 1990); Alex Byrne, “Inverted Qualia,” in The Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/archives/win2006/entries/qualia-inverted/ (2006); David Chalmers, The Conscious Mind: In Search of a Fundamental Theory (Oxford: Oxford University Press, 1996), pp. 263–66; Kirk, Raw Feeling, ch. 1. 5 For an example of this kind of reading, see Robinson, Perception, ch. 4. 6 Appeal might be made to PI §258 here to support such a line of argument, see for instance ibid., pp. 95–98. 7 BT §101, p. 493. 8 BT §94, p. 438. 9 BT §94, p. 438. It should be noted that these passages are taken from discussions in which Wittgenstein is attacking the idea of a phenomenological language. I have chosen these passages because they state particularly clearly the distinction between a phenomenological and ordinary physical language Wittgenstein will eventually reject. 10 Ibid., p. 32. 11 Cf. PG §68, §133; Z §320, §331; PI §497. 12 Rush Rhees published an edited version of Wittgenstein’s notes; see Ludwig Wittgenstein, “II: Notes for Lectures on ‘Private Experience’ and ‘Sense Data’,” The Philosophical Review 77:3 (1968): 275–320. An unedited version can be found in Ludwig Wittgenstein, Philosophical Occasions, 1912–1951 (Indianapolis, IN: Hackett, 1993), pp. 200–88. 13 They form the starting point for Shoemaker’s classic discussion of inverted spectra in “The Inverted Spectrum.” More recently Block has returned to these passages to mount an attack on Wittgenstein from the perspective of a friend of qualia. My discussion of Wittgenstein’s treatment of inverted spectra in this section is indebted to Block’s “Wittgenstein and Qualia.” 14 Block, “Wittgenstein and Qualia.” 15 Ibid. 16 For further discussion of this point, see ibid., section 4.

58

Julian Kiverstein

17 PESD, p. 316. 18 I do not mean to imply that this exhausts the ways in which the concept of pain is used. In addition it is used in explaining or justifying behavior, or as a motivation or threat in persuading either oneself or another to act to mention just of the few of the many other ways this concept can be employed. My thanks to D.K. Levy for insisting on this point. 19 A possibility along these lines is discussed ibid. 20 See for an objection along these lines to Kripke’s skeptical reading of Wittgenstein’s account of rule-following, Simon Blackburn, “The Individual Strikes Back,” Synthese (1984): 281–303. 21 PESD, p. 291; also see PI §257. 22 For further discussion of this point to which I am indebted, see Merrill Hintikka and Jaakko Hintikka, Investigating Wittgenstein (Oxford: Blackwell, 1986), chs. 8–11. 23 Friedrich Waismann, Wittgenstein and the Vienna Circle, trans. Joachim Schulte and Brian McGuinness (Oxford: Basil Blackwell, 1979), p. 45. 24 There is some controversy about how exactly Wittgenstein conceived of the individual objects which the constituents of elementary propositions name. Answering this question is made all the harder by Wittgenstein’s failure to provide any concrete examples in the Tractatus. Hintikka and Hintikka argue for a reading of simple objects as objects we are acquainted with in experience (i.e., sense data); see Investigating Wittgenstein, ch. 3. James Griffin treats simple objects as physical point masses, in Wittgenstein’s Logical Atomism (Oxford: Clarendon Press, 1964). Others have treated Wittgenstein as an anti-realist, arguing that we can only identify the referent of an elementary proposition within the context of a proposition. For a discussion of the range of different possible readings, see David G. Stern, Wittgenstein on Mind and Language (New York: Oxford University Press, 1995), pp. 62–65. 25 TLP 2.062, 4.211, 5.134, and 5.135. 26 Wittgenstein wrote this essay in the summer of 1929, his first piece of philosophical writing following his abandonment of philosophy in 1918. He was apparently extremely dissatisfied with the results, refusing to read the paper at the session for which it was written. No doubt his dissatisfaction was in part due to doubts about the notion of a phenomenological language he employs in this essay. 27 RLF, p. 170–1. 28 RLF, p.165. 29 PR §39. 30 PR §1. 31 Stern, Wittgenstein on Mind and Language, p. 138. Noë defends a similar reading, R. A. Noë, “Wittgenstein, Phenomenology, and What it Makes Sense to Say,” Philosophy and Phenomenological Research 54 (1994): 1–42. 32 Stern, Wittgenstein on Mind and Language, p. 138. 33 Waismann, Wittgenstein and the Vienna Circle, p. 46. 34 PR §1. 35 PR §51. 36 Quoted in Stern, Wittgenstein on Mind and Language, p. 139. 37 Noë, “Wittgenstein, Phenomenology, and What it Makes Sense to Say.” 38 PR §1. 39 BT, pp. 78–9. 40 PG, § 133, p. 184. 41 See Z §§331 and 357 and Michael Forster, Wittgenstein on the Arbitrariness of Grammar (Princeton, NJ: Princeton University Press, 2004), pp. 33–39. 42 RFM, III, §75.

Wittgenstein, qualia, and the autonomy of grammar

59

43 It might be argued that the appeal to practical purpose is incompatible with the autonomy of grammar claim. Am I not explaining our use of certain rules by reference to the fact that these rules are useful? Doesn’t this amount to saying our language games are accountable to reality—to the fact that each of the moves in a language game serves some practical purpose? I don’t think the appeal to practical purpose has this consequence. There is, on the one hand, the question of what justifies our employing a particular rule. In answer to this question I have made appeal to Wittgenstein’s talk of rules of grammar as having an application in our lives (see passage from PI §520 quoted above). On the other hand, there is the question of why we follow one rule rather than its contrary. Wittgenstein is quite clear that our employing a particular grammatical principle isn’t a matter of arbitrary choice—in addition to PI §520 see Z §358, MWL, p. 70 and LWL, p. 49 and ch. 3. My suggestion is that while practical purpose cannot justify our employing the grammatical principles we do, it nevertheless acts as a constraint explaining in part why our applying the rules we do isn’t arbitrary. For fuller discussion to which I am indebted, see Forster, Wittgenstein on the Arbitrariness of Grammar, pp. 33–39. 44 Frank Jackson, “Epiphenomenal Qualia,” The Philosophical Quarterly 32:127 (1982): 127–36. 45 Block, “Inverted Earth.” 46 J. Levine, “On Leaving Out What It’s Like,” in Consciousness: Psychological and Philosophical Essays, ed. Martin Davies and Glyn Humphreys, Readings in Mind and Language series (Oxford: Blackwell, 1993). 47 See Brian Loar, “Phenomenal States,” in Action Theory and Philosophy of Mind, ed. J. Tomberlin, Philosophical Perspectives series (Atascadero, CA: Ridgeview, 1990); David Papineau, Thinking about Consciousness (Oxford: Oxford University Press, 2002); John Perry, Knowledge, Possibility, and Consciousness, vol. 79 (7), The Jean Nicod Lectures (Cambridge, MA: MIT Press, 2001); Michael Tye, Consciousness, Color, and Content (Cambridge, MA: MIT Press, 2000). 48 Chalmers, The Conscious Mind. 49 E.g. Peter Carruthers, Phenomenal Consciousness: A Naturalistic Theory (Cambridge: Cambridge University Press, 2000); Tye, Consciousness, Color, and Content; Loar, “Phenomenal States.” 50 Perry, Knowledge, Possibility, and Consciousness. 51 Papineau, Thinking about Consciousness. 52 Cf. ibid., pp. 118–19. 53 Ibid., p. 129. 54 Ibid., p. 116n. 55 Ibid., pp. 129–39. 56 Ibid., p. 130. 57 Ibid., p. 130. 58 Ibid., pp. 131–32 and ch. 7. 59 Hintikka and Hintikka, Investigating Wittgenstein, ch. 11. 60 See PI II, xi. 61 Wittgenstein repeatedly concedes this much, see for instance PI, §272 where Wittgenstein writes: “The essential thing about private experience is really not that each person possesses his own exemplar, but that nobody knows whether other people also have this or something else. The assumption would thus be possible—though unverifiable—that one section of mankind had one sensation of red and another section another.”

3

Understanding (other) minds Wittgenstein’s phenomenological contribution Søren Overgaard and Dan Zahavi

“Not to explain, but to accept the psychological phenomenon—that is what is difficult.”1

3.1 Introduction Back in the fifties and sixties, it seemed obvious to many analytical philosophers that a new method and style of philosophizing had been found, one that was in the process of revolutionizing philosophy. Through careful analysis of actual linguistic usage, philosophy would once and for all escape the muddles of traditional metaphysical theorizing. Facing a delegation of continental philosophers at Royaumont, P. F. Strawson declared that “the actual use of linguistic expressions remains [the philosopher’s] sole and essential contact with the reality which he wishes to understand.”2 And a couple of decades earlier, at Cambridge, Wittgenstein had proclaimed that the problems of philosophy would be solved “by looking into the workings of our language.”3 Philosophical inquiry was no longer to be occupied with phenomena or appearances (Erscheinungen); philosophical investigations were to be “grammatical investigations” that would remind us of “the kind of statement we make about phenomena.”4 Thus, as Wittgenstein stated, “[w]e are not analyzing a phenomenon (e.g. thought) but a concept (e.g. that of thinking), and therefore the use of a word.”5 Since then, however, the philosophical landscape has changed considerably. Assisted by the rise of the philosophy of mind, metaphysics has regained its position in the mainstream of analytical philosophy. And very few contemporary analytical philosophers of mind would subscribe to Strawson’s dictum that the analysis of linguistic usage is the only way for philosophy to proceed. In fact, they unconcernedly discuss mental phenomena as such—consciousness, intentionality, qualia, etc.—without paying much attention to what we say about such phenomena. Nor do the writings of the champions of (one or the other variety of) the philosophy of “what we say” receive much attention in contemporary philosophy of mind (or in

Understanding (other) minds

61

other philosophical debates).6 One reason for this may be that once the conviction that linguistic analysis is the only respectable philosophical method has fallen out of favor, philosophers interested in understanding the nature and workings of the human mind will obviously consider philosophical texts that discuss precisely this phenomenon more relevant than texts discussing various details of our quotidian mental vocabulary. If you are interested in reality, rather than in what we say about reality—and if you don’t believe the only philosophical route to reality goes through what we say about it—why then waste your time on philosophers who explicitly avow that they are primarily concerned with what we say? Obviously, there is a lot that could be said in response to such a hasty and sweeping dismissal of a very heterogeneous group of philosophers. The response we will offer in this chapter is the following. It may (or may not) be that Wittgenstein’s grammatical reminders are of little relevance to contemporary philosophy of mind. However that may be, we find a different kind of descriptive approach in much of Wittgenstein’s later work—in particular the four volumes on the philosophy of psychology. Taken together, the descriptions of the latter kind offer the rough outline of an account of the mind that contemporary philosophy of mind and social cognition should not ignore. This account is one of Wittgenstein’s enduring contributions to philosophy. Moreover, it is one that will not surface as long as Wittgenstein is interpreted strictly as a grammarian, for its basis is phenomenological. Our argument for these controversial claims will be developed in three steps. First, we will suggest that there is reason to doubt that what Wittgenstein (c. 1929–30) called “phenomenology” has much to do with phenomenology of the continental variety. Then, second, we will argue that in the later Wittgenstein, amid grammatical remarks, one finds occasional observations of an obviously phenomenological character. In the final part of the chapter, in which we focus on the four volumes of remarks on the philosophy of psychology, we will then argue that the later Wittgenstein’s phenomenological remarks collectively sketch an account of the mind or mental life that constitutes a compelling corrective to currently fashionable positions in philosophy of mind and social cognition.

3.2 What phenomenology might be Maurice Drury recalls a remark Wittgenstein made sometime around 1930. The prominent logical positivist and member of the Vienna Circle, Moritz Schlick, was reading a paper entitled “Phenomenology” to the Moral Science Club at Cambridge. Wittgenstein said to Drury: “You ought to make a point of going to hear this paper, but I shan’t be there. You could say of my work that it’s ‘phenomenology’.”7 This remark, together with the occurrence of remarks on “phenomenology” in works such as the Philosophical Remarks and The Big Typescript, has mesmerized commentators. It has been widely believed that insofar as convergence were to be found between

62

Søren Overgaard and Dan Zahavi

Wittgenstein’s philosophy and some strand of Continental thought, then Wittgenstein’s phenomenological period would seem the most promising place to look for such convergence.8 Some have even gone as far as to claim that the phenomenology Wittgenstein considered around 1929–30 was so close to the position advocated by continental phenomenologists that the fate of the former would decide the fate of the latter. Merrill and Jaakko Hintikka express this view as follows: Wittgenstein’s conception [of phenomenology] turns out to be closely similar to those of phenomenologists. Their basic idea is precisely the same as [the] Wittgensteinian doctrine [ … ], viz. that we can uncover the conceptual structure of the world by attending to our immediate experience. Hence, if Wittgenstein’s view is deemed absurd, then so must be deemed the whole Husserlian project. Wittgenstein himself expresses his phenomenological viewpoint as late as 1931–32 by saying that “sense-data are the source of our concepts.”9 The Hintikkas go on to claim that Wittgenstein’s subsequent abandonment of central tenets of his phenomenology illustrates that “there cannot be any systematic science of phenomenology of the kind that, e.g., Husserl attempted to build.”10 Although Ray Monk disagrees with the Hintikkas about when to date the onset of Wittgenstein’s flirtations with phenomenology,11 he wants to draw a similar conclusion from Wittgenstein’s ultimate rejection of phenomenology. He cites the following passage from the Philosophical Remarks as an example of how close Wittgenstein’s ideas were, for a brief interval, to the views propagated by continental phenomenologists: Does anything now force me into interpreting the tree I see through my window as larger than the window? If I have a sense for the distance of objects from my eye, this is a justified interpretation. But even then it’s a representation in a different space from visual space, for what corresponds to the tree in visual space is, surely, obviously smaller than what corresponds to the window. (PR, p. 101) Monk remarks that “an obvious and striking similarity between [Wittgenstein’s] project and Husserl’s method of ‘bracketing’” is “almost embarrassingly apparent” in this passage.12 Unfortunately, Monk does not elaborate his interpretation of Husserl’s method of bracketing, so one can only guess in what ways he thinks the passage in question resembles Husserlian ideas. Perhaps his point is that both thinkers envisage a realm of experiential phenomena somehow distinct from the physical world, as well as a method of phenomenological purification designed to capture these elusive phenomena. The point that this is leading up to, however, is crystal clear: Monk wants to

Understanding (other) minds

63

show that whereas Wittgenstein’s flirtation with such “absurdities” was merely a “momentary lapse,”13 the phenomenological tradition—to the detriment of continental European philosophy—remained committed to them.14 Although we are not directly concerned with Wittgenstein’s transition from the Tractatus to the later philosophy, we do need to say something in response to the claim that the demise of Wittgenstein’s phenomenology has far-reaching ramifications for phenomenology in the Husserlian tradition. For if Wittgenstein tested and rejected the project of continental phenomenology in 1929–30, then it seems our idea that there is a phenomenological strand in his later thought must be wrong. Ultimately, everything depends on the precise extent to which Wittgenstein was engaged in the same sort of project as, say, Husserl or Merleau-Ponty. Such a question obviously cannot be properly addressed within the confines of a single chapter. However, in the absence of any convincing (book-length) exegetic account that demonstrates otherwise, we think there are at least two reasons for being wary of assuming such close proximity. First, a story needs to be told about the emergence of Wittgenstein’s phenomenology in the transition between his early and his later thought, before his remarks can be interpreted correctly. Whatever the precise details, that story would crucially involve Wittgenstein’s second thoughts on such things as the color-exclusion problem and its implications for the Tractatus’ conception of logical form. Briefly, Wittgenstein’s phenomenology should be viewed as an attempt to unveil the true logical form of propositions describing, for example, visual phenomena, by attending to these phenomena themselves. As Wittgenstein puts it in “Some Remarks on Logical Form,” The idea is to express in an appropriate symbolism what in ordinary language leads to endless misunderstandings. That is to say, where ordinary language disguises logical structure [ … ], we must replace it by a symbolism which gives a clear picture of the logical structure [ … ]. Now we can only substitute a clear symbolism for the unprecise [sic] one by inspecting the phenomena which we want to describe, thus trying to understand their logical multiplicity. (RLF, p. 29–30) What Wittgenstein envisages is thus a “logical investigation of the phenomena themselves,”15 a “phenomeno-logic” of, for example, vision. Visual space is a logical space with its own logical multiplicity that rules out as sheer nonsense the logical product of the two propositions “The point P in the visual field is blue at time t” and “The point P in the visual field is red at time t.” This is not a standard contradiction, in which, for all truth-values of the two propositions, the product is false. Rather, the assignment of the truth-value T to both propositions is in this case nonsense, since it implies “a greater logical multiplicity than that of the actual possibilities.”16 Wittgenstein continues: “a perfect notation will have to exclude such structures by definite

64

Søren Overgaard and Dan Zahavi

rules of syntax. [ … ] Such rules, however, cannot be laid down until we have actually reached the ultimate analysis of the phenomena in question.”17 On this kind of picture, phenomenology seems tightly bound up with Russell and the early Wittgenstein’s conception of philosophical analysis as uncovering the real logical form of propositions.18 It is far from obvious that this kind of project is the same as that of Husserl or Merleau-Ponty. That does not rule out interesting connections and points of overlap with phenomenology in the continental sense. But surely, it does place the burden of proof squarely on those who claim that the two kinds of phenomenology converge to such an extent that the fate of the one type simply decides that of the other. Casual references to Husserl’s method of bracketing, with no explanation given, or roundly general claims to the effect that Wittgenstein and Husserl are both interested in uncovering the conceptual structure of the world by attending to immediate experience surely will not do.19 Second, the very passages in which the Hintikkas and Monk argue for a close proximity between the Husserlian tradition and Wittgenstein’s phenomenology suggest that matters are much less clear-cut. As an example of Wittgenstein’s commitment to a phenomenological viewpoint, the Hintikkas cite his remark that “sense-data are the source of our concepts.” Whatever precisely he had in mind, Wittgenstein does seem to have associated phenomenology with the study of sense-data in isolation from everything else, in particular from physical objects.20 He even alludes to the need for a special, “primary” or “phenomenological language” to capture such data in the required isolation.21 And the passage quoted by Monk may perhaps illustrate what might be involved in an examination of visual sense-data as opposed to spatio-temporal objects. But it is less clear that there is any real convergence with Husserlian or Merleau-Pontian phenomenology in any of this. In order to see how heavily the onus of proof weighs down on those who think there is such a convergence, let us briefly say something about what we take continental phenomenology to involve. One thing Husserl and the author of the Philosophical Remarks would seem to agree on is that it is the business of phenomenology to offer faithful descriptions of experiential phenomena. This descriptive agenda is encapsulated in Husserl’s so-called “principle of all principles.” This principle stipulates that “everything originarily (so to speak, in its ‘personal actuality’) offered to us in ‘intuition’ is to be accepted simply as what it is presented as being, but also only within the limits in which it is presented there.”22 Note that this is a remark about the practice of the phenomenologist; it does not restrict the range of phenomena that the phenomenologist might study. So, for example, it does not say that only perceptual experiences and their objects may be described phenomenologically. Wishes, fantasies, acts of remembering, and so forth, are all legitimate phenomena in the phenomenological sense. What the principle of all principles stipulates is that, whatever the phenomenologist chooses to study, she must take care to describe it within the limits in which it presents itself.

Understanding (other) minds

65

That said, perception does of course have a special interest to the phenomenologist, because it is an experience that—unlike a recollection or an act of imagination—gives us the objects as present, “in flesh and blood,” before the perceiver. And what are the phenomena of immediate perceptual experience, according to Husserlian phenomenology? As the following quote from Husserl’s Ideas suggests, they are not terribly outlandish:23 I am conscious of a world endlessly spread out in space, endlessly becoming and having endlessly become in time. I am conscious of it: that signifies, above all, that intuitively I find it immediately, that I experience it. By my seeing, touching, hearing, and so forth, and in the different modes of sensuous perception, corporeal physical things with some spatial distribution or other are simply there for me, “on hand” in the literal or the figurative sense, whether or not I am particularly heedful of them and busied with them in my considering, thinking, feeling, or willing. Animate beings too—human beings, let us say—are immediately there for me: I look up; I see them; I hear their approach; I grasp their hands; talking with them I understand immediately what they objectivate [vorstellen] and think, what feelings stir within them, what they wish or will.24 The list goes on, of course. But there is no need to elaborate on it, as the general lesson should already be clear: What is “immediate experience” or “immediate perception” supposed to mean if not precisely the kind of perceptions we enjoy most of our waking lives, and have enjoyed for as long as we remember—long before we discovered philosophy? And how could a description of such experience be correct, unless it crucially involved reference to, among other things, visual, auditory, and tactile perceptions of spatio-temporal objects of various kinds? Surely, insofar as the task is to describe perceptual experience, without introducing any unwarranted assumptions, my perception of my coffee-cup should precisely be described as a perception of a coffee-cup: a spatial, material thing, standing on a table, in a room, that is, embedded in a wider context of physical things. The method of bracketing does not change any of this. As Husserl emphasizes, there is a fundamental sense in which “everything remains as of old” within the phenomenological epoché.25 What the brackets signify is merely that the phenomenologist has to stick to describing the phenomenon—say a perceived object—strictly as it presents itself. That is, the brackets do not isolate the phenomenon from the spatio-temporal world as such (for perceived objects typically present themselves precisely as belonging to that world). But they are intended to preserve and protect the phenomenon from the manifold metaphysical assumptions we might be tempted to bring to bear on it—such as, for example, the assumption that the real content or object of immediate experience must be something that belongs to a realm distinct from the ordinary world. Thus, the epoché has its place in

66

Søren Overgaard and Dan Zahavi

the context of Husserl’s attempt to implement the principle of all principles in the practice of phenomenology. Note how well this ties in with remarks of the later Wittgenstein, presumably directly against his earlier phenomenological views: But remember the meaning in which we learn to use the word “see”. We certainly say we see this human being, this flower, while our optical picture—the colours and shapes—is continually altering, and within the widest limits at that. Now that is just how we do use the word “see”. (Don’t think you can find a better use for it—a phenomenological one!) (RPP I, §1070) What Wittgenstein here expresses in the linguistic mode converges precisely with what Husserl would express in the material mode. We say that we see human beings, flowers, and so forth, because those kinds of things are precisely the kinds of things that we do see. We should not think we can come up with a better account of what we see—say, an account in terms of sense-data. Recall, finally, the passage from the Remarks about the size of the tree seen through a window. Wittgenstein remarks that it might be a “justified interpretation” (berechtigte Deutung) to conceive of the tree as larger than the window. But “it’s a representation in a different space from visual space, for what corresponds to the tree in visual space is, surely, obviously smaller than what corresponds to the window.”26 If this passage is read as a straightforward description of immediate experience it looks very dubious, to put it mildly.27 When looking out windows, do we ceaselessly ponder which interpretation of the sizes of the various objects we see is the justified one? Who would under normal circumstances say that what they experience in visual space as a small object is something they interpret as a large object in some other space? Surely, the following statement by Merleau-Ponty is truer to the phenomenological facts of the matter: It is frequently said that I restore the true size [of a perceived object] on the basis of the apparent size by analysis and conjecture. This is inexact for the very convincing reason that the apparent size of which we are speaking is not perceived by me. It is a remarkable fact that the uninstructed have no awareness of perspective and that it took a long time and much reflection for men to become aware of a perspectival deformation of objects.28 It is precisely when we bracket our preconceptions—philosophical or otherwise—about what must be going on in this sort of perception that we become able to attend to what actually is going on.29 The phenomenological facts lend no support to an account in terms of interpretations of visual sense-data.

Understanding (other) minds

67

3.3 The later Wittgenstein and phenomenology Let us now turn to issues that are more central to our agenda. Is there a sense in which the later Wittgenstein can be said to offer phenomenological descriptions? A few authors have ventured to argue that there is; but their claims have not been met with approval from the majority of Wittgenstein scholars.30 And one thing is beyond dispute: on the rare occasions when the later Wittgenstein makes explicit references to phenomenology, he is unequivocally dismissive of it. In the Remarks on Colour, for instance, he states that while the temptation to believe in something like phenomenology may be great, there really is no such thing.31 To understand properly the meaning of such statements, however, one would again have to inquire what Wittgenstein means by phenomenology. If, for example, he means something like the investigation of some realm of sense-data distinct from the world of ordinary objects, then, as we have just seen, phenomenologists would agree with the later Wittgenstein that there is no such thing. In this section of the chapter, however, we will refrain from inquiring further into Wittgenstein’s use of the term “phenomenology.” What we want to do, instead, is to offer some support for the claim that in the philosophy of the later Wittgenstein, there are phenomenological descriptions in the Husserlian or Merleau-Pontian sense. We will consequently claim that there is something of a blind spot in orthodox accounts of the philosophical project of the later Wittgenstein. Consider how Peter Hacker, in his classical study Insight and Illusion, contrasts the philosophy of the later Wittgenstein with more traditional philosophical projects: Clarification of concepts, description of conceptual relationships, is not theory construction. The resolution of philosophical problems lies in the description of the grammar of expressions (such as “mind”, “thinking”, “speaking a language”) not in building theories (about the mind, thought or language).32 Just a couple of pages before the quoted passages Hacker distinguishes sharply between Wittgenstein’s grammatical approach and Michael Dummett’s quest for a philosophical theory of meaning.33 Thus, Wittgenstein’s grammar should not only be contrasted with traditional metaphysical theories (about mind, matter, or whatever), but equally with modern analytical aspirations to theory building within the philosophy of language. There is no reason to disagree with any of this; but if we try to present these contrasts schematically, then it becomes clear that there is one option tacitly left out by Hacker (see Table 3.1). Perhaps a reason why Hacker feels no need to mention a fourth possibility is that when he was composing the first edition of Insight and Illusion, there was no doubt in anyone’s mind (in Oxford and Cambridge at any rate) that linguistic analysis was the present and the future of philosophy. History

68

Søren Overgaard and Dan Zahavi

Table 3.1 Giving descriptions of…

Building theories about…

The world, experiences, “phenomena”

<empty>

Traditional metaphysical theories

Language

Description of language use (the “grammar” of the later Wittgenstein)

Theories of language or “meaning” (Dummett and others)

appears to have proven that assumption wrong, as we pointed out in the introduction to this chapter. Whatever the reason for Hacker’s neglect, however, there is a genuine fourth option, and the phenomenological tradition is characterized by its efforts to develop this alternative. We have already indicated that phenomenology is concerned to describe carefully and accurately the world as it is given to us in our experiences, using a method of bracketing to keep prejudices and unwarranted theoretical constructions at bay. Our claim is that Wittgenstein, in his later philosophy, covers the whole left-hand side of the schematic illustration, weaving descriptions of language use together with descriptions of experiential phenomena. Presumably no one would disagree that the descriptive agenda is one shared by the later Wittgenstein. He declares in the Philosophical Investigations that “[w]e must do away with all explanation and description alone must take its place.”34 And the point receives its briefest possible formulation in the prescription “don’t think, but look!”35 However, on behalf of the orthodox view it might be pointed out that Wittgenstein would be critical of the phenomenological emphasis on experience. The phenomenon par excellence for the later Wittgenstein, as Marie McGinn has put it (in a quasi-Heideggerian style of hyphenation), is “language-in-use”36 rather than experience. Languagein-use—that is, the kinds of things we say about things, the circumstances under which we say it, and the behavior with which we accompany our words—is something that can be studied from a third-person perspective, whereas the phenomenological approach essentially relies on a first-person perspective. And Wittgenstein is critical of the philosophical appeal to the first person: “The examples which philosophers give in the first person should be investigated in the third.”37 There seems to be a pronounced conflict here between Wittgenstein’s views on philosophical method and the basic principles of phenomenology. But we want to suggest that the real target of Wittgenstein’s criticism is an introspective approach—in the sense of a self-analyzing, self-observing enterprise38— and this critical agenda does not place him in opposition to phenomenology. For one thing, it is a mistake to think that the practice of phenomenology involves “a special technique of introspection.”39 In fact, Husserl categorically rejects the attempt to equate the notion of phenomenological intuition with a type of inner experience or introspection40 and even argues that the

Understanding (other) minds

69

very suggestion that phenomenology attempts to restitute the method of introspection (innerer Beobachtung) is preposterous and perverse.41 Husserl’s stance on this issue is fully shared by the other phenomenologists. Heidegger, for example, not only denies that his own analysis of the existential structures of Dasein is a psychological analysis;42 he also writes that the attempt to interpret Husserl’s investigations as a kind of descriptive psychology completely fails to do justice to their transcendental character. In fact, as Heidegger adds, phenomenology will remain a book sealed with seven or more seals to any such psychological approach.43 What motivates this categorical dismissal? Phenomenology is certainly interested in the phenomena and in their conditions of possibility, but phenomenologists would typically argue that it would be a metaphysical fallacy to locate the phenomenal realm within the mind, and to suggest that the way to access and describe it is by turning the gaze inwards (introspicio). As Husserl had already pointed out in the Logical Investigations, the entire facile divide between inside and outside has its origin in a naïve commonsensical metaphysics and is phenomenologically suspect and inappropriate when it comes to understanding the nature of intentionality.44 But this divide is precisely something that the term “introspection” buys into and accepts. To speak of introspection is (tacitly) to endorse the idea that consciousness is inside the head and the world is outside. In opposition to this idea, MerleauPonty stresses that “Inside and outside are inseparable. The world is wholly inside and I am wholly outside myself.”45 In a similar vein, Heidegger denies that the relation between Dasein and world can be grasped with the help of the concepts “inner” and “outer.”46 Husserl, Heidegger and Merleau-Ponty are not particularly interested in scrutinizing the details of their own subjective lives. Phenomenology is not concerned with establishing what a given individual might currently be experiencing. Phenomenology is not interested in qualia in the sense of purely individual data that are incorrigible, ineffable, and incomparable, nor is it interested in psychological processes (in contrast to behavioral processes or physical processes). It is closer to the truth to say that the philosophical interest of the phenomenologists pertains to the world—not the world as described by quantum mechanics or molecular biology, but the world “as it strikes us in everyday life,” to borrow Gregory McCulloch’s phrase.47 But the phenomenological thematization of the world as we experience it in everyday life does involve a special reflective stance. Phenomenology is interested in the very dimension of givenness or appearance and seeks to explore its essential structures and conditions of possibility. When we go about our everyday business, we do not pay much attention to the experiential structures that allow us to perceive and interact with houses, tables, coffee-cups, and people. We are usually occupied with other, more practical matters: say picking up the cup to drink the coffee it contains. Whereas a phenomenologist would argue that the philosophical move must involve a reflective turn, and thereby a new perspective on one’s experiential life, it

70

Søren Overgaard and Dan Zahavi

would be wrong to think it involves introspective scrutiny of what goes on inside oneself. To attend to coffee-cups as they appear in this or that experience is still to attend to coffee-cups. The phenomenological investigation of the field of presence is beyond any divide between psychical interiority and physical exteriority, since it is an investigation of the dimension in which any object—be it external or internal—manifests itself.48 Phenomenology aims to disclose structures that are inter-subjectively accessible, and its analyses are consequently open for corrections and control by any (phenomenologically attuned) subject. In fact, as Merleau-Ponty points out in Phenomenology of Perception, the world of experience, the phenomenal field, is not some inner world, nor is the phenomenon a state of consciousness or a mental fact the experience of which requires a special act of introspection. Merleau-Ponty ends up declaring that phenomenology is distinguished in all its characteristics from introspective psychology and that the difference in question is a difference in principle. Whereas the introspective psychologist considers consciousness as a mere sector of being, and tries to investigate this sector in the same way the physicist tries to investigate his, the phenomenologist realizes that consciousness ultimately calls for a transcendental clarification that goes beyond common sense postulates and brings us face to face with the problem concerning the constitution of the world.49 Ultimately, therefore, we should see the field of givenness, the phenomena, as something that questions the very subject– object split, as something that stresses the co-emergence of self and world. The phenomenological analysis consequently amounts to more than a mere compilation of descriptive findings. Its systematic import can only be fully appreciated the moment the link to the overarching transcendental considerations is made visible. For one thing, this entails a showdown with metaphysical realism and scientism. Phenomenology endorses a this-worldly conception of objectivity and reality and seeks to overcome the skepticism that argues that the way the world appears to us is compatible with the world really being completely different.50 Phenomenology also rejects the view— currently endorsed by many naturalists—that natural science is the measure of all things, of what is that it is, and of what is not that it is not, to quote Sellars.51 For phenomenology, science is not simply a collection of systematic interrelated justified propositions. Science is performed by somebody; it involves a specific theoretical stance towards the world. This stance did not fall from the sky; it has its own presuppositions and origin. Scientific objectivity is something to strive for, but it rests on the observations and experiences of individuals; it is knowledge shared by a community of experiencing subjects and presupposes a triangulation of point of views or perspectives. This is, of course, not to say that there is no third-person perspective, but merely that such a perspective is exactly a perspective from somewhere. It is a view that we can adopt on the world. It is a perspective founded upon a first-person perspective, or to be more precise it emerges out of the encounter between at least two first-person perspectives; that is, it involves inter-subjectivity.

Understanding (other) minds

71

In short, if we wish to understand knowledge, truth, objectivity, meaning and reference we will have to investigate the forms and structures of intentionality that are employed by the cognizing and acting subjects. Failing to do so, failing to effectuate the reflective move of transcendental thought, would, in the eyes of the phenomenologists, be to succumb to a naïve objectivism. How different is this general take on the role of philosophy from that espoused by Wittgenstein when he declares that the business of philosophy is with matters of meaning and not with matters of fact; that its province is the domain of sense and nonsense, and not the domain of empirical truth or falsehood? For Wittgenstein, philosophy investigates and describes the bounds of sense: that is, the limits of what can coherently be thought and said. The boundary between what does and what does not make sense, between what is meaningful and what transgresses the bounds of sense, is determined by the concepts we use and the practices we engage in. Thus, rather than engaging in first-order claims about the nature of things (which it leaves to various scientific disciplines) philosophy concerns itself with the conceptual and pragmatic preconditions for any such empirical inquiries.52 It should be clear by now that when we say that there are phenomenological descriptions in the later Wittgenstein, we do not thereby subscribe to Kripke’s thesis that the method of the later Wittgenstein “is deeply introspective.”53 In fact, we think that as a general claim this is false. What we mean is rather that we find in the writings of Wittgenstein a significant number of remarks that can be characterized neither as descriptions of the use of words, nor as third-person descriptions of behavior, practices, or anything of that sort; remarks that are most plausibly interpreted as essentially appealing to first-person experiences of a particular kind of object or phenomenon (or the absence of such). In the next section of this chapter, we will quote a number of passages from the later Wittgenstein that we think offer the contours of a phenomenological account of social cognition. But let us at this point offer a couple of examples of the kind of remarks that we believe contain phenomenological claims. The first example is from the Remarks on Colour—the work in which Wittgenstein rejects phenomenology in no uncertain terms. Returning to the discussion of seeing distant objects, Wittgenstein now writes as follows: Isn’t this similar to the fact that we often see a distant object merely as distant and not as smaller? Thus we cannot say “I notice that he looks smaller, and I conclude from that that he is farther away,” but rather I notice that he is farther away, without being able to say how I notice it. (RC III, §171)54 This passage from the late work on color is not only significant because it seems to contradict Wittgenstein’s statement on the same topic in the Philosophical Remarks, and because it expresses the very same view that Merleau-Ponty

72

Søren Overgaard and Dan Zahavi

advocates. It is also significant because of the type of remark it is—that is, the kind of issue it addresses. Wittgenstein is not talking about the use of a word (say, “small”) here. The reason why we cannot say that the distant person looks smaller, etc., is not that there is no conceivable use for this kind of expression. In fact, we sometimes say this sort of thing to children— “Look, Aunt Dorothy is smaller than my thumb!”—much to their amusement. The reason why we can’t say it here is obviously that it doesn’t fit the experience. Someone who generally had to make inferences to establish that objects were distant (rather than small) would have a very different experiential life from ours. As Wittgenstein correctly notes, we usually see the person as far away, directly, without basing this on any sort of conclusion. This is clearly a phenomenological point. A similar case is the following passage. “Do I have two visual impressions, under normal circumstances, when I am seeing, so that I notice that my threedimensional visual impression is composed of two visual images? Certainly not.—So I cannot take three-dimensionality away from seeing.”55 These sentences are hardly interpretable as grammatical descriptions of our use of the word “seeing.” To be sure, we would never say that we have two visual impressions, but nor would we know what to do with the statement that “I cannot take three-dimensionality away from seeing,” unless it was meant as a statement about experience. Wittgenstein might be trying to point out a logical or essential connection between the concepts “three-dimensionality” and “seeing,” but he does so by appealing to phenomenological facts about seeing. We must consult our visual experience, rather than language-in-use. Consider also the following remarks on the feeling of joy: Now granted—although it is extremely doubtful—that the muscular feeling of a smile is a constituent part of feeling glad;—where are the other components? Well, in the breast and belly etc.!—But do you really feel them, or do you merely conclude that they must be there? Are you really conscious of these localised feelings? And if not—why are they supposed to be there at all? (RPP I, §456) Wittgenstein’s immediate target is William James’s account of emotion; but his remarks are also pertinent to recent theories of feelings and emotions, such as that of Antonio Damasio.56 If the feeling of joy is supposed to be composed of sensations in the face, the stomach, and the chest, then such sensations should be what we feel when we feel joyful. But Wittgenstein directs our attention to our actual experience of joy, and asks us whether we are really conscious of these various sensations. And if not, aren’t we simply constructing theories about what must be there in the experience, instead of paying attention to the experience itself ? Our final example is a related set of remarks from the Philosophical Investigations:

Understanding (other) minds

73

“My kinaesthetic sensations advise me of the movement and position of my limbs.” I let my index finger make an easy pendulum movement of small amplitude. I either hardly feel it, or don’t feel it at all. [ … ] And this sensation advises me of the movement?—for I can describe the movement exactly. [ … ] I may be able to tell the direction from which a sound comes only because it affects one ear more than the other, but I don’t feel this in my ears; yet it has its effect: I know the direction from which the sound comes [ … ]. (PI II, viii, p. 185/158)57 Under attack here is the prima facie plausible idea that there are certain sensations or feelings that inform me about the position and movement of my limbs, and the direction of a sound. Wittgenstein does not deny that there is a physiological story in which, for example, sound waves affecting my ears unequally play a significant role in the explanation of my ability to tell where a particular sound comes from. But he is denying that in the experiential story, I am conscious of particular sensations in my ear, on the basis of which I conclude what direction the sound is coming from. Nor do I determine my own movement on the basis of special kinaesthetic sensations. Both ideas falsify the experience by making it too indirect. They postulate sensations as the primary objects of attention in an attempt to explain how we know what we know; but in both cases, we know what we know directly, we do not know it because we notice some sensation first.58 As a good phenomenologist, Wittgenstein’s recommended antidote in both cases is again to consult our own experience: are we really aware of any such mediating sensations? Or are we instead being offered constructive explanations of what must be there?59

3.4 Wittgenstein’s phenomenology of inter-subjectivity Common to the examples we have just offered is their negative aim. That is, Wittgenstein employs phenomenological description in an attempt to counter tempting accounts of what goes on when we recognize a familiar object, see something, feel sad or happy, discover something about another person’s mental state, imagine something, are immediately aware of our own posture and movement, and so on and so forth. This negative phenomenological strategy is one Wittgenstein applies to a long list of topics. Often the aim is not only negative, but straightforwardly eliminative: the aim is to make us realize that we are tempted to postulate sensations, feelings, or inferences that are not actually part of the experience. Arguably, nobody can match Wittgenstein as a “negative” phenomenologist. In fact, however, Wittgenstein’s contribution to a phenomenological account of the mental has a positive side to it as well. Though perhaps such remarks are less frequent, Wittgenstein

74

Søren Overgaard and Dan Zahavi

sometimes offers phenomenological descriptions of what we do experience in this or that situation. In particular, as we will show in this section, he contributes significantly to an expressive account of mental life, and thereby offers an essential corrective to the current debates about social cognition. Wittgenstein’s most consistent phenomenological theme is probably that of seeing aspects or seeing as.60 In these discussions, it seems plausible to say that he is addressing certain peculiar features of our concept of seeing. But it is clear that he is also constantly appealing to the phenomena of visual experience, to what we see as we see it. There are frequent references to what we experience (erleben) and are conscious of.61 This ought to be no surprise. In fact, it would be quite difficult to understand what Wittgenstein’s discussion might be about, if he did not—by way of ambiguous pictures such as Jastrow’s famous duck/rabbit—appeal to our own first-person experience of aspect change. It is no coincidence that when Wittgenstein introduces the notion of noticing an aspect in the second part of the Philosophical Investigations, he does so precisely by reference to a particular experience of his: “I contemplate a face, and then suddenly notice its likeness to another. I see that it has not changed; and yet I see it differently. I call this experience ‘noticing an aspect’.”62 The ideas we want to broach here are connected with Wittgenstein’s discussions of seeing as, although they do not directly involve ambiguous pictures. Consider Wittgenstein’s well-known contrast between attributing pain to a stone and to a wriggling fly: “Look at a stone and imagine it having sensations.”63 If we try to do that, Wittgenstein argues, we will find that this is a very difficult thing to do. A stone simply does not present itself to us as a possible candidate for feeling itches or pain. “And now look at a wriggling fly and at once these difficulties vanish and pain seems able to get a foothold here, where before everything was, so to speak, too smooth for it.”64 The agenda might be grammatical in the sense that Wittgenstein wants to make a contribution to a description of our use of pain. But again, there is no question that Wittgenstein is asking us to consult our experience, rather than our actual use of any word. To repeat, we are of course not saying that Wittgenstein relies on introspection. Rather, he is suggesting that we consider things (stones and wriggling flies) as they strike us in everyday life, and this is precisely what makes his approach phenomenological in the relevant sense. So according to Wittgenstein, there is something about the perceptual appearance of animate creatures that makes mental predicates applicable to them, as it were. This is not a very revolutionary claim in itself, perhaps. But Wittgenstein offers a striking set of phenomenological elaborations of the point, concentrating on the case of human beings. Such phenomenological remarks are often surrounded by firm demands that only grammatical or conceptual investigations be accepted in philosophy, but the remarks themselves seem completely unbound by such demands. Consider the following passages:

Understanding (other) minds

75

[ … ] In general I do not surmise fear in him—I see it. I do not feel that I am deducing the probable existence of something inside from something outside; rather it is as if the human face were in a way translucent and that I were seeing it not in reflected light but rather in its own. (RPP II, §170) “We see emotion.”—As opposed to what?—We do not see facial contortions and make the inference that he is feeling joy, grief, boredom. We describe a face immediately as sad, radiant, bored, even when we are unable to give any other description of the features.—Grief, one would like to say, is personified in the face [ … ]. (RPP II, §570)65 We don’t see the human eye as a receiver; it seems, not to let something in, but to send out. The ear receives; the eye looks. [ … ] When you see the eye, you see something go out from it. You see the glance of the eye [ … ]. (RPP I, §1100) Certainly I too say that I see the glance that you throw someone else. And if someone wanted to correct me and say I don’t really see it, I should hold this to be a piece of stupidity. (RPP I, §1101) In passages such as these, Wittgenstein is trying to dislodge and replace a familiar conception of mind and body, according to which a person’s mental life is something essentially confined to an “inner realm,” accessible only to its possessor, while the body is reduced to a merely external container. According to Wittgenstein, fear, joy, grief, and so on, are mental phenomena that can be immediately perceptible in human facial expressions.66 There is no gap between the external face and the internal mental phenomenon, which must be bridged by means of inferences. In fact it can be quite the other way around. As Wittgenstein observes, we are sometimes able right away to describe a face as happy or sad, whereas we would be hard pressed to describe its features in purely external physical or geometrical terms.67 The difficult task here is not to get from clearly discernible physical features to hidden mental phenomena; it is rather to get from the clearly discernible mental phenomena to the purely physical features, which can only be reached by means of a process of abstraction. Wittgenstein is actually waging a war on two fronts here. As he emphasizes, it is “misleading to describe the genuine expression as a sum of the expression and something else, though it is just as misleading [ … ] if we say that the genuine expression is a particular behaviour and nothing besides” (PESD, p. 263). Wittgenstein is trying, then, to undermine the type of account that would drive a wedge between behavioral expression and mental

76

Søren Overgaard and Dan Zahavi

state, and at the same time he is combating the behaviorist alternative. But his agenda is not merely negative and critical. Although what he says does not amount to any theory or doctrine of mind and body, he does put something in the place of the doctrines he is trying to dislodge. Wittgenstein is trying to persuade us to think of mental life as something that is, in various ways and to varying degrees, expressed in bodily behavior.68 “Genuine expression” is not an external sign indicating (correctly) the obtaining of some hidden internal state. It is nothing that stands between one subject and the mental state of another subject.69 Rather, it is the peculiar way in which a mental phenomenon is inter-subjectively apparent.70 This move is intended to collapse the picture of mind and body that the behaviorist and her Cartesian opponent both take as their point of departure. Wittgenstein’s alternative picture is deceptively simple: “only of a living human being and what resembles (behaves like) a human being can one say: it has sensations; it sees; is blind; hears; is deaf; is conscious or unconscious.”71 Human beings (and other animate creatures) are the sorts of things that express mental phenomena; material bodies are not, nor are immaterial minds.72 We have to see our way to understanding the embodied or bodily subject—the whole human being—as the subject of mental states. Now, it is true that Wittgenstein does articulate this point by way of reference to what one can say. But it is important to emphasize that what he is offering is not simply a note on ordinary linguistic usage that carries no implications for reality. Rather, Wittgenstein also claims to be saying something about the nature of the mind and mental states; and, as we have been suggesting, in so doing he relies heavily on phenomenological descriptions of the expressiveness of the living, bodily, human being: And so, too, a corpse seems to us quite inaccessible to pain. [ … ] If anyone says: “That cannot simply come from the fact that a living thing moves about in such-and-such a way and a dead one not”, then I want to intimate to him that this is a case of the transition “from quantity to quality.” (PI §284) It is exactly the crucial quality of expression that Wittgenstein’s imaginary interlocutor misses on this occasion, failing to notice in the behavior of a living thing anything beyond mere physical movement. Wittgenstein reminds such an opponent of the phenomenological facts, the facts as we all experience them: “Think of the recognition of facial expressions. Or of the description of facial expressions—which does not consist in giving the measurements of the face.”73 One question that could be raised at this point concerns the extent to which Wittgenstein’s opponents are precisely imagined. That is, it may seem as if Wittgenstein is merely battling against obsolete positions such as behaviorism and Cartesian dualism. We think nothing could be further from the

Understanding (other) minds

77

truth. It is correct that very few contemporary philosophers and cognitive scientists would buy into the immaterialism that is an essential part of the Cartesian position. However, that does not mean that Wittgenstein’s phenomenological observations have no critical potential vis-à-vis contemporary theorizing about the mind. For what Wittgenstein directs his phenomenological aim at is not primarily the thesis of immaterialism. Rather, he criticizes the notion that mental phenomena are essentially “inner” and “hidden”; his target is the idea of the body “as a façade behind which the mental powers are at work.”74 And it is perfectly possible to be a staunch materialist and still maintain something like this. To put it differently, Wittgenstein’s phenomenological considerations are designed to work against what McCulloch has usefully dubbed “behaviour-rejecting mentalism”75—a position one still subscribes to if one believes everything mental really reduces to neuronal activities in the brain. Consider, for example, the following statement: The brain is all we have for the purpose of representing the world to ourselves and everything we can use must be inside the brain. Each of our beliefs must be possible for a brain in a vat because each of us is precisely a brain in a vat; the vat is a skull and the “messages” coming in are coming in by way of impacts on the nervous system.76 This suggests a robust Cartesian materialism, according to which the mind is simply identical with the brain, and the rest of the body is considered a mere container (a vat) for the brain. There is a straightforward sense in which, on this sort of picture, mental phenomena really are hidden behind the bodily façade. Similar quasi-Cartesian assumptions underlie most of the positions on offer in a contemporary debate concerning social cognition.77 Most contributors to the debate adopt one form or another of two frequently opposed theoretical outlooks, called the “theory-theory of mind” and the “simulation theory of mind.” Generally speaking, theory-theorists claim that our attribution of mental states to others (and indeed to ourselves) rests on a quasiscientific theory the purpose of which is to explain observable behavior. In other words, whenever we interpret others as being angry or in pain, say, we are implicitly applying a theory that explains the publicly accessible behavior (frowning, screaming, etc.) in terms of inaccessible mental states and processes. There is some discussion among theory-theorists as to whether the theory of mind is innate, or whether it is acquired in infancy and early childhood; but most theory-theorists agree that what is going on when we attribute mental states to others is a type of inference to best explanation.78 Simulation theorists disagree with most of this. In particular, they deny that social cognition essentially involves theoretical hypotheses and inferences. A crucial problem with the theory-theory approach is that it seems to represent our fundamental attitude to others as a detached, observational

78

Søren Overgaard and Dan Zahavi

and theoretical one. Simulationists hold that this cannot be right. Rather, our interpretation of others as subjects of pain, anger, joy, and so on, must draw upon emotional and motivational resources that are ignored on the neutral, quasi-scientific account of the theory-theorists. Though all avowed simulationists (unsurprisingly) agree that social cognition crucially involves a kind of simulation, there are some important differences between individual theorists. In its purest form, however, simulation theory claims that when I understand the mind of another, I perform a kind of imaginative projection of myself into his or her “mental shoes.” My access to other minds thus relies on my introspective access to my own mind.79 My extensive, intimate first-hand knowledge of my own feelings, emotions, thoughts, etc. is what I must exploit when I set out to grasp the mind of another. When I see another in a particular situation, I imagine what I would feel and how I would react if I were in that situation; that is how I grasp the other’s mental state. In the context of Wittgenstein’s phenomenology of social cognition, the most interesting thing about this debate is not the many ways in which theory-theory and simulation theory are opposed to each other, but rather the fundamental agreement underlying their dispute. Both parties agree on an important part of what theory-theorist Alan Leslie states in the following passage: One of the most important powers of the human mind is to conceive of and think about itself and other minds. Because the mental states of others (and indeed of ourselves) are completely hidden from the senses, they can only ever be inferred. Thinking about these unobservable states is a subtle business indeed, but in one way or another an essential part of our social life.80 A simulation theorist would of course deny that we can only ever infer the mental states of others, believing that we have a non-inferential, projectivesimulative grasp of them. But a simulationist like Goldman would agree that others’ mental states are unobservable. In other words, both parties to the debate would deny that it is possible to experience other minded creatures and their pains, emotions, etc. Indeed, this is why we need to rely on and employ either theoretical inferences or internal simulations in the first place. Both accounts consequently share the view that the minds of others are hidden, and they consider one of the main challenges facing a theory of social cognition to be the question of how and why we start ascribing such hidden mental entities or processes to certain publicly observable bodies. As should be clear from what we have already said, Wittgenstein would question this very setup. He would argue that the appeal to either theory or simulation is unwarranted since it is motivated by a too impoverished conception of what is experientially available. The phenomenological points assembled above remind us that sometimes we do see the emotions and sensations of others, that their feelings, thoughts, and desires are not essentially

Understanding (other) minds

79

hidden from us, but are rather expressed in their behavior. But in order to fully appreciate the importance of Wittgenstein’s insights to this debate, it is helpful to spell out in some detail how his phenomenological descriptions undermine the two sides of the debate. First of all, contra the theory-theory approach, Wittgenstein denies that we generally need to infer that another is in a particular mental state. When another bursts into a fit of rage, the anger is there, alive in the frowns and flushing of his or her face, and there is nothing I need to infer. Confronted with the phenomenological facts, it would be ridiculous to insist that “anger is something inner” and hence unobservable.81 And besides, the inferences are simply postulated; in a case such as the one under consideration, we make no inferences.82 Although there are many cases in which we do infer that someone is angry (from her glances, the sudden tremor in her voice, etc.), the case in which a person is exploding with rage is not among them. And it must be emphasized that the fact that there are (many) cases in which we do infer the obtaining of particular mental states is of no help to the theory-theorist. The latter’s claim, after all, was that, since mental states are completely unobservable, we can only ever infer them. This is clearly false. But in fact Wittgenstein’s opposition to the theory-theory outlook runs much deeper than these points suggest. For, in at least partial agreement with the simulationist critique, Wittgenstein would claim that the quasi-scientific framework theory-theory imposes on social cognition entirely misconstrues the fundamental nature of our involvement with other people. Theorytheorists claim that we ascribe unobservable pains, desires, and so on, to others in an attempt to explain and predict their overt behavior. Our involvement with others, on this sort of view, is thus thoroughly theoretical, observational, and detached. Quite like a scientist in her laboratory, we are basically aimed at acquiring knowledge about our environment, including the bits of it that move about in particular characteristic ways. To begin to appreciate how grotesque this view really is, consider these remarks: It is a help here to remember that it is a primitive reaction to tend, to treat, the part that hurts when someone else is in pain [ … ]. (Z §540) But what is the word “primitive” meant to say here? Presumably that this sort of behaviour is pre-linguistic: that a language-game is based on it, that it is the prototype of a way of thinking and not the result of thought. (Z §541) The game doesn’t begin with doubting whether someone has a toothache, because that doesn’t—as it were—fit the game’s biological function in our life. In its most primitive form it is a reaction to somebody’s cries and gestures, a reaction of sympathy or something of the sort. We comfort him, try to help him. (CEIA, p. 381)

80

Søren Overgaard and Dan Zahavi

Contra the intellectualist framework of theory-theory, others are not primarily objects of observation and theorizing. Rather, they are expressive beings that concern us. I do not assume or postulate that others are minded, because I think this helps explain and predict their behavior. Another’s cries and gestures are not behavioral data that I seek to explain by reference to postulated unobservable entities. Rather, they are expressive phenomena that call for primitive reactions of sympathy and the like. “I react immediately to someone else’s behaviour,” for it is immediately meaningful to me, and it immediately concerns me, as the expressive behavior of a human being.83 As Wittgenstein famously says in the Investigations, “My attitude towards [another] is an attitude towards a soul. I am not of the opinion that he has a soul.”84 The attitudes and primitive reactions of sympathy or empathy that Wittgenstein is concerned with in these passages come before opinions, theories, hypotheses, and the rest.85 Both temporally and systematically, emotional attitudes and responses are surely prior to inferences to best explanations. Before we are theoretical observers of others, we are empathetically involved with them. Although Wittgenstein agrees with simulation theory in part of its criticism of theory-theory, he has some rather decisive objections to the former as well. In its pure form the simulation theory holds that I grasp the state of mind of another by imagining myself in his or her place, and considering how I would feel in that situation. But this again gets the phenomenological data wrong. Do I really need to imagine myself in the “mental shoes” of another who is writhing in pain after a terrible accident? Do I need to project myself into the situation of a person who explodes with rage—and consider how I would feel under those circumstances—to grasp that he or she is angry? The simulationist thinks so; but only because she agrees with the theory-theorist that the pain and anger cannot be perceived by me. But this is contradicted by Wittgenstein’s phenomenological findings, as we have seen already. Nor can we lend any phenomenological credibility to the claim that, in general, I do consider my own feelings in some counterfactual scenario, whenever I grasp the feelings of another. Consciousness in another’s face. Look into someone else’s face, and see the consciousness in it, and a particular shade of consciousness. You see on it, in it, joy, indifference, interest, excitement, torpor, and so on. The light in other people’s faces. Do you look into yourself in order to recognize the fury in his face? It is there as clearly as in your own breast. (And what does one want to say? That someone else’s face stimulates me to imitate it, and so that I feel small movements and muscular tensions on my own part, and mean the sum of these? Nonsense! Nonsense,—for you are making suppositions instead of just describing. If your head is haunted by explanations here, you will neglect to bear in mind the facts which are most important.) (RPP I, §927)86

Understanding (other) minds

81

This long passage effectively exposes the simulationist claims for what they are: explanatory constructions without any phenomenological foundation. Once we grasp that mental phenomena such as joy, interest, excitement, and so on, can be expressed in a person’s face, we realize that it can only be in special cases that the detour over our own feelings, thoughts, and sensations, is needed. And when we are no longer under the illusion that some such detour must be involved, we should be able to accept the phenomenological fact that we don’t (usually) look into ourselves in order to grasp the fury in another’s face. According to Wittgenstein, we should not conceive of inter-subjective understanding as a two-stage process, in which we first perceive a material object, and then subsequently attribute psychological meaning to it. Such a view presents us with a distorted picture, not only of behavior but also of the mind. It is no coincidence that we use psychological terms to describe behavior and that we would be hard-pressed to describe the latter in terms of bare movements. Thus, Wittgenstein’s suggestion is that we should instead simply restore the concept of a human being to its proper place, namely “as a seamless whole of whose unity we ought not to have allowed ourselves to lose sight in the first place.”87 More generally, Wittgenstein would charge both theory-theory and simulation theory with underestimating the extent to which social cognition is embedded in a shared form of life.88 Both theoretical outlooks represent the problem as one of trying to access other persons’ minds, whereas in reality we already share a world with them. This point is connected with the primitive reactions and attitudes mentioned above. It is against the backdrop of shared life-patterns in a common world that we understand the expressions of others;89 it is not that we have to penetrate somehow into their interior world before our world can acquire inter-subjective significance. As Marie McGinn eloquently puts it, The fact of the existence of others is something that is woven into my habitual ways of acting and responding to the world, in a way that makes it part of the structure of my form of life; the fact that I am one among others does not enter my life as a discovery or as an item of knowledge [ … ], but constitutes the form of the world I inhabit.90 Wittgenstein’s phenomenological account of our understanding of other minds thus attacks fundamental assumptions in current debates in the philosophy of mind and social cognition. When philosophers and cognitive scientists think about the mind and about our understanding of each other’s minds, we too often jump to explaining, without paying sufficient attention to the mental phenomena. We assume that the mental phenomena of others are hidden deep inside their skulls. All we are left with are shaky behavioral clues that must be supplemented, either by a theory of mind or by acts of simulation, if we are ever to get to know about the minds of others. Wittgenstein

82

Søren Overgaard and Dan Zahavi

reminds us of the mental phenomena. He directs us to the difficult task of accepting them for what they are. And he exposes, perhaps better than any of the continental phenomenologists, speculative, explanatory constructions for what they are.

3.5 Concluding remarks As should be clear by now, we fully agree with a recent commentator that Wittgenstein “describes whatever is necessary to clarify the nature of that which we do not understand and only to the extent required to engender clarity. What unites his descriptions is that they serve this end, not that they all have some common object [such as language-in-use].”91 Wittgenstein not only describes language-in-use as well as extra-linguistic behavior, and ultimately a whole life form—all tasks that may seem primarily to call for a third-person approach; he also directs his descriptive focus on things as they appear to us in our everyday experiences, from our first-person points of view. In particular, he offers a phenomenological account of social cognition, which, as we have argued, is among his important contributions to philosophy.

Notes 1 RPP I, §509. 2 P. F. Strawson, “Analysis, Science, and Metaphysics,” in The Linguistic Turn: Recent Essays in Philosophical Method, ed. Richard Rorty (Chicago: University of Chicago Press, 1967), p. 320. 3 PI §109. 4 PI §90. 5 PI §383. 6 Advocates of the Wittgensteinian or Post-war Oxonian approach to philosophy therefore speak of a decline of analytical philosophy, for which Quine in particular is held responsible, cf. P. M. S. Hacker, Wittgenstein’s Place in TwentiethCentury Analytic Philosophy (Oxford: Blackwell, 1997), ch. 8. 7 M. O’C. Drury, “Conversations with Wittgenstein,” in Recollections of Wittgenstein, ed. Rush Rhees (Oxford: Oxford University Press, 1984), p. 116. 8 The classic statement of this view is found in Herbert Spiegelberg, “The Puzzle of Wittgenstein’s Phänomenologie (1929–?),” in The Context of the Phenomenological Movement (The Hague: Nijhoff, 1981). But the literature offers a veritable plethora of views. Some claim that Wittgenstein was a phenomenologist from beginning to end, in fact a purer or truer phenomenologist than those philosophers whom we normally refer to as phenomenologists, Byong-Chul Park, Phenomenological Aspects of Wittgenstein’s Philosophy (Dordrecht; London: Kluwer Academic, 1998). Others argue that Wittgenstein was a phenomenologist from the late twenties onwards, Nicholas F. Gier, Wittgenstein and Phenomenology: A Comparative Study of the Later Wittgenstein, Husserl, Heidegger, and MerleauPonty, SUNY series in philosophy (Albany, NY: SUNY Press, 1981). Others again, that he was a phenomenologist (in quite the Husserlian sense) from the Tractatus period until the late twenties: Merrill Hintikka and Jaakko Hintikka, Investigating Wittgenstein (Oxford: Blackwell, 1986). And finally, there are of course those who claim that Wittgenstein was never a phenomenologist in the

Understanding (other) minds

9 10 11 12 13 14 15 16 17 18

19

20 21 22 23 24 25 26 27

28 29 30

31 32

83

continental sense, e.g., H. P. Reeder, “Wittgenstein Never Was a Phenomenologist,” Journal of the British Society for Phenomenology 20 (1989): 257–76. Hintikka and Hintikka, Investigating Wittgenstein, p. 148. Ibid., p. 151. Monk claims Wittgenstein embraced a kind of phenomenology “only for about eight months, between February and October 1929,” Ray Monk, “The Temptations of Phenomenology,” (unpublished), p. 17. Ibid., p. 14. Ibid., p. 4. Ibid., p. 2. Ibid., p. 2. RLF, p. 35. RLF, p. 35. For a comprehensive, well-argued account of the emergence of Wittgenstein’s phenomenology in the transition from logical atomism to his later thought, see David G. Stern, Wittgenstein on Mind and Language (New York; Oxford: Oxford University Press, 1995), chs. 4–6. See also R. A. Noë, “Wittgenstein, Phenomenology, and What it Makes Sense to Say,” Philosophy and Phenomenological Research 54 (1994): 1–42. Strangely, the Hintikkas develop their account of Wittgenstein’s transition which they take to have such decisive consequences for Husserlian phenomenology, without providing a single reference to Husserl’s works: see Hintikka and Hintikka, Investigating Wittgenstein, pp. 137–59. Cf. PR, pp. 88, 98–99, 103, 283, 286. PR, pp. 51, 88. Edmund Husserl, Ideas Pertaining to a Pure Phenomenology and to a Phenomenological Philosophy, trans. F. Kersten (Dordrecht; London: Kluwer, 1982), p. 44. Contra Gier, Wittgenstein and Phenomenology, pp. 105–6. Husserl, Ideas Pertaining to a Pure Phenomenology and to a Phenomenological Philosophy, p. 51. Ibid., p. 216. PR, p. 101. But note that this may not be the right way to read it. Wittgenstein, after all, is only interested in the logical form of visual phenomena; the visual space in question is hence a logical space. This leads us back to the first reason for hesitating to assume close proximity between Wittgenstein and continental phenomenology. In general, one may perhaps say that to the extent Wittgenstein’s phenomenological remarks are read against their proper background, it becomes difficult to interpret them as addressing the same issues as continental phenomenologists address. If they are detached from their context, however, and read straightforwardly as descriptions of the phenomena of experience, they can easily be compared with the accounts given by phenomenologists. But then they are generally flatly contradicted by the latter. Maurice Merleau-Ponty, The Primacy of Perception and Other Essays on Phenomenological Psychology, the Philosophy of Art, History, and Politics (Evanston, IL: Northwestern University Press, 1964), pp. 14–15. Cf. PI §66. Note, for example, the unusually sharp terms in which Hilmy dismisses Nicholas Gier’s interpretation, e.g., S. Stephen Hilmy, The Later Wittgenstein: The Emergence of a New Philosophical Method (Oxford: Basil Blackwell, 1987), pp. 180–81, 246, 295. RC II, §3; I, §53. P. M. S. Hacker, Insight and Illusion: Themes in the Philosophy of Wittgenstein (Oxford: Clarendon, 1986), p. 178.

84 33 34 35 36 37

38 39

40 41 42 43 44 45 46 47 48 49 50 51 52 53 54

55 56

57 58 59

Søren Overgaard and Dan Zahavi Ibid., p. 176. PI §109. PI §66. Marie McGinn, Wittgenstein and the Philosophical Investigations (London: Routledge, 1997), p. 14. LW II, p. 44; cf. RPP I, §1011. This might seem to suggest a commitment to the type of approach Dennett has later dubbed “heterophenomenology,” Daniel C. Dennett, Consciousness Explained (London: Penguin, 1993), pp. 72–81 and passim. Yet, Wittgenstein is not all that hostile to a more traditional phenomenological approach, as we will argue in the following. Cf. PI II, xi, p. 204/174;; LW I, §459. Ibid., p. 44. Not only analytical philosophers such as Dennett make this mistake, however. In his attempt to show that Wittgenstein was never a phenomenologist, Harry Reeder lists the introspective method as one of the fundamental principles of phenomenology, in “Wittgenstein Never Was a Phenomenologist,” p. 258. This and many other common misinterpretations of Husserl are discussed in Dan Zahavi, Husserl’s Phenomenology (Stanford, CA: Stanford University Press, 2003). Edmund Husserl, Aufsätze und Vorträge, 1911–1921, vol. 25, Husserliana (Dordrecht: Nijhoff, 1986), p. 36. Edmund Husserl, Ideen zu einer reinen Phänomenologie und Phänomenologischen Philosophie, vol. 5, Husserliana (The Hague: Nijhoff, 1952), p. 38. Martin Heidegger, Being and Time, trans. Joan Stambaugh (Albany, NY: SUNY Press, 1996), pp. 42–47. Martin Heidegger, Grundprobleme der Phänomenologie (1919/20) (Frankfurt am Main: Klostermann, 1993), pp. 15–16. Edmund Husserl, Logical Investigations, trans. J. N. Findlay (London: Routledge, 2001), pp. 281–82, 304. Maurice Merleau-Ponty, Phenomenology of Perception (London: Routledge, 1962), p. 407. Heidegger, Being and Time, p. 58. Gregory McCulloch, The Mind and Its World (London: Routledge, 1995), p. 131. Cf. Heidegger, Being and Time, pp. 384–85. Merleau-Ponty, Phenomenology of Perception, p. 60. Heidegger, Being and Time, pp. 32, 117; Edmund Husserl, Cartesianische Meditationen und Pariser Vorträge, vol. 1, Husserliana (The Hague: Nijhoff, 1950), p. 210. Cf. Wilfrid Sellars, “Empiricism and the Philosophy of Mind,” in Science, Perception and Reality (London: Routledge & Kegan Paul, 1963), p. 173. Cf. PI §§89–90. Saul A. Kripke, Wittgenstein on Rules and Private Language: An Elementary Exposition (Oxford: Blackwell, 1982), p. 48. Elisabeth Rigal brought the contrast between the account of the Philosophical Remarks and Wittgenstein’s later account to our attention: see “The Duality of Wittgenstein’s Phenomenological Actuality,” in Wittgenstein and Contemporary Theories of Language, ed. P. Henry and A. Utaker, Working Papers from the Wittgenstein Archives at the University of Bergen (Bergen: WAB, 1992). RPP I, §420. Damasio claims that an emotion is a collection of changes in bodily states, and a feeling is a process of monitoring such bodily states, Antonio R. Damasio, Descartes’ Error: Emotion, Reason and the Human Brain (London: Papermac, 1996), ch. 7. Cf. LW I, §§386–87, LW II, pp. 4–5. RPP I, §§785–86. Cf. PI II, viii, p. 185/158.

Understanding (other) minds

85

60 E.g., PI II, xi, pp. 193–229/165–94. 61 E.g., RPP I, §§1022, 1034; LW I, §§510–11, 515. 62 PI II, xi, p. 193/165. Stephen Mulhall seems to overlook these limitations of a strict grammatical method (one that does not appeal to phenomena, but only to the use of words) in his efforts to show that Wittgenstein defends the notion of continuous aspect seeing. Mulhall might be right that seeing-as is purely a question of grammar, that is, of our concepts informing our experience, but there is no way to make such a point without appealing massively to the way things present themselves to us in our experience (which, of course, Mulhall also does). Cf. On Being in the World: Wittgenstein and Heidegger on Seeing Aspects (London: Routledge, 1990), pp. 137–55. 63 PI §284. 64 PI §284. 65 Cf. Z §225. 66 Cf. RPP I, §1070. 67 Cf. LW II, p. 62. 68 E.g., LW I, §947; LW II, pp. 32, 67. 69 Cf. LW II, p. 92. 70 Some ramifications of Wittgenstein’s expressive account of the mind for the philosophical problem of other minds are elaborated in S. Overgaard, Wittgenstein and Other Minds (New York: Routledge, 2007), especially ch. 7. 71 PI §281. 72 PI §§286, 573. 73 PI §285. 74 LW I, §978. 75 Gregory McCulloch, The Life of the Mind: An Essay on Phenomenological Externalism (London: Routledge, 2003), p. 13. 76 John R. Searle, Intentionality: An Essay in the Philosophy of Mind (Cambridge: Cambridge University Press, 1983), p. 230. 77 For an extensive discussion, see Dan Zahavi, “Expression and Empathy,” in Folk Psychology Re-assessed, eds. D. Hutto and M. Ratcliffe (Dordrecht: Springer, 2007). 78 Cf. Alison Gopnik, “Theories and Modules: Creation Myths, Developmental Realities, and Neurath’s Boat,” in Theories of Theories of Mind, ed. Peter Carruthers and Peter K. Smith (Cambridge: Cambridge University Press, 1996); A. M. Leslie, “Children’s Understanding of the Mental World,” in The Oxford Companion to the Mind, ed. R. L. Gregory and O. L. Zangwill (Oxford: Oxford University Press, 1987). Related views have been put forward by philosophers not usually thought of as participants to the theory of mind debate. Cf. Paul M. Churchland, Matter and Consciousness: A Contemporary Introduction to the Philosophy of Mind (Cambridge, MA: MIT Press, 1984), ch. 4; Daniel C. Dennett, “Intentional Systems,” The Journal of Philosophy 68 (1971): 87–106. An interesting point that the writings of these philosophers illustrate is the following: if attributing mental states to others (and indeed to oneself) is really a matter of inferring to best explanation, then rationality would dictate abandoning this type of explanation if and when a theoretical framework is available, which is able to yield better explanations and predictions of behaviour than are available within the theory-of-mind framework. For then the mentalist explanation would only be second best. 79 Cf. A. I. Goldman, “Folk Psychology and Mental Concepts,” Protosociology 14 (2000): 4–25. 80 Leslie, “Children’s Understanding of the Mental World,” p. 139. 81 LW II, p. 85. 82 Z §537; RPP II, §570, quoted above. 83 LW II, p. 84.

86

Søren Overgaard and Dan Zahavi

84 PI II, iv, p. 178/152. 85 LW II, p. 38. Wittgenstein’s emphasis on attitudes and primitive reactions has some critical potential vis-à-vis traditional skepticism concerning other minds. See Michel ter Hark, Beyond the Inner and the Outer: Wittgenstein’s Philosophy of Psychology (Dordrecht: Kluwer, 1990), ch. 5. 86 Cf. LW I, §769. 87 John McDowell, “Criteria, Defeasibility and Knowledge,” in Meaning, Knowledge, and Reality (Cambridge, MA: Harvard University Press, 1998), p. 384. 88 PI II, i, p. 174/148; xi, p. 226/192. 89 LW I, §§406, 861–62. 90 Marie McGinn, “The Real Problem of Others: Cavell, Merleau-Ponty and Wittgenstein on Scepticism about Other Minds,” European Journal of Philosophy 6 (1998): 54. 91 Daniel D. Hutto, Wittgenstein and the End of Philosophy: Neither Theory Nor Therapy (Basingstoke: Palgrave Macmillan, 2003), p. 119. Unfortunately, Hutto seems to think that what Wittgenstein describes, then, is our practices, i.e., what we do as well as what we say, cf. pp. 125–26. Like most other commentators, Hutto thus overlooks the fact that Wittgenstein also describes what we experience.

4

Wittgenstein, the secondary use of words, and child psychology Michel ter Hark

4.1 “For me the vowel e is yellow,” Wittgenstein writes in the second part of his Philosophical Investigations.1 The example is rather like Rimbaud’s vowel symbolism: “A noir, E blanc, I rouge, U vert, O bleu; voyelles.”2 On the same page Wittgenstein asks whether you would be inclined to say that Wednesday was “fat” and Tuesday “lean,” or the other way round, and says he is definitely inclined towards the former.3 Such uses of words, he says, might be described as “secondary” relative to their ordinary familiar uses. In both volumes of his Remarks on the Philosophy of Psychology as well the first volume of his Last Writings on the Philosophy of Psychology, many more passages occur concerning the theme of secondary sense. Moreover, in Philosophical Investigations §282, Wittgenstein speaks of a “secondary concept” of pain. Challenged by his interlocutor that he might be making the connection between pain and pain-behavior too tight, he responds that he wants to remind us of the fact that “only of a living human being and what resembles (behaves like) a living human being” can one say that it has pain, can hear or see.4 To be sure, we also say of inanimate things that they are in pain, for instance children making believe that their dolls are in pain. “But this use of the concept of pain is a secondary one.”5 In what follows I will first attempt to explain how the issue of the secondary use of words is related to the main theme of Wittgenstein’s philosophy of language: the idea that the meaning of a word is its use in language. Another question is, how is secondary use related to the more common notion of metaphor? I will conclude this first part by applying Wittgenstein’s discussion of perceptual “metaphors” to contemporary scientific investigations of synesthesia. In the second part of the essay, I will turn to the concept of pretend-play, raised by him so casually in §282. Seeing this remark in the light of the preceding analysis of secondary use will enable me to apply Wittgenstein’s insights to recent theories in developmental psychology about make-believe or pretend-play.

88

Michel ter Hark

4.2 Scattered remarks in Wittgenstein’s manuscripts in the 1930s, culminating in the manuscripts underlying his late published work on the philosophy of psychology,6 show him deeply concerned with a variety of phenomena which he summarizes as “experience of meaning.” These phenomena range from the feelings of the word “if,” proper names “fitting” their bearers to colored vowels and embodied weekdays. Given Wittgenstein’s insistence throughout the first part of his Philosophical Investigations on the publicity and communicability of meaning, this interest in psychological dimensions of meaning ought to be surprising. The meaning of a word is its use in the language in much the way the meaning of a hammer is its use in driving nails into timber, wood, etc. Comparing the former with the latter should disabuse philosophers from thinking of meaning as a locatable entity, and a fortiori, as located in the mind or brain. But how can Wittgenstein account for experiencing meaning along these lines? In one of his manuscripts, discussing the tone of word, he seems to come back to this restriction in Philosophical Investigations §43—viz. that the meaning of a word is its use in the language—when saying: “And if one wants to speak of meaning here, it does not consist in the use of words;”7 “The meaning of a word, I said, is its use. But an important addition has to be made to this.”8 In Philosophical Investigations Wittgenstein first discusses the secondary use of the expression “the meaning of” and “to mean it” in a language-game in which we ask someone to utter the word “March” and to mean it on one occasion as an imperative and on another as the name of the month.9 He also says that the name “Schubert” has a special kind of feel for him, as if the name fits its bearer, his music, his portrait, and even his handwriting.10 Elsewhere he asks us to say “Bank” and to mean first the river bank then the money bank. Here the experience of meaning would take place only if the inflection of voice seems to determine whether we mean one thing or another.11 The experience of meaning can also show itself negatively, for instance when we utter a word repeatedly in rapid succession it seems to lose all its meaning. Finally, one can become acutely aware of the experience of meaning failing to occur: Suppose I had agreed on a code with someone; “tower” means bank. I tell him “Now go to the tower”—he understand me and acts accordingly, but he feels the word “tower” to be strange in this use, it has not yet “taken on” the meaning. (PI II, xi, p. 214/182) Historically Wittgenstein may have been stimulated to write about meaning experience by his intense reading of William James. Consider for instance this passage of James: If we look at an isolated printed word and repeat it long enough, it ends by assuming an entirely unnatural aspect. Let the reader try this with

Secondary use of words and child psychology

89

any word on this page. He will soon begin to wonder if it can possibly be the word he has been using all his life with that meaning. It stares at him from the paper like a glass eye with no speculation in it. Its body is indeed there, but its soul is fled.12 For the exercise of goal-directed linguistic activities, meaning experiences may not seem that significant. Yet Wittgenstein’s focus on words isolated from the sentences in which they normally occur, when read or spoken in various forms of communication, may have a heuristic purpose only. For he says that familiar words have a physiognomy, that they are an actual likeness of their meaning.13 Words not only have a use but, as one might say, this use has impressed itself upon one, much in the way in which human faces impress themselves. That is why we feel at home in the use of our words, are attached to them and would not be willing to replace them by arbitrary alternative signs. Thus meaning experience is important after all. But how does it show itself ? “By the way we choose and value words.”14 What is puzzling about meaning experience is why we speak of experiencing the meaning or “meaning it.” What does the concept of experience have to do with the concept of meaning? Meaning as determined by linguistic rules is not something that can be experienced.” It is, however, precisely this question which puts us on the wrong track: This is a different kind of question.—It is the phenomenon which is characteristic of this language-game that in this situation we use this expression: we say we pronounced the word with this meaning and take this expression over from that other language-game. (PI II, xi, p. 216/184) Wittgenstein rejects the why-question as misleading and it is clear that he thinks both causal questions and linguistic questions pertain to ambiguity and metaphor. Let me first focus on the causal question. Wittgenstein does not deny that interesting causal questions about experiencing meaning can be raised, but his point is that they leave “the oddity untouched.”15 It is important to explore in more detail why this is so, because this may reveal more about the kind of phenomenon experiencing meaning is, and hence what secondary sense is. To show why causal explanations are irrelevant here, Wittgenstein asks us to consider first a related phenomenon. On walking in the environs of a city I am under the impression that: [T]he city must lie over there, even though a moment’s reflection can convince me that it is not so. Now I shall assume that I know how the illusion came about: The similarities between this landscape and another led me to draw the wrong conclusion, etc.—But I hadn’t explicitly drawn

90

Michel ter Hark the wrong conclusion, and furthermore this doesn’t explain why these similarities led me to draw this precipitous conclusion. (LW I, §80)

Wittgenstein speaks here of a “special kind of illusion” suggesting a difference with illusions in the normal sense of the word. Standard illusions arise because experience distorts what there is. On the same walk the green hills in the distance may look blue. The apparent properties of hills vary relative to the position of the percipient, the distance and media between him and the object, the lighting, the state of his health, etc. In such a case one has reasons for saying that the hills are blue, because they appear blue. The illusion typically persists as long as one finds oneself at a certain distance from the hills. But one has no such epistemic grounds for saying that one is under the impression that the city lies over there. It is not as if one saw the contours of a city loom up in the distance. Nothing has changed in one’s perceptual situation. One is not deceived by appearances as when a grey haze makes the green field look grey. Accordingly, one does not take the city to lie to the right and no revision of one’s beliefs by adding new evidence is needed to see that it is not the case. The illusion is typical, one might say, because one knows that one is under an illusion. In that sense it is like the experience of knowing that one dreams. This is why Wittgenstein says that all one has to do in order to know that the city lies to the left “is to wake up, as it were.”16 How literal is this “optical illusion” in the case of meaning experience? Wittgenstein reminds us of the fact that ambiguous words prompt the experience of meaning only after one utters the word, e.g. “bank,” in a specific way. Hence, we determine whether a word means this rather than that, the object in the park or the financial institution, by a specific tone of voice. Normally the meaning of a word is fixed by (ostensive) definitions and explanations all of which presuppose processes and activities spread over time. But no such procedures take place when experiencing the meaning of “bank” by uttering it in a particular way. Yet it seems as if saying “bank,” isolated from its normal context of use, co-ordinates the object (e.g., the object in the park) to the word. We know very well that by merely saying an ambiguous word in a specific way we cannot make it mean one object rather than another—only the context of use can accomplish this—yet it seems so. Hence, we know that we are experiencing an illusion. The similarities and associations involved in causal explanations of such phenomena are not one’s reason for drawing certain conclusions and feeling forced to use certain words. An autobiographical example illustrating this point is Wittgenstein’s speaking of embodied weekdays: “[ … ] given the two ideas ‘fat’ and ‘lean’, would you be rather inclined to say that Wednesday was fat and Tuesday lean, or the other way round? (I am definitely inclined towards the former).”17 He concedes that this way of speaking might be explained in terms of associations from childhood, but he adds that that is a hypothesis and that

Secondary use of words and child psychology

91

whatever the explanation, the inclination is there. From his manuscripts it becomes clear that he hypothesized that this specific association was due to two female teachers that visited him alternately during the week when he was a child.18 The point he wants to make here is that he had the inclination to speak of “fat Wednesday” long before he discovered the possible explanation of this way of speaking. In rejecting a linguistic why-question Wittgenstein wants to draw attention to the fact that the concept of meaning involved when speaking of meaning experience and secondary use is radically different from the use of the word “meaning” in normal explanations of the meaning of words, e.g. synonymy, ambiguity, metaphor, etc. The very question how it is possible that we use the same word, e.g. “meaning,” on both occasions might tempt us to explain meaning experience and secondary use in terms of ambiguity or metaphor. But then the following questions would have to admit a positive answer: “Can he use a different one? Does he use the same word for something different? Could he give another explanation of it?”19 Were we using the word to mean something different from what we mean by it in ordinary explanations of meaning we would be dealing with word ambiguity. The meaning of the word “meaning” would then be justified by saying that it has two meanings, as much as the word “bank” can mean either a river bank or a money bank. But for that to be the case one should have learned the word “meaning” in two different ways as indeed one learnt “bank” in two different ways. However, just as one has learnt that proper names not only refer to persons but also in addition to the properties of their bearers, one has learnt that the concept of meaning can also refer to a specific experience prompted by uttering a word with special emphasis. The same point is also argued for in another way. Reflecting upon his inclination to call Wednesday “fat” and Tuesday “lean”, he asks whether “fat” and “lean” have some different meaning here from their usual one. He answers that they have a different use but explicitly rejects the suggestion that he should have used different words instead. The characteristic feature of the experience is that he wants to use “these words (with their familiar meanings) here.”20 The (familiar) meaning of the word has not changed. This is why Wittgenstein notes: “I could only explain the meanings in the usual way. I could not point to the examples of Tuesday and Wednesday.”21 A few passages later he explicitly rejects the idea that secondary use is metaphorical. It is noteworthy that his example concerns what is today called “color–vowel synesthesia”: If I say “For me the vowel e is yellow” I do not mean: “yellow” in a metaphorical sense,—for I could not express what I want to say in any other way than by means of the idea “yellow”. (PI II, xi, p. 216/184) The examples of embodied weekdays and colored vowels may be extravagant but as an earlier discussion in the Brown Book shows, they can be supplemented

92

Michel ter Hark

with examples that are so mundane and have become so stereotyped as to be hardly perceived as secondary (or metaphorical) at all. Wittgenstein opens his discussion with the expression “mental strain”: A man holds a weight with out-stretched arm; his arm, his whole body is in a state of tension. We let him put down the weight, the tension relaxes. A man runs, then rests. He thinks hard about the solution of a problem in Euclid, then finds it, and relaxes. He tries to remember a name, and relaxes on finding it. (BB, p. 129) Right after the example of embodied weekdays he says: “We speak of a deep sorrow, a deep sound, a deep well.”22 And, one might add, a “deep sleep” or the “deep realms of the imagination.” Again, in other places, he gives the example of “a bitter food” and “a bitter grief.”23 In the Brown Book, Wittgenstein does not yet speak of secondary use, and even uses the term “metaphor” for the preceding examples. But this seems merely to be a matter of terminology, because he rejects an explanation of such word-uses typical of genuine metaphors. Asking the why-question in these cases easily leads one to suppose what Wittgenstein believes to be a grammatical prejudice: “Surely a similarity must strike us, or we shouldn’t be moved to use the same word.”24 Understanding metaphors typically depends on the apprehension of likenesses between the “primary” and the “secondary” subjects of a metaphorical expression. But what do a deep well, a deep sorrow and a deep sleep have in common, Wittgenstein asks rhetorically? And in which respect can the vowel “e” be said to be similar to the color yellow? According to Tilghman, these unanswerable questions show that Wittgenstein’s point is that secondary uses cannot be paraphrased.25 But this is not the only point he makes. The relationship between the primary use and the secondary use of a word, Wittgenstein says in his later writings, is “not like the one between ‘cutting off a piece of thread’ and ‘cutting off someone’s speech’, for here one doesn’t have to use the figurative expression.”26 Another distinctive feature of secondary use is that we are somehow forced to use this word here, are attached to it, and that no other word will do. What then is the relationship between the word’s use in its home and in its transferred use? In both cases the explanation of the word is that of its primary meaning. It can only have a secondary meaning for someone if he knows its primary meaning. That is, the secondary use consists in applying the word with this primary use in new surroundings. (LW I, §797)27 The relation of dependency between the secondary use and the primary use of words is logical rather than contingent as in the case of metaphors that can be paraphrased. If the secondary use of a word consists in applying a

Secondary use of words and child psychology

93

word with a familiar use in new surroundings then it is logically impossible for one to command the former without commanding the latter. Only a person who has learnt to apply a word in a primary use is able to use and understand it with a secondary use. The logical dependency of the secondary use upon the primary use therefore pertains to learning and it would be a misconception to construe the relationship between the primary and secondary use as merely a relation between the literal and metaphorical meanings of a word.28 The very feature of the secondary use of words, e.g. “meaning,” is that a person is inclined to use a word, for which there is a specific learning history, in a situation where one has not learnt to use it according to that history. One is inclined or even forced to use the word “meaning” which one has learnt for other purposes. The word “meaning” has been learnt while giving explanations and definitions of words, translating, crossword puzzles, etc. Now, when a person speaks of the experience of meaning he makes use of the familiar concept of meaning in order to express a particular experience. As Wittgenstein puts it elsewhere: “we employ the expression that we have learnt for a technique in an expression of experience.”29 This remark may be connected with another one: The feeling of the unreality of one’s surroundings. This feeling I have had once, and many have it before the onset of mental illness. Everything seems somehow not real; but not as if one saw things unclear or blurred; everything looks quite as usual. [ … ] But why do I choose precisely the word “unreality” to express it? Surely not because of its sound. (A word of very like sound but different meaning would not do.) I choose it because of its meaning. But I surely did not learn to use the word to mean: a feeling. No; but I learned how to use it with a particular meaning and now I use it spontaneously like this [ … ]. (RPP I, §125) A literary example may illustrate the same point. In John Banville’s novel The Sea, the art historian Max Morden returns to the seaside village where he once spent a childhood holiday: How flat sounds are at the seaside, flat and yet emphatic, like the sounds of gunshots heard at a distance. It must be the muffling effect of so much sand. Although I cannot say when I have had occasion to hear a gun or guns being fired.30 “Flat” in “flat sounds” is explained not in terms of a literal or quasi-literal paraphrase but in terms of an expression that involves a shift from the primary use of “flat”: the sound of guns being fired. “Flat sounds” therefore is a secondary use of words. Max is impelled to express his experience by taking this word from another language-game. A question that still needs answering is what Wittgenstein calls the “new use” or the “new surroundings”

94

Michel ter Hark

in which a word is used secondarily. This point especially has escaped notice in the literature on Wittgenstein. If secondary use is often a matter of expressing feelings and particular experiences, how is it related to these feelings and experiences? The following passage contains an answer to this question: [ … ] Why do we call this “meaning” and “signifying” if it is not a question of meaning and signifying?—What do I call “meaning” (or “signifying”) in this game: I say “By that word I just now meant … ”. But what am I calling that?—an experience? And what experience? For can I describe it otherwise than just by the expression: I “mean” this word in this way? (LW I, §63) What we call “meaning’ in the language-game of meaning experience is internally related to the verbal expression, “By that word I meant … .” Similarly, what we call the experience of flatness when listening to the sounds of the sea is internally related to saying or pronouncing: “How flat sounds are at the sea”. These verbal expressions belong to the experience of meaning as much as what Wittgenstein earlier called the primitive expression of pain belongs to the experience of pain. This means that the concept of experience here is different from straightforward sense experiences like pain or seeing shapes. Wittgenstein therefore speaks of a “modified” concept of experience.31 That is to say, the experience of colored vowels, like the experience of embodied weekdays, is constituted by saying (or thinking of) the relevant words or uttering them in a specific way.32 We do not first conceive of the vowel “e” in an inchoate way and then find that “yellow” is the appropriate word for it. On the contrary, the experience of colored vowels consists of the fact that the (appropriate) word comes out. Or more clearly in the case of “fat Wednesday,” the word “fat” comes often accompanied by “gestures expressive of fatness and a certain comfort.” The implication of this internal relation between the relevant experience and the verbal and behavioral expression is that for someone to be capable of having the experience of colored vowels or embodied weekdays it is necessary that she is familiar with the primary meaning of color words and words for describing food or animals, respectively. What is the philosophical importance of Wittgenstein’s reflections on secondary use? Put differently, how are we seduced, by the very form of language, into forming a false picture of these “curious” experiences? “‘But what is this queer experience?’—Of course it is not queerer than any other; it simply differs in kind from those experiences which we regard as the most fundamental ones, our sense impressions for instance” (PI II, xi, p. 215/183). The difference in kind Wittgenstein refers to is that in the case of meaning, experiencing a metaphor becomes the expression of an experience. Unlike other metaphors, however, this metaphor is not an indirect description of an

Secondary use of words and child psychology

95

experience, one that can be stated more directly or more literally. In this case a metaphor has become the primary expression of an experience in the sense in which Wittgenstein uses the term “primary” when speaking about the expression of bodily sensations such as pain. Instead of “metaphor” one might say “secondary use.” If a philosopher or a scientist does not recognize this difference in kind, what happens? She will then assimilate secondary use either to metaphor or to primary use. If the former she would be inclined to reject this form of words as at best a circumlocution, one that could be eliminated. If the latter she would be inclined to look for special sorts of facts to which they correspond. Recent discussions in neuroscience about synesthesia reflect these mistaken patterns of reasoning. The topic of synesthesia enjoys a controversial reputation, with many scientists dismissing it as a metaphoric way of speaking— especially when behaviorism reigned in psychology and philosophy. Synesthesia, so the standard definition goes, is the tendency of experiences in one sense modality to trigger anomalous experiences in another sense modality.33 Neuroscientists have recently argued that synesthetic experiences are “genuinely sensory” and “perceptually real” rather than a linguistic phenomenon.34 Ramachandran and Hubbard tried to determine whether synesthesia is a genuine sensory phenomenon by asking subjects explicitly: “Is this just a memory, or do you actually see the color as if it were right in front of you?” Some subjects responded that they saw colors perfectly clearly, but a more frequent kind of reaction was, “I kind of see it, kind of don’t,” or, “No, it is not like a memory. I see the number as being clearly red, but I also know it isn’t; it’s black. So it must be a memory, I guess.”35 Below are some other quotes from synesthetes: When I’m reading I can see that what I’m looking at is in black and white, but I also see the “correct” colours for the letters and symbols. [ … ] Tuesday is yellow. I don’t “see” it anywhere in particular; rather, I have a general awareness of yellowness in relation to the word.36 In their search for an empirical and sensorial explanation of synesthesia Ramachandran and Hubbard dismissed these expressions as irrelevant, thereby overlooking their very special nature. What these and other reports of synesthetes confirm is that Wittgenstein’s analysis in terms of a typical sort of illusion applies here too. Synesthetic experiences seem to be involuntary, in the sense that they occur without conscious effort.37 Synesthetes seem to experience specific vowels and specific colors as a type of entity. A specific vowel does not merely correspond with a color, but is yellow, blue, or brown. To them the color yellow fits the vowel “e,” for instance, but they are not deluded in this. One cannot describe their experience by saying that they mistake the vowel “e” printed in black for yellow. Indeed, as the quotes above indicate, they know, at the moment of having their synesthetic experiences, that the vowels are printed in black. Hence, they know they are

96

Michel ter Hark

having illusions of seeing colored vowels. If this description is correct then synesthesia might be further described in terms of secondary use and meaning experience. In the light of the discussion of Wittgenstein above, the putative dichotomy regarding synesthesia—it is, either, merely a metaphor or a genuine sensory experience—turns out to be false. Synesthesia is not just a dispensable metaphor yet it is constituted by language; synesthesia is a genuine experience yet not an experience like any other. It is an experience characterized by one’s being impelled to use a word taken over from another language-game where it has a specific primary use. Hence, it is an experience internally related to a prior command and use of language. The discussion of meaning experience and secondary use above may not be complete or definitive, yet a few conclusions can be drawn. A secondary use of words—in which often meaning experience is very prominent—is a use of words in which we make use of another, familiar or primary area of language in order to express a particular experience. In the primary area we have learned to use the words for specific purposes, but in the secondary area we do not use them according to these purposes. Here we do not use the word “meaning” in the sense we have learned but we say, “For me this word means …”. We borrow the familiar use of the word for other purposes. This means that one can use a word in a secondary sense only if one has learnt to use it in a primary sense.

4.3 I now turn to Wittgenstein’s remarks on pretend-play and make-believe. Two logical features of this concept show the similarity with concepts involving meaning experience. Right after his stretch of remarks on embodied weekdays he says: “Only children who know about real trains are said to be playing trains. And the word trains in the expression ‘playing trains’ is not used figuratively, nor in a metaphorical sense.”38 The concept of pretendplay, including the activities involved in play, is a secondary one logically depending on the natural or ingenuous performance of the relevant activities and primary use of the relevant concepts. The concepts used in a secondary game are not used figuratively or metaphorically. Before proceeding I must note some obvious differences between pretence as a secondary concept and meaning experience. Unlike meaning experience, pretend-play does not imply the secondary use of words (with a primary meaning) for expressing feelings. And unlike meaning experience, the language-game of pretend-play does not have linguistic expressions internally related to feelings and experiences. Why is “playing trains” not a metaphor? If it is a secondary use of concepts referring to real trains and traffic then the child not only must possess the concept of real trains but must also specifically use it in this primary sense in her game of playing with trains. Therefore, one cannot say that when the child says she is “playing with trains” it does not really mean “trains” but only something similar to them, as it would were it a metaphor.

Secondary use of words and child psychology

97

The same point is made in relation to a quite different example which I merely mention here: If a person says he is calculating in his head, is he not really calculating, does he mean something else by calculating? You could never explain to a person what is meant by “calculating in the head” if he hadn’t already been taught the concept calculating. (LW I, §801) In the rest of this section, I will explore the relevance of Wittgenstein’s emphasis upon the importance of the command of the primary use of words and activities for pretend-play and make-believe as these are discussed in recent experimental developmental psychology. I will show that psychologists in the wake of Leslie’s seminal paper have nevertheless failed to see pretend-play and make-believe as secondary.39 As a consequence, they have sought to explain pretence in terms of mental representations whereas it needs to be explained as a secondary use of language and related activities. Leslie’s seminal paper starts with the assumption that the child’s pretence and make-believe are so deviant, even nonsensical, that it poses a serious threat to her cognitive development.40 Pretend representations are sharply contrasted with “reality-oriented” representations which are taken as paradigmatic of what evolution is all about. About the (by now) standard example, “This banana is a telephone,” he says, “such nonsense violates the basic design principle of primary representation that it represent in a literal fashion.”41 Following McCune-Nicolich, he claims that children who pretend have a “double knowledge” about the situation.42 That is to say, they are pretending that a chair is a train, but at the same time they know that it is really a chair. This double-knowledge is interpreted as involving the presence of two simultaneous mental representations: literal or primary representations and pretend representations, both referring to the same actual situation. This poses a potential threat for the child, namely that of representational abuse. How can a child who holds a primary representation of a real object, i.e. a chair, at the same time think of it as if it were a train? How is it that the child’s representational system is not undermined by this—is it a chair or is it a train? Both representations cannot be primary as they contradict each other semantically. On the language of thought hypothesis espoused by Leslie, mental representations also underlie the use of language, and hence the child’s primary use of words should equally be undermined by its use of pretend language. To account for the child’s ability to pretend-play and make-believe without contaminating its system of primary representations, Leslie argues that the pretence representations must be “quarantined off ” in some way. Pretence representations are produced by copying primary representations into a meta-representational context. This second-order context gives a report or a quotation of the first-order expression. In this way, it makes opaque the

98

Michel ter Hark

expression that was previously transparent. For example, when the child represents the world seriously or in reality-oriented play it may have the representation The cup is full of water. But in pretending that the cup is full of water its representation will have the content I pretend “the cup is full of water.” The quotation marks indicate that the expression contained in them is de-coupled from its primary context, thereby suspending its normal semantics. Meta-representations are freed from their usual meanings, from their normal input–output relations, so that one object can substitute for a different object without the child confusing actual semantic relations. It is noteworthy that no actual cases of representational abuse are cited or discussed in the recent literature. Recently, Lillard has given a fairly complete description of possible forms of abuse.43 She distinguishes three ways in which the “real–pretend boundary” could be lacking. One is that the barrier does not exist, and children assume that the pretend world and the real world are one. This she considers not likely, the reason being that if children did not have a basic awareness of the barrier, their understanding of the real world would be undermined by pretence. She concedes that such reality confusion does appear to occur with fantasies that parents want children to believe are real, like Santa Claus, but that “it does not appear to happen in other cases, when parents are not trying to dupe their children.”44 The second way that the real-pretend boundary might be diffuse is that it might be penetrable, like a “semipermeable membrane.”45 Real-world knowledge indeed seeps into pretendplay, Lillard argues, but this speaks only to the fact that it is a subset of children’s behavior, not that children do not differentiate what is real from what is pretend. Another possibility is that pretend-world elements travel in the reverse direction. For example, children “might expect that if, in the pretend world, a block was a cookie, then in real life that which is called ‘cookie’ should actually be a block. Although it has not been directly studied, the available reports suggest no such confusion.”46 Despite the absence of evidence for representational abuse, psychologists feel nevertheless forced to offer an empirical explanation of why it does (probably) not occur. Talk of a membrane is as metaphorical as talk of a decoupling mechanism, yet the idea is that such metaphors can be unpacked in more straightforward mechanistic concepts thereby emphasizing the empirical nature of the explanation. The absence of evidence for representational abuse, however, could also be a sign of the conceptual nature of the problem. Since the problems dealt with concern at least partly the (semantic) question, “what is called ‘pretence’,” conceptual issues do seem to be involved.47 If the problem is conceptual its roots are to be found in a mistaken or deficient view of the way concepts function, how they are related to one another and how they are taught. Consider now Wittgenstein saying: “Only children who know about real trains are said to be playing trains. And the word trains in the expression ‘playing trains’ is not used figuratively, nor in a metaphorical sense.”48 The first sentence says that the ability to make believe requires the

Secondary use of words and child psychology

99

command and use of the primary use of the relevant words. Whereas the second sentence says that to recognize that words may be used in a secondary sense is to see that they are not meaningless in these secondary uses, even if one cannot give an account of what they mean with words used in their primary senses—as, by contrast, can be done with figurative language. The problem of representational abuse therefore may well reflect a failure to see the significance of the fact that the understanding of words used in a primary sense is necessary for a child to be able to pretend-play and make-believe. But one must distinguish between a stronger and a weaker relation here.49 In the case of “fat” and “lean,” as in the case of “calculating,” the secondary use logically could not exist without the primary use. On the other hand, in the case of pretence the relation is weaker. For it is possible that children of a tribe who know nothing of trains pick up the game of trains from others and play it without realizing that the game is copied from something. But as Wittgenstein remarks about this case: “One might say that the game did not make the same sense to them as to us.”50 For instance, the child’s game would fail to be comparable with a standard or example, hence questions as to the quality of its play could not arise. More importantly, because “train” would not be a transferred use of language (from the primary use) its being mentioned by those children would in fact be arbitrary and a different word might be substituted. And because the word—whether “train” or some other—would not have been transferred from the primary use it could be explained, at least among those children, simply by referring to what they actually did when playing their game. This latter point is again in conflict with the dependency relation implied by secondary use. For another way in which the latter depends on the former is that its meaning is explained by reference to paradigms involved in the primary use, not the secondary use. Put otherwise, in this hypothetical case a tribal child does not mean what a non-tribal child does when he makes-believe that the chair is a train. In the normal sense of “to make-believe” the connection with the primary use of words is not broken. Paraphrasing Wittgenstein, only if a child has learned what trains are can it be made to grasp, by means of this concept, what “playing with trains” is. Skeptical questions—e.g. does the child really mean “train”?; does it not confuse “train” and “chair”?—can arise only on the assumption that the child has learned the meaning of the word “train” independently of the ability to use it with real trains. The problem of representational abuse therefore is not a problem caused by the epistemic predicament of the child’s system of mental representations, but by overlooking the fact that a certain region of language, viz. pretend language, cannot be understood by considering that use alone but only by taking into account uses of language on which it logically depends. Thus, by overlooking the role of language learning and training it may seem that a child, without having learnt the concept of “pain” might start by learning, “My doll is in pain.” But this use of language is a secondary use and hence dependent on language in which “pain”

100

Michel ter Hark

is used with its the primary use, much as “It seems to me to be red,” is dependent on reacting without hesitation to, “What color is this?” by answering, “Red.” Representational abuse then is precluded by the logic of pretend language rather than by quarantining mechanisms in the mind or brain. That pretend elements do not seep in across the “barrier,” as Lillard herself admits, is not due to the structure of a sort of mental membrane but to the practice of learning and using pretend language against the background of paradigms involved in the primary use of words. On the other hand, the fact that real-world knowledge does seep in across the “barrier,” yet without causing confusion, does not “merely reflect that knowledge from one domain is applied in another domain.”51 For the question remains why this transfer does not lead to an erosion of the barrier. That no such confusion occurs is a reflection of the fact that the transfer leads to a new, and hence, different use of language, a secondary use, to be clear. The secondary use of a word differs from its primary use; but, specifically by separating secondary use from metaphor, Wittgenstein can explain that and why no confusion of concepts can occur. In its new and secondary use the word does not acquire a new meaning, as happens with ambiguity and metaphor, but a new use. The secondary use “cannot come into conflict with the rest of the use of the word,” because its command depends on the full awareness of (i.e., is internally related to) what the word in its primary use means.52 Leslie’s mechanism is not only useless in maintaining the conditions of sense, it even undermines them. The postulated de-coupling mechanism protects the primary representation from contamination by making a copy of the primary representation with its normal semantics suspended. But Leslie’s use of the term ‘copy’ here trades on an ambiguity. On the one hand, he refers to the analogy with editing a copy of a word-processing file which implies that the copy and the original are identical in content but physically separated. But then the copy, contrary to what Leslie intends, should have the very same semantic properties as the original. On the other hand, decoupling is said to effect a mental copy, a representation of a primary representation but without the semantic properties of the latter. Conceived this way it cannot have the same meaning as the primary representation.53 Such an account of pretence makes it impossible to explain why the child—if asked what she meant by “pain” when pitying her doll—will not refer to her doll but to the primary use of this concept. Yet it is precisely in virtue of this dependency relation with the primary use that the child is inclined to treat her doll with care and to find the word “pain” appropriate here.54 Leslie’s explanation of the concept of pretence in terms of mental representations therefore undermines the conditions of sense. To see more clearly the nature of this criticism it is necessary to have a closer look at other criticisms of Leslie’s position, especially those of Lillard. Proponents of the view that pretend-play is meta-representational have drawn a functional link between pretence and theory of mind, hence a capacity for understanding mental representation.55 On the other hand, Lillard has

Secondary use of words and child psychology

101

contested this linkage before the age of five. For a child to possess the concept of pretence in the full sense of that term, she argues, at least two conditions have to be met. The child’s pretence must involve having mental representations, for: “If one knew absolutely nothing about horses, and did not even know that they existed, one could not pretend to be one [ … ].”56 In addition, the child needs to be aware of, to think about what one is doing, e.g., dealing with a pretend cookie. Without such awareness, Lillard concludes, the child would be merely mistaken, rather than pretending.57 In one of her experiments she has shown that young children do not yet command the concept of pretence in the full sense, but instead understand it as external manifestation only. Four- and five-year-olds were shown a troll and were told, for example: This is Moe, and he’s from the Land of Trolls. Moe’s hopping around, kind of like a rabbit hops. Moe doesn’t know that rabbits hop like that; he doesn’t know anything about rabbits. But he is hopping like a rabbit.” Then children were asked, “Would you say he’s pretending to be a rabbit, or he’s not pretending to be a rabbit?” Over four trials as well as several variations on this experiment, about 65 percent of 4- and 5-year-olds have consistently claimed that the character was in fact pretending to be a rabbit. “In effect,” Lillard points out, “children were asked whether mental representation or action was the more important factor to consider in judging whether a character was engaging in pretence.”58 To conclude from these experiments that children are not aware that pretence involves mental states and hence are not in command of the concept of pretence is not so much inconclusive—as some have supposed— as misleading.59 The experimental outcome seems to be in conflict with naturalistic situations in which children won’t be readily inclined to ask other children to join in doll-play when they know they have never heard of dolls and do not even know they exist. On the other hand, children in real life might pick up a game from others without realizing that their game was copied from something real. It would be legitimate in such cases to say that the children are playing a game, and also that their game looked like a game in which the playing is based on knowledge of real situations.60 Finally, had the test question been formulated in non-mental yet functionally equivalent terms the outcome might well have been different. For instance, “Moe has never heard of rabbits, yet he hops like a rabbit without having copied this behavior from real rabbits.” By using mental terms and by limiting the test questions to a simple “Is Moe pretending to be a rabbit or not?” the experiment does not test for this sort of understanding. Indeed, as has also been objected by German and Leslie, the outcome of the experiment may be due to pragmatic factors.61 In particular, children may be unwilling to leave unresolved precisely what the protagonist was supposed to be doing and why. Perhaps children simply chose the only interpretation on offer. If the

102

Michel ter Hark

outcome of the experiment appears incorrect, what about Lillard’s claim that the early grasp of pretence is behavioristic and that not until mental representations come in can the child have acquired the full concept? Her mistake is to think that the knowledge required for concept possession is having a theory of mental representations, the way they are produced, and how they issue in behavior. The command of “fictive mental states” depends on knowledge acquisition that does not involve what we would usually consider to be mental state terms. Doll-play, for instance, is a form of make-believe that requires the preceding command of concepts like caring, pity, pain, or kissing as they are applied outside the game in their home base. Similarly, when children make believe as if inanimate objects feel pain or are happy, while at the same time denying that these objects can think, this need not mean that their grasp of pretence is in terms of action only.62 The conceptual stage-setting for the child’s behavior in this case is that she supposes that the inanimate objects have a language and it is in virtue of this capacity that she attributes psychological properties to them. Indeed, as Wittgenstein notes, in fairy tales children have no trouble with the idea that the pot can see and hear things, but in fairy tales pots also speak, walk about, etc.63 Or consider the (young) child’s use of pictures. For instance, a child points to a picture in her illustrated fairy tales and says “That is an angry woman, and that is her eye.” This a secondary use of language. Children learn this use of pictures and hence the secondary use of words (“angry” as applied to a picture-face) long before they have the capacity to talk in terms of intentions, let alone mental representations. And, as every parent knows, this priority of the secondary use of words over the talk of intentions and mental representations in no way leads the child into “representational abuse,” i.e. confusing the word “eye” as applied to human faces and picture-faces. A fortiori it would be nonsense to say that the child cannot really see the figure as an angry woman because for that to be possible it must have the intention to represent the lines and dots on the picture as eyes, noses, or whole faces. But to say that the child must be engaged in making a comparison between what it literally sees and what it represents it as being is to have things upside down. Describing pictures by what they represent is not a process of comparing or construing an analogy but a secondary use of language.64 It might seem as if I am now joining German and Leslie, who recently have objected to Lillard that her assumption that concept possession depends upon knowledge is mistaken.65 If anything, they reply, a representational theory of mind is a consequence of possessing concepts like pretence and belief, not a prerequisite. Rather than depending on knowledge, concepts may exist prior to knowledge and may depend on mechanisms. Like the mechanisms of color vision that allow colors to be attended and thus allow color concepts to be grounded without prior knowledge of a theory of colors, psychological concepts may be grounded on mechanisms without prior knowledge of a theory of mind. German’s and Leslie’s position is the old abstractionism in disguise. According to this view, concepts are formed from scratch by abstraction—by

Secondary use of words and child psychology

103

discriminative attention (i.e., ostensive definition) to features given in senseexperience. His choice of color concepts is unsurprising here for they have always been the abstractionist’s favoured examples. But as Geach in the footsteps of Wittgenstein has argued, abstractionism fails to account for even so simple a concept as chromatic color.66 More importantly, sensory concepts, e.g., color concepts, do not have the privileged position in concept formation that abstractionists typically think they have. They do not have an epistemological primacy over others, for the description of sensations is a highly sophisticated use of concepts, and is secondary to the description of material objects. Indeed, a child may have the concepts of book or chair before it has any color concepts at all. Similarly, to pretend or to infer pretence on the basis of the other person’s behavior, although not requiring the preceding command of a theory of mind to be applicable, is a highly sophisticated use of concepts, and is secondary to non-pretend uses of psychological and physical object language.

4.4 What is Wittgenstein’s contribution as regards the notion of secondary use? It is not that he has contributed to a theory of metaphor as suggested by Tilghman and more recently Malcolm Budd.67 The notion is part and parcel of Wittgenstein’s general account of philosophical error. The meaning of words cannot be read off from sentences in which they occur; it can only be determined by investigating their use. The meaning of a word is (often) not the object for which it stands. What is special about his treatment of secondary use and meaning experience is that he here confronts psychological dimensions of use and that he has by no means the intention to deny or reduce them, as one would expect of a behaviorist and also of a logicist. Rather he shows that these subtle psychological dimensions too have their logic and hence are communicable. What is perhaps most striking about his discussion of secondary use is its relevance for actual discussions in such diverse areas as neuroscience and developmental psychology. Both the literature on synesthesia and scientists investigating make-believe play have fallen prey to traps set by language, at least so I have argued. In particular they have thought that because the “secondary use” of words is different from the primary use of words the former must have a different meaning in comparison with the latter. Thus, synesthetic language about colored vowels gets interpreted as referring to private sensations of color and ultimately to specific processes in circumscribed areas in the brain. The language used by the child during make-believe gets interpreted as referring to private mental representations differing from the mental representations supposedly underlying non-pretend language. The ingenuity of Wittgenstein’s response (or, better, anticipation of these developments) is to show that secondary use is a matter of a differing use of words with a prior, established primary use. For developmental psychology especially, the consequence seems to be that a

104

Michel ter Hark

long mentalistic research program should reorient itself and focus more on what it now completely ignores: the child’s mastery of language in dyadic and triadic interactions.68

Notes 1 2 3 4 5 6 7 8

9 10 11 12

13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

29

PI II, xi, p. 216/184. “Les Voyelles,” in Arthur Rimbaud, Poésies (Paris: Mercure de France, 1929). PI II, xi, p. 216/184; cf. BB p. 137. PI §281. PI §282. LW I–II, 1946–47. MS 180b, p. 6. MS 180b, p. 8. The earliest discussion of experiencing meaning occurs in the Big Typescript (BT, p. 33), where Wittgenstein mentions James’ example of the psychological meaning of the word “if,” the so-called “if-feeling.” This notion of meaning, he says, is different from meaning in the sense of grammatical rules determining the use of a word. He speaks of two completely different ways of using the word “meaning,” thereby anticipating his later discussion. I discussed this fragment in Michel ter Hark, Beyond the Inner and the Outer: Wittgenstein’s Philosophy of Psychology (Dordrecht: Kluwer, 1990), p. 161. Cf. PI II, xi, p. 215/184;. ibid. B. R. Tilghman discussing this example fails to see this; see But Is it Art?: The Value of Art and the Temptation of Theory (Oxford: Blackwell, 1984), p. 154. William James, The Principles of Psychology, 2 vols. (London: Macmillan, 1890), p. 726. But Wittgenstein was also inspired by Goethe’s discussion of proper names. On Wittgenstein’s relation to Goethe and James see ter Hark, Beyond the Inner and the Outer. On James and Wittgenstein, see also Laurence Goldstein, “What Does ‘Experiencing Meaning’ Mean?,” in The Third Wittgenstein: The Post-Investigations Works, ed. Danièle Moyal-Sharrock (Aldershot: Ashgate, 2004). Cf. PI II, xi, p. 218/186. PI II, xi, p. 218/186. LW I, §80. LW I, §76. PI II, xi, p. 216/184; cf. BB, p. 137. MS 131, p. 173. LW I, §67. PI II, xi, p. 216/184; cf. LW I, §795. PI II, xi, p. 216/184. BB, p. 137. RPP I, §68. BB, p. 130. Tilghman, But Is it Art?, pp. 162–64. LW I, §799. Cf. PI II, xi, p. 216/184. It has to be noted that Wittgenstein’s primary use does not coincide with a notion of literal meaning as this is often used in the context of discussions about metaphor. Primary use is the familiar use of a word. This may include “literal” uses but also metaphorical uses. Of course, Wittgenstein would deny that there is such a thing as the literal meaning of a word independent of the variety of uses. RPP II, §1128.

Secondary use of words and child psychology

105

30 John Banville, The Sea (New York: Knopf, 2005). 31 Cf. PI II, xi, p. . 32 See also the Brown Book: “if we talk of an experience of conceiving at all, we have to apply this word to the experience of using certain words, gestures, etc.” (BB, p. 137) 33 See for instance the recent study in A. N. Rich et al., “A Systematic, Large-Scale Study of Synaesthesia: Implications for the Role of Early Experience in Lexical– Colour Associations,” Cognition 98:1 (2005): 53–84. There is much to say about this definition but this is not the place. A minor point is that colours and graphemes are in the same sense modality. 34 Thomas J. Palmeri et al., “The Perceptual Reality of Synesthetic Colors,” PNAS 99:6 (2002): 4127–31; V. S. Ramachandran and E. M. Hubbard, “Synesthesia—A Window into Perception, Thought and Language,” Journal of Consciousness Studies 8 (2001): 3–34; V. S. Ramachandran and E. M. Hubbard, “Hearing Colors, Tasting Shapes,” Scientific American 288:5 (2003): 52–59; A. N. Rich and J. B. Mattingley, “Anomalous Perception in Synaesthesia: A Cognitive Neuroscience Perspective,” Nature Reviews Neuroscience 3:1 (2002): 43–52. 35 Ramachandran and Hubbard, “Hearing Colors, Tasting Shapes,” p. 18. 36 Rich, and Mattingley, “Anomalous Perception in Synaesthesia.” 37 Ibid., p. 43. 38 LW I, §800. 39 A. M. Leslie, “Pretence and Representation: The Origins of ‘Theory of Mind’,” Psychological Review 94 (1987): 412–26. 40 Ibid. 41 Ibid., p. 9. 42 L. McCune-Nicolich, “Toward Symbolic Functioning: Structure of Early Pretend Games and Potential Parallels with Language,” Child Development 52 (1981): 785–97. 43 A. Lillard, “Making Sense of Pretense,” in Children’s Early Understanding of Mind: Origins and Development, ed. Charlie Lewis and Peter Mitchell (Hove: Lawrence Erlbaum, 1994). 44 Ibid., p. 216. 45 See also P. L. Harris, “Imagining and Pretending,” in Mental Simulation: Evaluations and Applications, ed. Tony Stone and Martin Davies (Oxford: Blackwell, 1995), on this point. 46 Lillard, “Making Sense of Pretense,” p. 217. 47 It is noteworthy that several authors refer to Austin’s definition of pretence, e.g., Lillard, “Making Sense of Pretense.” Austin’s article on pretence is a purely conceptual analysis; see “Pretending,” in Essays in Philosophical Psychology, ed. Donald Gustafson (London: Macmillan, 1967). 48 LW I, §800. 49 On this point, see Oswald Hanfling, “I Heard a Plaintive Melody,” in Wittgenstein Centenary Essays, ed. A. Phillips Griffiths, Royal Institute of Philosophy Lectures (Cambridge: Cambridge University Press, 1991). Hanfling’s discussion, however, is not about the psychology of pretence. 50 PI §282. 51 Lillard, “Making Sense of Pretense,” p. 216. 52 Z §181. 53 As a purely formal object it is even less similar to a primary representation than the more traditional mental images are to sensations. 54 Paul Harris tries to save Leslie’s hypothesis by supposing that quarantining involves a membrane that permits only one-way traffic, from the primary to the secondary representation but not the reverse. But this is not a description of an actual mechanism but merely a metaphor. More importantly, the metaphor in fact

106

55

56

57 58 59 60 61 62 63 64 65 66 67 68

Michel ter Hark

hides the conceptual or logical point concerning the dependency of concepts and not Leslie’s hypothesis. See “Imagining and Pretending.” S. Baron-Cohen, “The Development of a Theory of Mind in Autism: Deviance and Delay,” Psychiatry Clinics of North America 14 (1991): 33–51; Leslie, “Pretence and Representation.”; A. M. Leslie, “Some Implications of Pretence for Mechanisms Underlying the Child’s Theory of Mind,” in Developing Theories of Mind, ed. Janet W. Astington et al. (Cambridge: Cambridge University Press, 1989). A. Lillard, “Young Children’s Conceptualization of Pretense: Action or Mental Representational State?”, Child Development 64 (1993), p. 373. Note that this formulation expresses precisely the logical point of the primary and secondary use of words as I understand it. Owing to the paradigm of mental representations, however, Lillard subsequently interprets knowledge in terms of mental representation. As a consequence, the “could not” is taken empirically rather than logically. Lillard, “Making Sense of Pretense,” p. 213. Ibid., p. 223. Jacqueline D. Woolley, “The Fictional Mind: Young Children’s Understanding of Imagination, Pretense, and Dreams,” Developmental Review 15:2 (1995): 183. Still, those children do not mean what a child would who makes believe that, e.g., its doll is in pain. T. P. German and A. M. Leslie, “Children’s Inferences from ‘Knowing’ to ‘Pretending’ and ‘Believing’,” British Journal of Developmental Psychology 19 (2001): 59–83. A. Lillard, “Wanting to Be it: Children’s Understanding of Intentions Underlying Pretense,” Child Development 69 (1998): 981–93. Cf. PI §281. I maintain elsewhere that this is the general tenet of Wittgenstein’s discussion of the concept of seeing-as: see ter Hark, Beyond the Inner and the Outer. See also Peter Geach, Mental Acts (London: Routledge, 1957). German and Leslie, “Children’s Inferences from ‘Knowing’ to ‘Pretending’ and ‘Believing’.” Geach, Mental Acts. Malcolm Budd, “The Characterization of Aesthetic Qualities by Essential Metaphors and Quasi-Metaphors,” The British Journal of Aesthetics 46 (2006): 133–43. For research in this direction, see J. Carpendale and C. Lewis, “Constructing an Understanding of Mind: The Development of Children’s Social Understanding within Social Interaction,” Behavioural and Brain Sciences 27 (2004): 79–96.

Part II

Language

5

Concepts and primitive language-games Eva Picardi

“Nichts ist doch wichtiger, als die Bildung von fiktiven Begriffen, die uns die unseren erst verstehen lehren.”1

The primitive language-game introduced in §2 of Wittgenstein’s Philosophical Investigations has been the object of much controversy.2 Primitive languagegames are characterized by Wittgenstein as objects of comparison (Vergleichsobiekte) employed in the process of giving a perspicuous description of the way in which our language actually works. Wittgenstein writes: Our clear and simple language-games are not preparatory studies for a future regimentation of language—as it were first approximations, ignoring friction and air resistance. The language-games are rather set up as objects of comparison which are meant to throw light on the facts of our language by way not only of similarities, but also of dissimilarities. (PI §130) A statement in a similar spirit occurs in Remarks on the Philosophy of Psychology: When one describes simple language-games in illustration, let’s say, of what we call the “motive” of an action, then more involved cases keep on being held up before one, in order to show that our theory doesn’t yet correspond to the facts. Whereas more involved cases are just more involved cases. For if what were in question were a theory, it might indeed be said: It’s no use looking at these special cases, they offer no explanation of the most important cases. On the contrary, the simple language-games play a quite different role. They are the poles of a description, not the ground-floor of a theory. (RPP I, §633) Here Wittgenstein characterizes language-games as the poles of a description: they do not form part of a theory that puts forward hypotheses for

110

Eva Picardi

testing and refinement in the course of further investigation. Their function is to enable us to arrange phenomena along a scale, as it were, in such a way that similarities and dissimilarities that escaped our notice are brought to the fore. To achieve this goal we may invent language-games. However, we should not jump to the conclusion that all language-games discussed by Wittgenstein in Philosophical Investigations and elsewhere in his writings are invented. But, one may wonder, is it really the case that the language-game from §2 (hereafter “language-game No. 2”) is a good starting point for casting light on our own language? Unlike Rhees3 and Brandom,4 I believe that the answer to this question is “Yes.” However, I am not sure that the reasons why this is so have been fully appreciated to date. This chapter is chiefly devoted to spelling out those reasons. The elucidation of what is involved in concept-mastery is the main, but by no means the only, function that the invention of simple language-games and their extensions plays in Wittgenstein’s Philosophical Investigations. Also, the idea of primitiveness deserves scrutiny if we are to take seriously Wittgenstein’s suggestion to conceive the Builders’ language-game No. 2 as exhausting the linguistic repertoire of a tribe. The idea of primitiveness lends itself to at least three readings: (a) a methodological construal, according to which primitive (i.e. very simple) language-games are invented models whose main function is to shed light on the complexity of our own language; (b) a genealogical reading, according to which primitive language-games can be seen as either rudiments of fully developed languages or isolated fragments of such languages; (c) a literal reading, which intimates that we take seriously the idea that there are such things as primitive behavior and primitive language. We need not think here of cavemen: small children and mentally hindered persons will do—at any rate, it is to them that Wittgenstein calls attention in his writings. It is on this last reading that I will concentrate in what follows. The contrast between the genealogical and the methodological reading is more a matter of emphasis than of substance, for the “genealogy” in question is a philosophical, not a historical one. They can both be subsumed under the heading of what has been called by Schulte5 the “morphological method.” As Schulte points out, Wittgenstein’s conception of morphology builds on ideas of Goethe and (to a lesser degree) Spengler. The basic idea of such a morphology is that of a representation that links together phenomena in such a way that they can be taken in at a glance and are arranged along a continuum. This way of presenting phenomena has an advantage over scientific theorizing by allowing us to see connections.6 It sharpens our appreciation of structure: hence the importance of finding and inventing intermediate links.7 Once we have created a perspicuous representation of the workings of our language, we achieve the sort of clarity that brings our philosophical perplexities to a (provisional) halt. The structure of my chapter is as follows. After spelling out what I consider to be the philosophical significance of the Builders’ game in Wittgenstein’s

Concepts and primitive language-games

111

philosophical writings after 1930, I will concentrate on the Builders’ game as it is presented and discussed in Philosophical Investigations. In particular, I will focus on the bearing of the idea of primitiveness on the notions of thinking and concept mastery that Wittgenstein sees as implicit in the Augustinian picture of language. In the second part of the chapter I will deal with the (different) criticisms raised by Rhees and Brandom concerning the Builders’ game. I will argue that the language-game No. 2 and its extensions are a good starting point for investigating the relationship between action, speech and thought. In the third part of the chapter I focus on the other end of the spectrum, away from primitiveness and well into the jungle of our sophisticated language-games. Following in the footsteps of the Builders, I suggest that we draw a parallel between Wittgenstein’s technique of extending a primitive language and Dummett’s employment of the notion of a conservative extension for exploring the idea of harmony between the different aspects of our use of words and sentences.

5.1 The philosophical significance of primitive language-games As is well known, language-game No. 2 is taken up again and again in Wittgenstein’s writings. Whenever in his writings Wittgenstein dwells on the idea of primitiveness—be it primitive behavior, primitive logic, primitive thinking—the Builders are sure to be summoned. Whenever he wonders what it is to extend a given language-game by means of a new move, mention of language-game No. 2 is likely. It stands to reason that the argumentative function of the Builders’ primitive language-game is not confined to exposing the shortcomings of Augustine’s philosophical account of the way he learned to speak (as distinct from his very insightful description of the circumstances under which his learning took place). Seen from a broader perspective, the argumentative function of the Builders’ language-game in the context of the Philosophical Investigations is much more complex than meets the eye. As I have pointed out, the Builders of language-game No. 2 turn up again and again in Wittgenstein’s writings. To the best of my knowledge, the ancestor of language-game No. 2 is to be found in section 7 of the Big Typescript (and its manuscript precursor, WA 4, p. 10). The Builders are not there yet, but Augustine and Plato are. Wittgenstein there says that certain simple games that fit Augustine’s description of the way names (substantives) are used actually occur in our language, and as an example he offers “Slab,” “Block,” and “Pillar,” uttered in the process of building a house. The only (significant) difference between the way in which the building game is presented in the Big Typescript and Wittgenstein’s later writings (including The Blue and Brown Books) is that before issuing the orders (“Slab,” “Pillar,” “Block”), the participants set up a convention concerning the names of the building stones—“That is called ‘a slab’,” “That is called ‘a block’.” Then Wittgenstein goes on to remark that from the way Augustine describes how he learned to use language one may

112

Eva Picardi

get a clue as to from where a certain primitive picture of meaning arises. What Augustine describes is a calculus, only not all that we call language is such a calculus. This picture is primitive in so far as it centers on the assumption that the naming of things constitutes the foundation and essence of language.8 The Builders make their last appearance in On Certainty, where Wittgenstein imagines a child who, having mastered language-game No. 2, goes through the steps of mastering what counts as an extension of it. In On Certainty the intricate connections between the concept of knowledge and the language-games played with the verb “to know” in knowledge ascriptions (both in the first and third person) is the focus of Wittgenstein’s investigations. Possibly, one day the Builders (or the child) will come up with utterances such as “I know that this is called ‘a slab’.”9 A question, among the many that Wittgenstein asks in that context, is whether and how the addition of the words “I know” to the simple assertion “This is called ‘a slab’” extends the expressive possibilities of an extant range of language-games. Some philosophers hold that the only argumentative function of the Builders’ language-games is that of providing a reductio ad absurdum of the picture of language based on the Augustinian model.10 Others have dwelled on the irony implicit in the use of the word ‘primitive’ in the opening sections of Philosophical Investigations.11 Still others have emphasized the therapeutic lesson implicit in Wittgenstein’s discussion of the various uses to which proper names are put in our language.12 More recently Stern has suggested that we read the text of Philosophical Investigations and the opening sections in particular, as a Pyrrhonian dialogue13 The voice of temptation and the voice of correctness—to use Cavell’s terminology14— present the reader with descriptions of our linguistic practice which embody different conceptual priorities. The narrator (possibly Wittgenstein) often corrects the voice of temptation, but does not unequivocally endorse a specific view. There is much to be said in favor of this interpretation. However, I doubt that it does full justice to what Wittgenstein saw himself as doing in the Philosophical Investigations, for it pushes into the background the positive results that philosophy is expected to yield. In a section of the Philosophical Grammar where Wittgenstein asks the question what it is for a sentence to belong to a language or for a language to be extended (erweitert) in a certain direction, he says: What does a man do when he constructs (invents) a new language; on what principle does he operate? For this principle is the concept of ‘language’. Does every newly constructed language broaden (alter) [erweitert] the concept of language?—Consider its relationship to the earlier concept: that depends on how the earlier concept was established.—Think of the relation of complex numbers to the earlier concept of number; and again of the relation of a new multiplication to the general concept of the multiplication of cardinal numbers, when two

Concepts and primitive language-games

113

particular (perhaps very large) cardinal numbers are written down for the first time and multiplied together. (PG §71) [ … ] But if the general concept of language dissolves in this way, doesn’t philosophy dissolve as well? No, for the task of philosophy is not to create a new, ideal language, but to clarify our use of our language, the existing language. Its aim is to remove particular misunderstandings; not to produce a real understanding for the first time. (PG §72; my emphasis) This section of Philosophical Grammar is interesting for several reasons: it contains an early mention of the idea of extending a language; it hints at the logico-mathematical provenance of this notion; and it offers a description of the task of philosophy very similar in letter and spirit to the one to be found in §133 of Philosophical Investigations. It is a humble and humbling task that Wittgenstein assigns to philosophy: the removal of specific misunderstandings without producing an alternative general picture of our use of language. To be sure, after 1930 the idea of constructing an ideal language is dropped. What is retained, however, is the technique of inventing simple languagegames and extending them in various directions, in order to cast light both on the specific concepts deployed in our complicated language-games and on what is involved in having a language in the first place. The gist of the reminder concerning the task of philosophy, applied to the opening sections of Philosophical Investigations, is that no attempt is being made to produce “genuine understanding [eigentliches Verstehen] for the first time,” otherwise the author would be a party to the game played by Augustine. Thus neither the first 137 sections of Philosophical Investigations nor the whole of it should be seen as offering a full-fledged picture of language alternative to the Augustinian or Tractarian ones. Wittgenstein contents himself, first, with pointing out what he takes to be the source of certain philosophical errors and, second, with removing specific misunderstandings by showing where, in detail, a certain picture has led or might lead us astray. The mere staging of Pyrrhonian dialogues falls short of even such a humble task. The philosophical problems surrounding the notions of meaning, acting, thinking, intending do not tend to dissolve by appeal to deflationary policies, as, say, Rorty’s,15 Diamond’s16 or Horwich’s17 varied interpretations of Wittgenstein’s meta-philosophical statements suggest. The achievement of clarity—the complete clarity of Philosophical Investigations §133—is a substantive ideal, ill-served by deflationary strategies. Sometimes the only clarity to be had consists in acknowledging that no satisfactory way of coming to grips with a given set of problems has been found. To lay bare the perplexing features of a philosophical problem, without indulging in philosophical theorizing is a habit of thought that philosophers will find very hard to adopt, as Wittgenstein knew from his own experience.

114

Eva Picardi

5.2 Augustine and the builders The Builders’ language-game, introduced in §2 of Philosophical Investigations and extended in the following sections18 by means of new words, new gestures, new paradigm is officially presented by Wittgenstein as an example of a system of communication (ein System der Verständigung) to which the Augustinian picture of the functioning of language is appropriate. Wittgenstein invites his (bewildered) readers to conceive of it as the complete primitive language (eine vollständige primitive Sprache) of a tribe. Languagegame No. 2 consists of four shouts (Rufe) uttered by builder A and the reactions of his assistant, builder B. B hands over a slab, a pillar, a block or a beam to A, depending on whether A calls out “Slab,” “Pillar,” “Block,” or “Beam.” Everything is described as flowing without a hitch, in spite of the huge indeterminacy lurking everywhere.19 In the first book of the Confessions Augustine reconstructs the stages he went through when, as a child, he learned to use language. The essential goal—this is the result of the anamnesis of the philosopher Augustine—was to figure out which object was pointed out to him by his elders when uttering a certain word (vox). Wittgenstein’s sums up the results of Augustine’s philosophical account in terminology strongly reminiscent of his Tractatus Logico-Philosophicus:20 “[ … ] the individual words in language name objects—sentences are combinations of such names. [ … ] Every word has a meaning. The meaning is correlated with the word. It is the object for which the word stands” (PI §1).The philosopher Augustine recalls that his attention, as a child, was focused on two orders of facts. On the one hand, he effortlessly attuned himself to the gestures (e.g., bodily posture, facial expression, the play of the eyes, the tone of voice) exploited by his elders in directing his attention to the object in the presence of which a certain word was uttered. On the other hand, he attended to the various positions (possibly syntactic roles) in which the word in question featured in the (small, I suppose) sample of sentences to which he was exposed. Putting together these two orders of facts he got clues as to the specific role played by a given word on a certain occasion of its use. Indeed, the child Augustine apparently got much more out of the ostensive teaching to which he was exposed. For he soon became alert to one of the chief advantages implicit in mastering the naming technique, i.e., exploiting it as a means for getting across his own desires (voluntates) to others, thus getting his elders to do things for him—fetch the objects for which he has learned to pronounce the names.21 But, as Wittgenstein points out, in order to act on others one needn’t name anything. In uttering “Help!,” “No!,” “Ow!,” “Fine!,” “Away!,” “Water!” no act of naming is performed, but, with luck, one manages to elicit a response. In language-game No. 2, too, there is no such thing as asking something’s name.22 The game of asking something’s name—the game of the name, as I suggest we call it—is a language-game in its own right, and must be mastered as such. It is in fact one of the

Concepts and primitive language-games

115

first extensions of language-game No. 2 described in §8 of Philosophical Investigations. The Builders enter the picture at this stage of Augustine’s narrative. What Wittgenstein wants us to realize is how vastly Augustine overestimates the significance of the ostensive training. Even a simple game such as the one played by Builders presupposes for its intelligibility a background of activities in which the ostensive training is embedded: [ … ] With different training the same ostensive teaching of these words would have effected a quite different understanding. “I set the brake up by connecting up rod and lever.”—Yes, given the whole of the rest of the mechanism. Only in conjunction with that is it a brake-lever, and separated from its support it is not even a lever; it may be anything, or nothing. (PI §6) One is of course here reminded of Frege’s context principle, which is in fact explicitly stated by Wittgenstein in §49.23 However what matters here is not so much the linguistic context (the different kinds of words and the different roles that the same word can play in the context of various sentences), but the web of activities in which the ostensive learning of words is embedded. The ostensive teaching of the word “slab” does not take place in a void, but is part of a certain activity (the building of a house) in the context of which the shouting of “Slab” brings about the desired understanding, i.e., that a slab (rather than, say, a brick or a pillar) is to be handed over by B to A (rather than, say, concealed or destroyed), depending on whether the latter utters “Slab.” The point here is that it is an understanding of “Slab” that is in part responsible for B’s acting as he does. I’ll return to this in §5.5 of this chapter. Thus, one of the relatively simple reasons why the picture of the essence of language prompted by the Augustine’s narrative is misleading is its onesidedness: it focuses on the phenomenon of ostensive teaching without paying attention to the specific context in which the teaching takes place. It also overlooks the distinction between ostensive training and ostensive definition. It is one thing for the child to be confronted with the words, “That is a slab,” uttered by its teacher in the process of pointing to certain stone salient to both of them. It is a very different thing for the child to reach the stage where it can quiz its elders about the names of things, of people or of specific features of things and people. It is easy to conflate ostensive learning (“That is a slab”) with ostensive definition (“That is called ‘a slab’”). Only somone who is able already to do something can inquire after the name of a certain color or of a certain move in a game of cricket or of a certain musical pattern.24 Plainly, the abilities and skills exploited by the teacher in ostensive training are grounded in the natural endowment of the human animal. Discriminative abilities, for instance, are enhanced and extended in the

116

Eva Picardi

process of learning how to make certain moves in very simple languagegames, e.g. acquired similarity standards are grafted onto our natural (possibly innate) similarity standards. However, by far the most telling criticism of Augustine’s picture is formulated in §32.25 I think this section is worth quoting in full, also because here we encounter for the first time the dramatis personae—i.e. the radical linguist and his native informants—that have haunted the imagination of students of Quine since 1960:26 Someone coming into a strange country will sometimes learn the language of the inhabitants from ostensive definitions that they give him; and he will often have to guess the meaning of these definitions; and will guess sometimes right, sometimes wrong. And now, I think, we can say: Augustine describes the learning of human language as if the child came into a strange country and did not understand the language of the country; that is, as if it already had a language, only not this one. Or again, as if the child could already think, only not yet speak. And “think” here means something like “talk to itself.” (PI §32) Augustine’s “mistake” here is that he helps himself to a certain picture of the connection between thought and speech that he has inherited from the philosophical tradition (and from Plato in particular), instead of pausing to reflect on the features of the learning situation, to which he has so keenly called attention. Seen in this light, the error that Wittgenstein detects in the Augustinian picture is its circularity, its lack of explanatory power. This reproach must strike one as very odd, for, after all, is not philosophy as such unable to provide explanations, by Wittgenstein’s own lights? And, at any rate, are there really philosophical explanations to be had concerning certain basic phenomena, such as the child’s acquisition of his first language? Of course not, and this is why pretending that one is offering an explanation is particularly misleading. Here we are moving on slippery ground, and it is essential that we understand the central intent in Wittgenstein’s argumentative moves. Augustine’s fault here is that he shifts the ground from a description of the way a child acquires his first language (sermo exterior) to the postulation of a language of thought (sermo interior). The language of thought is an internal language in which the soul talks to itself, whose possession enables the child Augustine to acquire whatever language with which it is presented. The child Augustine already thinks in such an immaterial medium. The acquisition of the external language, whose main function is to make communication possible, is made possible by its being already equipped with an internal abstract mechanism that allows him to decode, as it were, the bits of language he is presented with into the language of thought, and to encode, when the need arises,

Concepts and primitive language-games

117

sentences in the language of thought into sentences of Latin or Greek. The appeal to memory in Book 10 of the Confessions is the key to Augustine’s account of cognition. This part of Augustine’s picture—the postulation of a spiritual mechanism of thinking that accounts for the production and the understanding of speech—is a leading theme, possibly the leading theme of Philosophical Investigations and of Wittgenstein’s philosophical work. The concentration on ostensive teaching and ostensive definition is pernicious because it suggests that once we know the object a given name stands for, we have got hold of its meaning. Meaning, thus construed, strikes us as something that constrains all future applications of the word. That the use of a word flows from its meaning is certainly a philosophical thesis that Wittgenstein wanted to controvert. And the same can be said about a certain conception of thinking as a mysterious mental process. Has he succeeded? Hardly, the flourishing during the last 50-odd years of causal theories of reference and the passionate debates over reference-fixing procedures seem to testify to the soundness of Augustine’s insights regarding the importance of the practice of naming. The cognitive turn advocated by Chomsky and the vast majority of contemporary cognitive scientists, too, seems definitely to speak in favor of Augustine’s distinction between sermo interior and sermo exterior. Chomsky’s distinction between I(nternal)-languages and E(xternal)languages seems modelled on Augustine’s.27 These problems are still with us, and Wittgenstein’s questions too. As a remedy to rampant mentalism one may be tempted to follow Quine’s lead and offer a behaviorist account of language as a social art and a dispositional account of meaning and understanding. But Quine’s behaviorism, coupled to his meaning holism, has unpalatable consequences. There are many formulations of meaning holism, the most popular of which is that the unit of meaning is not the sentence, but the entire language to which it belongs. This formulation sounds very tame, in spite of its vastly implausible consequences concerning, e.g., what understanding a given sentence requires. It is as if one could never understand a sentence belonging to a given language, short of understanding the entire language to which it belongs. Since a natural (i.e., historical) language is constantly changing (expanding in some respects and shrinking in other respects) there is no telling what understanding a given sentence requires. Quine’s semantic holism has as corollaries the thesis (better, conjecture) of the indeterminacy of radical translation, the rejection of the analytic/synthetic distinction, the thesis of the indeterminacy of reference, and the rejection of all intentional vocabulary as unintelligible. All these theses are alien both to the letter and to the spirit of Wittgenstein’s philosophy. It is good to bear in mind that, possibly, there is no more to the meaning of a word or sentence than what is contained in the explanation of its meaning (i.e., in a description of the technique of its use appropriate to a given context), but also no less.28 The focus of Wittgenstein’s argumentative strategy, as it unfolds in the Philosophical Investigations, is to probe the connection between thinking and

118

Eva Picardi

speaking. His painstaking description of the Builders’ primitive languagegames helps us to pinpoint the critical steps where in our description of linguistic practice the temptation to appeal to a hidden mechanism becomes almost irresistible. Even those who eventually will yield to the voice of temptation should find his survey instructive.

5.3 The Builders’ language-game: alternative descriptions Wittgenstein claims that Augustine can be seen as presenting us with a primitive picture of language or with a picture of a primitive language. Plainly these are two very different things. So far we have explored the shortcomings of the primitive, i.e., oversimplified picture of language. But what about the second reading, what if we take Augustine as offering a picture of a primitive language? How is such a primitive language to be described? The description of language-game No. 2 presents us with a host of difficulties, no matter whether we try to describe it from the outside, following in the footsteps of Quine’s radical translator, or from the inside—as Wittgenstein suggests that we should—by making an imaginative effort to put ourselves into the shoes of the people whose life and language is so different from ours. Nearly half a century devoted to a close study of the vicissitudes of Quine’s field linguist, as described in chapter 2 of Word and Object, has made philosophy students fully conversant with minimalist scenarios such as those inhabited by Wittgenstein’s Builders. Therefore when Wittgenstein wonders whether “Slab” should be seen as a sentence (“Bring me a slab”) or as a word,29 a student of Quine is reminded of the adventures of the field linguist, who tries to project a pattern into the utterances of his informants. We know from Quine that utterances bearing on ostensible features of the environment are the entering wedge into the language of an exotic tribe both for the child and for the field linguist. Such utterances are first construed holophrastically, as exhibiting no syntactic structure. Using the method of prompted assent and dissent the radical linguist sets up a correlation between the natives’ verdictive behavior (assent, dissent, abstain) and salient features of the observable situation. It is only with the linguist’s appeal to analytical hypotheses, that the distinction between whole sentences and their component words enters the picture. In so doing the linguist projects onto the language of the natives his own parochial apparatus of definite and indefinite articles, singular and general terms, pronouns and generality expressions. In Quine’s opinion, projection is the only choice at our disposal. Seen from this perspective, Wittgenstein’s question whether “Slab” should be conceived as a sentence or as a word sounds familiar. However, Wittgenstein’s emphasis lies elsewhere. Like Quine he invites us to notice that it is because our language contains the possibility of other sentences that we are inclined to say that “Slab” means “Bring me a slab” (rather than, say “Bring him a slab,” “Bring some slabs”). Moreover our

Concepts and primitive language-games

119

language does not contain just orders, but also assertions, requests, questions, etc. What right have we got to project our own language onto a language made up only of orders? None, to be sure. The question is whether we have other means at our disposal, different from massive projection. Both Quine’s and Davidson’s answers to this question are negative. Unlike Quine, whose radical linguist quizzes the informants with the method of prompted assent and dissent, Wittgenstein urges that we take “[t]he common way of acting of mankind” (die gemeinsame menschliche Handlungsweise) as a frame of reference for finding our way into the language of a remote tribe, if, for instance, we want to find out whether the people in question give, understand, and obey orders.30 The trouble is that the fictitious tribe we are asked to imagine has such a limited repertoire, that we have the impression that such a Gedankenexperiment challenges our imagination. And, above all, what is it that Wittgenstein wants us to appreciate concerning our own language in inviting us to imagine such a fantastic tribe? Wittgenstein’s stance is clear: we should not see primitive language-games as incomplete versions of our own. Rather they are forms of language appropriate to the lives of primitive people, people, i.e., whose life is very different from ours. We should try to understand what is going on in the language from the inside, as it were, refraining from projecting the mastery of our own language onto theirs. It is important, however, that such languages are seen as capable of being extended, like the language of the Builders, or like “those games by means of which children learn their native language.”31 As Wittgenstein himself points out, the Builders’ language-game No. 2 is better described as a system of communication (ein System der Verständingung) that does not exhaust the whole of what we call language. One is here reminded of the differential cries by means of which certain varieties of monkeys alert their young to the presence of danger. Mere differential behavior in the presence of different stimuli does not count as language. We may very well say that animals (and monkeys in particular) communicate with one another, only not all ways of conveying information count as language. Animals are indeed mentioned by Wittgenstein himself in connection with primitive language-games, It is sometimes said that animals don’t talk because they lack the mental capacity. And this means, “they don’t think, and this is why they do not talk.” But—they simply do not talk. Or to put it better, they do not use language—if we except the most primitive forms of language (die primitivsten Sprachformen).—Commanding, questioning, recounting, chatting, are as much a part of our natural history as walking, eating, drinking, playing. (PI §25) To be sure, some species of animals display sophisticated forms of communication, but they don’t use language, and it can be misleading here to take

120

Eva Picardi

literally Wittgenstein’s talk of the most primitive linguistic forms. However, if instead of relating thinking to the exercise of various linguiatic skills, we relate it to the ability to extract information from the environment and get it across (communicate it) to other members of the species, then our concept of thinking and the criteria for detecting thought will be broadened. But this is not what Wittgenstein means when he alludes to certain aspects of animal behavior. All the same, isn’t Wittgenstein here playing on the ambiguity between a loose use of the word “language” and a restricted sense of language, according to which English is a language but, say, Morse code is not? The answer to this question will be found in §§493–94 of Philosophical Investigations, where Wittgenstein writes: We say, “The cock calls the hen by crowing”—but doesn’t a comparison with our language lie at the bottom of this?—Isn’t the aspect quite altered if we imagine the crowing to set the hens in motion by some kind of physical causation? But if it were shewn how the words “come to me” act on the person addressed, so that finally, given certain conditions, the muscles of his legs are innervated, and so on—should we feel that that sentence lost the character of a sentence? (PI §493) I want to say: It is primarily the apparatus of our ordinary language, of our word-language, that we call language; and then other things by analogy or comparability with this. (PI §494) Word-language has the explanatory priority on other systems of communication or various ways of imparting information: the suggestion is emphatically not that we start with apes’ cries and end up with sentences of English. Apparently, it is because it can be compared with our way of using language to impart orders that the Builders’ exchanges can be described as moves in a language-game. And the comparison makes sense only under the assumption that the motions of builder B are brought about by the shouts of builder A not just in virtue of a purely physical mechanism. What in the doings of the Builders of §2 of Philosophical Investigations rules out that they are automata? Nothing, of course. It is Wittgenstein who has described the Builders’ exchanges as informed by an understanding of the uttered sounds. This was the whole point of the criticism of Augustine’s conception of the ostensive training as taking place in a void, that we reviewed in the section above. A brings about a certain reaction in B not just by calling out “Slab,” but by calling out “Slab” in the process of carrying out an activity whose point is clear to both parties (the building a house). The idea of “bringing about” employed here is not to be read in purely

Concepts and primitive language-games

121

causal terms. To turn the shouts of A into a piece of linguistic practice something more is required. But what, exactly? Should we say that A intended to evoke a response in B by uttering as he did? Was B’s response prompted, at least in part, by B’s recognition of A’s intention, as Grice suggests?32 Well, not quite. The intention must be embedded in a situation, not the other way around. We are tempted to say that builder B acts as he does because he understands the content of the specific utterance, where the “because” is not that of causal explanation, nor (yet) the “because” characteristic of purposive action, performed with the intention of achieving a goal. The distinction between right and wrong, the idea of producing a response that counts as correct or incorrect, i.e., as conforming (or failing to conform) to the imparted order has not been explicitly introduced in the language of the Builders. They are drilled into it as they go along—much as it happens in the early stages of language learning. The criticism that Wittgenstein levels against Augustine’s description of what takes place when a child learns to speak is that the ostensive teaching is of itself unable to bring about a specific understanding of the word in question, unless we say much more about the specific context of the non-linguistic activities in which the training takes place.33 Augustine’s description strikes as unproblematic because we tacitly supply the relevant context. Although Wittgenstein does not say it in so many words I think that what makes his description of the Builders’ game a description of a language-game is that we see the regular pattern exemplified in the doings of the Builders as an instance of purposive behavior prompted by a (rudimentary) understanding of the uttered shouts. What full mastery of language requires is that we use sentences in the awareness of their significance to others: the notions of reason and purpose must enter the picture, and with them the notion of right and wrong. To be sure, the space of reasons and the space of causes—to use Sellars’ apt phrase—in such primitive scenarios are not sharply demarcated, but the basic ingredients are there. As Malcolm has pointed out, the word “primitive,” as so often in Wittgenstein’s writings, signals that we are confronted with a sort of behavior that is pre-linguistic.34 This behavior is such that “a language-game is based on it, that is the prototype of a way of thinking and not the result of thought.”35 This observation of Wittgenstein provides us with an apt transition to the objection levelled by Rush Rhees, and more recently by Robert Brandom, against taking the Builders’ language-game as a good starting point for casting light on our own language.

5.4 Are all language-games played with concepts? According to Brandom the Builders’ language-game is not a good starting point for investigating the workings of our language, for the beings engaged in these activities fail to display the conceptual abilities characteristic of language users:

122

Eva Picardi Practices that do not involve reasoning are not linguistic or (therefore) discursive practices. Thus the “Slab” Sprachspiel that Wittgenstein introduces in the opening sections of the Philosophical Investigations should not, by these standards of demarcation, count as a genuine Sprachspiel. It is a vocal but not yet a verbal practice. By contrast to Wittgenstein, the inferential identification of the conceptual claims that language (discursive practice) has a center; it is not a motley. [ … ] Claiming, being able to justify one’s claims, and using one’s claims to justify other claims and actions are not just one among other sets of things one can do with language. They are not on a par with other “games” one can play. They are what in the first place make possible talking, and therefore thinking: sapience in general.36

Brandom is not the first to deny that the Builders’ language-game is a genuine instance of linguistic practice. Rush Rhees, for instance, has denied that the language-game(s) of the Builders can even for a moment be conceived as a language. Their mechanical character makes them unfit to shed light on language in the first place, let alone a language in which a conversation can be had.37 A dog can be trained to react differentially to four distinct shouts, but we would not be inclined to say that the dog engages in a language-game with its master, but, at most, that the dog has been trained to react differentially to certain auditory stimuli.38 What makes the Builders’ game different from a retrieving game—one may suggest—is that A and B can exchange roles, as every hearer is also a potential speaker and vice versa. Similarly, builder B can issue orders to builder A. But automata can be built that perform in either role, and yet we would not be tempted to say that they engage in a language-game. The sounds uttered by automaton A merely cause automaton B to enter a certain physical state, and vice versa. As I have tried to show in the previous section, these objections are not on the right track. There is no obstacle, in the light of Wittgenstein’s description of language-game No. 2, to construing the four utterances of builder A as a piece of purposive linguistic action based on his expectation of their significance to builder B. The Builders are not just engaging in a vocal practice, but they are deploying linguistic skills of a very rudimentary type. Wittgenstein himself is ready to admit that there is a sense in which the Builders in language-game No. 2 do not master concepts yet. If thinking is equated with following rules according to logical standards, i.e., being able to engage in inferential practices, then the Builders do not think—nor do young children and mentally-hindered people. The suspicion arises that Brandom has set the standards of the conceptual too high—at the level of full-fledged propositional contents, capable of featuring as premises and consequences of an argument—thus neglecting a whole range of intermediate cases. It may come as a surprise to Brandom that Wittgenstein agrees with him as to the lack of concepts proper in the Builders’ game (No. 2). In Remarks on the Foundations of Mathematics, Wittgenstein concedes as much. Here we read:

Concepts and primitive language-games

123

It is not in every language-game that there occurs something that one would call a concept. “Concept” is like a picture with which one compares objects. Are there concepts in language-game (2)? Still, it would be easy to add to it in such a way that “slab”, “block” etc. became concepts. For example, by means of a technique of portraying or describing those objects. There is of course no sharp dividing line between languagegames which work with concepts and others. What is important is that the word “concept” refers to one kind of expedient in the mechanics of language-games. (RFM VII, §71) But why should the addition (Erweiterung) of a technique of portraying bricks and slabs turn language-game No. 2 into a game played with concepts? Here is a suggestion: once such an extension is made, builder A can issue to builder B an order phrased in such a way that the notion of acting in accordance to a standard is brought to the fore. After all, the main point of taking orders rather than assertions as the starting point of a philosophical reflection on language was to emphasize the many similarities between the practice of acting in accordance with a rule or standard and that of complying with orders. In addition to shouting “Slab,” builder A may now utter words to the effect of “Bring me something that fits this!” where the demonstration is directed at a schematic drawing of a slab. What both Builders must master is the use of the drawing of a building stone as a standard that can be employed both to direct action and to assess whether things done are in conformance to the standard. In the absence of a full grasp of the difference between right and wrong, between what counts as a correct move and what counts as incorrect in a given game, we are not entitled to describe a language-game as one “played with concepts.” Now, extending language-game No. 2 with a technique of portraying slabs and using such drawings as a standard does not tell us yet whether an utterance of builder A is to be interpreted as “More slab!” or “More slabs!,” more slab material or more individual items of the same size and shape. Placing labels on the building stones (the momentous extension of languagegame No. 2 described in §8) does not yet settle this question. One may comply with the first order without complying with the second. Possibly, these questions may be raised and answered in further extensions of language-game No. 2, where definite and indefinite articles and generality expressions are introduced. The advent of full conceptualization—one may surmise—goes hand in hand with the mastery of concepts such as “all,” “many,” “some,” “just as many,” together with a grasp of the concept of sameness and difference.39 We usually take it for granted that creatures who take part in the practice of applying concepts, following rules, drawing inferences, etc., engage ipso facto in the activity of thinking. The question Wittgenstein asks in those

124

Eva Picardi

passages where he says that the concept of thinking is a widely ramified one, is whether operating with concepts is a necessary condition of thought.40 It is in the context of this discussion that another facet of the idea of primitiveness comes to the fore. To be sure, performing linguistically and thinking are closely related. However, if what we have said so far concerning concepts is on the right track, we should expect that the “symptoms” of thinking can be detected also in the absence of a full-fledged mastery of concepts.

5.5 Thinking While Brandom spells out in full what he means by “thinking,” we would look in vain in Wittgenstein’s writings for a definition of thinking—in fact he positively refrains from offering one.41 What he offers is a disclaimer: “If a normal human is holding a conversation under normal circumstances, and I were to ask what distinguishes thinking from not thinking in such a case—I should not know what to answer. And I could certainly not say that the difference lay in something that goes on or fails to go on while he is speaking.”42 In the context of similar reflections on thinking that in volume I of Remarks on the Philosophy of Psychology the Builders put in an appearance. Wittgenstein wonders: Do we say that anyone who talks sense is thinking? E.g. the builder in language-game (2)? Might we not imagine the building and the calling out of the words etc. in a surrounding in which we should not connect them with any thinking? For “thinking” is akin to “considering”. (RPP I, §559) And he goes on to say: The expression, the behaviour, of considering. Of what do we say: It is considering something? Of a human being, sometimes of a beast. (Not of a tree or a stone). One sign of considering is hesitating in what you do (Köhler). (Not just any hesitation). (RPP I, §561) Wittgenstein’s train of reasoning here seems roughly as follows. First of all, his puzzled interlocutor, who, confronted with very primitive languagegames like the Builders’ language-game reacts in dismay by saying, “You can’t possibly count something as primitive as that as language (let alone a language)!” He is reminded how little it takes to turn a primitive languagegame into a language-game of our own language. The interlocutor is then brought to appreciate the paramount importance of Umgebung (overall context) for recognition. A small extension of a given language-game suffices to turn it into a familiar language-game played with concepts. For, as we

Concepts and primitive language-games

125

have seen, no sharp line exists between games played with concepts and those played without them, much depends on the context. Then the question is raised as to how close is the connection between performing linguistically, even in the absence of full mastery of concepts, and thinking. Here, again, the paramount importance of the Umgebung (overall context) is underlined. In the Remarks on the Philosophy of Psychology, we read: And here something about my language-game No. 2 should be said.— Under what circumstances would one really call the sounds of the builder, etc., a language? Under all circumstances? Certainly not!—Was it wrong then to isolate a rudiment of language [ein Sprachrudiment] and call it language? Should one perhaps say that this rudiment is a languagegame only in the context [Umgebung] of the whole that we usually call our language? (RPP II, §203) Now in the first place this surrounding is not the mental accompaniment of speech; it is not the “meaning” [das Meinen] and “understanding” [Verstehen] which one is inclined to consider as essential to language. (RPP II, §204) It would only be dangerous to me if someone were to say: “You are just tacitly assuming that these people think; that they are like people as we know them in that respect; that they do not carry on that language-game merely mechanically. For if you imagined them doing that, you yourself wouldn’t call it speaking.” What am I to reply to this? Of course it is true that the life of those men must be like ours in many respects and that I said nothing about this similarity. But the important point is precisely that I can imagine their language, and their thinking too, as primitive; that there is such a thing as “primitive thinking” which is to be described via primitive behaviour. (RPP II, §§205) This passage raises a number of tantalizing questions as regards the connection between thought, speech, and behavior. However, having fully appreciated the thrust of §34 of Philosophical Investigations, we are now in a better position to understand why it would be dangerous to agree with someone who claimed that it is on the assumption that we are all equally endowed with an inner mechanism responsible for the spontaneity of thinking that we grant that the Builders engage in a linguistic practice. Between automatic performances and fully intentional ones there is a vast range of intermediate cases. As Schulte remarks, Wittgenstein does not call language-game No. 2 a fragment of our language but a rudiment.43 What we have here is a basic

126

Eva Picardi

structure which is far from being fully developed. Its not being fully developed suggests that it could develop into something more complex. However, Wittgenstein also insists that “there is such a thing as ‘primitive thinking’ which is to be described by way of primitive behaviour.” Here the emphasis is on the connection between thinking and speaking. We should not jump to the conclusion that thinking and speaking are connected in such a way that there is no thought without language. As Golfarb pointed out long ago in his discussion of the Builders’ language-game: In simply presenting the possibility of taking the builders to have “rudimentary thought”, the sort of thought appropriate to their lives, Wittgenstein means to undermine the sharp dichotomy “either they are just like people as we know them or they are automata.” More important, though, he is posing two general questions: From what stance are standards for what counts as “thinking” supposed to come? Upon what features is the power we are ascribing to thought supposed to be based? These questions have a particular force, I think, against those who hold that whatever handle we can get on the notion of thought comes only via considerations about language. For Wittgenstein is suggesting that they may well have stacked the deck in implicitly (or even subliminally) using a conception of thought to frame the characterization of what is to be analysed, and thereby coloring the data that have to be accounted for.44 I fully concur with Golfarb’s observations. Wittgenstein’s insistence on the descriptive/morphological stance and the attendant rejection of theorybuilding in philosophy should not blind us to what I have called the literal reading of primitiveness, i.e., to the fact that there really exist primitive behavior and primitive thinking. We must not make the mistake of considering such games as impoverished versions of our fully developed forms of linguistic usage. This strand of Wittgenstein’s argument does not simply amount to (although it is obviously connected with) a warning not to use words such as “incomplete” or “primitive” in a teleological fashion, as if we were confronted with the early stages of a development which had been impeded or truncated. In the passage quoted above, Wittgenstein alludes to the languages-games in which children engage; and in §646 of the Remarks on the Philosophy of Psychology, volume I, he suggests that we think of the language used by people who are mentally hindered as exhibiting a form of primitive order: The feeble-minded are pictured in our imagination as degenerate, essentially incomplete, as it were in rags. Thus as in a state of disorder, rather than in a more primitive order (which would be a far more fruitful way of looking at them).

Concepts and primitive language-games

127

The suggestion implicit here is that we should take these performances as they stand, and refrain from comparing them to more refined ones. Thinking is indeed a very ramified concept.

5.6 Extending a language-game The word “Erweiterung” that Wittgenstein uses in connection with languagegames45 has been translated into English as “expansion.” In this chapter I have translated “Erweiterung” as “extension,” for I want to preserve also in the terminology the connection with the mathematical notion of an extension, i.e., a procedure that allow us to extend a calculus or a formalized language by adding new symbols together with a set of rules and definitions governing their inferential behavior. This connection is drawn by Wittgenstein himself, in various of his writings—a connection that Fogelin correctly emphasizes in the context of his discussion of the notion of family resemblance.46 With reference to Part V of the Philosophical Grammar,47 where Wittgenstein draws a connection between the notion of proposition and that of number, Fogelin writes: The cardinal number, the rational numbers, the real numbers, etc. are each well defined—that is, each extension of the number system is carried out with rigor. What is not well defined, or better, what is not the subject of a definition at all is the extension of the concept that might take place in the future. This shows that the notion of family resemblance is not just the old notion of vagueness putting on airs. A second reason why the example of numbers is important is that it shows that the extension of a concept can involve categorical shifts. If, for example, we treat the relationships between real numbers as relationships between sequences, the categorical difference between cardinal numbers and real numbers becomes evident.48 It is in a spirit similar to Fogelin’s that I propose that we approach languagegame No. 2 and its extensions in Philosophical Investigations.49 As Juliet Floyd has pointed out, Wittgenstein took the concept of a perspicuous representation to be of fundamental importance both in discussing mathematics and in discussing language.50 Much the same applies, I suggest, to the concept (or procedure) of extending a language. I take it that one of the guiding assumptions of Wittgenstein’s philosophical work is that expressive enrichment and conceptual enrichment go hand in hand. More specifically, the process of extending a given language-game into a new one and that of constructing new concepts often go hand in hand. The extensions to which a given concept lends itself are not fully determined by the rules that governed its use in the language-games in which it had been deployed so far. Often the question arises whether we are dealing with a new concept or with a novel application of an old concept (perhaps the rules governing the concept in question had been only incompletely stated). If, instead of concentrating our

128

Eva Picardi

attention only upon the rules that implicitly or explicitly have governed our application of a given concept so far, we pay attention also to the ways in which the concept in question can be fruitfully linked to its neighboring ones, we may be able to frame the alternative (new concept or new application of a given concept?) in a different way. Plainly, there are ways of extending a language or a game that destroy the game. Let’s consider, for instance, Black’s “epichess” the variant of chess where each player has a right to move any piece as if it were another of equal or different value, provided that the other player accept such a move.51 The paradigmatic example of a non-conservative extension of a language or calculus is one whereby, with the introduction of a new symbol governed by certain rules, any sentence becomes derivable in it that was not derivable before. Prior’s logical connective “tonk” provides an example of such an extension, and there is a vast literature concerning the connection between the meaning of connectives, as given by the rules governing their inferential potential, and the meaning of the connectives as given by truth-tables.52 But outside logic things are not so clear-cut. Suppose, for instance that the players of epichess have not so far availed themselves of the new rule, and have played conventional chess. Should we say that the mere existence of such an agreed-upon possibility has altered their conception of chess? For epichess now contains an agreed-upon rule that is such that, had the players made use of it, they would have disrupted the point of the game (its Witz as Wittgenstein calls it in PI §564). It is unclear to what extent a follower of Wittgenstein could or ought to avail herself of counterfactual reasoning. What seems advisable, however, is to steer clear of two opposite claims: that every possible extension of a language, no matter whether it is actually carried out, changes the entire language, as a radical holist would have it;53 or that none ever does, for whenever we make a change in the rules governing a given word, we should choose a new word, instead of loading the existing ones with new applications, and, possibly, new meanings. The extensions I am interested in are those that logicians call “conservative extensions,” i.e., extensions that increase the expressive power of a language without calling into question the existing fragment of the language.54 The extensions of language-game No. 2 in §§8, 41–42 and 48 of Philosophical Investigations are conservative, and the same applies to the extensions discussed in On Certainty. There is no denying, however, that the addition of new moves or new rules casts a different light on the previous ones. For example, the possibility envisaged in §42 of extending languagegame No. 2 with empty names, and getting away with it, casts a different light on the way the game of the name was introduced in §8. In §41 of Philosophical Investigations Wittgenstein writes: In §15 we introduced proper names into language (8). Now suppose that the tool with the name “N” is broken. Not knowing this, A gives B the sign “N”. Has this sign meaning now or not?—What is B to do when he

Concepts and primitive language-games

129

is given it?—We have not settled anything about this. One might ask: what will he do? Well, perhaps he will stand there at a loss, or show A the pieces. Here one might say, “N” has become meaningless; and this expression would mean that the sign “N” no longer had a use in our language-game (unless we gave it a new one). “N” might also become meaningless because, for whatever reason, the tool was given another name and the sign “N” no longer used in the language-game.—But we could also imagine a convention, whereby B has to shake his head in reply if A gives him the sign belonging to a tool that is broken.—In this way the command “N” might be said to be given a place in the language-game even when the tool no longer exists, and the sign “N” to have meaning even when its bearer ceases to exist. At last the Builders display the signs of hesitation, reflection, consideration, and successful problem-solving characteristic of human thinkers. The feeling of Verfremdung we experience when we first encounter the Builders is brought about, among other things, by the circumstance where everything runs without a hitch or a flaw: reflection, doubt, hesitation are alien to the Builders, as Wittgenstein describes them in the opening sections. Nor are we told what counts as failure to comply with the order “Slab,” e.g., how much time can elapse between the issuing of the order and its execution, whether the order counts as obeyed if builder B brings something else in addition to a slab or if he brings pieces of slab material or more than one slab. But in §§41–42 we witness a dramatic change. First of all, it is remarkable that a sign for denial (a shake of the head) is introduced in the language of the Builders not in a context where an order has been carried out incorrectly or disobeyed, but under circumstances that make it impossible to carry it out. An empty name, i.e., device for singular reference that at the time of the Tractatus Logico-Philosophicus was not even admitted into the language, is used in Philosophical Investigations to illustrate how to make a new move in a language-game. The role of vacuous proper names had been discussed at length by Frege, Russell and the author of Tractatus Logico-Philosophicus. Very important metaphysical and epistemological consequences were supposed to follow from the answer we give to questions concerning the truth-conditions of sentences with empty proper names. As an antidote against philosophically inflationary treatments of proper names, Wittgenstein directs our attention to the ways in which the Builders handle orders with vacuous names. Sometimes it may be convenient to handle such sentences as if they failed to convey an order at all, sometimes as if the conveyed order is such that, as a matter of fact, it can no longer be complied with. All the pathos surrounding Frege’s, Russell’s and young Wittgenstein’s theories of empty names evaporates. Not only in §41 do the builders cope successfully with a co-ordination problem—to use Lewis’s apt phrase55—but they also exploit the accident in the invention of the new language-game described in §42:

130

Eva Picardi But has for instance a name, which has never been used for a tool also got a meaning in that game?—Let us assume that “X” is such a sign and that A gives this sign to B—well, even such signs could be given a place in the language-game, and B might have, say, to answer them too with a shake of the head. (One can imagine this as a sort of joke between them).

In so extending their repertoire of possible moves the Builders have achieved a remarkable degree of sophistication: they now seem capable of engaging in a game of pretending, i.e., they employ sign X in the full knowledge that it has never been used as a name of any of their tools. They go through the moves of the language-game described in §41 but in this context, B’s shake of the head acquires a new role, different from the one it played in the language-game described in §41. To an as-if order issued by A, B replies with an as-if denial. The point of the whole procedure, we are told, is Belustigung (amusement). They play the game just for fun. Should we say that the Builders described in §§41–42 of Philosophical Investigations are in the process of acquiring a taste for modal distinctions—between, say, an order that is in principle impossible to satisfy and an order that, as a matter of fact, can no longer be satisfied? It does not really matter what we say here: only consider how little it takes to turn the Builders of §2 of Philosophical Investigations into thinkers conversant with the actual, the possible and the contingent. It is easy, however, to misunderstand the significance of what Wittgenstein is accomplishing in §§41–42. The reminder of the variety of ways in which we actually use empty proper names in ordinary language has no tendency to show that different ways of using them would be illegitimate or conceptually flawed.56 We may cope with the presence of empty names in different ways, different extensions of language-games 2 and 8 are possible, indeed, actual. We do not have clear-cut intuitions concerning how to handle sentences with empty names, neither in ordinary discourse nor in modal extensions of the language. My tentative suggestion here is to say that language-games with empty names depend for their intelligibility on the language-game played with names for existing things and people. Moreover, it is hard to imagine that the extension of the game described in §42 could be made before, or independently of, the extension described in §41. We should not lose sight of the interconnectedness of language-games, the conceptual tools required in certain extensions of the language depend on the availability of the conceptual resources acquired at a lower level. Thus, in Dummett’s terminology, there is a relation of relative priority among language-games, there is (partial) order in the motley.

Notes 1 CV, p. 85. “Nothing is more important for teaching us to understand the concepts we have than constructing fictitious ones” (24.10.1948).

Concepts and primitive language-games

131

2 For a survey of these interpretations see David G. Stern, Wittgenstein’s Philosophical Investigations: An Introduction (Cambridge: Cambridge University Press, 2004), ch. 4. 3 Rush Rhees, “Wittgenstein’s Builders,” in Discussions of Wittgenstein (London: Routledge & Kegan Paul, 1970). 4 Robert Brandom, Articulating Reasons: An Introduction to Inferentialism (Cambridge, MA: Harvard University Press, 2000). 5 Joachim Schulte, “Chor und Gesetz. Zur ‘morphologische Methode’ bei Goethe und Wittgenstein,” in Chor und Gesetz: Wittgenstein im Kontext (Frankfurt am Main: Suhrkamp, 1990). 6 Cf. PI, §122. 7 This idea is forcefully expressed by Wittgenstein in Part I, §35 of Philosophical Grammar, where he qualifies as too primitive the usual account of what makes it possible for a concept-word to apply to objects, i.e., their sharing an underlying common property. Here Wittgenstein introduces for the first time the notion of family resemblance, together with the idea that the members of a family should be arranged in a chain-like fashion (kettenartig), in such a way that the interpolation of intermediate links (Zwischenglieder) allows us to see connections and similarities between remote family members, that might otherwise have escaped our notice, Ludwig Wittgenstein, Philosophische Grammatik (Oxford: Basil Blackwell, 1969), p. 75. Wittgenstein’s statement in §109 of Philosophical Investigations that in philosophy description should take the place of explanation and theoryconstruction can be seen as a meta-philosophical application to his own way of doing philosophy of his insights concerning the way in which questions concerning our mastery of concepts should be formulated. 8 BT, p. 31; PG, p. 57. 9 OC §566. 10 E.g., Anthony Kenny, Wittgenstein (London: Allen Lane, 1973); and, to a large extent, Gordon Baker and P. M. S. Hacker, Wittgenstein: Understanding and Meaning, vol. 1, An Analytical Commentary on the Philosophical Investigations (Oxford: Blackwell, 1980). Cp., for a recantation of that reading, Gordon Baker, “Philosophical Investigations Section 122: Neglected Aspects,” in Wittgenstein’s Philosophical Investigations: Text and Context, ed. Robert L. Arrington and Hans-Johann Glock (London: Routledge, 1991). 11 E.g., Joachim Schulte, “The Builders’ Language—The Opening Sections,” in Wittgenstein at Work: Method in the Philosophical Investigations, ed. Erich Ammereller and Eugen Fischer (London: Routledge, 2004). 12 E.g., Eike von Savigny, “Diagnosis and Therapy: Christening Simple Objects. PI 38–64,” in Wittgenstein at Work: Method in the Philosophical Investigations, ed. Erich Ammereller and Eugen Fischer (London: Routledge, 2004). 13 Stern, Wittgenstein’s Philosophical Investigations, pp. 21 and 165. 14 Stanley Cavell, The Claim of Reason: Wittgenstein, Skepticism, Morality and Tragedy (Oxford: Clarendon Press, 1979). 15 Richard Rorty, “Is Truth a Goal of Inquiry?,” The Philosophical Quarterly 45: 180 (1995): 281–300. 16 Cora Diamond, The Realistic Spirit: Wittgenstein, Philosophy, and the Mind (Cambridge, MA: MIT Press, 1991). 17 Paul Horwich, From a Deflationary Point of View (Oxford: Clarendon Press, 2004). 18 E.g., §§8, 41–42, 48. 19 In the Brown Book, on the other hand, where the Builders’ game is described in full for the first time, questions concerning misunderstandings due to the underdetermination of the described situation (calling out certain words as orders and acting according to them) are raised and answered. For a discussion of this aspect

132

20 21 22 23

24 25

26

27 28

29 30

Eva Picardi

of Wittgenstein’s philosophy, cf. Charles Travis, The Uses of Sense: Wittgenstein’s Philosophy of Language (Oxford: Oxford University Press, 1989). Cf. TLP 2.03, 3.201–3.221, 4.0311. This, incidentally, is one of the truly mythological bits of Augustine’s narrative. PI §24. As formulated in his Grundlagen der Arithmetik, “Nur im Zusammenhange eines Satzes bedeuten die Wörter etwas” [Only in the context of a sentence do words have a meaning], Gottlob Frege, Die Grundlagen der Arithmetik: Eine logisch mathematische Untersuchung über den Begriff der Zahl, Centennial ed. (Hamburg: Meiner, 1986), §62, cp. also §106 and Foreword, p. XXI. For a discussion of Frege’s dictum in the context of Wittgenstein’s Philosophical Investigations, cf. Baker and Hacker, Wittgenstein: Understanding and Meaning, pp. 258–83 and passim. PI §30. Cf. Eva Picardi, “Sensory Evidence and Shared Interests,” in Interpretations and Causes: New Perspectives on Donald Davidson’s Philosophy, ed. Mario De Caro (Dordrecht: Kluwer Academic, 1999). I take it that McDowell’s idea of second nature points in a similar direction, too; cf. John McDowell, Mind and World (Cambridge, MA: Harvard University Press, 1994). The translation is said to be radical since there is no manual of translation from the exotic language into the language of the field-linguist, nor into a third language which is connected via a manual of translation with that of the fieldlinguist. However, this thought experiment merely serves to dramatize a situation with which we are all familiar. For, according to Quine, radical translation starts at home: see Word and Object (Cambridge, MA: MIT Press, 1960), § 16. Quine himself has drawn attention to Wittgenstein’s Blue and Brown Books in the course of his discussion of the thesis of the indeterminacy of radical translation (p. 77, n. 2). There are more similarities in the ways Quine and Wittgenstein approach questions of meaning and understanding than in the answers they offer to such questions. For instance, Quine’s criticism of ostensive learning as a means for settling the reference of words occurring in observation sentences leads him to formulate the doctrine of the indeterminacy of reference. No such doctrine is to be found in Wittgenstein’s Philosophical Investigations, for neither the notion of truth nor of reference plays a central role in his account of language use. Cf. Eva Picardi, “Davidson and Quine on Observation Sentences,” in Language, Mind and Epistemology: On Donald Davidson’s Philosophy, ed. Gerhard Preyer et al. (Dordrecht: Kluwer Academic, 1994); Eva Picardi, “Empathy and Charity,” in Quine: Naturalized Epistemology, Perceptual Knowledge and Ontology, ed. Lieven Decock, and Leon Horsten (Amsterdam: Rodopi, 2000). See also the papers collected in Wittgenstein and Quine, ed. Robert L. Arrington and Hans-Johann Glock (London: Routledge, 1996). Noam Chomsky, “Language and Nature,” Mind: A Quarterly Review of Philosophy 104 (1995): 1–62. PI §560. It goes without saying that I do not read Wittgenstein’s dictum “Roughly: understanding a sentence means understanding a language” (BB, p. 5) as a manifesto of the doctrine of semantic holism and its corollaries, but as an injunction not to lose sight of the system of signs to which a given word belongs and to refrain from supposing that what infuses life into a sentence is “something in an occult sphere, accompanying the sentence. But whatever accompanied it would for us just be another sign” (BB, p. 5). All the same, there are important similarities between Quine and Wittgenstein, to which I will call attention in the next sections. PI §20. PI §206.

Concepts and primitive language-games

133

31 PI §7. 32 H. P. Grice, Studies in the Way of Words (Cambridge, MA: Harvard University Press, 1989). 33 Augustine’s seemingly naïve account is a mixture of powerful philosophical assumptions and of very acute observations, concerning the non-linguistic means of expression (bodily movements, the tone of voice, the direction of the gaze, etc.) on which we rely in communicating with others. Both in the Philosophical Investigations and in his later works Wittgenstein will follow Augustine in emphasizing the role of non-linguistic means of expression in mutual understanding. 34 Norman Malcolm, Wittgensteinian Themes: Essays, 1978–1989 (Ithaca, NY: Cornell University Press, 1995). 35 Z §541. 36 Brandom, Articulating Reasons, pp. 14–15. 37 Cf. Rhees, “Wittgenstein’s Builders.” 38 Wittgenstein himself in the Brown Book (BB, p. 90) mentions the retrieving game, a game in which a dog, but not a cat, can be trained. He also forestalls the objection that Builder B seems to act automatically on the shouts of Builder A. But he nonetheless insists on the legitimacy of taking the Builders’ language-game as a starting point of a philosophical inquiry into the working of our language. He then goes on to point out that it is far from obvious what the criteria are for saying that someone acted automatically or after conscious reflection. And if we complicate the examples, the distinction between automatic and non-automatic appears no longer clear-cut (BB, pp. 87–88). 39 I have here helped myself to Quine’s account of the ontogenesis of reference, offered in Word and Object, ch. 3. 40 E.g., Z 110. 41 Z §§93–146. 42 Z §93. 43 Schulte, “The Builders’ Language—The Opening Sections.” 44 Warren Goldfarb, “I Want You to Bring Me a Slab. Remarks on the Opening Sections of the Philosophical Investigations,” Synthese 56 (1983): 271. 45 Cf. Ludwig Wittgenstein, Philosophische Untersuchungen: Kritisch-Genetische Edition (Frankfurt am Main: Suhrkamp, 2001), p. 63. 46 Cf. In §135 of Philosophical Investigations Wittgenstein points out that the answers to questions relating to what we mean by “proposition” and what we mean by “game” should proceed along the same pattern, and draws a comparison between the concept of proposition and the concept of number. 47 PG, p. 112. 48 Robert Fogelin, Wittgenstein (London: Routledge and Kegan Paul, 1976), pp. 118–19. 49 E.g., PI §§8, 41–42, 48. 50 Cf. Juliet Floyd, “Wittgenstein, Mathematics and Philosophy,” in The New Wittgenstein, ed. Rupert Read and Alice Crary (London: Routledge, 2000). 51 Max Black, “More about Metaphor,” in Metaphor and Thought, ed. Andrew Ortony (Cambridge: Cambridge University Press, 1979), p. 23. 52 A. N. Prior, “The Runabout Inference-Ticket,” Analysis 21:2 (1960): 38–39. 53 What Wittgenstein says about contradiction in the Lectures on the Foundations of Mathematics shows that he is not a meaning holist concerning logic. For he suggests that we take local remedies to cope with the contradiction, Ludwig Wittgenstein, Wittgenstein’s Lectures on the Foundations of Mathematics, Cambridge, 1939: From the Notes of R. G. Bosanquet, Norman Malcolm, Rush Rhees and Yorick Smythies (Hassocks, West Sussex: Harvester Press, 1976), especially lectures xxxi–xxxiii. We may reconsider, for instance, the meaning (i.e., the rules) of negation. The presence of a contradiction does not infect the language or

134

Eva Picardi

calculus in its entirety. To be sure, not everybody is happy with this solution, for often we have no clue as to what is responsible for the contradiction. My paradigm of a radical meaning holist is Quine, Word and Object. An advocate of second strategy is Frege, e.g. “Logik in der Mathematik,” in Nachgelassene Schriften, ed. Hans Hermes et al. (Hamburg: Meiner, 1969). 54 For a discussion of the notion of a conservative extension in connection with meaning holism see M. A. E. Dummett, The Logical Basis of Metaphysics (London: Duckworth, 1991), chs. 9 and 10. The notion of a conservative extension is, in its turn, employed by Dummett to characterize the notion of harmony. The harmony in question concerns the two main aspects of our use of a sentence, i.e., the grounds for justifying its assertion and the consequences that follow from our acceptance of it. More precisely, “When an expression, including a logical constant, is introduced into the language, the rules for its use should determine its meaning, but its introduction should not be allowed to affect the meanings of the sentences already in the language. If, by its means, it becomes possible for the first time to derive certain such sentences from other such sentences, then either their meanings have changed, or those meanings were not, after all, fully determined by the use made of them. In either case, it will not be true that such a derivation demonstrates that the conclusion holds good according to previously acknowledged criteria. The introduction of the new constant has created new criteria for the truth of statements not containing it” (ibid., p. 220). 55 David Lewis, Convention: A Philosophical Study (Cambridge, MA: Harvard University Press, 1969). 56 Cf. RPP I, §592.

6

On the teaching and learning of words Michael Luntley

6.1 In a number of central and well-known passages in the Philosophical Investigations Wittgenstein describes situations in which a pupil is instructed in the meaning of words. The nature of the transaction between teacher and pupil in these cases is often thought to be problematic. On a familiar way of reading these passages, the teacher struggles to convey to the pupil what a word means and finds that their expression of what they know is subject to a radical skepticism that underdetermines the content of what they are trying to convey.1 On this reading, there is potential for an unbridgeable gap between what the teacher is able to say when explaining a word’s meaning and what the pupil needs to learn. The existence of this gap undermines our ordinary notion of word meaning. It presents a skeptical challenge to our ordinary understanding of meaning. The standard way of avoiding the skeptical challenge is to close the gap, to show that it can be bridged by reconstruing the teacher/pupil transaction. Kripke’s re-construal consists in deflating the account of knowledge that is transferred.2 A better strategy is to say that in place of a transaction in which the teacher transfers knowledge to the pupil, we are invited to construe the teacher/pupil relationship as one in which the teacher initiates the pupil into a practice.3 This is, perhaps, the favored response. If we construe the teacher/pupil relationship in this way, we avoid the slack that otherwise exists between the knowledge the teacher is able to express and the information that the pupil needs to acquire in learning the meaning of a word. Such, in outline, is a familiar way of reading the teacher/pupil relationship as it figures in Wittgenstein’s discussion of the concept game from §66 onwards and in the discussion of teaching someone the meaning of “add 2” from §185 onwards. In this essay, I shall argue that this familiar interpretation of the teacher/pupil relationship is wrong. The familiar interpretation assumes an empiricist model of learning that, although clearly suggested in many passages in Wittgenstein’s later writings, is incompatible with a rationalist model of learning that makes much better sense of the key passages. To argue, as I shall, that Wittgenstein has a rationalist model of learning appears to fly

136

Michael Luntley

in the face of the numerous places where he emphasizes the role of training in language learning, where “training” is something that predates ratiocination. I need, therefore, to provide some account of how the frequent references to training are to work in a broadly rationalist model of learning. I shall argue that, properly understood, the accounts of training are not only compatible with my preferred rationalist reading, but provide the starting point for a serious detailed account of the pupil’s mental activity implied by the rationalist model. In short, despite initial appearances, the reading I provide does not take issue with what Wittgenstein actually says in his later writings. I take issue only with how he is often read. On my reading, Wittgenstein’s writings have a substantial contribution to make to contemporary debates about rationalism and nativism about concepts in the philosophy of mind. I start in the next section with an overview of the distinction between an empiricist and rationalist model of learning. I offer this as an analytical tool for framing a reading of Wittgenstein that throws into relief a number of central passages and issues in his later work.

6.2 A rationalist model of learning is one in which learning experiences work, for the main part, in virtue of the reasoning abilities of the learner. In the case of learning the meaning of words, a rationalist account of learning experiences is one in which the learner “figures out what other people are thinking when they use words.”4 Quite what goes into the notion of “figuring out what other people are thinking,” let alone what other people are doing, or how the world is working, etc., is a matter for debate and scientific investigation. Fodor sees this as a case of hypothesis formation and testing.5 Many developmental psychologists hold similar views.6 The key idea to a rationalist account of learning is that learning happens when the learner brings reasoning skills to the learning experience. Consider Bloom again responding to Quine’s argument for the indeterminacy of translation: “the naturalness of the rabbit hypothesis and the madness of the alternatives is not a logical necessity; it is instead the result of how the human mind works.”7 The key point is that the learner’s mind has work to do to bring about learning and the work undertaken is reasoning. For the rationalist, the learner is an active contributor to learning by virtue of the exercise of cognitive skills properly called reasoning. A central issue for a rationalist model of learning is the specification of the reasoning skills the learner deploys to achieve learning. In the case of early language learning, if the reasoning skills deployed are those requiring concept possession, then the rationalist will typically be thought to be faced with the following dilemma: either the account of language learning begs the question by modeling first language learning on second language learning, or they have to endorse the innateness of primitive concepts.8 Wittgenstein is often thought to deploy such a challenge in his critique of Augustine.

On the teaching and learning of words

137

An empiricist model of learning is one that prioritizes training as the means by which the pupil acquires the basic materials for understanding. By “training” I mean an account of the learning experience in which, as it were, the working element concerns what is provided to the learner in experience, rather than what the learner brings to the experience. In this respect, an empiricist model of learning has a passive role for the learner. What does the work is the input provided by experience and the learner’s non-reasoning responses to the way experience impinges on them. The normal account for an empiricist model of learning is to see learning experiences as shaped by associationist principles and imitation. The point about associationist principles is that the connection between word and object, for example, is not based on reasoning or reason-sensitive properties, but on reinforcement by frequency of pairings of word and object. And imitation is not a reasoned attempt to reproduce an action, but a simple replication. An empiricist model requires that the learner have some set of receptivities, but these are receptivities that produce mimesis and dispositions to respond to significant frequencies of pairings of elements in experience. In order to keep any element of reasoning skills by the learner out of the picture, an empiricist model will need to have some account of “significant frequencies of word/ object pairings.” One way of taking Quine’s argument about the indeterminacy of translation is as a reductio ad absurdum of the idea that there is any account available of what constitutes significant frequencies of word/object pairings. At the heart of the empiricist model of early language learning then is the notion of learning via acquisition of dispositions to respond to significant pairings of word and object, including those responses in which the learner produces the word in the object’s presence. This highlights two key problems for the empiricist. First, there is the indeterminacy problem – how to screen for pairings that are significant? Second, even if the indeterminacy problem is solved, there is a further problem concerning the conceptual poverty of the model. This is the disposition problem. The disposition problem is that a disposition to utter a word in the presence of an object or feature falls short of using a word to represent an object or feature. Using a word to represent requires understanding and that is a normative notion that involves knowing the circumstances in which it is correct to use the word. Knowing the correctness conditions for word use is qualitatively different to having dispositions to use a word. In summary, the empiricist seems to have very little to go on to get language learning off the ground. Quine’s work either shows how little the empiricist can reconstruct, or it’s a reductio of the empiricist enterprise. In contrast, the rationalist seems to have too much to make the account of language learning genuinely explanatory. In the case of early language learning the choice between an empiricist and rationalist model of learning comes to the question: Which comes first, training or reasoning? It can seem obvious that one should endorse the

138

Michael Luntley

empiricist model. Wittgenstein certainly seems to endorse it: “A child uses [ … ] primitive forms of language [ … ] when it learns to talk. Here the teaching of language is not explanation, but training.”9 Also, “the child, I should like to say, learns to react in such and such a way; and in so reacting it doesn’t so far know anything. Knowing only begins at a later level.”10 And, “I want to regard man here as an animal; as a primitive being to which one grants instinct but not ratiocination [ … ]. Language did not emerge from some kind of ratiocination.”11 And, The child learns this language from the grown-ups by being trained to its use. I am using the word “trained” in a way strictly analogous to that in which we talk of an animal being trained to do certain things. It is done by means of example, reward, punishment and suchlike. (BB, p. 77)12 One of the issues that need to be considered concerns the scope of behaviors that fall under the heading “training.” In the passage from the Blue and Brown Books Wittgenstein uses “training” in a very narrow sense, for it applies also to the habituated routines in animal training. I shall adopt this narrow concept of training in which it involves habituation of responses to things in which the subject acquires a disposition to take, e.g., frequencies of word/object pairings as significant. Wittgenstein uses the word in richer ways too. It is beyond the scope of the present essay to provide a detailed map of all these uses, but we will not understand the significance of richer notions of training unless we mark clearly how they differ from the narrow conception. To do that, we need to understand the limitations of the narrow concept, so that is where I start. Suppose these passages are endorsements of an empiricist model of learning as training. There are three broad points that underpin the intuitive appeal of an empiricist model of language learning. First, an empiricist model provides a simple developmental trajectory for the individual learner. This trajectory has intuitive appeal, for it offers the prospect of learning getting off the ground with minimal resources. The sophisticated resources of conceptual understanding that are deployed in more mature learning encounters are, on the empiricist model, the products of the early stages in this trajectory.13 Second, the empiricist prioritizing of training provides an account of the passivity of cultural immersion when the learner is acculturated into the habits of the tribe. It makes sense of an idea that, once again, appears intuitively compelling: before one can rationally examine the ways of the community, one needs to be initiated into them. Third, and most important of all, a rationalist account of early learning is often thought to carry a commitment to nativism about concepts and that can only be avoided by the empiricist model. The case for avoiding nativism is the most compelling reason in support of an empiricist model of early learning. Modeling early language learning as a rationalist working out that a word stands for a certain kind of thing

On the teaching and learning of words

139

presupposes that the child already has the capacity for distinguishing the kind of thing in question. And that runs the risk of modeling early language learning on second language learning. The argument for this has the following form. Assume that working out that a word stands for Fs can only be undertaken if the subject already has the capacity for singling out Fs. The assumption that the subject has the capacity for singling out Fs is required in order to solve the indeterminacy problem. If you then assume that you cannot single out Fs without possessing the concept F, this requires the child have the capacity for conceptually singling out Fs prior to learning that the label for things of that kind is “F.” It would then follow that a rationalist model of learning in which the child has to deploy mental capacities for actively working out what a word means presupposes that early language learning is just like second language learning. The key assumption is the assumption that singling out Fs requires possession of the concept F. Call this the Fregean assumption, it is the idea that concepts are the way things are presented to us in experience. Given the Fregean assumption, the child could not work out that “F” stands for F-ness if they did not already possess the conceptual space for F. And if they have a conceptual repertoire with a space for the concept F, then all they really lack is the label for the concept. And that is the model of second language learning. Wittgenstein’s criticism of Augustine seems of a piece with this response to rationalism. I shall argue, however, that the key Fregean assumption is false and there is no reason for thinking that Wittgenstein made it. Now, having set up the distinction between empiricist and rationalist accounts of learning, we rapidly generate a dilemma. If you take seriously the idea of training in the empiricist model, in which training is something prior to ratiocination, then training can amount to no more than acquiring dispositions to respond to F-ness with the word for F-ness. But acquiring a capacity to respond to F-ness in this way falls short of representing F-ness. You can acquire a capacity to respond to Fs, even a response that involves saying “F,” regardless of whether or not you understand anything about Fs. This is because dispositions are grounded in nomologically salient features – stuff that is causally salient, but in that case the significance of word/object pairings is nomological, not normative. Training cannot prima facie be an account of language learning, however much it might contribute to such an account. On the other hand, a model that accounts for the achievement of linguistic mastery in terms of the child’s capacities for working out for herself what words mean appears to beg the question. It requires that the child has a rich set of cognitive capacities and it is a challenge to characterize that set without assuming something that too closely approximates the idea that the child had the appropriate concepts at the outset and needs only to learn how to label those concepts. I think this simple dilemma needs to be avoided. The normal strategy in readings of Wittgenstein avoids the dilemma by replacing the concept of training with a richer notion of immersion in a

140

Michael Luntley

practice. This is, in effect, to endorse a modified empiricism about learning. I think that misconstrues the depth of the problems facing an empiricist model of learning. In place of that strategy, I want to suggest that a better response is to provide a modified rationalism. What I want to do is show how a rationalist model makes best sense of Wittgenstein’s text. Furthermore, the way it makes sense of his text means that Wittgenstein has a significant contribution to make to ongoing debates about rationalism in the philosophy of psychology.14 A more sophisticated model is required if we are to make sense of early word learning, but the key to getting a model that works (and that makes sense of Wittgenstein) is to find space for the central component of the rationalist model. This is the conception of the learner as an agent, a subject with a capacity for mental activity, the activity of working out what to think and do with words. Furthermore, the account of that activity has to avoid the Fregean assumption that leads to nativism about concepts. In the next section I briefly sketch what I take to be a common reading of Wittgenstein on the teaching and learning of words.

6.3 Consider the following two examples involving the teacher/pupil relationship in the Philosophical Investigations: the teaching of the word “game” and the teaching of arithmetical series, especially the rule “add 2.”15 In both cases the teacher/pupil relationship is problematic, for it threatens to undermine our ordinary concept of meaning. Our ordinary concept of meaning is undermined because of a concern that what the teacher is able to articulate in explaining the meaning of a word is insufficient to constrain the pupil in a manner thought appropriate for grasp of meaning. It seems plausible to suppose that, in both cases, we start with the assumption that the teacher knows what the words mean. If knowing what the word means is a matter of grasp of theoretical knowledge, in each case the teacher is unable to explicitly articulate this knowledge. If we construe the teacher/pupil transactions in terms of a transfer of explicit instruction, the transaction fails, for it leaves the pupil unconstrained. They would have to guess or make a leap to grasp the knowledge that has been given only a fragmentary articulation by the teacher. In short, if we think of the teacher/pupil transaction as a transfer of explicit instruction,there would seem to be a gap between what the teacher says and what the pupil has to learn. The notion of a gap appears to flow from what I’ll call the negative thesis: it is not possible to teach early language meaning by explicit articulation. In the “game” example, the gap between what the teacher says and what the pupil has to learn is due to the lack of explicit criteria for the application of “game.” The teacher gives examples and then can say no more than, “this and similar things are called games.” This marks the point where the pupil has to “catch on” and appreciate the “and so on” that seems to apply in all

On the teaching and learning of words

141

cases. The point is more marked in the later examples with arithmetic series. In these cases, any attempt at articulating theoretical knowledge of meaning will fall short of the infinitary nature of meaning’s rules. The pupil can only be given a fragment of the appropriate series and the rest is covered by the “and so on.” The idea of the gap between what the teacher says and what the pupil has to learn is central to this common reading of Wittgenstein. The idea of a gap is a variation on the indeterminacy problem – How are we to tell what the significant pairing is when confronted with a football match and the word “game”? So, if we suppose that the teacher knows the meaning of “game” or of “add 2,” there is a gap between what they say and what the pupil has to learn. Given that in both examples the teacher has to resort to “and so on” in explaining meaning, then the “and so on” can only provide an adequate articulation of the teacher’s knowledge of meaning if the pupil is able to employ it correctly. But given that the “and so on” in itself does not specify how the examples given are to be continued, then the pupil can only employ the “‘and so on” correctly if they have already grasped the meaning of the words at issue. In other words, there has to be a gap between what the teacher says and the pupil has to learn on pain of accepting that the pupil can work out for themselves what is meant by “and so on,” something they could only do if they already knew the concept at issue. That means that without begging the question, there must be a gap and it must be unbridgeable by reasoning on the pupil’s behalf on pain of begging the question about the teaching and learning of words. So, from one conception of a rationalist perspective on learning (call it codified rationalism) in which the pupil works out what to do by employing theoretically codifiable concepts, what is said by the teacher falls short of what the pupil has to learn and, unless we are willing to ascribe to the pupil the very conceptual mastery the process is meant to teach them, it would seem that there is no reasoning by which the pupil can close the gap. The existence of the gap then undermines the obvious way of construing a rationalist model of learning (codified rationalism) in which the teacher/ pupil transaction is construed as a transfer of explicit instruction.16 The idea of the gap also undermines the empiricist model of learning as training. So long as training is narrowly conceived in terms of acquisition of dispositions to respond to significant pairings of word and object, then no amount of training will succeed in producing a normatively configured pattern of word use. If training is understood as Wittgenstein often uses the word – as something that precedes ratiocination and that provides primitive responses not yet of the level of knowledge – then training can be no more than a habituation in dispositions to respond to significant pairings. If there is one thing that is now clear in readings of Wittgenstein, it is that a dispositional analysis of word use falls short of the notion of grasp of correct word use. Training, as conceived on the empiricist model of learning, falls short of provision by the teacher of the appropriate resources from which the

142

Michael Luntley

pupil can obtain an understanding of correct use. It is this point that turns the dialogic problem of the teacher/pupil transaction into the radical skepticism of the single person case in which we realize that our past use of a word, construed as the basis for training, leaves present and future use indeterminate with respect to patterns of correct use. And once the idea of patterns of correct use is indeterminate, there is nothing left to the idea that the teacher (or our earlier self) really knew what they meant with “add 2.” The familiar skepticism about meaning that is found in Wittgenstein’s rule-following arguments can then be represented as one horn of the dilemma between a rationalist and empiricist model of learning. The rationalist model of learning, in which the pupil learns by working out for herself what a word means, begs the question by equipping the pupil with just the sorts of cognitive capacities with which learning is meant to equip them. The empiricist model of learning, in terms of training, is then the default choice, but that threatens the intuitive normativity to meaning and the idea of rules of correct use. Both horns of the dilemma are susceptible to modification in providing an acceptable model of learning. It is, however, worth noting how the simply drawn choice in the dilemma throws the options for modification into focus. A modification of a rationalist model of learning will be a model that ascribes to the pupil some cognitive capacities that are actively involved in their learning, but where those capacities fall short of full-blown conceptual capacities. The critical move here is to deny the Fregean assumption that the ability to discriminate Fs requires possession of the concept F. I shall indicate my preferred rationalist modification below. What makes it a rationalist model is the role the notion of mental agency plays in the account of learning; the pupil’s capacity for deploying cognitive skills is a key ingredient to language learning. The normal modification in readings of Wittgenstein is to modify the empiricist model by introducing a richer concept of training. The simple dilemma places an obvious constraint on what can count as a modified empiricism: if the concept of training is modified to cover instances of reasoning, the model has turned into a rationalist model and, without a clear account of what constitutes reasoning, is in danger of begging the question just as the codified rationalist does. My suspicion is that the common reading of Wittgenstein produces just such a confused modified empiricism. The common reading of Wittgenstein avoids our simple dilemma by treating training as “immersion in a practice” where a practice is a normatively configured pattern of behavior with intrinsic standards of correctness. I think this is a fudge. It amounts to endorsing our common sense account of the teaching of words – you give examples and the pupil catches on – with no account of our entitlement to this common sense view. It uses the word “training” for that activity which works: training pulls it off, but no account is offered of how or why. Simple training in the empiricist model of acquisition of dispositions to respond is a clear concept, but to elide that into training as immersion in practice is, arguably, to merely add to the idea of

On the teaching and learning of words

143

simple training the very sort of reasoning capacities in using words that the whole learning encounter is meant to produce. The dilemma was that the negative thesis entailed a gap between what the teacher says and what the pupil has to learn from the transaction. The appeal to practice looks to offer a distinct kind of activity that is supposed to close down the gap. But the danger is that only someone already equipped with the sorts of reasoning skills that come from acquiring concepts could take part in the activity offered to close down the gap. One alternative for the empiricist would be to define sophisticated forms of receptivity, including receptivities to others and social behavior, that would help bridge the gap. It is not clear such a move is not a case of putting the cart before the horse by building into the receptivities required the very conceptual understanding that is supposedly acquired by learning. In the absence of a detailed working out of such a strategy that avoided the obvious circularities, the move to take immersion in practice as the activity that closes the gap can only result in a familiar quietism: learning takes place only for those subjects apt to be initiated into practices. Who are these subjects? They are those who catch on!17

6.4 There is a naïve response to the simple dilemma. Accept the negative thesis is true; that is to say, accept the rejection of a codified rationalism which models learning as a transfer of theoretical knowledge of what the teacher means when introducing a word. Why accept that acceptance of the negative thesis entails the gap, the idea that what the teacher says falls short of what the pupil needs to learn? What stops us from saying that what the teacher says is exactly what they know with regard to the meaning of the word and what the pupil needs to know to learn to understand the meaning of the word? In short, why accept that there is a problem with the teacher/pupil relationship at all? I think this naïve response is the one Wittgenstein intends us to take. Making proper sense of the naïve response requires a modified rationalism. It requires that we acknowledge the cognitive skills – the mental agency – that the pupil brings to the encounter. It requires what I shall call a stratified rationalism, for it is a rationalism that accommodates levels of activity that contribute to how we make sense of ourselves where some of those levels fall short of conceptually articulated activity.18 My aim now is to bring this stratified rationalism into focus and to argue that it provides the best reading of Wittgenstein’s own understanding of the teacher/pupil relationship in early language learning. We can approach the idea of stratified rationalism in the following way. If the negative thesis is correct, then the sort of knowledge at issue that the teacher imparts to the pupil in early language learning is not theoretical knowledge. We could say that is practical knowledge. That does not take us very far unless we have a clear conception of practical knowledge. It had

144

Michael Luntley

better involve more than simple training – acquisition of dispositions to respond, but it had also better be more than a label for whatever works. The concept needs to provide some explanatory leverage on the cognitive components of successful teaching and learning events. The critical issue is whether accepting the negative thesis that knowledge of meaning is not theoretical knowledge entails that there is a gap between what the teacher says and what the pupil has to come to know. If there is a gap, the naïve response is not available. If we can provide content to the notion of stratified rationalism by accepting cognitive capacities that fall short of full-blown conceptually articulated reasoning capacities, we can then begin to make sense of how to endorse the naïve response. What is central to this is making sense of the idea of doing something for a reason, the placing and seeing of ourselves as acting in the space of reasons. This is seeing our actions as things that stand in chains of justification as well as explanation. After a successful teaching/learning transaction, the pupil acquires reasons for correct use of the word. They learn how to use the word “game” with reason. The resulting language use must involve a use of words done for reasons, a use that helps render ourselves intelligible by showing how our behavior is susceptible to justification. The concept of doing something for a reason is central to how we make space for the naïve response. I suggest that doing something for a reason can involve a repertoire of cognitive skills, including those that make use of language and inference, that make up the components of how we render ourselves intelligible. The basic notion of doing something for a reason is acting so as to render oneself intelligible. And there can be more to the notion of making oneself intelligible than acting in a way that is representable as the conclusion of a practical syllogism. There is no reason to suppose that because a form of learning is not representable in terms of the acquisition of a body of theoretical knowledge, it must thereby be treated as a training that does not centrally involve the operation of reasoning, the mental activity of working out what to think. Of course, if your only model of reasoning is that of an activity that can be represented as something that can be regimented as following from conceptually articulated premises in a deductive inference, then reasoning could not figure in language learning without begging the question. That was the problem with codified rationalism. Equally, with such a restrictive sense of doing something for a reason, the absence of an explicit theoretical representation of what you learn when you learn the meaning of “add 2” would entail that you could not, in the restricted sense at issue, go on and act for a reason as you use the phrase in manifesting your understanding. Similarly, if what you know when you know the meaning of “game” is not representable as a body of theoretical knowledge, then there is no body of information that could form the basis of an explicit inference to provide a reason for your applying the word to a new case. But all that this shows is that it is a bad idea to work with such a limited conception of what it is to do something for

On the teaching and learning of words

145

a reason. The negative thesis shows that there are no explicit reasons available such that the pupil’s learning can be treated as acquiring reasons construed as premises of inferences for carrying on with “add 2” or “game.” But this lack of explicit reasons is only a problem if you restrict the notion of doing something for a reason to something that can be regimented into deductive inferences. And why do that? In contrast, if doing something for a reason is, however else one might characterize it, at least a matter of doing something so as to render oneself intelligible, then it is a doing something that must require some sensitivity to rational control and evaluation.19 It will be minimally a doing of things, including the use of language, which is directed and purposeful. Such a doing of things requires a repertoire of cognitive capacities for making sense of activities, capacities for organizing and directing activities so that they make sense. Organizing and directing activities need not be a matter of having a codified script in advance of the doing from which the doing is directed. It need not, therefore, require a representation of the directedness that has to be first articulated and formulated in an explicitly codified way in order that it might then direct behavior.20 The point ought to be obvious, but it is typically missed. Getting this point right lies at the root of avoiding the charge of a simple nativism regarding capacities for reasoning. The chief worry with codified rationalism and the thing that motivates the search for a credible empiricism about learning is the threat of nativism about concepts. Nativism requires that the pupil already have a conceptual repertoire with a space for the concept F in order to be able to be taught the concept F. But what I am calling stratified rationalism requires only that the pupil has capacities to direct her behavior with respect to things. The capacity to direct behavior with respect to Fs does not require a language that codes for F, nor does it require the concept F. It requires a capacity to find Fs salient plus a general capacity to attend to saliences. These are more basic capacities than conceptual capacities. They are capacities that fill out a conception of the pupil as a mental agent, someone with capacities that are reasoning capacities, for they are capacities that contribute to the project of rendering oneself intelligible. They are, however, capacities that fall short of conceptually structured reasoning.

6.5 Despite the industry that has grown up around a reading of Wittgenstein as posing a radical skepticism about meaning, I want to suggest that Wittgenstein holds that what we say when we teach someone the meaning of a word is a perfectly good explanation of meaning. What the teacher says does not short-change the pupil, leaving a gap that has to be bridged. What the teacher says accurately gives voice to what the pupil knows about the meaning of the word. If the pupil has the appropriate capacities as a mental actor, s/he is given everything s/he needs in order to understand the meaning

146

Michael Luntley

of the word “game” or the instruction “add 2.” Consider the detail of what Wittgenstein says in the former example.21 Having raised the issue about the meaning of the word game in §66, Wittgenstein continues: How should we explain to someone what a game is? I imagine that we should describe games to him, and we might add: “This and similar things are called ‘games’.” And do we know any more about it ourselves? Is it only other people whom we cannot tell exactly what a game is? – But this is not ignorance. (PI §69) Note two points about this passage. First, there is the clear acceptance of the limitation of a theoretical codification of games. We can give examples, but then we add “This and similar things are called ‘games’.” It might seem that this is an indication of the gap between what the teacher says and what the pupil has to learn, for without an explicit formulation of what constitutes the relevant similarity, this remark might be thought to give little or no guidance to the pupil. If you think that there is a gap here, then it can seem natural to suppose that the way to close it down is to invoke the notion of training as immersion in a practice. The relevant practice might be conceived as the rich filigree of uses of the word referred to by the idea of a family resemblance of uses. That, I think, is the fudge. Without enormous detail on the relevant repertoire of receptivities and how such repertoires could be brought about by training, this is the move that refuses to say how learning is achieved but simply says that we get by. There is, however, no case for thinking that the inevitable appeal to the “this and similar things … ” construction indicates a gap between what the teacher knows and the pupil has to learn. The precise wording of the passage is not totally transparent, but I take the last three sentences – two questions and an assertion – as clear indication that Wittgenstein thinks there is no gap. The first question, “And do we know any more about it ourselves?” raises the issue whether the teacher and other knowledgeable users of the word know any more than this. That is, do knowledgeable users know any more than a few examples plus the gesture to some undefined notion of similarity? I take from this the idea that if that is all they know, then what they know falls short of what we took knowledge of the meaning of the word to consist in. It would be a kind of ignorance. The second question then raises a different point. The first question asked whether we know any more than this, the second asks a contrasting question: perhaps we know but cannot tell others what we know. The idea behind this question is, I take it, that the poverty of the teacher’s articulation leaves the others in ignorance, for we have failed to make available to them what we know. The final assertion is, I take it, the denial of both these forms of ignorance. The reference of the “this” could be to either type of ignorance – the ignorance of the knowledgeable users raised by the first question or the

On the teaching and learning of words

147

ignorance of the others raised by the second question. I think the final sentence is saying, in effect, there is no case for ignorance of either kind. The idea that the teacher/pupil relationship is plagued by ignorance of either type, whether of teacher or pupil, is simply mistaken. The point is, perhaps, even more general. It is also natural to read the last sentence as referring back to what the teacher has made manifest in what they said. They have given some examples and then said, “this and similar things are called ‘games’.” If you think that this indicates a gap, then saying this is to say less than what the teacher knows. It is to offer less than the knowledge that gives them mastery of the word. It is not to offer the pupil knowledge, but to short-change the pupil and to offer them, if you like, ignorance, something less than what they need. But doing this is not to offer them less than what they need. Perhaps it is to provide them with precisely what they need to understand games? Further evidence for reading §69 in this manner comes two sections later. The lack of explicit criteria for games means that there is an openendedness to the concept. This open-endedness prompts the metaphor of family resemblance. One way of treating this is to think that concepts have blurred boundaries; the open-endedness is due to the roughness of the boundary. Again, this would suggest a form of ignorance, as if we did not know enough to render our concepts sharp. In §71 Wittgenstein raises the question whether concepts have blurred boundaries and in raising the rhetorical question, “Isn’t an indistinct one [picture] often exactly what we need?” suggests that there is nothing wrong with the lack of precisely articulated boundaries. But that suggestion is not because rough boundaries and the lack of precise knowledge is enough to get by; the suggestion works because he has a quite different conception of the sort of knowledge that is at stake in the teacher/pupil transaction. If the transaction were a transfer of knowledge, then any looseness of articulation would mean the transfer was less than complete. The point would apply to training as immersion in a practice too: any immersion would be less than complete, so if the transaction were a transfer of a practical mastery in an activity, what is on offer would still be less than a complete practical mastery. Either way, the transfer would be incomplete and there would be a gap between what the teacher has on offer and what the pupil has to acquire. But that is not Wittgenstein’s model. After raising the issue about imprecise boundaries Wittgenstein says: this is just how one might explain to someone what a game is. One gives examples and intends them to be taken in a particular way. – I do not, however, mean by this that he is supposed to see in those examples that common thing which I – for some reason – was unable to express; but that he is now to employ those examples in a particular way. Here giving examples is not an indirect means of explaining – in default of a better. (PI §71)

148

Michael Luntley

There are three points I want to emphasize in this passage. First and most obviously, Wittgenstein confirms the view that the idea that the teacher intends the pupil to take examples in a certain way should not be confused with the idea that the pupil is supposed to see the common essence of games that the teacher was unable to articulate. There is no common essence in the sense of something that could in principle be articulated, or at least seen, and that informs further usage. Second, the last sentence makes clear that the reading offered above of §69 is correct. Giving examples is not an indirect way of explaining the word. It is not a method of explaining that shortchanges the pupils and leaves them with less than they need. It is not a second best, something that we have to make do with in place of the full complete account of what is known. Wittgenstein is quite clear. What is on offer in this transaction is exactly what needs to be on offer. What the teacher says is what the teacher knows. And what the teacher knows is indicated in the third key point of this passage. What the teacher knows is how to employ the word in the right way. In contrast to the essentialist account of meaning in which the pupil has to fathom out the hidden commonality to games, what the pupil takes from the examples is knowing how to employ the examples in the particular way that the teacher did. Wittgenstein emphasizes the verb, the activity that the pupil gains from the teacher. The key question is what activity is signaled by knowing how to employ the word in a particular way? I think it is clear that the idea of knowing how to employ the word cannot be modeled either as theoretical knowledge or as practical knowledge understood in terms of immersion in a practice. Both types of knowledge are incomplete in terms of what is on display by the teacher. The teacher cannot articulate a complete theoretical account of games or display a complete practical mastery of how the word is used. So the notion of employing the examples in a particular way can be neither of these things. It is, however, an activity on behalf of both the teacher and the pupil. It is also, in contrast to either knowing a theory or knowing how to act (in the sense of training or practice immersion) something that is essentially open-ended. If it were not essentially open-ended, the activity would not be shareable in what otherwise seem fragmentary exchanges. Knowledge of meaning is on offer in the teacher/pupil transaction. It cannot be knowledge of theory (codified rationalism), for there is no theory of games. It cannot be a knowing how to act in either the sense of training in dispositions to respond to word/object pairings or immersion in practice. The former fails to deliver the normativity to meaning, the latter would leave the transaction indirect and what is on offer by the teacher incomplete. The simplest way of expressing the knowledge on offer is to say that it is a knowing how to render self and others intelligible, in the case at hand by deploying the word “game.” It is a knowing how to organize and direct one’s behavior so as to render it intelligible to self and others and to see others’ behavior as similarly organized. Knowing how to render self and

On the teaching and learning of words

149

others intelligible just is to find one’s behavior and that of others as reasonable, as done for a reason. But unless one assumes that something done for a reason can only be a doing that can be regimented as the conclusion of a syllogism, so that reasons are always explicitly articulable premises of inferences, then rendering oneself intelligible is essentially open-ended. The notion of doing something for a reason is just the notion of a doing that can be rendered intelligible. And what counts as an instance of that depends on what one can do in the activity of rendering oneself intelligible. One renders oneself intelligible not by fixing one’s behavior to a script or recipe for intelligible behavior, but in virtue of basic capacities for mental agency, including the capacity to direct one’s behavior with respect to things. This suggests a radical revision of the options for conceiving of the teacher/pupil transaction. The aim of the teacher in my model of the teacher/pupil transaction is not to convey information, e.g., theory. Neither is it the aim of the teacher to convey behavior, a practice or a simple set of dispositions to respond. In both these cases, what is on offer would be less than what the pupil needs to learn. The pupil would be short-changed; there would be a gap. The aim of the teacher is, I think, to get the pupils to join in the activity of making themselves and others intelligible. The aim is to get them to join in the activity of employing words so as to make themselves and others intelligible. But that is an open-ended project, for there is no telling what any of us might do and what we might need to do and say in response in order to make sense of it. The aim of the teacher is not to transfer a something or other, some form of either theoretical knowledge or behavioral pattern. The aim is to invite the pupil to join in a common enterprise of understanding. If this is right, then what we say when explaining the meaning of words is, as Wittgenstein allows, precisely enough if we assume that the pupil is a subject with a capacity for reasoning, a capacity for directing and organizing their behavior to make sense of themselves and others. If we assume that, then we say plenty when we say, “This and similar things are called ‘games’.” There is no more a puzzle about the gap between what the teacher says and what the pupil learns than there is a gap between what talkative adults say and do and what human infants learn from this. In both cases, what the pupil learns is to join in an activity of reasoning and, by so doing, to extend that activity.22 Wittgenstein endorses the negative thesis but not the idea of the gap between what is said and what is understood. Without that gap, there is no need for training or immersion in practices as social patterns of behavior. There is a need for activity, for agency. But the place of that activity is not to fill a gap by taking part in public behavior. It is to engage the mental activity of reasoning. The activity of reasoning here starts with the capacity to direct one’s behavior with respect to things. That is the basic mental activity that the pupil needs to engage in the learning of words on offer by the teacher. That capacity is less than the capacity for full-blown conceptual reasoning, but it

150

Michael Luntley

is nevertheless the basis for such reasoning. It requires that we conceive of the pupil as a subject qua mental actor. I call this a stratified rationalist conception of the subject, for it is a subject conceived without need for mastery of theory or concepts as such. It is a subject who is capable of more than training in dispositions to respond to word/object pairings and the extra is precisely that component that makes the model a rationalist model: the capacity to direct behavior. This is the Russellian core to a stratified rationalism, for the capacity to direct behavior requires the capacity for attending to things and properties, what Russell called acquaintance. This is the direct cognitive capacity to relate to things independent of and prior to thinking about them.23 It is this capacity that provides a model that challenges the Fregean assumption.

6.6 If this is right, we need to re-assess Wittgenstein’s critique of Augustine. For sure Wittgenstein objects to the overly intellectualist account that he finds in the passage quoted in §1. Augustine seems to conceive of the infant as a subject with a complete set of beliefs and desires that need only labels to bring these into the arena of public language. It is one thing to object to the sophistication of the picture of the infant mind that Augustine presents, but Wittgenstein nowhere objects to the core of Augustine’s model – the idea that the infant mind is an agent with a capacity for directedness. For sure, it would be more plausible to allow that in early days this capacity is blunt and crude, but that still requires an agent, albeit a clumsy and unfocused agent. The learning of words and of concepts might sharpen the directedness of the subject’s mental activity, but that does not get started at all if the pupil is not in the first place a subject with a capacity to direct its behavior. Learning provides the resources for tuning and focusing that activity, but learning only works with subjects with the capacity for that activity. If the above is correct, we need to mark a clear distinction between learning and training. Learning requires that the pupil comes equipped with capacities for mental agency. These involve, centrally, the capacity to direct their behavior with respect to saliences. That is the capacity for attention.24 Training is a more primitive concept. Training concerns the acquisition of dispositions to respond. There are plenty of places in Wittgenstein’s later work where such a notion of training seems to be at play, especially in contrast to a learning that involves a directed response rather than simply a dispositional response to things. We need then to consider what role training, conceived as acquisition of dispositions, plays within Wittgenstein’s overall account of learning. Training cannot fill the gap between what the teacher knows and what the pupil needs to learn as that figures in the empiricist model. But training serves some role in the teaching and learning of words. I think the obvious suggestion is to see the role of training as providing the background on which stratified rationalism operates. More precisely, the

On the teaching and learning of words

151

background provided by training consists of the patterns of habituation to things against which saliences become items of personal level experience. Let me sketch the sort of model that I have in mind for the role of training. Put briefly, the model that I want to outline, and that I think we can find in Wittgenstein, is a model that sees the role of training as laying the foundations for language learning, but not contributing to language learning per se. At its simplest, the model requires two sorts of capacities of the child: a capacity to find things salient plus a general capacity to direct behavior with respect to saliences. The capacity to find things salient is non-Fregean. There are two accounts of how this could be so. The discrimination of saliences might be done wholly sub-personally in terms of capacities for acting with respect to the affordances of things. But as such discriminations are not available within personal level experience, it is difficult to see how they contribute to learning conceived as something that the subject does, rather than something that happens to them. I suggest, therefore, that we restrict the notion of salience to things discriminated within personal experience. There is therefore, a substantive issue about what makes it the case that an item is salient within experience such it can then be a target on which the child can direct its behavior. My suggestion here is that training in the sense of habituation of dispositions to respond (which might take place wholly subpersonally) produces habits or expectations about the way the world works. Those expectations can be thwarted. When thwarted, they produce a primary case of a salience, for the thwarting of expectations typically registers within experience regardless of whether the subject has a concept for discriminating the thwarting. The second capacity is the capacity to direct behavior with respect to saliences. This is the key rationalist component of the model and is what makes the concept of a salience important in language learning. The capacity to direct behavior with respect to saliences does not require a conceptual capacity to discriminate saliences. Rather, this capacity picks out the key mental activity of attention that the child brings to learning experience. The capacity for attention is a personal level focusing on saliences so as to direct behavior with respect to that salience.25 On my model, this activity operates upon saliences revealed by the way that training in discriminating things generates habits or expectations about properties in the world that, when thwarted, produce saliences. So, in outline, the model is this: training provides dispositions to respond that generate expectations regarding things. Expectations are theoretical posits from the third person point of view of the theorist, they are not ingredients of the subject’s personal level point of view. Nevertheless, when one’s dispositions to respond are thwarted – expectations are not met – this produces a salience for the subject at the personal level point of view. So training lays down the background against which things (thwarted expectations) become salient. Once a subject has such saliences, the next ingredient concerns their mental capacities. If the subject has the general capacity to

152

Michael Luntley

direct their behavior with respect to saliences, they have the key ingredient for learning. They have the minimum level of mental agency required to get learning going. The capacity to direct behavior with respect to an F does not require possession of the concept F. Indeed, possession of the concept F is, on the approach I am suggesting, to be explained in terms of the subject’s capacity to find F salient plus the general capacity to direct behavior with respect to such saliences. The more the latter capacity is employed, the more their directing behavior with respect to F displays the generality of a conceptual orientation to F. Doubtless, an important ingredient for the eventual development of a conceptual orientation is the capacity to label the directing of behavior with respect to F and thereby secure a generality of application of that orientation. But for that to work there needs first to be the general capacity to direct behavior with respect to saliences, regardless of whether the subject has a conceptual orientation to saliences. The hypothesis here is that what distinguishes concept using creatures from non-concept using creatures is not that the former have labels for their saliences that the latter lack. The difference is that the former have the willful capacity to direct behavior with respect to saliences, the latter do not. Cats can have thwarted expectations, but for them such saliences do no more than prompt the need to behave differently. For the human infant, such saliences can be attended to and treated as that about which behavior is directed. An example will illustrate the point. Suppose a young child becomes habituated to finding bright spherical objects to be soft. The child acquires a capacity for discriminating soft objects in terms of their shape and color. This is not yet conceptual. It involves a training, a habituation of acting with respect to brightly colored spherical objects as if they are soft. Although that is simple training, it sets up expectations with regard to the new brightly colored sphere. When the child picks up a prickly pear, they encounter a surprise. That is a salience. At this point, there is still no language learning. However, if the child has the general capacity to direct its behavior with respect to saliences, it has the capacity to act with respect to things thrown into relief by disruptions in patterns acquired in training. It is, I suggest, that general capacity that is the key distinguishing feature of Augustine’s conception of the human infant with which Wittgenstein never takes exception. Wittgenstein objects to the sophistication and sheer number of beliefs and desires that Augustine appears to attribute to the infant. That attribution might be a mistake. But Wittgenstein does not object to the central notion of the infant as a willful subject, a subject with a capacity to direct behavior with respect to saliences. And, of course, if that general capacity is permitted, we have the ingredients to explain how ostensive definition works. In early language learning, the child is working out for herself what the adult is saying, not because they already have concepts, let alone the concept for the thing being pointed at, but because they already have the basic mental capacity to direct behavior with respect to saliences. The infant mind

On the teaching and learning of words

153

starts, presumably, with very vague and ill-focused capacities to find things salient and, therefore, vague and ill-focused capacities to direct her behavior. But the training is a training in the focus of saliences and, with the general capacity in play, thereby a training that provides ever more focused opportunities for directing behavior. At what point such directing of behavior gets labeled and acquires the generality of application that warrants calling it conceptual need not be precise. But there are, I submit, clear cases in which we would say deploying that capacity was not yet to deploy a conceptual capacity and clear cases in which we would say it was a conceptual capacity. Furthermore, although the attention that provides the direction is preconceptual, it is an ingredient in the overall project of making oneself intelligible. How we are placed within the space of reasons is, in part, a function of how and to what we attend. The model I have given provides a sketch of how the capacity to direct behavior can contribute to an account of concept acquisition in early language learning. It marks a distinction between conceptually structured directions and those that are not. The rationalism is stratified. In that respect, I suggest not only that Wittgenstein had a rationalist model of language learning, but one which has much to contribute to contemporary philosophy of psychology.

Notes 1 E.g., “Each new application [of a word] we make is a leap in the dark,” Saul A. Kripke, Wittgenstein on Rules and Private Language: An Elementary Exposition (Oxford: Blackwell, 1982), p. 55. Or “understanding an expression is a kind of ‘cottoning on’; that is, a leap, an inspired guess at the pattern of application which the instructor is trying to get across,” Crispin Wright, Wittgenstein on the Foundations of Mathematics (London: Duckworth, 1980), p. 216. 2 Kripke’s account is deflationary because it provides a reduced account of meaning in terms of assertibility conditions rather than truth conditions, Saul A. Kripke, Wittgenstein on Rules and Private Language. 3 The clearest statement of this response is perhaps that provided by John McDowell, “Wittgenstein on Following a Rule,” Synthese 58 (1994): 325–63. Goldfarb has an importantly distinct version of this move in which he appeals to Wittgenstein’s notion of the background, see Warren Goldfarb, “Kripke on Wittgenstein on Rules,” The Journal of Philosophy 82:9 (1985): 471–88. See p. 486 for the appeal to the background. Goldfarb makes this appeal as part of a rejection of the idea of the gap that informs what I am calling the familiar reading. It is in this respect that his reading is importantly distinct and more in line with the reading I provide below. 4 Paul Bloom, How Children Learn the Meanings of Words (Cambridge, MA: MIT Press, 2000), p. 10. 5 E.g., Jerry Fodor, “The Present State of the Innateness Controversy,” in Representations: Philosophical Essays on the Foundations of Cognitive Science (Brighton: Harvester, 1981). For a recent overview of Fodor’s position, see his “Doing Without What’s Within: Fiona Cowie’s Critique of Nativism,” Mind: A Quarterly Review of Philosophy 110:437 (2001): 99–148. Strictly speaking, for a rationalist of Fodor’s persuasion there is no such thing as learning, simply the triggering of innate concepts.

154

Michael Luntley

6 Cf. Alison Gopnik et al., The Scientist in the Crib: Minds, Brains, and How Children Learn (New York: William Morrow, 1999). 7 Bloom, How Children Learn the Meanings of Words, p. 4. 8 Of course, contemporary defenders of nativism about concepts do not see this as a dilemma, cf. Fodor, “Doing Without What’s Within.” For an explicit endorsement of an Augustinian account of learning in defiance of Wittgenstein’s apparent critique, see also Bloom, How Children Learn the Meanings of Words. 9 PI §5. 10 OC §538. 11 OC §475. 12 See also related passages: “I really want to say that scruples in thinking begin with (have their roots in) instinct. Or again: a language-game does not have its origin in consideration. Consideration is part of a language-game.” (Zettel §391; cf. RPP II §632). Also, “Instinct comes first, reasoning second. Not until there is a language-game are there reasons.” (RPP II §689). 13 This point is often simply taken for granted in philosophical treatments of language learning. For example, Russell says, “A child learning to speak is learning habits and associations which are just as much determined by the environment as the habit of expecting dogs to bark and cocks to crow [ … ]. [This involves] processes almost all of which are not deliberate, but the results of causes operating according to more or less ascertainable laws.” Bertrand Russell, The Analysis of Mind, Muirhead Library of Philosophy (London: Allen & Unwin, 1921). For contemporary endorsements of the same line, cf. Donald Davidson, “Truth Rehabilitated,” in Rorty and His Critics, ed. Robert Brandom (Oxford: Blackwell, 2000), pp. 70–71; M. A. E. Dummett, The Logical Basis of Metaphysics (London: Duckworth, 1991), p. 95. 14 The outline stratified rationalism that I attribute to Wittgenstein offers the prospect of explanatory detail in a rationalist account of concept acquisition. Fodor sees concept acquisition as a kind of triggering by, not an induction on, experience. But that still leaves the question why the concept F is triggered by experiences of Fs, rather than anything else. This is what Fodor calls the doorknob/ DOORKNOB problem, cf. Jerry Fodor, Concepts: Where Cognitive Science Went Wrong (Oxford: Clarendon Press, 1998). Solving this problem requires detail on how experience provides the basis for concept acquisition. The account I sketch begins to address these issues and without endorsing a straightforward nativism, for on the stratified rationalist account concepts are genuinely learnt, their acquisition requires mental activity on behalf of the subject. 15 PI §§65 ff. and 185 ff., respectively. 16 This is why it can seem that the best option for the rationalist is nativism, cf. ibid. 17 As noted, Goldfarb’s appeal to background is not really a move to close the gap, but one that denies the gap exists in the first place, see especially Goldfarb, “Kripke on Wittgenstein on Rules,” p. 484. Making that move is different to the appeal to a concept of initiation into a practice, but it is still a prolegomenon to a detailed account of why there is no gap and how we are to construe the capacities of the pupil to ensure that no gap arises. What I offer below is a start on that detail. 18 The very idea of a stratified rationalism is a direct challenge to McDowell’s treatment of concepts and what he calls, following Sellars, the space of reasons. See John McDowell, Mind and World (Cambridge, MA: Harvard University Press, 1994). If so, that means that on my reading, McDowell is not only wrong in his account of what it is to make sense of ourselves by grasping how we are located in the space of reasons, he is also wrong in his reading of Wittgenstein. When McDowell appeals to “practice” in resolving the rule-following argument, he is appealing to a concept of practice that is fully conceptually articulated; see

On the teaching and learning of words

19 20

21

22

23 24

25

155

“Wittgenstein on Following a Rule.” I think practice can include more levels of activity than that. The current reading of Wittgenstein is an extension of the reading provided in my Michael Luntley, Wittgenstein: Meaning and Judgement (Oxford: Blackwell, 2003). In that book, the concept of practice was not stratified. I concentrated there on the conceptually articulated sense of practice although I suggested that more work needed to be done on the concept of practice. It is what McDowell would call acting within the space of reasons. Giving due weight to this point is central to understanding the behavior of experts and how their behavior is to be rendered intelligible. In a recent pilot study on expertise I was concerned with cataloging and accounting for the situated reasoning of experienced professionals, such as classroom teachers, who organize and direct their behavior without a codified account of what they are doing. They act without a script for their directedness by deploying attentional skills that enable them to “lock-on” and adapt their behavior to the particular contingencies of the environment with which they deal. This is reasoning – the rational enterprise of making oneself intelligible. Space does not permit detailed discussion of the later example which has dominated the literature, but the reading I give of §69 and §71 is of a piece with a reading that needs to be developed of the key passages around §§208–10 in which Wittgenstein appears to deny the existence of the gap in the “add 2” case. I leave those passages to another time, but see, for example, the rejection of the need for a “deeper explanation” in §209. The notion of joining in does not require a constructivist construal of learning and of the meaning of words. Learning by joining in is, of course, a social activity, but the activity that the pupil joins in with is not intrinsically social; it is the activity of reasoning. On my model, that is something individualistic, for it is based in the pupil’s capacity to direct and organize her action. That is the individualistic basis of reasoning. Learning by reasoning takes place both individualistically and socially when two or more people are involved, e.g. teacher and pupil. But that does not entail social constructivism, it entails the platitude that social action involves two or more people coordinating their behavior. The best example of contemporary Russellian account of this capacity is to be found in John Campbell, Reference and Consciousness (Oxford: Clarendon Press, 2002). I am assuming a slightly more restricted use of the concept of attention than Campbell, for he sometimes seems to use “attention” for reactions to affordances, whereas I think it best to restrict it to personal level directing of attention to saliences. Some of Wittgenstein’s remarks about training are, perhaps, best construed as accounts of how the capacity for attention is acquired. If so, training picks out both basic reactions to affordances and those more sophisticated reactions in which one attends to saliences. I suspect that Wittgenstein’s concept of training covers both phenomena. But that is just to note that Wittgenstein does not get the last word in all this.

7

Meaning, rules, and conventions Hans-Johann Glock

Kripke’s Wittgenstein on Rules and Private Languages placed the so-called “rule-following considerations” at the center of philosophical attention. Kripke does not purport to provide an accurate account of the primary texts, but to propound “Wittgenstein’s argument as it struck Kripke.”1 As regards its content, Kripke’s interpretation is characterized by two features. First, like many other commentators, he adopts a communitarian reading of Wittgenstein, according to which rule-following and language are inherently social; second, like Fogelin before him,2 he portrays Wittgenstein as constructing a skeptical paradox in the style of Hume. As regards Wittgenstein studies proper, Kripke’s book stimulated an unprecedented interest in his remarks on rule-following, leading, e.g., to McDowell’s communitarian yet nonskeptical reading.3 It also helped to rekindle interest in Wittgenstein’s philosophy of mathematics, for instance through Crispin Wright.4 But it was vigorously attacked by more orthodox interpreters like Baker and Hacker.5 At the same time, Kripke’s book was the starting point for a debate about “Kripkenstein” on rule-following, a debate which is now conducted largely in blissful disregard of Wittgenstein’s own writings.6 My aim is to show that these writings, as well as those of Wittgensteinians, are relevant to one important question that has emerged from the Kripkenstein debate: Is there an essentially normative dimension to language and linguistic meaning? Both Wittgenstein scholars and Kripkenstein enthusiasts have tended to ignore one important fact. While the skeptical interpretation of Wittgenstein is wrong and the community interpretation contentious at best, as regards the normativity of meaning Kripke has highlighted and ably defended a genuinely Wittgensteinian idea. Furthermore, that idea is of undeniable importance to contemporary debates. The normativity of meaning is rightly perceived as a serious challenge to naturalism. Normativitism, generally speaking, is the view that human thought and behavior differ from inanimate nature and the behavior of mere animals in that they are subject to norms or rules, standards which prescribe and evaluate. Such norms have always appeared as a threat to naturalism, since they seem to defy reduction to the causal regularities recognized by the natural sciences. They have also inspired attempts to avoid

Meaning, rules, and conventions

157

both epistemological naturalism, the view that there is no knowledge outside of natural science, and ontological supernaturalism, the view that there are supernatural entities such as God, Platonic forms, or Cartesian souls. The most important attempts of this kind hail from Wittgenstein and philosophers influenced by him, e.g., Sellars, Brandom, McDowell, Putnam, von Savigny, and Hacker. The basic idea is that human beings are special not because they are connected to a reality beyond the physical world of space, time, and matter, but because they can only be adequately understood from a normative perspective alien to the natural sciences. For this reason, there is knowledge outside of natural science, knowledge of language and meaning, for example, even though it does not deal with supernatural entities. In this chapter I shall not pronounce on whether such a third way is feasible. Nor shall I try to defend the normativity of meaning against all the extant animadversions.7 Instead I show how the issue emerged from Wittgenstein’s philosophy of language (§7.1). I shall also reject interpretations that distance him from semantic normativity (§7.4). My main aim is substantive, however, namely to argue that Wittgenstein’s reflections on meaning and rules constitute a durable legacy for contemporary philosophy. I show how the current debate about rule-following has moved away from the original Wittgensteinian impetus, and distinguish different dimensions of semantic normativity (§§7.2–7.3). In §7.5, I demonstrate how semantic normativity can be explained by reference to a particular type of linguistic mistake, and in §§7.6–7.7 I defend it against two prominent Davidsonian objections.

7.1 Historical background The idea that language is a rule-guided activity is as old as the study of language. It underlies the discipline of grammar, whether in its traditional or modern form (descriptive and generative grammar). It was given an additional boost by the development of modern logic at the turn of the twentieth century (Frege, Russell): logical calculi evolve around formation-rules (syntax) and transformation rules (rules of inference). Nevertheless, within analytic philosophy, the idea that there is an irreducibly normative dimension to linguistic meaning goes back to Wittgenstein. In a text from the early thirties which is based on dictation by Wittgenstein, Waismann wrote: Of course words impinge on human life, of course they produce various effects. What we want to say is merely this: we do not regard language from the perspective of a mechanism, but from that of a calculus, a procedure according to norms. [ … ] In the picture of a psychic mechanism, in which language pulls the lever that moves the body of the hearer, the concept of meaning does not yet appear. There is only room for this concept if one regards language from a normative perspective. Indeed, the concept of meaning is felt to be an alien element in the considerations of natural science. And our analysis shows why this is so.8

158

Hans-Johann Glock

In some respects this passage is very modern. In it we find the three ideas that characterize an important stand of contemporary normativism: first, meaning is essentially normative; second, this normativity defies reduction to the kind of causal relations investigated by natural science; third, the reason is not that meaning or meaningful language involve supernatural phenomena; rather, the normative dimension reveals itself to a special perspective, a unique yet indispensable way of looking at phenomena that can also be described scientifically. In other respects, this passage is antiquated. Few contemporary proponents of the normativist “third way” between naturalism and supernaturalism express themselves in such an unguarded manner. More importantly, while the Wittgensteinian normativity claim features in Kripke’s account, it is only distantly related to various ideas that have since come to dominate the debate. When the Wittgensteinians talk about the normative dimensions of meaning, they really mean meaning, as opposed to propositional content, for instance. They are specifically concerned with the conventional or literal meaning possessed by words of a language. Their position is the result of two steps. First, Wittgenstein attacked the referential conceptions of meaning, variants of which occur in Frege, Russell and his own Tractatus. The meaning of a word, he came to think, is not an object it stands for, but “its use in the language.”9 Second, Wittgenstein criticized the behavioristic and causal theories of meaning he encountered in Russell,10 Ramsey,11 Ogden, and Richards.12 These theories also explain the meaning of an expression by reference to its use. But the operative conception of use is purely causal. Leaving aside various complications, the meaning of a word is equated either with the conditions which cause a speaker to utter it, or with the effects which such an utterance has on hearers. The analysis which, according to Waismann, reveals the incongruity between meaning and natural science is Wittgenstein’s attack on such theories. Wittgenstein grants that the initial acquisition of language is a kind of training which includes the conditioning of verbal reactions to external stimuli. But, like Chomsky after him, he also points out that fully developed linguistic behavior is not strictly conditioned. Furthermore, the causes and effects of uttering an expression determine neither whether the expression has meaning, nor what meaning it has. The meaning of an expression is not identical with the conditions that determine whether or not it is uttered in specific situations, since that depends not just on its meaning, but on extrinsic factors. For instance, the fact that few people would dare call Prince Philip a racist has nothing to do either with the meaning of the term or with Prince Philip, but with social norms. Similarly, whether an expression is meaningful and what meaning it has does not depend on either its actual or even its intended effect, whether on a particular occasion or in general. If I say “Milk me sugar!” for example, this may well have the result that my hearer stares at me and gapes. But it does

Meaning, rules, and conventions

159

not follow that this combination of words means “Stare at me and gape!” It doesn’t even follow if this entertaining effect can be repeated. Indeed, it does not even follow if I utter these words with the intention of bringing about this reaction.13 Wittgenstein’s alternative to the causal conception of use is a normative one. The meaning of an utterance depends not on how it is actually understood, but on how it ought to be understood by members of a linguistic community. What is semantically relevant about the use of an expression is neither the causes nor the effects of its utterance, nor the intentions with which speakers utter it, but its correct use. The linguistic meaning of an expression depends on general rules. Wittgenstein uses the expression “grammatical rules” here. This is misleading. For what he has in mind are not (just) school-grammatical or syntactical rules, but semantic and logical rules (see §7.5). These rules provide standards for the correct use of expressions. They thereby determine what it makes sense to say with the help of these expressions. The Wittgensteinian normativity of meaning is therefore a specific claim about the connection between linguistic meaning and a certain type of linguistic rule. According to Wittgenstein, the meaning of an expression is determined by rules for its correct use.14 This idea also provides a Wittgensteinian middle way between the Scylla of epistemic naturalism and the Charybdis of ontological supernaturalism. Language is in some ways analogous to a game like chess. On the one hand, a chess-piece is a piece of wood that can be described by physics. On the other hand, one cannot explain what a chess-piece or what the game of chess is in purely physical terms. But the difference between a chess-piece and a simple piece of wood is not that the former is associated with an abstract entity or with a process in a separate mental realm. Rather, it is that there are rules for its use.15 The chess-piece has a role in a rule-guided practice. That practice in turn presupposes agents with special and distinctively human capacities. But while these capacities cannot adequately be characterized in physical terms, they do not transcend the natural world. They are perfectly intelligible features of animals of a unique kind; and their causal prerequisites and evolutionary emergence can be explained by science.

7.2 The development of the normativity claim after Wittgenstein Paul Boghossian has provided a formulation of the normativity claim which has become standard: Suppose the expression “green” means green. It follows immediately that the expression “green” applies correctly only to these things (the green ones) and not to those (the non-greens). The fact that the expression means something implies, that is, a whole set of normative truths about my behaviour with that expression: namely, that my use of it is correct in application to certain objects and not in application to others [ … ]. The

160

Hans-Johann Glock normativity of meaning turns out to be [ … ] simply a new name for the familiar fact that meaningful expressions possess conditions of correct use.16

This formulation is compatible with the Wittgensteinian normativity claim. But the current debate is also marked by three shifts: 1. from rules to correctness; 2. from meaning to content; 3. from language to mind. In what follows I shall comment on these developments in turn. (AD 1) Historically “norm” means something like standard or rule. And for many philosophers the term and its cognates remain by definition tied to rules. Recently, however, “normative” and “normativity” have been employed without this restriction, namely to characterize phenomena that can be assessed for correctness, whether or not this is done by reference to rules. Thus Boghossian accepts the Wittgensteinian idea that an expression has a meaning only if there are conditions for its correct use. There is “an a priori connection [ … ] between the meaning of a word and its correct use.”17 But contrary to what I shall call rule-based normativity, he denies that there is a connection between an expression having such conditions and it being subject to rules.18 A similar line is taken by Davidson. In his last writings, he explicitly embraced the idea that there is an irreducibly normative dimension to linguistic behavior. There must be a “distinction between using words correctly and merely thinking one is using them correctly.”19 At the same time, however, Davidson opines that “the yearning for [linguistic] norms is a nostalgic hangover from the dependence on a Platonic conception of meaning.”20 Thus Boghossian and Davidson come together in maintaining that while linguistic meaning requires correctness conditions, it does not require rules of use. (AD 2) In recent contributions one encounters not just the term “normativity of meaning” but also the term “normativity of content.” This is in line

Figure 7.1

Meaning, rules, and conventions

161

with a more general tendency, namely to employ the terms “content” and “meaning” interchangeably, and to speak of the content or meaning of a linguistic act and of a thought.21 In the wake of Frege, many contemporaries assume as a matter of course that thoughts or propositions are not just the bearers of truth-values, and the objects of intentional attitudes like belief, but also the meanings of declarative sentences. This is one Fregean assumption that escaped Wittgenstein’s critical attentions. In a radical redeployment of Frege’s distinction between sense and meaning, the Tractatus had insisted that only sentences have sense and that only names, i.e., the ultimate components of sentences, have meaning.22 Even in his later oeuvre Wittgenstein tended to reserve the expression “meaning” for words and to speak of sense in the case of sentences. At the same time, however, he did not distinguish consistently between the sense of a sentence and the thought expressed by it. Thus when he discussed the obeying of an order,23 he did not keep apart understanding the meaning of the imperative sentence from understanding the order expressed by it. But I may understand the meaning of “Move your bicycle!” without grasping that the order was issued to me. The Fregean assumption is untenable and has had a baleful influence on contemporary discussions of normativity. A proposition, propositional content or thought is something that is or could be said or believed—a sayable or thinkable. However, propositions in the sense of what is or could be said are what we express or convey by the use of sentences, not the meanings of these sentences. Unlike the meaning of a sentence, what is said (believed, etc.) can be true or false, implausible or exaggerated.24 Furthermore, far from being identical with sentence-meanings, what is said on a particular occasion depends on sentence-meaning and context of utterance. Thus, if someone utters “Where I live the unemployment rate is currently lower than here” what the speaker has said depends both on the meaning or sense of the type-sentence and on who utters it, where and when. Consequently we must distinguish issues concerning meaning from issues concerning content, even if it should transpire that there is a close connection between the normativity of the one and that of the other. In cases in which indexicality is not an issue, or in which knowledge of the context of the utterance can be assumed, understanding the sense of the sentence and understanding what is said coincide. Furthermore, according to Wittgenstein to understand the content of an order can no more be equated with obeying it than understanding the sense of the sentence used to issue the order. Such understanding does not depend on how the order is in fact understood, but on how it ought to be understood. (AD 3) The final shift is closely connected to the second, even though it cuts across the distinction between meaning and content. In a more recent paper, Boghossian distinguishes between a mentalistic and a linguistic version of the normativity thesis, the normativity of mental content on the one hand, and the normativity of linguistic meaning on the other. According to

162

Hans-Johann Glock

the normativity of mental content, “it is constitutive of belief to aim at the truth. The truth is what you ought to believe [ … ].”25 By the same token, according to what Boghossian calls the normativity of linguistic meaning, it is constitutive of assertion to aim at the truth, and the truth is what you ought to assert. Within my scheme of classification, by contrast, the dichotomy between content and meaning does not coincide with that between mind and language. The two normativity claims distinguished in this passage both concern truth and hence content rather than meaning. Content attaches not just to utterances or statements but equally to unexpressed mental states. By the same token, any normative dimension to content concerns not just language but also the mind.26 What Boghossian distinguishes is a mentalistic claim about the content of mental states—namely beliefs, whereas the linguistic claim is about the content of linguistic acts—namely assertions.

7.3 Dimensions of semantic normativity In the previous section we have seen that the so-called “normativity of meaning” actually covers a wide range of issues. We may wish to label the overall topic “semantic normativity,” using the term “semantic” in a highly general sense which is potentially misleading but widespread, one which includes all types of intentionality. The above considerations suggest that this general topic must be divided up as shown in Figure 7.1 Accordingly, we must distinguish between various normativity claims. All of them involve the idea that a certain phenomenon which is semantic in a highly general sense entails the existence of certain normative phenomena. One such claim concerns linguistic meaning rather than content, and it is epitomised in the above quote from Boghossian. It maintains that if a word w is meaningful, then there must be conditions for its correct application. Bare normativity of meaning (BNM) w is meaningful ) there are conditions for the correct use of w. A second, stronger claim is that semantic normativity is rule-based. It is accepted by followers of Wittgenstein and speech-act theorists like Searle, yet rejected by some who subscribe to BNM, such as Davidson and Boghossian. According to rule-based normativity, meaningful expressions entail not just the existence of correctness conditions, but of rules of use, because the correctness conditions of NM are laid down by such rules. Rule-based normativity of meaning (RNM)

e is meaningful ) there are rules for the use of e.

A third claim concerns prescriptions or oughts.

Meaning, rules, and conventions

163

Prescriptive normativity of meaning (PNM)

e is meaningful ) there are prescriptions for e’s use.

As we shall see, (RNM) and (PNM) must be distinguished, since not all rules must be prescriptions. Two further claims concern content rather than meaning. One is Dummett’s proposal that the concept of assertion entails the existence of certain rules, notably the rule that an assertion should aim to be true. Normativity of content: assertion (NCA) a asserts that p ) there are rules for asserting that p. The other concerns the content of beliefs rather than assertoric utterances, and is accepted by Boghossian, who rejects NCA. Normativity of content: belief (NCB)

a believes that p ) there are rules for believing that p.

7.4 Unruly Wittgensteinians This is not the place to discuss the substantive merits of (NCA) and (NCB).27 But an exegetical observation is in order. For Dummett there is a general convention whereby in making an assertion we aim at truth, just as there is a general convention according to which in making a move in chess we aim at the stipulated point of the game, e.g. checkmating in chess.28 In comparing language and linguistic activities to games, Dummett is obviously taking a cue from Wittgenstein. It is noteworthy, however, that Wittgenstein himself did not take the analogy that far. On the contrary. He repeatedly denied that language in general and the language-game of asserting in particular serve a single overarching purpose, whether it be communication or the expression of thoughts.29 This last observation has two consequences. First, Wittgenstein did not intend the game analogy to support normativity claims concerning assertion. Second, for him, the rules implied by meaningful speech do not include a prescription of truthfulness. So one can exaggerate the extent to which Wittgenstein regarded language as rule-governed. Ironically, some followers of Wittgenstein would go much further, and distance him from the claim that any rules are implied by either content or meaning. While they acknowledge that Wittgenstein insisted on a normative dimension to meaning (BNM), they deny that he regarded language as rulegoverned (RNM). They also suggest that (RNM) is, in any case, incorrect.30 According to these “unruly” Wittgensteinians, his comparison of language to rule-following activities such as games should be seen as a dispensable, if not misleading, heuristic device, which perhaps betokens a school-masterly attitude he imbibed while teaching in Lower Austria in the 1920s.

164

Hans-Johann Glock

This view is correct insofar as Wittgenstein became suspicious of his own earlier picture of linguistic rules. However, his reaction was not to abandon the notion of a linguistic rule, but to clarify it. This is precisely the role of the later discussion of rule-following. In the Tractatus linguistic rules form the logical syntax, a complex calculus of inexorable norms hidden behind the surface of natural languages.31 By the mid-thirties Wittgenstein had moved away from this “calculus-model” of language.32 He denied that rules can guide linguistic behavior and determine in what it makes sense to say without being known to us. He also rejected the idea that rules are inexorable or independent of human activities, while retaining the idea that human practice is in many respects rule-governed. In particular, Wittgenstein changed his conception of linguistic rules by comparing language no longer to a calculus of hidden and rigid rules, but to a game. That comparison, however, is not just a dispensable heuristic device. Wittgenstein continued to insist that linguistic understanding involves mastery of techniques concerning the application of rules. To follow a rule, to issue a report, to give an order, to play a game of chess, are habits (customs, institutions). To understand a sentence means to understand a language. To understand a language means to be master of a technique.33 We need to avoid not just the calculus model of hidden rules, but also an intellectualist picture of rule-following, according to which it necessarily involves the consultation and interpretation of rule-formulations. For this picture leads to Wittgenstein’s notorious regress: to interpret the initial ruleformulation, we replace it by another one, which in turn needs to be interpreted, and so on. As a result, we would end up with a “paradox: [ … ] a rule could not determine a course of action, since every course of action could be made out to accord with the rule.” Does Wittgenstein conclude that following a rule is strictly speaking impossible, as rule-sceptical and unruly readers have suggested? Not at all! He immediately goes on to diagnose the paradox as based on a misunderstanding. [We] posit one interpretation after another; as if each one contented us at least for a moment, until we thought of yet another standing behind it. What this shows is that there is a way of grasping a rule which is not an interpretation, but which is exhibited in what we call “obeying a rule” and “going against it” in actual cases. (PI §201) On the basis of this down-to-earth conception of rules, Wittgenstein continued to insist on the link between grammatical rules and meaning. In 1930 he remarked in a lecture:

Meaning, rules, and conventions

165

What justifies us in using any particular word? Suppose I say “This gown is black.” The word black is arbitrary in one sense; another sound or scratch would serve. And the correlation of the word “gown” to a particular object is in itself arbitrary and has no consequences. But if a proposition is to have sense we must commit ourselves to the use of the words in it. It is not a matter of association; that would not make language work at all. What is essential is that in using the word I commit myself to a rule of use. A word only has meaning in a grammatical system, and what characterises it is the way in which it is used. (LWL p. 36) To some ears, talk of a grammatical system may still smack of an overly rigid conception of linguistic rules. But in stressing commitment to a practice, this passage already avoids both the idea of unconscious rule-following and the intellectualist mistake. Furthermore, right to the end of his career Wittgenstein expressed his central insight that meaning is a matter of use through invoking the idea of rules. “A meaning of a word is a kind of employment. That is why there exists a correspondence between the concepts ‘rule’ and ‘meaning’.”34 Small wonder, then, that Wittgenstein maintained that “following according to the rule is FUNDAMENTAL to our language-game.”35 Rule-following plays a central role in Wittgenstein’s overall oeuvre because of two abiding convictions: first, language is a rule-guided activity; second, the a priori status of logic, mathematics and philosophy derives from such rules.36 More specifically, philosophical problems are to be resolved by attending to rules of use. We are interested in language as a procedure according to explicit rules, because philosophical problems are misunderstandings which must be removed by clarification of the rules according to which we are inclined to use words. We consider language from one point of view only.37 But why is this “one point of view only”? First, because language can also be viewed from a non-normative, purely causal “perspective of a mechanism,” as the Waismann passage quoted in section 7.1 states.38 Second, our linguistic practice is rarely guided by explicitly stated or rigid rules. As long as this is kept in mind, however, nothing prevents us from explicating rules for philosophical purposes.39 And the same holds for stressing the rule-guided character of language in general. Accordingly, there are overwhelming reasons for ascribing to Wittgenstein not just (BNM)—the idea that meaning is normative—but also (RNM)—the idea that meaning is linked to rules. The question is whether he provides us with compelling arguments in their support.

7.5 Introducing semantic rules “by mistake” In this section I sketch a Wittgensteinian argument in favor of rule-based normativity of meaning. The argument revolves around the need to distinguish

166

Hans-Johann Glock

between a statement being false and a statement being meaningless or nonsensical, a need which contemporary discussions of semantic normativity tend to ignore. To illustrate this, let me turn to a passage from Blackburn: I intend no particular theoretical implications by talking of rules here. The topic is that there is such a thing as the correct or incorrect application of a term, and to say such a thing is no more than to say that there is truth and falsity. I shall talk indifferently of there being correctness and incorrectness, of words being rule-governed, and of their obeying principles of application. Whatever this is, it is the fact that distinguishes the production of terms from mere noise, and turns utterance into assertion—into the making of judgement.40 Blackburn simply identifies meaningful terms having correctness conditions with the sentences in which they occur having truth-conditions. Boghossian has pointed out that this is unwarranted, since assertion conditions provide a different kind of correctness conditions. In my view, this failure is symptomatic of a more fundamental shortcoming. It is a failure to heed a crucial lesson of Wittgenstein’s idea that grammatical rules furnish standards for the correct use of words. The contrast between correctness and incorrectness applies to language in a variety of ways, and their connection to the notion of meaning differs in crucial respects. In particular, one can use a word meaningfully, in a sentence with a sense that expresses a thought, without that thought being either true or justified. Conversely, as Wittgenstein insisted on throughout his career, “the question of meaning” is antecedent to “the question of truth.”41 A sentence must be meaningful in order to be even in the running for expressing a truth, i.e., in order to be what we nowadays call “truth-apt.” As regards the issue of normativity, this difference between truth and meaning emerges most clearly when we consider the different types of mistakes that people can make when they use words. In focusing on mistakes, I am taking my cue from Wittgenstein’s aforementioned suggestion that rules are manifest in what we call “following the rule” and “going against it.”42 Unlike Wittgenstein, however, I place a certain trust in the distinction between syntactic and semantic rules.43 I shall use examples from recent politics, which provides rich pickings for this sort of analysis. First, in the run-up to a recent by-election to the House of Commons the Conservative candidate Uma Fernandes said: (1) I know that I cannot win this seat. There is nothing wrong with this statement, either linguistically or factually. And yet she made a dreadful mistake, because it is not politic for a candidate to admit to such dim prospects, and she has certainly damaged her career.

Meaning, rules, and conventions

167

Next consider Tony Blair’s statement in the House of Commons: (2) Iraq is capable of deploying weapons of mass destruction within 45 minutes of an order by Saddam Hussein. There is nothing wrong linguistically with this sentence, but the statement is mistaken, because it is not just unjustified but false. Now consider George W. Bush’s statement: (3) You teach a child to read, and he or her will be able to pass a literacy test. Here we are confronted with a linguistic mistake. But it is of a syntactic rather than semantic kind. In fact, (3) is analytic, at least on a certain understanding of “teach” and “literacy.” In spite of involving a solecism, therefore, this example shows Bush at the very height of his semantic powers. That is more than can be said of the following statement: (4) I am mindful not only of preserving executive powers for myself, but for my predecessors as well. Here Bush’s travails are semantic: (4) betokens not lack of factual knowledge, but an insufficient grasp of the meaning of at least one of the two terms “preserving” and “predecessors.” There is a fundamental contrast between saying something false or unjustified—as in (2)—and saying something which is meaningless—as in (4). Some uses of words are mistaken solely because of what these words mean, irrespective of any other facts, of syntactical rules or of social expectations. Conversely, one can apply a word in a way which is semantically correct, without applying it correctly in the sense of applying it to say something true. Proponents of rule-based normativity try to capture these points by treating meaning as determined by rules. In the case of a general term, for instance, it is plausible to hold that these rules specify not what it actually applies to, but the conditions under which it applies to an object x. Whether it actually applies then depends on whether x satisfies these conditions; it is a matter of fact rather than meaning. Wittgenstein regarded explanations of meaning as rules of this kind. And he was not particularly fussed about the form they take. Thus he regarded as expressions of semantic rules not just a meta-linguistic definition such as: (5) The term “drake” means male duck, but also a standard definition such as: (5’) A drake is a male duck. Only (5) has a meta-linguistic form, i.e., mentions a linguistic expression. Nevertheless, both (5) and (5’) are standardly used to explain what the term

168

Hans-Johann Glock

“drake” means. And because they function as explanations, they also have a normative role vis-à-vis the use of the term “drake”: we can criticize and justify our use of words by reference to them. Wittgenstein had a functional conception of semantic rules. That is to say, whether a sentence expresses a grammatical rule or an empirical proposition depends not on its linguistic form, but on its role or function within our linguistic practice. As a result, his conception of a semantic rule is extremely catholic. Grammatical rules in his sense comprise any explanation of meaning.44 What counts as an explanation of meaning is again a matter of function, not form. Grammatical rules in this sense include: i) ii) iii) iv) v) vi) vii)

definitions whether in formal (“‘Bachelor’ means ‘unmarried man’”) or material mode (“Bachelors are unmarried men”); analytical propositions (“All bachelors are unmarried”); color-charts and conversion-tables;45 ostensive definitions;46 explanation by exemplification;47 expressions of the geometry of color such as “Nothing can be red and green all over;”48 propositions of arithmetic and geometry.49

I am sympathetic to this functional and catholic picture of semantic normativity. In order to link up directly with the current discussion, however, I shall focus on statements of the following form: (6) It is correct to apply “drake” to x iff x is a male duck. Serious issues lurk here. One reason for favoring (6) might be that, unlike (5), it promises an analysis of meaning in terms of truth- or application-conditions. But this comes at a price. The “correct” in (6) is that of truth rather than meaningfulness, and it operates with the extensional “iff.” As a result, one can know (6) without knowing what “drake” means, namely if one regards the latter’s connection to “male duck” as one of material equivalence. But I shall leave these complications aside. For most opponents of rule-based normativity concede that: (III)

“drake” means what it does ) (6).

That is to say, they accept my Normativity of Meaning (NM) while rejecting my Rule-Based Normativity of Meaning (RNM).50

7.6 Objections to rule-based normativity of meaning Davidson and his followers have leveled two principal objections against (RNM), which I shall discuss in turn.

Meaning, rules, and conventions

169

The first objection is that the alleged semantic norms really boil down to mere “descriptive facts” concerning linguistic behavior.51 In my view, this objection is ultimately based on a refusal to recognize the difference between two kinds of statements involving norms. Thus Wittgenstein insisted that grammatical propositions must be distinguished from empirical statements to the effect that a community follows certain linguistic rules, e.g., “All Englishmen use these signs in this way.”52 For they are used normatively, to explain, justify, and criticize uses of words. By this token, we must distinguish (6), which expresses a norm, from something like: (7) In English, it is correct to apply “drake” to an object x iff x is a male duck. (7) is an empirical proposition to the effect that a community follows a norm, what von Wright (perhaps influenced by Wittgenstein) calls a “normproposition.”53 Such a norm-proposition states a fact about the behavior of Anglophones. It does so, however, only because the Anglophone speech community accepts a certain norm, a standard for the correct use of “drake.” That is to say, (7) is a true factual statement only because (6) expresses a norm of English; (6) and (7) cannot be equivalent, since the normative status of (6) explains why (7) is a factual truth. Davidsonians have a comeback here which is not available to less radical opponents of semantic rules.54 They could say that even (7) overstates the normativist case. All we really have are statements such as: (7’) Among a certain population, most people tend to apply “drake” to an object x iff x is a male duck. (7’) is indeed bereft of any normative implications. But the insistence that linguistic communication requires no more than statistical regularity, and perhaps not even that, is untenable. We not only tend to apply “drake” to male ducks, we also explain the term as applying to such creatures, and we criticize and justify our use of the term by reference to statements such as (5)–(7). Applying “drake” to a male duck or a female fox counts not just as unusual, but as a mistake. Davidson has insisted that this notion of a linguistic mistake is shallow and uninteresting.55 On a charitable interpretation, what he means is that these rules need not be conventions, i.e., shared arbitrary rules, in order for communication to succeed. As I have argued elsewhere, this claim is mistaken.56 In any event, it is not to the current point, since semantic principles such as (6) need not be shared among all members of a linguistic community in order to have a normative status. Some opponents of rule-based normativity are prepared to avoid the conclusion that (6) has a normative status by querying the idea that correctness is a genuinely normative concept. In my view, this is pretty much like

170

Hans-Johann Glock

querying whether the Pope is really Catholic. Correctness is a normative notion, indeed it is a normative notion par excellence. Still, I can detect one possible ground for doubt. A serviceable explanation of correctness which is in line with works of reference like the Oxford English Dictionary runs as follows: (IV) x is correct iff x is in accordance with an acknowledged standard. This standard may be part of a specific system of evaluation appropriate to x, as when we speak of correct dress. Or it may be a general standard, e.g., that x should be in line with the demands of fact, truth or reason. Either way, we presuppose a standard specifying conditions under which something counts as correct. At this point it is tempting to argue that to call x correct is merely to state that it satisfies these conditions, and hence that the concept has no normative implications after all. This is a fallacy, however. In characterizing something as correct we do not just characterize it as possessing the features specified in the relevant standard. We also characterize it as meeting a standard of positive evaluation. The term “correct” is not purely descriptive or factual, but evaluative. What is more, unlike other evaluative terms, e.g., “good,” it carries the further implication that failure to meet this standard is not just undesirable, but provides grounds for intervention. After all, the verb “to correct” means to set right, rectify, or amend, and the Latin verb corrigere lies at the etymological root of the English expression. In other words, applying terms used in the standard may be purely descriptive, but applying the term “correct” is not. And to call something incorrect is not only to evaluate it negatively; if something is incorrect, then one has grounds for correcting it. To appreciate this difference, consider the following statements: (8) Glock (9) Glock (10) Glock (11) Glock (12) Glock

is not wearing a cap and gown is not dressed correctly is dressed badly had to change his dress, because it was incorrect didn’t have to change his dress, because it was correct

No normative force attaches to (8). But in certain contexts, for instance, a graduation ceremony, not wearing a cap and gown counts as not being dressed correctly. And (9), which states this fact, does have normative force. It is not just a negative evaluation like (10). According to my family, (10) is what Quine calls a standing sentence. It holds true most of the time. But (9) doesn’t. I am happy to say that my attire rarely gives license to correction. But this is precisely the force of (9). In the envisaged case it entitles the stewards to insist that I either put on cap and gown or absent myself from the procession. If “correct” and “incorrect” no longer served these normative

Meaning, rules, and conventions

171

purposes, their meaning would have altered by any standards. For example, if (11) and (12) became unacceptable while: (11’) (12’)

Glock had to change his dress, because it was correct; Glock didn’t have to change his dress, because it was incorrect;

were to become acceptable locutions, then “correct” and “incorrect” would have reversed their meaning. This lesson carries over to semantic normativity, whether it be that of meaning or that of content. To assess an assertion or belief as false is to present it as fit for epistemic correction. Equally, to assess an utterance as meaningless, nonsensical, or a malapropism is to present it as fit for semantic correction. To be sure, on particular occasions a speaker may aim to make false assertions or to commit linguistic mistakes. In that case what she does may be correct relative to a different standard—a prudential or moral one, for example. If that standard is more important in the situation, what she does may also be correct in the all things considered sense. But this does not detract from the fact that she is violating semantic or epistemic norms. In semantic statements such as (6), “correct” is manifestly not used in the all things considered variety. Conversely, the general norms of reason presupposed in that usage are not constitutive of the meaning of specific terms, with the possible exception of terms like “reasonable” or “rational.”

7.7 Objections to prescriptive normativity of meaning At this point the Davidsonians have a second objection in store: even if a statement such as (6) has some kind of normative status, it is not a rule, since it does not guide behavior.57 It does not oblige or even entreat us to apply the term “drake” to an object in a particular situation. To this objection there are three responses. First, the complaint is based on an unduly narrow conception of what counts as a rule. It ignores the “definitory aspect of rules.”58 Some rules do not prescribe a certain form of behavior, but instead lay down what a thing must be like to satisfy a certain description. Prime examples are the beloved EU norms concerning, for example, what qualifies as chocolate or the rules of games like chess. These norms lay down what counts as chocolate or as castling, but they do not request anyone to produce or buy chocolate, or to castle. This point is connected to a well-known distinction between regulative and constitutive rules.59 Regulative rules advise one on how to pursue in an optimal fashion an activity that can be specified independently of the rule. By contrast, constitutive rules are partly definitive of the activity in question. What goes on at dinner parties can be described as eating independently of the rules of etiquette. By contrast, what goes on when people move certain pieces across a board cannot be described as castling, checking, mating, or, more generally, as playing chess, independently of rules such as “The king

172

Hans-Johann Glock

cannot castle out of a check” or “The king is mate if it is in check in such a way that no move will remove it from being in check.” This contrasts with a regulative rule of chess such as “Do not relinquish control of the center.” Even someone who disregards all of these regulative rules does not cease to play chess, he will just play chess badly.60 Wittgenstein’s grammatical rules are precisely constitutive rather than regulative. The grammar of a language is the overall system of its grammatical rules.61 It comprises the rules which define that language by determining what it makes sense to say in it, rather than prescribing an optimal way of speaking the language.62 As regards the substantive issue, it can hardly be denied that constitutive rules are rules or norms, indeed that they make up a paradigmatic type. Otherwise one would have to insist that rule books do not really contain rules and that the law of contract does not really contain norms. Opponents of rule-based normativity might insist, however, that leaving labels aside, the constitutive rules of language lack the dynamic, languageguiding oomph of genuine rules. In other words, they might attack prescriptive normativity of meaning or (PNM). But even this fallback position is untenable. For, and this is the second response, semantic rules do have action-guiding force. By laying down conditions under which terms of a language can be applied, semantic principles also lay down conditions under which certain utterances are correct. For instance, by laying down what counts as a drake, explanations such as (5) or (5’) also lay down what counts as using the term “drake” correctly, a point which comes out in statements such as (6). And such statements support prohibitions. We can justify: (13) You ought not to apply “drake” to a female fox, by reference to (6). And we often say: (14) You shouldn’t call that …, because it’s not … . Finally, the complaint that semantic norms are not genuine rules ignores that even among regulative rules there are different types. On the one hand there are prescriptive norms, also known as “norms of obligation”63 or “mandatory norms,”64 norms of the form: (V)

x ought to u in condition C

Examples include: “Promises ought to be kept,” “Parents must look after their children,” “Nobody should commit adultery,” “When insulted, one should issue a challenge to duel.” But there are also “permissive norms,” norms of the form: (VI)

x is entitled to u in condition C

Meaning, rules, and conventions

173

Examples include: “Children under the age of 8 may remain with their parents,” “The chancellor is empowered to appoint a special envoy,” “The defendant is entitled to a legal counsel of his choice,” etc. It is common and unproblematic to use sentences like (5)–(6) as permissive norms. I can appeal to a definition such as (5) in defending or simply explaining my use of “drake.” If I am challenged why I call a male duck a drake I can defend myself by saying that “drake” means male duck. Furthermore, such definitions also function as transformation or inference rules. This point also hails from Wittgenstein’s discussion of grammatical rules. Ostensive definitions, for instance, function as substitution rules which licence the substitution of a demonstrative together with a gesture indicating a sample for the definiendum. They specify that anything which is this can be characterized as being A. An ostensive definition of red, for instance, entitles one to pass from “My bike is this ☞ color” to “My bike is red.”65 The idea that such transformation rules are constitutive of meaning was subsequently taken up by the inferential role semantics pioneered by Sellars and developed by Brandom and Boghossian, among others; (5) licences transformations of empirical propositions, e.g., from “This creature is a male duck” to “This creature is a drake” and vice versa. In other words, we can appeal to (5) in support of an inference such as: (15) Donald is a male duck, therefore Donald is a drake.

7.8 Conclusion One of the main arguments in favor of a normativist conception of language is the fact that linguistic utterances are subject to both regulative and constitutive rules in the same way as chess moves. Someone who calls his superior an “idiot,” or who provides lengthy explanations when he requires urgent help is imprudent, but he does not commit a linguistic mistake. By contrast, someone who violates rules of syntax, e.g., by failing to conjugate properly or by violating rules of word order, or lexical rules, e.g., by calling an ambiguous statement ambidextrous, commits a linguistic mistake.66 It might be thought that there are limits to the analogy with chess, on the grounds that someone who violates a constitutive rule of chess is no longer playing chess, while someone who violates the alleged constitutive rules of English, for example, is still speaking English. But this objection is dubious on both grounds. It is simply implausible to insist that a game involving a single incorrect move no longer counts as chess, just as it is implausible to insist that the medieval board-game was not really chess, because the pawns could not move two squares at a time in their first move. Conversely, although someone who commits a mistake may still be said to be speaking English, the mistake itself does not count as English. It is quite common to react to a serious mistake by saying “That’s not English.” And someone who commits many such mistakes can be said not to be speaking English. What

174

Hans-Johann Glock

is true is that the margin of tolerance is greater and less well defined in the case of natural languages than in the case of chess, for the simple but important reason that the constitutive rules of natural languages do not form a precise and stable system. That language is not spoken according to such rigid and precise rules is a lesson of Wittgenstein’s attack on the calculus-model. It is a lesson ignored by Davidsonians and Chomskians, who deny that any two people ever speak the same language. They rightly note that: if no two speakers ever share the very same vocabulary items or observe exactly the same rules of grammar there will be no formally precise, syntactic and semantic characterisation of the language spoken by a given community.”67 But this only entails that linguistic communities do not share a language on the assumption that languages are individuated by such formally precise characterizations. This is clearly not how we individuate natural languages like Basque or French. Consequently there is no reason to suppose that, strictly speaking, members of a linguistic community do not share a language, provided that we use “language” in the established way rightly emphasized by Wittgenstein.68 Wittgenstein is generally at odds with proponents of an individualist conception of language, such as Davidson and Chomsky. He insists not just on the rule-governed nature of language, but also on the idea that these rules are conventions, i.e., rules which are both arbitrary and inter-subjectively shared. The grammatical rules which underlie language and logic are arbitrary: they do not mirror or pay heed to any putative essence or form of reality and cannot be correct or incorrect in a philosophically relevant way.69 This position has often been regarded as conventionalist. At the same time scholars such as Baker have rejected that label.70 Yet Wittgenstein himself had no scruples to call grammatical rules “conventions” (Übereinkunft, Konvention).71 The term Übereinkunft is even more blatant in its inter-subjective connotations than Konvention. Furthermore, in discussing the adequacy of these labels, Wittgenstein maintains that grammatical rules are akin to social norms: not their history, but their status and function is that of conventions.72 And that alone counts in philosophy, at least according to Wittgenstein. The individualists might agree with Wittgenstein that it is arbitrary that we use a particular sound-pattern to mean something specific. Yet they could rightly insist that this does not entail that it is a matter of conventions. It is arbitrary what color tag I use to label my files for university administration, or whether before going out I start by putting on my left or my right shoe. Yet this does not indicate the existence of a shared convention. In fact, Davidson tries to exploit the arbitrariness of meaning against the idea that meaning is governed by such conventions. What is more, this exploitation seems to be licenced by Wittgenstein’s own central insights. Expressions do not have an intrinsic meaning. Rather, what they mean is determined by how they are used.

Meaning, rules, and conventions But let’s were, by scientific meaning

175

not forget that a word hasn’t got a meaning given to it, as it a power independent of us, so that there could be a kind of investigation into what the word really means. A word has the someone has given to it. (BB, p. 28)

But now the individualist argues as follows: if the meaning of an expression e in a language L is determined or constituted by the use speakers make of e, the meaning of e cannot at the same time prescribe a certain use to the speakers of L. Accordingly, the idea that the use of e is governed by lexical norms is incompatible with the idea that meaning is determined by use. It is a lapse into the Platonist picture according to which “the meanings of words are magically independent of the speaker’s intentions.”73 It would seem, however, that there is a straightforward answer to this complaint. We must distinguish between the use an individual speaker a makes of e, and the use that the linguistic community C makes of it. Communal use constitutes meaning, while individual use is responsible to it. By this token, the existence of semantic norms and their independence from the utterances and intentions of individual speakers is no more mysterious than the existence of legal norms and their independence from the acts and motives of individual agents. This kind of an answer can be found in the hermeneutic tradition. But it is also the answer implied by Wittgenstein’s conception of language as a communal practice governed by conventions, a practice into which children are induced by way of training and explanation. It emerges, therefore, that Wittgenstein’s reflections on the connection between meaning, rules and conventions are a lasting legacy in several respects. First, he put semantic normativity on the philosophical agenda. Second, he alerted us to various dimensions of semantic normativity (although he did not distinguish them as such). Third, his idea that meaning is antecedent to truth and his reflections on grammatical rules provide ways of explaining and defending not just the bare normativity claim, but also the claim that semantic normativity is rule-based and has a prescriptive dimension. Fourth, by criticizing the calculus-model of language he warns us against too rigid and exaggerated a conception of the normative aspects of language. Finally, this realistic conception of linguistic rules allows one to defend the idea that speaking a language is a communal practice against the animadversions of individualists.

Notes 1 Saul A. Kripke, Wittgenstein on Rules and Private Language: An Elementary Exposition (Oxford: Blackwell, 1982), p. 5. 2 Robert Fogelin, Wittgenstein (London: Routledge and Kegan Paul, 1976). 3 John McDowell, Mind, Value, and Reality (Cambridge, MA: Harvard University Press, 1998).

176

Hans-Johann Glock

4 Crispin Wright, Wittgenstein on the Foundations of Mathematics (London: Duckworth, 1980). 5 Gordon Baker and P. M. S. Hacker, Scepticism, Rules and Language (Oxford: Blackwell, 1984). 6 Alexander Miller and Crispin Wright, eds., Rule-Following and Meaning (Chesham: Acumen, 2002). 7 See Hans-Johann Glock, “The Normativity of Meaning Made Simple,” in Philosophy—Science—Scientific Philosophy, ed. C. Nimitz and A. Beckermann, Main Lectures and Colloquia of Fifth International Congress of the Society for Analytical Philosophy, Bielefeld, 22–26 September 2003 (Paderborn, Germany: Mentis, 2005); D. Whiting, “The Normativity of Meaning Defended,” Analysis 67:4 (2007): 133–40. 8 Friedrich Waismann, Logik, Sprache, Philosophie (Stuttgart: Reclam, 1976), pp. 184–85, 190–91. On the genesis of Logik, Sprache, Philosophie see the Nachwort by the editors. For further evidence that the dictations to Waismann are an accurate account of Wittgenstein’s quickly evolving views from the 1930s, see the editor’s preface in Ludwig Wittgenstein and Friedrich Waismann, The Voices of Wittgenstein: The Vienna Circle (London: Routledge, 2003). 9 PI §43. 10 Bertrand Russell, The Analysis of Mind, Muirhead Library of Philosophy (London: Allen & Unwin, 1921). 11 F. P. Ramsey, “Facts and Propositions,” in Philosophical Papers, ed. D. H. Mellor (Cambridge: Cambridge University Press, 1990). 12 C. K. Ogden and I. A. Richards, The Meaning of Meaning: A Study of the Influence of Language upon Thought and of the Science of Symbolism (London: K. Paul, Trench, Trubner, 1923). 13 PI §§493–98; Hans-Johann Glock, “Wie Kam die Bedeutung zur Regel,” Deutsche Zeitschrift für Philosophie 48 (2000): 429–47; P. M. S. Hacker, Wittgenstein: Mind and Will, vol. 4, An Analytical Commentary on the Philosophical Investigations (Oxford: Blackwell, 1996), ch. 1. 14 MWL, p. 51; PG §§62–64; OC §§61–62; LWL pp. 35–37. 15 See PI §108. 16 Paul Boghossian, “The Rule-Following Considerations,” in Rule-Following and Meaning, ed. Alexander Miller and Crispin Wright (Chesham, Bucks: Acumen, 2002), p. 148. 17 Paul Boghossian, “The Normativity of Content,” Philosophical Perspectives 13 (2003), p. 45, n. 4. 18 Paul Boghossian, “The Rule-Following Considerations,” p. 151. 19 Donald Davidson, “The Social Aspect of Language,” in The Philosophy of Michael Dummett, ed. Brian McGuinness and Gianluigi Oliveri, Synthese Library (Dordrecht: Kluwer, 1994), pp. 10–11. 20 Donald Davidson, “Reply to Akeel Bilgrami,” in Reflecting Davidson: Donald Davidson Responding to an International Forum of Philosophers, ed. Ralf Stoecker (Berlin: W. de Gruyter, 1993), p. 145. See also Akeel Bilgrami, “Norms and Meaning,” in Reflecting Davidson: Donald Davidson Responding to an International Forum of Philosophers, ed. Ralf Stoecker (Berlin: W. de Gruyter, 1993). 21 E.g., G. Forbes, “Scepticism and Semantic Knowledge,” in Rule-Following and Meaning, ed. Miller and Wright, p. 16; Boghossian, “The Rule-Following Considerations,” p. 144. 22 TLP 3.3; see Hans-Johann Glock, “Sense and Meaning in Frege and the Tractatus,” in From the Tractatus to the Tractatus (and Other Essays), ed. G. Oliveri, Wittgenstein Studies (Frankfurt am Main: Peter Lang, 2000). 23 E.g., BB, pp. 3–15, 130–32.

Meaning, rules, and conventions

177

24 R. Cartwright, “Propositions,” in Analytical Philosophy: First Series, ed. Ronald Butler (Oxford: Blackwell, 1962); Alan R. White, Truth (London: Macmillan, 1971), p. 14. 25 Boghossian, “The Normativity of Content,” pp. 38–39. 26 E.g., Boghossian, “The Rule-Following Considerations,” p. 144. 27 See, respectively, Glock, “The Normativity of Meaning Made Simple,” pp. 235– 40; Severin Schroeder, Das Privatsprachen-Argument (Paderborn, Germany: Schöningh, 1998), §21. 28 M. A. E. Dummett, Frege: Philosophy of Language (London: Duckworth, 1981), pp. 296–98. 29 PG, pp. 184, 193; PI §§491, 501. 30 E.g., Stanley Cavell, “The Availability of Wittgenstein’s Later Philosophy,” The Philosophical Review LXXI (1962): 67–93; Oswald Hanfling, Wittgenstein’s Later Philosophy (Basingstoke: Macmillan, 1989); Bede Rundle, Wittgenstein and Contemporary Philosophy of Language (Oxford: Blackwell, 1990), pp. 5–8, 33–34, 64–69. 31 See TLP 3.3 ff., 4.002, 5.4541, 5.5562. 32 See Hans-Johann Glock, A Wittgenstein Dictionary (Oxford: Blackwell, 1996), pp. 67–72. 33 PI §199. 34 OC §§61–62. 35 RFM, p. 330. 36 See Hans-Johann Glock, “Necessity and Normativity,” in The Cambridge Companion to Wittgenstein, ed. Hans D. Sluga and David G. Stern (Cambridge: Cambridge University Press, 1996). Cp. J. Kalhat, “Has the Later Wittgenstein Accounted for Necessity?,” Philosophical Investigations 31: 1 (2008): 1–23. 37 PG, p. 68; see also BT, pp. 424–25. 38 See also PI §§491–93. 39 AWL, pp. 47–48; Z §467. 40 Simon Blackburn, “The Individual Strikes Back,” Synthese (1984): 28–29. 41 See MS 106, p. 46. 42 PI §201. 43 Hans-Johann Glock, “The Normativity of Meaning Made Simple,” pp. 235–43. Cp. Gordon Baker and P. M. S. Hacker, Wittgenstein: Rules, Grammar and Necessity, vol. 2, An Analytical Commentary on the Philosophical Investigations (Oxford: Blackwell, 1985), pp. 52–58. 44 PG, p. 68, 143; MWL, p. 276. 45 BB, p. 4; LFM, p. 118. 46 BB, pp. 12, 90; PR, p. 78. 47 PI §§69–79. 48 PR, pp. 51–52; LWL, p. 8. 49 WVC, pp. 38, 61–63, 156; PR, pp. 143, 170, 216, 249; LWL, pp. 8, 55; PG, pp. 319, 347; RFM, p. 345. 50 E.g., K. Glüer, “Dreams and Nightmares. Conventions, Norms, and Meaning in Davidson’s Philosophy of Language,” in Interpreting Davidson, ed. Petr Koþtátko et al., CSLI Lecture Notes (Stanford, CA: CSLI Publications, 2001), p. 59. 51 Davidson, “Reply to Akeel Bilgrami,” p. 146; Glüer, “Dreams and Nightmares.” 52 AWL, p. 154. 53 G. H. von Wright, Norm and Action: A Logical Enquiry (London: Routledge & Kegan Paul, 1963), p. viii. 54 E.g., Donald Davidson, “The Social Aspect of Language,” p. 3. 55 Ibid., pp. 2, 12. 56 Hans-Johann Glock, Quine and Davidson on Language, Thought and Reality (Cambridge: Cambridge University Press, 2003), pp. 250–59. See also Schroeder, Das Privatsprachen-Argument, §§16–18.

178

Hans-Johann Glock

57 See K. Glüer, and Peter Pagin, “Rules of Meaning and Practical Reasoning,” Synthese 118 (1999): 208–27; Robert J. Stainton, Philosophical Perspectives on Language (Peterborough, ON: Broadview Press, 1996), pp. 139–42. 58 Baker and Hacker, Scepticism, Rules and Language, p. 259. 59 In the current context, the distinction was first drawn by John R. Searle, Speech Acts: An Essay in the Philosophy of Language (Cambridge: Cambridge University Press, 1969), ch. 2.5. But his formulations suffer from shortcomings highlighted by Joseph Raz, Practical Reason and Norms (Princeton, NJ: Princeton University Press, 1990), pp. 108–11. These shortcomings are not essential to the distinction as such. 60 David Ryan informed me that as far as games are concerned, what philosophers call regulative rules would not even be called rules, but would instead be referred to as strategies or tactics. In this context, he referred to the canonical Hoyle’s Book of Games, e.g., Lawrence Dawson and Edmond Hoyle, Hoyle’s Games Modernized (London: Routledge, 1934). 61 PR, p. 51; LWL, pp. 46–59; PG, pp. 60, 133, 143; PI §496. 62 See PG, pp. 184–85. 63 Von Wright, Norm and Action. 64 Joseph Raz, “Reasons for Action, Decisions and Norms,” in Practical Reasoning, ed. Joseph Raz, Oxford Readings in Philosophy (Oxford: Oxford University Press, 1978). 65 PR, p. 78; PG, pp. 88–91, 202; BB, pp. 12, 85–90, 109. 66 Schroeder, Das Privatsprachen-Argument, §12. 67 B. Smith, “Language, Conventionality of,” in Routledge Encyclopedia of Philosophy, ed. Edward Craig (London: Routledge, 1998). 68 See Hans-Johann Glock, Quine and Davidson on Language, Thought and Reality, pp. 250–59. 69 PG, pp. 184–85; see Hans-Johann Glock, A Wittgenstein Dictionary, pp. 45–50. 70 Gordon Baker, Wittgenstein, Frege and the Vienna Circle (Oxford: Basil Blackwell, 1988), ch. 7. 71 PI §355; AWL, pp. 89–90; PG, pp. 68, 190. 72 AWL, pp. 169–70. 73 Donald Davidson, “The Structure and Content of Truth,” Journal of Philosophy 87 (1990), p. 310.

8

Wittgenstein’s critique of referential theories of meaning and the paradox of ostension, Philosophical Investigations §§26–481 David G. Stern

8.1 Prologue Readers of Wittgenstein who take their orientation from reference works, elementary introductions or the canonical literature on his later work often take it for granted that there is relatively little positive philosophical argument in the Philosophical Investigations, and that the few main lines of argument—principally the rule-following considerations and the private language argument—play a purely negative, or critical, role. These, in turn, are typically construed as Wittgenstein’s principal weapons in a frontal attack on a single target, usually identified as the philosophy of the Tractatus, the Augustinian Picture, or the presuppositions that supposedly underlie traditional philosophy. However, reading Wittgenstein in this way, an approach that can find strong support in not only the work of most of Wittgenstein’s constructively critical readers, such as Strawson or Kripke, but also the work of some of his most well-known sympathetic expositors, such as Malcolm or Hacker, does not do justice to the range and variety of the arguments at work, and the positions under discussion, in the Philosophical Investigations. The targets Wittgenstein takes on are much more varied, and much less monolithic, than they are usually supposed to be. On the reading advocated here, both the “straight” and the “sceptical” solutions to the paradoxes of rule-following are equally misguided if they are presented as an exposition of the author’s overall argumentative intentions, for they both misunderstand the character and methods of the Philosophical Investigations. They mistakenly identify the viewpoint defended in a particular strand of argument—in one case, the reasons Wittgenstein’s narrator gives us for thinking that the problem of rule-following can be solved, in the other, a skeptical problem that the narrator claims the interlocutor faces—as equivalent to the views that are advocated by the author, or the book as a whole. Wittgenstein, I contend, provided neither a straight solution nor a skeptical solution to the philosophical problems discussed in the Philosophical Investigations; rather, he aimed to dissolve those problems, by means of a dialogue between opposing voices. However, to dissolve an argument is not to dismiss it; an successful dissolution requires an intimate knowledge of the

180

David G. Stern

argumentative terrain if one is to convince one’s readers of the limitations of the particular arguments one criticizes. Consequently, Wittgenstein’s dialogues explores a much wider variety of arguments, and argumentative strategies, than the very limited range of positions and approaches usually discussed in the expository literature. Like Plato’s dialogues, they are simultaneously literary and philosophical, and the reader’s initial impression that the speakers and positions under discussion can be identified without much trouble should not be taken at face value. We shall see that it is a third voice in that dialogue, one which responds to that debate without endorsing either side, that comes closer to expressing the author’s viewpoint than either of his leading protagonists do. The principal concern of the Philosophical Investigations is not the specific arguments about rule-following or private language that have been the focus of most interpreters’ attention, but rather the larger argumentative strategies that inform them, and the very wide variety of quite particular trains of argument that are provided throughout the book. Once we appreciate this, we will also see that Wittgenstein’s work is directly relevant to a much broader range of current concerns.

8.2 The “method” of §2 In the Philosophical Investigations, topics are repeatedly introduced in the following way. Stage 1. A brief statement of a philosophical position that Wittgenstein opposes, which usually emerges out of an exchange with another voice. Thus, in §1, we are presented with a conception of meaning that arises out of Wittgenstein’s reading of a passage from Augustine’s Confessions: “Every word has a meaning. The meaning is correlated with the word. It is the object for which the word stands (PI §1).2 Stage 2. The description of a quite specific set of circumstances in which that position is appropriate: That philosophical conception of meaning has its place in a primitive idea of the way language functions. But one can also say that it is the idea of a language more primitive than ours. Let us imagine a language for which the description given by Augustine is right. (PI §2) In §2 of the Philosophical Investigations, the passage just quoted leads in to the famous story of Wittgenstein’s builders, a tribe who only have four words, each of which is used by a builder to instruct his assistant to bring one of four kinds of building blocks. Stage 3. The deflationary observation that the circumstances in question are quite limited, and that once we move beyond them, the position becomes inappropriate:

Philosophical Investigations §§26–48

181

Augustine, we might say, does describe a system of communication; only not everything that we call language is this system. And one has to say this in many cases where the question arises “Is this an appropriate description or not?” The answer is: “Yes, it is appropriate, but only for this narrowly circumscribed region, not for the whole of what you were claiming to describe.” (PI §3) To drive the point home, Wittgenstein later adds other uses of signs that don’t fit Augustine’s description: §8 describes an expansion of the language in §2 to include numerals, demonstratives and color samples, and §15 adds names for particular objects. This three-stage argument scheme suggests a more general recipe for unsettling philosophical preconceptions. First, describe a case the preconception fits as well as possible, “a language-game for which this account is really valid,” then change just enough about the case in question, either by adding or removing some aspect, or by changing the context or our point of view, so that we run up against the limitations of the preconception.3 This method of §2, as Wittgenstein calls it in §48, is used repeatedly in the remarks that follow. It is also characteristic of Wittgenstein’s use of this argument-scheme that all three stages follow each other so quickly. In §§1–3 and §§46–48, each stage of the argument is presented quite explicitly; in many other cases the argument is only sketched, and Stage 3 may be left as an exercise for the reader. Because he aims, not to solve philosophical problems, but to undo, or dissolve them, Wittgenstein frequently presents the materials for a Stage 3 reply immediately before setting out Stage 2. The aim of the reply in Stage 3 is not to articulate a philosophical answer to the proto-philosophical question with which we began, but to get us to give up the question. The story of the grocer and his different ways of using words in §1d plays this role in the argument of §§1–3. Similarly, §47’s multiplication of examples of alternative conceptions of complexity comes between Socrates’ Stage 1 discussion of simples, the “primary elements” out of which the world is made in §46, and the use of the method of §2 in §48 to attack the very idea of a primary element. One might reply, in defense of the first moves toward philosophical theorizing Wittgenstein is criticizing, and the theories they give rise to, that the approach to philosophy he opposes aims at a “view from nowhere,” a position that is correct for all possible circumstances and contexts, not just a position that fits a few carefully selected cases. In defense of Wittgenstein’s method of §2, one can say that if we grant, for the sake of argument, that such philosophical accounts do work at all, they must be applicable to specific cases, and ultimately these must include not only the best cases, but the problem cases, too. Furthermore, Wittgenstein will suggest, the “view from nowhere” is a distinctively philosophical fiction, a fiction that always starts out from a quite specific somewhere, and begins its theorizing with particular examples of familiar objects and activities. The philosophy Wittgenstein

182

David G. Stern

takes as his target begins, in other words, with our taking familiar matters out of context, and taking them as the model for a universal account, true everywhere and at any time, of how things must be. Confronted with §2-type examples, the Socratic philosopher dismisses the concrete cases as irrelevant, insisting that what matters is to get clear about the rules that determine which cases the term really applies to, and what they have in common. In 1944, when Wittgenstein was putting the first part of the Philosophical Investigations into its final form, he told a friend that he was reading Plato’s Theaetetus, and that “Plato in this dialogue is occupied with the same problems that I am writing about.”4 Wittgenstein owned a fivevolume German translation of Plato by Preisendanz, and refers to passages in Plato quite frequently in his writings. The philosophical discussion in the Theaetetus begins with Socrates’ asking Theaetetus “what is knowledge?” His first answer is as follows: TH.:

I think the things Theodorus teaches are knowledge—I mean geometry and the subjects you enumerated just now. Then again there are the crafts such as cobbling, whether you take them together or separately. They must be knowledge, surely. SOC.: That is certainly a frank and indeed a generous answer, my dear lad. I ask you for one thing and you have given me many; I wanted something simple, and I have got a variety. [ … ] You were not asked to say what one may have knowledge of, or how many branches of knowledge there are. It was not with any idea of counting these up that the question was asked; we wanted to know what knowledge itself is.—Or am I talking nonsense?5 We can see much of Wittgenstein’s later philosophy as an extended defense of Theaetetus’ initial answer—the best we can do in answering questions about the essence of a word such as “knowledge” is to give examples, with the aim of showing that Socrates is talking nonsense, and so “bring words back from their metaphysical to their everyday use.”6 In the Blue Book, Wittgenstein explicitly opposes his approach to Socrates’: When Socrates asks the question, “what is knowledge?” he does not even regard it as a preliminary answer to enumerate cases of knowledge [ … ] the discussion begins with the pupil giving an example of an exact definition, and then analogous to this a definition of the word “knowledge” is asked for.7 As the problem is put, it seems that there is something wrong with the ordinary use of the word “knowledge.” It appears we don’t know what it means, and that therefore, perhaps, we have no right to use it. We should reply: “There is no one exact usage of the word ‘knowledge’; but we can make up several such usages, which will more or less agree with the ways the word is actually used.” (BB, pp. 20, 27)

Philosophical Investigations §§26–48

183

On the other hand, there are also deep affinities between Wittgenstein’s and Plato’s dialogues: each of the definitions of knowledge Socrates proposes in the Theaetetus proves unsuccessful. The three-stage argument of §§1–3 marks the beginning of an extended critique of the notion, prominent in §1, that every word has a meaning, the object for which it stands. This critique, which occupies much of §§1–64, can be divided into two distinct units. The first unit, §§1–38, is an attack on the idea that the meaning of a word consists in its standing for a familiar object of one kind or another. One of the principal approaches under discussion in §§1–38 is the view that ostensive definition—explaining a word’s meaning by pointing at an object—is the basis of meaning. The second unit, §§39–64, looks at an approach that promises to avoid some of the difficulties that have arisen in §§1–38, the view that words stand for simple objects. §§39–45 consider the possibility that familiar things could be simples; §§47–64 examine the notion that ultimately words get their meaning by standing for simple objects, objects that we reach by analyzing familiar objects, which are complex, into their ultimate, primitive components, or “ur-elements,” on an overly literal translation. A three-stage argument starts from questions about something apparently un-philosophical—language learning, or giving a name to a thing; but this draws the narrator’s alter ego into Stage 1, a preliminary formulation of a philosophical thesis. Stages 2 and 3 bring us back to earth, by first proposing as prosaic and simple an example as possible, and then pointing out its limitations. However, the method of three-stage argument can only get us so far; replying to Socrates with a list of different kinds of knowledge, or the denotationalist with a list of different uses of words, may perhaps make us suspicious of the idea that we must be able to give a unitary specification of knowledge, naming, or the use of words, but it runs the danger of giving the impression that the only thing that is wrong with such theories is that they need refining. For this reason, Wittgenstein’s narrator replies to many different Socratic lines of thought in §§1–38 and §§65–88. However, the problems and paradoxes that Wittgenstein’s narrator produces provoke his alter ego to dig deeper, and look for something hidden, a structure that supposedly underlies, or a hidden process that somehow animates, our everyday lives and language. Wittgenstein speaks of this movement, from the Socratic questions that typically initiate philosophical inquiry, to the counter-examples and paradoxes that such questions inevitably produce, and from there to the Platonic vision of a reality behind the phenomena as a “tendency to sublime the logic of our language.”8 For this reason, the Socratic questions about naming and reference in the opening sections of the Philosophical Investigations lead up to the formulation of a paradox about ostensive definition: “an ostensive definition can be variously interpreted in every case.”9 This results in a discussion of “sublime names”: names that must stand for their objects, and so cannot be variously interpreted.10

184

David G. Stern

8.3 Seeing the Philosophical Investigations as a dialogue A key paradox that occurs over and over again, in one form or another, throughout the first 200 sections of the Philosophical Investigations is that nothing is intrinsically meaningful, for all determination of meaning, by such means as definitions, rules, thoughts, or images, is dependent on interpretation. Given any candidate meaning-determiner, it is always, in principle, open to a further, deviant, interpretation. No act of defining or intending, grasping a rule or deciding to go on in a certain way, can give the supposed meaning-determiner the power to determine our future actions, because there is always the question of how it is to be interpreted. Only if we ignore the context can we think that some isolated act or event can have a determinate meaning regardless of its context. A change in the context of application can yield a change in meaning, and therefore meaning cannot be identified with anything independent of context. Leading examples include the wayward child who learns to add small numbers correctly but systematically miscalculates, all the while insisting that he is going on the same way;11 the drawing of an old man walking up a steep path, resting on a stick, that a Martian might describe as a man sliding downhill, and deviant ways following arrows or signposts;12 or interpreting a drawing of a cube.13 The resolution of these paradoxes, like the resolution of a three-stage argument, turns on considering the wider context in which our words are used. In response to a proto-philosophical theory, a three-stage argument draws our attention to the circumstances it fits, and those it does not fit. Similarly, in replying to a philosophical paradox, Wittgenstein’s narrator points out that the paradox does not arise in our everyday lives, and draws our attention to the way in which it turns on a failure to pay attention to the circumstances in which those words are ordinarily used. Ordinarily, the paradox does not arise, because it is already clear how the words or actions in question are to be understood. Thus Wittgenstein’s reply to the paradox of ostension turns on the point that “ostensive definition explains the use— the meaning—of the word if it is already clear what kind of role the word is to play in language.”14 A parallel paradox about explanation—that any explanation of the meaning of a word hangs in the air unless supported by another one—receives a similar response: “One explanation may rest on another explanation which has been given, but no explanation needs another—unless we need it to avoid a misunderstanding.”15 Analogous paradoxes about understanding, interpreting, and following a rule are the principal concern of §§134–242; by the time Wittgenstein sums up the paradox of rule-following and provides his response in §201, he expects the paradox to be so familiar that his treatment takes the form of a summary that begins by referring back to the previous discussion. The previous paragraph attributes a clear-cut set of answers to the paradoxes of ostension, explanation, and rule-following to Wittgenstein. There can be no denying that these answers are present in the text; they can be

Philosophical Investigations §§26–48

185

compared to a prominent and repeated pattern in the weave of the Philosophical Investigations’ argumentative fabric. Most interpreters attribute this argumentative strand to Wittgenstein without any pause. But the connection between this train of thought and the author’s intentions is far from clear and so I will usually qualify this by attributing them to Wittgenstein’s narrator. If we take Wittgenstein’s narrator to be a behaviorist, or an ordinary language philosopher who maintains that the rules of our language guarantee that we are mostly right, then the skeptical paradoxes—namely, that ostension, explanation, and rule-following can always be undermined by sceptical possibilities—receive what Kripke calls a “straight” solution: we really can provide a positive answer to the paradoxes, because the expressions in question can be defined in terms of public behavior, or the rules of grammar that govern our use of language. If, on the other hand, we follow Kripke in taking Wittgenstein to be a sceptic who endorses the paradoxes he has formulated, then the appeal to what the community ordinarily does in its use of these terms is only a negative answer to the sceptical problem (Kripke calls this a “sceptical” solution): recognizing that we cannot solve the problem, we instead appeal to what we ordinarily do as a way of indicating the best reply available, albeit one that does not really solve the paradoxes. Rather than seeing these arguments as exchanges between Wittgenstein and his interlocutor, I propose that we approach them as an exchange between a number of different voices, none of which can be unproblematically identified with the author’s. For these reasons, in discussing passages of dialogue in the Philosophical Investigations, I prefer to speak of dialogues between Wittgenstein’s narrator and an interlocutory voice, rather than Wittgenstein and the interlocutor. In some places, and particularly in those parts of the text that are most critical of the Tractatus—principally §§39–142—the narrator’s voice sets out the case against philosophical positions set out in Wittgenstein’s first book, in opposition to voices that express Tractarian convictions. In §§140–693, the narrator’s voice is frequently used to set out behaviorist, verificationist, and anti-essentialist objections to traditional philosophical views, in opposition to an anti-behaviorist voice that expresses mentalist, verification-transcendent, and essentialist intuitions and convictions. In addition to these opposing voices, voices that play different parts at different points in the text, we also meet with a third voice. This third voice, which is not always clearly distinct from the narrator’s voice, provides an ironic commentary on their exchanges, a commentary consisting partly of objections to assumptions the debaters take for granted, and partly of platitudes about language and everyday life they have both overlooked. Most readers treat both of these voices as expressions of Wittgenstein’s views, with the result that they are unable to reconcile the trenchant and provocative theses advocated by the narrator and the commentator’s rejection of all philosophical theses.

186

David G. Stern

The following passage provides a convenient summary of the commentator’s approach, not only to disputes over realism and idealism, but also to the exchanges between narrator and interlocutor, the voice of correctness and the voice of temptation: For this is what disputes between idealists, solipsists, and realists look like. One side attacks the normal form of expression as if they were attacking a statement; the others defend it, as if they were stating facts recognized by every reasonable human being. (PI §402b) Most of the Philosophical Investigations does consist of a debate for and against “the normal form of expression.” The narrator is usually taken to be arguing for Wittgenstein’s own philosophical position, ordinary language philosophy, while the interlocutor attacks our ordinary way of speaking, arguing that it does not do justice to his intuitions and his arguments. While it is certainly possible to construe large swatches of the Philosophical Investigations as exchanges between Wittgenstein and his interlocutor, or a doctrinaire behaviorist and a querulous anti-behaviorist, we should not identify the outlook of the author with every passage that we attribute to his leading narrator. The closest the author of the Philosophical Investigations comes to expressing his own views is not in the person of his narrator, the aggressively anti-Socratic protagonist we meet in the book’s three-step arguments, but rather in the moments when he steps back from this serio-comedy and offers us a striking simile, or draws our attention to platitudes that philosophers don’t take seriously. For this reason, all this talk of Wittgenstein’s argument, and of the positions he opposes, while unavoidable, is potentially deeply misleading. For it implies that he thinks of the views he is opposing as intelligible, albeit mistaken. It also makes it tempting to suppose that Wittgenstein’s distinctive contribution to philosophy turns on a clear distinction between unproblematic, everyday uses of language, and their mirror image, the metaphysical uses of language that are characteristic of traditional philosophy. However, if Wittgenstein is correct, the accounts offered by all the participants in his dialogues, although they certainly appear at first sight to make good sense, are actually “disguised nonsense,” and so cannot, in the end, be true or false.16 Ultimately, Wittgenstein’s aim is to get us to recognize that the proto-philosophical accounts of meaning and mind that his interlocutor proposes, and his narrator opposes, only appear to make sense, and that neither the descriptions of simple situations his narrator offers in Stage 2, nor the sublime truths about the essence of the world and language his interlocutor aims for, will do justice to those ideas and intuitions with which philosophical discussion begins. On this reading, Wittgenstein is neither saying that a solution to the skeptical paradoxes, or a private language is possible, nor proving that such things are impossible. Rather, Wittgenstein holds that such words do no useful work at all:

Philosophical Investigations §§26–48

187

What we “are tempted to say” in such a case is, of course, not philosophy; but it is its raw material. Thus, for example, what a mathematician is inclined to say about the objectivity and reality of mathematical facts, is not a philosophy of mathematics, but something for philosophical treatment. The philosopher treats a question; like an illness. (PI §§254–55*) If we follow the author’s advice, rather than those of the protagonists in his dialogues, we will give up both behaviorism and anti-behaviorism. The result of his discussion of philosophical problems is not supposed to be an endorsement of one of the views he discusses; rather, “a combination of words is being excluded from the language, withdrawn from circulation.”17 Nevertheless, in order to “pass from a piece of disguised nonsense”—such as the initial expression of a philosophical account of meaning we find in §1—“to something that is patent nonsense,” we must first try to make sense of it, and in so doing, come to see that we cannot.18 There are few better ways of beginning to do this than to try to think of cases the proposed account does fit as well as possible, and then seeing how it fails to fit when the context or circumstances change. Wittgenstein sums up this predicament and his response to it in the following words: The great difficulty here is not to represent the matter as if there were something one couldn’t do. As if there really were an object, from which I derive its description, but I were unable to show it to anyone.—And the best that I can propose is that we should yield to the temptation to use this picture, but then investigate how the application of the picture goes. (PI §374)

8.4 The paradox of ostensive definition: §§26–38 The method of §2, the three-stage argument, and the notion of a language-game first emerge in the early 1930s, and reach their high point in the Brown Book, where each of the 73 numbered remarks in Part I introduces another languagegame. The central idea that motivates this method is that of drawing our attention to the context in which our use of language takes place. In some cases that context consists of the rules for the use of certain terms, and in these cases there is little difference from the method of paradox; but often, and especially in work after the early 1930s, the context is a much broader one. Certainly, there is no clear distinction between the notion of a calculus, a formal system of rules, and a language-game in Wittgenstein’s work from the early 1930s, and at first the two expressions were used interchangeably. However by the time he wrote the early version of the Philosophical Investigations, the two terms are quite clearly contrasted with each other: “in philosophy we often compare the use of words with games and calculi which have fixed rules, but cannot say that someone who is using language must play such a game.”19

188

David G. Stern

In §6, where Wittgenstein’s narrator asks us to imagine a tribe whose whole language consists in the game of §2, he suggests that the teaching of the words for the blocks will presumably consist in such activities as the teacher’s repeatedly pointing to the various kinds of blocks and saying the appropriate words. However, he observes ostensive definition is not yet possible for a child learning the vocabulary of §2: “I do not want to call this ‘ostensive definition,’ because the child cannot as yet ask what the name is.”20 That words have the meaning they do is never simply the product of a single connection between word and thing, for even when one makes such a connection, its effect depends on the larger context: But if the ostensive teaching has this effect,—am I to say that it effects an understanding of the word? Don’t you understand the call “Slab!” if you act upon it in such-and-such a way?—Doubtless, the ostensive teaching helped to bring this about; but only together with a particular training. With different training the same ostensive teaching of these words would have effected a quite different understanding. “I set the brake up by connecting up rod and lever.”—Yes, given the whole of the rest of the mechanism. Only in conjunction with that is it a brake-lever, and separated from its support it is not even a lever; it may be anything, or nothing. (PI §6c–d) §8 introduces a new language-game by making additions to the game described in §2: words for numerals, colors, “this” and “there.” The subsequent discussion emphasizes the great differences between these words, and the even greater differences between the variety of words commonly used in everyday language. Wittgenstein’s narrator stresses how the differences between these words are not simply a matter of the different ways in which they are used, but also the larger context of use and training within which they are located. One of the morals of the intricate discussion in §§19–20 of the question whether “the call ‘Slab!’ in example (2) [is] a sentence or a word” is that the answer depends on the resources of the language in question, and the interests and concerns of the speaker. Only after an extended introductory discussion of the diversity and multiplicity of language, is Wittgenstein ready to turn to the topic of ostensive definition.21 §15 first introduces an extension of the game of §2, which includes names for particular objects, by putting distinctive marks on certain objects; the builder can then use those marks as a way of getting his assistant to bring the object: It is in this and more or less similar ways that a name denotes and is given to a thing.—It will often prove useful in doing philosophy to say to ourselves: naming something is like attaching a label to a thing. (PI §15*)

Philosophical Investigations §§26–48

189

However, this train of thought is allowed to lapse until §26, when it is explicitly taken up again, in the context of a discussion of the view that learning language consists in naming objects, where “object” covers such diverse entities as people, shapes, colors, pains, moods, numbers, and so forth. Wittgenstein writes: “To repeat—naming is something like attaching a label to a thing. One can say that this is preparatory to the use of a word. But what is it a preparation for?” (PI §26). An interlocutory voice immediately answers this question at the beginning of §27, in a tone that implies the answer is entirely unproblematic: “We name things and then we can talk about them: can refer to them in talk.” This provides the jumping-off point for a correction by Wittgenstein’s narrator: —As if what we did next were given with the mere act of naming. As if there were only one thing called “talking about a thing”. Whereas in fact we do the most various things with our sentences. Think of exclamations alone, with their completely different functions. (PI §27a) The first point the narrator makes here is skeptical: that, by itself, the act of naming settles nothing, for much depends on what happens afterwards (and before). The second point is that we have many ways of talking about things, and many ways of talking that are not a matter of talking about things at all. But Wittgenstein’s principal point at this stage in his discussion of naming is neither the multiplicity of ways we talk about objects, nor the many ways in which we do things with words, but how these various uses of words depend on a taken-for-granted context. §§26–28 build up to Wittgenstein’s paradox of ostensive definition: “an ostensive definition can be variously interpreted in every case.” Given any definition of a word that involves pointing to something, one can always come up with more than one interpretation of the definition, because there will be more than one way of interpreting the act of pointing. The dialogue begins with a platitude and puzzle that could have been spoken by the Queen in Alice Through the Looking Glass: Now one can ostensively define a proper name, the name of a colour, the name of a material, a numeral, the name of a point of the compass and so on. The definition of the number two, “That is called ‘two’”—pointing to two nuts—is perfectly exact. (PI §28) The sensible voice that replies to the puzzle fails, as Alice usually does, to see the point: “But how can two be defined like that? The person one gives the definition to doesn’t know what one wants to call ‘two’; he will suppose that ‘two’ is the name given to this group of nuts!” (PI §28a).

190

David G. Stern

This sets the Queen up for her claim that the puzzle is quite general: the problem doesn’t only arise for the unusual case of ostensively defining a number, but even for the most seemingly straightforward cases of ostensive definition, such as naming objects or people: —He may suppose this; but perhaps he does not. He might make the opposite mistake; when I want to assign a name to this group of nuts, he might understand it as a numeral. And he might equally well take the name of a person, of which I give an ostensive definition, as that of a colour, of a race, or even of a point of the compass. That is to say: an ostensive definition can be variously interpreted in every case. (PI §28a) At this point, the Queen makes the next move on Alice’s behalf: “Perhaps you say: two can only be ostensively defined in this way: ‘This number is called “two”’” (PI §29). This is one of the ways that Wittgenstein’s dialogues frequently segue into a monologue of sorts: the voice criticizing a philosophical position or intuition begins to say what the philosopher will say next. Here, this is only an opportunity for the Queen to spell out the consequences of what has just been said: For the word “number” here shows what place in language, in grammar, we assign to the word. But this means that the word “number” must be explained before the ostensive definition can be understood.—The word “number” in the definition does indeed show this place; does show the post at which we station the word. And we can prevent misunderstandings by saying: “This colour is called so-and-so”, “This length is called so-and-so”, and so on. That is to say: misunderstandings are sometimes averted in this way. (PI §29a) At first it may look as if the Queen has conceded too much to Alice: hasn’t she just said that we can settle how an ostensive definition is to be understood, by specifying the appropriate category or type the name belongs to? In practice, she concedes, misunderstandings are sometimes averted in this way. In principle, however, the problem of multiple interpretations has not gone away, for the same questions can always be raised about the classifying terms that are used in specifying the word’s post: But is there only one way of taking the word “colour” or “length”? Well, they just need explaining. —Explaining, then, by means of other words! And what about the last explanation in this chain? (Do not say: “There isn’t a ‘last’ explanation”. That is just as if you chose to say: “There isn’t a last house in this road; one can always build an additional one”.) (PI §29a*)22

Philosophical Investigations §§26–48

191

This should remind us of the end of §1: “explanations come to an end somewhere.” We can give explanations, but somewhere, usually sooner rather than later, our explanations will give out. The beginning of §30 provides a preliminary summary of the moral to be drawn: So one might say: the ostensive definition explains the use—the meaning—of a word when the overall role of the word in language is clear. Thus if I know that someone means to explain a colour-word to me, the ostensive explanation “This is called ‘sepia’” will help me to understand the word.—And you can say this, so long as you don’t forget that all sorts of problems attach to the word “to know” or “to be clear.” One has already to know (or be able to do) something in order to be capable of asking a thing’s name. But what does one have to know? (PI §30*) In a way, the question at the end of §30 is a restatement of the questions that opened this part of the discussion in §26. A minimal answer here would be the one canvassed above: we have to know what kind of thing you are talking about—a number, a color—its “place in language, in grammar” as Wittgenstein puts it in §29. At this point in the discussion, Wittgenstein’s narrator is particularly didactic and insistent, devoting §31 to answering the question posed at the end of §30, by way of a discussion of a simple example taken from chess. Chess is one of Wittgenstein’s favorite objects of comparison, a familiar game that can “throw light on the facts of our language by way not only of similarities, but also of dissimilarities.”23 Like many of Wittgenstein’s imaginary language-games, chess is a game that has a clearly stated set of rules, but one can learn the game without explicitly stating them. §31 considers a variety of cases in which someone might be shown a chesspiece and told “This is the king,” and uses this as a way of asking what you have to know before I can explain the use of a chess-piece to you by pointing to it and saying “This is the king.” In the first paragraph, Wittgenstein describes a straightforward, if unlikely, scenario in which you have explicitly learned the rules, but have not yet been told what a king looks like. The second paragraph considers the case in which you pick up the game without ever hearing or stating the rules. Here, you would have learned what a king looks like as you picked up the game, but perhaps because I own one of those novelty chess sets, I might need to tell you which piece in this set is the king. In either case, the explanation can only work if it connects my previous knowledge with the current circumstances. This explanation again only tells him the use of the piece because, as we might say, the place for it was already prepared. Or even: we shall only say that it tells him the use, if the place is already prepared. And in this case it is so, not because the person to whom we give the explanation

192

David G. Stern already knows rules, but because in another sense he has already mastered the game. (PI §31b)

In both of these cases, the words “This is the king” only do their work because I already know the rules of the game, and only need to find out which chess-piece is the king. Finally, Wittgenstein turns to the case where someone is learning how to play chess, and the teacher uses “This is the king” to introduce the piece by pointing to it, indicating how it moves, and so forth. Here, at last, one might think, the words do more work than in the previous cases. However, Wittgenstein observes that this case, just like the others, presupposes a great deal of prior knowledge, knowledge of how games are played and taught, if not knowledge of chess itself: In this case we shall say: the words “This is the king” (or “This is called the ‘king’”) explain the use of the words “the king” only if the learner already “knows what a piece in a game is.” That is, if he has already played other games, or watched “with understanding” other people playing—and similar things. Further, only under these conditions will he be able to ask relevantly in the course of learning the game: “What do you call this?”—that is, this piece in a game. We may say: only someone who already knows how to do something with it can significantly ask a name. (PI §31cd*) To sum up: ostensive explanation cannot be the foundation for learning a first language, because ostensive explanation presupposes a knowledge of how names work, and more generally, a grasp of their place in language, a grasp which will include familiarity with how they are used in a variety of cases. On the other hand, ostensive explanation is likely to be a large part of language learning for someone learning a second language, for here the preconditions for ostensive explanation are already in place. In §32, this leads Wittgenstein back to Augustine, and the suggestion that the Augustinian conception of language learning is modeled on the way we learn a second language, not a first. With the statement of the paradox of ostension, and its connection, in §32, with the issues raised at the very beginning of the book, the main business of the first half of the first “chapter” of the Philosophical Investigations has been completed: we have been led from the initial expression of philosophical intuitions and proto-theories about language to a clear statement of one of the paradoxes that arise if such trains of thought are pursued. §§33–35 explore a variety of possible responses to this paradox, all of which turn on the idea that something in the speaker’s or hearer’s mind allows him or her to pick out just what’s intended—such as guessing correctly, attending to the correct aspect, having a certain characteristic experience. In each case,

Philosophical Investigations §§26–48

193

Wittgenstein’s narrator points out that the supposedly disambiguating experience won’t do the job, for such experiences are neither necessary—for we can think of cases in which the experience in question occurs, but we don’t pick out what’s intended—nor sufficient—for we can think of cases where the experience in question doesn’t occur, but we do pick out what’s intended. Once again, this is a further application of the familiar three-step argument we met in the first ‘chapter’ of the Philosophical Investigations. The point of the train of argument in §§33–35 partly lies in the details, in the careful consideration of the very wide range of cases in which we may attend to a color, or a shape. But it also serves to lay the groundwork for the next stage of Wittgenstein’s discussion of naming. §§36–37 sum up the principal negative and positive conclusions that emerge from what has been said so far. Negative: when we are unable to find any one bodily action that guarantees one is pointing to a shape, rather than a color, we find ourselves thinking that something in the mind, something spiritual, must do the trick. Positive: the relation between name and thing named is in clear view, and consists of a variety of quite familiar activities that we engage in when using names. Wittgenstein is quite happy to refer to both physical activities, such as writing the name on the thing, or saying the name when pointing at it, as well as mental activities, such as calling the picture of the object to mind. He is not denying that we do call pictures of objects to mind, or that the particular ways of looking and attending discussed in §§33–35 do occur, and can help us identify what we’re talking about; what he is denying is that the mind has the peculiar, quasi-magical ability to disambiguate, to guarantee that words and things are rigidly attached to each other, in the way that the paradox of ostension can lead us to wish for. §§36–37 thus wrap up the discussion of the interlocutor’s initial attempts to overcome the paradox of ostension, and lead in to §38, which lays the ground for the discussion of the strategies philosophers turn to when they fail to identify everyday solutions to their problems in understanding how names work and so are driven to postulate spiritual solutions. Wittgenstein speaks here of a “tendency to sublime the logic of language” (PI §38b) and says “philosophical problems arise when language idles” (PI §38d*).24 Philosophers have often found this diagnosis puzzling, or unsatisfying: why, they ask, does Wittgenstein regard philosophical uses of language as illegitimate? Why can’t we simply say that the philosopher’s use of language is another language-game, no less legitimate than the ones Wittgenstein offers as examples? The nub of Wittgenstein’s reply is that the philosophical uses of language he opposes are not really uses of language at all, and that appealing to philosophers’ intuitions about the use of language is like appealing to our fantasies, or our dreams, if we want to understand our relation to the world around us. While there are places where Wittgenstein’s voice of correctness does lay down the rules of grammar, and says that philosophical mistakes consist in breaking them, this is not the considered position of the author of the Philosophical Investigations. Talk of language’s idling, or going

194

David G. Stern

on holiday, should not be understood as a matter of breaking the rules of sense, or offending against certain pragmatic principles, but rather as summing up the way philosophical language can conjure up an illusion of sense, a fantasy that creates the appearance of solidity. Wittgenstein supports this diagnosis by a detailed account of how we go wrong, how we may be led from the desire to specify what naming consists in to the Russellian illusion that we best capture the essence of naming when I apply the word “this” to an object directly in the centre of my visual field. Failing to identify a single uniform connection in plain view that links word and thing, we are misled into thinking that there must be something “spiritual,” a mental or intellectual process that does the job. But this “spiritual” process is, as Wittgenstein puts it, a “queer conception,” for it is an unexplained explainer that has been introduced to fill a gap of our own making. One way of understanding Wittgenstein’s talk of subliming here is as a way of talking about the elevated and mysterious character of the appeal to spirit. One can also draw a connection here with the Romantic notion of the aesthetic sublime, a realm above or beyond the ordinary that only the gifted visionary can reach; certainly the suggestion that the philosopher is reaching for a spurious profundity is appropriate.25 However, the word Wittgenstein uses here is sublimieren. While it does, like the cognate English word, mean to elevate, or to purify, German, unlike English, has a different word for the romantic sublime, namely erhaben, which would have been a much more natural choice if Wittgenstein had meant to emphasize this connection. There is a stronger connection here with the notion of subliming as a physical process, one that Wittgenstein, as a trained engineer, would have found quite familiar: the direct transition from a solid to a gas, without a liquid phase—a good example would be the clouds produced when dry ice (frozen carbon dioxide) vaporizes. One way of reading this talk of subliming our language is to take it to be about a mistaken attempt to purify, or refine, the motley material of our everyday activities into something pure and simple, much as a distillation extracts pure alcohol from a fermented liquid, or pure crystals can be formed directly from a cloud of vapor.26 This reading enables us to draw a connection with the talk of the interlocutor’s requirement of “crystalline purity.”27 A related, but rather different possibility is that the transition in question is not from (impure) solid to (pure) solid, by way of a gaseous process of distillation, but rather from something everyday and solid to something cloudy and diffuse.28 On this reading, Wittgenstein is characterizing what the philosopher does as a matter of turning something solid—our ordinary use of words, the “working of language”—into a gas— the mysterious spiritual processes supposedly at work, “a haze which makes clear vision impossible.”29 This is connected with his disparaging parenthetic reference in §109 to “the conception of thought as a gaseous medium.” Wittgenstein’s talk of condensing “a whole cloud of philosophy into a drop of grammar” would then be a way of summing up the reverse process that takes us away from these cloud-castles and back to something more

Philosophical Investigations §§26–48

195

definite.30 The “distillation” reading is an apt summary of what the interlocutor is trying to do; the “gaseous” interpretation captures the narrator’s assessment of what the interlocutor actually does.

8.5 Subliming names: §§39–48 In §39 and the sections that follow, the Stage 1 formulation of the position under attack is “A name ought really to signify a simple.”31 §39 opens the discussion by sketching an argument for the conclusion that a name must correspond to an object in order to have a meaning, an object that must exist for the name to have a meaning. The motivation for saying that “[a] name ought really to signify a simple” is that otherwise it would be contingent whether a given proposition makes sense or not, for it might be the case that the things it appears to refer to don’t actually exist. §39 sets out this Tractarian line of reasoning: if the object a name refers to doesn’t exist, then the name would have no meaning. But the proposition does have meaning, and so the object in question must exist. However, §40 replies that this confuses the meaning of the name with the bearer: when Mr. N.N. dies, his name is still meaningful. In other words, this account manifestly fails to fit the way we use names of everyday people and equipment, which still have a meaning when their object ceases to exist. §§41–42 point out that we can construct language-games that follow the Tractarian model, but that this is only one possibility. This problem leads to the idea that we must first identify those names that do stand for simples. The critique of this idea by Wittgenstein’s narrator involves at least two applications of the three-stage argument. In the first he examines the idea that ordinary names signify simples;32 in the second he turns to the view that we must first analyze ordinary names into real names, names that signify primary elements, simple objects that must exist.33 This movement back and forth between turning to familiar words and objects as the terminus of theorizing and looking for something hidden and unfamiliar is a recurrent feature of the Philosophical Investigations. Pears aptly sums up this aspect of the interlocutor’s predicament as follows: [T]here are two equally slippery paths leading to the limbo in which so many philosophical theories terminate: one is to postulate something remote and recondite to play an impossible role, and the other is to give an equally impossible role to some perfectly familiar kind of thing. [ … ] They both make the same impossible demand: when a word is attached to a thing, it must immediately slot into a pre-existing grid or lock on to rails extending indefinitely into the future.34 §41 begins a reply to the idea that ordinary names signify simples: we are asked to imagine an extension of the language-game of §2 that includes names, and to consider the case where the builder calls out the name of a

196

David G. Stern

broken object. What will the assistant do? First, we get a Stage 2 response, a description of a situation that fits the position under discussion: Well, perhaps he will stand there at a loss, or show A the pieces. Here one might say: “N” has become meaningless; and this expression would indicate that the sign “N” no longer had a use in our language-game (unless we gave it a new one). “N” might also become meaningless because, for whatever reason, the tool was given another name and the sign “N” no longer used in the language-game. (PI §41a) Stage 3 follows immediately—the narrator describes a different case, which the stage 2 interpretation does not fit: —But we could also imagine a convention whereby B has to shake his head in reply if A gives him the sign belonging to a tool that is broken.—In this way the command “N” might be said to be given a place in the language-game even when the tool no longer exists, and the sign “N” to have meaning even when its bearer ceases to exist. (PI §41) However, the defender of the view that names must stand for simples will reply that this is just shadow boxing: if we imagine a case where a word has meaning when the bearer ceases to exist, then that just goes to show that the word wasn’t really a name after all. A real name signifies something that is not composite, something absolutely simple. In §518, Wittgenstein quotes a closely related passage from the Theaetetus in which Socrates argues that imagination must be a matter of imagining something. As it provides an example of how one might begin to argue for the view that words must be about something really real, something that must exist, it can provide us with an introduction to the concerns that motivate Wittgenstein’s discussion of simples in §§39–64: Socrates to Theaetetus: And if you imagine mustn’t you imagine something?—Th.: Necessarily.— Soc.: And if you imagine something, mustn’t it be something real?—Th.: It seems so”35 (PI §518*). Socrates gets Theaetetus to agree with him that if I imagine something, then there is some thing I imagine. But what is that thing? It can’t just be my idea of what I imagine, for I can also imagine things that are real. But then what are we to say about the case where I imagine something that doesn’t exist? The Tractatus, like Socrates at this point in the Theaetetus, aims at a systematic explanation of what must be the case in order for true and false statements to be possible. The Tractarian solution is that we can’t really talk or think about nonexistent objects: when we say something doesn’t exist, or imagine something unreal, closer analysis must show we are talking or thinking about certain simpler objects, which do exist, and denying that they are arranged in a specific way. In order to avoid

Philosophical Investigations §§26–48

197

a regress, one is forced to conclude that there must be some primary elements, which cannot be further analyzed, that refer to things that must exist. The meanings of these directly referring terms just are the objects they refer to. For Russell, these terms were demonstratives, which picked out the contents of immediate experience; for the early Wittgenstein, they were the names that referred to simple objects. These simples cannot come into existence or cease to be, for they are the unchanging ground that makes change possible. In §518, the quotation from Plato is followed by a couple of cryptic and leading questions that amount to a compressed outline of the familiar threestage argument: And mustn’t someone who is painting be painting something—and someone who is painting something be painting something real?—Well, what is the object of the painting: the person in the picture (e.g.), or the person that the picture represents? (PI §518b*) The first sentence, a recasting of Socrates’ questions in Wittgenstein’s words, also does the work of Wittgenstein’s Stage 1 formulation of the problem: we are asked to consider the case of a painting, say a painting of a person, as a good example of what Socrates has in mind. Wittgenstein will often propose that we examine a claim about imagining, or another mental state, by asking us to consider a parallel case that involves a picture, or some other public object. He does not deny that we imagine, or have mental lives, but thinks that we are often tempted to accept incoherent theories about the mind because we attribute mysterious powers to the mind, and that the incoherence only becomes clear when we consider an illustration of the account in question on the public stage. The last sentence of §518 gives us a choice between two different construals of the phrase “the object of the painting.” The first gives us a way of understanding Socrates’ argument on which he is making a simple and uncontroversial point: if we look at a picture of a man, we may say that the object of the picture is the person contained within the picture, the person depicted there; in this sense, the object of the picture certainly exists, even if the picture is the product of the artist’s imagination. This is Stage 2: if we take the object of the picture to be the person in the picture, then any painting of a person must be a painting of someone, and in a similar way we can make sense of the claim that if I imagine, I imagine something, for the talk of imagining something is just another way of talking about my act of imagination. On the other hand, it is easy to see how a Stage 3 reading is possible: if we take the picture to be the person represented by the picture, the living (or dead) person that the picture represents, then there is no guarantee that the picture must be a picture of someone, for it is entirely possible to paint a picture that has no object in this sense. The point is spelled out quite explicitly a few remarks later, in §522–23:

198

David G. Stern If we compare a proposition to a picture, we must think whether we are comparing it to a portrait (a historical representation) or to a genrepicture. And both comparisons make sense. When I look at a genrepicture, it “tells” me something, even though I don’t believe (imagine) for a moment that the people I see in it really exist, or that there have really been people in this situation. For suppose I ask “What does it really tell me?” (PI §522*) “What the picture tells me is itself,” I should like to say. That is, its telling me something consists in its own structure, in its forms and colours. (PI §523*)

Thus in §§518–523, Wittgenstein leads us to a way of looking at pictures, imagination, and other forms of representation on which Socrates’ problem about imagining something real is dissolved. Paraphrasing the Blue Book on knowledge, we could say “There is no one exact usage of the word ‘picture’ or ‘imagine’; but we can make up several such usages, which will more or less agree with the ways the word is actually used.”36 Wittgenstein’s main line of response to Socrates’ concerns about simples in §§46–64 is very similar. For the notion of simplicity is always relative to a context: what counts as simple depends on what we identify as the constituents or components of the complex in question. In §§46–48, Wittgenstein’s initial response to the view that real names refer to primary elements, he proposes that we yield to the temptation and see where it leads. In Stage 1, a quotation of Socrates’ statement of the view that names really stand for primary elements plays the role taken by Augustine’s description of learning to speak in §1: Now what about this matter of names really standing for something simple?— Socrates says in the Theaetetus: “If I make no mistake, I have heard some people say this: there is no explanation of the primary elements— so to speak—out of which we and everything else are composed; for everything that exists in its own right can only be named, no other determination is possible, neither that it is nor that it is not [ … ]. But what exists in its own right has to be [ … ] named without any other determination. Consequently, it is impossible to talk of any primary element by way of an explanation; it, after all, admits of nothing but mere naming; its name is all it has. But just as what consists of these primary elements is itself composite, indeed is an interwoven structure, so its [the structure’s] names become explanatory speech through this kind of interweaving, for the essence of speech is an interweaving of names.”37

Philosophical Investigations §§26–48

199

Both Russell’s “individuals” and my “objects” (Tractatus LogicoPhilosophicus) were such primary elements. (PI §46*) Anscombe’s translation of the opening sentence of §46—“What lies behind the idea that names really signify simples?”—is misleading, if it suggests that Wittgenstein thinks something does lie behind this idea, and he is about to tell us what it is. Wittgenstein, looking over Rhees’ convoluted but fairly literal translation—“What is the position with regard to whether names really stand in for what is simple?”—wrote instead: “Now what about this matter of names really standing for something simple?”38 This way of speaking, like the German, indicates that a topic is being picked up, rather than a position taken. Immediately afterward, §47 tries to show us that the idea in question, that a name ought really to signify a simple, makes no sense at all. For “simple,” if it means anything at all, surely means “not composite,” not made up of smaller parts. The philosopher of the simple is looking for something absolutely, contextlessly, simple, but “composite” is itself a context-dependent term. §47 both gives a number of striking examples of different cases in which “composite” can be understood in different ways, and sums up Wittgenstein’s response to the philosophical question about the nature of the simple: If I tell someone without any further explanation: “What I see before me now is composite,” he will have the right to ask: “What do you mean by ‘composite’? For there are all sorts of things that that may mean!”—The question “Is what you see composite?” makes good sense if it is already established what kind of complexity—that is, which particular use of the word—is in question [ … ]. We use the word “composite” (and therefore the word “simple”) in an enormous number of different and differently related ways [ … ]. To the philosophical question: “Is the visual image of this tree composite, and what are its component parts?” the correct answer is: “That depends on what you understand by ‘composite’.” (And that is of course not an answer but a rejection of the question.) (PI §47c, §47e, §47f) As in §1d, the nub of Wittgenstein’s way of rejecting the view under discussion is set out before the method of §2 is invoked. In §48, the barrage of objections we found in §47 is distilled into a language-game that plays the role of Stage 2, a language-game that is clearly proposed by the voice of correctness, the voice that aims to make what sense it can of the idea of real simples: Let us apply the method of §2 to the account in the Theaetetus. Let us consider a language-game for which this account is really valid. The

200

David G. Stern language serves to describe combinations of coloured squares on a surface. The squares form a complex like a chessboard. There are red, green, white and black squares. The words of the language are (correspondingly) “R”, “G”, “W”, “B”, and a sentence is a series of these words. They describe an arrangement of squares in the order:

And so for instance the sentence “RRBGGGRWW” describes an arrangement of this sort:

Here the sentence is a complex of names, to which corresponds a complex of elements. The primary elements are the colored squares.39 At this point, an interlocutory voice asks “But are these simple?” to which the voice of correctness replies, “I do not know what else you would have me call ‘the simples,’ what would be more natural in this language-game.” But this immediately leads to Stage 3: however natural the proposed interpretation seems, a change of context can make another interpretation more natural. But under other circumstances I should call a monochrome square “composite,” consisting perhaps of two rectangles, or of the elements colour and shape. But the concept of compositeness might be extended in such a way that a smaller area is said to be “composed” of a greater area and another one subtracted from it. (PI §48a*) Of course, the idea proposed here is counter-intuitive at first sight, but as Wittgenstein points out, there are precedents for it: we do think of a plane’s speed flying into a headwind as the result of subtracting the headwind from the plane’s cruising speed, for instance. As usual, the point of the Stage 3 argument here is not that another construal is more natural, but that another construal is possible, given different circumstances, circumstances that may be quite far-fetched. Wittgenstein’s narrator goes on to offer a diagnostic explanation for the Socratic (and Tractarian) desire to say that simples must exist, that they can

Philosophical Investigations §§26–48

201

only be named, and that their existence is a condition of significant discourse. He suggests that this conviction arises out of a misunderstanding of the role of paradigmatic examples (such as color samples) or rules (such as a table correlating colors and their names) in our use of language. In §50, he offers the example of the standard meter in Paris—the specially constructed measuring rod that was once used as the standard for the construction of all other measures of length in the metric system. This rod, he claims, is the one thing of which one can neither say that it is, nor it isn’t, one meter long. Not because its length is ineffable, but because these words (namely, “the standard meter is one meter long”) make no sense. To say that the standard meter is one meter long is comparable to saying “But I know how tall I am!” and putting my hand on top of my head to prove it40 or maintaining that “this is here,” said pointing at something right in front of me, always makes sense.41 The words “the standard meter is one meter long” have the sound of a logical truth, but the truth is nonsense, precisely because we are no longer making use of the standard meter as a paradigm in order to measure something else, but rather turning it back on itself. If we take into account the circumstances in which we use these words, we see that what we are trying to do in such uses of words is to step outside the usual circumstances in which we locate and measure things, with the aim of stating something more profound and fundamental. Of course, we can imagine unusual circumstances in which “the standard meter is one meter long” could make sense—perhaps in talking to someone who has no idea what a standard meter is, or in which “this is here” might do some work—perhaps pointing out a familiar pattern that had so far gone unnoticed—but the point of such philosophical truths, as the philosopher wants to use them, is that they are supposed to be stating something that must be true, not drawing our attention to something unexpected. Here we are inclined to misunderstand the role assigned to the object in the language-games as though it were a mysterious feature of the object itself, and the doctrine of simples is an ontological crystallization of this methodological role.

8.6 Epilogue Most interpreters take the opening remarks of the Philosophical Investigations as an outline of Wittgenstein’s answers to questions about the nature of language. Consequently, they take the material we have just discussed to have immediate and clear-cut consequences for positions in contemporary analytic philosophy, such as the theory of reference. However, on the dialectical reading advocated here, Wittgenstein’s opening is best understood as raising questions and introducing us to a number of voices in the discussion that follows, rather than as offering solutions to familiar problems. Wittgenstein’s primary concern is not with the sophisticated statement of philosophical problems and theoretical solutions that they lead to, but how those problems arise. He begins not with systematic philosophy, or the history of philosophy,

202

David G. Stern

but with the patterns of thought, the ways of speaking, which can lead us into formulating such philosophical theories. Because the opening sentences of the Philosophical Investigations are often construed as a critique of certain fundamental premises in the theory of reference, they provide an excellent point of departure for further consideration of this question about the implications of a dialectical, or polyphonic, reading for current work in the theory of reference. Wittgenstein’s first sentence in the Philosophical Investigations tells us that the opening quotation from Saint Augustine contains a definite picture of the essence of language. Yet on first reading, Augustine’s words can strike the reader as an entirely natural and unproblematic description of how he learned to speak: My primary reaction to the citation from the Confessions, read by itself, is to think that what it expresses is obvious—it seems trivial, prosaic, well-nigh unobjectionable. It is just a harmless elaboration of the observations that early in life children learn what things are called, and learn to express their wants and needs verbally.42 Warren Goldfarb contends that Wittgenstein expected his readers to be shocked by his initial construal of Augustine, and those commentators who “would have us meekly acquiesce to this sentence” have been too ready to take what he has to say at face value.43 Instead, Goldfarb proposes, the aim of these opening words is not to show us that Augustine’s conception of language is wrong and Wittgenstein’s right, but to throw us off balance, by challenging our preconceptions about what is involved in having a conception of language. Ultimately, the aim of this challenge is to get us to see that the same words can be understood in both a commonplace and a philosophical way, and thus the unclarity of the very idea of our having a conception of language. Talk of naming objects, pointing things out, and states of mind can be quite unproblematic, yet in certain contexts these notions “come to have a weight that our ordinary understanding of them does not support.”44 For Augustine’s words can also be read as intimating a number of different conceptions of how language works, conceptions that can provide a starting point for philosophical theorizing about language and meaning. Similarly, Cavell proposes that we approach the multiplicity of voices, the lack of clear demarcation between voices, the frequent shifts in topic, the fragmentary arguments, the multitude of questions, suggestions, instructions, stories, and far-fetched imaginary examples as an invitation to explore the deep difficulties that arise when we start to philosophize, rather than construing them as all advancing a single, authorial point of view. On the other hand, nothing can stop a philosopher from taking Wittgenstein’s arguments and drawing his or her own positive conclusions; indeed, the book both invites and resists such a reading. What, precisely, is the “particular picture of the essence of human language” that Wittgenstein speaks of in the first sentence of his own words in

Philosophical Investigations §§26–48

203

the book, or “the conception expressed in that quotation,” as Malcolm puts it?45 To many philosophers, §1b provides a clear outline of the mistaken view of the nature of language that Wittgenstein finds in Augustine, a view that is presupposed in §1a, and that forms one of the principal targets in what follows. One line of interpretation concentrates on the question of the nature of this particular picture, and its place in the overall project of the Philosophical Investigations. While I cannot do justice to the full range of such readings of §1b, the main disagreements concern the scope and character of the picture in question. On a big picture reading of Augustine’s words, they serve to introduce the paradigm, or world-view, that Wittgenstein opposes throughout the book.46 On a small picture reading, they set out a quite specific philosophical theory, the topic of the opening sections of the book. While there is some support for each of these approaches in §1b, Wittgenstein’s words are more equivocal than they may seem at first sight. He does say that the particular picture suggested to him by Augustine’s words is that: (1) “individual words in language name objects”; and (2) “sentences are combinations of such names.” However, Wittgenstein qualifies all this by an “it seems to me,” one of the expressions he repeatedly uses as a warning that he is introducing, or entertaining an idea, rather than giving it his full support. These words are followed by a long double-dash, used by Wittgenstein to indicate either a change of topic, or a new voice (unlike a single dash, which often indicates a brief pause within a single train of thought).47 Only then are we introduced to a further idea, which, we are told, has its roots in this picture: that (3) “every word has a meaning [Bedeutung],” and it “is the object for which the word stands.” Thus we are introduced to at least two views of the Augustine passage: the particular picture, which concerns the role of words in language, and a further development of this conception, which introduces the notion of a word’s meaning. Exponents of a small picture reading of these words are often attracted to the idea that the Tractatus is the real target. For instance, according to Jaakko and Merrill Hintikka, the passage from Augustine is Wittgenstein’s starting point because it sets out the view that ostension—naming objects and pointing to them—is the “prime vehicle of language teaching and learning.”48 This view supposedly follows from an “idea embraced by Wittgenstein in the Tractatus [ … ] the basic links between language and the world are simple two-place relations of naming.”49 On the Hintikkas’ construal, the links in question are ineffable; nothing more can be said about them within the Tractarian framework. A related reading maintains that the picture is a little bigger, for the Tractarian conception of naming can be specified—and criticized—in more detail. Thus, according to Hacker, it is part of the Tractatus view of naming that it requires a mental act “that injects meaning or significance into signs, whether in thought or in language.”50 But such crude and simple views of language and mind fail to do justice to the subtlety of the Tractatus, a book that takes as its fundamental thought the idea that certain words do not stand for anything.51 Neither of

204

David G. Stern

these remarkably naïve views of naming nor the role of mind in meaning are of much independent philosophical interest, either. It is implausible, to say the least, that these are the principal topics raised by the Philosophical Investigations’ opening.52 Consequently, most of those who begin by spelling out the rather limited views they find in §1b go on to say that their real interest lies in the bigger picture that they give rise to, or otherwise support. For instance, Baker and Hacker hold that this passage introduces the Augustinian picture, which proves to be nothing less than an entire world-view or philosophical paradigm, the principal target throughout the Philosophical Investigations, and this interpretation has seemed obviously right to many. However, much of the attractiveness of this view depends on the way its expositors trade on the open-ended and vague character of the picture in question, moving between a relatively small and specific view in the initial exposition, and the much larger theoretical positions it supposedly underwrites. Glock’s Wittgenstein Dictionary entry on the “Augustinian picture of language” is a good example of this: [Wittgenstein] treated Augustine’s view not as a full-blown theory of language, but as a proto-theoretic paradigm or “picture” which deserves critical attention because it tacitly underlies sophisticated philosophical theories. [ … ] The Augustinian picture comprises four positions: a referential conception of word meaning; a descriptivist conception of sentences; the idea that ostensive definition provides the foundations of language; and the idea that a language of thought underlies our public languages.53 Indeed, less sympathetic readers of Baker and Hacker have noted that “most of the leading ideas of the [Philosophical Investigations are] interpreted as so many aspects of an extended critique of the Augustinian Picture.”54 However, the big picture reading has a slender textual basis: the very expression “Augustinian” picture is an artifact that occurs nowhere in Wittgenstein’s writing, and the particular picture of §1b is only one of many quite specific philosophical mistakes and temptations that Wittgenstein discusses in the pages that follow. Indeed, Wittgenstein’s explicit references in the Philosophical Investigations to the conception of language that he finds in Augustine are outnumbered by his references to other passages from the Confessions, each of which serves to introduce other philosophical pictures. Running these pictures, and others, into a single over-arching big picture runs contrary to Wittgenstein’s emphasis on the diversity and multiplicity of the ways in which we go wrong. Furthermore, the significance of the passage from Augustine for the Philosophical Investigations is not simply that the opening quotation is a convenient point of departure for a critique of mistaken views about meaning and language learning. For the very words from Augustine that Wittgenstein quotes do not lend unambiguous support to the particular

Philosophical Investigations §§26–48

205

interpretations he advances in §1 ff., either. In fact, Wittgenstein’s reading of the Augustine passage is itself remarkably one-sided. While the quotation from Augustine does begin with his learning that certain sounds his elders made were signs for things, and ends with his learning to use these signs to express his own desires, much of it does not fit Wittgenstein’s description at all. Augustine does not claim that this gives a model for understanding all word meaning, and actually connects understanding words with grasping their proper places, not only in the structure of sentences, but also the context in which they are used. He also draws our attention to the role of facial expression, the play of the eyes and tone of voice in expressing intention. Given the amount of thought Wittgenstein clearly gave to the question of how best to start, and given his intimate acquaintance with Augustine, is it not much more likely that he expected the reader to be initially impressed by the reading offered in §1b, but then come to see its limitations? These are multiple: §1b is not only a very partial description of what Augustine says, and the picture that it does sketch is only a first example of how we can go wrong in philosophy, not the outline of an overarching big picture. After all, in §1c and §2a, Wittgenstein suggests that it can help us to understand the over-simple conception of language attributed to Augustine in §1b if we see it, not as a mis-description of our language, but rather as a description of a language simpler than our own. Surely Wittgenstein expected the careful reader to eventually see that §1b is an over-simple, one-sided and provocative reading of §1a. Yet many of Wittgenstein’s readers have either, like Hintikka or Hacker, taken §1b to be an unproblematic exposition of what Augustine says or presupposes or, like Kenny, taken it as proof that Wittgenstein missed Augustine’s point. Wittgenstein’s opening words, like Augustine’s, are not as simple as they seem; they offer orientation in several different directions. Clearly, they do introduce us to a family of views about how language is learned, and what words mean, that will be Wittgenstein’s principal target in the first 64 sections of the book. Along these lines, Anscombe observes that the main purpose of the opening of the Investigations is negative, namely to persuade us not to look at the connection between a word and its meaning either as set up or as explained by (a) ostensive definition, or (b) by association, or (c) by mental pictures, or (d) by experiences characteristic of meaning one thing rather than another, or (e) by a general relation of reference or naming or designation or signifying which has (logically) different kinds of objects as its terms in different cases.55 However, while there is a good deal of argument along these lines in the sections that follow, much of it, like §1b, is directed at positions that are remarkably naïve. Baker and Hacker’s explanation of this is that the real target is the big picture that lies behind the particular topics that are discussed. There is something right about this: Wittgenstein is trying to get at

206

David G. Stern

the preconceptions and unexamined assumptions that lead philosophers to argue as they do. However, Wittgenstein does not see those preconceptions as something else, over and above these particular lines of argument, along the lines of the Augustinian picture, the true motives hidden behind what philosophers say and do, but rather as present in the moves that begin philosophical reflection. For this reason he begins with particular examples of deceptively simple philosophical arguments, of the kind that are more likely to come up in a classroom discussion than a philosophical article or book, yet his main aim is not so much to show that they are wrong, as to get us to think about what is involved in beginning to look for a philosophical account of language and meaning. As Robert Fogelin puts it, “for the most part the work is not a criticism of the results of philosophizing, but an interrogation of its source.”56 This question about the character of Wittgenstein’s critique of philosophy, in turn, is connected with the question of what Wittgenstein means by saying that past philosophy is nonsense. On Hacker’s reading, Wittgenstein appeals to the grammar of our language to show that what philosophers say makes no sense. While it is true that there is a strand in the book’s dialogue— Cavell’s “voice of correctness”—that does make use of grammar in just this way, I take it that the point of the book is not to get us to cheer on this side of the debate. Rather, to say that philosophy is nonsense is just to say that it falls apart when we try to make sense of it: The results of philosophy are the discovery of one or another piece of plain nonsense and of bumps that the understanding has got by running up against the limits of language. These bumps make us see the value of this discovery. (PI §119) But this must emerge out of the reader’s involvement in the dialogue of the Philosophical Investigations, our being tempted into particular philosophical theories, and our coming to see that those particular attempts at theorizing are nonsensical, rather than as a more general principle, or overarching method, that any one voice in the dialogue is advocating.

Notes 1 This chapter is based on ch. 1, §§ 1 and 3, and ch. 4, §§ 1, 3, and 4 of David G. Stern, Wittgenstein’s Philosophical Investigations: An Introduction (Cambridge: Cambridge University Press, 2004). The permission of Cambridge University Press to make use of this material is hereby gratefully acknowledged. References in footnotes to “ch. m” or § “m.n” are to other parts of that book. 2 For references to the Philosophical Investigations, I provide section numbers, or page numbers, in parentheses. All translations are based on G. E. M. Anscombe’s. Where I depart from her translation, I add an asterisk to the cited section number. 3 PI §48.

Philosophical Investigations §§26–48

207

4 M. O’C. Drury, “Conversations with Wittgenstein,” in Recollections of Wittgenstein, ed. Rush Rhees (Oxford: Oxford University Press, 1984), p. 149. 5 Theaetetus 146c–e; Plato, Complete Works (Indianapolis, IN: Hackett, 1997), pp. 162–63. 6 PI §116. 7 See Theaetetus 146 ff. 8 PI §39, cf. §89. 9 PI §§26–31; see section 8.3 of this chapter. 10 PI §§39–64; see section 8.4 of this chapter. 11 PI §143, §185. 12 PI §§85–86. 13 PI §§139–41. 14 PI §30. 15 PI §87. 16 PI §464, §524. 17 PI §500. 18 PI §464; cf. §524c. For further discussion of Wittgenstein on philosophy and nonsense, see 2.2–2.3 in Stern, Wittgenstein’s Philosophical Investigations. 19 PI §81a. 20 PI §6b. 21 PI §§21–25. 22 I have introduced line breaks to emphasize the changes of voice. 23 PI §130. 24 PI §38b & 38d*. Here I follow Rhees’ prewar translation, in TS 226 of the Wittgenstein papers, rather than Anscombe’s somewhat more literal “goes on holiday.” The advantage of this translation is that it makes it a little clearer that Wittgenstein’s point is that the language in question is not doing any work at all. 25 Usually, it is taken for granted that “sublime” means “elevated” or “pure.” Cavell is the leading advocate of a Romantic reading, Stanley Cavell, This New Yet Unapproachable America: Lectures after Emerson after Wittgenstein, The Frederick Ives Carpenter Lectures 1987 (Albuquerque, NM: Living Batch, 1989), pp. 56–58. Two interpreters that do give due consideration to the question of how to understand Wittgenstein’s use of sublimieren are Gordon Bearn, Waking to Wonder: Wittgenstein’s Existential Investigations (Albany, NY: SUNY Press, 1997), pp. 309–12; Eike von Savigny, “No Chapter ‘on Philosophy’ in the Philosophical Investigations,” Metaphilosophy 22 (1991): 148–49. There are also parallels with the psychoanalytic use of the term: Freudian sublimation is a matter of giving displaced artistic expression to repressed drives that would otherwise be inexpressible; Wittgensteinian sublimation is a matter of providing a metaphysical response to philosophical problems that cannot be given a more direct solution. In both cases, the result, however beautiful it may be, does not solve the problem that gave rise to it. 26 This appears to be the reading that von Savigny favours. 27 PI §97, §§107–08. 28 Bearn advocates this construal: Bearn, Waking to Wonder. 29 PI §5a. 30 PI II, xi, p. 222/189. 31 PI §39. 32 PI §§40–45. 33 §§48–64. 34 In a footnote and the surrounding text, Pears draws a connection with the concerns of §218, but the point is considerably more general. David Pears, The False Prison: A Study of the Development of Wittgenstein’s Philosophy, vol. II (Oxford: Clarendon Press, 1987), pp. 209–10.

208

David G. Stern

35 Theaetetus 189a. For further discussion of this passage see David G. Stern, Wittgenstein on Mind and Language (New York: Oxford University Press, 1995), pp. 54–55. 36 See BB, p. 27. 37 Theaetetus 202a–b. Like Anscombe’s translation, this is not a direct translation of the Greek, but of the German translation (by Preisendanz) that Wittgenstein used. 38 TS 226, p. 31. 39 Readers of the “third” 2001 edition of the Philosophical Investigations should be warned that the use of seven black squares and two white ones in the second diagram in §48 is an error. In fact, none of the revised graphics incorporated in the reissued second edition (1997) are taken up in the 2001 edition. Presumably, the text of the 2001 edition is based on a photocopy of a pre-1997 printing of the second edition; this would explain why black ones replaced the colored blocks, used in most printings of the text prior to 1997. I discuss the relationship between the varying reproduction of the blocks in these editions and the source text in a review of the Bergen edition of the Wittgenstein papers, David G. Stern, “Critical Review of Wittgenstein 2000,” European Journal of Philosophy (forthcoming). 40 PI §279. 41 PI §117b. 42 Warren Goldfarb, “I Want You to Bring Me a Slab. Remarks on the Opening Sections of the ‘Philosophical Investigations’,” Synthese 56 (1983): 268. 43 Ibid.: 268. 44 Ibid.: 268. 45 Norman Malcolm, Ludwig Wittgenstein: A Memoir, 2nd ed. (Oxford: Oxford University Press, 1984), pp. 59–60. 46 Gordon Baker and P. M. S. Hacker, Wittgenstein: Meaning and Understanding, vol. 1, Essays on the Philosphical Investigations (Oxford: Blackwell, 1985), 1–27. 47 PI §1b. 48 Merrill Hintikka and Jaakko Hintikka, Investigating Wittgenstein (Oxford: Blackwell, 1986), p. 179. 49 Ibid. 50 Ibid., p. 75. 51 TLP 4.0312. 52 Cf. Robert Fogelin, Wittgenstein, 2nd ed. (London: Routledge & Kegan Paul, 1987), pp. 108–09; Brendan Wilson, Wittgenstein’s Philosophical Investigations: A Guide (Edinburgh: Edinburgh University Press, 1998), p. 1; Goldfarb, “I Want You to Bring Me a Slab.” 53 The last idea is not present in §1b, but something like it is attributed to Augustine in §32, Hans-Johann Glock, A Wittgenstein Dictionary (Oxford: Blackwell, 1996), p. 41. 54 Peter Carruthers, “Critical Study: Baker and Hacker’s Wittgenstein,” Synthese 58 (1984): 451. 55 G. E. M. Anscombe, “A Theory of Language,” in Perspectives on the Philosophy of Wittgenstein, ed. Irving Block (Oxford: Blackwell, 1981), p. 154. 56 Fogelin, Wittgenstein, p. 110.

Part III

The absolute

9

The architecture of meaning Wittgenstein’s Tractatus and formal semantics Martin Stokhof 1

9.1 Introduction With a few notable exceptions formal semantics, as it originated from the seminal work of Richard Montague, Donald Davidson, Max Cresswell, David Lewis, and others, in the late 1960s and early 1970s, does not consider Wittgenstein as one of its ancestors. That honor is bestowed on Frege, Tarski, Carnap. And so it has been in later developments. Most introductions to the subject will refer to Frege and Tarski (Carnap less frequently)— in addition to the pioneers just mentioned, of course—and discuss the main elements of their work that helped shape formal semantics in some detail. But Wittgenstein is usually conspicuously absent whenever the history of the subject is mentioned. Of course, if one thinks of Wittgenstein’s later work, this is obvious: nothing, it seems, could be more antithetic to what formal semantics aims for and to how it pursues those aims than the views on meaning and language that Wittgenstein expounds in, e.g., Philosophical Investigations, with its insistence on particularity and diversity, and its rejection of explanation and formal modeling. But what about his earlier work, the Tractatus? At first sight, that seems much more congenial, as it develops a conception of language and meaning that is both general and uniform, explanatory and formal. In view of that, the general lack of reference to Tractatus is curious. The central claim of this chapter is that, actually, this is an oversight. Perhaps Wittgenstein was no conscious influence on the minds of Montague et al. at the time, but he did play a major role in establishing the fundamental principles and philosophical assumptions that helped shape formal semantics and make it such a successful enterprise, in linguistics and in philosophy. The actual channel through which this transmission of concepts and ideas has taken place is not what we will be focusing on here. That is another story, and a complicated one, which requires more historical knowledge and skills than we can muster. Rather, we will be content with discussing some systematic analogies (and differences, for there are those as well, of course) between Wittgenstein’s conception of language and meaning in the Tractatus, and the one that was prevalent in formal semantics at its

212

Martin Stokhof

inception and that continues to exert a major influence in the field until the present day. The reasons for being interested in these connections are twofold. First of all, it appears that formal semanticists do not always appreciate what philosophical assumptions are behind their enterprise. This is deplorable, since an awareness of that part of its legacy could help formal semantics answer questions regarding its proper status as a scientific discipline. The diversity that is characteristic of the state in which formal semantics finds itself today, raises the question how this came about, and how it can be justified. And part of the answer may well have to do with diverging ways of dealing with the problems caused by these philosophical assumptions. The second reason why tracing some systematic connections between Wittgenstein’s early work and the origins of formal semantics is of interest stays closer to home, i.e., closer to Wittgenstein. At many points in his later works Wittgenstein formulated penetrating criticisms of his earlier ideas. Inasmuch as formal semantics incorporates some of the latter the former might apply to it as well. These are related but distinct considerations, that each in their own way put formal semantics to the test: Is it really an empirical discipline? Or does it remain rooted in its philosophical ancestry? The approach taken in what follows is by and large systematic. In section 9.2 we will review the main characteristics of the Tractatus’s architecture of meaning (what meaning is, how it is structured, how it relates to language and to the world), isolating three aspects that are particularly relevant for a comparison with formal semantics. Section 9.3 contains a brief sketch of the way in which some ideas from the Tractatus were transmitted through the work of Rudolf Carnap. Then, in section 9.4 we turn to a detailed analysis of the conceptions that are prominent in the work of the pioneers of formal semantics. The focus will be on the work of Richard Montague and Donald Davidson, but where relevant we also refer to the writings of other authors such as David Lewis and Max Cresswell. After this exposition we turn to an exploration of resemblance and difference between the Tractatus and formal semantics in section 9.5. Finally, in section 9.6 we will address the two issues identified above: the consequences for the nature of formal semantics as a scientific discipline, and the potential relevance of Wittgenstein’s own criticisms on the Tractatus framework for formal semantics.

9.2 Wittgenstein’s architecture of meaning Although the nature of the main work of Wittgenstein’s youth remains controversial, it does seem, at least at first sight, that the Tractatus contains a substantial theory of meaning. It outlines the nature of meaning, its structure, it analyzes the way in which meaning relates to linguistic expressions, it details the role of the world in all this, and so on. The aim of this section is to sketch the place this theory of meaning has in the overall conception of

The architecture of meaning

213

the Tractatus, and to derive from that some of its general and central characteristics. It is not our intention here to enter into the debate surrounding the resolute interpretation of the Tractatus that has been proposed in recent years by the so-called New Wittgensteinians.2 Nevertheless it should be pointed out that the idea of extracting a general theory of meaning from the Tractatus does take a stand in such a debate, since it presupposes an interpretation of the book as a whole that allows for this, and such an interpretation will at certain points be at odds with the resolute one. As a matter of fact, there are a number of ways of reading the Tractatus other than the resolute one, that make this possible. Fortunately for our present purposes, it will not be necessary to discuss the various options and decide between them on the basis of exegetical or systematic argumentation. Rather, as long as the resulting picture of Wittgenstein’s ideas about meaning remains indifferent to these variations, we can pick one that serves our needs. And this is what we will do in this section. In a few words our reading is as follows.3 A central concern of Wittgenstein’s during his entire life was ethics. The fundamental question, the basic moral challenge, of how to live, was one which occupied Wittgenstein always. The time of his work on the Tractatus was no exception. Reports by contemporaries, from philosophy and from other walks of life, the surviving notebooks that he kept in that period, as well as some of the things he said about the Tractatus after it was completed, leave no doubt that at that time, as always, ethics was of fundamental importance to Wittgenstein. And the concern for ethics was all-pervading, not just one of a number of interests that Wittgenstein explored, but a need and an attitude that were present in everything he did. To argue this in detail would take us too far afield, but a telling testimony is the famous letter Wittgenstein wrote to Ludwig von Ficker, in which he states the following: the point of the book is an ethical one. I once wanted to include in the preface a sentence that is now actually not there, but that I will write to you now since it might be key for you: I wanted to write that my work consists of two parts: of the one that is present here and of everything that I have not written. Precisely this second part is the important one. For the ethical is delimited as it were from the inside by my book; and I am convinced that strictly speaking it can only be delimited in this way. In short I think: everything of which many nowadays are blethering, I have defined in my book by being silent about it [ … ].4 This suggests that the Tractarian system serves a purpose that goes beyond itself. What Wittgenstein wants to do is to safeguard ethics from all kinds of theorizing, from logical analysis, metaphysical speculation, and theological dogma. Ethics is about action, about what to do (or what to refrain from doing), and it should be dealt with in that sphere. Ethics does not belong in the cognitive realm of thought and reasoning, proof and disproof, ethical

214

Martin Stokhof

views can not be discussed, argued, they can not even be expressed in meaningful language. To show this, rather than merely state it, Wittgenstein needs a theory of meaning and thought from which it follows that ethics falls outside their domain. That is why in the preface Wittgenstein describes the aim of the Tractatus as “to draw a limit to thought, or rather—not to thought, but to the expression of thoughts.” The totality of expressions of thoughts is the totality of everything that can meaningfully be expressed, i.e., the realm of the meaningful. The idea is that once the limits of that are drawn, we can characterize the realm of ethics ex negativo as being that with which the meaningful can not be concerned.5 So the theory of meaning (and language and the world) that the Tractatus offers is there for a specific purpose. This observation is not just relevant for an understanding of the Tractarian project as a whole, it also informs some of the more conspicuous features of the theory itself. Three of these will be discussed in some detail in what follows, viz., universalism, intensional referentialism, and compositionality, all central characteristics of Wittgenstein’s architecture of meaning. “Universalism” here refers to the task that the Tractatus sets itself, viz., to give an account of how language operates in general, i.e., not the way any specific language works, but what underlies the possibility of any language to express meaning. “Intensional referentialism” refers to the combination of an intensional ontology (consisting of possible situations built from a substance of objects) and a truth-functional, extensional semantics. Finally, compositionality (in various domains) plays a key role in the entire the Tractatus-framework. But let us first look briefly at the global structure of the theory. The central role is played by the picture theory of meaning, according to which the meaning of a sentence is the situation it pictures. This picturing relation has two basic ingredients. The first is the existence of a depicting relation between simple expressions, names, and simple entities, objects.6 The second is identity of the logical forms of the picturing sentence and the pictured situation. The idea of logical form as the most fundamental structure of both the world and language is central to Wittgenstein’s thinking here: every contingent feature of what makes an actual language meaningful, every contingent characteristic of some situation that is captured in language has to be abstracted away from, if we are to uncover the most fundamental principles that make meaning possible in the first place. At its most general level it is the logical form that world and language share that allows the two to enter into the picturing relation. Likewise, the depicting relation that puts names and objects in a one-to-one relationship is devoid of any descriptive content, simply because such content would be contingent and hence would make the meaningfulness itself a contingent matter as well. The combined result of such a purely referential relation and identity of abstract, logical form is a concept of meaning that is as general as can be. Contingency, then, plays a crucial, Janus-faced role in Wittgenstein’s conception of meaning. On the one hand, as we just saw, he strives for an account

The architecture of meaning

215

that is as general as possible, and hence abstracts away from any contingent features of the various elements that enter into the meaning relationship. On the other hand, contingency is the most fundamental characteristic of the meaningful. For, if the realm of the meaningful coincides with the totality of the contingent, then, a fortiori, ethics is excluded, since whatever ethical value is, it is necessary.7 Of course, with ethics a number of other domains of the necessary, such as logic, mathematics, and certain aspects of science, fall outside the realm of the meaningful as well. These Wittgenstein deals with successively in the last part of the Tractatus, explaining in each case why what looks necessary and meaningful in fact is not. Logic, being the transcendental scaffolding of world and language that makes the meaningful possible, is ineffable in that it can not be meaningfully described, but is shown by meaningful description of contingent situations. Mathematics does the same. Ethics is also dubbed transcendental8 and is also ineffable, but it is shown, not through language, but through action. It is in the action itself (and not in the consequences that follow it or in the intentions that precede it, for such connections are temporal and causal, and hence logically contingent) that ethical value resides. And what distinguishes the ethically just from the ethically unjust is not reflected in some contingent change in what happens in the world, but is revealed in the way in which the world as a whole, and not just some specific part of it, appears to the subject. Such a necessary and a priori distinction between the meaningful and the meaningless requires a theory of meaning that is itself necessary and a priori. This means that it has to hold not just for some particular language (say, English), or some group of languages (say, Altaic languages). Nor is it sufficient to characterize the meaningful for all natural languages. No, the very purpose that the theory serves, the goal it aims to achieve, requires that it be completely general, that it hold for any conceivable form of language, any possible form of symbolic expression. In other words, the Tractatus’s conception of meaning is universalistic through and through. When the Tractatus mentions the language it means all logically possible forms of symbolic expression; when it refers to the world it indicates any logically possible configuration of situations; and when it mentions the meaning it abstracts away from any contingent features that make particular expressions mean particular things. The Tractatus’s theory of meaning is thus transcendental in the broad sense of dealing with conditions of possibility. Constructing meaning in terms of the picturing relation, it comprises an ontology, a theory of linguistic (symbolic) structure, and an account of the relation between the two. These three elements consist of a reconstruction of the fundamental properties of the world, of language and of their interconnection, that are needed for any concrete language to mean any concrete situation. Those fundamental properties Wittgenstein finds in logic, which (in the Tractatus at least) provides an unquestioned and unique foundation. Thus the transcendental task that Wittgenstein sets himself in the preface of the Tractatus,

216

Martin Stokhof

viz., that of determining the limits of what can be meaningfully expressed, turns out to be to uncover the fundamental logical form that characterizes both language and the world and that makes meaningful expression possible. Recall that this critical task itself derives from another one, that of securing the ineffability of ethics. Thus, it appears that the ontological and linguistic theories that the Tractatus offers are derived from this fundamental starting point, and hence, give us an account, not of the world or of language directly, but of what they have to be like for meaningful expression to be possible (in such a way that ethics remains outside its reach). One consequence of the universalism of the Tractatus is that many of the things it states about meaning, language, and the world, are extremely abstract and often quite difficult to connect with what we understand by these notions in everyday contexts. For example, the concept of an object, being simple in the Tractatus’s sense, i.e., devoid of any contingent features, defies attempts to provide concrete instantiations precisely because of the abstract logical nature of simplicity that is characteristic for it. Yet, this feature is needed in order for the theory to be universal.9 The same holds for the linguistic category of names, which, appearances notwithstanding, have no counterpart in a natural language, but are a purely logical kind of entity.10 Another consequence of the universalism of the Tractatus is that Wittgenstein can not avail himself of something akin to the distinction between object language and metalanguage that modern readers are thoroughly familiar with from Tarski’s work. The point is not that Wittgenstein did not have the distinction, or something analogous such as the distinction between use and mention, available as an analytic tool. Rather such a tool is useless because it constantly shifts the fundamental question to the next level. Using a suitable metalanguage we may well be able to give a complete specification of the semantics of some object language, but that works only if such a semantics for the metalanguage is already in place.11 That in its turn can be specified in terms of yet another metalanguage, but the question what meaning tout court is, remains unanswered. For example, if we are to give a description of the fundamental way in which names and objects are related, then the language in which we formulate this description presupposes what it describes. For Wittgenstein this is not a viable way to go, his universalistic goal forces him to determine the nature of the meaningful, both its scope and its structure, in one fell swoop and in a definitive way: it may not rest on assumptions about meaning or meaningfulness.12 This means he has to accept that any description of the fundamental principles of meaning is itself meaningless.13 In order to pull this off Wittgenstein make heavy use of two conceptual tools: the saying/showing distinction, and the principle of compositionality. The latter will be dealt with later, the former stands out as one of the most characteristic features of the Tractarian system. At the root of the distinction between saying and showing is an absolute distinction between the necessary and the contingent. That distinction is not parametrized

The architecture of meaning

217

or contextualized, it holds without any qualification. The realm of the contingent being identified with that of the meaningful, anything necessary ipso facto can not be expressed in meaningful language: “What can be shown, cannot be said.”14 This applies also to the fundamental structure of the realm of meaningful itself: given that this structure does not change, the principles determining it are necessary, hence can not be captured in a theory. “Logic must take care of itself” is the first entry in the 1914–16 Notebooks,15 and the claim is repeated in the Tractatus16 There can be no theory of logic, because logic is the most fundamental structure of all, and is presupposed by any theory.17 So nothing can take care of logic. Then how does logic take care of itself ? Here the notion of showing enters the stage. For Wittgenstein the fundamental, logical principles that characterize the meaningful, can not be described but, being the very principles in virtue of which language is meaningful, i.e., the principles that are operative in each and every meaningful expression, they are displayed, shown in the Tractatus sense of the term, by these expressions and the use we make of them.18 Thus, the logic of language, which is also the logic of the world in virtue of the identity of logical form that is a necessary requirement for picturing, is accessible in this sense that analysis of meaningful language in the end will show this logic, laid out in completely analyzed expressions.19 This reads the larger part of the Tractatus as some kind of transcendental deduction that starts from the given of meaningful language and then derives the fundamental conditions of its possibility, basically by abstracting away from anything contingent that enters into actual meaningful language.20 As alluded to at the beginning of this section, this is a specific way of reading the Tractatus, one that is not universally shared. But it is a reading which not only provides a substantial account of meaning, one which Wittgenstein arguably contested at many points later on in his life, it also fits in with Wittgenstein’s professed aim in writing the Tractatus, viz., that of making a specific point about ethics. It would seem that the connection with ethics is not strictly necessary to defend this reading of its sections on language and meaning, logic, and ontology. There are also reasons internal to what Wittgenstein says about these subjects that support it, such as the fundamental idea that “logical constants do not represent.”21 The fundamentals of logic, being at the same time the fundamentals of meaning, are not entities in the world, but conditions thereof. They show the structure of any world,22 hence “logic is transcendental.”23 However, as argued elsewhere,24 the connection with ethics is essential for a coherent interpretation of the Tractatus as a whole. Let us now turn to the second feature identified above, that of intensional referentialism. If one approaches the Tractatus conception of meaning primarily from the point of view of language and investigates what kind of semantic theory it proposes, then what one finds looks like basically a referential theory of meaning. All sentences of a language are composed of elementary ones, which, in their turn, are composed solely of names. This

218

Martin Stokhof

depends on names (and objects) displaying type variety, i.e., possessing different combinatory possibilities that allow, e.g., for function–argument structure (cf. n. 6). Names have meaning in virtue of going proxy for objects:25 their meaning is what they refer to.26 Elementary sentences then depict elementary situations in virtue of this direct referential relation and identity of their logical form, which is determined by the logical forms of their constituents, i.e, the names and the objects respectively.27 All other sentences are built from elementary ones by means of truth functional operations.28 This preserves the referential character of elementary sentences in a strong way: there is no room for any intensional expressions or constructions, such as modal or tense operators, intensional verbs, and the like. Every aspect of meaning is exhausted by the referential relation between names and objects, and the truth functional operations. But if one approaches things from the other angle, that of the world, a different picture emerges. The Tractatus ontology is a very rich one, in which we find not just actual entities and actual situations (Tatsachen), but also all kinds of possibilities. The notion of an object as such transcends questions of existence, rather, the totality of objects functions as “the substance of the world,”29 by which Wittgenstein means that every possible world is somehow made up from this totality.30 Consequently, the notion of a situation (Sachlage), including that of a state of affairs (Sachverhalt) being an elementary situation, covers not just the factual, but everything that is logically possible. So the ontology of the Tractatus is made up from intensional entities. Possible situations are pictured by sentences, they are their meanings. So the semantics of language includes intensional entities as meanings, yet does not exploit this intensionality in the sense of allowing expressions in the language to refer to it (as modal or temporal expressions or intensional verbs would do). Only in its definition of truth does the Tractatus conception of meaning make use of the intensional character of meanings. Depicted situations being always possible, truth can be defined simply as the property a sentence has when it depicts a situation that is not just possible, but also actual.31 For the rest, Wittgenstein has no use.32 It is not completely clear why Wittgenstein chose to go this way. The reason he uses intensional entities in the ontology is clear: the required onceand-for-all characterization of the realm of the meaningful necessitates the inclusion of all possible situations as possible meanings. So on the ontological side the ingredients for an intensional semantics are available. Presumably, Wittgenstein was of the opinion that an intensional (modal, temporal) logic would not really be a logic in the required sense, i.e., that what underlies all that is possibly meaningful. For it introduces conceptual relations, which within that logic are necessary, but which are not purely logical according to Wittgenstein’s standards (if only because there are alternatives). Some evidence is perhaps forthcoming from the way in which Wittgenstein treats a particular instance of conceptual necessity, viz., that of color exclusion. He claims that the necessity of the truth of the sentences “a is red” excluding

The architecture of meaning

219

that of, say, “a is blue,” must be a logical one, and that hence these sentences cannot be elementary.33 This implies that they are not completely analysed, and that further analysis is needed. Obviously, an intensional semantics would deal with the exclusion in another way, e.g., by restricting the allowed interpretations by means of a postulate. One reason why this would not appeal to Wittgenstein is the conception of logic: it is not just that for him logic is extensional, it is also syntactic in not quite the sense we attach to that phrase nowadays. The distinction between semantic, i.e., model-theoretic accounts of logical validity and syntactic, i.e., proof theoretic ones, at the time was not yet conceptualized in the way it is today. For Wittgenstein the logic of an expression ultimately is displayed in its form: it is by inspecting the form, and only the form, of a completely analyzed expression that all its logical relations with other expressions should be surveyable. No additional rules or postulates may enter into that. And that rules out an intensional semantics in the modern sense of the term. Such a mixture of syntactic and semantic aspects, confusing as it is from a modern perspective, also characterizes Wittgenstein’s use of truth functions. They are not the purely semantic entities (functions from (n-tuples of) truth values to truth values) that we associate with the term. Truth functions are defined in terms of operations on sentences,34 which would make them syntactic in our conception, yet they also have what we would regard as a semantic effect in that they determine the truth conditions of complex sentences in terms of the truth conditions of the sentences from which these sentences are constructed. It is in these terms that Wittgenstein specifies the structure of the meaningful, i.e., the meaningfulness of sentences, and not, as one would expect if one is familiar with modern intensional semantics, in terms of operations on meanings, i.e., situations, directly. Finally compositionality. As was indicated above, for Wittgenstein to be able to achieve his aim, viz., that of determining the realm of ethics ex negativo by determining the realm of the meaningful, he needs, not only a conception of meaning that excludes anything necessary, including ethics, but also an account of the structure of the meaningful. Given how elementary sentences have contingent meanings by picturing contingent states of affairs, what then needs to be shown is how complex sentences have meaning in such a way that the feature of contingency is preserved. Here compositionality comes in. One of Wittgenstein’s strongly held convictions is that the logic of the meaningful needs no “middle men,” there is nothing that serves to form meaningful combinations of signs that is not a meaningful sign itself. Hence, it is names that by themselves form configurations, the elementary sentences.35 Which means that the elementary sentence is a purely compositional combination of signs. It is solely in virtue of their logical form that names determine, in a necessary, a priori way, all their possible combinations, i.e., the totality of all elementary sentences.36 For complex sentences the task of determining the totality of possible meaningful combinations is fulfilled by truth functional logic, which secures compositionality. The fact

220

Martin Stokhof

that Wittgenstein does not rely on logical constants (connectives, quantifiers) as separate categories of expressions, with appropriate separate semantic interpretations, should not disturb us.37 Wittgenstein’s treatment of the logic of complex expressions is actually technically quite complicated, and it would take us too far afield to go into the details, inasmuch as these can be ascertained in the first place. Suffice it to say that by all intentions, Wittgenstein’s treatment is compositional in the sense that besides elementary sentences and the transcendental, i.e. basic and independent logical structure of language and the world, nothing plays a role in determining the realm of the meaningful. Meaning is determinate, non-contextual, and has a self-sufficient logic. Does Wittgenstein’s appeal to truth functional combination as the only machinery to derive complex sentences indeed preserve contingency? Well, almost, but not quite: tautologies and contradictions are the obvious exceptions. But given that they are the only ones, that is not a problem for Wittgenstein: for tautologies and contradictions do serve a function, since they show the logic of the meaningful in a very special way, viz., by being the limits of meaningful combinations of signs.38 Thus the structure of the realm of the meaningful is shown in three ways: in the way in which sentences have meaning by picturing situations; in truth-functional combinations that determine ever new complex meaningful sentences; and in the dissolving of meaningful combination in tautologies and contradictions. And in each case compositionality is key. Time to take stock. Disregarding many details, we can characterize Wittgenstein’s architecture of meaning as an attempt to give a purely transcendental, non-contextual and formal account of the possibility of all meaningful expression. It characterizes the realm of the meaningful by tying language (and thought) and the world together as isomorphic copies based on a purely logical structure common to both. That structure is extensional and compositional, and hence devoid of any material content or links with context or situation. It is like the grand design of a cathedral that rises above its worldly surroundings striving upwards towards eternity. Lofty, overwhelming, and an almost irresistible invitation to transcendence. But actual cathedrals took decennia, often centuries to built, their plans were constantly changed and adjusted, and very often they were only partially executed.

9.3 Quasi-historical interlude: the road from Vienna to Los Angeles In the next section we will sketch how formal semantics arose at the end of the sixties, beginning of the seventies of the last century, partly as a reaction to prevalent opinions about natural language in philosophy, partly inspired by contemporary developments in linguistics. But it was made possible also by the availability of new logical techniques, and if one looks for historically traceable links between Wittgenstein’s ideas from the Tractatus and formal semantics, that is one area to look for one. A real historical investigation of

The architecture of meaning

221

the various often indirect ways in which Wittgenstein’s ideas trickled down and influenced, mostly unconsciously, the work of the pioneers of formal semantics is beyond the scope of this paper. A few sketchy indications must suffice, and these are, moreover, only made for their systematic importance. That Wittgenstein and the Tractatus had great prestige among the members of the Vienna Circle is well-known. Moritz Schlick critically admired Wittgenstein’s work, and Wittgenstein held Schlick in high regard, though also not uncritically. Waismann spent many years taking part in and recording conversations with Wittgenstein, Schlick and other members of the Circle. Some of these acknowledged at some point the influence of Wittgenstein’s work. For our present purposes the most important one among these is Rudolf Carnap. That Carnap’s work, in particular the Logische Aufbau der Welt from 1928,39 was inspired by some of Wittgenstein’s ideas from the Tractatus is well known.40 And in his later work, like Meaning and Necessity,41 Carnap also refers to the Tractatus as a source of inspiration. For the theme of the present paper it is not so much Wittgenstein’s presumed or real influence on Carnap with regard to such issues as physicalism,42 philosophy, or scientific method, including the issue of verification and meaning, that are relevant, but rather the ontological backdrop of meaning and reference. In Meaning and Necessity Carnap develops what he calls “the method of state descriptions and ranges,” as a means to give explicit definition of various concepts, such as truth and equivalence, intension, and extension. When he introduces the method he states that “some ideas of Wittgenstein were the starting point for the development of [it].”43 Carnap likens his notion of state description to Wittgenstein’s concept of state of affairs (and to Leibniz’s concept of possible world). And as another example of Wittgenstein’s influence he mentions the idea that “to know the meaning of a sentence is to know in which of the possible cases it would be true and in which not.”44 The state description and range method was used by Carnap to give, among other things, a semantics for various modal notions, and to give a formal construction of the concepts of intension and extension that avoided some difficulties that Carnap saw with Frege’s notions of sense and reference. The details of the difference of opinion between Frege and Carnap need not concern us here. What is important to note is the way in which Carnap “linguistifies” what in Frege’s and in Wittgenstein’s theories was an ontological concept, that of a possibility. As we saw in section 9.2, Wittgenstein introduced in the Tractatus an ontology that consists of possible states of affairs (which are the atoms) that make up possible situations that together fill logical space. The world is identified with “the totality of existing states of affairs,”45 suggesting that any combination of states of affairs forms a possible world. Note that this concept thus is inherently partial in nature. Carnap re-analyzes this ontological perspective in linguistic terms. A state description is defined in terms of what Carnap calls “a semantical system,” which is basically a formal language (e.g., that of first order predicate logic

222

Martin Stokhof

with iota and lambda-operators) along with an interpretation specified in some metalanguage (e.g., English, suitably enriched). A state description in a semantical system is any set containing for all the atomic sentences of the system either that sentence itself, or its negation. The concept of a sentence holding in a state description is defined recursively on the basis of atomic sentences belonging to it or not. Then the range of a sentence is defined as the set of all state descriptions in which it holds, and in terms of these concepts Carnap then continues to define various logical and intensional notions. The consequence of this linguistic re-analysis is that many ontological issues become linguistic ones, something that of course fits in well with the main lines of thought characteristic for logical empiricism. And in view of the isomorphic relation that the Tractatus constructs between language and world, such a redefinition might even be considered to be in the spirit of that work as well. But there is one crucial difference. For Wittgenstein in the Tractatus the language that is isomorphic to the world is a concept that is both singular and universal: it is a language, not languages, but this singular language is universal, it is language-as-such, the essence of any form of symbolic representation. In Carnap’s case languages are plural and particular: there is a wide variety of semantical systems and for each of them the notion of a state description is extensionally different. In such a set-up reducing ontological concepts to linguistic ones implies a form of relativism, that is emphatically not present in the Tractatus framework. To be sure, Carnap welcomed this consequence and continued to make use of it, e.g. by analyzing questions concerning abstract entities in terms of what he came to call “linguistic frameworks.”46 But from the perspective of what Wittgenstein wanted to achieve with the Tractatus, such a relativism is not helpful or interesting at all. Interestingly, although Carnap’s analyses have proved to be very important and influential, especially in contemporary American analytic philosophy and logic, it was precisely this linguistic re-analysis of the ontological notion of a possible world that in the end proved to be an obstacle for obtaining a satisfactory, uniform semantics for the large variety of systems that by that time were characterized syntactically by means of various (additional) axioms. With state descriptions representing possible worlds, Carnap’s semantics for the modal operators effectively comes down to quantification of the total set of worlds, which yields an adequate semantics for the strong system S5.47 In order to obtain satisfactory characterizations of the semantics of weaker systems (S4, B, K, etc.) one needs a weaker notion, of quantification over some subset of worlds properly related. This relational semantics, also known as possible world semantics, turned out to require a “re-ontologization” of state descriptions. Textbook wisdom associates the first satisfactory formulation of possible worlds semantics for modal logic with the name of Saul Kripke, but a number of people have been responsible: Carew Meredith, Arthur Prior, David Kaplan, Stig Kanger, Richard Montague, and Jaakko Hintikka, to name a few.

The architecture of meaning

223

But by the time possible world semantics became a standard the link with, Wittgenstein and the Tractatus was by and large broken,48 and when possible worlds semantics was given a significant role in formal semantics this was done with hardly any awareness of the role the Tractatus had played. It was Frege and Carnap that people referred to in this connection, not Wittgenstein.

9.4 Pioneering natural language semantics In the late sixties and early seventies several people, from philosophy and from logic, started to develop systematic and formally explicit theories that attempted to deal with the semantics of natural languages along lines that were familiar from logic. In doing so, they felt they were going against prevalent opinion, which held that natural languages, being as complex, ambiguous, and vague as they are, do not lend themselves to rigorously formal treatment. And a conclusion that some, though not all, people who held this conviction connected with it was that, for that very reason, natural languages were not suitable instruments for philosophical analysis. The pioneers of formal semantics all rejected the first point, and most of them also the second one, though here opinions sometimes vacillated somewhat. A clear statement of the goal of the enterprise, that reflects these motivations, comes from a 1970 paper by Richard Montague, aptly titled “Universal Grammar”: There is in my opinion no important theoretical difference between natural languages and the artificial languages of logicians; indeed, I consider it possible to comprehend the syntax and semantics of both kinds of languages within a single natural and mathematically precise theory. On this point I differ from a number of philosophers, but agree, I believe, with Chomsky and his associates.49 The message conveyed here is clear: other than a number of philosophers, Montague thinks that the tools and techniques used to deal with the languages of logic can be applied to natural languages because in the end, when it comes down to fundamental properties, there is no difference between the two. Both can be captured with the same means. Notice that Montague appeals to Chomsky as a supporter of this cause, the idea being that Chomsky had provided evidence that the syntax of natural languages can be dealt with with formal precision, which was regarded as an obvious requirement for a formally precise semantics.50 In its turn this assumption depends on another one, regarding compositionality, about which more below. Similarly Davidson, in his seminal 1967 paper “Truth and Meaning” claims that Tarski’s worries51 that “we would have to reform a natural language out of all recognition before we could apply formal semantic methods”52 are ill-founded and he, too, appeals to Chomsky, and on a more positive note than Montague, when he says that “recent work by Chomsky

224

Martin Stokhof

and others is doing much to bring the complexities of natural languages within the scope of serious theory.”53 For Davidson the differences between formal languages and natural languages are outweighed by the similarities inasmuch as we can get a proper semantic theory for a natural language, such as English, starting from interpreted formal languages as follows: [P]ick one as much like English as possible. Since this new language has been explained in English and contains much English we not only may, but I think must, view it as part of English for those who understand it. For this fragment of English we have, ex hypothesi, a theory of the required sort. Not only that, but in interpreting this adjunct of English in old English we necessarily gave hints connecting old and new. Wherever there are sentences of old English with the same truth conditions as sentences in the adjunct we may extend the theory to cover them. Much of what is called for is to mechanize as far as possible what we now do by art when we put ordinary English into one or other canonical notation. The point is not that canonical notation is better than the rough original idiom, but rather that if we know what idiom the canonical notation is canonical for, we have as good a theory for the idiom as for its kept companion.54 The passage deserves to be quoted in full since it illustrates, not just the idea that formal languages can be used to provide the semantics of a natural language, but also a view on the process of formalization in semantics that is quite characteristic.55 The idea is that a proper semantic theory for a natural language can be obtained by defining a mechanized mapping of the sentences of that language into formulae of some suitable, interpreted formal language. Such a procedure has to meet requirements of completeness and explicitness that make it mechanical in the sense of not dependent on any particular knowledge and skills one might need in order to apply it. In these respects it then mimics what are assumed to be characteristic features of semantic competence. A couple of things should be noticed at this point. First of all, the pioneers clearly shared with generative linguistics the conception of a natural language as an infinite object, i.e., as an infinite set of structured expressions. This is already evident from their use of formal languages, which are defined to be that way, as models for natural languages. And some authors are even more explicit in that they agree that the goal of a semantic theory is to account in a finite way for the learnability of an infinite language and for the language users’ competence in dealing with a potentially infinite number of new sentences and new meanings. In Davidson’s early papers, for example, this is a recurrent theme.56 A second thing to be noticed is that compositionality plays a key role. If the infinite number of meanings of an infinite number of natural language

The architecture of meaning

225

expressions is to be defined in a finite manner, so as to account for both learnability and usability, then compositionality seems, if not necessary, at least sufficient.57 The idea is that by defining compositional semantic analogues of syntactic rules one can account for these features of natural languages, since by making use of the syntactic structure one is able to define the meaning of a complex expression in terms of the meanings of its components, in the end all the way down to those of its lexical elements. The link between compositionality and competence brought formal semantics close to Chomskyan linguistics, but not all the way. The pioneers coming from philosophy and logic consciously or unconsciously subscribed to the Fregean idea of logic, and, by extension of logic as applied to natural language, as a normative discipline, to be distinguished sharply from psychology. Frege held that logic does not describe how we actually think, but prescribes how we ought to think. Analogously, many formal semanticists held that their theories exhibited the ideal structure of natural language meaning, and that what they were aiming at was a complete and transparent representation of the logical relationships between sentences, not a description of how these sentences are actually produced and understood. So for them formal semantics was empirical (and in that sense different from logic) only to a certain extent, i.e., only insofar as it deals with an empirically given language. For formal semanticists coming from Chomskyan linguistics, on the other hand, the idea of semantics being an account of underlying psychological processes was not that strange.58 Finally, note that there is an assumption at work when one claims that formal languages can model natural languages, viz., that by abstracting away from the surface differences between formal languages and natural languages nothing of significance is lost. More in particular, the meanings of the natural language expressions remain essentially intact. One difference that is particularly relevant in the context of the present discussion is that whereas natural language expressions are geared towards use in context, and actively employ contextual resources to convey meaning, the expressions of formal languages are context-independent, and deliberately so. It will be clear that abstracting away from this difference favors a highly de-contextualized concept of meaning. It also makes the relation between form and meaning in a natural language very indirect, up to a point where the same meaning can be expressed by any arbitrary form, provided some constraints regarding category and type are met. This clearly points towards a notion of meaning that is not only not contextual, but universal and absolute. In the picture that emerges the central goal of formal semantics is to deal with objective meanings of natural languages, conceived of as infinite structures, in a precise, mechanical and finite way. What should be noted is that the equation “precise = formal” is applied without much further ado. It is assumed, rather than argued that precision needs formalization, or at least the use of a formal language as a counterpart to the natural language that is being studied. The assumptions on which the concept of formalization rests,

226

Martin Stokhof

are complex and diverse, and it is not too often that one finds a more elaborate defence than a quick pointer to ambiguity and vagueness as properties of natural language expressions that necessitate it. The use of a distinction between grammatical form and logical form is more often a sleight of hand than a well-argued application of a clearly conceived distinction.59 However, an investigation of them would be beyond the scope of this paper. But it is worth noting in the present context, since sometimes one gets the impression that formal semanticists are as adverse to language reform and hence as ambivalent as Wittgenstein was when he claimed that natural language is all right60 yet also propagated the idea of analysis by means of reformulation in a language that obeys logical syntax61 into a completely perspicuous representation, from which the logical properties could simply be read off. And of course this is a point where Wittgenstein’s later criticisms of the Tractatus straightforwardly apply also to formal semantics. Another equation that formal semantics applies is “formal = logical.” Again, not much argumentation can be found in the original writings of the pioneers, which in this case probably is understandable. At the time logic pretty much had the field to itself.62 The semantics of procedural programming languages still had to be invented, and other formal modeling techniques were also still in their infancy. The application of concepts and tools from fields other than logic, then, had to wait another decade. And then there is a third equation that is applied, one that might appear more puzzling. It reads “logical = model-theoretic.” Again, if one looks at the seminal papers of Montague, Davidson, and others, not much argumentation for this equation can be found. The idea of a proof-theoretic approach is never raised, discussed and rejected, and it would only be much later that the first attempts to do semantics of natural languages by some more syntactic methods would appear.63 There may be various reasons for this, such as a shared rejection of the various forms of verificationism that had characterized logical empiricism; the fact that possible worlds semantics had constituted such a breakthrough in modal logic; or perhaps even Gödel’s incompleteness results. That the expressive power of natural languages equals at least that of second order predicate logic seems universally accepted at the time of the pioneers. It was only much later, when work on computational applications started to get serious momentum, that questions concerning complexity and effectiveness gained equal footing with considerations concerning expressive power. This general scheme, that aims to realize a shared ambition, viz., that of a formally rigorous, model-theoretic semantics for natural languages, still allows for a variety of ways in which it can be made concrete and applied. One major dimension along which one may find differences between implementations is the kind of semantics that is considered needed, or allowed, in particular with regard to the use of intensional entities, such as possible world, properties, propositions, and the like. Interestingly, argumentation here is not just empirical, but also, and in some cases predominantly, philosophical.

The architecture of meaning

227

For some, such as Montague the full range of intensional entities is in principle available to be used in modeling natural language meanings, as is testified by his ample use of intensional type theory.64 Similarly, Cresswell uses set-theory to construct a rich ontology, which includes a variety of intensional objects, including events.65 Davidson, on the other hand, rejects a wholesale appeal to intensional objects and uses an extensional ontology to define his truth conditional semantics.66 And then there is also what looks like an “agnostic” attitude expressed by David Lewis: “In order to say what a meaning is, we may first ask what a meaning does and then find something that does that.”67 This may look like a completely pragmatic approach, but appearances may be deceiving here, since Lewis says at the beginning of his paper that his intention is “to propose a convenient format for semantics general enough to work for a great variety of logically possible languages” and that answering such questions as “What sort of thing is a meaning?” is part of that.68 This suggests that for Lewis, too, the choice of a framework for semantics really is a philosophical issue, a suggestion that is reinforced by his staunch realism with regard to modal concepts, which is in part defended by means of detailed analyses of natural language constructions.69 That the way in which the pioneers go about developing their respective frameworks for formal semantics of natural language is rife with philosophical considerations, is also apparent from the fact they are not just concerned with the notion of meaning in a strict sense, but treat issues concerning semantics as closely linked to broader philosophical ones. For example, Davidson’s choice of semantic framework is in part based on ontological considerations, and he uses that very same framework also to tackle philosophical questions, such as the ontology of individuals, time, and events, causation, and so on.70 Perhaps less well known are Montague’s views on this matter: I have reason to hope that the ambiguities I have pointed out will confirm two points sometimes mistakenly supposed incompatible: there is philosophic interest in attempting to analyse ordinary English; and ordinary English is an inadequate vehicle for philosophy.71 Obviously Montague, like Davidson, is of the opinion that semantic and philosophical issues are intimately related. True to “the linguistic turn” he claims that the analysis of language is key in dealing with philosophical problems. The passage just quoted appears at the very end of the paper, but in the opening paragraphs, too, we can witness the same seamless going back and forth between analysis of language and ontological and epistemological questions: “It has been maintained that we need not tolerate such entities as pains, events, tasks, and obligations” is the opening statement, which introduces the ontological problem of certain abstract entities.72 That problem is immediately tied to language, e.g., through the observation that we may not need such entities as pains for sentences like “Jones has a pain” (which can

228

Martin Stokhof

be analyzed as a simple subject—predicate judgment), but that we do need them for sentences like “Jones just had a pain similar to the one he had yesterday”(since the referential function of the indefinite noun in this case can not be explained away in terms of a predicate, without loss of meaning). Hence the analysis of such sentences enforces an ontology containing such abstract entities as pains, obligations, events, and the like. Thus semantic analysis has ontological consequences. And vice versa, the proper way to go about, according to Montague, is “to investigate the nature of the entities in question, construct an exact and convenient language in which to speak of them, and analyse the pertinent notion of logical consequence.”73 Time to take stock again. Although formal semantics of natural language aimed right from the start to be applied in the study of empirical phenomena, and soon after its inception was so applied, broadly and successfully, its origins are remarkably strung with strands of philosophical conviction. Its architecture of meaning, like that of Wittgenstein’s in the Tractatus, is built, primarily, it seems, on philosophical, conceptual considerations regarding meaning in general, many of which are derived from other branches than those dealing with natural language meaning as an empirical phenomenon. And the requirements imposed, too, are by and large motivated by considerations of explicitness and elegance that bespeak their origins in logic. Unlike in the Tractatus, which is after all a single man’s work, formal semantics displays a certain variety, e.g., with regard to ontology, but again it is remarkable that the arguments adduced on such issues are mainly of a philosophical bent, even where they explicitly address facts from a given natural language.

9.5 Resemblance and difference In this section we will confront the characteristic features of the Tractatus, discussed in section 9.2, with the main assumptions of formal semantics outlined in the previous section, and see to what extent they line up, in particular with regard to the philosophical assumptions underlying them. Again we will be concerned primarily with formal semantics in its initial stage, when the enterprise was taking shape. As we already pointed out, the ideas of the original pioneers quickly caught on and drew more and more people to this emerging field, not just from philosophy or logic but increasingly also from linguistics. To a certain extent that changed the discipline, at least in the sense that discussion of principles soon gave way to description and explanation of empirical semantic phenomena in concrete natural languages.74 However, that principles are not actively discussed does not imply that they are not operative. In section 9.6 we will have the opportunity to briefly mention some instances of their continuing influence. But for now we restrict our attention to the pioneering stage. Before we turn to that discussion, however, it might be useful to state once again that our considerations here are systematical, not historical. In fact, there is very little or no awareness among formal semanticists that the

The architecture of meaning

229

Wittgenstein of the Tractatus could count as a predecessor of their trade. The pioneers all refer to Frege, to Carnap, to Tarski, they acknowledge the influence of these thinkers and often discuss their ideas in detail. But Wittgenstein? Montague does not mention him, Cresswell only refers to him in connection with the truth conditional view of sentence meaning. References to his work in Davidson are sparse and never address issues concerning semantic theory proper. The same holds for many other authors. And it does not get better over time. References in the textbook introductions that start to appear from the eighties onwards all display the same pattern. The early Dowty does not discuss the history of the discipline at all;75 Gamut discusses Frege (in some detail), Tarski, and Carnap, but mentions Wittgenstein only in connection with the truth conditional view;76 de Swart mentions Frege and Tarski,77 Chierchia and McConnell-Ginet add a reference to Carnap,78 but in both Wittgenstein is conspicuously absent. And this is curious, indeed. Frege never proposed a formalization of natural language, Tarski explicitly entertained serious doubts about the applicability of his truth theory to natural language, and Carnap used the wrong conceptual apparatus. Yet the one figure that is not mentioned actually is the closest predecessor of formal semantics. But let us leave the riddle of how a discipline constructs its history, and turn to our systematic issues. The case of two of the three characteristics of Wittgenstein’s architecture of meaning that we introduced in section 9.2, viz., those of compositionality and intensional referentialism, is relatively clear. It is mainly the third feature, that of universalism, that may not be that obviously represented in formal semantics. Let us start with compositionality. Both in the Tractatus and in formal semantics compositionality was a key feature, one that was so characteristic for the entire undertaking that it is hardly conceivable what it would be without. As was pointed out above, formal semantics imports compositionality from the formal languages of logic, which simply come with a semantics that is defined in that way. The most common route is to model natural language meanings indirectly, via translation of natural language expressions into those of a logical language which has a compositional semantics. If the translation relation itself is compositional (which it is, if it is a homomorphic mapping), then the compositional semantics of the formal language translations induces a compositional semantics of the natural language source expressions. A good example of this “indirect interpretation strategy” is what was for a long time the standard model of formal semantics, viz., Montague’s PTQ-model.79 What is important to note, since it brings the Tractatus system and formal semantics closer together, is that indirect compositional interpretation can also be transformed into direct compositional interpretation, which is then one crucial feature that the Tractatus and formal semantics share.80 But, as we have also seen, there are differences as well. For Wittgenstein compositionality is first and foremost a fundamental characteristic of the

230

Martin Stokhof

realm of the meaningful, one that allows a determination of that realm “from within,” i.e., by specifying the base case, of elementary sentences, and the production rules, i.e., the compositional, truth functional combinations. In formal semantics compositionality is imported from logic and then used to explain features of competence, in particular creativity, i.e., the ability to produce and understand novel expressions. For Wittgenstein this is not a real concern. Although he does mention the aspect of creativity (though not by that name) in the Tractatus,81 the concept of individual competence has no role to play in the Tractatus scheme of things, because of the general nature of the latter: it is not individual languages, let alone individual users of those languages, that Wittgenstein is concerned with, but language as such. Conversely, no requirement to characterize meaning from within arises when we consider the semantics of a particular language. In fact, using the object language/ metalanguage distinction we can describe the semantics of any particular language “from the outside.” Hence Wittgenstein’s specific reason for doing things compositionally does not apply. And indeed what we see are different reasons being given, mainly centering around creativity and learnability. What is important to note is that although at this point there is a clear difference between Wittgenstein and formal semanticists, it plays out against the same backdrop, viz. a view of language as an infinite object, an entity in its own right, that Wittgenstein and the formal semanticists share. That concept of language is typically dissociated from actual use, actual users, actual competence. It leads to non-contextual notions of meaning, which then need to be supplemented with contextual elements, such as speaker’s meaning, implicatures, and the like, and it strongly favors individuality, i.e., a conception of competence as the property of an individual: it is individuals as such that are competent, and hence it is individuals that have (grasp, produce) meanings. Language being conceived as an infinite object and competence being considered as an individual property, a problem arises: How can one give a finite characterization of such an object? How can one learn it? How can one use it to its fullness (at least potentially)? That is not three different problems but three variants of one and the same problem, and compositionality is called to the rescue in each case.82 As regards intensional referentialism, we first of all have to take into account the difference of opinion among formal semanticists with regard to the use of intensional entities as meanings. As we saw, for Davidson such entities have no role to play, which means that he disagrees at this point with the Tractatus framework, in which the meaning of a sentence, being the situation it depicts, is such an entity. Those who work in the possible worlds framework, however, couldn’t agree more.83 There is definitely a sense in which Davidson is more akin in spirit to the Tractatus, since, as we saw above, Wittgenstein does forgo the use of intensional meanings in defining the semantics of complex sentences. There he uses extensional truth functions, not the intensional situations the ontology supplies. And it is probably because

The architecture of meaning

231

of that when the Tractatus is referenced at all, it is Wittgenstein’s equation of knowledge of meaning with knowledge of truth conditions that is cited.84 Another question with regard to intensional referentialism concerns reference as such. As we saw the existence of a referential relation between names and objects is one of the central elements in how elementary sentences depict states of affairs, i.e., in how they have meaning. Likewise, a determination of the references of basic expressions as a preliminary to the determination of the meanings of complex expressions is an integral part of the various possible worlds type approaches in formal semantics.85 Some approaches in formal semantics, notably Davidson’s, exemplify a holistic approach to the role of reference with regard to meaning. There reference is not a relation that can be determined independent of the meanings (truth conditions) of sentences. Thus Davidson insists that we can not explain reference within a semantic theory, but that we can only employ it.86 That seems congenial to the contextuality principle with regard to reference of names that Wittgenstein formulates in the Tractatus,87 where he states that “Only sentences have meaning; only in the connection of a sentence does a name have reference.” For Wittgenstein, too, it holds that the references of names can not be explained independently from the meanings of the sentences in which they occur, which serve to elucidate them. Does this imply that the reference relation is not fixed, but can be reinterpreted in any way that is consistent with the meanings assigned to sentences? This is a question that makes sense in the modern model-theoretic perspective, where interpretation is defined relative to a model, and where different models may share some properties (say, assign the same truth conditions, or sets of possible world, to a certain class of sentences) yet differ in other respects (for example, in what reference is assigned to what basic expressions). In the Tractatus that distinction is not available. In modern terminology: Wittgenstein is interested in absolute interpretation, not in relative interpretation. But then reference takes on a different meaning, and needs to be thought of more as an indication of a semantic role than as denoting a semantic value.88 The distinction between absolute and relative interpretation connects the discussion of intensional referentialism with that of the third characteristic of the Tractatus’s theory of meaning, universalism. We venture the hypothesis that formal semantics is committed to universalism in a way that is quite comparable to the Tractatus. For what formal semantics shares with the Tractatus is a concern with laying bare the underlying logical structure of language as such. There is a difference, of course. In the Tractatus the universalism is directly visible, since it is the only thing that matters. In formal semantics, which aims to provide semantic descriptions and explanation of concrete languages, the universalism is present in an indirect way, viz., the tools that are being developed and principles that are being adopted to realize that descriptive and explanatory aim. Wittgenstein has no such aims, his interest is in a transcendental deduction of the basic principles underlying any form of symbolic representation. As we indicated above, ineffability is a

232

Martin Stokhof

necessary consequence. Whether that applies to formal semantics as well depends on the extent to which it is committed to universalism. So let us take a closer look at that. First we try to substantiate the claim that formal semantics is committed to universalism, in some form or other. And second we will investigate whether the availability of modern conceptual tools, such as the object language/metalanguage distinction, makes a difference for the consequences of that universalism. As said, in formal semantics the universalistic claims are not in the actual descriptions and explanations that applications of the frameworks are supposed to produce. By and large these are language specific, and even when they generalize over certain language families, or even aim to identify universal properties of natural language semantics, they do so from essentially an empirical point of view. And as such their goals and hence the requirements they should meet are categorically different from what applies in the case of the Tractatus. So where are the universalistic assumptions located? It is in the framework as such. The definition of a framework for doing semantics, such as Montague’s universal grammar, or Davidson’s truth conditional semantics, embodies the claim that it is by the use of this particular formal apparatus, and hence by means of these assumptions and principle about meaning that apparatus embodies, that all meaning can be described. Typically, the canonical frameworks for formal semantics that were developed in the pioneering days start out with a completely general definition of what meaning is, what its structure is like, and so on, only subsequently to apply it in the description and explanation of concrete semantic phenomena.89 And that this is not all that different from what Wittgenstein does in the Tractatus. The difference is that in the Tractatus the emphasis is exclusively on the framework (and then only its most general aspects) and that actual applications simply do not figure in the text as such. But the relationship between framework and possible application in the Tractatus and in formal semantics appears to be quite comparable. And the pioneers themselves would seem to agree to an ascription of universalism in some form. Recall the first sentence of the opening paragraph of Montague’s “Universal Grammar,” which we quoted at the beginning of section 9.4: “There is in my opinion no important theoretical difference between natural languages and the artificial languages of logicians,” clearly a statement that conveys universalistic aims.90 Davidson, too, is quite explicit about the universal nature of his framework. This is already evident from the completely general way in which he argues that a Tarskian theory of truth gives us the basic structure of semantic theory: in that derivation no contingent features of particular languages or formalisms play a role (other than illustrative).91 And the way in which he uses that framework in settling other issues, such as conceptual relativism, or ontological issues, also bespeaks a universalistic approach.92 But what about the distinction between absolute and relative interpretation? Actually, there seem to be three positions one might take here. First of

The architecture of meaning

233

all, one may aim to give the interpretation of a language, i.e., to give a complete description of the actual meanings of all the expressions of a specific language.93 Second, one may be intent to specify suitable interpretations of a language, i.e., to give a partial description, which fully specifies some of its meanings, but leaves others unspecified (i.e., determine their type but not their content). Third, one may drop the language parameter, and look for general features that characterize meaning in general. Note that these three options can not be simply ordered in terms of generality, since the presence or absence of the language parameter also changes the character of the undertaking, making it empirical or conceptual, respectively. It is clear that the Tractatus’s goal is the third one, and a framework like Montague’s aims for something like the second option. One of the main differences between the first and the second is that semantic specification is total or partial, respectively. Note further that in a standard, referential setting the first option does not make much sense (as an aim for semantic description, that is), since it would involve determining the actual references of all the expressions which implies the actual truth values (not just truth conditions) of all the sentences. So the only way to stick to the idea of semantics aiming for absolute interpretation in the first sense is to loosen up on reference. And this is what Davidson does. He insists that semantics should be what he calls an absolute theory, in which no relativization to a model (interpretation, domain, world) occurs.94 In the Tractatus, too, the semantic theory is an absolute one; it is the one true theory. And here, too, reference plays a different role than it does in theories that take the second position. This shift in the role of reference that the first and third options bring about is not without conceptual complications, as we have seen. For one thing, a semantics that does not account for reference seems strangely incomplete. So presumably, that is one reason why the second option is still very much in favor. Another reason concerns the distinction between lexical semantics and structural semantics. Intuitively, it seems clear that we should aim for an absolute theory: if the semantics of, say, English is what we are after, then it is a complete description that we want. Everything else would be substandard. However, most theories will in fact take the second position, and usually this is motivated by making a distinction between an account of the content of lexical expressions, lexical semantics, and a specification of operations that build complex meanings from simpler ones, structural semantics. The difference between the two seems intuitive and clear enough, and almost innocent, but it remains to be seen whether it really is a distinction that can be made in a theory-neutral way. The difference between word and sentence, for example, might have some intuitive, theory-independent ground, yet any explication of it would be theory-laden, and as an intuitive distinction it would not support full-blown compositionality as a real phenomenon, but at best only as a theoretical device. For one thing, it is quite obvious that in formal semantics the distinction is in fact inherited from the way in which formal languages operate. It is

234

Martin Stokhof

because the semantics of natural languages is modeled after that of formal languages (among others through the very use of those formal languages in their description) that we end up with this way out. For formal languages the distinction makes sense: it is the function of a formal language to enable us to abstract away from what we are not interested in, and to just focus on what we do want to treat. After all, that is how formal languages are designed to operate. But a semantics for a natural language that is true to its calling can not follow suit, at least not ultimately. Of course, here too, we can do only so much at a time, so in that sense the use of a formal language can function in the same way. But do note that this is not at all the same as modelling a natural language after a formal one. So it seems that what is more or less the standard way of doing formal semantics of natural language is heavily influenced by unconscious borrowings from formal language. Certain properties of the latter are projected on the former by being built into the general framework.95 And precisely in that lies the (implicit) universalism. Let us now turn to the second issue raised above, ineffability. What formal semantics shares with the Tractatus is a concern with laying bare the underlying logical structure of language. Of course, the various frameworks that have been proposed differ on what that structure looks like, but that is a material difference, and hence immaterial. For Wittgenstein universalism implies ineffability: logical structure can not be described, it can only be shown. Formal semantics could rely on Tarski’s distinction between object language and metalanguage to get around this, but that works only (or, more carefully, might work only) if we distinguish between formal languages, to be employed as metalanguages, and natural languages, being the object languages that we want to describe. Two remarks are in order. First of all, such a use of the object language/metalanguage distinction flies in the face of universalism, that, as we just saw, also characterizes formal semantics. After all, Montague claimed that there was in his opinion no difference between formal and natural languages to begin with. Second, if we grant the distinction, another deep and serious problem arises, the threat of circularity. If we use a formal language to specify the semantics of natural language (all of them), then we can do so only if somehow the semantics of the formal language is given independently. But what is the semantics of a formal language other than its description given in, ultimately, a natural language? Of course this problem does not diminish the usefulness of the object language/metalanguage distinction for giving concrete descriptions of particular languages or classes of languages. But it is a serious problem for a framework that wants to be universal, i.e., one that wants to talk about meaning as such. The problem is completely obvious if the universalism encompasses all languages, formal and natural. But it is arguably also a problem in case we limit our universalistic aims to just natural languages, because in that case we would need to have access to the semantics of the formal language(s) that we use, via some other route than a natural language. But what could that be?

The architecture of meaning

235

Here, it seems, formal semanticists and Wittgenstein (of the Tractatusdays) are in the same boat, the difference being that the latter is aware of the consequences of his universalism and embraces ineffability wholeheartedly, whereas the former in general seem oblivious to the conundrum that is the result of their universalistic aspirations. In general, for in Davidson’s case one could argue that his distinction between a semantic theory, being a Tarskian theory of truth that functions as a specification of the semantics of natural language expressions, and a theory of interpretation, being a theory that accounts for how we can arrive at such a semantic theory, indicates an awareness of the problem. Whether it also solves it, is another question, one that would lead us too far afield to answer here.

9.6 Consequences This final section is concerned with a very brief discussion of two questions concerning the status of modern-day formal semantics. From the pioneering days that have been the main subject in the foregoing sections, formal semantics has developed into a thriving branch of linguistics. As was indicated earlier, the initial phase was very much concerned with laying down the principles of the enterprise, concentrating on conceptual and formal issues, but in the next stage the emphasis shifted to description and explanation of concrete empirical phenomena. During that development the theoretical frameworks changed as well, of course. Was the initial phase one of uncovering structure, and defining the tools needed to deal with that, what came after that was very much one of reinstating content, and modifying and adapting the formal machinery inherited from the pioneers to do so. Gradually, uniformity gave way to diversity, and radical changes were proposed and implemented. That has led to the present-day situation in which the term “formal semantics” denotes a single, coherent field only in name. A wealth of different frameworks and approaches are pursued almost in parallel: dynamic semantics, discourse representation theory, optimality theoretic semantics, game theoretic semantics, to mention just a few. The first question is whether the philosophical elements of formal semantics in its early days, that we have discussed above, have survived the development of formal semantics into its present state. For example, we have seen that universalism is connected with an a prioristic, transcendental point of view. Can that simply be dropped, or would that have further consequences? Alternatively, if it is intrinsic to universalism, what does that imply for the status of semantics? The second question is whether there are any lessons still to be learnt by semantics from Wittgenstein’s work after the Tractatus. Wittgenstein formulated some penetrating criticisms of the Tractatus, e.g., in Philosophical Investigations, some of which are directly aimed at the by now familiar aspects of universalism, compositionality, and intensional referentialism. The core question here is whether Wittgenstein’s objections have consequences

236

Martin Stokhof

for the theoretical frameworks of formal semantics. Obviously, one’s answer to the second question depends in part on what the answer to the first question is taken to be. As for that first question, one thing that should be noted is that universalism has had an influence on the methodology of formal semantics. Characteristic for the way in which semantic description and explanation is carried out is a specific way of obtaining and assessing the relevant data, viz., via introspection. It is by consulting one’s own judgments on what sentences mean, how they are related (by implication, equivalence, exclusion), whether they are ambiguous, and so on, that one proceeds in one’s semantic research. Of course consultation of other speakers is not excluded, but in principle it is not needed either. This methodology is in part due to the individualistic nature of competence, which in its turn is partly the result of the central role played by compositionality. But arguably the a priori nature of semantic theory plays a role here as well. The adoption of introspective methodology has had far-reaching consequences, for one thing it has made the introduction of formal semantics as part of overall linguistic theory easier, since the very same methodology was characteristic for linguistics in the Chomskyan tradition. But it also perpetuated a particular view of language, the structuralist conception of language as an entity in its own right, that may be the result of historical developments and social structures but that can be studied profitably in isolation. Recently this has changed somewhat, also in formal semantics, and statistical methods and corpus-based studies are now part of the methodological toolkit of the formal semanticist as well. Another effect of the individualism spurred by compositionality and strengthened by universalism is an increasingly close association of semantics with cognitive neuroscience. If what semantics studies is competence, and competence is individual, then there has to be a relation between the descriptions that the semanticist gives of meaning and the investigations that the cognitive neuroscientist conducts of what happens in the brain. The nature of the match (identity, some transformation, etc.) may be debatable, but the existence of a link seems given. That is to say, it is a given once semantic competence is conceptualized the way it is. And that conceptualization is derived in part from the semantic framework as such. The issue of reductionism looms large over all of this, but not as an option, or even a possibility, but rather as something that will inevitably happen. Note that it can only be turned into something that we can debate if we can step back and ask anew what meaning is, and why would it be the case that meaning is best modeled as some kind of object that “fits in a user’s head”? And there may not be a real answer to that question either. It seems that the question we really should be asking is what we want semantics to do. The interesting (and perhaps confusing or upsetting) thing about that question (especially the last one: what do we want semantics to do?), is that it really allows for quite different answers in different contexts:

The architecture of meaning

237

what semantics could or should do in a computational setting (e.g., a translation system that is geared towards a dedicated domain, or a man–machine interface that uses natural language) will not be the same thing as what it could or should do when we study certain aspects of language processing. And within that broad domain it will probably be different according to whether we talk about the higher order level of cognitive psychology, or about what goes on at the basic neuronal level. And if we are interested in linguistic aspects (e.g., in a typological study) or in variation from a sociological perspective, or when we are trying to find out what the syntax– semantics interface is all about, we get other sets of answers every time. This is disconcerting because it seems to rob semantics of its object: meaning appears to be not a natural kind. But as our discussion of formal semantics has illustrated, the idea that it is a natural kind, i.e., a natural phenomenon that has an intrinsic unity, is philosophically loaded. We can assume it, we can even think it is obvious, but from a philosophical perspective it is not a neutral assumption, it comes with implicit claims that need to be defended. At least, we should be able to decide what proper identity conditions for being a meaning, or application conditions for being semantic, might look like. Such specifications have been given, although not necessarily in these terms, but we can observe that the best and most explicit theories around are quite specific philosophical theories. But of course philosophically there is an alternative. A use-based theory actually provides a framework for thinking about meaning that is much less restrictive. This observation brings us to the second question, whether the criticisms Wittgenstein adduces in his later work on his own ideas in the Tractatus, should have had an impact on formal semantics. Of course we cannot survey these criticisms here, we will confine ourselves therefore to the general question of the influence they may have had, or should have had. With regard to the pioneering days, it should be noted that Wittgenstein’s later work, which at the time consisted mainly of Philosophical Investigations, was not as widely known in the late sixties and early seventies as it is today. Also, the reception of Wittgenstein’s later work in mainstream philosophy, certainly in logic and related disciplines was initially rather hostile and dismissive. In retrospect there are affinities between Wittgenstein’s later work and ideas that Quine and Sellars were developing. But Wittgenstein’s general philosophical attitude for a long time remained strange to most philosophers. In particular, the philosophers who were actively engaged with formal semantics in the pioneering days, and who also deeply influenced the linguists that were taking part, were all very much formal-minded, logic-oriented, problem-solving types of philosophers. Hence the influence of Wittgenstein’s later work was virtually non-existent. So, if Wittgenstein’s criticisms of the Tractatus did not in fact change formal semantics at the time, should it have? In our discussion above a couple of issues were hinted at that provide connections. If we look at the main elements in Wittgenstein’s rejection of his earlier ideas, we can pick out

238

Martin Stokhof

the following as particularly relevant for formal semantics: the rejection of reference as the (only) ultimate semantic relation; the rejection of a viable distinction between systematic, context independent meaning (semantics proper) and situational aspects of meaning (pragmatics); the rejection of individualism; the rejection of universalism. Each of these negatives is mirrored by a positive, of course: the acknowledgment of a multitude of ways in which language can be meaningful; the acknowledgment that meaning is fundamentally situational, tied to context, purpose, practice; the acknowledgment that the self-sufficient subject is a myth and that our subjectivity is at least partly socially constituted; and finally, the acknowledgment that language is varied, changing, developing, and adapting, and that no a priori determination of its purported essence is possible, or needed. We can not discuss each of these points here, but then neither do we need to. It will be clear that they touch upon fundamental features of the framework of formal semantics, and hence one way or the other must be taken seriously (as some of them already have been). In some cases an appeal to theoretical abstraction might be defensible, but others seem to call for serious changes if we are ever to develop an approach to natural language meaning that really can deal with its variety, its complexity, and its ubiquity. For language is everywhere: in each of us individually, in our social and cultural surroundings, in our interactions with our physical environment; language is interwoven with thought, with action, with emotion. There is in a fundamental sense no escape from it, and that is the sense in which universalism is inescapable and has to be confronted. Of course, the bottom line question then is: Does a formal approach to natural language meaning still makes sense? Is it still possible? That depends. On the terms on which it started, presumably not. But as one form of a Wittgensteinian perspicuous presentation, why not?

Notes 1 I would like to thank Michiel van Lambalgen for his helpful comments. 2 This radically therapeutic interpretation started with Cora Diamond’s work and was further developed by, among others, James Conant, Thomas Ricketts, and Rupert Read. For a collection of papers that gives a representative overview of views and arguments, see Rupert Read and Alice Crary, eds., The New Wittgenstein (London: Routledge, 2000). 3 For a detailed exposition along these lines, see Martin Stokhof, World and Life as One: Ethics and Ontology in Wittgenstein’s Early Thought (Stanford, CA: Stanford University Press, 2002), chs. 1 and 2. 4 The letter is from October or November 1919, Ludwig Wittgenstein, Briefe an Ludwig von Ficker (Salzburg: Otto Müller, 1969); my translation. 5 TLP 6.42. 6 The quotes indicate that these terms have a special meaning in the Tractatus, and it is in that sense that we use them here. One of the more important properties of objects and names in the Tractarian system is that they display variation in form, another term used in a technical sense, viz., to indicate the possible configurations

The architecture of meaning

7 8 9 10

11 12

13 14 15 16 17

18 19 20 21 22 23

239

in which they may enter (TLP 2.0122, 3.21). Both names and objects may differ among themselves in the combinations with other names and objects they may engage in. Although Wittgenstein does not say much about what the various forms of objects and names might be, it seems that something akin to function application is included. TLP 6.41. TLP 6.421. For a detailed discussion of the role of simplicity in the Tractarian system, see Stokhof, World and Life as One, ch. 3. The development of semantics is described as going through two phases—that of uncovering structure and that of reinstating content—in Hans Kamp and Martin Stokhof, “Information in Natural Language,” in Philosophy of Information, ed. Johan van Benthem and Pieter Adriaans (Amsterdam: Elsevier, 2008). In the first phase the driving force is to lay bare the formal principles, to discover the necessary features, and the way that is done is classic: by abstracting away from what is material and contingent. The Tractatus is typical of that phase, as is the work that was done in formal semantics in the pioneering days as we shall see later. In formal semantics the uncovering structure phase was followed by one in which the content that was abstracted away is put back into place, but, by and large, within the framework that resulted from the first phase. In Wittgenstein’s case things did not go like that. His failure to apply the Tractatus principles to the (relatively) concrete problem of color exclusion, made Wittgenstein change his ways in a much more radical fashion. He turned the entire enterprise upside down, and we can read his later work as an ongoing attempt to let what is necessary and a priori appear on its own terms from amid the contingent. Which is an innocent enough assumption if we are engaged in linguistic description, but one that directly flies in the face of universalism. Cf., in this connection, Wittgenstein’s rejection of the axiomatic method in logic (TLP 6.127–6.1271). The idea that self-evidence might be used as a criterion for determining the set of axioms of logic was already rejected by Wittgenstein early on, cf., “‘self-evidence’ is wholly deceptive and always has been” (N 8.9.14). He insists on the surveyability of logical form: all its logical properties and relations must be surveyable from the form of an expression (TLP 6.113 ff.). This, according to Wittgenstein, is not guaranteed by the axiomatic method, but it is a property of the truth table method he outlines in the Tractatus. The requirement of surveyability must not be confused with that of decidability in the modern sense of the word, although there is a kinship between the two. TLP 6.54. TLP 4.1212. Cf. N 22.8.14. TLP 5.473. This is also the basis for Wittgenstein’s rejection of Russell’s theory of types; cf. TLP 3.331–33.334. Another point at which the ineffability of logic manifests itself is in Wittgenstein’s claim that logical constants do not refer (TLP 4.0312, 5.4). This is another insight that Wittgenstein developed early on. It is already present in Ludwig Wittgenstein, “Notes on Logic,” in Notebooks, 1914–1916, ed. G. H. von Wright and G. E. M Anscombe (Oxford: Blackwell, 1979). TLP 4.121. TLP 3.24–3.25. An illustrative example of the kind of abstraction that Wittgenstein uses is in his derivation of the notion of a symbol at TLP 3.32 ff. TLP 4.0312. TLP 6.124. TLP 6.13.

240 24 25 26 27

28 29 30 31 32 33 34 35

36

37

38 39 40 41 42 43 44 45 46 47

Martin Stokhof

Cf. Stokhof, World and Life as One, p. 241 ff. TLP 3.22. TLP 3.203. Wittgenstein draws the apt comparison with a tableau vivant: “One name stands for one thing, another for another thing, and they are combined with one another. In this way the whole group—like a tableau vivant—presents a state of affairs” (TLP 4.0311). TLP 6, 6.00. TLP 2.021. TLP 2.014. TLP 2.221, 2.222. In fact Wittgenstein goes to great lengths to explain away what looks like intensionality in language, a case in point being his extensional reconstruction of such intensional constructions as “A believes that p” in the Tractatus 5.542 ff. TLP 6.3751. TLP 5.5. TLP 4.22. And the same holds, of course, for their ontological counterparts, the objects; cf. TLP 2.0272, 2.03. Note that this presupposes that names and objects display type variety, which allows function-argument structure and other kinds of combinations. It is because they are a priori (objects as substance) that the totality of states of affairs (and elementary sentences, in the case of names) is indeed determined, i.e., not contingent. The notion of the form of objects (and names) lends this determination a quasi-algorithmic flavor. This is reminiscent of a purely combinatory categorial syntax, as was developed later by people such as Lesniewski and Adjuciewicz. Note that the lambdacalculus, though similar in some respects, is not a perfect model, since there one needs to define the properties of function-application (in terms of conversion and reduction). In modern logic and in formal semantics, too, some form of syncategorematic treatment of certain classes of expressions is quite common, which departs from strict compositionality, but not in any principled way. Often the strictly compositional formulation is more cumbersome, and hence some slack in the formulation is accepted, for pragmatic reasons. As long as a strictly compositional formulation is possible, there is obviously no harm in this. TLP 4.466. Rudolf Carnap, Der logische Aufbau der Welt (Hamburg: Meiner, 1928). Cf. Friedman Michael, “Carnap’s Aufbau Reconsidered,” Noûs 21, no. 4 (1987): 521–45. Rudolf Carnap, Meaning and Necessity: A Study in Semantics and Modal Logic, 2nd ed. (Chicago: University of Chicago Press, 1956). Cf. David G. Stern, “Wittgenstein versus Carnap on Physicalism: A Reassessment,” in Cambridge Companion to Logical Empiricism, ed. A. Richardson and T. Uebel (Cambridge: Cambridge University Press, 2007). Carnap, Meaning and Necessity, p. 9. Ibid., p. 10. TLP 2.04. Cf. Rudolf Carnap, “Empiricism, Semantics and Ontology,” Revue International de Philosophie 4 (1950): 20–40. A semantics like this is already hinted at in the Tractatus’s analysis of tautologies and contradictions as being true and false respectively for all truth possibilities of elementary propositions (TLP 4.46), given that truth possibilities are defined in terms of the existence and non-existence of states of affairs. Cf. Jack B. Copeland, “The Genesis of Possible Worlds Semantics,” Journal of Philosophical Logic 31 (2002), p. 100–01. Copeland’s paper gives an excellent overview of the very

The architecture of meaning

48 49

50

51 52 53 54 55

56

57

58

241

complicated and sometimes obscure history of the development of possible worlds semantics. For an historical survey that focuses on formal results, cf. Rob Goldblatt, “Mathematical Modal Logic: A View of Its Evolution,” in Handbook of the History of Logic, ed. Dov Gabbay and John Woods (Amsterdam: Elsevier, 2005). Copeland does mention that Meredith was influenced by the Tractatus: “The Genesis of Possible Worlds Semantics,” p. 117. Cf. Richard Montague, “Universal Grammar,” Theoria: A Journal of Studies in the Arts, Humanities and Social Sciences 36 (1970): 221. Page references are to the version in Richmond Thomason, ed., Formal Philosophy: Selected Papers of Richard Montague (New Haven, CT: Yale University Press, 1974). Not that Montague held the actual work of Chomsky in very high regard: “It appears to me that the syntactical analyses of particular fragmentary languages that have been suggested by transformational grammarians [ … ] will prove to lack semantic relevance,” Richard Montague, “Universal Grammar,” p. 223, n. 2. E.g., Alfred J. Tarski, “The Semantic Conception of Truth and the Foundations of Semantics,” Philosophy and Phenomenological Research 4 (1944): 341–75. Cf. Donald Davidson, “Truth and Meaning,” Synthese 17 (1967): 29. Page references are to the version in Donald Davidson, Inquiries into Truth and Interpretation (Oxford: Oxford University Press, 1984). Cf. Davidson, “Truth and Meaning,” p. 30. Cf. ibid., p. 29. Cresswell gives another statement of this view: “I shall introduce a series of formal languages, modelled on the languages of symbolic logic, which gradually increase in complexity until they reach a point at which they can profitably be thought of as models for natural languages.” As in Davidson’s case, the formal languages are interpreted languages. See Max Cresswell, Logics and Languages (London: Methuen, 1973), p. 1. Cf., in particular, Donald Davidson, “Theories of Meaning and Learnable Languages,” in Proceedings of the 1964 International Congress for Logic, Methodology, and Philosophy of Science, ed. Bar-Hillel Yehoshua (Amsterdam: North-Holland, 1965); Davidson, “Truth and Meaning.”; Donald Davidson, “Semantics for Natural Languages,” in Linguaggi Nella Società e Nella Tecnica, ed. Visentini Bruno (Milan: Edizioni di Communità, 1970). Although widely accepted, the idea of language as an infinite object is by no means a conceptual necessity. For an alternative view, see Jeroen Groenendijk, and Martin Stokhof, “Why Compositionality?,” in Reference and Quantification: The Partee Effect, ed. Greg Carlson and Jeff Pelletier (Stanford, CA: CSLI, 2005). In linguistics, too, other voices are being heard, e.g., those of people working on so-called “construction grammars”; cf. Michael Tomasello, Constructing a Language: A UsageBased Theory of Language Acquisition (Cambridge, MA: Harvard University Press, 2003). For some discussion of the history of the idea in a generative framework, see also Geoffrey Pullum and Barbara Scholz, “Contrasting Applications of Logic in Natural Language Syntactic Description,” in Proceedings of the Twelfth International Congress for Logic, Methodology and Philosophy of Science, ed. Petr Hájek et al. (London: King’s College Publications, 2005). Traditionally, the link between usability and compositionality is traced back to Frege, e.g., in “Der Gedanke: Eine logische Untersuchung,” Beiträge zur Philosophie des deutschen Idealismus 2 (1918): 58–77. The attribution is controversial, however, cf. Theo Janssen, “Frege, Contextuality and Compositionality,” Journal of Logic, Language and Information 10 (2001): 115–36. So there is an obvious tension between the Fregean and the Chomskyan roots of formal semantics. For a while people tried to argue that Fregean anti-psychologism could somehow be connected to the Chomskyan competence/performance

242

59 60 61 62

63

64

65 66

67 68 69 70 71 72 73 74 75 76

Martin Stokhof

distinction. But that turns out to construct competence in such a way that it is not related to human psychological or biological structure at all. That is certainly at odds with Chomsky’s insistence that linguistics is a branch of psychology (or biology). If competence is psychologically or biologically real the Chomskyan position can not be any other than that (the larger part of) semantics does not belong to linguistics (in the narrow sense). But that was hardly an attractive position for formal semanticists. Wittgenstein’s position in the Tractatus is complicated. On the one hand he clearly intends his analysis of “thoughts,” which are characterized as “logical pictures” (TLP 3), as completely general, and hence dismisses psychological considerations as irrelevant (TLP 4.1121). On the other hand, from correspondence with Russell in 1919, it is clear that he did think that actual thoughts do conform to his general principles, Ludwig Wittgenstein, Notebooks, 1914–1916, 2nd ed. (Oxford: Blackwell, 1979), appendix III. For more discussion, see Martin Stokhof, “Hand or Hammer? On Formal and Natural Languages,” Journal of Indian Philosophy (forthcoming). TLP 5.5563. TLP 3.325. Perhaps with exception of algebraic methods, which were already heavily used in the study of modal logic, cf. Goldblatt, “Mathematical Modal Logic.” But the idea that they could be fruitfully applied in natural language semantics came only much later, with the work of Keenan and Faltz on Boolean constructions. One development that spurred such attempts was the concept of “formulae as types” proposed by Martin-Löf, cf. Aarne Ranta, “Intuitionistic Categorial Grammar,” Linguistics and Philosophy 14:2 (1991): 203–39. The earlier work of Prawitz on natural deduction and constructive semantics remained largely unknown. Cf. Richard Montague, “English as a Formal Language,” in Linguaggi Nella Società e Nella Tecnica, ed. Visentini Bruno (Milan: Edizioni di Communità, 1970); Richard Montague, “The Proper Treatment of Quantification in Ordinary English,” in Approaches to Natural Language, ed. Jaakko Hintikka et al. (Dordrecht: Reidel, 1973). Cf. Max Cresswell, Logics and Languages, ch. 7. It is no accident that both Montague and Cresswell, before turning to natural language, had worked on modal logic. Cf. Donald Davidson, “Theories of Meaning and Learnable Languages.”; Donald Davidson, “Truth and Meaning”; Donald Davidson, “In Defense of Convention ‘T’,” in Truth, Syntax and Modality, ed. Leblanc Hugues (Amsterdam: NorthHolland, 1973). Cf. David Lewis, “General Semantics,” Synthese 22 (1970): 173. Page references are to the version in Semantics of Natural Language, 2nd ed. (Dordrecht: Reidel, 1972). Lewis, “General Semantics,” p. 169. Cf. David Lewis, Counterfactuals (Oxford: Blackwell, 1973). Cp. for discussion Donald Davidson, “The Method of Truth in Metaphysics,” Midwest Studies in Philosophy 2 (1977): 294–304. Richard Montague, “On the Nature of Certain Philosophical Entities,” The Monist 53 (1970): 186. Ibid., p. 148. Ibid., pp. 148–49. For a detailed description and analysis of this development, see Kamp and Stokhof, “Information in Natural Language.” David Dowty, et al., Introduction to Montague Semantics (Dordrecht: Reidel, 1981). L. T. F. Gamut, Logic, Language and Meaning, vol. 2: Intensional Logic and Logical Grammar (Chicago: University of Chicago Press, 1991).

The architecture of meaning

243

77 Henriëtte de Swart, Introduction to Natural Language Semantics (Stanford, CA: CSLI, 1998). 78 Gennaro Chierchia and Sally McConnell-Ginet, Meaning and Grammar, 2nd ed. (Cambridge, MA: MIT Press, 2000). 79 Cf. Montague, “The Proper Treatment of Quantification in Ordinary English.” 80 For a framework in which natural language expressions are interpreted directly, cf. Montague, “English as a Formal Language.” Montague defined a general, algebraic framework in “Universal Grammar.” There, both syntax and semantics are defined as algebras, with translation (between languages) and interpretation (of languages) being characterized in terms of homomorphic mappings. 81 TLP 4.025 ff. 82 For a more detailed discussion of the various assumptions that go into making up such a view, see Groenendijk and Stokhof, “Why Compositionality?” 83 We can safely disregard the difference between the “total” nature of possible worlds and the partial nature of the Tractatus’s situations. 84 TLP 4.024. 85 An interesting formulation, that actually makes a connection with the Tractatus, can be found in Thomason’s introduction to Montague’s collected papers: “One of the chief developments that made it possible to put physics on a rigorous mathematical basis was agreement on the form a system of units should take. [ … ] Many philosophers seem to have regarded the establishment of a theory of meaning as requiring an analogous foundation in terms of a system of units. Wittgenstein’s Tractatus is, among other things, an attempt to regard meanings as dependent on just two basic notions: entity and truth value. [ … ] Montague’s framework (adopted not only by him but by many other practitioners of “possible-worlds semantics”) is a generalisation of the Tractatus framework. The type of entities and the type of truth values are retained as basic semantic units, but another type is added, that of possible worlds.” “Introduction,” in Formal Philosophy: Selected Papers of Richard Montague, ed. Richmond Thomason (New Haven, CT: Yale University Press, 1974), p. 41.

86 87 88

89

90

The passage is interesting because Thomason completely ignores the intrinsically intensional nature of the Tractatus system. As we suggested above, apparently, that aspect is eclipsed by the truth-conditional characterization of meaning in the text. Also note that what Thomason ascribes to Wittgenstein here is in fact Frege’s position: the latter did acknowledge truth values as entities, the former rejects this explicitly (TLP 4.431, 4.441, 5.4). Cf. Donald Davidson, “Reality without Reference,” Dialectica 31 (1977): 247–53. TLP 3.3. Consequently, the notion of an object becomes more like that of an arbitrary object in the sense of Kit Fine. Much more needs be to be said on this issue, also in connection with the Tractatus’s treatment of identity. For more discussion, see Stokhof, World and Life as One, p. 162 ff. Cresswell’s approach might appear to be an exception, since he advocates going from basic logical language to one that “models” a concrete natural one. However, note that this process is mainly one of adding content (and some additional structure) to a basic frame that is the same for all such models. Cp. Cresswell, Logics and Languages. Actually, Richmond Thomason notices that what he calls the “generality” of Montague’s approach is likely to be misunderstood: “Philosophers and linguists are apt to object to the form taken by Montague’s semantic theories, on the ground that content has been sacrificed

244

Martin Stokhof to generality. [ … ] this feature of his work is not merely a defect of style. To a considerable extent the rarefied character of his semantic writings is a consequence of his fundamental approach,” Thomason, “Introduction,” p. 47.

91 92 93 94 95

Apparently, Thomason, too, reads a universalistic aim in Montague’s framework, and recognizes that there is a discrepancy with the particularistic nature of the applications of the framework. This gap he then attempts to bridge by making use of conceptual distinction such as that between lexical semantics and structural semantics. Cf. Davidson, “Truth and Meaning.” Cf. Donald Davidson, “On the Very Idea of a Conceptual Scheme,” Proceedings and Addresses of The American Philosophical Association 47 (1974); Davidson, “The Method of Truth in Metaphysics.” The parametrization could also involve some specific subclass of languages, but we will ignore that in what follows. Cf. Davidson, “Reality without Reference,” p. 216. Does this apply to the Tractatus as well? There the abstraction to just truth functional operations seems motivated by different concerns. But on the other hand, Wittgenstein’s only model of a language that transparently shows its logical form was that of a logical language.

10 De-transcendentalizing religion Anat Biletzki

“ … it is after all a way of living … ”1

10.1 Introduction We begin with the common and veritably well-accepted view of religion as housing the transcendent. Without attending to comparative studies of religion (where pundits may be able to point to religions that deny any transcendent being) and without a seemingly mandatory analysis of the concept of transcendence, we take the risk of claiming straightforwardly that, by and large, religion recognizes a transcendent being or reality which is characterized as such by its independence of physical, material, ordinary existence and, indeed, its surpassing of this type of existence. This is not to say, however, that religious transcendence belies existence (of God, or a transcendent reality); that would implicate one immediately in a certain atheism. It is to say, rather, that religion recognizes a different reality holding a distinct stratum of existence—the transcendent. It is also to say, finally, something metaphysical; i.e., to hold a metaphysical position.2 Somewhat related to the metaphysics of transcendence (called now religion), are questions of the epistemology of religious belief. The philosophy of religion is justly taxed by issues in this arena and subsequently analytical treatment of the concept of religious belief (and associated with it—religious faith) becomes obligatory, pursuant to the abandonment of revelation as the (only) grounding of such belief. But whether this discussion asks about evidence (for God’s existence), or internal coherence (between propositions describing God), or the status of religious attitudes (towards God)—it presupposes transcendence (and then must include, in the asking, more intricate questions about the evidence for transcendence, or the contradictions in concepts of transcendence, or—most straightforward—human obligations vis-à-vis a transcendent being). Not all discussion about religion must acknowledge such transcendence. But such talk, if it denies this exalted status of the divine or the holy, does so

246

Anat Biletzki

from the outside, so to speak, by inspecting religious practice, analyzing religious belief, and describing religious believers from the vantage point of psychology, sociology, anthropology, or other varieties of cultural reportage. Put differently, any de-transcendentalization of religion is executed by theoreticians attempting to explain the human propensity for transcendence and, in that sense, taking as their starting point a secular world-view and using it to both investigate and elucidate religion. Wittgenstein is not of that ilk; he takes religion seriously, from within, as it were. In other words, he presents a version of religious practice and belief that purports to maintain the traditional—i.e., holy, divine, sacred—significance of religious terms while relinquishing their need for transcendence. Whether such a view of religion is consistently feasible is a question for philosophers of religion. Whether it truly describes religion, i.e., religious persons, institutions, beliefs and practices, is to be debated by believers and non-believers alike. That it has provided a novel scrutiny of religion— de-trancendentalizing religion—is the subject of this chapter.

10.2 Interpreting Wittgenstein on religion Religion and Wittgenstein’s dealings with it appear to be a subject relatively less well-traveled than others in the interpretive Wittgenstein enterprise of the past 85 years, though now seemingly attracting more intensive scrutiny. This is not a statistical comment, but one garnered from the encounter with the Wittgensteinian community and reinforced by the—perhaps impressionistic—conduct of religious studies of different hues. Fifty-five years after Wittgenstein’s death is a fitting time to continue asking interpretative questions, but to also consider the implications of such asking, since the issue of Wittgenstein on Religion is addressed not only to what Wittgenstein said about religion, but also to his influence, i.e., to the use made of his thought in religious studies, theology, philosophy of religion, and, more widely perceived, sociology and anthropology as they relate to religious practice.3 What did Wittgenstein have to say about religion? Where, and when, did he say what he said (and not necessarily wrote) about religion? Now quantitatively, if we suppose that that aspect matters, and as opposed to Wittgenstein’s writings on, e.g., mathematics, which took up a massive amount (almost or over half) of his total output, it has seemed that his explicit writings on religion can be pinpointed in certain well-identified paragraphs and pages. That is to say—if writings on religion are identified as places where religion (or religious ritual, or God) is mentioned explicitly, then the total religious output may seem meager indeed. More strikingly, the larger part of these writings are either notes recorded by students or notes collected from the Nachlass into edited versions, not intended by Wittgenstein for publication.4 In chronological order we usually point to passages in the Notebooks, very few sections in the Tractatus Logico-Philosophicus, some thoughts in A Lecture on Ethics, the remarks in Remarks on Frazer’s Golden Bough, three

De-transcendentalizing religion

247

lectures in Lectures on Religious Belief, scant comments in Philosophical Investigations, and some others in On Certainty; out of chronological order there are a few remarks in the editions of later work such as Zettel or Remarks on Colour, and the several allusions to religion in Culture and Value. Yet even if added up, it all amounts—still quantitatively—to less than 100 pages. Two expansions of this estimate may be admitted. If the Wittgensteinian mystical is seen as essentially including the realm of religion, or his assuredly related talk of ethics, then the field of Wittgensteinian religion broadens, not only quantitatively, but interpretively. And if we also consider another slant, a wider reading of Wittgenstein that is not restricted to specific concepts (prayer, ritual, religion, God, the ethical, or the mystical), but rather the perspective—a religious perspective—from which to read all of Wittgenstein (work or life), then our understanding of Wittgenstein and of his philosophy as relevant to religion can, perhaps, be further enriched. There is another significant, methodological, point in any account of Wittgenstein on religion, born of the use that has been made of his thought. When embarking on the interpretation of his words on religion—and closely related to them, his thoughts on (what we would term) social science—the line between interpreting Wittgenstein on religion and using his thoughts on religion becomes a fine one. “One [question] is: what view did Wittgenstein himself take of religion? The other: what view of it can be taken in the light of his philosophy?”5 That is to say, the line between interpretation for interpretation’s sake alone—the Wittgenstein scholar’s project—and interpretation for use (of Wittgenstein’s philosophy) —the theologian’s or philosopher-ofreligion’s agenda—is elusive and perhaps unimportant. In any case, however, both types of practitioner are interpreting Wittgenstein; the line need not be insisted on, and, to the contrary, it seems that it can be intentionally flouted—since the question is not only what Wittgenstein said but rather what was his influence, i.e., the use made of his thought. Indeed, interpretation of Wittgenstein has often been both the groundwork for, and the result of, interests of use; this is especially so in the area of interpretation of religion. Which of the Wittgensteinian bio-philosophical periods is most pertinent to religion? It may be instructive to notice the differences in the interpretive balance concerning this question. Reviewing the robust interpretations and interpretive uses of Wittgenstein—early, late, middle, and very late—on religion, there seems to be a clear disproportion. More commentators seem to have dealt, at least until recently, with the later Wittgenstein, and therein mostly with religion as a language game or as a form of life, than any other interpretive gesture. Another way of phrasing this, concerning the Wittgensteinian influence on philosophy of religion rather than the interpretation per se of Wittgenstein on religion, is by flagging the label of fideism so often associated with Wittgenstein (even when criticized as such). I will be suggesting, in what follows, that this is as it should be. In other words, it is Wittgenstein’s later thought that provides an original, even revolutionary,

248

Anat Biletzki

standpoint on religion; analysis done via a later-Wittgensteinian view detranscendentalizes6 religion without emptying it of meaning. This is not to say, however, that perusing his early thought on religion is not fruitful; it does provide tantalizing facets of Wittgenstein’s thought on religion. But this has nothing to do with de-transcendentalization; on the contrary, it might be perceived precisely as transcendence. I refer here, of course, to the positing of religion in the area that Wittgenstein called the mystical, religion as beyond the limit of language, thought and world. Now, this is a heady way of reading Wittgenstein, it is recognition of the transcendence involved in any religious discourse—for what is transcendence if not a being beyond and outside the world? Another, less facile interpretative route, and one that may be somewhat related to de-transcendentalization, is the one that focuses on the construct of faith, explaining Wittgenstein’s special stance on religion as being similar to Kierkegaard’s.7 Using the Kierkegaardian connection (with sometime associations with Nietzsche, Tolstoy, and James) we reach the point at which, deserting traditional transcendence, religious faith is identified as both experiential and existential. Such faith is severed from any cognitive, propositional lay belief but need not be described as faith in a different, transcendent reality. It does, however, have to be made sense of when it is taken as instructing the human being to engage in acts as extreme as, e.g., Abraham’s intended killing of his son.8 That sense transcends, for Kierkegaard, the ethical as we know it; for Wittgenstein, believing is, accordingly, a matter of “submitting to an authority.”9 This belief and that submission may be at a distance from the transcendental and can well be expressed as the existential experience of faith. But the conscious, explicit move to de-transcendentalization has yet to be made; Wittgenstein does, indeed, make it.

10.3 Religion as a form of life and language game There are, then, novel perspectives from which to view Wittgenstein’s words and thought on religion. They do not necessarily oppose one another, in certain formulations they may be complementary, they are almost invariably unusual, and, consequently, it should not surprise us that they do not voice the traditional and standard belief in a transcendental being (ruler of the universe, omnipotent, omniscient, eternal one, almighty). Still, and perhaps due to this novelty, many diverse gazes at religion by (mostly the early) Wittgenstein, or by his interpreters, do not concentrate on the de-transcendentalization of religion as their focal point. It will be the thrust of the following discussion, which aims at what I view as the principal and most productive way of perceiving what Wittgenstein said about religion, to identify an explicit detranscendentalization of religion. This has been clearly recognized, though not so named, by a large part of the interpretive Wittgensteinian community. But after relating this mainstream analysis—of religion in Wittgenstein as being a language game and a form of life, of what has been generally and

De-transcendentalizing religion

249

pejoratively called Wittgensteinian fideism—I will attempt to ground the detranscendentalization not only in these two Wittgensteinian constructs, but rather on the more general context of his thought, which will then supply us with an even broader framework, so to speak, which cannot countenance transcendence and does not need to do so. Of the articles and books written on and specifically devoted to Wittgenstein on religion, the greatest part seem to adopt the two key terms, language game and form of life, as linchpins of their analysis. This is, to begin with, an interesting interpretive ploy, for Wittgenstein, in his sporadic mention of religion, and even in places where he deals with it more systematically (a word which can be used to describe Wittgenstein only with nervous irony, but such may perhaps be said of the Lectures on Religious Belief and even the Remarks on Frazer’s Golden Bough), does not usually call our attention to religion, or the religious way of life, or religious beliefs, or even religious practice as language games or a form of life.10 He does make one such reference when asking “Why shouldn’t one form of life culminate in an utterance of belief in a Last Judgement?”;11 and he does seductively say: “[ … ] it is after all a way of living [ … ].”12 The more formal labeling— religion as a form of life or religion as a language game—is done by the interpreters, while Wittgenstein supplies the descriptions of and insights about religion directly, and asks questions, sometimes supplying answers, which function as grounds for the designations (of language game and form of life). This is engrossing for the study of interpretation, for it is clear that such an analysis of religion under the rubric of language game or form of life takes into account an amalgam of Wittgenstein’s remarks on religion with his theoretical (again, an ironic adjective) terminology, which can then provide the basis for interpretation. It is, in fact, the strategy I will be using shortly; but I am getting ahead of myself. What goes on in the religious language game or form of life? Religion consists of religious language as an activity of religious practice, having very clear and obvious roots. “All rites are of this kind. Such actions may be called Instinct-actions.”13 These instinctive-actions, elsewhere called primitive reactions, are what give rise to religious language. It is not speculative reasoning or any sort of belief, even belief in a transcendent being, that ground religion; it needs no cognitive or theoretical grounding and no prior, independently existing entity or reality on which to hinge its practices. “Religious language originates in certain primitive reactions, not in speculative reasoning, and for this reason the practice of religion cannot rightly be thought to involve belief in the existence of any sort of queer, transcendent beings.”14 If this construal of primary, original religious language is correct, then religious belief, if that be our word of choice, is not belief in facts—not only not empirical, regular, scientific facts, but not even special kinds of other-worldly, spiritual, ethereal, facts. Belief in the Last Judgment, the immortality of the soul, the existence of God, is not belief in special kinds of facts, or in facts about the future. Rather such belief is a set of attitudes.

250

Anat Biletzki If someone who believes in God looks round and asks “Where does everything I see come from?,” “Where does all this come from?,” he is not craving for a (causal) explanation; and his question gets its point from being the expression of a certain craving. He is, namely, expressing an attitude to all explanations. (CV, p. 85)15

In fact, insisting that religious talk, which is an expression of religious belief, cohere with other kinds of talk or asking that it provide evidence for any description of reality—any reality—adds up to superstition. “The point is that if there were evidence, this would in fact destroy the whole business [ … ]. I would say, if this is religious belief, then it’s all superstition.”16 Consequently, there is, in religion, no call for propositions, but only attitudes and practices—a form of life. And such a unique form of life includes its own criteria for rationality, or coherence, or even intelligibility. “Whether a thing is a blunder or not—it is a blunder in a particular system.”17 So trying to evaluate the truth or validity of propositions and arguments expressed in the religious language game is fraught with the risks of irrelevance; these propositions and arguments can only be understood from within the practice of religion—including prayer, worship, ritual—i.e., within the language game. This is the mainstream interpretation of Wittgenstein on religion.18 Dubbed Wittgensteinian fideism (as early as 1967),19 this reading of Wittgenstein emphasizes elements of the later Wittgenstein that are obvious and well-known, though not necessarily in the context of religion. Forms of life are the given—“What has to be accepted, the given [ … ] are forms of life”—and this given constitutes and is constituted by ordinary language in its various contexts.20 As philosophers we are not (though we sometimes succumb) to investigate language or forms of life with a view to criticism; we can only describe them perspicuously. And this description will lay out the grammar of distinctive forms of life, albeit without subjecting them to any evaluation. Most strikingly, the criteria for and norms of intelligibility, reality, and rationality (i.e., what is understood in a certain form of life, what is its reality, and what part of it is rational) are internal to a form of life and therefore, obviously, cannot be determined from without. Similarly, yet differently, instead of concentrating on form of life, some interpreters choose to emphasize religion as a language game, but the essential characteristics are analogously present. A language game is behavior, but its origins are in primitive behavior, in reactions, and these are not the result of thought. Furthermore, any justification, explanation or validation can occur only within language games.21 The thrust of these characteristics puts language games on a par with forms of life in the service of discussions on religion.22 But notice that these accounts of both form of life and language games do not necessitate talk of religion; rather, they have to do with Wittgenstein’s philosophy of language and his perception of the philosopher’s task while using these pivotal terms. Yet this is what engenders Wittgensteinian fideism—

De-transcendentalizing religion

251

insisting on the uniqueness and criterial independence of forms of life and language games and, in particular, the religious form of life and the language game of religion. Consequently, and possibly most controversial, according to Wittgensteinian fideism the philosophical argument which portrays religion as incoherent or irrational is the paradigm of philosophical confusion; philosophy can and should only describe this unique form of life or language game. Before we continue, let us make note of the level of commitment involved in this—the most common—interpretation of Wittgenstein on religion and its subsequent relevance to various intellectual, philosophical, or religious projects. The disagreements that take place within and against this family of interpretations do so on different, yet nevertheless interrelated tiers of argument. One may argue with the bent of interpretation that reads Wittgenstein thus and try to show—using context, exegesis, or other analytic methods— that “this is not what Wittgenstein said,”or would say, about religion. This would count as an interpretative argument and is only one ingredient of our current project. One can, instead, accept that Wittgenstein does indeed consider religion as a language game or a form of life (by any other name), and argue with him, i.e., with his characterization of religion. This is closely aligned to a more general argument whereby one may address the fideistic stance on religion, whether or not Wittgenstein so thought; in other words, one is then arguing with the Wittgensteinian influence on thought of religion, irrespective of interpretive fidelity. We turn now to critique which is an amalgam of the last two paths; and traveling these routes means questioning, in depth and in detail, the Wittgensteinian view accepted by the general consensus of religion-as-a-language-game-or-form-of-life.

10.4 (Constructive) critique It transpires that some claims of this Wittgensteinian interpretive turn of religion, treated as almost unassailable truths, can be put to doubt (and sometimes even to ridicule); and such doubt harbors deep questions for the Wittgensteinian analysis of religion. For instance, the turn to primitive reactions, that is to say, to pre-linguistic reactions, when purporting to explain the meaning of religious terms might simply be obfuscating matters. When religious linguistic activity, such as prayer, is explained as being built upon primitive ritual, and more so, when religious belief is posited as being no more than participation in such activity—Wittgenstein says, “where that practice and these views occur together, the practice does not spring from the view, but they are both just there”23—one is hard-pressed to understand why this should be so; and why it should be a better explanation than the traditional, cognitively understood one, that bases prayer on belief (and not the other way around). Or, for another instance, the view of religion as a language game that has pushed this interpretation on to accept that religious beliefs are never errors might simply be misconceived. When we subsequently maintain that people who change their minds about religion do so not as a

252

Anat Biletzki

result of being convinced that they were wrong—“‘Perhaps now he sees he was wrong.’ What sort of remark is this?”24—this seems patently erroneous. All one need do is recall encounters with those who have, by their own admission, given up on religious belief by realizing their mistakes. These claims, and their refutations, are the groundwork for proto-empirical questions concerning what it is that makes a believing person religious, a ritualistic practice characteristically religious, a religious belief one to be differentiated from other beliefs (in being, specifically, religious), and the mutual relationships among all of these strata of religion. They are not to be shrugged off and can, in fact, be used to argue with the Wittgensteinian view of religion. Yet the elusive nature of such questions, and their dependence on variable and subjective responses, render them speculative; it is to the more substantive questions at the core of this type of Wittgensteinian religion that we must turn. These are born of the standard Wittgensteinian view of religion as a unique language game or form of life, a view which can be faulted on two specific theoretical lines. The first addresses—and discounts—the admonition that one can only understand the religious game from within, as a participant, and therefore that one cannot criticize it from an external (philosophical, logical, rational) position (which is, itself, embedded in another game, of course). The second pounces on the idea of the uniqueness of a form of life that decides its own measure of rationality, intelligibility, and logicality, investigating it with the aim of showing that such singularity is, in itself, incoherent because impossible and definitely not true to the facts of life. Together they lead to a third, no less crucial challenge, in this construal of religion: its unequivocal denial that there could be any philosophical criticism of any religious statement. Not for naught have traditionalist philosophers of religion, i.e., those not taken by Wittgenstein’s novel perspective, said that “they can make nothing of this” while others have talked of the connection made between the constructs of language game and form of life on the one hand, and Wittgenstein’s thought and description of religion on the other, as being “often absurd.”25 We are faced, it seems, with a conundrum. If we accept that religion is a form of life (or language game), and if we follow a standardized reading of what typifies forms of life (or language games) and then attempt to apply those types to the token of religion, we emerge with an awkward characterization of religion. Its practice involves non-cognitive foundations—and we think it needn’t. Its practitioners appear necessarily unreflective—and we think they aren’t. It becomes isolated from other aspects of life—and we think it usually isn’t. It enjoys immunity from criticism—and we think it shouldn’t. Wittgensteinian fideism, introducing religion as a form of life or a language game, although tempting in its fresh insights into religion, seems to have involved us in both descriptive and normative puzzles. Still, solving these puzzles does not necessarily lead to denying the connection between religion and these heuristically illuminating Wittgensteinian terms. We opt, instead, to continue with additional, more subtle, exposition, which is based on

De-transcendentalizing religion

253

more Wittgensteinian interpretation. For what criticisms of the standardfideistic reading actually bring to light (perhaps unintentionally) is the unWittgensteinian strictness of these erstwhile readings. It is as if the adoption of Wittgenstein’s constructs of language game and form of life, and their application to the practice of religion, have made religion an isolated game, replete with its insular rules, having nothing to do with real life, with life on the outside of religion, as it were. But surely this very molecular, very atomistic view of religion cannot be truly Wittgensteinian. So there is something—maybe unexpected—to be said for such critique: it brings out the paradoxical nature and ironic tension of the very terms language game and form of life by trying to pinpoint the standard Wittgensteinian reading of religion as depicting, in detail and in excruciating firmness, one specific language game or form of life. But, it turns out, this standard reading has missed out on the Wittgensteinian to-and-fro between detail and holism, between description and insight, and between the particular and the generally profound. Or, to use Wittgenstein’s analogy, religion as a language game must be thought of as a strand in the thread composed of several interconnected language games: “Something runs through the whole thread— namely the continuous overlapping of these fibres.”26 The original masthead interpretation seems to have missed the point it was making. This can be evinced in the meanderings of this most central of interpretations of Wittgenstein on religion. First there were the giants of original Wittgensteinian religious interpretation—Rush Rhees, Norman Malcolm, D. Z. Phillips, Peter Winch, and others—who instigated the standard Wittgensteinian interpretation of religion as based on practice (rather than beliefs), as (logically, rather than historically) originating in, and never really deserting, primitive religious custom, as a language game and a form of life which determine their own criteria of meaning, intelligibility, and rationality. The second stage of this interpretive game (continuing, in a sense, until current times) consisted of resistance to the standard reading, such opposition taking place, as we have said, on different levels: sometimes argument with the accuracy of this interpretation and sometimes dispute with the view of religion prescribed by it. The latter draws on the fideistic interpretation in order to make opposing claims about religion (and can then be viewed as the epitome of Wittgensteinian influence on schools of thought). But what we have been calling a critique does not have to end with resistance—it can, instead, be thought of as a third (non-chronological) stage, in which Wittgenstein’s thought on religion is appropriated for a more nuanced, yet still Wittgensteinian (fideistic or not), view. It is fascinating to note the give-and-take of these discussions; for several protagonists—on the standard (religion as language game or form of life) or its critical sides—make way, along the way, in qualifying their remarks, in recognizing complexities, and in making sense of the religious language game or form of life as a player in a general Wittgensteinian ambiance. As early as 1970 D. Z. Phillips, thought of as the main protagonist of fideism,

254

Anat Biletzki

talks of misgivings, saying, almost touchingly, “I write [ … ] as one who has talked of religious beliefs as distinctive language-games, but also as one who has come to feel misgivings in some respects about doing so.”27 These misgivings have to do, as we have seen, with the isolation of religious language games that invites skepticism as to the place of religion within the whole of life, its meaning, its importance, and its relevance. Phillips travels a treacherous road, explaining the reasons for the misgivings, the additional confusions wrought by treating the misgivings in a certain philosophical fashion, and the intricate view of the relation between religion, life, and philosophy that must be formulated if one is to remain Wittgensteinian in one’s view of religion: It can be seen that the meaning and force of religious beliefs depend in part on the relation of these beliefs with features of human existence other than religion. Without such dependence, religion would not have the importance it does have in people’s lives. It is an awareness of these important truths which in part accounts for the philosophical objections to talking of religion as a distinctive language-game. But these objections are confused.28 So Phillips is, in the end, using Wittgenstein’s own demand, that we search our philosophical talk for confusions hiding in it, to explain (away) the misgivings. Phillips is being Wittgensteinian to the hilt.29

10.5 A general Wittgensteinian ambiance I have used the expression “general Wittgensteinian ambiance” intentionally, since it is this perhaps intangible framework that will enable us to continue with the idea of religion as a form of life and language game, but nevertheless to escape from the supposed constricting contradictions that have arisen in its wake. Michael J. Coughlan, alluding to the Wittgensteinian quote above, puts it differently, yet so succinctly. If we are to speak of a religious language-game at all, we are to think of it, not as an individual strand of the thread, but as a quality of the thread as a whole—its color, or its texture, perhaps.30 Having followed the traditional fideists, their traditionalist counterparts, and the ongoing discussion that is minutely tinged in its analysis of religion as a language game and form of life, we herewith take a further, needed step in order to establish the point of de-transcendentalization. Indeed, fideism itself, as we have rendered it, is far removed from any unavoidably transcendental view of religion; Wittgensteinian fideism, even the subtle kind, is similarly lacking of transcendentalism. But we can take Coughlan’s cue and act upon it less metaphorically; that is to say, we can apply the two constructs of language game and form of life to religion, but also go on to incorporate Wittgenstein’s

De-transcendentalizing religion

255

words on religion in another and more encompassing view he holds on language and life—their social and communal color and texture. The further step to be taken consists of the insistence that language games in general, and the language game of religion in particular, are (a part of) an essentially active, social human form of life. We can conscript, for example, On Certainty for this generally acknowledged Wittgensteinian emphasis. In On Certainty we encounter hinge propositions—those propositions that we are certain of, no matter what. Some can seem like empirical propositions but their fascinating epistemological status is not our point here. Rather, it is propositions in which we believe—religious propositions—that join this family of certain propositions: “If someone believes something, we needn’t always be able to answer the question ‘why he believes it’.”31 More pertinently–“Is God bound by our knowledge? Are a lot of our statements incapable of falsehood? For that is what we want to say.”32 And it is here, while investigating the grounds of these fundamental propositions, without reneging on their groundlessness, that Wittgenstein adduces the well-known “As if giving grounds did not come to an end sometime. But the end is not an ungrounded presupposition; it is an ungrounded way of acting.”33 And the equally celebrated: “Giving grounds, however, justifying the evidence, comes to an end;—but it is not a kind of seeing on our part; it is our acting, which lies at the bottom of the language game.”34 Our point of emphasis, however, is that this acting is not an individualistic, solitary act. On the contrary, it is a way of active life that we have learned by being participants in a society, a community. Whether it be the knowledge of the age of the earth, the playing of a game of chess, or kneeling down in prayer, these are activities that we engage in having learned them through our upbringing, which could not be anything but a part of a social way of life.35 Continuing in this line of ascribing essential communality to Wittgenstein in his perception of religion, we can turn to another intriguing construct, which can be likewise mobilized for its illumination on religion—seeing-as (or aspect seeing). Famously used in psychological discussions of perception, this Wittgensteinian term, and the conversation that has accompanied it, is thought to dismantle the naïve view of our passive reception of objects that are objectively there, in favor of the observation that we can perceive objects by perceiving them as something.36 When relocated to issues of religion, seeing-as becomes useful in describing the religious person’s world-view, or beliefs, as a way of seeing the world differently, i.e., religiously. Now, granted, this reading of a Wittgensteinian term, while enhancing our understanding of the experiential features of religiosity, can point to a private, individual seeing-as.37 But it can also, more daringly, support a view of religion as necessarily social, since seeing-as is not “independent of custom, upbringing and training of a religious community in which certain religious concepts are used.”38 Indeed, these two slants do not contradict each other, for it is precisely our specific, social upbringing in the matter of religion that produces the experiences we call religious belief.

256

Anat Biletzki

So in religion there is the certainty born of communal bedrock and the socially produced seeing-as. The general Wittgensteinian ambiance—which includes the two specific constructs, language game and form of life, but also certainty and seeing-as39—is a persistent stress on language as both public, i.e., not private, and societal. It is Kai Nielsen, he who notoriously charged the Wittgensteinian view of religion with fideism, whose concise formulation now serves us better: In being socialized—in learning, as we all must if we are at all to navigate in the world, to be human—we come to have practices in which words and actions are interwoven. In this activity, in learning to play these language-games, we come to understand words by coming to know their uses in the stream of life, and with this we come to know how to use words in the course of our various practice-embedded activities.40 This does not contradict the experiential (perhaps Kierkegaardian) essence of individual faith, but it does add a vital and necessary component, without which we cannot really practice, i.e., cannot really participate in, the religious language game. Notably, in Culture and Value, where one meets Wittgenstein’s most pinpointed remarks on religion,41 this double-insight invigorating our point, of the personal experience cohering with the necessary social context, is made explicit Perhaps one could “convince someone that God exists” by means of a certain kind of upbringing, by shaping his life in such and such a way. Life can educate one to a belief in God. And experiences too are what bring this about (CV, pp. 85–86). Religious language, then, is not different from other language games—it is words and actions playing one particular game in our human world. It is a language game which uses words like “God,” “holy,” or “sacred,” but it is no less sacrilegious for that, nor is it less mundane. It is as de-transcendentalized as all other language games, precisely because it is a part of a human form of life predicated on a human community. The only remaining question is then—is it, in a sense, more important or more significant, than other forms of life? Some of Wittgenstein’s remarks—about himself, at the very least— seem to intimate that it is.

10.6 A short epilogue: can there be a religious, non-transcendental point of view? How does this Wittgensteinian characterization of religion, pinpointed as socially and publicly oriented, square with Wittgenstein’s (in)famous, oftquoted quip to Drury: “I am not a religious man but I cannot help seeing every problem from a religious point of view”? That it has affinities with the

De-transcendentalizing religion

257

other cases of religious interpretation that have been mentioned but not focused on—religion as (beyond) the limit of language, thought, or world; religion as an existential faith—is translucent; that it is, in some ways, more evasive and amorphous, and therefore more Wittgensteinian than those readings, is, itself, an evasive and amorphous estimate. I pause, for a final moment, on the expression a religious point of view, and on Wittgenstein’s delicate maneuverings to distance holding that point of view from being a religious man.42 What, however, is a—personal—religious point of view (as opposed to a view of religion, which was our not unrelated query)? And in what way can it be understood as de-transcendentalized? The most celebrated discussion of this quote and query is Norman Malcolm’s Wittgenstein: A Religious Point of View? and Peter Winch’s attendant response.43 Malcolm presents a reasonably straightforward analysis of the parallelism that Wittgenstein adheres to between philosophy and religion, and is subsequently refuted, almost point by point, by Winch on that same comparison. It is the latter rather than the former who gives a newfound interpretation of the religious point of view by gesturing towards the religious—i.e., the passionate, committed, and wonderful—fashion of Wittgenstein’s way of doing philosophy. This way of contemplating Wittgenstein unearths something that lies somewhere between the (biographical) extreme of a religious man (which most think he was not) and the other (theoretical) extreme of a theological thinker; perchance, a religious philosopher. This way of looking at Wittgenstein also provides a consistent interpretation of a religious philosopher saying things like “How can I be a logician before I am a human being?”44 or answering Russell’s question about what troubles him, logic or his sins, by the stupefying both.45 Still, where is the de-transcendental fulcrum? We have said that from the early Wittgenstein, who saw religion as beyond the limit (of thought, of language, of world) and can therefore be easily labeled “transcendental,” up until the later, whom we have identified as identifying religion as a form of life or language game, there is a guiding and developing thread of detranscendentalization. But how does the religious point of view that must color everything fit in, and what is its role in the de-transcendentalization? Hilary Putnam gives us the beginning of a reply: If the project of describing “the absolute conception of the world”, the project of describing “the things in themselves”, the project of dividing our common world into what is “really there” and what is “only a projection”, has collapsed, then that seems to be all the more reason to take our lives and our practice seriously in philosophical discussion.46 Fine and well for philosophy, but what about religion? Here we find a parallelism that goes beyond both Malcolm, with his methodological analogies between religion and philosophy, and Winch, with his likenesses of temperament between the two: “what Wittgenstein [ … ] is saying is this: that religious

258

Anat Biletzki

discourse can be understood in any depth only by understanding the form of life to which it belongs.”47 Our form of life, i.e., our actions, behaviors, beliefs, and practices, is crucial for understanding both philosophy and religion. Yet in order to ground this form of life as non-transcendental, or, at the least, as not necessarily requiring belief in anything transcendental, Putnam’s insight should be taken one further step. If the project of explaining the world, of philosophically explaining the world, by a transcendental authority has collapsed, then the only way of explaining religion is by describing it as we describe all our other language games and forms of life. With these astute, insightful, and perceptive Wittgensteinian constructs we then acquire the understanding that forms of life, and more specifically, religious forms of life, are, if linguistic—and they are—then always and only socially and publicly constituted.

Notes 1 CV, p. 64. 2 We will forgo, in this article, discussion of immanence (and its related pantheism), which may be posited as the main contender against transcendence in religious thought. Although usually described as diametrically opposed to transcendence, immanence, placing God within rather than outside any manifestation of human and material existence, still preserves the exclusive and special status of God’s existence and reality. Our use of “de-transcendentalization” is at a distance from all such conceptualization. 3 In just the past few years we’ve witnessed several such “accountings” of Wittgenstein on Religion. See, e.g., Robert L. Arrington and Mark Addis, eds., Wittgenstein and Philosophy of Religion (London: Routledge, 2001); John W. Cook, The Undiscovered Wittgenstein: The Twentieth Century’s Most Misunderstood Philosopher (Amherst, MA: Humanity Books, 2005); D. Z. Phillips, “Wittgenstein, Wittgensteinianism, and Magic: A Philosophical Tragedy?,” Religious Studies 39 (2003): 185–201. 4 In John Hyman’s words: “Wittgenstein’s influence in the philosophy of religion is due to scattered remarks, marginalia and students’ notes. He never intended to publish any material on the subject, and never wrote about it systematically.” John Hyman, “The Gospel According to Wittgenstein,” in Wittgenstein and Philosophy of Religion, ed. Arrington and Addis, p. 10. 5 W. D. Hudson, “The Light Wittgenstein Sheds on Religion,” Midwest Studies in Philosophy 6 (1981): 275–92. Hudson goes on to explain that “it is [ … ] more important to answer the latter than the former of the two questions, for a thinker may shed more light upon a subject by his general insights than his expressed opinions on that particular subject reflect” (p. 275). 6 Our object here is “transcendent” rather than “transcendental” (in the Kantian sense). The title and further discussion use “de-transcendentalize” and “detranscendentalization” as derivative of “transcendent” since the accompanying verb and noun—de-transcendentize and de-transcendentization—are unhappy in English. 7 Some relevant interpretative work is to be found in R. H. Bell and R. E. Hustwit, eds., Essays on Kierkegaard and Wittgenstein (Wooster, OH: College of Wooster, 1978); Stanley Cavell, “Existentialism and Analytic Philosophy,” Daedalus (1964):

De-transcendentalizing religion

8

9 10 11 12 13 14 15

16 17 18

19 20 21 22

259

946–74; R. B. Goodman, “How a Thing Is Said and Heard: Wittgenstein and Kierkegaard,” History of Philosophy Quarterly 3 (1986): 335–53; R. E. Hustwit, “Understanding a Suggestion of Professor Cavell’s: Kierkegaard’s Religious State as a Wittgensteinian ‘Form of Life’,” Philosophy Research Archives 4 (1978): 329– 47; J. Lippit and D. Hutto, “Making Sense of Nonsense: Kierkegaard and Wittgenstein,” Proceedings of the Aristotelian Society 98 (1998): 263–86; Elizabeth Wolgast, “A Religious Point of View,” Philosophical Investigations 27 (2004): 129–47. For a more current reading of this Kierkegaardian–Wittgensteinian association, marking the connection with the story of Abraham via trust, authority, and certainty, see Galia Pat-Shamir, “To Live a Riddle: The Case of Binding Isaac,” Philosophy and Literature 27 (2003): 269–83. CV, p. 45. Brian Clack, “Wittgenstein and Magic,” in Wittgenstein and Philosophy of Religion, ed. Arrington and Addis, p. 87. LC, p. 58. CV, p. 64. RFGB, p. 72. John W. Cook, “Wittgenstein and Religious Belief,” Philosophy 63 (1988), p. 442. The position on religion which sees belief as an attitude, an expression of emotion, or something to that effect, labeled “expressivism” or “emotivism,” has been attributed to Wittgenstein by some interpreters. I do not deal with it here separately since it can be viewed as a sub-set of the “religion as a language game” group. (But see Phillips who diametrically opposed this reading: “Wittgenstein, Wittgensteinianism, and Magic: A Philosophical Tragedy?”) It is attractive in providing several natural readings of Wittgenstein’s comments on religion, and fits in well with his criticism of Frazer. For a succinct survey and analysis of this branch, see Clack, “Wittgenstein and Magic.” For general criticism of this whole group, denying the essentialism involved in their views and bringing Wittgenstein back to the personal, experiential core of religion, see Duncan Richter, “Missing the Entire Point: Wittgenstein and Religion,” Religious Studies 37 (2001): 161–75. LC pp. 56, 59. LC, p. 59. Main representatives of this group of interpreters are Rush Rhees, D. Z. Phillips, Peter Winch, Norman Malcolm, Peter Geach, but there are so many others. The bibliographical list is daunting—but some seminal writings include: Peter Geach, “Nominalism,” Sophia 3 (1964): 3–14; Norman Malcolm, “Is it a Religious Belief that ‘God Exists’?,” in Faith and the Philosophers, ed. John Hick (London: Macmillan, 1964); D. Z. Phillips, “Religious Beliefs and Language Games,” Ratio 12 (1970): 24–46; Rush Rhees, “Wittgenstein on Language and Ritual,” in Essays on Wittgenstein in Honour of G.H. von Wright, ed. Jaakko Hintikka and G. H. von Wright, Acta Philosophica Fennica (Amsterdam: North-Holland, 1976); Peter Winch, “Understanding a Primitive Society,” American Philosophical Quarterly (1964): 307–24. Kai Nielsen, “Wittgensteinian Fideism,” Philosophy 61 (1967): 191–209. RPP I, §630. An explicit rendition of language-games in this context is found in Cook, “Wittgenstein and Religious Belief.” I refrain, here, from voicing a preference for form of life as a more suitable construct than language games in the context of religion. Suffice to point out that the literature is rich with deliberations concerning the precise meaning that Wittgenstein gives to “language game” (primitive or not, rule-governed or not, part of a language or all of it, etc.) and that its appropriateness for a rendition of religion is more controversial.

260

Anat Biletzki

23 RFGB, p. 119. 24 LC, p. 71. 25 Kai Nielsen, “Wittgensteinian Fideism,” p. 191. It behooves us not to forget that these can be perceived either as being involved in an argument with Wittgenstein, or as arguing with the conclusions that Wittgensteinian fideism compels. For us, however, the argument is about what can be said of religion as a form of life— and not whether this is exactly what Wittgenstein said. Be that as it may, it is a conversation about Wittgenstein’s sway on religious thought. 26 PI §67. 27 D. Z. Phillips, “Religious Beliefs and Language Games,” p. 24. 28 Ibid., p. 40. My emphasis. 29 More recent labels (such as “expressivism”), misgivings and charges of misunderstandings can also be found in Clack, “Wittgenstein and Magic”; D. Z. Phillips, “Wittgenstein, Wittgensteinianism, and Magic: A Philosophical Tragedy?”; Brian Clack, “Response to Phillips,” Religious Studies 39 (2003): 203–09. A good assessment of the discussion is R. Amesbury, “Has Wittgenstein Been Misunderstood by Wittgensteinian Philosophers of Religion?,” Philosophical Investigations 26 (2003): 44–72. 30 M. J. Coughlan, “Wittgenstein, Language and Religious Belief,” in God in Language, ed. R. P. Scharlemann and G. E. M. Ogutu (New York: Paragon House, 1987). 31 OC §550. 32 OC §436. 33 OC §110. 34 OC §204. 35 Interpreters looking to On Certainty for religious illumination include D. M. High, “On Thinking More Crazily than Philosophers: Wittgenstein, Knowledge and Religious Beliefs,” International Journal for Philosophy of Religion 19 (1986): 161–75; D. Martin, “‘On Certainty’ and Religious Belief,” Religious Studies 20 (1984): 593–613; D. Martin, “God and Objects: Beginning with Existence,” International Journal for Philosophy of Religion 41 (1997): 1–11; W. D. Hudson, “The Light Wittgenstein Sheds on Religion.” Most pertinent to our point is Iakovos Vasiliou, although I am not certain that he would go as far in the unequivocal conscription of community, Iakovos Vasiliou, “Wittgenstein, Religious Belief, and on Certainty,” in Wittgenstein and Philosophy of Religion, ed. Arrington and Addis. 36 This is often illustrated by the famous duck-rabbit drawing, which is neither a drawing of a duck nor one of a rabbit, but rather the same drawing which can be seen-as a (drawing of a) duck or seen-as a (drawing of a) rabbit. (PI II, xi, pp. 193–229/165–94). 37 This emphasis on the experience of faith is usually associated with the Kierkegaardian reading of Wittgenstein, which, as mentioned above, though not necessarily transcendental is not automatically de-transcendental either. 38 N. K. Verbin, “Religious Belief and Aspect Seeing,” Religious Studies 36 (2001): 18. But see J. Kellenberger for an opposing view, that insists on “seeing-as” as a vehicle of religious, personal discovery rather than community: “‘Seeing-As’ in Religion: Discovery and Community,” Religious Studies 38 (2002): 101–08. 39 And, of course, we’ll always have the private language argument. 40 Kai Nielsen, “Wittgenstein and Wittgensteinians on Religion,” in Wittgenstein and Philosophy of Religion, ed. Arrington and Addis, p. 140. 41 I make a note of the fact that G. H. von Wright, editor of this collection of notes, remarks that they “do not belong directly with [Wittgenstein’s] philosophical works.” The question of criteria for notes being philosophical—or not—is beyond our interests here; needless to say, we view these as relevant to Wittgenstein’s thought on (philosophy of) religion.

De-transcendentalizing religion

261

42 Indeed, I see sustenance in that quote for the claim that the question of Wittgenstein’s religious beliefs or proclivities has nothing to do with Wittgenstein’s philosophical views in general and philosophical views on religion in particular. 43 Norman Malcolm, Wittgenstein: A Religious Point of View? (London: Routledge, 1993), with a response by Peter Winch. 44 Quoted by Rush Rhees in Ludwig Wittgenstein: Personal Recollections (Oxford: Blackwell, 1981), p. 211. 45 Bertrand Russell, Autobiography (London: Routledge, 1991), p. 330. 46 Hilary Putnam, “Wittgenstein on Religious Belief,” in Renewing Philosophy (Cambridge, MA: Harvard University Press, 1992), p. 135. 47 Ibid., p. 154.

11 Morality without agency D.K. Levy

While serving in World War I, Wittgenstein received the news that his brother Paul, a talented pianist and Austrian officer, had been wounded, and his arm amputated. Wittgenstein wrote in his notebook in October 1914: Again and again I cannot but think of poor Paul who so suddenly lost his vocation! How terrible. Which philosophy would be needed to overcome this. If this can be done except by suicide. [ … ] Thy will be done.1 Which philosophy indeed? What, if anything, might philosophy offer to someone who has lost his vocation? Philosophy for Wittgenstein was therapeutic, even in the Tractatus Logico-Philosophicus.2 However it was therapy for confusion and tangles created by language, not for the loss of an arm or one’s vocation. In the most dire form these confusions might take, the confusions appear solely as questions seeming to demand impossible answers.3 How could philosophy—conceived superficially as riddles—overcome a challenge that urges suicide? I shall argue first that Wittgenstein’s philosophy—specifically in his ideas concerning ethics—can help resolve a challenge such as he imagined his brother Paul faced. My argument faces an immediate difficulty. Though Wittgenstein was famously deeply concerned with ethical or moral matters, he also maintained that the will was powerless to effect change in the world, because will and world were wholly independent. But if the will is powerless, then what is left for the expression of someone’s ethical engagement with the world? It is as if, by this claim, Wittgenstein erased ethics, contrary to what we know of his interests and contrary to what I am arguing. Therefore I shall argue second that we can honor Wittgenstein’s serious interest in ethics by distilling and elaborating from his conception of the will as powerless an account of morality that is not dependent on the power to act, i.e., on agency, but which still gives a place for engaging one’s moral challenges. This account is, I suggest, a plausible reconstruction of Wittgenstein’s view of ethics as it is sparingly set out in his writings. The account I attribute to Wittgenstein is of additional philosophical interest for its revisionary implications for contemporary moral theories. An

Morality without agency

263

emphasis on instrumental action as the exemplar of moral response can distort moral philosophy. Moral philosophy, I shall claim, does not wait on theories of action or on what they presuppose, namely agency. To support this claim, I propose—following the reconstruction of Wittgenstein’s account of morality without agency—to suppose that we lack the power to act. Then we can see how little changes in the possibilities for moral understanding and response. If an intelligible account of morality can be recovered without action or agency, then there is a prima facie reason to doubt that action or agency is central to morality. I shall proceed in four sections. First, I shall set out Wittgenstein’s view of ethics based on his three central themes. Second, I shall reconstruct his view in a way that makes it more plausible and reduces its dependence on peculiarities of his philosophy. Third, I shall use Wittgenstein’s reconstructed ethical views to engage contemporary views. Finally, I will return to the challenge that Wittgenstein’s brother faced.

11.1 In this section, I set out three themes that I think are central to Wittgenstein’s conception of ethics. These are the valuelessness of facts, the unworldliness of self, and the powerlessness of will. These themes are presented briefly in the Tractatus, where they are shadows of much deeper discussions in his Notebooks 1914–1916. I work from the more familiar Tractatus back to the Notebooks.4 11.1.1 TLP: valuelessness of facts I cannot reprise the whole logical apparatus of the Tractatus which in part forces Wittgenstein to say what he does about ethics. I will focus instead on the three themes above as consolidations of Wittgenstein’s views on ethics, value, and the will. The first theme—the valuelessness of facts—is stated plainly: 6.4 All propositions are of equal value.” This is unpacked as follows: The sense of the world must lie outside the world. In the world everything is as it is and happens as it does happen. In it there is no value—and if there were, it would be of no value. If there is a value which is of value, it must lie outside all happening and being-so. For all happening and being-so is accidental. What makes it non-accidental cannot lie in the world, for otherwise this would again be accidental. It must lie outside the world. (TLP 6.41) The idea is that what occurs in the world is necessarily contingent. Every fact can obtain or not. If, per impossibile, a putative fact were necessarily true

264

D.K. Levy

then it would not be a fact.5 Wittgenstein’s hidden premise is that value is not contingent. If something is valuable, then it must be so without regard to contingencies. Value does not occur through chance. That is a premise. If it is granted then the inference follows directly. Facts are contingent; value is not; therefore, value cannot be a fact. Wittgenstein makes clear that the sense of value he intends is ethical when he continues: “Hence also there can be no ethical propositions. Propositions cannot express anything higher” (TLP 6.42). If we take ethics to concern value, and the content of a putative ethical proposition to be one that describes value, then there is an obstacle to the existence of ethical propositions. This is the compressed point of TLP 6.42. According to the Tractatus, a proposition’s content is a picture of a fact. According to the reasoning above, value cannot be a fact. Therefore, there can be no propositions describing value, i.e., there can be no ethical propositions. The further consequence of this chain of reasoning is that there can be no talk of ethics, nor ethical judgments expressible in language. Note that nothing in this implies that there is no such thing as value, only that it does not inhere in the factual. Language only concerns the factual. So language cannot express value or ethics. 11.1.2 TLP: unworldliness of the self The second theme of relevance to Wittgenstein’s view of ethics concerns what I term the unworldliness of the self. The theme expresses Wittgenstein’s view that the subject who thinks and wills is, in some sense, not in the world. Indeed, as regards the factual—i.e., the world describable by language—the subject does not exist. Wittgenstein says: “The thinking, presenting subject; there is no such thing” (TLP 5.631). The consideration he gives for his claim depends on distinguishing the phenomenal subject from the noumenal subject. Naturally, my body is in the world, and no doubt the movements of my body occur as phenomena in the world. But as a thinking (i.e., noumenal) subject, I am a perspective that is not identical with my body. Wittgenstein elaborates the non-identity as follows: If I wrote a book “The world as I found it”, I should also have therein to report on my body and say which members obey my will and which do not, etc. This then would be a method of isolating the subject or rather of showing that in an important sense there is no subject: that is to say, of it alone in this book mention could not be made. (TLP 5.631) So, while I could describe the things I see, including my body, I could not describe the perspective in which the seeing occurs. A subject’s perspective is an orientation on something essentially perspective-less, viz., facts. If that is

Morality without agency

265

right, then perspective, here constituting the noumenal subject, is not a fact and therefore not existent in the world. In this sense, the self does not exist in the world. While the world is the totality of facts, the contents of a subject’s experiences—i.e., his perspective—is a definite ordering of solely a few such facts. This thought may be developed to suggest that it is within one’s perspective that the world is “my world.” There is therefore really a sense in which in philosophy we can talk of a non-psychological I. The I occurs in philosophy through the fact that the “world is my world”. The philosophical I is not the man, not the human body or the human soul of which psychology treats, but the metaphysical subject, the limit—not a part of the world. (TLP 5.641) The subject does exist, but is unworldly. It is the limit of what can be described by language. The analogy Wittgenstein used was with an eye and what it sees. The eye cannot see itself seeing and yet it is the eye’s attributes (e.g., its location and orientation, depth of field, etc.) that determine and limit what is seen.6 The idea is familiar from Kant. The expression of thought presupposes a thing whose thought is expressed. Yet nothing in the expression of a particular thought describes, in a linguistic sense, the thing thinking that thought. Suppose I say, “I am hungry.” There can be bodily facts concerning hunger that make my statement true. But my body does not have the thought that it is hungry. My body has the chemical processes that make it a fact that I am hungry. However, I am not my body, for my body can change without changing my thoughts, e.g., when I get a hair cut or tonsillectomy. Therefore, seemingly true self-referring thoughts do not describe the subject thinking those thoughts. The distinction between phenomenal and noumenal subjects mirrors a distinction between phenomenal and noumenal wills. This latter distinction is important for ethics: “Of the will as the bearer of the ethical we cannot speak. And the will as a phenomenon is only of interest to psychology” (TLP 6.423). The phenomenal will, which can effect change in the world, is a matter for psychology and is nothing to do with ethics. The noumenal will, by contrast, is the bearer of the ethical, by which Wittgenstein means the bearer of ethical predicates, e.g., callous, vain, courageous, pious, good, or evil. Since the noumenal will, with which ethics is concerned, is unworldly like the noumenal subject, it too cannot be described. This underscores the barrier to ethical propositions, because judgments that predicate ethical properties of the noumenal will—e.g., Smith is cruel—do not describe anything in the world, i.e., anything factual.

266

D.K. Levy

11.1.3 TLP: powerlessness of will We can now move to the third theme, which is the one most central to developing the idea of morality without agency. It is the powerlessness of the will. The crux of the theme is set out directly. A necessity for one thing to happen because another has happened does not exist. There is only logical necessity. (TLP 6.37) The world is independent of my will. (TLP 6.373) Even if everything we wished were to happen, this would only be, so to speak, a favour of fate, for there is no logical connexion between will and world, which would guarantee this, and the assumed physical connexion itself we could not again will. (TLP 6.374) To be clear, there are no causal necessities on Wittgenstein’s view. Even if the will were in the world—contra Wittgenstein’s view of the noumenal will— and causally efficient, the relation between willing and the realization of the will’s objective cannot be necessary, but solely contingent. And, the noumenal will is not, in any case, in the world. This does not show the powerlessness of the will, for the will may be causally effective, albeit contingently so. However, when combined with the preceding themes, the will proves irrelevant for ethics: 1 Assuming that value is not contingent; 2 given that the relation between willing and the realization of its objective is contingent; 3 then the relation between willing and the realization of its objective cannot be of value. The same conclusion can be reached by noting that the effects of a causally efficient will are solely states of affairs, themselves combinations of facts. But the considerations above established that facts have no value. So the effects of a causally efficient will have no value. The conclusion, to be clear, is that, ethically speaking, the will is powerless. The will as bearer of ethical is not in the world. Even if it were, it could not in willing remain as bearer of the ethical. Its ethical status must originate elsewhere. Wittgenstein puts the point this way: If good or bad willing changes the world, it can only change the limits of the world, not the facts; not the things that can be expressed in language.” (TLP 6.43).

Morality without agency

267

11.1.4 From TLP back to notebooks Taking interim stock, Wittgenstein’s account seems to arrive at the following limited place for ethics. There is ethics and its purview is value. However ethics cannot be expressed because value is not in the world. The bearer of the ethical and therefore the locus of value is the willing subject who, while having worldly phenomenal counterparts, is also not in the world. The activity of the will, when directed at a worldly object, is not the activity of the will in its capacity as the bearer of the ethical. So, with regard to ethics, the will is powerless in any worldly sense we might associate with actions or the effects of those actions. What scope is left then for the activity of the will by which it is the bearer of the ethical? The Tractatus contains very little about ethics, though Wittgenstein once claimed it was composed with an ethical purpose.7 The core logical theory of the Tractatus was complete by 1916, composed in notebooks Wittgenstein used while on campaign in World War I. Wittgenstein added entries concerning ethics to these notebooks through most of the second half of 1916. The same three themes outlined above in the Tractatus are also present in these notebooks, but in more detail. The discussion of ethics in the notebooks is vexing and inconclusive, and perhaps as a consequence little was transferred to the Tractatus. However, review of the central themes as presented in the notebooks is helpful in elaborating and refining what Wittgenstein thought was left for the ethical alongside his logical theory. 11.1.5 NB: valuelessness of facts Addressing the valuelessness of facts, Wittgenstein emphasizes that the world itself is not a locus of value. Everything in the world, as constituents of facts, is on the same level. “A stone, the body of an animal, the body of a man being, my body, all stand on the same level.”8 For Wittgenstein, worldly events are transitions among states of affairs and their factual constituents. As facts, they are on one and the same level. Wittgenstein’s conclusion is the denial of facts as valuable or morally charged, “The world in itself is neither good nor evil.”9 Wittgenstein thought that ethics could not be in the world, but must be of the world—like logic. Ethics is a condition of the world, meaning that ethics is absolute in that it is independent of how the world happens to be at any moment. As a condition of the world it requires only “that the world is.” The difficulty remains that facts—and the world which they constitute—are neither good nor evil nor any ethical form in between. Yet, there is good and evil. So where is it? Ethics is conditional on a subject, e.g., a person, because the subject is the one capable of ethical understanding, whose engagement with the world can create higher, ethical meaning. Good and evil is born in and borne by the subject. The subject’s role in the world’s ethical form is total for Wittgenstein,

268

D.K. Levy

“Good and evil only enter through the subject. [ … ] good and evil [ … ] are not properties of the world.”10 The subject originates good and evil. Any significance beyond mere description of facts—higher meaning—depends on the subject. He imbues the world with the higher meanings including good and evil. The dependence is clear on Wittgenstein’s view, but how does the subject do it? 11.1.6 NB: unworldliness of the self The subject proved mysterious above in its unworldly character. Wittgenstein tried for greater clarification when writing his notebooks. Definitely, “The I is not an object.”11 More, he refines a common sense thought—viz. the subject’s body is in the world—when giving a fuller version of the Tractatus passage: The philosophical I is not the human being, not the human body or the human soul with the psychological properties, but the metaphysical subject, the boundary (not a part) of the world. The human body, however, my body in particular, is a part of the world among others, among beasts, plants, stones, etc., etc. (N 2.9.16) Wittgenstein finds himself caught between two conflicting positions. On one, “It is true that the perceiving [vorstellende] subject is not in the world, that there is no perceiving subject.”12 On the other, ethics depends on the subject. He re-iterates, “Were there no will, neither would there be that centre of the world, which we call the I, and which is the bearer of the ethical”13 and “[ … ] the subject is not a part of the world but a presupposition of its existence [ … ] good and evil [ … ] are predicates of the subject [ … ] not properties of the world.”14 The tension in Wittgenstein’s position is acute because ethics requires a subject who is engaged with the world but who is not in the world. “Good and evil only enter through the subject. And the subject is not part of the world, but a boundary of the world.”15 Wittgenstein augments this, “The perceiving subject is surely empty illusion. But the willing subject exists.”16 Later summarizing, he writes, “The subject is the willing subject.”17 11.1.7 NB: powerlessness of the will Everything ethical seems to depend on the willing subject. This is awkward since the will is also powerless because, Wittgenstein notes, “The world is independent of my will.”18 World and will are disjunct, “The world is given me, i.e. my will approaches the world completely from outside as something ready-made.”19 The subject can neither make nor control the world, but his perspective seems dependent on its happenings, for “we have the feeling of being dependent

Morality without agency

269

on an alien will.”20 The idea is familiar from and analogous to the mythology of fate. “God in this sense would be simply fate or, what is the same thing, the world independent of our will.”21 This is an important point, because fate—curiously—requires a willing engagement that is also idle. Someone subject to fate seems to act in making his future and in some sense does so. But the events consequent on his actions are pre-determined. One’s willing seems paradoxically both redundant and necessary. Yet the paradoxical situation is neither personally excluding nor exculpating: Oedipus takes no solace in what he has done even when he knows that he was fated to do so by oracular pronouncement. Wittgenstein has arrived at an increasingly difficult position since he is dependent on the willing subject’s engagement with the world to account for good and evil; the willing subject seems furthermore personally obliged willingly to engage the world; yet the willing subject has no control over the worldly events wrought by his will. Wittgenstein admits on, for him, incontestable grounds that what one wills does not determine, even partially, happenings in the world, “For it is a fact of logic that wanting does not stand in any logical connexion with its own fulfilment.”22 Logically, wanting states of affairs in no way necessitates their obtaining. Plainly, what one wants may not happen. Even when one’s body is physically implicated in the realization of a state of affairs, the physical relations on which this occurrence depended were not themselves the objects of the willing. One does not will the laws of nature any more than one always gets what one wants. Indeed, since one’s body is in and part of the world, and given the world’s independence from will, even one’s body may not be subject to a sovereign will: [ … ] I can imagine carrying out the act of will for raising my arm, but that my arm does not move. [ … ] the act of will does not relate to a body at all, and so that in the ordinary sense of the word there is no such thing as the act of the will. (N 20.10.16) Wittgenstein’s position, already difficult, is made seemingly intolerable by foreclosing the body as the—last and most natural—point of contact for the will to act in the world. Wittgenstein is forced to a tentative conclusion in which he accepts the comprehensive powerlessness of the will, viz., “The will is an attitude of the subject to the world.”23

11.2 The three themes I have canvassed above are, I suggest, a cumulative expression of Wittgenstein’s seeming view of ethics. The position arrived at is vexing, because the will is powerless to do good or evil. Yet, there is ethics and its concomitants—there is good and evil—and people confront these in their lives.

270

D.K. Levy

11.2.1 Two Wittgensteinian clues There are, I suggest, at least two clues in Wittgenstein’s writings that intimate how to understand ethical challenges and responses within the vexing constraints set out above. The first is latent in Wittgenstein’s remark that the will is an attitude of the subject to the world. The second lies in his remarks on punishment: TLP 6.422 and N 30.7.16. Concerning his tentative conclusion above, “The will is an attitude of the subject to the world,”24 Wittgenstein notes that, “The act of will is not an experience.”25 The act of will, i.e., willing, therefore does not happen to the willing subject. Rather, it must originate in or from the subject. As originator, a subject’s engagement with a state of affairs, which is expressed in his willing, is perforce personal and perspectival. The engagement is personal because, as originator, it is solely a function of his will. It does not happen to him, else it would not be willing. The engagement is perspectival because it is within the subject’s perspective that the import of matters appears. For taken as things, all things are on the same level, of equal import, “As a thing among things, each thing is equally insignificant.”26 Wittgenstein continues, “Things acquire ‘significance’ only through their relation to my will.”27 The personal and the perspectival are united in willing because the subject’s perspective is itself the product of willing. The idea latent in these remarks is that while the will does not alter the facts in a state of affairs, the will is responsible for the perspective in which that state of affairs has the import it does. By “import” I mean what Wittgenstein called “higher meaning” or value. Ethical significance, we can say, is in the perspective expressed by the subject’s attitude to a state of affairs. That may be fine as far as it goes, but how does this show itself in what we could recognize as ethical matters? Wittgenstein’s remarks concerning punishment—the second clue—hold an answer to this question: The first thought in setting up an ethical law of the form “thou shalt … ” is: And what if I do not do it? But it is clear that ethics has nothing to do with punishment and reward in the ordinary sense. This question as to the consequences of an action must therefore be irrelevant. At least those consequences will not be events. For there must be something right in that formulation of the question. There must be some sort of ethical reward and ethical punishment, but this must lie in the action itself. (And this is clear also that the reward must be something acceptable, and the punishment something unacceptable.) (TLP 6.422) In this remark, Wittgenstein distinguishes punishment and reward in ordinary and extraordinary senses. The ordinary sense is connected with actions and their consequences. This sense cannot be the ethical sense, both because

Morality without agency

271

the will is powerless and the relations between actions and consequences contingent. There is also an extraordinary sense of punishment and reward. It is the ethical one. This, I suggest, is an example of an attitude to the facts that is an ethical response. For one’s attitude to what occurs—to the states of affairs that unfold—can be one in which what occurs is a punishment or a reward. Just this—my attitude—can give these occurrences a higher significance. That they are a punishment is not a further fact.28 This is evident by observing that the very same facts could be taken as a reward or as neither, without my thereby thinking that the state of affairs had changed. Moreover, it is not that the same facts will in the case of punishment or reward be ones for which one is causally responsible and in another not, because for Wittgenstein, ex hypothesi, one is not causally responsible for any. Furthermore, one may be ignorant of the causal aetiology of the state of affairs, in which case one’s responsibility could not enter into one’s attitude. In one perspective, with one attitude, these are just more facts. But in and with another, what occurs is punishment. Indeed, one can think the same of what occurs to someone else. And just this is already, so-to-speak, to moralize the facts, to give them an ethical or higher meaning. This provides an answer (and an example) to the question of how the attitude of a willing subject could show itself as a recognizable ethical response. 11.2.2 Ethical attitudes To repeat, the point is that to think in these terms, punishment and reward, is to express an ethical attitude. It is an ethical response. To be sure, it requires the will. But it is not an action, nor is it an alteration of states of affairs. For Wittgenstein at least, it is not a judgment either, if by that is meant something that could find expression in language, because (as noted above) what is expressed would be solely factual not ethical. It is morality without agency. Why might one think that some events are a punishment? Ordinarily, a punishment is for what one has done. However, one can imagine punishment for what has been thought or desired too. Most vices do not require action. Consider cardinal vices such as lust, cowardice, foolishness, and venality. One can lust without acting on it, if only through weakness or fear of getting caught. One can be cowardly through the unwillingness to confront one’s faults in self-examination. An interest in the foolish or banal does not require acting foolishly or for banal ends. Foolishness can show itself through attention to trivia. Venality can be evidenced through contemplating a corrupt act such as bearing false witness, even if one does not after all do so. Similarly for the virtues. Temperance or courage need not require action. Temperance may be precisely wanting solely what one needs, even when provided with more. Indeed, patience is precisely attenuating desire.

272

D.K. Levy

One can take what happens as a punishment for lust, even when there is no causal link between one’s lust and what occurs. For example, I am persistently unsatisfied—and unhappy with it—because of my lust, e.g., that it is unfulfilled. No doubt this may show itself in action. But allowing, as we should with or without Wittgenstein, that there is no necessary link between will and its object, my unhappiness is independent of whether events actually eventuate from my lust. Neither is my unhappiness caused by my lust. My lust might cause nothing, but I will nevertheless be unhappy. My unhappiness is internal to my lust. Indeed, the lust is rightly termed internal to the unhappiness I experience, precisely because it is not dependent on actual events. Similarly, achieving patience, taken as a virtue, does not alter the facts, for if it did—e.g., by avoiding what must be endured—it would not be patience. But internal to patience is contentment, for discontented patience is no virtue. A common objection to this conception of punishment is that punishment requires a punisher who punishes. Certainly this is a familiar idea, especially when the punisher is God.29 However, it hardly seems analytic of punishment. What seems essential to punishment is that one suffers a penalty that is also just. Justice is essential, not a competent authority that inflicts it. For example, a man who tries to swindle but is in the process swindled himself may understand the loss he suffers as just, viz., as a punishment, even if the counter-swindlers are not competent authorities. Similarly for the man who suffers for his hubris, such as a politician disgraced but not convicted. Summarizing the position, the world acquires higher meaning by the attitudes one takes to some states of affairs. Attitudes are activities of the will. The states of affairs with higher meaning are one’s own—personal and perspectival—because one understands oneself as implicated in them. One is not implicated in the sense of having caused or eventuated the states of affairs. Rather, it is a willing by the subject that is the object of the attitude in which a state of affairs is understood with a higher meaning. For instance, within a subject’s attitude to a state of affairs, the subject understands the state of affairs as his punishment. More, the subject understands the state of affairs as his punishment for some willing of his, e.g., his vanity or lust. This position makes sense of Wittgenstein’s claim in TLP 6.422 that punishment or reward in an ethical sense must be internal to the action. Consider as an example the attitude expressed and the self-judgment made autobiographically by Sven Lindqvist: “I am being punished for my ignorant moralism. I thought those who could no longer love each other were just lazy. But work alone is not enough. Without grace, there is no love.”30 What Lindqvist laments is his approach to restoring the love in his relationship, viz., it was graceless. He is being punished for thinking that such things were inessential. He suffers for it. But he is not grieving for his lost relationship here—though he could—but how he was lacking and what it has meant for his life. Nothing in understanding this as a moral response requires thinking that it is the facts, the state of affairs as they are now that are his object. It is

Morality without agency

273

enough that his attitude to matters is one of punishment for his gracelessness and his ignorant disdain for the place of grace in love. In so doing, Lindqvist gives some further, higher meaning to the state of affairs in which he now lives—i.e., alone—namely, that he is being punished. The precise meaning will emerge in the life it makes possible, impossible or difficult for Lindqvist. It is in that sense personal to the subject, in the way that a subject’s life flows from his ethical bearing, viz., by what he takes as important, values, and so on. For it would hardly be a revelatory claim if one’s discontents flowed from falling short of one’s own values, and if contentment came from the opposite. This, I suggest, is part of what is meant when Wittgenstein says that punishment is internal to the action, not its consequences. What I am pointing to is a connection between moral responses and how life goes. I return to this below. 11.2.3 A worry A worry may present itself now, if it is thought on this view that punishment lies within the subject’s gift. That seems antithetical to the idea of punishment, for one is not free to choose for what one is punished. This worry arises from not distinguishing attitude and the willing that is its object. However, if the two are distinguished, the worry will abate. One has an attitude to events such that one sees them as punishments or rewards and much else beside, e.g., righteous vengeance, wishes come true, tragedy, or harbingers. These are states of affairs that one can understand with ethical attitudes. The states of affairs are, however, unchanged by those attitudes. More, those attitudes are not mandated by or internal to those states of affairs that, as a matter of contingencies, obtain. A state of affairs does not mandate, e.g., that I understand it as my punishment. Distinct from one’s attitude, are those qualities that can be internal to one’s willing. One’s dissatisfaction is, e.g., internal to one’s lust. I term it “internal” because it is not dependent on what actually comes of the lust. For nothing may come of it, for all the reasons that willing can fail to realize its objects. Therefore, like the attitude described above, willing is also not dependent on the states of affairs that, contingently, obtain. As above, e.g., the dissatisfaction internal to my lust is not mandated by any state of affairs, whether consequent or not on my lust. Unlike attitude, however, lusting and dissatisfaction—both of which inhere in the willing subject—are internally related. Similarly, e.g., for decency, it must be internal to the action. For if it were not, then I could act decently by accident—but that is absurd. The worry has given rise to a tension. On the one hand, it looks as if there is no internal relation between the ethical attitude with which states of affairs are understood and the states of affairs. On the other hand, the character of one’s willing and the willing seem internally related, i.e., the character is internal to the willing.

274

D.K. Levy

Dissolving this tension and the worry requires further subtlety in clarifying the distinction between attitude and willing. It seems a conceptual point that punishment must be punishment for something. So it is a constraint on the intelligibility of understanding a state of affairs as punishment that it is a punishment for something. Understanding a state of affairs as punishment requires that one also is aware of for what one is being punished. By contrast, the character internal to willing is not dependent on the willing subject being aware that the character is internal to the willing. For plainly one can become anxious, unhappy or dissatisfied without knowing its source, e.g., that dissatisfaction is internal to one’s lust. If the contrast is sound, then the tension and worry are dissolved for the same reason. Dissatisfaction, say, is, for some, internal to lusting. Over this, they have no control or choice—yet neither is it mandated by a state of affairs. Whether someone comes to see one state of affairs or another as a punishment for his lusting is within his gift insofar as any of one’s attitudes are. Putting it another way, it is within one’s gift which states of affairs are understood as the punishments (or the expression) for one’s lusting (or willing). One’s punishments are one’s own, self-inflicted so to speak. The worry above is misplaced because one does not choose for what one is punished, even if the state of affairs that is the punishment is self-recognized.

11.3 In this section I will illustrate how Wittgenstein’s view of ethics, as elaborated and extended above, provides a disruptive corrective to prevailing positions in contemporary moral philosophy. Naturally, to achieve this it would be better to lessen any dependence on Wittgenstein’s eccentric view of logic or reality in the Tractatus. This can be done by restating the claims that are central in Wittgenstein’s view of moral response above, viz. morality without agency. These stand on their own without reference to the larger Wittgensteinian picture which was their motivation. There is a weak claim: moral response is possible without action (or omissions). The example of taking some events as punishments or rewards is sufficient for this claim. There is a stronger claim too: the moral character of willing is internal to the willing, not to the actions that may happen to eventuate. That was something illustrated with examples involving virtues and vices. If these claims are good, then each is a prima facie consideration against focusing on action, especially taken instrumentally, as the exemplar of moral response. This consideration is applied below. 11.3.1 What to do? It is a commonplace in moral philosophy that it is practical. This is sometimes expressed by saying that morality concerns the question, “What should

Morality without agency

275

I do?” This is expressed with broader application as, “How should I live?” Naturally, this can be made impersonal by substituting “one” for “I.” Further attention to the practical emphasis expressed in the sense of “do” above, reveals that the concern with morality is not self-evident. Often people are held accountable for what they did not do or did not consider doing, e.g., bribery. I do not mean in the sense of an omission versus a commission, though that is true. Rather, someone may be morally responsible for ignorance, false belief, insensitivity, or inattention. Second, by the answers it will admit, the question proposes that the exemplar of moral response is the formulation of a state of affairs at which one aims, i.e., what one will do in response to the question. In contrast, if the will is powerless yet there is still moral response, then there are moral responses that are not approximations of the exemplar. If the will is not instrumental in realizing states of affairs because causally inert, then answers concerning what to do can come to nothing more than wishing.31 The complaint is more pointed if there are responses whose aims cannot be formulated in terms of states of affairs. How can we formulate—in a practical way—the state of affairs in which grace is part of love or not? Indeed, patience is a response to the question, “What should I do?” except that its aim is no change in the state of affairs prompting the question. Indeed patience specifically precludes doing anything. In short, the question of what to do pre-conceives the role of a subject’s will as instrumental, as responding with a “doing.” The weak claim, following the illustrations above, denies that an instrumental role exhausts when the will can be the bearer of the ethical. The stronger claim denies that the will ever has an instrumental role when it is the bearer of the ethical. A natural objection is that the sense of the original question regarding morality as what to do can be restored by asking instead, as the exemplar of moral response, “What should I want?” Suppose that is right. That would be a significant revision of the commonplace. For the revised question is not at all equivalent to, “What should I do?” Wants need not eventuate in action, nor even be likely to do so. Indeed, it may be inconceivable that they should do so or impossible (contingently) to formulate practically a state of affairs that is the satisfaction of what is wanted.32 11.3.2 Accidental or essential to action The nature of the objection to the practical and instrumental way of conceiving morality is evident if one were to consider what to do if one wished to be decent. Suppose one answers the question, “How should I live?” with the desire to be decent. It is not at all obvious to which state of affairs— particular or kind—this ought to incline one. An immediate obstacle is that one’s wanting to be decent is not equivalent to wanting decency, wherever it may come. Insofar as one wants decency as an answer to the question of how to live then one wants the decency in oneself. Even here, this is not

276

D.K. Levy

equivalent to wanting many states of affairs in which one does decent things, for it is likely better if the world were such that fewer acts of decency were required. This example recalls the distinction above between what is internal and external to an action, what essential and what accidental. The state of affairs consequent on an action is external. Seeking to be decent is a state of the will, a standing disposition to decency. When that state of the will eventuates in action, if ever, then the action will be a decent one regardless of how it turns out. Decency is internal and essential to that state of the will.33 Very likely we will see how someone’s life has the shape it does, by the dispositions he comes to have. Being decent may not often eventuate in episodes of decency, but it will shape the possibilities for the episodes in someone’s life that do eventuate. In this way, his life becomes his own, because of what is possible for him—motivationally speaking. I suggest this as a contemporary gloss on the idea that the subject appears in philosophy by the way in which the world becomes “my world,” by the possibilities for my life in the world. Clarity about morality depends, I suggest, on answers to questions about the sense in which a life is personal and about the difficulties internal to life, particularly sustaining what is personal in it. 11.3.3 Theory of action in ethics No doubt when we talk about what is personal in life we mean the various ways in which one is responsible for the course of one’s life, even where one does not control what happens in it. Responsibility is central also to justification, culpability and apology. However, the idea of responsibility has been developed as if it were a problem in the metaphysics of voluntary action— as if clarity about what makes an action free were also clarity about the character of the action.34 The viability of the morality without agency view above shows that voluntarism in the metaphysics of action is not to the point. Insofar as the will is the bearer of the ethical it hardly matters (whether due to counterfactual matters of fact) that the will does not eventuate in action and could not have eventuated in action. That Jones did not, e.g., consider saving a drowning man—even if Black would have prevented him from doing so—is sufficient for Jones to understand his subsequent disquiet and subsequent events as punishment for his indifference. That, I have argued, is all that is required for his response to be intelligible as a moral response. Clarity about the will instrumentally conceived as acting to achieve aims does not in this way lead to clarity about morality. 11.3.4 Consequentializing morality A recent idea in ethical theory is that all moral theories can be conceived as consequentialist.35 The strategy is variously operationalized but structurally

Morality without agency

277

it involves subsuming non-consequentialist elements of a moral theory into a consequentialist value assessment procedure. The strategy is deemed successful when a consequentialist counterpart of a non-consequentialist theory is constructed that is co-extensive with regard to deontic predicates, i.e., permissibility, obligation, etc. A consequentialist theory is a counterpart of a non-consequentialist one just when each dis-/approves or obliges the same actions and omissions. On this view, moral theories have their identity conditions, more or less, because of the extension of the deontic predicates they employ. The morality without agency position elaborated from Wittgenstein above is a counter-example, I suggest, because it is not conceived in terms of deontic predicates. Nor is it obvious that the elements offered as fundamental to the possibility of intelligible moral response are re-describable by reference to deontic predicates. If ought implies can, and there is no can, there may be no ought, and hence no applicable deontic predicate. More is required to prove such a counter-example, but it is indicative. 11.3.5 Not action-guiding This brings to the fore the peculiar character of the moral philosophical position set out above, viz., it is not action-guiding, as such. This feature is probably in most stark contrast with dominant contemporary theories.36 Indeed, because the position is in some ways anti-voluntarist, the notion of action-guidance when choosing is redundant. It is this feature that is essential to the pressure it puts on the instrumental conception of the will in moral response. The view is evidently not consequentialist since the moral relation between will and consequence is an external one. But neither is it Kantian insofar as it is not one that focuses on deontic notions. The current moral philosophical notion of right is not at work here. The right cannot be defined in terms of the good, because the good does not inhere in states of affairs. The good cannot be defined in terms of the right, because on this view being good is prior to being right, i.e., deontic terms get little hold. Punishment comes of vice, not of doing what is not right—except where one adds, trivially, all vice is not right. The place of the virtues is however different here from that in recent virtue ethics. Recall the example of patience in which virtue is internal to willing, but not thereby action-guiding. The foregoing briefly shows the disruptive points of contact with prevailing positions in contemporary moral philosophy. It also shows that the moral philosophical position in Wittgenstein’s writing is significantly revisionary.

11.4 The test of the enduring influence of Wittgenstein’s work or his ideas concerning ethics is not the impact or implications for philosophical theory. If there is such a test, then it is the difference it could make in living, particularly

278

D.K. Levy

understanding one’s life. This explains the enduring significance of early works of moral philosophy, e.g., Plato’s Apology. I therefore return to the topic with which I began and which recurred at several points above. That topic concerns the place moral thinking (philosophy) can have in someone’s life. The point I have pressed, following Wittgenstein, is that the place is not— need not be—practical or instrumental. While it is another commonplace that morality is about how to live, the relation between that commonplace and the commonplace that morality is about what to do is not easily illuminated. For the answer to “How should I live?” is not a simple compound of answers to “What should I do?” Someone’s life is not constituted solely by his doings. Things also happen to him. His life unfolds by how he responds to the things that happen to him. The responses may not be the practical exercise of a capacity to act, because responses might solely be attitudes, e.g., hope, patience, courage, cowardice. The question how to live—if this is seriously posed—is not answered with a life-plan setting one’s goals. Such an answer avoids the question’s purport by transmuting the challenges of life into seeking instruments for achieving one’s goals. This falsely presumes that a good life is within one’s power to eventuate. But just as there is no logical relation assuring that willing will realize its objects, neither is there any assurance that one’s goals in life will come to pass. In life, one is frequently powerless. Answering the question how to live demands different kinds of understanding. One is the understanding of which are the best means to a good life. The other is the understanding of what can be made of a life, of life’s possibilities for being, e.g., fine or ignoble. The contrast can be put coarsely, but fairly, as between understanding the ends of highest value and the sense or purpose of life. Which is needed for an answer to the question, “How should I live?” Returning to Wittgenstein’s brother, perhaps we can express the challenge he faces, and Wittgenstein laments, as, “After what has happened, how should I live now?”37 Wittgenstein’s ethical philosophy can answer this question. For what Paul Wittgenstein needs is a new understanding of his purpose in life, of what is possible for him. This makes no presumption concerning whether Paul can achieve this purpose. Rather, the possibilities for his life depend on what about his situation he can take as reward and what as punishment—or any of the other modes of moral assessment, virtuous or vicious. A statement of what he might already have known—viz., the ends of the highest value—is no answer. For nothing about the change in Paul’s situation will have changed which are the highest ends in life. If he knew them before, he knows them now. If they were of highest value before, they remain so. A re-statement of these ends or knowledge of how to achieve them is, therefore, moot for meeting his new situation. Paul Wittgenstein faces a challenge arising from the loss of his vocation, part of his purpose. Of course, he need not be the author of that loss. He

Morality without agency

279

may also see the change as a punishment, for vanity perhaps. If he does, then that too will constrain his response to how to live after the loss of his arm. What is important about this example—because of what it illuminates about moral philosophy—is that worrying about the purpose of life is the worry that makes understanding difficult. Worrying which ends to seek as means to a good life is not difficult any more than problems in higher mathematics are difficult, practically speaking. Finding the route to a proof is difficult, just as it is difficult to find the means to an end. But worries about purpose and means are not difficult in the same sense. The sense of difficulty in the first is practical in the sense that any answer conditions what may come of one’s life. The difficulty in the second, is practical solely instrumentally. One point of contrast is that resolving the second is amenable to luck, not so the first. Vocations are not even the hardest of the practical challenges one can confront. If I am a bad teacher or a bad philosopher, I could find another job or try harder. But if I find that no matter how hard I will it, I fail to be decent, I can’t just get another job or another life.38 I have to live with it and that can be punishing. That is a true practical challenge. Successful resolution will itself be its own reward, just as continuing failure is itself the punishment.39 Wittgenstein’s ethical view was, roughly, that to live a good life is to seek harmony with what occurs, though not, as it were, to be the cause of those occurrences. In arguing that morality without agency is a view that is illuminating and revisionary, I am not claiming that morality is not practical without qualification. What I have tried to show is that the challenges for which moral philosophy may provide understanding are not practical because they demand instrumental thinking or action. Rather, I have argued that if they are practical, the practicality is caught up with living—seeking life’s meaning and purpose—and with resolving the difficulties internal to each life. Seeing this clearly and seeing it through means understanding that agency is of the least consequence. For this understanding, Wittgenstein’s personal writings remain instructive.

Notes 1 2 3 4

GT 28.10.14. TLP 3.324, 4.112. TLP 6.5 ff. These themes are discussed at greater length, in chronological order, to a different purpose in my “Wittgenstein’s Early Writings on Ethics,” in E. V. Di Lascio et al., eds., Lecture on Ethics: Introduction, Interpretation and Complete Text (Macerata, Italy: Quodlibet, 2007). 5 A necessarily true proposition—i.e., a tautology—does not have an articulated logical form and so cannot picture a specific fact with the same form; see TLP 4.4463. 6 TLP 5.633. 7 Letter to Ludwig von Ficker, reprinted in G. H. von Wright, Wittgenstein (Oxford: Basil Blackwell, 1982), p. 83

280 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

30 31 32 33 34 35 36

37 38 39

D.K. Levy

N 12.10.16. N 2.8.16. N 2.8.16. N 7.8.16. N 20.10.16. N 5.8.16; my emphases. N 2.8.16. N 2.8.16. N 5.8.16. N 4.11.16. N 5.7.16. N 8.7.16. N 8.7.16. N 8.7.16. N 29.7.16. N 4.11.16. N 4.11.16. N 9.11.16. N 8.10.16. N 15.10.16. Wittgenstein makes the same point in his Lecture on Ethics “there will simply be facts facts and facts but no Ethics,” MS 139a, p. 9. Wittgenstein himself sometimes understood matters in terms of God’s judgements. For instance, he likens feeling guilty to the idea that God is unhappy with our conduct, albeit solely as a simile, not as an ethical attitude. See his Lecture on Ethics, MS 139a, pp. 15–16. Sven Lindqvist, Desert Divers, trans. Joan Tate (London: Granta Books, 2000), p. 112. This position might yet be compatible with a broadly Kantian account of the good will as willing solely right ends, despite the apparent practicality of Kant’s concern with practical reason. More suggestively, but equally problematic, the identity of wanting or willing may have no dependence on the state of affairs at which either aims, because willing need have no internal relation to any state of affairs. Of course the willing state may embody a flawed conception of decency, in which case actions may fail to be decent. But in this case, the failure will still be such independent of how things happen to turn out. The mini-industry in the metaphysics of responsibility has a recent origin in the “Frankfurt counter-examples” in Harry Frankfurt, “Alternate Possibilities and Moral Responsibility,” Journal of Philosophy 66 (1969): 829–39. An early discussion of this strategy is in James Dreier, “Structure of Normative Theories,” The Monist 76 (1993): 22–40. Hursthouse defends virtue ethics as a genuine ethical theory precisely by attempting to show that it is action-guiding—the hallmark of authenticity possessed by consequentialism and deontology; Rosalind Hursthouse, On Virtue Ethics (Oxford: Oxford University Press, 1999), ch. 1. Although, given the Wittgenstein brothers’ propensity to suicide, the question might have been, “How can I live now?” These examples are mentioned in Rush Rhees, “What Are Moral Statements Like?,” in Without Answers, ed. D. Z. Phillips (London: Routledge & Kegan Paul, 1969). Cp. TLP 6.422.

Afterword

12 Wittgenstein passages Denis Paul

When one studies Wittgenstein’s post-Tractatus work, twin enthusiasms develop. The first is to wonder at the impressive progress of his ideas, from February 1929 to April 1951. The second follows later, as one gets deeper into his Nachlass—the extraordinary way in which he edited and re-edited his expressions of those ideas. He only gave this passion up at the very end, when he no longer had any time for it. Even Remarks on Colour is much edited. The second volume of Last Writings on the Philosophy of Psychology known as “Das Inner und das Äußere” is too scattered in its notebooks for one to know what he might have hoped to do with it. Only On Certainty, and in particular from §300 to the end, flows so well as to convince one that it is as it was meant to be. The most extreme example of successive editing is what was printed as Part I of Philosophische Grammatik, edited for Blackwell by Rush Rhees.1 Its publication date is 1969, but I received unbound page proofs from Blackwell in January 1970 and it was only late in that year that it actually saw print. In 1976 Anthony Kenny published a criticism of Rhees’s editing.2 I cannot claim to have read it but I must at least have heard about it on the grapevine, and it reinforced suspicions of my own that began early in 1976 when a sabbatical six months enabled me to devour the Cornell microfilms, on which, in spite of their imperfections, nearly all the relevant manuscripts and typescripts had been recorded. My further researches during that half-year made me even more suspicious, for I had also obtained (from Rhees, in Elizabeth Anscombe’s absence) permission to read originals in the Wren Library of Trinity College, Cambridge, starting with Band X. This, rather puzzlingly, consisted of two parts, the first being the end of a long series of dated first or second drafts, its last date 4.6.32 on page 41 of that volume, the second the opening of an undated revision. In between, TSS 212 and 213 (the Big Typescript, derived from 212) had been dictated, and the revision in question was of the first 404 pages of 213 minus a segment that Wittgenstein apparently did not think needed revising so much, in print as the opening two sections of Rhees’s Part II of Philosophische Grammatik, called “Logischer Schluss” and “Allgemeinheit.” All the remainder of 213 similarly escaped minute revision.

284

Denis Paul

In other words, the Big Typescript falls into two parts, a much revised part consisting of its first 404 pages, minus the sections I have mentioned and a few smaller ones, and the little-revised remainder. Rhees ought to have made clear that his Part I was the upshot of repeated Wittgenstein revisions, and given a clear account of the form those revisions had taken, not calling his volume anything but Philosophische Grammatik simply. Then he should have printed the totality of the Big Typescript in a separate volume, entitled as such and not confusing us by referring to Philosophische Grammatik at all. (The Anhang to his Part I was cobbled together from the smaller sections mentioned above and from various other sources. As it is, in the middle of Rhees’s volume, it is a serious distraction to the reader.) In any case, from his so-called Part II Rhees had dropped three major Big Typescript sections, called “Philosophie,” “Phänomenologie” and “Idealismus etc.,” and they are extremely interesting and important, but they are now easily accessible, not only in the Bergen Electronic Edition but in two editions of the complete Big Typescript: Michael Nedo’s in the form of Volume 11 of the Wiener Ausgabe and a recent Blackwell edition, edited and translated on opposing pages by Luckhardt and Aue. What primarily concerns me here, however, is what I mean by calling Philosophische Grammatik’s Part I the upshot of repeated revisions, because misunderstandings about this go back to the Kenny article in Acta Philosophica Fennica.3 What I had heard about this seemed to me entirely reasonable in 1976, until, at the Wren Library, I moved from the second part of Band X to the first part of Band XI. There I next found a further revision in the form of a large loose-leaf manuscript called Grosses Format, MS 140, and finally a manuscript written on both sides of about 100 half-sized pages called Kleines Format. This had, and still has, no assigned G. H. von Wright number, a fact that no one has ever explained to me. It was having all four manuscripts at one large Wren table that enabled me to find the missing clue. This was that in all of them there were markings between paragraphs directing the reader, or Wittgenstein himself in making a dictation, from one manuscript to another. I also had my printed volume with me and was able to compare it with these directions and found that they tallied almost exactly. Unfortunately I did not publish this discovery and was content to mark details in the margins of my printed copy. I still have the most important of these, one directing from page 38 of Grosses Format to page 180 of Kleines Format (taking one to the opening of §41) and another, after three paragraphs of §42, taking one back to page 38 of Grosses Format. Anscombe arrived at the Wren Library as I was examining these manuscripts and was much reassured to be shown that Kenny’s criticisms were unfounded. That seemed to me to be all I needed to do with my discovery. I did, however, report it to Isaiah Berlin, who was as pleased as Anscombe to be told that Kenny was wrong. Some years later I was told by the Wren that Kleines Format was missing, in spite of my having returned it in the normal manner. I had done so under

Wittgenstein passages

285

the Wren’s original coding, but now they used only von Wright numbering and so had no way of tracing it. The only alternative to its being somewhere in the Wren but inaccessible is that Anscombe might have been so intrigued as to take it home with her, which as a trustee she had every right to do, in which case one can only hope that it will turn up among her papers when they are catalogued. In 1994, when he was launching the first volume of his Wiener Ausgabe edition, Nedo told me that a typescript vindicating Rhees’s editing had been found in Moritz Schlick’s Nachlass. With the help of Alois Pichler of the Wittgenstein Archives at the University of Bergen and Brigitte Parakenings of the Philosophical Archive at the University of Konstanz I have been able to obtain a copy. It is catalogued at Konstanz as D.5, but it turns out to be only a copy of the original, held by the Vienna Circle Archive at Haarlem in the Netherlands and catalogued there as Mulder V. It stops short where §42 ends, just after the Kleines Format insertion, followed by three short paragraphs from elsewhere. Nevertheless, it is a vindication as far as it goes, for it coincides with the printed version with very few inclusions and omissions. It is perhaps significant that the next manuscript written on the same large paper as Grosses Format is the first sketch for the Brown Book, written in German (MS 141). That, and its subsequent English version, forms an opening sketch for Philosophical Investigations. So as well as corroborating Rhees’s Philosophische Grammatik editing, it is also possible that the Schlick Mulder V dictation betrays Wittgenstein’s loss of interest in that particular version. The fact still remains, however, that it was the last version before Wittgenstein turned his attention to writing the Philosophical Investigations. Now I had always assumed that Kenny’s criticisms of Rhees were a consequence of his not knowing the editorial signs that indicate this version (a virtual version, as Nedo calls it), but his article in Wittgenstein: The Philosopher and His Works,4 volume 17 of the Working Papers of the University of Bergen, puts the matter in a different light. First, he reveals that he had discussed Rhees’s editing with him, as a result of being commissioned to translate the book, in which case it is inconceivable that Rhees would not have shown him these editorial marks, or at least some of them. But then, second, he uses a phrase that reveals his attitude: It is not sure that Wittgenstein never intended the passages chosen by Rhees to go together, but the text Rhees published, on the basis of a certain stage of Wittgenstein’s revision, is only one of many possible orderings that could claim Wittgenstein’s authority.5 On the contrary, it is quite sure that Wittgenstein seriously intended them to go together, and did so up to a very significant moment, namely when he lost interest in dictating Mulder V and turned towards the ideas of Philosophical Investigations. That he could call this moment a mere certain stage shows that Kenny’s long-standing discontents sprang not from ignorance of the

286

Denis Paul

editorial markings but from a conviction that they were not enormously important. Having tied my flag to the “markings” or “virtual” version, and backed Rhees’s judgment in publishing it in 1970, I must admit that there was one further version of the Philosophische Grammatik ideas, and that it was an extremely important and interesting one that should have been published as soon after 1970 as possible. It is to be found in the first 315 pages of Band XII, MS 116. In their Blackwell edition of the Big Typescript Luckhardt and Aue call it Wittgenstein’s last attempt to revise the Big Typescript. In my book I call it his last rewriting of Philosophische Grammatik.6 That is no difference worth arguing about. Luckhardt and Aue describe how, in his 1937 autumn and winter visit to Norway, Wittgenstein mentions having his typescript with him, and says that there was still much use to be made of it.7 To my shame, when I read this passage (in my last long stint at the Wren in 1980), I did not recognize it as referring to the Big Typescript. Yet I had read the opening of Band XII during an earlier visit and immediately recognized the ideas in it. They read very freshly, and certainly not as the last stage of a protracted revision. Indeed they seem to shake themselves clear of the Big Typescript quite quickly. Not only could Rhees well have published Band XII while he was still able to (including the few final just-post-war pages) but the current trustees could do so now. The full (facsimile) electronic edition, which includes it, is superb for specialists, and its cheaper non-facsimile version ought to have attracted serious non-specialists by now, but there remains a more general market of people who want to find out why Wittgenstein is so attractive to philosophers. Band XII would make just such a volume, and it is a pity no room can be found for it in the Wiener Ausgabe. In it Wittgenstein experiments with a new convention of paragraph spacing and numbering, which I hope would be represented in print better than it has been in the electronic transcriptions. It takes in ideas that belong to the Investigation, for example on page 27 it has, written above “Werkzeugkasten,” “eigentlich meine ich toolshed.”8 It shares with Band XV (written in Bergen) a certain sanity and down-toearthness in its treatment of privacy. He assumes (on page 284) that the term “the Children of Israel” originally referred to a tribe rather than to the descendants of one particular man, and offers as a comparison the possibility of calling the expansion of p the children of p. It is wide-ranging and relaxed. There are many passages that non-philosophers might find interesting without having any idea of what they have to do with philosophy. On the first page of his article in Wittgenstein: The Philosopher and His Works Kenny reinforces my long-held conviction that Part II of Philosophical Investigations was a separate work and should have been published as such, even if bound in one volume for convenience—after all we were agog in 1953 for whatever Wittgenstein had written.9 In error, he refers to it as MS 144 when in fact the trustees used a missing pair of typescripts (top and carbon) whose former existence is well documented. They were dictated from MS

Wittgenstein passages

287

144 and were mainly close to it except in one important respect: the trustees (Anscombe told me while I was correcting her translation of both parts) had on their own responsibility decided to put Part II, xiv at the end because it impressed them so much. On the second page of his article Kenny tells us that von Wright had been ill and taken no part in editing either part, or in the decision to combine them—and still less, it follows, in the misplacement of Part II, xiv which must have been done by Anscombe and Rhees on their own.10 He also refers to the microfilming of the Cornell edition, begun in 1967, under the supervision of von Wright and Norman Malcolm. Because so many of the reels are marked “Swansea” I had assumed that Rhees had been responsible, and indeed he must have played a significant part. The spelling “Color” ought to have alerted me to Malcolm’s part. The very last of the Cornell films is separate from the others, made in 1978 of an Investigation typescript given by me in that year to Trinity College, having been given to me by Yorick Smythies in 1957. (In between comes a rather tortuous story to which I have confessed in my book.) Malcolm wrote me a very appreciative letter of thanks from his new home in Hampstead, and I always regret not visiting him there. I did not realize how much he could have helped me in my attempt to recount Wittgenstein’s philosophical progress from his phenomenological language episode to Philosophical Investigations and beyond. The last paragraph of Kenny’s opening section prompts another quibble, this time with von Wright. He is quoted as announcing, in 1969, “the forthcoming publication of the Big Typescript and of On Certainty” and as feeling able to say “With the publication of these posthumous works the full body of Wittgenstein’s philosophy has been made accessible to the public.” With no Remarks on Colour, no second volume of the Last Writings on the Philosophy of Psychology, and no background to the so-called Part II of Philosophical Investigations? Remarks on Colour is important as a link between the phenomenological language quest and the perplexing ideas of Part IIxi. The second volume of the Last Writings on the Philosophy of Psychology is important as a down-to-earth re-appraisal of Philosophical Investigations’ private language argument and, without its manuscript background, Part II is a weirdly poetical and over-condensed concoction. Nor did von Wright mean the Cornell microfilms as giving accessibility to the public, even though anybody who could find a university library that subscribed to them could read nearly everything I have mentioned (except, alas, for Kleines Format). And I should add: anybody who could put up with their nannyish censoring. In his second section, as well as remarks I have already criticized, Kenny mentions a publication that he praises very properly, the Prototractatus, published by Routledge. This firm had published the Tractatus and to Wittgenstein’s extreme displeasure had refused to give him any royalties when they reprinted it. As they held the legal rights, von Wright could do nothing to find another publisher, while Routledge were only too anxious to make up for their past meanness, and the result was a magnificent volume

288

Denis Paul

that not only set a standard for future Nachlass publications but, as Kenny observes, showed how expensive and difficult they were liable to be. Kenny then moves on to the tangled story of the Tübingen attempts and Michael Nedo’s part in them. I got to know him at the Wren Library in 1980 when he was already making beautifully accurate microfilms that are still there (and at Bergen), as well as indulging in software experiments that Kenny thinks were a waste of time. They may well have been, and there is no doubt that huge lengths of time were lost before Nedo’s efforts began to appear in print in the form of the Wiener Ausgabe, whose first volume appeared in 1994. The first thing I noticed about it was the excellence of its typography and typesetting. Nedo had chosen Baskerville, of which I had unhappy memories from my own printer, and he explained the technical problems that had led to these unsatisfactory results, and the efforts he had made to find a printer who was able to overcome them. Apparently it was in 1992 that Nedo was at last able to offer the trustees the first six Bände ready for printing, to form the first three volumes of the Wiener Ausgabe, which they simply could not refuse. In 1993 they gave Springer Verlag permission to print these, going up to the first ten (strictly, nine-and-a-half, as I have explained) Bände and the typescripts from TS 208 to TS 213. The last of those forms the 11th volume, with volumes 12 and 13 intended to deal with Wittgenstein’s manuscript changes to 213. It is a pleasure to me to find from Kenny that Anscombe played a strong part in getting all that through, in view of the critical things I have said about her in my book. These Springer volumes are very handsome and usable, and shortly before her death Anscombe wrote to Kenny claiming that their publication vindicated her efforts on Nedo’s behalf and the other trustees’ acquiescence in them. He disagrees, calling them coffee-table volumes, not worth the time and money they have cost, but the time and money are gone and are irrecoverable, and what matters is our (scholars’) value in buying them. They are volumes for reading in the lap, not resting on a coffee table, and for supplementing what one can read on one’s computer screen. Now, of course, their volumes 11, 12, and 13 will have the Blackwell Big Typescript for competition, and I think there is room for both, but it is volumes 1–5 that I primarily want to gun for, and these bring me back to my opening theme, the double significance we find in Wittgenstein’s total 1929–51 corpus. These first nine-and-a-half Bände are only partially second drafts. There may have been some small first-draft notebooks from 1929–30, destroyed by Wittgenstein, and certainly one from 1930–31 is missing. Others have survived from then, and in my book I specify passages where Bände VI and VII alternate between first and second drafts. From Band VII to the end of the nine-and-a-half there are second drafts taken not from notebooks but from earlier Bände. Bearing all that in mind, a reader of the first five Wiener Ausgabe volumes is in as close touch with the evolution of Wittgenstein’s early post-Tractatus ideas as one can hope to be. There is certainly no need for a “multi-volume hard-copy edition” to reduplicate them (and in using

Wittgenstein passages

289

that phrase what Kenny is gunning for is an improved electronic edition, with which I am certainly in agreement). My strong recommendation is that any serious Wittgenstein scholar should read them minutely before encountering the dictations and revisions that follow them. Had Luckhardt and Aue done so they would not have committed an error (on their page VIe) where they confuse the very opening of Band I, written on 2.2.29, with a second use of it to opposite effect written in Band VII on 16.7.31. The latter comes into the Big Typescript near the end, in the “Zur Mengenlehre” section.11 This second use is the first example of a series of such quotations from 1929, sometimes to opposite and sometimes to varied effect, given the term “Wiederaufnahme” by Wolfgang Kienzler.12 Here, the 1929 meaning is that the irrational numbers are already presupposed by the rationals, the 1931 meaning is that they are no more presupposed by them than chess is presupposed by draughts. As a result of failing to notice this, they cite the 2.2.29 paragraph as showing that the Big Typescript goes all the way back to February 1929. Its opening paragraph (which remains, with slight changes, the opening paragraph of all the Philosophische Grammatik variants) first appeared in Band VI, MS 110, page 187, 20.6.31, copied from a missing small notebook. That is far from the earliest passage that has found its way into the Big Typescript. A surprising number come from the second half of Band IV, MS 108, written after TS 208 had been dictated in Vienna, but only two paragraphs come from the first half of Band IV, and they come from close to the end of that half, on pain, and were thus written in Vienna. In the Big Typescript they come in “Idealismus etc.,” in the subsection called “Schmerzen haben.” So it would be quite reasonable to say that the ideas of the Big Typescript begin with the Vienna break of Easter 1930. However, one must remember that those ideas underwent many changes, only approaching stability in the course of Band VII, MS 111. In my book I have pinpointed those changes minutely. Having made my recommendation that all serious Wittgenstein scholars should begin their study of the 1929–51 Nachlass by reading the first five Wiener Ausgabe volumes, supplemented by the electronic edition for the relevant small notebooks, I should like to return to the needs of general readers who would like printed volumes that could give them an insight into what it is about Wittgenstein that makes philosophers so attentive to him. I have already suggested that Band XII would make one such volume, moving from Philosophische Grammatik ideas towards his new start with Philosophical Investigations. Another would be three Bände (XIV, XV, and XVI) written mainly in Norway in the autumn and winter of 1937 but continuing in Vienna, Dublin, and Cambridge and recording Wittgenstein’s horror at the Anschluss (a word he could not bring himself to use). Philosophically, they do two important things: by giving us the opening sketches for the concepts of following a mathematical rule and of what came to be known as the private language argument they make both of those seem

290

Denis Paul

much more reasonable than they are in Philosophical Investigations. And biographically they give us vivid insights into his relationships with Francis Skinner and his Norwegian neighbors and into his ideas about religion, as well as his agonies about Austria. As to the question of what studying Wittgenstein might do for current and future philosophers, and for the direction of their philosophy, my view is that they will find so many different ideas that it is impossible for us to predict which will be most useful to them. That is one reason why I want them to read as much of him as possible. He disliked the idea of a one-sided diet, and a one-sided Wittgenstein diet is as bad as any other kind. So far I have given a rather optimistic account of what I think the minimum is, and I must face the fact that only a minority of philosophy students will be able to stay my course. So I shall end by specifying what I think is the real minimum. Lest I give the impression that typescripts are only for addicts of Wittgenstein’s self-editings, I must mention one that certainly doesn’t come into that category, the Blue Book, dictated to Alice Ambrose and Skinner. It electrified us in Oxford when it circulated there in samizdat in Wittgenstein’s final years. As a companion to it I think the Ambrose lecture volume is another must.13 The Brown Book was also dictated to Ambrose and Skinner, but it was part of a series of editing that led to Philosophical Investigations and is certainly not for the beginner (nor did its samizdat version electrify us at Oxford). A student who wants to read something leading to Philosophical Investigations and has access to the electronic edition should read MS 142, begun at the end of 1936 and given to his sister Margarethe Stonborough, from whose possessions it was mysteriously lost and then recovered. Then there is a little treasure, MS 166, which appears to have been Wittgenstein’s English lecture notes for a lecture he gave in his mathematics series of 1939 when Alan Turing was due to be absent, where it would have come between Lectures VI and VII, as edited by Cora Diamond.14 It is a vivid early sketch of the private language argument and is in the electronic edition, but it could well go into a printed edition of a selection from the English lecture notes. (One made by Rhees, restricted to the private language argument, which such a volume need not be, was defective.) These preparations for his lectures often show us Wittgenstein with his hair down and help bring his character to life, and a sensitively edited volume of them would be illuminating psychologically as well as filling in gaps in his philosophical development. After the war came a pair of typescripts that Josef Rothhaupt has shown to be a preparation for the additions made to pre-war Philosophical Investigations, called Bemerkungen I–II. They seem to exist in different copies, presumably top and carbon, and those in the electronic edition do not coincide with the pair that Anscombe showed me in 1952, having heard nothing from Wittgenstein about any relationship with Philosophical Investigations. They were housed in two identical box-files, and in these they answered to her story about them, that Wittgenstein wanted them to show how philosophical ideas could lead on to each other in different orders (they consist nearly but not quite of

Wittgenstein passages

291

the same paragraphs). Published in that spirit, and with photographs of the box-files, they would be useful for non-specialists and beginners. Including many paragraphs already familiar, they lack the didactic drive of Philosophical Investigations, and are thus more relaxing to read. Wittgenstein’s last three items, Remarks on Colour, the second volume of the Last Writings on the Philosophy of Psychology, and On Certainty are already published. They do need improved editing before being reprinted, and in particular On Certainty would be improved, if copyright allowed, by having the American conversations with Malcolm included.15 Non-specialists and beginners should certainly attempt all three. In my book I have explained how Remarks on Colour awakened in me an interest in colour from my very first discovery of its final, condensed notes (misguidedly printed as “Part I”), and the coolness of the academic response to it has always surprised me. The “Das Inner und das Äußere” notes, basically comprising the second volume of Last Writings on the Philosophy of Psychology are scattered through various late notebooks, one of which is possibly lost, and their quality is varied, but their best passages are very impressive. To me they are important for a quality I call their cagey ambiguity, which enables them to round off old “privacy” ideas in a way that makes them entirely sane and acceptable. An example comes in volume II, part V where he says “The ‘inner’ is an illusion” but makes it clear from further context that the way we describe it results in an illusion.16 This comes in the opening pages of MS 174, which, with a short passage written close to his death, printed in volume II, part VI offer us his final ideas on the “inner.”17 That leaves me with the quandary of what to recommend of Philosophical Investigations Part II and its contributory manuscripts (MSS 130–38). Certainly not to attempt §xi of Part II, or its drafts. In my book I have explained how the problems of seeing as or aspect have a source in the phenomenological language episode of 1929, but even with that explanation I do not think xi should be attempted by a beginner, let alone its drafts. The shorter Part II sections are another matter, however. Their condensed style makes them provocatively attractive, sometimes to the point of incomprehensibility, and this strange dual character could well spark a search for their manuscript origins. I should like to end with a reminiscence. In 1957 I met Imre Lakatos in London, having known him in Cambridge soon after his escape from Hungary. He asked me, with contempt, why I read Wittgenstein. Because it helped me think clearly, I said. If I wanted to think clearly, he said, why did I not drink strong coffee, in his thickest Hungarian accent. Ever since, I have stuck to my guns and read more and more, convinced that it was helping me think. Fairly recently, however, I have noticed a difference. Earlier, if I conducted an argument of my own, I was always clear which Wittgenstein ideas, and frequently which passages, were influencing me. Now I seem to have become detached. I simply do my best to think. I have no longer any idea what Wittgenstein has to do with the process—but I am sure that, in my

292

Denis Paul

depth, he has much. I cannot claim that this is any improvement in effect. I merely observe that it is the case. Nor do I dare recommend it as anything for anyone else to strive for. Still, I do dare to guess that in 100 years or so it is how philosophy will have advanced. All interest in the progress of Wittgenstein’s ideas and his tortuous editing of them may well have passed away—but never the effect they have had on people’s thought.

Notes 1 Ludwig Wittgenstein, Philosophische Grammatik (Oxford: Basil Blackwell, 1969). 2 Anthony Kenny, “From the Big Typescript to the Philosophical Grammar,” in Essays on Wittgenstein in Honour of G.H. von Wright, ed. Jaakko Hintikka and G. H. von Wright, Acta Philosophica Fennica (Amsterdam: North-Holland, 1976). 3 Ibid. 4 Anthony Kenny, “A Brief History of Wittgenstein Editing,” in Wittgenstein: The Philosopher and His Works, ed. Alois Pichler and Simo Säätelä, Working Papers from the Wittgenstein Archives at the University of Bergen (Bergen: WAB, 2005). 5 Ibid., p. 344; my emphasis. 6 Denis Paul, Wittgenstein’s Progress 1929–1951, vol. 19, Working Papers from the Wittgenstein Archives at the University of Bergen (Bergen: WAB, 2007). 7 MS 119, 23.10.37. 8 See PI §11, to slightly different effect. 9 Kenny, “A Brief History of Wittgenstein Editing,” p. 341. 10 Ibid., p. 342. 11 WA 11, p. 488. 12 Wolfgang Kienzler, Wittgensteins Wende zu seiner Spätphilosophie: 1930–1932 (Frankfurt am Main: Suhrkamp, 1997). 13 Ludwig Wittgenstein, Wittgenstein’s Lectures, Cambridge, 1932–1935: From the Notes of Alice Ambrose and Margaret Macdonald (Oxford: Basil Blackwell, 1979). 14 Ludwig Wittgenstein, Wittgenstein’s Lectures on the Foundations of Mathematics, Cambridge, 1939: From the Notes of R. G. Bosanquet, Norman Malcolm, Rush Rhees and Yorick Smythies (Hassocks, West Sussex: Harvester Press, 1976). 15 Norman Malcolm, Ludwig Wittgenstein: A Memoir, 2nd ed. (Oxford: Oxford University Press, 1984), pp. 71–75. 16 LW II, p. 84. 17 LW II, p. 88.

Bibliography

Ambrose, Alice, and Morris Lazerowitz (eds.) Ludwig Wittgenstein: Philosophy and Language. Muirhead Library of Philosophy. London; New York: Allen & Unwin, 1972. Amesbury, R. “Has Wittgenstein Been Misunderstood by Wittgensteinian Philosophers of Religion?” Philosophical Investigations 26 (2003): 44–72. Ammereller, Erich, and Eugen Fischer (eds.). Wittgenstein At Work: Method in the Philosophical Investigations. London: Routledge, 2004. Anscombe, G. E. M. “A Theory of Language.” In Perspectives on the Philosophy of Wittgenstein, edited by Irving Block, Oxford: Blackwell, 1981. Arrington, Robert L. (ed.). Wittgenstein’s Philosophical Investigations: Text and Context. London: Routledge, 1991. Arrington, Robert L., and Mark Addis (eds.). Wittgenstein and Philosophy of Religion. London: Routledge, 2001. Arrington, Robert L., and Hans-Johann Glock (eds.). Wittgenstein and Quine. London: Routledge, 1996. Astington, Janet W., Paul L. Harris, and David R. Olson (eds.). Developing Theories of Mind. Cambridge: Cambridge University Press, 1989. Austin, J. L. “Pretending.” In Essays in Philosophical Psychology, edited by Donald Gustafson, London: Macmillan, 1967. Baker, Gordon. Wittgenstein, Frege and the Vienna Circle. Oxford: Basil Blackwell, 1988. ——. “Philosophical Investigations Section 122: Neglected Aspects.” In Wittgenstein’s Philosophical Investigations: Text and Context, edited by Robert L. Arrington and Hans-Johann Glock, London: Routledge, 1991, pp. 35–68. Baker, Gordon, and P. M. S. Hacker. Wittgenstein: Understanding and Meaning. Vol. 1, An Analytical Commentary on the Philosophical Investigations. Oxford: Blackwell, 1980. ——. Scepticism, Rules and Language. Oxford: Blackwell, 1984. ——. Wittgenstein: Meaning and Understanding. Vol. 1, Essays on the Philosophical Investigations. Oxford: Blackwell, 1985. ——. Wittgenstein: Rules, Grammar and Necessity. Vol. 2, An Analytical Commentary on the Philosophical Investigations. Oxford: Blackwell, 1985. Banville, John. The Sea. New York: Knopf, 2005. Baron-Cohen, S. “The Development of a Theory of Mind in Autism: Deviance and Delay.” Psychiatry Clinics of North America 14 (1991): 33–51. Bartley, William Warren. Wittgenstein. London: Quartet, 1974.

294

Bibliography

Bearn, Gordon. Waking to Wonder: Wittgenstein’s Existential Investigations. Albany, NY: SUNY Press, 1997. Bell, R. H., and R. E. Hustwit (eds.). Essays on Kierkegaard and Wittgenstein. Wooster, OH: College of Wooster, 1978. van Benthem, Johan F.A.K, and Pieter Adriaans (eds.). Philosophy of Information. Amsterdam: Elsevier, 2008. Bilgrami, Akeel. “Norms and Meaning.” In Reflecting Davidson: Donald Davidson Responding to an International Forum of Philosophers, edited by Ralf Stoecker, Berlin: W. de Gruyter, 1993, pp. 121–44. Black, Max. “More about Metaphor.” In Metaphor and Thought, edited by Andrew Ortony, Cambridge: Cambridge University Press, 1979, pp. 19–43. Blackburn, Simon. “The Individual Strikes Back.” Synthese: An International Journal for Epistemology, Methodology and Philosophy of Science (1984): 281–303. Block, Irving (ed.). Perspectives on the Philosophy of Wittgenstein. Oxford: Blackwell, 1981. Block, Ned. “Inverted Earth.” In Action Theory and Philosophy of Mind, edited by J. Tomberlin, Philosophical Perspectives series, Atascadero, CA: Ridgeview, 1990. ——. “Wittgenstein and Qualia.” In Philosophical Perspectives, edited by John Hawthorne, Oxford: Blackwell, forthcoming. Bloom, Paul. How Children Learn the Meanings of Words. Cambridge, MA: MIT Press, 2000. Boghossian, Paul. “The Rule-Following Considerations.” In Rule-Following and Meaning, edited by Alexander Miller and Crispin Wright, Chesham, Bucks: Acumen, 2002. ——. “The Normativity of Content.” Philosophical Perspectives 13 (2003): 31–45. Brandom, Robert. Articulating Reasons: An Introduction to Inferentialism. Cambridge, MA: Harvard University Press, 2000. —— (ed.). Rorty and His Critics. Oxford: Blackwell, 2000. Bruno, Visentini (ed.). Linguaggi Nella Società e Nella Tecnica. Milan: Edizioni di Communità, 1970. Budd, Malcolm. Wittgenstein’s Philosophy of Psychology. London: Routledge, 1989. ——. “The Characterization of Aesthetic Qualities by Essential Metaphors and Quasi-Metaphors.” The British Journal of Aesthetics 46 (2006): 133–43. Butler, Ronald (ed.). Analytical Philosophy: First Series. Oxford: Blackwell, 1962. Byrne, Alex. “Inverted Qualia,” in The Stanford Encyclopedia of Philosophy (2006) (http://plato.stanford.edu/archives/win2006/entries/qualia-inverted/). Campbell, John. Reference and Consciousness. Oxford: Clarendon Press, 2002. Carlson, Greg, and Jeff Pelletier (eds.). Reference and Quantification: The Partee Effect. Stanford, CA: CSLI, 2005. Carnap, Rudolf. Der logische Aufbau der Welt. Hamburg: Meiner, 1928. ——. “Empiricism, Semantics and Ontology.” Revue International de Philosophie 4 (1950): 20–40. ——. Meaning and Necessity: A Study in Semantics and Modal Logic. 2nd ed. Chicago: University of Chicago Press, 1956. Carpendale, J., and Lewis, C. “Constructing an Understanding of Mind: The Development of Children’s Social Understanding within Social Interaction.” Behavioural and Brain Sciences 27 (2004): 79–96. Carr, David. The Paradox of Subjectivity: The Self in the Transcendental Tradition. Oxford: Oxford University Press, 1999.

Bibliography

295

Carruthers, Peter. “Critical Study: Baker and Hacker’s Wittgenstein.” Synthese: An International Journal for Epistemology, Methodology and Philosophy of Science 58 (1984): 451–79. ——. Phenomenal Consciousness: A Naturalistic Theory. Cambridge: Cambridge University Press, 2000. Carruthers, Peter, and Peter K. Smith (eds.). Theories of Theories of Mind. Cambridge: Cambridge University Press, 1996. Cartwright, R. “Propositions.” In Analytical Philosophy: First Series, edited by Ronald Butler, Oxford: Blackwell, 1962, pp. 81–103. Cavell, Stanley. “The Availability of Wittgenstein’s Later Philosophy.” The Philosophical Review LXXI (1962): 67–93. ——. “Existentialism and Analytic Philosophy.” Daedalus (1964): 946–74. ——. The Claim of Reason: Wittgenstein, Skepticism, Morality and Tragedy. Oxford: Clarendon Press, 1979. ——. This New Yet Unapproachable America: Lectures after Emerson after Wittgenstein. The Frederick Ives Carpenter Lectures 1987. Albuquerque, NM: Living Batch, 1989. Chalmers, David. The Conscious Mind: In Search of a Fundamental Theory. Oxford: Oxford University Press, 1996. Chierchia, Gennaro, and Sally McConnell-Ginet. Meaning and Grammar. 2nd ed. Cambridge, MA: MIT Press, 2000. Chomsky, Noam. “Language and Nature.” Mind 104 (1995): 1–62. Churchland, Paul M. Matter and Consciousness: A Contemporary Introduction to the Philosophy of Mind. Cambridge, MA: MIT, 1984. Clack, Brian. An Introduction to Wittgenstein’s Philosophy of Religion. Edinburgh: Edinburgh University Press, 1999. ——. “Wittgenstein and Magic.” In Wittgenstein and Philosophy of Religion, edited by Robert L. Arrington and Mark Addis, London: Routledge, 2001. ——. “Response to Phillips.” Religious Studies 39 (2003): 203–9. Cook, John W. “Wittgenstein and Religious Belief.” Philosophy 63 (1988): 427–52. ——. “Wittgenstein and Magic.” In Wittgenstein and Philosophy of Religion, edited by Robert L. Arrington and Mark Addis, London: Routledge, 2001. ——. The Undiscovered Wittgenstein: The Twentieth Century’s Most Misunderstood Philosopher. Amherst, MA: Humanity Books, 2005. Copeland, Jack B. “The Genesis of Possible Worlds Semantics.” Journal of Philosophical Logic 31 (2002): 99–137. Coughlan, M. J. “Wittgenstein, Language and Religious Belief.” In God in Language, edited by R. P. Scharlemann and G. E. M. Ogutu, New York: Paragon House, 1987, pp. 149–65. Craig, Edward (ed.). Routledge Encyclopedia of Philosophy. London: Routledge, 1998. Cresswell, Max. Logics and Languages. London: Methuen, 1973. Damasio, Antonio R. Descartes’ Error: Emotion, Reason and the Human Brain. London: Papermac, 1996. Davidson, Donald. “Theories of Meaning and Learnable Languages.” In Proceedings of the 1964 International Congress for Logic, Methodology, and Philosophy of Science, edited by Bar-Hillel Yehoshua, Amsterdam: North-Holland, 1965. ——. “Truth and Meaning.” Synthese: An International Journal for Epistemology, Methodology and Philosophy of Science 17 (1967): 304–23.

296

Bibliography

——. “Semantics for Natural Languages.” In Linguaggi Nella Società e Nella Tecnica, edited by Visentini Bruno, Milan: Edizioni di Communità, 1970. ——. “In Defense of Convention ‘T’.” In Truth, Syntax and Modality, edited by Leblanc Hugues, Amsterdam: North-Holland, 1973. ——. “On the Very Idea of a Conceptual Scheme.” Proceedings and Addresses of The American Philosophical Association 47 (1974): 5–20 ——. “Reality without Reference.” Dialectica 31 (1977): 247–53. ——. “The Method of Truth in Metaphysics.” Midwest Studies in Philosophy 2 (1977): 294–304. ——. Inquiries into Truth and Interpretation. Oxford: Oxford University Press, 1984. ——. “The Structure and Content of Truth.” Journal of Philosophy 87 (1990): 279–328. ——. “Reply to Akeel Bilgrami.” In Reflecting Davidson: Donald Davidson Responding to an International Forum of Philosophers, edited by Ralf Stoecker, Berlin: W. de Gruyter, 1993, pp. 145–47. ——. “The Social Aspect of Language.” In The Philosophy of Michael Dummett, edited by Brian McGuinness and Gianluigi Oliveri, Dordrecht: Kluwer, 1994, pp. 1–16. ——. “Truth Rehabilitated.” In Rorty and His Critics, edited by Robert Brandom, Oxford: Blackwell, 2000. Davidson, Donald, and Gilbert Harman (eds.). Semantics of Natural Language. 2nd ed. Dordrecht: Reidel, 1972. Davies, Martin, and Glyn Humphreys (eds.). Consciousness: Psychological and Philosophical Essays. Vol. 2, Readings in Mind and Language. Oxford: Blackwell, 1993. Dawson, Lawrence, and Edmond Hoyle. Hoyle’s Games Modernized. London: Routledge, 1934. De Caro, Mario (ed.). Interpretations and Causes: New Perspectives on Donald Davidson’s Philosophy. Dordrecht: Kluwer Academic, 1999. de Swart, Henriëtte. Introduction to Natural Language Semantics. Stanford, CA: CSLI, 1998. Decock, Lieven, and Leon Horsten (eds.). Quine: Naturalized Epistemology, Perceptual Knowledge and Ontology. Vol. 70, Poznan´ Studies in the Philosophy of the Sciences and the Humanities. Amsterdam: Rodopi, 2000. Dennett, Daniel C. “Intentional Systems.” The Journal of Philosophy 68 (1971): 87–106. ——. Consciousness Explained. London: Penguin, 1993. Diamond, Cora. “Secondary Sense.” In The Realistic Spirit: Wittgenstein, Philosophy, and the Mind, Cambridge, MA: MIT Press, 1991. ——. The Realistic Spirit: Wittgenstein, Philosophy, and the Mind. Cambridge, MA: MIT Press, 1991. ——. “Ethics, Imagination and the Method of Wittgenstein’s Tractatus.” In Bilder der Philosophie: Reflexionen über das Bildliche und die Phantasie, edited by R. Heinrich and H. Vetter, Vienna; Munich: Oldenbourg, 1991, pp. 55–90. Dowty, David, Robert Wall, and Stanley Peters. Introduction to Montague Semantics. Dordrecht: Reidel, 1981. Dreier, James. “Structure of Normative Theories.” The Monist 76 (1993): 22–40. Drury, M. O’C. “Conversations with Wittgenstein.” In Recollections of Wittgenstein, edited by Rush Rhees, Oxford: Oxford University Press, 1984. Dummett, M. A. E. Frege: Philosophy of Language. London: Duckworth, 1981. ——. The Logical Basis of Metaphysics. London: Duckworth, 1991. Edwards, James C. Ethics without Philosophy: Wittgenstein and the Moral Life. Tampa, FL: University Presses of Florida, 1985.

Bibliography

297

Engelmann, Paul. Letters from Ludwig Wittgenstein: With a Memoir. Translated by L. Furtmüller. Edited by Brian McGuinness. Oxford: Basil Blackwell, 1967. Finch, Henry LeRoy. Wittgenstein. The Spirit of Philosophy. Rockport, MA: Element, 1995. Floyd, Juliet. “Wittgenstein, Mathematics and Philosophy.” In The New Wittgenstein, edited by Rupert Read and Alice Crary, London: Routledge, 2000, pp. 232–61. Fodor, Jerry. “The Present State of the Innateness Controversy.” In Representations : Philosophical Essays on the Foundations of Cognitive Science. Brighton: Harvester Press, 1981. ——. Concepts: Where Cognitive Science Went Wrong. Oxford: Clarendon Press, 1998. ——. “Doing Without What’s Within: Fiona Cowie’s Critique of Nativism.” Mind 110:437 (2001): 99–148. Fogelin, Robert. Wittgenstein. London: Routledge & Kegan Paul, 1976. ——. Wittgenstein. 2nd ed. London: Routledge & Kegan Paul, 1987. Forbes, G. “Scepticism and Semantic Knowledge.” In Rule-Following and Meaning, edited by Alexander Miller and Crispin Wright, Chesham, Bucks: Acumen, 2002, pp. 16–27. Forster, Michael. Wittgenstein on the Arbitrariness of Grammar. Princeton, NJ: Princeton University Press, 2004. Frankfurt, Harry. “Alternate Possibilities and Moral Responsibility.” Journal of Philosophy 66 (1969): 829–39. Frege, Gottlob. “Der Gedanke: Eine logische Untersuchung.” Beiträge zur Philosophie des deutschen Idealismus 2 (1918): 58–77. ——. “Funktion und Begriff.” In Funktion, Begriff, Bedeutung: Fünf Logische Studien, edited by Günther Patzig, Göttingen: Vandenhoeck & Ruprecht, 1962. ——. “Was ist eine Funktion?” In Funktion, Begriff, Bedeutung: Fünf Logische Studien, edited by Günther Patzig, Göttingen: Vandenhoeck & Ruprecht, 1962. ——. Funktion, Begriff, Bedeutung: Fünf logische Studien. Edited by Günther Patzig. Göttingen: Vandenhoeck & Ruprecht, 1962. ——. The Basic Laws of Arithmetic: Exposition of the System. Translated and edited by Montgomery Furth, Berkeley, CA: University of California Press, 1964. ——. “Logik in der Mathematik.” In Nachgelassene Schriften, edited by Hans Hermes, Friedrich Kambartel, and Friedrich Kaulbach, Hamburg: Meiner, 1969. ——. Nachgelassene Schriften. Edited by Hans Hermes, Friedrich Kambartel, and Friedrich Kaulbach. Hamburg: Meiner, 1969. ——. Posthumous Writings. Translated by Peter Long, Roger White, and Raymond Hargreaves. Edited by Hans Hermes, Friedrich Kambartel, and Friedrich Kaulbach. Oxford: Blackwell, 1979. ——. “Comments on Sense and Meaning.” In Posthumous Writings, edited by Hans Hermes, Friedrich Kambartel, and Friedrich Kaulbach, Oxford: Blackwell, 1979. ——. “My Basic Logical Insights.” In Posthumous Writings, edited by Hans Hermes, Friedrich Kambartel, and Friedrich Kaulbach, Oxford: Blackwell, 1979. ——. “17 Key Sentences on Logic.” In Posthumous Writings, edited by Hans Hermes, Friedrich Kambartel, and Friedrich Kaulbach, Oxford: Blackwell, 1979. ——. “Notes for Ludwig Darmstädter.” In Posthumous Writings, edited by Hans Hermes, Friedrich Kambartel, and Friedrich Kaulbach, Oxford: Blackwell, 1979. ——. “Logic.” In Posthumous Writings, edited by Hans Hermes, Friedrich Kambartel, and Friedrich Kaulbach, Oxford: Blackwell, 1979.

298

Bibliography

——. Die Grundlagen der Arithmetik: Eine logisch mathematische Untersuchung über den Begriff der Zahl. Edited by Christian Thiel. Centennial ed. Hamburg: Meiner, 1986. Friedman, Michael. “Carnap’s Aufbau Reconsidered.” Noûs 21(4), no. 4 (1987): 521– 45. Gabbay, Dov, and John Woods (eds.). Handbook of the History of Logic. Vol. 6: Logic and the Modalities in the Twentieth Century, Amsterdam: Elsevier, 2005. Gamut, L. T. F. Logic, Language and Meaning. Vol. 2: Intensional Logic and Logical Grammar, Chicago: University of Chicago Press, 1991. Geach, Peter. Mental Acts. London: Routledge, 1957. ——. “Nominalism.” Sophia 3 (1964): 3–14. Gellner, Ernest. Words and Things: A Critical Account of Linguistic Philosophy and a Study in Ideology. London: Gollancz, 1959. German, T. P., and A. M. Leslie. “Children’s Inferences from ‘Knowing’ to ‘Pretending’ and ‘Believing’.” British Journal of Developmental Psychology 19 (2001): 59–83. Gibbard, Allan. “Thoughts and Norms.” Philosophical Perspectives 13: 83–98. Gier, Nicholas F. Wittgenstein and Phenomenology: A Comparative Study of the Later Wittgenstein, Husserl, Heidegger, and Merleau-Ponty. SUNY series in philosophy, Albany, NY: SUNY Press, 1981. Glock, Hans-Johann. A Wittgenstein Dictionary. Oxford: Blackwell, 1996. ——. “Necessity and Normativity.” In The Cambridge Companion to Wittgenstein, edited by Hans D. Sluga and David G. Stern, Cambridge: Cambridge University Press, 1996, pp. 198–225. ——. “Wie kam die Bedeutung zur Regel.” Deutsche Zeitschrift für Philosophie 48 (2000): 429–47. ——. “Sense and Meaning in Frege and the Tractatus.” In From the Tractatus to the Tractatus (and Other Essays), edited by G. Oliveri, Frankfurt am Main: Peter Lang, 2000, pp. 53–68. ——. Quine and Davidson on Language, Thought and Reality. Cambridge: Cambridge University Press, 2003. ——. “All Kinds of Nonsense.” In Wittgenstein At Work: Method in the Philosophical Investigations, edited by Erich Ammereller and Eugen Fischer, London: Routledge, 2004, pp. 221–45. ——. “The Normativity of Meaning Made Simple.” In Philosophy—Science— Scientific Philosophy, edited by C. Nimitz and A. Beckermann, Paderborn, Germany: Mentis, 2005. Glüer, K. “Dreams and Nightmares. Conventions, Norms, and Meaning in Davidson’s Philosophy of Language.” In Interpreting Davidson, edited by Petr Koþtátko, Peter Pagin, and Gabriel Segal, Stanford, CA: CSLI Publications, 2001, pp. 53–74. Glüer, K., and Peter Pagin. “Rules of Meaning and Practical Reasoning.” Synthese: An International Journal for Epistemology, Methodology and Philosophy of Science 118 (1999): 208–27. Goldblatt, Rob. “Mathematical Modal Logic: A View of Its Evolution.” In Handbook of the History of Logic, edited by Dov Gabbay and John Woods, Amsterdam: Elsevier, 2005. Goldfarb, Warren. “I Want You to Bring Me a Slab. Remarks on the Opening Sections of the Philosophical Investigations.” Synthese: An International Journal for Epistemology, Methodology and Philosophy of Science 56 (1983): 265–82.

Bibliography

299

——. “Kripke on Wittgenstein on Rules.” The Journal of Philosophy 82:9 (1985): 471–88. Goldman, A. I. “Folk Psychology and Mental Concepts.” Protosociology 14 (2000): 4–25. Goldstein, Laurence. “What Does ‘Experiencing Meaning’ Mean?” In The Third Wittgenstein: The Post-Investigations Works, edited by Danièle Moyal-Sharrock, Aldershot: Ashgate, 2004. Goodman, R. B. “How a Thing is Said and Heard: Wittgenstein and Kierkegaard.” History of Philosophy Quarterly 3 (1986): 335–53. Gopnik, Alison. “Theories and Modules: Creation Myths, Developmental Realities, and Neurath’s Boat.” In Theories of Theories of Mind, edited by Peter Carruthers and Peter K. Smith, Cambridge: Cambridge University Press, 1996, pp. 169–83. Gopnik, Alison, Andrew Meltzoff, and Patricia Kuhl. The Scientist in the Crib: Minds, Brains, and How Children Learn. New York: William Morrow, 1999. Gregory, R. L., and O. L. Zangwill (eds.). The Oxford Companion to the Mind. Oxford: Oxford University Press, 1987. Grice, H. P. Studies in the Way of Words. Cambridge, MA: Harvard University Press, 1989. ——. “Meaning.” In Studies in the Way of Words, Cambridge, Mass.: Harvard University Press, 1989. Griffin, James. Wittgenstein’s Logical Atomism. Oxford: Clarendon Press, 1964. Griffiths, A. Phillips (ed.). Wittgenstein Centenary Essays. Vol. 28, Royal Institute of Philosophy Lectures. Cambridge: Cambridge University Press, 1991. Groenendijk, Jeroen, and Martin Stokhof. “Why Compositionality?” In Reference and Quantification: The Partee Effect, edited by Greg Carlson and Jeff Pelletier, Stanford, CA: CSLI, 2005, pp. 83–106. Gustafson, Donald (ed.). Essays in Philosophical Psychology. London: Macmillan, 1967. Hacker, P. M. S. Insight and Illusion: Themes in the Philosophy of Wittgenstein. Oxford: Clarendon, 1986. ——. Wittgenstein: Mind and Will. Vol. 4, An Analytical Commentary on the Philosophical Investigations. Oxford: Blackwell, 1996. ——. Wittgenstein’s Place in Twentieth-Century Analytic Philosophy. Oxford: Blackwell, 1997. Hanfling, Oswald. Wittgenstein’s Later Philosophy. Basingstoke: Macmillan, 1989. ——. “I Heard a Plaintive Melody.” In Wittgenstein Centenary Essays, edited by A. Phillips Griffiths, Cambridge: Cambridge University Press, 1991. ——. “Secondary Sense and “What They Have in Common”.” In Wittgenstein and the Human Form of Life, London: Routledge, 2002. ter Hark, Michel. Beyond the Inner and the Outer: Wittgenstein’s Philosophy of Psychology. Dordrecht: Kluwer, 1990. Harris, P. L. “Imagining and Pretending.” In Mental Simulation: Evaluations and Applications, edited by Tony Stone and Martin Davies, Oxford: Blackwell, 1995. Hawthorne, John (ed.). Philosophical Perspectives. Oxford: Blackwell, forthcoming. Heidegger, Martin. Grundprobleme der Phänomenologie (1919/20). Edited by HansHelmuth Gander. Frankfurt am Main: Klostermann, 1993. ——. Being and Time. Translated by Joan Stambaugh. Albany, NY: SUNY Press, 1996.

300

Bibliography

Henry, P., and A. Utaker (eds.). Wittgenstein and Contemporary Theories of Language. Vol. 5, Working Papers from the Wittgenstein Archives at the University of Bergen. Bergen: WAB, 1992. Hick, John (ed.). Faith and the Philosophers. London: Macmillan, 1964. High, D. M. “On Thinking More Crazily than Philosophers: Wittgenstein, Knowledge and Religious Beliefs.” International Journal for Philosophy of Religion 19 (1986): 161–75. Hilmy, S. Stephen. The Later Wittgenstein: The Emergence of a New Philosophical Method. Oxford: Basil Blackwell, 1987. Hintikka, Jaakko, Julius Moravcsik, and Patrick Suppes (eds.). Approaches to Natural Language. Dordrecht: Reidel, 1973. Hintikka, Jaakko, and G. H. von Wright (eds.). Essays on Wittgenstein in Honour of G. H. von Wright. Vol. 28, Acta Philosophica Fennica. Amsterdam: NorthHolland, 1976. Hintikka, Merrill, and Jaakko Hintikka. Investigating Wittgenstein. Oxford: Blackwell, 1986. Hopkins, James. “Wittgenstein and Physicalism.” Proceedings of the Aristotelian Society 75 (1975): 121–46. Horwich, Paul. From a Deflationary Point of View. Oxford: Clarendon Press, 2004. Hudson, W. D. “The Light Wittgenstein Sheds on Religion.” Midwest Studies in Philosophy 6 (1981): 275–92. Hugues, Leblanc (ed.). Truth, Syntax and Modality. Amsterdam: North-Holland, 1973. Hursthouse, Rosalind. On Virtue Ethics. Oxford: Oxford University Press, 1999. Husserl, Edmund. Cartesianische Meditationen und Pariser Vorträge. Edited by S. Strasser. Vol. 1, Husserliana. The Hague: Nijhoff, 1950. ——. Ideen zu einer reinen Phänomenologie und Phänomenologischen Philosophie. Edited by Marly Biemel. Vol. 5, Husserliana. The Hague: Nijhoff, 1952. ——. Ideas Pertaining to a Pure Phenomenology and to a Phenomenological Philosophy. Translated by F. Kersten. Dordrecht; London: Kluwer, 1982. ——. Aufsätze und Vorträge, 1911–1921. Edited by T. Nenon and H. R. Sepp. Vol. 25, Husserliana. Dordrecht: Nijhoff, 1986. ——. Logical Investigations. Translated by J. N. Findlay. Edited by Dermot Moran. London: Routledge, 2001. Hustwit, R. E. “Understanding a Suggestion of Professor Cavell’s: Kierkegaard’s Religious State as a Wittgensteinian ‘Form of Life’.” Philosophy Research Archives 4 (1978): 329–47. Hutto, D., and M. Ratcliffe (eds.). Folk Psychology Reassessed. Springer, 2007. Hutto, Daniel D. Wittgenstein and the End of Philosophy: Neither Theory Nor Therapy. Basingstoke: Palgrave Macmillan, 2003. Hyman, John. “The Gospel According to Wittgenstein.” In Wittgenstein and Philosophy of Religion, edited by Robert L. Arrington and Mark Addis, London: Routledge, 2001. Jackson, Frank. “Epiphenomenal Qualia.” The Philosophical Quarterly 32:127 (1982): 127–36. James, William. The Principles of Psychology. 2 vols. London: Macmillan, 1890. Janssen, Theo. “Frege, Contextuality and Compositionality.” Journal of Logic, Language and Information 10 (2001): 115–36. Kalhat, J. “Has the Later Wittgenstein Accounted for Necessity?” Philosophical Investigations 31: 1 (2008): 1–23.

Bibliography

301

Kamp, Hans, and Martin Stokhof. “Information in Natural Language.” In Handbook of Philosophy of Information, edited by Johan F.A.K. van Benthem, and Pieter Adriaans, Amsterdam: Elsevier, 2008. Kang, Jinho. “On the Composition of the Prototractatus.” The Philosophical Quarterly 55 (2005): 2–20. Kellenberger, J. “‘Seeing-As’ in Religion: Discovery and Community.” Religious Studies 38 (2002): 101–8. Kenny, Anthony. Wittgenstein. London: Allen Lane, 1973. ——. “From the Big Typescript to the Philosophical Grammar.” In Essays on Wittgenstein in Honour of G.H. Von Wright, edited by Jaakko Hintikka and G. H. von Wright, Amsterdam: North-Holland, 1976, pp. 41–53. ——. “A Brief History of Wittgenstein Editing.” In Wittgenstein: The Philosopher and His Works, edited by Alois Pichler and Simo Säätelä, Bergen: WAB, 2005, pp. 341–55. Kienzler, Wolfgang. Wittgensteins Wende zu seiner Spätphilosophie: 1930–1932. Frankfurt am Main: Suhrkamp, 1997. Kirk, Robert. Raw Feeling: A Philosophical Account of the Essence of Consciousness. Oxford: Clarendon Press, 1994. Koþtátko, Petr, Peter Pagin, and Gabriel Segal (eds.), Interpreting Davidson. Vol. 129, CSLI Lecture Notes. Stanford, CA: CSLI Publications, 2001. Kripke, Saul A. Wittgenstein on Rules and Private Language: An Elementary Exposition. Oxford: Blackwell, 1982. Leslie, A. M. “Pretence and Representation: The Origins of ‘Theory of Mind’.” Psychological Review 94 (1987): 412–26. ——. “Children’s Understanding of the Mental World.” In The Oxford Companion to the Mind, edited by R. L. Gregory and O. L. Zangwill, Oxford: Oxford University Press, 1987, pp. 139–42. ——. “Some Implications of Pretence for Mechanisms Underlying the Child’s Theory of Mind.” In Developing Theories of Mind, edited by Janet W. Astington, Paul L. Harris, and David R. Olson, Cambridge: Cambridge University Press, 1989. Levine, J. “On Leaving Out What It’s Like.” In Consciousness: Psychological and Philosophical Essays, edited by Martin Davies and Glyn Humphreys, Readings in Mind and Language series, Oxford: Blackwell, 1993. Lewis, Charlie, and Peter Mitchell (eds.). Children’s Early Understanding of Mind: Origins and Development. Hove, East Sussex: Lawrence Erlbaum, 1994. Lewis, David. Convention: A Philosophical Study. Cambridge, MA: Harvard University Press, 1969. ——. “General Semantics.” Synthese: An International Journal for Epistemology, Methodology and Philosophy of Science 22 (1970): 18–67. ——. Counterfactuals. Oxford: Blackwell, 1973. Lewy, C. “A Note on the Text of the Tractatus.” Mind 76 (1967): 416–23. Lillard, A. “Young Children’s Conceptualization of Pretense: Action or Mental Representational State?” Child Development 64 (1993): 372–86. ——. “Making Sense of Pretense.” In Children’s Early Understanding of Mind: Origins and Development, edited by Charlie Lewis and Peter Mitchell, Hove, East Sussex: Lawrence Erlbaum, 1994. ——. “Wanting to Be it: Children’s Understanding of Intentions Underlying Pretense.” Child Development 69 (1998): 981–93. Lindqvist, Sven. Desert Divers. Translated by Joan Tate. London: Granta Books, 2000.

302

Bibliography

Lippit, J., and D. Hutto. “Making Sense of Nonsense: Kierkegaard and Wittgenstein.” Proceedings of the Aristotelian Society 98 (1998): 263–86. Loar, Brian. “Phenomenal States.” In Action Theory and Philosophy of Mind, edited by J. Tomberlin, Philosophical Perspectives series, Atascadero, CA: Ridgeview, 1990. Luntley, Michael. Wittgenstein: Meaning and Judgement. Oxford: Blackwell, 2003. Malcolm, Norman. “Is it a Religious Belief That ‘God Exists’?” In Faith and the Philosophers, edited by John Hick, London: Macmillan, 1964. ——. Ludwig Wittgenstein: A Memoir. 2nd ed. Oxford: Oxford University Press, 1984. ——. Wittgenstein: A Religious Point of View?. Edited by Peter Winch. London: Routledge, 1993. ——. Wittgensteinian Themes: Essays, 1978–1989. Edited by G. H. von Wright. Ithaca, NY: Cornell University Press, 1995. Martin, D. “‘On Certainty’ and Religious Belief.” Religious Studies 20 (1984): 593–613. ——. “God and Objects: Beginning with Existence.” International Journal for Philosophy of Religion 41 (1997): 1–11. McCulloch, Gregory. The Mind and Its World. London: Routledge, 1995. ——. The Life of the Mind: An Essay on Phenomenological Externalism. London: Routledge, 2003. McCune-Nicolich, L. “Toward Symbolic Functioning: Structure of Early Pretend Games and Potential Parallels with Language.” Child Development 52 (1981): 785–97. McDowell, John. Mind and World. Cambridge, MA: Harvard University Press, 1994. ——. “Wittgenstein on Following a Rule.” Synthese: An International Journal for Epistemology, Methodology and Philosophy of Science 58 (1994): 325–63. ——. Meaning, Knowledge, and Reality. Cambridge, MA: Harvard University Press, 1998. ——. “Criteria, Defeasibility and Knowledge.” In Meaning, Knowledge, and Reality. Cambridge, MA: Harvard University Press, 1998, pp. 369–94. ——. Mind, Value, and Reality. Cambridge, MA: Harvard University Press, 1998. McGinn, Marie. Wittgenstein and the Philosophical Investigations. London: Routledge, 1997. ——. “The Real Problem of Others: Cavell, Merleau-Ponty and Wittgenstein on Scepticism about Other Minds.” European Journal of Philosophy 6 (1998): 45–58. McGuinness, Brian. Approaches to Wittgenstein: Collected Papers. London: Routledge, 2002. McGuinness, Brian, and Gianluigi Oliveri (eds.). The Philosophy of Michael Dummett. Vol. 239, Synthese Library. Dordrecht: Kluwer, 1994. Merleau-Ponty, Maurice. Phenomenology of Perception. London: Routledge, 1962. ——. The Primacy of Perception and Other Essays on Phenomenological Psychology, the Philosophy of Art, History, and Politics. Edited by James M. Edie. Evanston, IL: Northwestern University Press, 1964. Miller, Alexander, and Crispin Wright (eds.). Rule-Following and Meaning. Chesham, Bucks: Acumen, 2002. Monk, Ray. “The Temptations of Phenomenology.” (Unpublished). Montague, Richard. “English as a Formal Language.” In Linguaggi Nella Società e Nella Tecnica, edited by Visentini Bruno, Milan: Edizioni di Communità, 1970, pp. 189–224. ——. “On the Nature of Certain Philosophical Entities.” The Monist 53 (1970): 159–94. ——. “Universal Grammar.” Theoria 36 (1970): 373–98. ——. “The Proper Treatment of Quantification in Ordinary English.” In Approaches to Natural Language, edited by Jaakko Hintikka, Julius Moravcsik, and Patrick Suppes, Dordrecht: Reidel, 1973, pp. 221–42.

Bibliography

303

Thomason, Richmond (ed.), Formal Philosophy: Selected Papers of Richard Montague. New Haven, CT: Yale University Press, 1974. Mounce, H. O. “Philosophy, Solipsism and Thought.” The Philosophical Quarterly 47:186 (1997): 1–18. Moyal-Sharrock, Danièle (ed.). The Third Wittgenstein: The Post-Investigations Works. Aldershot: Ashgate, 2004. Mulhall, Stephen. On Being in the World: Wittgenstein and Heidegger on Seeing Aspects. London: Routledge, 1990. Nielsen, Kai. “Wittgensteinian Fideism.” Philosophy 61 (1967): 191–209. ——. “Wittgenstein and Wittgensteinians on Religion.” In Wittgenstein and Philosophy of Religion, edited by Robert L. Arrington and Mark Addis, London: Routledge, 2001. Nimitz, C., and A. Beckermann (eds.). Philosophy—Science—Scientific Philosophy. Main Lectures and Colloquia of Fifth International Congress of the Society for Analytical Philosophy, Bielefeld, 22–26 September 2003. Paderborn, Germany: Mentis, 2005. Noë, R. A. “Wittgenstein, Phenomenology, and What it Makes Sense to Say.” Philosophy and Phenomenological Research 54 (1994): 1–42. Ogden, C. K., and I. A. Richards. The Meaning of Meaning: A Study of the Influence of Language upon Thought and of the Science of Symbolism. London: Kegan Paul, Trench, Trubner, 1923. Oliveri, G. (ed.). From the Tractatus to the Tractatus (and Other Essays). Vol. 2, Wittgenstein Studies. Frankfurt am Main: Peter Lang, 2000. Ortony, Andrew (ed.). Metaphor and Thought. Cambridge: Cambridge University Press, 1979. Overgaard, S. Wittgenstein and Other Minds. New York: Routledge, 2007. Palmeri, Thomas J., Randolph Blake, Rene Marois, Marci A. Flanery, and William Whetsell, Jr. “The Perceptual Reality of Synesthetic Colors.” PNAS 99:6 (2002): 4127–31. Papineau, David. Thinking about Consciousness. Oxford: Oxford University Press, 2002. Park, Byong-Chul. Phenomenological Aspects of Wittgenstein’s Philosophy. Dordrecht; London: Kluwer Academic, 1998. Pat-Shamir, Galia. “To Live a Riddle: The Case of Binding Isaac.” Philosophy and Literature 27 (2003): 269–83. Paul, Denis. Wittgenstein’s Progress 1929–1951. Vol. 19, Working Papers from the Wittgenstein Archives at the University of Bergen. Bergen: WAB, 2007. Pears, David. The False Prison: A Study of the Development of Wittgenstein’s Philosophy. Vol. II. Oxford: Clarendon Press, 1987. Perry, John. Knowledge, Possibility, and Consciousness. Vol. 79 (7), The Jean Nicod Lectures. Cambridge, MA: MIT Press, 2001. Petr Hájek, Luis Valdés-Villanueva, and Dag Westerståhl (eds.). Proceedings of the Twelfth International Congress for Logic, Methodology and Philosophy of Science. London: King’s College Publications, 2005. Phillips, D. Z. “Religious Beliefs and Language Games.” Ratio 12 (1970): 24–46. ——. “Wittgenstein, Wittgensteinianism, and Magic: A Philosophical Tragedy?” Religious Studies 39 (2003): 185–200. —— (ed.). Wittgenstein and the Possibility of Discourse. 2nd ed. Oxford: Blackwell, 2006. Phillips, D. Z., and Mario von der Ruhr (eds.). Religion and Wittgenstein’s Legacy. Burlington, VT: Ashgate, 2005.

304

Bibliography

Picardi, Eva. “Davidson and Quine on Observation Sentences.” In Language, Mind and Epistemology: On Donald Davidson’s Philosophy, edited by Gerhard Preyer, Frank Siebelt, and Alexander Ulfig, Dordrecht: Kluwer Academic, 1994, pp. 97–116. ——. “Sensory Evidence and Shared Interests.” In Interpretations and Causes: New Perspectives on Donald Davidson’s Philosophy, edited by Mario De Caro, Dordrecht: Kluwer Academic, 1999, pp. 171–85. ——. “Empathy and Charity.” In Quine: Naturalized Epistemology, Perceptual Knowledge and Ontology, edited by Lieven Decock and Leon Horsten, Poznan´ Studies in the Philosophy of the Sciences and the Humanities, Amsterdam: Rodopi, 2000, pp. 121–34. Pichler, Alois, and Simo Säätelä (eds.). Wittgenstein: The Philosopher and His Works. Vol. 17, Working Papers from the Wittgenstein Archives at the University of Bergen. Bergen: WAB, 2005. Plato. Complete Works. Edited by John Cooper. Indianapolis, IN: Hackett, 1997. Preyer, Gerhard, Frank Siebelt, and Alexander Ulfig (eds.). Language, Mind and Epistemology: On Donald Davidson’s Philosophy. Dordrecht: Kluwer Academic, 1994. Prior, A. N. “The Runabout Inference-Ticket.” Analysis 21:2 (1960): 38–39. Pullum, Geoffrey, and Barbara Scholz. “Contrasting Applications of Logic in Natural Language Syntactic Description.” In Proceedings of the Twelfth International Congress for Logic, Methodology and Philosophy of Science, edited by Petr Hájek, Luis Valdés-Villanueva, and Dag Westerståhl, London: King’s College Publications, 2005. Putnam, Hilary. “Wittgenstein on Religious Belief.” In Renewing Philosophy, Cambridge, MA: Harvard University Press, 1992. Quine, W. V. Word and Object. Cambridge, MA: MIT Press, 1960. Ramachandran, V. S., and E. M. Hubbard. “Synesthesia—A Window into Perception, Thought and Language.” Journal of Consciousness Studies 8 (2001): 3–34. ——. “Hearing Colors, Tasting Shapes.” Scientific American 288:5 (2003): 52–59. Ramsey, F. P. “Facts and Propositions.” In Philosophical Papers, edited by D. H. Mellor, Cambridge: Cambridge University Press, 1990. ——. Philosophical Papers. Edited by D. H. Mellor. Cambridge: Cambridge University Press, 1990. Ranta, Aarne. “Intuitionistic Categorial Grammar.” Linguistics and Philosophy 14:2 (1991): 203–39. Raz, Joseph. “Reasons for Action, Decisions and Norms.” In Practical Reasoning, edited by Joseph Raz, Oxford: Oxford University Press, 1978. ——. Practical Reason and Norms. Princeton, NJ: Princeton University Press, 1990. —— (ed.). Practical Reasoning. Oxford Readings in Philosophy. Oxford: Oxford University Press, 1978. Read, Rupert, and Alice Crary (eds.). The New Wittgenstein. London: Routledge, 2000. Redpath, Theodore. “Wittgenstein and Ethics.” In Ludwig Wittgenstein: Philosophy and Language, edited by Alice Ambrose and Morris Lazerowitz, London: George Allen and Unwin, 1972, pp. 95–119. Reeder, H. P. “Wittgenstein Never Was a Phenomenologist.” Journal of the British Society for Phenomenology 20 (1989): 257–76. Rhees, Rush. “Some Developments in Wittgenstein’s View of Ethics.” The Philosophical Review 74:1 (1965): 17–26.

Bibliography

305

——. “What Are Moral Statements Like?” In Without Answers, edited by D. Z. Phillips, London: Routledge & Kegan Paul, 1969, pp. 103–09. ——. “Wittgenstein’s Builders.” In Discussions of Wittgenstein, London: Routledge & Kegan Paul, 1970, pp. 71–84. ——. Discussions of Wittgenstein. London: Routledge & Kegan Paul, 1970. ——. “Preface to the Blue and Brown Books.” Oxford: Basil Blackwell, 1972, pp. v–xiv. ——. “Wittgenstein on Language and Ritual.” In Essays on Wittgenstein in Honour of G.H. Von Wright, edited by Jaakko Hintikka and G. H. von Wright, Amsterdam: North-Holland, 1976, pp. 450–84. —— (ed.). Ludwig Wittgenstein: Personal Recollections. Oxford: Blackwell, 1981. —— (ed.). Recollections of Wittgenstein. Revised ed. Oxford: Oxford University Press, 1984. ——. Discussions of Wittgenstein. Wittgenstein Studies. Bristol: Thoemmes, 1996. Rich, A. N., J. L. Bradshaw, and J. B. Mattingley. “A Systematic, Large-Scale Study of Synaesthesia: Implications for the Role of Early Experience in Lexical–Colour Associations.” Cognition: International Journal of Cognitive Science 98:1 (2005): 53–84. Rich, A. N., and J. B. Mattingley. “Anomalous Perception in Synaesthesia: A Cognitive Neuroscience Perspective.” Nature Reviews Neuroscience 3:1 (2002): 43–52. Richardson, A., and T. Uebel (eds.). Cambridge Companion to Logical Empiricism. Cambridge: Cambridge University Press, 2007. Richter, Duncan. “Missing the Entire Point: Wittgenstein and Religion.” Religious Studies 37 (2001): 161–75. Rigal, Elisabeth. “The Duality of Wittgenstein’s Phenomenological Actuality.” In Wittgenstein and Contemporary Theories of Language, edited by P. Henry and A. Utaker, Working Papers from the Wittgenstein Archives at the University of Bergen, Bergen: WAB, 1992, pp. 62–83. Rimbaud, Arthur. Poésies. Paris: Mercure de France, 1929. Robinson, Howard. Perception. London: Routledge, 1994. Rorty, Richard. “Is Truth a Goal of Inquiry?” The Philosophical Quarterly 45:180 (1995): 281–300. —— (ed.). The Linguistic Turn: Recent Essays in Philosophical Method. Chicago: University of Chicago Press, 1967. Rundle, Bede. Wittgenstein and Contemporary Philosophy of Language. Oxford: Blackwell, 1990. Russell, Bertrand. The Analysis of Mind. Muirhead Library of Philosophy. London: Allen & Unwin, 1921. ——. Autobiography. London: Routledge, 1991. von Savigny, Eike. “No Chapter ‘on Philosophy’ in the Philosophical Investigations.” Metaphilosophy 22 (1991): 307–19. Scharlemann, R. P., and G. E. M. Ogutu (eds.). God in Language. New York: Paragon House, 1987. Schroeder, Severin. Das Privatsprachen-Argument. Paderborn, Germany: Schöningh, 1998. Schulte, Joachim. Chor und Gesetz: Wittgenstein im Kontext. Frankfurt am Main: Suhrkamp, 1990. ——. “Chor und Gesetz. Zur ‘morphologische Methode’ bei Goethe und Wittgenstein.” In Chor und Gesetz: Wittgenstein im Kontext, Frankfurt am Main: Suhrkamp, 1990.

306

Bibliography

——. “The Builders’ Language—The Opening Sections.” In Wittgenstein at Work: Method in the Philosophical Investigations, edited by Erich Ammereller and Eugen Fischer, London: Routledge, 2004, pp. 22–41. Searle, John R. Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press, 1969. ——. Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press, 1983. Sellars, Wilfrid. Science, Perception and Reality. London: Routledge & Kegan Paul, 1963. ——. “Empiricism and the Philosophy of Mind.” In Science, Perception and Reality, London: Routledge & Kegan Paul, 1963. ——. Empiricism and the Philosophy of Mind. Cambridge, MA: Harvard University Press, 1997. Shields, Philip R. Logic and Sin in the Writings of Ludwig Wittgenstein. Chicago; London: University of Chicago Press, 1993. Shoemaker, Sydney. “The Inverted Spectrum.” Journal of Philosophy 79:7 (1982): 357–81. Sluga, Hans D., and David G. Stern (eds.). The Cambridge Companion to Wittgenstein. Cambridge Companions to Philosophy series, Cambridge: Cambridge University Press, 1996. Smith, B. “Language, Conventionality of.” In Routledge Encyclopedia of Philosophy, edited by Edward Craig, London: Routledge, 1998. Spiegelberg, Herbert. The Context of the Phenomenological Movement. Vol. 80, Phenomenologica. The Hague: Nijhoff, 1981. ——. “The Puzzle of Wittgenstein’s Phänomenologie (1929–?).” In The Context of the Phenomenological Movement, The Hague: Nijhoff, 1981. Stainton, Robert J. Philosophical Perspectives on Language. Peterborough, ON: Broadview Press, 1996. ——. “Comparing Qualia across Persons.” Philosophical Topics 26 (2000): 385–405. Stalnaker, Robert. “Comparing Qualia across Persons” Philosophical Topics 26 (2000): 385–405. Stern, David G. Wittgenstein on Mind and Language. New York; Oxford: Oxford University Press, 1995. ——. Wittgenstein’s Philosophical Investigations: An Introduction. Cambridge: Cambridge University Press, 2004. ——. “Wittgenstein Versus Carnap on Physicalism: A Reassessment.” In Cambridge Companion to Logical Empiricism, edited by A. Richardson and T. Uebel, Cambridge: Cambridge University Press, 2007. ——. “Critical Review of Wittgenstein 2000.” European Journal of Philosophy (forthcoming). Stoecker, Ralf (ed.). Reflecting Davidson: Donald Davidson Responding to an International Forum of Philosophers. Berlin: W. de Gruyter, 1993. Stokhof, Martin. World and Life as One: Ethics and Ontology in Wittgenstein’s Early Thought. Stanford, CA: Stanford University Press, 2002. ——. “Hand or Hammer? On Formal and Natural Languages.” Journal of Indian Philosophy (forthcoming). Stone, Tony, and Martin Davies (eds.). Mental Simulation: Evaluations and Applications. Oxford: Blackwell, 1995. Strawson, P. F. “Analysis, Science, and Metaphysics.” In The Linguistic Turn: Recent Essays in Philosophical Method, edited by Richard Rorty, Chicago: University of Chicago Press, 1967.

Bibliography

307

Tarski, Alfred J. “The Semantic Conception of Truth and the Foundations of Semantics.” Philosophy and Phenomenological Research 4 (1944): 341–75. Thomason, Richmond. “Introduction.” In Formal Philosophy: Selected Papers of Richard Montague, edited by Richmond Thomason, New Haven, CT: Yale University Press, 1974, pp. 1–71. Tilghman, B. R. But Is it Art?: The Value of Art and the Temptation of Theory. Oxford: Blackwell, 1984. ——. Wittgenstein, Ethics and Aesthetics: The View from Eternity. Swansea Studies in Philosophy. Basingstoke: Macmillan, 1991. Tomasello, Michael. Constructing a Language: A Usage-Based Theory of Language Acquisition. Cambridge, MA: Harvard University Press, 2003. Tomberlin, J. (ed.). Action Theory and Philosophy of Mind. Vol. 4, Philosophical Perspectives. Atascadero, CA: Ridgeview, 1990. Travis, Charles. The Uses of Sense: Wittgenstein’s Philosophy of Language. Oxford: Oxford University Press, 1989. Tye, Michael. Consciousness, Color, and Content. Cambridge, MA: MIT Press, 2000. Vasiliou, Iakovos. “Wittgenstein, Religious Belief, and on Certainty.” In Wittgenstein and Philosophy of Religion, edited by Robert L. Arrington and Mark Addis, London: Routledge, 2001. Verbin, N. K. “Religious Belief and Aspect Seeing.” Religious Studies 36 (2001): 1–23. ——. “Diagnosis and Therapy: Christening Simple Objects. PI 38–64.” In Wittgenstein at Work: Method in the Philosophical Investigations, edited by Erich Ammereller and Eugen Fischer, London: Routledge, 2004, pp. 42–63. Waismann, Friedrich. Logik, Sprache, Philosophie. Stuttgart: Reclam, 1976. ——. Wittgenstein and the Vienna Circle. Translated by Joachim Schulte and Brian McGuinness. Edited by Brian McGuinness. Oxford: Basil Blackwell, 1979. ——. “Notes on Talks with Wittgenstein.” The Philosophical Review 74:1 (1965): 12–16. White, Alan R. Truth. London: Macmillan, 1971. Whiting, D. “The Normativity of Meaning Defended.” Analysis 67:4 (2007): 133–40. Wilson, Brendan. Wittgenstein’s Philosophical Investigations: A Guide. Edinburgh: Edinburgh University Press, 1998. Winch, Peter. “Understanding a Primitive Society.” American Philosophical Quarterly (1964): 307–24. ——. Ethics and Action. Studies in Ethics and the Philosophy of Religion. London: Routledge & Kegan Paul, 1972. Wittgenstein, Ludwig. Tractatus Logico-Philosophicus. Translated by C. K. Ogden. International Library of Psychology, Philosophy, and Scientific Method. London: Routledge & Kegan Paul, 1922. ——. Tractatus Logico-Philosophicus. Translated by David Francis Pears and Brian McGuinness. 2nd ed. London: Routledge & Kegan Paul, 1963. ——. Briefe an Ludwig Von Ficker. Salzburg: Otto Müller, 1969. ——. Prototractatus: An Early Version of Tractatus Logico-Philosophicus. Translated by B. F. McGuinness and D. F. Pears. Edited by B. F. McGuinness, T. Nyberg, and G. H. von Wright. London: Routledge & Kegan Paul, 1971. ——. Lectures & Conversations on Aesthetics, Psychology and Religious Belief. Edited by Cyril Barrett. Oxford: Blackwell, 1978. ——. “Notes on Logic.” In Notebooks, 1914–1916, edited by G. H. von Wright and G. E. M. Anscombe, Oxford: Blackwell, 1979.

308

Bibliography

——. Wittgenstein’s Lectures, Cambridge, 1930–1932: From the Notes of John King and Desmond Lee. Edited by H. D. P. Lee. Chicago: University of Chicago Press, 1982. ——. Last Writings on the Philosophy of Psychology: The Inner and the Outer, 1949–1951. Translated by C. G. Luckhardt and Maximilian A. E. Aue. Edited by G. H. von Wright and Heikki Nyman. Vol. II. Oxford: Blackwell, 1982. ——. Geheime Tagebücher, 1914–1916. Edited by Wilhelm Baum. 3rd ed. Vienna: Turia & Kant, 1992. ——. “Remarks on Frazer’s Golden Bough.” In Philosophical Occasions, 1912–1951, edited by Alfred Nordmann and James Carl Klagge, Indianapolis, IN: Hackett, 1993. ——. Culture and Value: A Selection from the Posthumous Remains. Translated by Peter Winch. Edited by G. H. von Wright, Heikki Nyman, and Alois Pichler. 2nd ed. Oxford: Blackwell, 1998. ——. Wittgenstein’s Nachlass: Bergen Electronic Edition. Oxford: Oxford University Press, 1998. ——. “A Lecture on Ethics.” The Philosophical Review 74:1 (1965): 3– 12. ——. “II: Notes for Lectures on ‘Private Experience’ and ‘Sense Data’.” The Philosophical Review 77:3 (1968): 275–320. ——. Philosophische Grammatik. Edited by Rush Rhees. Oxford: Basil Blackwell, 1969. ——. Preliminary Studies for the “Philosophical Investigations,” Generally Known as the Blue and Brown Books. 2nd ed. Oxford: Basil Blackwell, 1972. ——. Philosophical Investigations. Translated by G. E. M. Anscombe. Edited by Rush Rhees and G. E. M Anscombe. 3rd ed. Oxford: Blackwell, 1972.

——. On Certainty. Translated by G. E. M. Anscombe and Denis Paul. Edited by G. H. von Wright and G. E. M. Anscombe. Oxford: Blackwell, 1974. ——. Philosophical Grammar. Translated by Anthony Kenny. Edited by Rush Rhees. Oxford: Blackwell, 1974. ——. Vermischte Bemerkungen: Eine Auswahl aus dem Nachlass. Edited by G. H. von Wright and Heikki Nyman. Frankfurt am Main: Suhrkamp, 1977. ——. Remarks on the Foundations of Mathematics. Translated by G. E. M. Anscombe. Edited by G. H. von Wright, Rush Rhees, and G. E. M. Anscombe. 3rd ed. Oxford: Blackwell, 1978. ——. Notebooks, 1914–1916. Edited by G. H. von Wright and G. E. M. Anscombe. 2nd ed. Oxford: Blackwell, 1979. ——. Wittgenstein’s Lectures, Cambridge, 1932–1935: From the Notes of Alice Ambrose and Margaret Macdonald. Edited by Alice Ambrose. Oxford: Basil Blackwell, 1979. ——. Remarks on the Philosophy of Psychology. Translated by G. E. M. Anscombe. Edited by G. E. M. Anscombe, G. H. von Wright, and Rush Rhees. Vol. I. Oxford: Basil Blackwell, 1980. ——. Remarks on the Philosophy of Psychology. Translated by C. V. Luckhardt and M. A. E. Aue. Edited by Heikki Nyman and G. H. von Wright. Vol. II. Oxford: Basil Blackwell, 1980. ——. Zettel. Translated by G. E. M. Anscombe. Edited by G. E. M. Anscombe and G. H. von Wright. 2nd ed. Oxford: Blackwell, 1981. ——. Last Writings on the Philosophy of Psychology: Preliminary Studies for Part II of Philosophical Investigations.

Bibliography Translated by C. G. Luckhardt and Maximilian A. E. Aue. Edited by G. H. von Wright and Heikki Nyman. Vol. I. Oxford: Blackwell, 1982. ——. Philosophical Occasions, 1912– 1951. Edited by Alfred Nordmann and James Carl Klagge. Indianapolis, IN: Hackett, 1993. ——. “The Big Typescript”. Edited by Michael Nedo. Wiener Ausgabe ed. Vol. 11. Vienna: Springer, 2000. ——. Philosophische Untersuchungen: Kritisch-Genetische Edition. Edited by Joachim Schulte, Heikki Nyman, Eike von Savigny, and G. H. von Wright. Frankfurt am Main: Suhrkamp, 2001. ——. The Big Typescript: TS 213. Translated by C. Grant Luckhardt and Maximilian A. E Aue. Edited by C. G. Luckhardt and M. A. E. Aue. Oxford: Blackwell, 2005. ——. Wittgenstein’s Lectures on the Foundations of Mathematics, Cambridge, 1939: From the Notes of R. G. Bosanquet, Norman Malcolm, Rush Rhees and Yorick Smythies. Edited by Cora Diamond. Hassocks, West Sussex: Harvester Press, 1976. Wittgenstein, Ludwig, C. K. Ogden, and Frank Plumpton Ramsey. Letters to C.K. Ogden: With Comments on the English Translation of the Tractatus Logico-Philosophicus. Edited by G. H. von Wright. Oxford: Basil Blackwell, 1973. Wittgenstein, Ludwig, and Friedrich Waismann. The Voices of Wittgenstein: The Vienna Circle. Edited by Gordon Baker. London: Routledge, 2003. Wolgast, Elizabeth. “A Religious Point of View.” Philosophical Investigations 27 (2004): 129–47. Woolley, Jacqueline D. “The Fictional Mind: Young Children’s Understanding of Imagination, Pretense, and Dreams.”

309

Developmental Review 15:2 (1995): 172–211. Wright, Crispin. Wittgenstein on the Foundations of Mathematics. London: Duckworth, 1980. Wright, G. H. von. Norm and Action: A Logical Enquiry. London: Routledge & Kegan Paul, 1963. ——. Wittgenstein. Oxford: Basil Blackwell, 1982. Yehoshua, Bar-Hillel (ed.). Proceedings of the 1964 International Congress for Logic, Methodology, and Philosophy of Science. Amsterdam: NorthHolland, 1965. Zahavi, Dan. Husserl’s Phenomenology. Stanford, CA: Stanford University Press, 2003. ——. “Expression and Empathy.” In Folk Psychology Re-assessed, edited by D. Hutto and M. Ratcliffe, Dordrecht: Springer, 2007. Zamuner, E., E. V. Di Lascio, and D. K. Levy (eds.). Lecture on Ethics: Introduction, Interpretation and Complete Text. Macerata: Quodlibet, 2007.

Index

absolute and relative interpretation 231, 232–33 abstraction 102–3 abuse, representational 98, 99, 100, 102 Acta Philosophica Fennica 284 action: essential to, or accidental 275– 76; theory of, in ethics, 213–14, 276 act of will 269 aesthetic subliming 194 Alice Through the Looking Glass, Queen in 189–90 ambiance, Wittgensteinian 254–56 Anglo-American academic tradition 1 animal communication 119 Anschluss 289 Anscombe, Elizabeth 283 Anscombe, G. E. M. 199, 205, 288, 290 anti-behaviourism 187 architecture of meaning 212–20; Tractatus Logico-Philosophicus, 211, 212 arguments 2 as-if denial 130 assertion, normativity of content (NCA) 163 attitude, belief as 259n audience, representing to/for 17 Augustine, Saint 111–12; Augustinian Picture 204; on Builders’ language game 114–18, 120; Confessions 114, 117, 180, 202; critique of 136, 150; quotation from 202, 203, 204 Austin, J. L. 42 autorepresentation 18, 20 axiomatic method, rejection by Wittgenstein 239n Baker, Gordon 204, 205 Band X 283 Band XI 284

Band XII 286 Banville, John 93 bare normativity of meaning (BNM) 162 behaviourism 95, 187 behaviour-rejecting mentalism 77 belief, normativity of content 162, 163 Belustigung (amusement) 130 Berlin, Isaiah 284 The Big Typescript 61, 111, 283, 284, 286, 287, 288 Blackburn, Simon 166 Blair, Tony 167 Block, Ned 35, 37, 49 Bloom, Paul 136 Blue Book: and Builders’ language game 111; dictating of, 290; on knowledge 198; and Socrates 182; and training 138 BNM (bare normativity of meaning) 162 Boghossian, Paul 159–60, 161, 162, 173 boundaries 147 bracketing Husserl’s method of, 62 brain activities 77 Brandom, Robert 110, 121, 122, 124, 173 Brown Book: Builders’ language game 111; dictating of 290; embodied weekdays and coloured vowels 91; Grosses Format 285; and secondary use 92; and training 138 Budd, Malcolm 103 Builders’ language game: alternative descriptions 118–21; argumentative function 112; and Augustine 114–18; philosophical significance 110–11; and primitive language games 122, 124 Bush, George W. 167

Index calculus model 112, 164, 174; lambdacalculus, 240n Carnap, Rudolf 211, 221, 222, 229 Cartesian materialism 77 causal question 89, 90 causal theories of reference 117 Cavell, Stanley 112 Chalmers, David 49 chess, and language 159, 191–92 Chierchia, Gennaro 229 child psychology 97–103 Chomsky, Noam 117, 158, 174, 223–24 codified rationalism 141, 143 coloured vowels 94 colour-exclusion proposition 47, 48 communitarian reading 156 comparisons 12–15, 40; language games as objects of comparison, 5 competence 19, 225, 230 compositionality: architecture of meaning 216, 219, 220; and competence 225; and consequences 235; and logic 230; resemblance and difference 229–30 Conant, James 238n concept game 135 concepts: compositionality principle see compositionality; first-order 23; as functions 10–11, 12, 14, 16; and language games 121–24; notion of ‘concept’ 12; pain 39, 87; phenomenal 49–56; saying/showing distinction 216–17 Confessions (Augustine) 114, 117, 180, 202 consciousness 80 consequences 235–38; morality, consequentializing, 276–77 content: and meaning 161–62; normativity of, 160, 163 context principle 7, 115, 231 contingency 214–15 contradictions 220 conventions 174 Cornell microfilms 283, 287 correctness conditions 166, 170–71 Coughlan, Michael J. 254–55 Cresswell, Max 211, 227, 243n Damasio, Antonio, 72 darkness 9–15, 17, 19 Davidson, Donald: architecture of meaning 212, 229; bare normativity of meaning 162; Builders’ language

311

game 119; formal semantics 211, 227, 231; individualist conception of language 174; linguistic behaviour 160; publications by 223; and Tractatus Logico-Philosophicus 230 decomposition 11 “Der Gedanke” 8 de Swart, Henriëtte 229 detectable spectrum inversion 36, 37, 38 de-transcendentalization 248 Diamond, Cora 113, 238n, 290 die gemeinsame menschliche Handlungsweise (common way of acting of mankind) 119 distillation process 194–95, 199 doll-play 101, 102 double-knowledge 97 Dowty, David 229 Drury, Maurice 61 Dummett, Michael 67, 111, 163 Effability Requirement (ER) 33, 37, 40, 41 elementary propositions 43, 44 elements: of moves 7; in propositions 7–9 elimination rules 5, 13 embodied weekdays 91, 92, 96 emotion 72 emotivism 259n empiricist learning model 137–41 empty names 129 English, speaking 12–13 environment, judging, 8 epistemological naturalism 157, 158, 159 epoché, phenomenological 65–66 ER (Effability Requirement) 33, 37, 40, 41 erhaben 194 error, risk of 8, 9 “Erweiterung” 123, 127 ethics: ethical attitudes 271–73; and other types of theorizing 213; theory of action in 213–14, 276; transcendental nature of 215 expectations 151 experience of meaning 88, 89, 94 explanatory gap 49 expressivism 259n extending of language games 127–30, 188; Builders’ language games 123; Philosophical Investigations §2, 188; and primitive language games 111, 112–13

312

Index

facial expressions 75 facts, valuelessness of 263–64, 267–68 fate 269 Fernandes, Uma 166 fictive mental states 102 fideism 247, 250 film-developing game 9–15, 17, 19 flat sounds 93, 94 Floyd, Juliet 127 F-ness 37, 139, 142, 145, 152 Fodor, Jerry 136 Fogelin, Robert 127, 206 formalization 225 forms of life, and religion 248–51, 252 Frege, Gottlob 142, 211, 229; context principle 7, 115; limitations of 10–12; logic 1; sense and meaning 160, 161, 221; truth 15–17; and Wittgenstein 5, 6 games: chess 159, 191–92; defined 46, 144; language see language games; teacher/pupil relationship 140–42; and words, 187 Gamut, L. T. F. 229 Geach, Peter 103 German language 194 German, T. P. 101, 102 Glock, Hans-Johann 204 Gödel 226 Goethe, Johann Wolfgang von 104n, 110 Goldfarb, Warren 126, 154n, 202 Goldman, A. I. 78 good and evil 267, 268, 269 grammar: and language 157; phenomenology as 46–49; prejudice, grammatical 92; statements of 47; and theory of meaning 67 grammatical rules: constitutive 171, 172, 173; as conventions 174; and factual claims 49; and language games 55; and meaning 164–65, 168; regulative 171, 172, 173; semantic and logical 159; syntactic and semantic 166, 219; and words 166; see also rules Grosses Format 284, 285 Habituation 152 Hacker, Peter 67, 68, 203, 204, 205 Heidegger, Martin 69 Hintikka, J. 54, 62, 203, 205, 222 Hintikka, M. 54, 62, 203, 205 Horwich, Paul 113

Hubbard, E. M. 95 Hussein, Saddam 167 Husserl, Edmund: on introspection 68– 69; on logic, 1 on phenomenology 62, 64, 65 Ideas (Husserl) 65 “if-feeling” 104n Ignorance 146–47 illusions, optical 90 immediate experience 34, 65 immersion in a practice 142, 143, 147, 148 imprecise boundaries 147 indexable expressions 50 ineffability 215, 216, 231–32, 234 inside/outside divide 69 Insight and Illusion (Hacker) 67 instinctive actions 249 intensional referentialism 217, 230, 231, 235 intentionality 71 inter-subjectivity, phenomenology 70, 73–82 introduction rules 5 introspection 57n, 68–69 Jackson, Frank 49–50 James, William 72, 88–89, 104n joining in, learning by 149, 155n joy 72 Kanger, Stig 222 Kant, Immanuel 265 Kaplan, David 222 keelhaul 19, 20 Kenny, Anthony 205, 283, 284–85, 287; on Rhees 285–86 Kierkegaard, Søren 248 Kleines Format 284, 285 knowledge: Blue Book on 198; doubleknowledge 97; of meaning 144; philosophical definition 182; practical 143, 148; theoretical 144, 148 Konvention 174 Kripke, Saul A. 71, 135, 156, 222; publications by 156, 158; on “straight” solutions and “sceptical” solutions 185 Lakatos, Imre 291 language: early learning 116, 117, 121, 137, 152–53, 154n; external and internal 117; historical background

Index 157–59; individualist conception of 174; and meaning 117; metaphysical uses 186; natural language semantics, pioneering 223–28; ordinary see ordinary language; phenomenological see phenomenological language; philosophical uses of 193; private 33– 34, 40, 287; as a rule-guided activity 157; of thought 116, 117; see also language games; words language games: Builders’ see Builders’ language game; colour 54; and concepts 121–24; error, risk of 8, 9; extending see extending of language games; misuse of language 19; moves in 7 11; as objects of comparison 5; Philosophical Investigations 5, 110; playing 13; primary and secondary 54; primitive see primitive language games; and religion 248–51; rulefollowing, 165 language-in-use 68 Last Writings on the Philosophy of Psychology 283, 287, 291 later Wittgenstein 237; architecture of meaning 211; impact of 1; phenomenology 67–73; see also Blue Book; Brown Book learning: empiricist model 137–41; by joining in 149, 155n; of language 116, 117, 121, 137, 152–53, 154n; rationalist model 136, 137, 139, 141, 142, 143; and training 150–51; words, 135–3 Lectures on Religious Belief 249 Leslie, A. M., 78, 97, 100, 101 102 Lewis, David 211, 227 Lillard, A. 98, 100–101 Lindqvist, Sven 272, 273 linguistic frameworks 222 linguistic mode, of Wittgenstein 66 linguistic turn 2, 227 logic 1, 24, 25; of complex expressions 220; and compositionality 230; and contingency 215; and language 157; logical form 63, 214, 216, 217; transcendental nature of 217 Logical Investigations inside/outside divide 69 logical roles, categories 10 main tasks 6 make-believe 87, 96, 97, 99; doll-play, 102

313

Malcolm, Norman 121, 203, 253, 287, 291; publications by, 257 material mode, of Husserl 66 mathematics, philosophy of 156 McConnell-Ginet, Sally 229 McCulloch, Gregory 69, 77 McCune-Nicolich, L. 97 McDowell, John 154n, 156 McGinn, Marie 68, 81 meaning: architecture of 212–20; communal use 175; and content 161– 62; experience of 88, 89, 94; and grammatical rules 164–65, 168; knowledge of 144; and language 117; meaningful and meaningless, distinction 215; normativity of 156, 157, 162, 163, 168–73; picture theory of 214, 215; Platonic concept 160; proto-philosophical accounts of 186; referential theories of, critique 158, 180–82; and representation 18–19; and sense 160; skepticism about 142; terminology 91; theory of, and grammar 67; in Tractatus LogicoPhilosophicus 215; and truth 166; and words 46, 135, 149, 161, 183 Meaning and Necessity (Carnap) 221 mental strain 92 Meredith, Carew 222 Merleau-Ponty, Maurice 66, 69, 70, 71– 72 meta-philosophical statements 113 metaphors 92, 94–95, 100; perceptual 87; pretend-play 96 metaphysics 60, 186 meta-representations 97, 98, 100 middle period of Wittgenstein’s writings 34–35, 41 mind, philosophy of 60, 61 minds of others, understanding 60–82 modified rationalism 143 Monk, Ray 62–63 Montague, Richard 211, 212, 222; natural language semantics 227, 229; universal grammar of 223, 232 morality: accidental or essential to action 275–76; agency, without 262– 79; consequentializing 276–77; ethical attitudes 271–73; Notebooks 267; recommended actions 274–75; Tractatus Logico-Philosophicus 263– 66; worry 273–74, 279 Morden, Max 93 morphology 110

314

Index

moves: elements of 7; in language games 11 Mulder V 285 Nachlass, of Wittgenstein 2, 283, 288, 289 naming: act of 189; empty names, 129; and labelling 189; language games 11; of objects and concepts 10 188–89; propositions 12; sublime names 183, 194, 195–201 nativism, 138–39, 145 naturalism 157, 158 natural language semantics 223–28 NCA (normativity of content: assertion) 163 NCB (normativity of content: belief) 163 Nedo, Michael 288 neuroscience, and semantics 236 new use/new surroundings 93–94 New Wittgensteinians 213 Nielson, Kai 256 Noë, R. A. 46 normativity: bare 162; claim of, development following Wittgenstein 159–62; of content 160, 163; defined 156; of meaning 156, 157, 162, 163, 168–73; mentalistic and linguistic versions 161, 162; prescriptive 163, 171–73; rule-based 162, 168–71; semantic 162–63 norm, defined 160 norm-propositions 169 Notebooks 1914–1916: logic 217; morality 267; valuelessness of facts 267–68 Notes for Lectures on Private Experience and Sense Data 35, 40 object language/metalanguage distinction 230, 232, 234 objects: of comparison, language games as 5; naming of 10; physical 34 “Occam’s Razor” (Tractatus LogicoPhilosophicus) 7 On Certainty 112, 128, 255, 291; expression of ideas in 283 ontological supernaturalism 157, 158, 159 open-endedness 147 optical illusion 90 ordinary language: phenomenological language distinguished 41–43, 44, 45, 47–48; subject-predicate form 44

ostension, paradox of 183, 184, 187–95, 203 pain: and inanimate objects 102; lack of display of 38, 39, 47; and language games 54; primitive expression of, and experience of 94; secondary concept 87, 99 paintings, as representations 16 Papineau, David 50, 51–52, 53 paradox of ostension 183, 184, 187–95, 203 Pears, David 195 phenomenal and noumenal subjects, distinguished 265 phenomenal character 41 phenomenal concepts 49–56 phenomenal principle 34, 35 phenomenological language: abandonment of goal of 45; ordinary language distinguished 41–43, 44, 45, 47–48; Philosophical Remarks 35; see also phenomenology phenomenology: continental 63; contribution of Wittgenstein 60–82; definitions/descriptions 61–66; and experience 46; as grammar 46–49; and inter-subjectivity 70, 73–82; and later Wittgenstein 67–73; middle period work as 34; and qualia 69; and science 70 Phenomenology of Perception (MerleauPonty) 70 Phillips, D. Z. 253, 254 philosophical analysis 43, 60 Philosophical Grammar, 46; extending of languages 113, 127 Philosophical Investigations: clarity of 113; concept, notion of 12; critique of referential theories of meaning and ostension paradox (§§26–48) 179–206; and description 68; as dialogue 184– 87; and Frege, 3; language games 5, 110; and meaning 88; “method” of §2 180–82; on noticing 74; opening move in 5–7, 17; paradox of ostension 183, 184, 187–95, 203; Part II 287, 291; rule-following 18, 179, 180; on sensations 41, 72–73; teacher/ pupil relationship in 140; three-stage argument scheme 180–83, 184, 187; on vowels 87 Philosophical Remarks 61, 62, 71; colour octahedrom introduced in 44; phenomenological language 35, 42

Index Philosophische Grammatik 283, 285, 286, 289; Part I, 284 physical objects, and sense data 34 Pichler, Alois 285 pictures of objects, imagining 193 picture theory of meaning 214, 215 Plato 3, 111, 160, 183, 197 PNM (prescriptive normativity of meaning) 163; objections to 171–73 powerlessness of will 266, 268–69 practical knowledge 143, 148 practice 142, 143, 147, 148 pre-linguistic behaviour 79, 121 prescriptive normativity of meaning (PNM) 163; objections to 171–73 Pre-Socratics 3 pretend-play 87, 96, 97, 99, 101 pretend representations 97 primitive language games 109–30; as objects of comparison 109; philosophical significance 111–13 principle of all principles (Husserl) 64, 66 Prior, Arthur 222 private language, Wittgenstein on 33– 34, 40, 287 procedural programming languages 226 prohibitions 172 proper names, vacuous 129, 130 propositions: colour-exclusion 47, 48; elementary 43, 44; elements in 7–9; essential features of 8; grammatical 169; and moves in language games 9; and normativity 161; and thoughts 6 proto-philosophical accounts of meaning 186 Prototractatus 287 public language 34 punishment and reward 270–71, 272 Putnam, Hilary 257 qualia (intrinsic properties of experience): defined 57n; experiences, qualitative character of 48; and phenomenology 69; proponents of 48, 51, 56; skepticism regarding existence of 33, 34, 35, 37, 41, 47, 49 Quine, W. V. 118, 119, 136, 137, 237 Ramachandran, V. S. 95 rationalism/rationalist model of learning 136, 137, 139, 141, 142, 143; concept acquisition 154n Read, Rupert 238n

315

reasoning 149–50 reductio ad absurdum 112, 137 reference-fixing procedures 117 referential theories of meaning, critique 158 religion: attitude, belief as 259n; as form of life and language game 248– 51, 252; general Wittgensteinian ambiance 254–56; non-transcendental 256–58; religious belief/faith 245; Wittgenstein on 246–48 Remarks of the Foundations of Mathematics 122–23 Remarks on Colour 67, 71, 283, 287, 291 Remarks on Frazer’s Golden Bough 249 Remarks on the Foundations of Mathematics and colour exclusion proposition 47 Remarks on the Philosophy of Psychology: primitive language games 109; secondary sense 87; thinking 124, 125, 126 representation 3–28; audience, to or for 17; autorepresentation 18, 20; comparisons 12–15; concepts see concepts; linguistic 17; metarepresentations 97, 98, 100; morphology 110; particular and general 22–28; pretend 97; “realityoriented” 97 representational abuse 98, 99, 100, 102 resemblance and difference 228–35 responsibility 276 restricted spectrum inversion, defense of 37–41 Rhees, Rush 110, 121, 122, 199, 253, 284, 287; editing by 283, 285 Ricketts, Thomas 238n riddles 262 RNM (rule-based normativity of meaning) 162, 163; objections to 168–71 Robinson, Howard 34 romantic sublime 194 Rorty, Richard 113 Rothhaupt, Josef 290 rule-based normativity of meaning (RNM) 162, 163; objections to 168–71 rule-following Philosophical Investigations 18, 179, 180 rules: elimination 5, 13; grammatical see grammatical rules; introduction 5; linguistic 164; semantic 165–68 Russell, Bertrand 154n; on logic 1

316

Index

saliences 151, 152 saying/showing distinction 216–17 “sceptical” solutions 185 sceptical questions 99 Schlick, Moritz 42, 61, 221; publications by 285 Schulte, Joachim 110, 125 science 70 The Sea (Banville) 93 seeing, 72, 74 self, unworldliness of 264–65, 268 Sellars, Wilfrid 70, 173, 237 semantic and syntactic, distinguished 166, 219 semantic normativity 162–63 semantics: and cognitive neuroscience 236; development 239n; formal, and Tractatus Logico-Philosophicus 211– 38; history 220; lexical and structural 233; methodology 236; natural language, pioneering 223–28; of natural languages 234; semantic rules, introducing “by mistake” 165–68 sensations, words referring to 41, 103 sense, and meaning 160 sense data, and physical objects 34 sentences 161, 217–18; complex, 219, 230; elementary 218, 219; meaningfulness of 219; state descriptions 222; and truth functions 219 sermo exterior (acquisition of first language) 116, 117, 121 sermo interior (language of thought) 116, 117 set-theory 227 significance 270 simples, names signifying 199 simulation theory of mind 77–78, 79, 80, 81 Smythies, Yorick 287 Socrates/Socratic philosophy 182, 183, 196, 197, 200 “Some Remarks on Logical Form” 63 Sophists 3 speaking, and thinking 117–18 spectrum inversion: detectable 36, 37, 38; and Papineau, 53; restricted, defence of 37–41; widespread 33, 36, 39, 47; Wittgenstein on possibility of 35–37 Spengler, Oswald 110 Springer Verlag 288

state descriptions 222 states of affairs 221, 269, 270, 273, 274, 275 Stern, David G. 45 Stonborough, Margarethe 290 “straight” solutions 185 stratified rationalism 143 Strawson, P. F. 60 structure 8 sublime names 183, 194, 195–201 sublimieren 194 sub-tasks 6, 7 such-and-such 6, 8, 11, 13, 20, 21 supernaturalist 157, 158, 159 synesthesia/synesthetes 95–96 Tarski, Alfred J. 211, 223, 229 tautologies 220 teacher/pupil relationship 140–49 Theaetetus 182, 196, 198 theoretical knowledge 144, 148 theory-theory of mind 77, 78, 79, 80, 81 thinking 18, 20, 124–27; and speaking 117–18 Thomason, Richmond 243n, 244n thoughts 6; chain of 50; generality of 22; introspective 51; language of (sermo interior) 116, 117 three-dimensionality 72 three-stage argument scheme, Philosophical Investigations 180–83, 184, 187 Tilghman, B. R. 92, 103 Tractatus Logico-Philosophicus 1, 8; architecture of meaning 211, 212; and Augustine 114; and Carnap 221; and formal semantics 211–38; and Frege 3; linguistic rules in 164; on logical form 63; and naming 203; “Occam’s Razor” 7; ontology of 218; and philosophical analysis 43; powerlessness of will 266; and subliming names 196; universalism of 216; unworldliness of self 264–65; valuelessness of facts 263–64 training 138, 141, 142; and learning 150–51 truth 15–17, 166 truth-bearers 23 truth-conditions 166 truth-values 10, 12, 16, 219 type B materialism 49, 50 Übereinkunft 174 universalism 216, 231, 232, 235, 236

Index unworldliness of self 264–65, 268 useless applications 39 valuelessness of facts 263–64, 267–68 Verfremdung 129 Vienna Circle 61, 221; Archive, Haarlem 285 “view from nowhere” 181 visual impressions 55, 72, 74, 83n vocations 279 voice 112, 185 von Ficker, Ludwig 213 von Wright, G. H. 169, 284, 285, 287 vowels 87, 95; coloured 91, 94 Waismann, Friedrich 42, 157–58, 221 why-question 89, 91 widespread spectrum inversion 33, 36, 39, 47 Wiener Ausgabe 288, 289 will, powerlessness of 266, 268–69 Winch, Peter 253, 257

317

Wittgenstein, Paul (brother of Ludwig) 262, 263, 278 Wittgenstein: A Religious Point of View? (Malcolm) 257 Wittgenstein on Rules and Private Languages (Kripke) 156, 158 Wittgenstein: The Philosopher and His Works (Kenny) 285, 286 words: ambiguous 90; familiar, physiognomy of 89; and games 187; instruction in meaning of 135; in language games 114; and meaning 46, 135, 149, 161, 183; primary and secondary use 91, 92–93, 95, 96, 100, 103, 104n; sensations, referring to 41; teaching and learning 135–53; and visual impressions 55; see also language worry, 273–74, 279 Wren Library of Trinity College 283, 284, 288 Wright, Crispin 156