¹ Donald Davidson, ‘On Saying That’, reprinted in his Inquiries into Truth and Interpretation (Oxford: Oxford University Press, 1984), 94.

² See esp. 108.

³ This way of putting it is an allusion to Wittgenstein, Ludwig, Philosophical Investigations, trans. Anscombe, G. E. M. (Oxford: Basil Blackwell, 1974)Google Scholar, Pt. I, §§7 and 22.

⁴ For endorsement of the Governing Picture, along with many interesting references, see Standish, Paul, Beyond the Self: Wittgenstein, Heidegger and the Limits of Language (Aldershot: Avebury, 1992),Google Scholar Ch. 2: see esp. the quotation from Taylor, Charles, Human Agency and Language (Cambridge: Cambridge University Press, 1985),CrossRefGoogle Scholar 231, which Standish gives on 74.

⁵ Wittgenstein, Ludwig, Tractatus Logico-Philosophicus, trans. Pears, D. F. and McGuiness, B. F. (London: Routledge & Kegan Paul, 1961)Google Scholar, 4.002. Cf. Donald Davidson, ‘Criticism, Comment, and Defence’, reprinted in his Essays on Actions and Events (Oxford: Oxford University Press, 1980), esp. 123 and §H; and Donald Davidson, ‘What Metaphors Mean’, reprinted in his op. cit., note 1.

⁶ There is something of this line of thought in Derrida: see e.g. Derrida, Jacques, ‘Signature, Event, Context’, in his Margins of Philosophy, trans. Bass, Alan (Brighton: Harvester Press, 1982)Google Scholar.

⁷ Murdoch, Iris, Metaphysics as a Guide to Morals (Harmondsworth: Penguin, 1993), 188–189Google Scholar.

⁸ See e.g. Wittgenstein, op. cit. note 3, Pt. I, §§23– 24, and 224, third and fourth paragraphs from the bottom. Concerning the point about sentences with the same surface grammar being used to make quite different moves, cf. the volte-face between the first and second editions of Hacker's Insight and Illusion, respectively Hacker, P. M. S., Insight and Illusion: Wittgenstein on Philosophy and the Metaphysics of Experience (Oxford: Oxford University Press, 1972)Google Scholar and Hacker, P. M. S., Insight and Illusion: Themes in the Philosophy of Wittgenstein (Oxford: Oxford University Press, 1986)Google Scholar; see in particular ix of the latter.

⁹ See e.g. Lovibond, Sabina, Realism and Imagination in Ethics (Oxford: Basil Blackwell, 1983)Google Scholar, §§6 ff. For a backlash, somewhat more in keeping with my reading, see Blackburn, Simon, ‘Wittgenstein's Irrealism’, in Wittgenstein: Eine Neuerberwehrung, Brandl, Johannes L. and Haller, Rudolf (eds.) (Vienna: Holder-Richler-Temsky, 1990).Google Scholar A third possible position, of course, is to see in this divergence of interpretation fuel for cynicism, either about Wittgensteinian exegesis or indeed about Wittgenstein.

¹⁰ In support of the opposed reading see e.g. Wittgenstein, op. cit., note 3, Pt I, §402—cited by Lovibond in her op. cit. note 9, 26—and Wittgenstein, Ludwig, The Blue and Brown Books: Preliminary Studies for the ‘Philosophical Investigations’; (Oxford: Basil Blackwell), 25Google Scholar.

¹¹ E.g. Wittgenstein, op. cit., note 10, 24–25.

¹² The locus classicus is W. V. Quine, ‘Two Dogmas of Empiricism’, reprinted in his From a Logical Point of View: Logico-Philosophical Essays (New York: Harper & Row, 1961).

¹³ See e.g. Wittgenstein, Ludwig, Philosophical Grammar, ed. Rhees, Rush and trans. Kenny, Anthony (Oxford: Basil Blackwell, 1974), 54Google Scholar.

¹⁴ See e.g. Wittgenstein, Ludwig, On Certainty, ed. Anscombe, G. E. M. and von Wright, G. H. and trans. Paul, Denis and Anscombe, G. E. M. (Oxford: Basil Blackwell, 1969), §§96–99Google Scholar. Indeed such thoughts are present in the very place where he draws the distinction between giving criteria and giving symptoms: see again his op. cit., note 10, 25.

¹⁵ Op. cit., note 12.

¹⁶ Michael Dummett, ‘The Significance of Quine's Indeterminacy Thesis’, reprinted in his Truth and Other Enigmas (London: Duckworth, 1978), 375. Note that, because Dummett couches his discussion in terms of sentences rather than in terms of their utterances, we must presume that attention is being restricted to sentences that can be classified as true or false without reference to individual utterances of them—sentences of the kind that Quine elsewhere calls ‘eternal’ (W. V. Quine, Word and Object (Cambridge: The MIT Press, 1960), §40).

¹⁷ Some of what I am about to say is anticipated by Dummett later in his discussion: see 411 ff. Cf. also Carnap, Rudolf, ‘Quine on Analyticity’, in Dear Carnap, Dear Van: The Quine-Carnap Correspondence and Related Work, Creath, R. (ed.) (Berkeley: The University of California Press, 1990)Google Scholar.

¹⁸ So too, on the traditional conception, given any synthetic sentence which is held true, and given any recalcitrant experience, the sentence can continue to be held true vis-à-vis that experience, in the way that Quine envisages; but again, sometimes, only by undergoing a change of meaning. If meaning is kept fixed, then there will never be any choice about whether or not a sentence which is held true should be rejected vis-à-vis any possible recalcitrant experience: if the sentence is analytic, it never should be; if the sentence is synthetic, it sometimes should be. This is why, on the traditional conception, the distinction is both exclusive and exhaustive, a fact that Dummett's discussion somewhat obscures.

¹⁹ I have borrowed material in this paragraph from Moore, A. W., ‘The Underdetermination/Indeterminacy Distinction and the Analytic/ Synthetic Distinction’, in Erkenntnis 46 (1997), 8–9.CrossRefGoogle Scholar Note that in his op. cit., note 12, Quine does describe the word ‘analytic’ as ‘un-understood’ (34) and later relates the idea of analyticity to another idea which he describes as ‘nonsense, and the root of much nonsense,’ (42): I am grateful to Alexander George for drawing my attention to these two passages.

²⁰ Cf. W. V. Quine, ‘Ontological Relativity’, reprinted in his Ontological Relativity and Other Essays (New York: Columbia University Press, 1969), 27.

²¹ Cf. Quine's concession in his op. cit., note 12, 25–26, that we sometimes stipulatively define novel terms, and that, whenever we do, ‘we have a really transparent case of synonymy’: I am grateful to Timothy Williamson for drawing my attention to this passage. (Not that I want to claim that Quine would be happy with everything I say in this section. Stating rules of representation covers far more than stipulatively defining novel terms. When I talk about ‘what Quine is really attacking’, I do not mean this exegetically. I am making a point about the force of his arguments.)

²² Will it also mean that we no longer have our current concept of an aunt? Or will it mean that our current concept of an aunt has undergone a change?—When, some time in the fifteenth century, the Pawn was first allowed to move forward two squares in chess, did this create a new game, what we now call chess? Or did that very game undergo a change?—It is relatively clear what is going on in these cases. Say what you will as long as you do nothing to threaten such clarity.

²³ Cf. again Wittgenstein, op. cit., note 14, §§96–99—esp. §98, where he denies that his famous river-bed analogy makes logic ‘an empirical science’. And cf. Wittgenstein, op. cit., note 3, Pt. I, §354. For some extremely helpful material on the ideas in this section see various essays in Arrington, Robert L. and Glock, Hans-Johann (eds.), Wittgenstein and Quine (London: Routledge, 1996)CrossRefGoogle Scholar, esp. P. M. S. Hacker, ‘Wittgenstein and Quine: Proximity at Great Distance’ and Christopher Hookway, ‘Perspicuous Representations’ the former of which is an abbreviated version of the even more helpful Hacker, P. M. S., Wittgenstein's Place in Twentieth- Century Analytic Philosophy (Oxford: Basil Blackwell, 1996), Ch. 7Google Scholar.

²⁴ Cf. above, note 16: precisely what an ‘eternal’ sentence is is a sentence that equips us to make this move.

²⁵ Cf. Wittgenstein, op. cit., note 3, Pt. I, §410.

²⁶ Newton, Isaac, Mathematical Principles of Natural Philosophy, trans. Motte, Andrew and Cajori, Florian (Berkeley: The University of Californian Press, 1947)Google Scholar, Bk. I, Law I.

²⁷ See Einstein, Albert, Relativity: The Special and the General Theory, trans. Lawson, Robert W. (London: Methuen, 1960), 11, 61, 71–72 and 99.Google Scholar For more on the aim to transcend perspective (of this and any other kind), and for an argument that this aim is achievable, see Moore, A.W., Points of View (Oxford: Oxford University Press, 1997)Google Scholar, esp. Chs. One-Four.

²⁸ Cf. above, note 22: even if allowing the Pawn to move forward two squares did create a new game, it would be unsatisfactory just to say that chess was invented in the fifteenth century.

²⁹ Another notable example would be a use of the sentence ‘Aunts are female’ to state a rule of representation. On the grammatical characterization, this would involve using both ‘aunts’ and ‘female’. On the pragmatic characterization, it would involve mentioning them both (roughly— though it is a very nice question how roughly (cf. Baker, G. P and Hacker, P. M. S., Wittgenstein: Rules, Grammar and Necessity, Volume 2 of An Analytical Commentary on the Philosophical Investigations (Oxford: Basil Blackwell, 1985)Google Scholar, Ch. VI, §3(ii))).

³⁰ This is a main theme of Moore, A. W., ‘How Significant is the Use/Mention Distinction?’, in Analysis 46 (1986)CrossRefGoogle Scholar, which contains a fuller discussion of these issues.

³¹ Gödel, Kurt, ‘On Formally Undecidable Propositions of Principia Mathematica and Related Systems I’, trans. van Heijenoort, Jean, in From Frege to Gödel: A Source Book in Mathematical Logic, 1879–1931, Jean van Heijenoort (ed.) (Cambridge: Harvard University Press, 1967)Google Scholar, e.g. 601.See also Quine, W. V., ‘Gödel's Theorem’, in his Quiddities: An Intermittently Philosophical Dictionary (Harmondsworth: Penguin, 1990), 84Google Scholar. Indeed I think the grammatical characterization allows us to go further and say that one of the great achievements, in turn, of analytic philosophy is to have made due appreciation of the distinction possible. Even the most rigorous writings in mathematics often flout it: cf. W. V. Quine, ‘Use Versus Mention’, in his Ibid., 232. (For an indication of the significance of the distinction on the pragmatic characterization, see Moore, op. cit., note 30.)

³² It is interesting at this point to consider Derrida again. His attitude to the distinction between using expressions and mentioning them is a curiously ambivalent one. He often seems to make play with words, and indeed to make philosophical points, precisely by flouting the distinction: see e.g. Derrida, op. cit., note 6, 320–321. (This is a complaint that Searle levels against him in his commentary on this essay, Searle, J. R., ‘Reiterating the Differences: A Reply to Derrida’, in Glyph 1 (1977), 203.)Google Scholar Furthermore there are places where Derrida seems to be overtly hostile to the distinction. Cf. his comment, ‘I try to place myself at a certain point at which⃛ the thing signified is no longer easily separable from the signifier,’ quoted in Wood, David and Bernasconi, Robert (eds.), Derrida and ‘Différance’ (Evanston: Northwestern University Press, 1988), 88Google Scholar. Cf. also Derrida, Jacques, The Post Card: From Socrates to Freud and Beyond, trans. Bass, Alan (Chicago: The University of Chicago Press, 1987), 97 ffGoogle Scholar. However, in his reply to Searle—Jacques Derrida, ‘Limited Inc a b c⃛’, trans. S. Weber and reprinted in his Limited Inc, ed. G. Graff (Evanston: Northwestern University Press, 1988)— he writes (81), ‘I agree that [the confusion of “use” and “mention”] might very well be [a radical evil].’

³³ For a tiny sample, see: Plato, , Theaetetus, trans. Levett, M. J. and Burnyeat, Myles and ed. Burnyeat, Myles (Indianapolis: Hackett, 1990)Google Scholar, 183a4–b8; Nietzsche, Friedrich, The Will to Power, trans. Kaufmann, Walter and Hollingdale, R. J. and ed. Kaufmann, Walter (New York: Random House, 1967), §616Google Scholar; and Derrida, Jacques, Dissemination, trans. Johnson, B. (London: Athlone Press, 1981), 168Google Scholar.

³⁴ Travis, Charles, ‘Sublunary Intuitionism’, in Grazer Philosophische Studien 55Google Scholar, a special issue entitled ‘New Essays on the Philosophy of Michael Dummett’, Johannes L. Brandl and Peter M. Sullivan (eds.), (1998), which includes an extensive (and superb) discussion of such cases.

³⁵ J. L. Austin, ‘Truth’, reprinted in his Philosophical Papers, eds. J. O. Urmson and G. J. Warnock (Oxford: Oxford University Press, 1970), 128.

³⁶ See e.g. Michael Dummett, ‘Truth’, reprinted in his op. cit., note 16; and Dummett, Michael, ‘Realism and Anti-Realism’, reprinted in his The Seas of Language (Oxford: Oxford University Press)CrossRefGoogle Scholar.

³⁷ I do not say that anti-realism challenges the idea that something can be true though we have no procedure for telling that it is. That way of putting it (though often found in the writings of anti-realists themselves) saddles anti-realism with the paradoxical consequence that there can be no unknown truths (see Fitch, F. B., ‘A Logical Analysis of Some Value Concepts’, in Journal of Symbolic Logic 28 (1963)CrossRefGoogle Scholar). The version in the main text does not (see Melia, Joseph, ‘Anti-Realism Untouched’, in Mind 100 (1991)Google Scholar. (Our having a procedure for telling whether x is true or false is to be understood as allowing for the following possibility: that, although x is in fact true, the actual carrying out of the procedure would render it false (and would accordingly put us in a position to tell that it was false).

³⁸ For different but related reservations about the idea of our making this move (in fact of our making any move), based on suspicion of the very contrast between situations in which there are impediments to our doing so and situations in which there are not, see Derrida, op. cit., note 6. It may be that Derrida is insufficiently open to the possibility that one and the same sentence can be used in one situation to make one move and in another situation, inimical to the making of that move, to make another.

³⁹ Is this idea perhaps negotiable? Could we eliminate the category of failed attempts to make this move, by simply extending our notion of falsity and assimilating all such cases to cases in which something false has been said (cf. Dummett, ‘Truth’, op. cit., note 36)? For reasons why we may not be able to do this see again Travis, op. cit., note 34.—Note: there is also the question, on which Strawson is sometimes alleged to have equivocated in Strawson, P. F., ‘On Referring’, reprinted in Meaning and Reference, Moore, A. W. (ed.) (Oxford: Oxford University Press, 1993)Google Scholar, whether someone who utters a declarative sentence in an unsuccessful attempt to make this move thereby makes a statement that is neither true nor false or fails to make a statement at all: see e.g. G. Nerlich ‘Presupposition and Entailment’, in American Philosophical Quarterly 2. That seems to me to be an unimportant point of terminology.

⁴⁰ Cf. Wittgenstein, op. cit., note 3, Pt. I, §136.

⁴¹ Moore, op. cit., note 27, Ch. Ten, §4: see in particular the discussion of what I call ‘partial realism’, 245–249. For a fascinating discussion which in effect raises problems for this position, see Williamson, Timothy, ‘Never Say Never’, in Topoi 13 (1994)CrossRefGoogle Scholar. Williamson ends his essay by asking whether these problems constitute a reductio ad absurdum of the position with which he has been concerned in the essay, namely intuitionism (143). We can also ask whether they constitute a reductio ad absurdum of partial realism: I think not.

⁴² For a superb discussion of related issues in connection with Wittgenstein, see Hacker, the second edition of Insight and Illusion, op. cit. in note 8, Ch. XI, §4.

⁴³ For discussrion of such questions see Diamond, Cora, ‘How Old Are These Bones? Putnam, Wittgenstein and Verification’, in Proceedings of the Aristotelian Society Sup. Vol. 73Google Scholar.

⁴⁴ Pre-eminent is Williamson, TimothyVagueness (London: Routledge, 1994)Google Scholar. An excellent collection is Keefe, RosannaSmith, Peter (eds.), Vagueness: A Reader (Cambridge: The MIT Press, 1997)Google Scholar. Each of these contains extensive bibliographies.

⁴⁵ The sorites paradoxes are a family of paradoxes modelled on the following, from which their name derives (the Greek adjective ‘sorites’ corresponds to the noun ‘soros’, meaning ‘heap’). One grain of sand does not make a heap; for any number n, if n grains of sand do not make a heap, then n + 1 grains of sand do not make a heap; therefore, there is no number of grains of sand that make a heap.

⁴⁶ Austin, in Austin, J. L., How To Do Things With Words, eds. Urmson, J. O. and Sbisà, Marina (Oxford: Oxford University Press, 1975)CrossRefGoogle Scholar, introduces the notion of a ‘performative’ by means of examples: these include an utterance of ‘I do’ in the course of a marriage ceremony, and an utterance of ‘I bet you sixpence it will rain tomorrow’ (5–6). He expressly denies that performatives are either true or false (6). Others have subsequently adopted his notion, but have amended it so as to grant performatives truth: on the amended account, performatives make themselves true (see e.g. Quine, W.V., ‘On Austin's Method’, reprinted in his Theories and Things (Cambridge: Harvard University Press, 1981)Google Scholar, esp. 90). In invoking performatives here, I am obviously presupposing this amended account.

⁴⁷ Cf. R.M. Sainsbury, ‘Concepts Without Boundaries’, reprinted in Keefe and Smith (eds.), op. cit., note 44, §6.

⁴⁸ Williamson, op. cit., note 44, §7.2

⁴⁹ Ibid., 195–197.

⁵⁰ Let us take for granted that childhood is age-determined, at least to this extent. In fact, this is something of an idealization. (See further below, note 56.)

⁵¹ Ibid., 196.

⁵² Cf. again Hacker's volte-face, referred to above in note 8.

⁵³ If the person making the utterance knows that the addressee has a younger sibling, Stephen say, then he or she could also state what might be called an applied rule of representation, by saying, ‘If you are a child, then Stephen is a child.’ While the truth of the original utterance is necessary, the truth of this utterance would enjoy a sort of conditional necessity—conditional on Stephen's being the addressee's younger sibling. What I am about to say in the main text about the original rule would hold of this applied rule too.

⁵⁴ It might be said, in defence of Williamson, that he expressly of forestalls this counter-argument by talking about ‘material’ conditionals (Ibid., 196), where a material conditional, unlike a conditional used to state a rule of representation, is precisely one whose truth or falsity is determined by the truth or falsity of its antecedent and consequent. But in that case it is question-begging to suppose that there are any relevant conditionals that are both material and true (or material and false).

⁵⁵ Cf. Michael Dummett, ‘Wang's Paradox’, reprinted in Keefe and Smith (eds.), op. cit., note 44, 109.

⁵⁶ Not that the age at which Ellen stops being a child need be the same as the age at which any other person stops being a child. Cf. above, note 50: it is something of an idealization to think of childhood as age-determined at all, even to an extent that precludes children having younger siblings who are not children, or adolescents reverting to childhood for that matter; but it would be a far greater idealization, with which we need have no truck, to think that, for any two people of the same age, one is a child if and only if the other is a child. Childhood is more contextual than that.

⁵⁷ Cf. what I said about ‘If you are a child, then any younger sibling of yours is a child’ in the previous section. But cf. also note 53: it is more accurate to speak here of conditional necessity than of necessity simpliciter. Ellen might never have been born. She might (God forbid) not survive her childhood. (The latter possibility is one from which I am prescinding throughout this section.)

⁵⁸ Cf. Bernard Williams, ‘Which Slopes are Slippery?’, reprinted in his Making Sense of Humanity and Other Philosophical Papers: 1982–1993 (Cambridge: Cambridge University Press, 1995), 217. Cf. also Ludwig Wittgenstein, Remarks on the Foundations of Mathematics, eds. G. H. von Wright and G. E. M. Anscombe and trans. G. E. M. Anscombe (Oxford: Basil Blackwell, 1978), Pt V, §13.

⁵⁹ In the previous section I tried to give some indication of the free play which our concept of a child allows for, alluding in particular to the way in which vague utterances that use this concept can function like performatives. Part of what it means to say that Ellen will gradually stop being a child, in semantic terms, is that this free play, applied over the course of her life, will first gradually increase, then gradually decrease. (Part of what this in turn means is that the sentence ‘Ellen is old enough for an utterance of “Ellen is a child” to function like a performative' is every bit as vague as the sentence ‘Ellen is a child’ This seems to me to rebut an argument in Williamson, op. cit., note 44, §7.3, in favour of the view that the concept of a child has some sort of cut-off point. (Williamson actually couches the argument in terms of the concept of a heap.) This argument can be broached by considering the two following situations. In the first, there are two co-operative omniscient speakers who are asked to say whether Ellen is a child at various future dates. They diverge in when they stop calling her a child. In the second, there are again two co-operative omniscient speakers who are asked to say whether Ellen is a child at various future dates, but they are also instructed to use any discretion they are allowed as conservatively as possible (that is, roughly, they are told to stop calling her a child as soon as they can). They too diverge in when they stop calling her a child. The argument turns on the thought that, whatever sense we can make of the first of these situations, the second is unintelligible. For if it is, then there must after all be some sort of cut-off point for the concept of a child. But in fact, granted what I said about the vagueness of the sentence ‘Ellen is old enough for an utterance of “Ellen is a child” to function like a performative’, the second situation is no more unintelligible than the first. Or if it is, then this is because the instruction given to the two speakers—to use any discretion they are allowed as conservatively as possible—is meant to apply even to itself, in which case I think we have reason to locate the unintelligibility of the situation, not in the fact that there is some sort of cut-off point for the concept of a child, but in the unintelligibility of the instruction.)

⁶⁰ My 1993 edition of The Chambers Dictionary defines ‘child’ as ‘a very young person (up to the age of sixteen for the purpose of some acts of parliament, under fourteen in criminal law)’.

⁶¹ This is another example of the free play that our concept of a child allows for (see above, n. 59). Cf. once again Sainsbury, op. cit., note 47, §6.

⁶² I am very grateful to Uri Henig for a number of helpful discussions, and to Peter Hacker, Oswald Hanfling and Timothy Williamson for comments on an earlier draft. I have also profited from Henig, Uri, Meaning and Negation, unpublished M. Litt. thesis, Oxford University (1994)Google Scholar.