Wiley
Unfamiliar Noises
Author(s): Richard Rorty and Mary Hesse
Source: Proceedings of the Aristotelian Society, Supplementary Volumes, Vol. 61 (1987), pp. 283-311
Published by: Wiley on behalf of The Aristotelian Society Stable URL: http://www.jstor.org/stable/4106839 Accessed: 10/07/2014 09:47
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
	1
	
STOR	
Wiley and The Aristotelian Society are collaborating with JSTOR to digitize, preserve and extend access to Proceedings of the Aristotelian Society, Supplementary Volumes.
http://www.jstor.org
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
Richard Rorty and Mary Hesse /—Richard Rorty HESSE AND DAVIDSON ON METAPHOR
We speak of one thing being like some other thing, when what we are really craving to do is to describe something that is like nothing on earth.—Vladimir Nabokov
Philosophers of science like Mary Hesse have helped us realize that metaphor is essential to scientific progress. This realization has encouraged Hesse and others to argue for 'the cognitive claims of metaphor'.1 She is concerned to give metaphorical sentences truth and reference—to find worlds for them to be about: 'imaginative symbolic worlds that have relations with natural reality other than those of predictive interest . . . Utopias, fictional exposes of the moral features of this world by caricature and other means, and all kinds of myths symbolic of our understanding of nature, society and the gods'.2 Like many other philosophers of this century (e.g., Cassirer, Whitehead, Heidegger, Gadamer, Habermas, Goodman, Putnam) she sees over-attention to the natural sciences as having distorted modern philosophy. Following Habermas, Hesse sees cognition as wider than the satisfaction of our 'technical interest' and as extending to 'the practical interest of personal communication and the emancipatory interest of critique of ideology'. In discourse which satisfies these interests, Hesse says, 'metaphor remains the necessary mode of speech'.3 So she believes that metaphor 'poses a radical challenge to contemporary philosophy' and that we need 'a revised ontology and theory of knowledge and truth' in order to do justice to metaphor as an instrument of cognition.4
1 This is the title of Hesse's article in Metaphor and Religion, ed. J. P. Van Noppen, Brussels, 1984.
2 Hesse, op. cit., p. 39.
3 ibid, p. 40.
* See ibid, p. 41.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Coiulil ions
284
I—RICHARD RORTY
I agree with Hesse that over-attention to natural science has skewed philosophy, but I do not think that her strategy is sufficiently radical to let us correct the error. For one way in which this skewing is evident is that we philosophers still tend to take 'cognition' as the highest compliment we can pay to discourse. We take 'cognitive claims' as the most important claims which can be made for a given sort of language. Were we not concerned to raise the rest of discourse to the level of science, we would not be so concerned to broaden our use of terms like 'truth', 'refers to a world' and 'meaning' so as to make them relevant to metaphor.
To correct the error of the tradition, to help ourselves see natural science as simply an instrument of prediction and control rather than as a standard-setting area of culture, we need instead to restrict the applicability of these semantical terms. We need to see that the applicability of such terms is not a measure of the cultural importance of a use of language, but merely of the extent to which language-use can be predicted and controlled on the basis of presently-available, widely-shared, theory. We should see semantical notions as applicable only to familiar and relatively uninteresting uses of words, and 'cognition' as the positivists saw it: confined to familiar and relatively uninteresting uses of language, to discourses for which there are generally accepted procedures for fixing belief. We should find other compliments to pay other sorts of discourse rather than trying to 'broaden' either semantic or epistemic notions.
In particular, we should follow Davidson rather than (as Hesse does) Black in our account of metaphor. For, by putting metaphor outside the pale of semantics, insisting that a metaphorical sentence has no meaning other than its literal one, Davidson lets us see metaphors on the model of unfamiliar events in the natural world—causes of changing beliefs and desires—rather than on the model of representations of unfamiliar worlds, worlds which are 'symbolic' rather than 'natural'. He lets us see the metaphors which make possible novel scientific theories as causes of our ability to know more about the world, rather than expressions of such knowledge. He thereby makes it possible to see other metaphors as causes of our ability to do lots of other things—e.g., be more sophisticated and interesting
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
285
people, emancipate ourselves from tradition, transvaluate our values, gain or lose religious faith—without having to interpret these latter abilities as functions of increased cognitive ability. Not the least of the advantages of Davidson's view, I shall be arguing, is that it gives us a better account of the role played in our lives by metaphorical expressions which are not sentences— scraps of poetry which send shivers down our spine, non-sentential phrases which reverberate endlessly, change our selves and our patterns of action, without ever coming to express belief or desires.
The issue between Black and Davidson has struck many people as factitious. Both philosophers insist that metaphors are unparaphrasable, and also that they are not merely ornamental. But Black thinks that a defence of these claims requires the notion of 'metaphorical meaning' and Davidson denies this. Clearly they are using 'meaning' in different ways, and so it is easy to suspect that the issue is verbal. But we can see that something important is at stake by looking at Black's claims that Davidson is 'fixated' on 'the explanatory power of standard sense' and that his account gives us 'no insight into how metaphors work'.5 These assertions show that Black and Davidson differ not just about how to use the term 'meaning' but about the ends which a theory of meaning should serve, about the point and reach of semantics.
Davidson is, indeed, 'fixated' on the explanatory power of standard sense. But this is because he thinks that semantical notions like 'meaning' have a role only within the quite narrow (though shifting) limits of regular, predictable, linguistic behaviour—the limits which mark off (temporarily) the literal use of language. In Quine's image, the realm of meaning is a relatively small 'cleared' area within the jungle of use, one whose boundaries are constantly being both extended and encroached upon.6 To say, as Davidson does, that 'metaphor
5 Max Black, 'How metaphors work: a reply to Donald Davidson' in On Metaphor, ed. Sheldon Sacks (Chicago, University of Chicago Press, 1979), pp. 189, 191.
6 See Quine, 'A Postscript on Metaphor' in On Metaphor, ed. Sacks, cited above, p. 160: 'Metaphor, or something like it, governs both the growth of language and our acquisition of it. What comes as a subsequent refinement is rather cognitive discourse itself, at its most dryly literal. The neatly worked out inner stretches of science are an open space in the tropical jungle, created by clearing tropes away'.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Gondii ions
286
I—RICHARD RORTY
belongs exclusively to the domain of use is simply to say that, because metaphors (while still alive) are unparaphrasable, they fall outside the cleared area. By contrast, if one regards meaning and use as co-extensive, one will be inclined to adopt what Hesse calls a 'network view of language'—one according to which, as she says, 'the use of a predicate in a new situation in principle shifts, however little, the meaning of every other word and sentence in the language'.8
Davidson's resistance to this 'network' view can be put in terms of an analogy with dynamics. In the case of the gravitational effects of the movements of very small and faraway particles (a phenomenon to which Hesse analogizes the insensible but continuous process of meaning-change), physicists must simply disregard insensible perturbations and concentrate on relatively conspicuous and enduring regularities. So it is with the study of language-use. The current limits of those regularities fix the current limits of the cleared area called 'meaning'.9 So where 'the explanatory power of standard sense' comes to an end, so does semantics.
If one holds a different conception of the limits of semantics and of philosophical explanation, as Black and Hesse do, this is probably because one has a different conception of the reach of philosophy. Davidson's metaphilosophical approach differs from theirs as Newton's metascientific approach to dynamics differed from Leibniz's; the one is an approach which describes regularities without venturing on hypotheses about the under-
7 Davidson, 'What Metaphors Mean,' Inquiries into Truth and Interpretation (Oxford, Clarendon Press, 1984), p. 247.
8 Hesse, op. cil., p. 31.
9 Akeel Bilgrami puts this point as follows:
. . . [0]ne should not go away with the impression that there is no more to the study of meaning than a specification of the assertions (or other speech-acts) that different sentences can be used to effect. If we were under this impression, the simple fact that a sentence can be used to effect any number of assertions in different contexts is a fact that would threaten the possibility of theorizing systematically about meaning. . . . [LJinguistic meaning is a theoretical core that is indispensable in the explanation of our use of language—and so, unsurprisingly, manifest in it. . . The point of the method of radical interpretation is to distil or abstract out of the assent behaviour of an agent (via a combination of observation of the world around the agent and an application of the constraint of charity) this theoretical core. ('Meaning, Holism and Use', Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson, ed. Ernest LePore (Oxford, Blackwell, 1986), pp. 120-121).
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
287
lying forces at work, while the other tries to go further in the direction of what Leibniz called 'metaphysics', Hesse's demand for a new ontology, and her praise of Ricoeur as the only theorist of metaphor who 'recognizes an ontological foundation for metaphor other than the naturalistic one',10 are indications of this difference.
The need to go further in a 'metaphysical' direction than Davidson wants to go is also felt by Michael Dummett, who denies that the task of the philosopher of language has been completed when we have described the process of constructing translation manuals, exhibited the ways in which we are able to predict (and, in some measure, control) linguistic behaviour. Thus when Davidson says that 'the ability to communicate by speech consists in the ability to make oneself understood and to understand', and that this ability does not require 'shared grammar or rules' or 'a portable interpreting machine set to grind out the meaning of an arbitrary utterance', Dummett suggests that this is true only of the idiosyncratic features of idiolects.11 When Davidson says that 'we should give up the attempt to illuminate how we communicate by appeal to conventions,'12 Dummett replies that 'Conventions, whether they be expressly taught or picked up piecemeal, are what constitutes a social practice; to repudiate the role of convention is to deny that a language is in this sense a practice'.13
This exchange brings out the fact that, whereas Davidson is content with an outside view, with discovering the sort of behavioural regularities in which a radical interpreter would be interested, Dummett wants to take up, so to speak, a position inside the speaker or the speaker's community. He wants to discover the rules or conventions which form the program of an interpreting machine. For only if there is something like that to find, Dummett thinks, can one 'throw light on what meaning is'.14 Dummett thinks that if we follow Davidson in jettisoning the notion of'a language', then 'our theories of meaning have no
10 Hesse, op. cit., p. 38.
11 Dummett, 'A nice derangement of epitaphs: some comments on Davidson and Hacking' in Truth and Interpretation, ed. LePore. p. 474.
12 Davidson, 'A nice derangement of epitaphs', in Truth and Interpretation, ed. LePore, cited above, pp. 445-446.
13 Dummett, op. cit., p. 474. 14Dummett, op. cit., p. 464.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
288
I—RICHARD RORTY
subject-matter'.15 Davidson, by contrast, thinks that there is nothing called 'meaning' whose nature is mysterious, and that philosophy of language need no more offer theories about the nature of such a mysterious thing than Newton's Principia needed to offer a theory about the nature of gravity. Gravity was not the subject-matter of that book, but rather various regular motions; meaning is not the subject-matter either of a radical interpreter's T-theory or of philosophy of language, but rather behaviour.
To be sure, the behaviour in question is typically, but not necessarily, behaviour which is sufficiently regular among large numbers of people to give those people a handle for notions like 'correctness', 'rule' and 'social practice'.16 But the utility of such normative notions within a community for controlling and changing the members' linguistic behaviour is independent of the utility of translation manuals for predicting that behaviour. Only when there are sufficient regularities for the insider's normative notions to apply will there be sufficient for the outsider's interpretative, semantical, notions to apply. But this co-extensiveness does not mean that the former notions 'ground' or 'explain' or 'complement', the latter, or that the two sets of notions are relevant to each other in any other way. So the job of the philosopher of language is, for Davidson, finished when the latter notions are explicated by reference to the radical interpreter's procedures.17
15Dummett, op. cit., p. 469.
16See Ian Hacking, 'The parody of conversation', in Truth and Interpretation, ed. LePore, p. 458 for the point that we only have correctness where we have lots of people (not just two) exhibiting the same regularities in the behaviour. Davidson would, I think, have no difficulty accepting this 'anti-private language' point—since it leaves open the possibility of understanding (translating) noises regularly made only by one person, and takes away only the possibility of saying that this person has used a language correctly or incorrectly.
171 have developed this notion of the 'outside' view of the field linguist, and the contrast between Davidson and Dummett's programs, in 'Pragmatism, Davidson and truth' in Truth and Interpretation, cited above, pp. 333-355. On my account of the matter, Davidson sees no need to supplement a T-theory for a language with what Dummett calls 'linking principles', principles which 'make the connection between the theoretical notions and what the speakers of the language say and do'. (Dummett, op. cit., p. 467). At p. 475 Dummett tells us that such linking principles
will be very complex, since they have to describe an immensely complex social practice: they will treat, among other things, of the division of linguistic labour, of the usually ill-defined sources of linguistic authority, of the different modes of
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
289
Only if one agrees with Dummett that what makes understanding possible is something like a portable interpreting machine will one be inclined to think Black's question 'how does metaphor work?' a good one. More specifically, only then will one assume that there is something called 'mastery of a language' which includes an ability to 'get the point' of metaphorical uses of bits of that language. Conversely, only if one thinks that there is such a thing as 'the point' of such a use will one be inclined to think of our ability to understand a metaphor as the result of the workings of such a machine. For only if one has already put irregular and unpredictable uses of language within the reach of notions like 'mastery of the language', will one think of reactions to metaphors as dictated by rules, or conventions, or the program of an interpreting machine. Only then will one think 'How do metaphors work?' a better question than 'What is the nature of the unexpected?' or 'How do surprises work?'
It is of course true that if you do not know English you will get no use out of such metaphors as 'Man is a wolf or 'Metaphor is the dreamwork of language'. Your reaction to these metaphors will be as limited as your reactions to any other utterly unfamiliar noise. But it is one thing to say that the ability to grasp the literal meaning of an English sentence is causally necessary if you are to get something out of its metaphorical use and another to say that this ability insures that you will do so. If Davidson is right, nothing could insure that. The difference between a literal use and a metaphorical use of an English sentence is, on Davidson's view, precisely that 'knowing English' (that is, sharing the current theory about how to handle the linguistic behaviour of English-speakers) is sufficient to understand the former. That is just why we call the use 'literal'. But nothing in existence prior to the metaphor's occurrence is sufficient to understand the metaphorical use. That is just why we call it 'metaphorical'. If
speech and the relations between the parent language and various dialects and slangs.
It is not clear to me how such descriptions can provide a criterion of correctness for a theory of meaning (in Davidson's sense), as Dummett says they can at p. 467. But it is apparent that Dummett thinks that there is some sort of criterion for the correctness of a translation manual other than its giving us what Quine calls the ability to 'bicker with the native like a brother', and that Davidson does not.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
290
I—RICHARD RORTY
'understanding' or 'interpreting' means 'bringing under an antecedent scheme', then metaphors cannot be understood or interpreted. But if we extend these two notions to mean something like 'making use of or 'coping with', then we can say that we come to understand metaphors in the same way that we come to understand anomalous natural phenomena. We do so by revising our theories so as to fit them around the new material. We interpret metaphors in the same sense in which we interpret such anomalies—by casting around for possible revisions in our theories which may help to handle the surprises.18
Davidson does, occasionally, say things which seem to support the view that metaphors have 'cognitive content'. For example: 'Metaphors often make us notice aspects of things we did not notice before; no doubt they bring surprising analogies and similarities to our attention. . . .,19 But notice that the same can be said about anomalous non-linguistic phenomena like platypuses and pulsars. The latter do not (literally) tell us anything, but they do make us notice things and start looking around for analogies and similarities. They do not have cognitive content, but they are responsible for a lot of cognitions. For if they had not turned up we should not have been moved to formulate and deploy certain sentences which do have such content. As with platypuses, so with metaphors. The only important difference is that the platypus does not itself come to express a literal truth, whereas the very same string of words which once formed a metaphorical utterance may, if the metaphor dies into literalness, come to convey such a truth. You may not have to kill the platypus to get a satisfactory theory of
18 See Davidson's 'A Nice Derangement of Epitaphs' for a parallel between metaphors and malapropisms. See also my 'Texts and Lumps', New Literary History, XVII (1985), pp. 1-16 for suggestions on how to avoid Diltheyan distinctions between linguistic and non-linguistic surprises. Hesse has commented on the latter paper in her 'Texts Without Types and Lumps Without Laws', ibid, pp. 31-48. In her paper she interprets Davidson as a 'reductionist' in regard to metaphor. My account of Davidson's view of metaphor in the present paper is an implicit reply to some of Hesse's criticisms of him in hers.
19 Davidson, 'What Metaphors Mean', p. 261. Davidson goes on to say that metaphors 'do provide a kind of lens or lattice,through which we view the relevant phenomenon'. I confess that I cannot see how to use Black's 'lens' and 'filter' metaphors in ways which fit in with Davidson's metaphors, so I am inclined to say that in this passage Davidson grants too much to the opposition.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
291
how it works, but you do have to kill off a metaphor to get a satisfactory theory of how it works. For such a theory will give you a widely-accepted paraphrase, and a metaphor for which such a paraphrase is widely available is just what we mean by a dead metaphor.
I take Davidson to be saying that the positivists were on the right track both when they urged that meaning and cognitive content are coextensive, and when they deprived metaphor of cognitive content. They went wrong only when they failed to add that metaphors were necessary for gaining knowledge, even though they did not (while alive) express knowledge. If this interpretation is right, Davidson should deny what Black affirms: that to say, for example, 'Metaphor is the dreamwork of language' is to 'express a distinctive view of metaphor', a 'new insight into what metaphor is', to say something which a reader could 'understand or misunderstand', etc.20 He should say that, when he began 'What Metaphors Mean' with that metaphor, he was instead inviting the reader to participate in a 'creative endeavour'.21 As he puts it, if we 'give up the idea that the metaphor carries a message' then we can see that the various theories about 'how metaphors work' do not 'provide a method for deciphering an encoded content. . . [but] tell us (or try to tell us) something about the effects metaphors have on us'.22 Davidson can cheerfully agree with the positivists that these effects are 'psychological' rather than 'logical'. But the acqusition of knowledge is, after all, a psychological matter.
One reason philosophers like Habermas and Hesse—philosophers who are suspicious of positivism—are likely to be suspicious of Davidson's attack on 'the thesis that associated with metaphor is a definite cognitive content that its author wishes to convey'23 is that this seems to give the highest flights of genius the same metaphysical status as thunderclaps and birdsongs. It takes them out of the sphere of what Grice calls 'non-natural meaning' and reduces them to the level of mere stimuli, mere evocations. But such suspicion shows how many background assumptions Habermas and Hesse share with their
20 See Black, op. cit. pp. 182-3.
21 Davidson, Essays on Truth and Interpretation, p. 245.
22 Ibid, p. 261.
23 Davidson, Essays on Truth and Interpretation, p. 262.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
292
I—RICHARD RORTY
positivist enemies. They share the Kantian presumption that there is some sort of inviolable 'metaphysical' break between the formal and the material, the logical and the psychological, the non-natural and the natural—between, in short, what Davidson calls 'scheme and content'.
For Davidson, the break between the realm of meaning and cognitive content (the realm in which it is useful to speak of norms and intentions), and the realm of'mere' stimuli, is just the pragmatic and temporary break between stimuli whose occurrences are more or less predictable (on the basis of some antecedent theory) and stimuli which are not—a break whose location changes as theory changes and as, concomitantly, fresh new metaphors die off into literalness.24 The genius who transcends the predictable thereby transcends the cognitive and the meaningful. This is not to the discredit of the genius, but, if to anybody's, to that of the sceptical 'man of reason'. For neither knowledge nor morality will flourish unless somebody uses language for purposes other than making predictable moves in currently popular language-games.25 (Hesse goes too far in saying that metaphor is 'the necessary mode of speech' when fulfilling, e.g., Habermas' 'emancipatory interest'. Plain argumentative prose may, depending on circumstances, be equally useful. But it is certainly true that apt new metaphors have done a lot for radical emancipatory programs in morals and politics.)
One way to see why, if one repudiates Davidson's bete noire—the scheme-content view of meaning and cognition—one will want to analogize metaphor to birdsong is to note that traditional empiricism notoriously ran together the claim that
24 Davidson's anti-Kantian naturalism is well expressed in a passage from 'A Nice Derangement of Epitaphs', pp. 445-446: '. . . we have erased the boundary between knowing a language and knowing our way around in the world generally'. Another way to put the point is to say that this boundary changes as metaphors pass over from the 'world' side to the 'language' side—pass from being evocative to being cliches. It is essential to Davidson's view that dead metaphors are not metaphors, just as it is essential for the opposing 'metaphysical' view, common to Black and Searle (and to the view of Hesse, Mark Johnson and George Lakoff that language is 'shot through' with metaphor), that dead metaphors still count as metaphors. See Searle, 'Metaphor' in Johnson, ed. Philosophical Perspectives on Metaphor, p. 225.
25 Davidson enlarges on this point at the end of his essay 'Paradoxes of Irrationality' in Richard Wollheim and James Hopkins, eds., Philosophical Essays on Freud (Cambridge, Cambridge University Press, 1982). I have enlarged on it further in 'The Contingency of Community', London Review of Books, July 24, 1986.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
293
sensory observation (of, e.g., birdsong) was a stimulus to knowledge and the claim that it conveyed knowledge. This confusion (exposed most thoroughly in Sellars' classic 'Empiricism and the Philosophy of Mind') was between the claim that overhearing, e.g., an unfamiliar noise caused you to acquire the belief that there was a quetzal in the forest and the claim that it 'conveyed the information' that there was a quetzal there. The empiricist slogan 'Nothing in the intellect that was not previously in the senses' traded on this confusion, on the ambiguity in 'source of knowledge' between 'cause of belief, and 'justification of belief.
The same ambiguity arises in the case of 'metaphor is an indispensable source of knowledge'. If we accept the Black-Hesse-Searle view that metaphors convey information, they will be able to function as reasons for belief. On Davidson's view, by contrast, 'live' metaphors can justify belief only in the same metaphorical sense in which one may 'justify' a belief not by citing another belief but by using a non-sentence to stimulate one's interlocutor's sense organs—hoping thereby to cause assent to a sentence. (As when someone holds up a probative photograph and asks 'Now do you believe?')
The relation between birdsong, poetic imagery (the poets' wood-notes wild) and the sort of metaphorical uses of sentences discussed by Black and Davidson may be clarified by considering the following spectrum of unfamiliar noises:
(1) A noise in the primeval forest, heard for the first time and eventually discovered to be the song of a bird hitherto unknown to science, the quetzal.
(2) The first utterance of an 'imagistic' and 'poetic' phrase— e.g., 'that dolphin-torn, that gong-tormented sea'.
(3) The first intentional use of an apparently false or pointless sentence—e.g., 'She set me ablaze', 'Metaphor is the dreamwork of language', 'Man is a wolf, 'No man is an island'.
(4) The first (startling, highly paradoxical) utterance of a sentence which, though still construed literally by reference to a theory which antedated it, comes eventually to be taken as truistic—e.g., 'No harm can come to a good man', 'Love is the only law', 'The earth whirls round the sun', 'There is no largest set', 'The heavens will fill with commerce', 'Meaning does not determine reference'.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
294
I—RICHARD RORTY
Consider what happens as each of these unfamiliar noises becomes more and more integrated into our practices, better and better coped with. (1) helps bring into existence a taxonomy of the avifauna of Central America. In time the call of the quetzal is one more occasion for the heavens filling with commerce, as wealthy bird-watchers fly in. The bird's call never acquires a non-natural meaning, but it does acquire a place in our causal stories about our interaction with the world. The question 'What does that noise mean?' now has answers (e.g., 'It means there is a quetzal around'; 'It means that our village can get in on the tourist industry').
The fragment of Yeats—(2)—also does not acquire a non-natural meaning. But it acquires a place in people's practices— not just in the Yeats industry but in the lives of all those who find themselves remembering it, being haunted by it. It becomes part of what such people are able to say (neither about gongs, dolphins, the sea or Byzantium, nor about anything else), but not part of what they know.26 People's linguistic repertoires are thus enlarged, and their lives and actions changed in ways they cannot easily articulate. But they have not acquired any beliefs which these particular words express. They would not claim to have acquired information from Yeats. Black's apparatus of 'filters'—which, in his 'Man is a wolf example, are supposed to highlight the wolfish features of humanity—is irrelevant to this sort of non-sentential fragment, a fragment which lacks what Black calls a 'primary subject'. Yeats is not interested in making us notice something about the sea, nor about anything else which he or we can usefully put a finger on.
Between (2) and (3) we cross the fuzzy and fluctuating line
26 There is a character in one of Charles Williams' novels for whom the most salient feature of the universe is Milton's line 'And thus the filial Godhead, answering, spake'. It is not that he cares about whether there is or could be such a thing as a filial Godhead. It is the noise itself which matters to him. This noise could not have had this effect, of course, unless he had been familiar with the role in the English language of noises like 'filial' and 'Godhead' (and, perhaps, with the use of similiar noises in Latin and German as well) nor unless he had some familiarity with Christian doctrine. But neither could the little phrase from Vinteuil's sonata have had its effect on Proust's narrator's life and actions if he had not previously listened to other pieces of music of roughly the same sort. The hair on the back of our neck would not stand up when it does if we had not lived the lives we have, but this is not to say that the noises which make them stand up have anything like non-natural meaning, even when these noises happen to be expressions of English, or notes on a musical scale.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
295
between natural and non-natural meaning, between stimulus and cognition, between a noise having a place in a causal network and having, in addition, a place in a pattern of justification of belief. Or, more precisely, we begin to cross this line if and when these unfamiliar noises acquire familiarity and lose vitality through being not just mentioned (as the Yeats fragment was) but used: used in arguments, cited to justify beliefs, treated as counters within a social practice, employed correctly or incorrectly.
The difference between (3) and (4) is the difference between fresh metaphorical sentences and fresh paradoxes. These two blend into one another, but a rough sorting can be made by asking whether the first utterer of what seems a blatantly false remark can offer arguments for what he says. If he can, it is a paradox. If not, it is a metaphor. Both are the sort of noises which, on first hearing 'make no sense'. But as metaphors get picked up, bandied about, and begin to die, and as paradoxes begin to function as conclusions, and later as premises, of arguments, both sorts of noises start to convey information. The process of becoming stale, familiar, unparadoxical and platitudinous is the process by which such noises cross the line from 'mere' causes of belief to reasons for belief.
Crossing this line is not the acquisition of a new metaphysical character, but simply the process of becoming, through increasingly predictable utterance, usefully describable in intentionalistic language—describable as an expression of belief. For a noise to become so describable is for it to assume a place in a pattern of justification of belief. This can, under propitious circumstances, happen to any noise; one can even imagine it happening to the examples I have placed under (1) and (2). It is pointless to ask what there is about the noise which brings about this double describability, as noise and as language. Whether it occurs is a matter of what is going on in the rest of the universe, not of something which lay deep within the noise itself.27 This double describability (as cause and reason,
27 Davidson says
It is no help in explaining how words work in metaphor to posit metaphorical or figurative meanings, or special kinds of poetic or metaphorical truth. Once we understand a metaphor we can call what we grasp the 'metaphorical truth' and (up to a point) say what the 'metaphorical meaning' is. But simply to lodge this
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
296
I—RICHARD RORTY
noise and language) is brought about not by the unfolding of latent content (like a Leibnizian monad), but by unpredictable shifts in causal relations to other noises (like a Newtonian corpuscle). If it does come about, we can look back and explain what features of the noise suited it for this process of familiarization, but there is no way to do so prospectively. For similar reasons, there is no way of telling geniuses from eccentrics, or creativity from idle paradox-mongering, or poetry from babble, prior to seeing how utterances are, over the course of centuries, received. To ask 'how metaphors work' is like asking how genius works. If we knew that, genius would be superfluous. If we knew how metaphors work they would be like the magician's illusions: matters of amusement, rather than (as Hesse rightly says they are) indispensable instruments of moral and intellectual progress.28
meaning in the metaphor is like explaining why a pill puts you to sleep by saying it has a dormitive power. I should prefer to say 'once the metaphor, or the paradox, ceases to seem metaphorical or paradoxical' rather than 'once we understand the metaphor'. Once we drop the idea of a meaning lodged deep within the metaphorical sentence, it is less misleading to say that we simultaneously de-metaphorize the sentence and endow it with a use. We thus endow it with something to be understood—a new literal sense.
I take it that Davidson would regard Black's talk of a 'filter' (adopted by Hesse), Goodman's talk of a 'scheme', and Johnson and Lakoff s talk of a 'gestalt' as so many 'dormitive power' explanations of 'how metaphors work'—so many attempts to find something hidden inside the sentence, as opposed to something lying outside it, which accounts for the transition from an unfamiliar noise to a familiar counter in a social practice. But see n. 19 above.
28 Michael Chase-Levenson and Samuel Wheeler made very valuable criticisms of an earlier draft of this paper, and I have made many revisions in response.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
Richard Rorty and Mary Hesse
//—Mary Hesse
TROPICAL TALK: THE MYTH OF THE
LITERAL
Rorty quotes a nice pun (or is it a metaphor?) from Quine, to the effect that 'cognitive discourse' is 'the neatly worked inner stretches of science [which] are an open space in the tropical jungle, created by clearing tropes away' [fn. 6].1 My aim in this paper is to question this notion of'cognitive discourse', and to try to reinstate the cognitive function of linguistic tropes, particularly metaphor.
Rorty follows Davidson2 in placing metaphoric use of language outside semantics, making a distinction between meaning and use. 'Meaning' is restricted to literal use, and is what semantics is about. Only literal sentences express knowledge. Metaphor is a non-cognitive speech-act, to be understood causally, as 'noise', expressing nothing. Nevertheless it is allowed to have functions, in drawing attention to or seeking literal similarities, in facilitating language-learning, and in oiling the wheels of language-change. But the only 'meaning' associated with a metaphor is its literal meaning (which is usually nonsense or false and anyway does not coincide with its use as metaphor), together with the literal similarities to which it draws attention. It is essential to Davidson's and Rorty's conception that when metaphors introduce new literal 'meanings' they do so causally, not semantically, and then become 'dead': dead metaphors are not metaphors, but a new stage of literal language. What semantics is about is, as it were, a frozen stage of natural language, and all it can deal with is a discrete series of such stages.
1 Page references to Rorty's 'Unfamiliar noises' are given in brackets in the text. 2'What metaphors mean', Inquiries into Truth and Interpretation, (Oxford, Clarendon Press, 1984), p. 245.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
298
II—MARY HESSE
We have seen something like this in logical empiricist attempts to deal with scientific theories, attempts which neglected the dynamics of history and theory-change. Philosophy of science, however, has recognised that the problem of change is a proper part of its brief. I shall return to the analogy between explanations of language and of scientific data below, since Rorty also relies on aspects of this analogy. Meanwhile it does seem on the face of it strange to hold that an explanation of natural language has to make a radical distinction between our cool, dry, literal speech (note that Quine explicitly equates this with scientific uses), and the lush proliferation of metaphor, metonymy, hyperbole, irony and the rest, when all these are intimately woven with the literal in all use of natural language except the self-consciously logical, behavioural and scientific. To place the literal within the sphere of 'philosophical' semantics, and the tropical within something else (linguistics, psychology, history of language?) is one of those dichotomies which might have gone the way of analytic/synthetic, discovery/justification, explanation/description, or reasons/causes. What are the arguments for the literal/metaphorical distinction?
The arguments in the literature have seldom got beyond the stage of swapping examples which are then tailored to fit the author's particular theory. One will not expect crucial inductive instances or knock-down arguments in such a context, but one looks at least for some deeper appeal to the nature and functions of language. Rorty's present paper is specially valuable in that it does begin to provide this, by relating the question of metaphor to Davidson's linguistic behaviourism and his theory of truth in general. Why is it necessary to regard the subject-matter of semantics as an idealized literal language which is almost as far removed from actual speech as is Principia Mathematica or the syllogisms of a medieval disputation? Let us look at Rorty's arguments and his interpretation of Davidson.
First, Rorty draws an analogy between Davidson's conception of semantics as a behaviourist explanation of language, and an instrumental view of scientific explanation as requiring only surface regularities, not deep theories. Roughly, Davidson is to Black, Dummett, etc. as Newton is to Leibniz [p. 286]. This reference to the history of science is perhaps unfortunate, since the historical Newton by no means eschewed deep theory, and
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
299
Leibniz's overt objections to Newton's occasional positivism could be said to be the harbingers of subsequent fruitful progress. To be coupled with Leibniz as requiring deep theories in linguistics is no dishonourable fate. However that may be, the intention of the analogy is clear: Rorty agrees with Davidson that a semantic explanation of language has as explanandum the surface regularities of speech. Davidson recognises the holistic character of these regularities—there is a linguistic network of literal speech, and it is this that is to be deduced from the premisses of the semantic theory, just as the regularities of the behaviour of matter are to be deduced from gravitational or electromagnetic laws. Two questions arise. What is meant by saying that semantics requires no deep theories (what Rorty calls 'metaphysical theories')? And if this is the case, why does it banish metaphor to something analogous to birdsong? [p. 291] To answer these questions we need to look at Rorty's second thesis, which is concerned with Davidson's rejection of the scheme/content distinction.
The notion that theory, and even language itself, structures the world according to certain classificatory schemes or conceptual frameworks became familiar in the wake of relativist tendencies in linguistics, social anthropology and history of science, associated among others with the work of Levy-Bruhl, Sapir and Whorf, and Kuhn. In all these areas problems of translation and understanding arose: if truth in language is relative to a particular linguistic framework, how can truth-values be accurately transmitted from language to language, theory to theory? Davidson holds that in seeking to understand the speakers of other languages we must assume that most of their utterances are true, and capable of translation correctly into our truths, because the test of a working language lies in its ability to express intentions and beliefs, and we all need successful intentions and beliefs within the same world. If we could not assume a common truth-content in the utterances of different cultures, we would have no reason to assume a language was being spoken at all. The problem is an extension of that made familiar in philosophy of science as meaning variance: theories imply different networks of meaning, but there must be agreement on common-or-garden truths describing the world around us, both for practical purposes and because the whole of
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
300
II—MARY HESSE
valid scientific knowledge is supposed to rest upon them.
Davidson does not, as the positivists do, adopt the piecemeal solution to this problem which requires 'observation' sentences to be translated one by one. Like Quine he believes language comes to the tribunal of experience as a whole—a network of interrelated sentences carries truth which becomes manifest in connected patterns of justified belief, and increasingly predictable as expressions of that belief. In a working language, truth is a function of successful public use. There are no metaphysical 'meanings' hidden in use that mediate between language and the world; there is only surface, behavioural, regularity of utterance. Neither are there schemes or conceptual frameworks to come between language and the world—all such differences are idle-wheels relative to linguistic functioning, to be bypassed in successful communication between speakers and in their commerce with the world.
The above paragraph is partly derived from Rorty's interpretation [pp. 295-296], which suggests how far Davidson has moved from the problem of translation to the assertion that philosophically speaking there is no such problem, because every working language is in direct touch with reality. In his paper 'Pragmatism, Davidson and Truth' Rorty applies this view also to value-judgments. Talk about goodness and redness are explicable in the same way, namely in the 'trivial sense' that there is a working web of inferential relationships between the sentences of the language. We can get no nearer to 'contact with reality' than that. Moreover if disagreements occur about what is morally right, or what is red, 'our disagreements with them will be explicable by various differences in our respective environments (or the environments of our respective ancestors)'.3
I do not want want to consider here whether Rorty has given an entirely fair account of Davidson's position. I suspect that he has arrived at more radical conclusions than Davidson would countenance by ironing out some of the paradoxes in Davidson's approach—for example the apparent neglect of the patent fact that different theories and different cultures do parcel up the contents of the world differently. But since Rorty's understanding
3 'Pragmatism, Davidson and truth', in Truth and Interpretation, (Oxford, Blackwell, 1986), p. 351.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
301
of Davidson underlies what he says about metaphor (and interprets Davidson as saying about metaphor), I shall try to restate the position by using Rorty's own analogy with scientific theory. Rorty's distinction between the literal and the metaphorical is not a metaphysical but a pragmatic one. It rests ultimately on the need for 'predictable utterances' which can 'assume a place in a pattern of justification of belief [pp. 295]. Although he generalizes the domain of belief from the factual to the normative, justification of factual (including scientific) belief still seems to be the persuasive model. This places the difference between his and my views of metaphor within a dispute about the nature of knowledge, and makes the argument by analogy with science crucial.
Davidson has moved away from the atomic empiricism of observation sentences, and taken on board the Duhem-Quine thesis that theory is an internally connected network that represents the world as a whole. It was however the existence of fairly discrete observation sentences that gave original plausibility to the thesis that we must share with all language speakers a basis of accepted inter-translatable truths,4 and that charity or humanity must dictate as far as possible our rendering of other languages into these truths. The truth of discrete observation sentences is what we need for direct commerce with the world, but the Duhem-Quine thesis means that the points of the language at which this commerce takes place are not determined either by the net or by the world—within a given language the 'best fit' of the network as a whole is constantly subject to negotiation and modification as successful language is used for communication and expression of belief. This means that 'best translation' between different language-nets also takes place as a whole, and by constant negotiation. Under these circumstances it is not clear just how 'charity' is supposed generally to dictate our translations of other people's 'observation sentences' into our true sentences, or what grounds we could have for denying 'truth' to their observation sentences. 'Here comes the lion' has
4 This is the thesis that Dummett rightly holds on to in his debates with Davidson about the need for truth criteria for theoretical sentences which have left observations behind, and for which meaning and truth are not patent. See M. Dummett, 'What is a theory of meaning? (II)', Truth and Meaning, ed. G Evans and J. McDowell (Oxford, Clarendon Press, 1976), p. 98 f.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
302
II—MARY HESSE
an obvious translation which may have immediate survival value in some circumstances, but if the other people's version of this saying occurs in the absence of lions but in the presence of a (to us) bewildering and undefinable atmosphere of social expectation and menace, who are we to hold that their (to them) perfectly simple and clear 'observation sentence' is false? In such a case the sentence and its translation belong to different and probably incompatible nets of language and belief. If there are grounds for holding that this second use of the sentence is not 'in close touch with reality', these grounds cannot be as straightforward as those in the case of 'No lions here' when a lion is patently bounding towards us.
There is another argument to be found in Quine and Davidson which may be used to suggest non-straightforward grounds for judgments of falsity of other people's beliefs. This argument rests crucially upon an acceptance of our science as not just the best account of the world we have, but as far it goes the best anyone could have. It is assumed that by and large the body of theoretical science is acceptable and agreed in an increasingly unified world-wide culture, and that the area of such agreement is constantly expanding.5 As an argument this begs several questions, for example about the notion of the 'unified world-wide culture'—is this a rational or a social phenomenon? And it presupposes the adequacy of the realist view of scientific theory as an accumulation of and convergence to truth: a view that is still highly controversial in the debates between 'realists' and 'anti-realists'. But even if that debate should be settled in favour of the realists, it still would not follow that an increasingly true and convergent theoretical science of the natural world would exhaust the true successful ways of talking about reality. It is a naturalistic prejudice that English locutions such as 'Here comes the lion' are the paradigm cases of successful description. The claim to manifest and common truth in all working languages at their interface with reality does not therefore seem well based on the scientific analogy, either in terms of the commonality of observation sentences, or of the
5 Most explicitly laid out in W. V. O. Quine, Ontological Relativity, (New York, Columbia University Press, 1969), ch. 5; and H. Putnam, 'What is realism?', Proc. Aristotelian Soc. 1975/6, p. 177.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
303
suggested potential of scientific theory as a vehicle of comprehensive truth.
If we move from judgments of'red' to judgments of'good', the scientific analogy is even less secure. There is no need to depend here on a radical fact-value distinction in science, and I would not wish to do so. Values do enter scientific theory in the guise of selection of unifying concepts, or judgments of'normality' and 'stability', and in many other ways.6 But it is just where these value judgments are most obviously being used to structure theory (in some parts of biology and in the social sciences) that there is least plausibility in the theses of value-independent data-reports and of theoretical accumulation, convergence and realism. Hence even if we do have a systematic network of utterances about the 'good', we can use neither the positivist basis of observation sentences, nor the potential global agreement over scientific theory, as firm evidence for our being 'in close touch with [normative] reality'. Davidson and Rorty have borrowed prestigious arguments from science and extended them to areas of discourse where the relevant analogies do not hold.
The picture of language that Rorty conjures up is however somewhat more naturalistic than Davidson's. Rorty describes environmentally caused 'noises' as gradually, over evolution and history, yielding an entirely causal and selectively beneficial network of literal talk, interrelated by what we call rational inferences. So far so good. As I shall indicate below, I do not believe in non-naturalistic hidden meanings or linguistic aprioris any more than Rorty does. But his picture implies more than this. It implies that the messy transition stages from noise to talk, stages that seem to be always with us, are imperfections on the way to some ideal, literal, rational discourse that is alone fitted to express knowledge. The aspiration is that of the 17th-century Royal Society's 'close, naked, natural way of speaking',7 and of the Enlightenment vision of scientific rationality informing all areas of discourse and practice, and eschewing in particular tropical talk.
61 have discussed this in 'Theory and value in the social sciences', Revolutions and Reconstructions in Philosophy of Science, (Brighton, Harvester Press, 1980), Ch. 8. 7T. Sprat, The History of the Royal Society, (London, 1667), section 20.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
304
II—MARY HESSE
But suppose that the messy transition stages are in many respects the norm of human communication. The question of the stage at which 'rational language' enters human evolution is a contestable one, and hardly one that is theory-independent. It is not obvious that the accumulation of scientific knowledge is the right analogy to draw on, nor even the instrumental criterion of'predictability' of utterance, itself a criterion drawn from instrumental science. An even stronger objection to the scientific analogy here is the possibility that science itself always exists in stages of transition from theory to theory, paradigm to paradigm, as anti-realists would argue. The relevance of this question already shows that the Quine-Davidson-Rorty analogy takes sides on the debate about scientific realism, and cannot be said to go beyond it as Rorty has claimed.8
Are there other arguments for the patency of literal truth? The only ones that are hinted at in the literature depend essentially on contemplating disaster for philosophy if this hypothesis were abandoned. For example, how can we recognize other utterances as languages at all and not noise? Answer: if we cannot find a basis of agreed truth, we become enmeshed in problems about meaning variance and the criteria of holistic translation. Again, how can we find a basic semantics upon which to rest our theories of entailment, modality, understanding, 'and a host of other semantic and mental notions'?9 Answer: if this is not to be found in a Davidson-type semantic theory, a great deal of philosophical reconstruction would need to be done. But none of these hypothetical answers are arguments. It may be that a great deal of philosophical work does need to be done, just as reconstructions of theories of meaning, explanation, modality, rationality and understanding have been undertaken in philosophy of science.
We must conclude that there are no decisive arguments for Davidson's manifest truth, or in general for a radical explanatory distinction between a literal, semantically amenable ideal of language, and the tropical jungle in which it is always enmeshed. Let us return to Davidson's rejection of the
8'Pragmatism, Davidson and truth', p. 351 ff.
9J. R. Searle, Expression and Meaning, (Cambridge, Cambridge University Press, 1979), p. 132.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
305
scheme/content distinction, and see where this conclusion leaves us.
According to Quine, Davidson and Rorty, schemes or frameworks belong to the uncleared jungle of language, and so do metaphor and other linguistic tropes. If an analogy is to be drawn with the development of science, then it would seem appropriate to notice that current discussion emphasizes the problems of theory growth, theory change, theory conflict, without presupposing that these processes have an attainable, definable, or even relevant, ideal telos. Differences of theoretical ontology and conceptual framework are of the essence of current history and sociology of science, and of much of the philosophy of particular sciences. Is this just a mistake, oramisidentification of essentially historical and scientific questions with philosophical ones? Such questions of demarcation are not of great philosophical interest. What would be of philosophical interest would be a successful account of how models and analogical frameworks work in scientific inference, and by extension, how metaphor works in discourse. It may well turn out that nothing but broad generalities are possible in such an account, supplemented by detailed study of particular cases. This would be interesting, not because it would yield a 'philosophical' theory, or even because a completely general logical or scientific theory would become available, but because it would show how science and language can be 'rational' and 'cognitive' without being constrained by the idealizations of formal logic or the semantics of the literal.
Underdetermined schemes and frameworks remain with us as intrinsic elements of our expressions of the real, and so do the metaphors and models within which we structure our representations of things, both on the scale of cosmic models of 'mechanism', or 'evolution', or 'information processing', and on the micro-scale of particular views of particular things: 'metaphor is birdsong', 'linguistic tropes are a jungle', 'rationality is logic', and so on. Rorty, however, describes another thesis related to the rejection of schemes, which need not be affected by their reinstatement as an intrinsic element of science and general discourse. This is the thesis, which he calls 'metaphysical', that there is something like a framework, or set of hidden linguistic 'meanings', that mediates between language and the world, and cuts off language from being the purely naturalistic response of a
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
306
II—MARY HESSE
particular evolutionary kind to the pressures of the environment. For him, and he claims for Davidson also, working languages have no deep semantic structure to explain their application to the world; they are just sets of surface regularities and interconnections of speech, which as a matter of behavioural fact enable communication to occur.
Rejection of such a metaphysical notion of 'meaning' does not, however, entail rejection of linguistic schemes as I have described them. Schemes are as behavioural as any other features of languages—they are the surface ways of classifying and ordering experience, part of the network of relatedness that should enter a full explanation of how language works. So far Rorty might well be able to agree that such 'schemes' do exist. Indeed it is here that a difference seems to open up between Davidson and Rorty. Rorty admits to embarrassment about Davidson's claimed allegiance to 'an objective public world which is not of our making'.10 For Rorty the relation between language and any such world is of a purely causal, knee-jerk, variety, and is inscrutable. This applies as well to literal as to metaphorical language. About a particularly holistic passage in Davidson he says:
I interpret this passage as saying that the inferential relations between our belief that S and our other beliefs have nothing in particular to do with the aboutness relation which ties S to its objects. The lines of evidential force, so to speak, do not parallel the lines of referential direction. . . . To know about the former lines is to know the language in which the beliefs are expressed. To know about the latter is to have an empirical theory about what the people who use that language mean by what they say—which is also the story about the causal roles played by their linguistic behaviour in their interaction with their environment.11 (My italics).
A couple of paragraphs earlier Rorty rejects 'the contrast between "objective realities" and "useful fictions", or that between the "ontological status" of the objects of, respectively,
10 'Pragmatism, Davidson and truth', p. 354.
11 ibid p. 353.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
307
physics, ethics and logic'. Taking physics as the best worked-out example here, the issue as to whether it refers to unobservable 'objective realities' is controversial, but there is no doubt that it does contain 'schemes' in the sense of fundamental models or paradigms. Are these more like the old 'objective realities' or like 'useful fictions'? Rorty would reject the question, but he cannot evade the fact of underdetermination: schemes in physics are subject to historical change—in this they are more like useful fictions than like objective realities. In rejecting schemes as mediating tertia, Rorty has not got rid of the essential relativism of the different knee-jerks with which different theories and different cultures respond to their environment. It might indeed be said that while Davidson rejects the scheme/content distinction in favour of common content, i.e. the objective public world, Rorty rejects the distinction in favour of schemes with no content. For Rorty the question of content which 'represents the world' is almost buried beneath schemes and pragmatic responses to environment plus intersubjective-communication, conceived as an unanalysable whole. There are only schemes and their behavioural networks.
Since schemes of this kind are with us in both literal and metaphorical talk, and do not imply the existence of metaphysical meanings, why does Rorty reject the possibility that metaphorical frameworks can express knowledge, while presumably accepting that literal and scientific frameworks can do so? The reason seems to be his belief in an important distinction between the tropical jungle and the scientific clearings, that is, a belief in a negotiable and specifiable ideal of literal language, transcending particular schemes and metaphors. But when Rorty refers to disagreements as due to our environments or the environments of our ancestors, this possibility of disagreement seems to apply to literal as well as to metaphorical talk. In both cases Rorty has replaced deep metaphysical common 'meanings' by an evolved ideal of the literal, rational and cognitive, somewhat like Peirce's long-term convergence on truth, or Habermas's products of the ideal speech situation, although Rorty does not, like them, put this forward as a theory of truth, but only of something like evolutionary success. But if, as I have suggested, there is no argument or evidence even for the evolutionary existence of such ideals, there is no argument for a
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
308
II—MARY HESSE
distinction between the knowledge-bearing properties of the literal and the metaphorical either.
When the friends of metaphor seek for metaphoric 'meaning' or 'cognitive content' they are not asking for anything other than what Rorty describes in connection with literal language. There are no substantial hidden meanings for the literal either, there are only holistic meaning-relations, which are adjusted to give best fit for practical purposes. To explain the meanings of either literal or metaphorical talk is just to trace out surface regularities and interconnections in the kind of generalised semantics for which I have suggested a programme.
In particular, schemes can express knowledge—they are not cognitive idle-wheels in either literal or metaphorical talk, certainly not in scientific theory. Insofar as schemes in science are often alternative and equally adequate ways of explaining the same data as of now, they must be said to be distinguishable from their common empirical content. But this does not imply that the cognitive character of schemes and content differ. Schemes are not unfalsiflable.12 Even if no decision can be made now between two radically different fundamental theories in the light of present evidence, such theories always have further implications (usually of a probabilistic or analogical rather than a deductive kind), and these implications will often provide empirical decision procedures between theories. Moreover this can be the case without any further assumption of a linear sequence of winning theories converging upon a unique ideal. Schemes in science therefore share the cognitive properties of all theories, that they are sensitive to changes in the stock of empirical data, and in the social purposes for which they are sought. Schemes represent the world adequately as far as they go, but they do not and need not represent the 'content' of reality with universal accuracy.
What distinguishes such cognitive views of scheme and metaphor from the Rorty-Davidson view is not the postulation of metaphysical meanings. It is rather the question whether cognitive functions like those of scientific schemes can be found in general discourse, where this includes expressions of belief
12See my 'The hunt for scientific reason', PSA 1980, Vol. 2, ed. P. D. Asquith and R. Giere (Philosophy of Science Association, 1981), p. 3.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
309
about norms and values, gods and demons, heavens and hells. Insofar as common language has resources for decision-making in these areas, the relevant schemes and metaphors are surely as cognitive as are scientific paradigms. For a philosophical naturalist, who believes that science has access to all the reality there is, there is no rational decision-making outside science, only the play of causal and cultural forces. But for Rorty, and perhaps for Davidson, the rational is always only the play of causal and cultural forces (of a particular evolutionary kind), and this applies equally to science and all other discourse. For Rorty there are in all cases cognitive decision-procedures, and he explicitly relates the 'cognitive' to the predictability of linguistic behaviour in all forms of discourse, normative and factual alike. The theory of metaphoric meaning I have projected would provide such predictability for metaphoric as well as literal talk. In the case of metaphor it will not be tight and univocal predictability, because the nature of metaphor is to be allusive, evocative, and partly ambiguous. But ambiguity is not total—we do know how to respond appropriately to tropical talk, we do not flounder about in morasses of uncommunication until we miraculously come upon the cleared space of the literal. (I hope you understand and will respond predictably to this sentence). It may seem that the projected semantics of metaphor is very far from being available and is merely a blank cheque on the future; on the other hand, an important philosophical principle should not be made to rest on the assumed practical impossibility of such a theory. Neither should the propriety or otherwise of calling such a theory 'semantics', or its subject-matter 'cognitive'.
In summary, the substantial issues between Rorty and the friends of metaphor seem to reduce to two. First, should semantics, with all its prestigious logical and cognitive vocabulary, be restricted to literal talk? Second, if the primary reason for this restriction (at least for Rorty) is a belief in the importance of an ideal, univocal science-type language for every form of discourse, do the arguments for this belief hold up? By developing the implied analogy with philosophical accounts of science, I have suggested that the arguments do not hold up, and that there is no reason for denying cognitivity to schemes and metaphors as expressions of belief and knowledge. Nothing in principle forbids a semantic explanation of language novelty,
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
310
II—MARY HESSE
change and conflict, and whether this is called philosophical semantics or by some other name is not a question worth arguing about.13 In no case does adoption of such a programme for semantics imply that there are hidden metaphysical meanings, any more than the search for explanatory unobservables in physics is a search for metaphysical entities.
Rorty dislikes dualisms, and so do I. In the paper on metaphor he refers to,14 I sought to dissolve a dualism that he still wishes to reinforce, namely that between the literal and the metaphorical. There I put forward the thesis that, in a carefully specified sense, 'all language is metaphorical'. Of course, even assuming this thesis is true, we still need some pragmatic account of the distinctions we do in fact make between plain and tropical language in ordinary talk. My pragmatic distinction is roughly in terms of familiarity, observability and ease of learning and communication, and relates particularly to the need of science and logic for the univocity and substitutability of technical terms. In other words, the relatively 'literal' is required for a particular kind of knowledge-interest (Habermas's 'technical interest'). Rorty's distinction is not too different—it is about predictability of linguistic behaviour and fitness to express belief-inferences. The difference is that he equates his distinction with that between fitness and unfitness to express knowledge and belief, and that is to beg the very question at issue about the cognitivity of metaphor.
If the claim that all language is metaphorical is true, the explanation of metaphor cannot require a basic literal language as premiss. In 'The cognitive claims of metaphor' I used Wittgenstein's notion of family resemblances between general
13 What philosophy is about, or rather what it is not about, is a question Rorty thinks important in the context of rejecting many traditional philosophical problems. For example, after the field linguist has done his best with the ethnographic and translation job, 'there is no further job for philosophy to do' ('Pragmatism, Davidson and truth', p. 341) and 'we can safely get along with less philosophizing about truth than we had thought we needed' (ibid p. 345). I would rather move the discussion on to what philosophy can do after the dissolution of these problems—namely to give the sort of general accounts of rationality that are not, or not yet, the subject-matter of the special sciences.
14 'The cognitive claims of metaphor', Metaphor and Religion, Theolinguistics 2, ed. J. P. van Noppen, Brussels, 1984,27. (See also M. A. Arbib and M. B. Hesse The Construction of Reality, (Cambridge, Cambridge University Press, 1987), ch. 8.) I am grateful to Michael Bravo for discussions of these two papers.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions
UNFAMILIAR NOISES
311
descriptive terms ('game' etc.) to argue that all applications of general terms, however apparently literal, depend on perceptions of similarities between their referents. In learning a language, we learn to structure our perceptions of similarity, so that the general terms of that language implicitly classify the furniture of the world in conformity with the classifications of our culture. Different natural languages generally presuppose different classifications, which, like theories, are underdetermined by the world. This is a fundamental fact about language: the world does not come naturally parcelled up into sets of identical instances for our inspection and description. What we call linguistic 'metaphor' is only a complex extension of the same process into novel and striking contexts, and does not differ in principle from any decision to recognize 'That's an X again'. Like Rorty, I would reject the question 'Is it really an X?'. It is an X if the classification presupposed in a given language can coherently contain the object as an X, when related to the rest of the language and its acceptability in its community. At the edges there will always be disputes about whether there is sufficient coherence, and whether the classification is to be accepted in this case, and therefore about whether this thing is to be taken as an X. In cases that we discriminate as 'metaphor' there is additional uncertainty due to unfamiliar similarities that are drawn attention to, and perhaps due to the recommendation implicit in metaphor that we should change some of the classifications buried in familiar language (change, for example, the classification 'metaphor is deviant' to 'all rational language is metaphorical'). Some classifications are better for some purposes than others: science for prediction and control, rhetoric for political and moral persuasion, novel and extended metaphors for aesthetic expression and philosophy, myths for philosophy and religion. But no domain of'knowledge', whether science or any other, can assure us that there is one 'true' classification, or one ideal literal language.
This content downloaded from 129.20.73.3 on Thu, 10 Jul 2014 09:47:47 AM All use subject to JSTOR Terms and Conditions