Version classiqueVersion mobile

La rhétorique – Rhetoricity

Jean-Paul Regis

Metaphor in the Anglo-American. Philosophical Tradition

Daniel J. Costello

Texte intégral

1Though we use and identify metaphor quite often, most of us are probably unaware of the fascination and perplexity that this particular use of language has generated among philosophers since Aristotle. In recent decades more than ever, metaphor has been recognized as a subject closely related to certain theoretical concerns of traditional importance to philosophy. In what follows, I provide a brief survey and assessment of a few of the more important perspectives on metaphor that have emerged in Anglo-American philosophy, in an attempt to raise questions about some common presuppositions concerning language-use in general. To this end, I begin with a preliminary effort to define the terms of my title.

2Most views of metaphor begin by distinguishing the metaphorical use of language from its ordinary literal use, which is taken to be strictly denotative. In ordinary language, we represent reality to ourselves and to each other; its denotative capacity allows us to characterize, or make reference to different states of the world. Metaphor is said to be a deviation from this standard use; it is discerned when we confront a nonstandard conjunction of standard literal terms, whose strikingly odd concurrence in a manifestly false or apparently nonsensical assertion deters us from assigning to them their usual meanings. This initially unfamiliar use of familiar linguistic expression may effect a novel comparison of normally unassociated objects or events, which may in turn lead us to identify previously unrecognized similarities in their respective properties. As such, we have come to accept as meaningful the hackneyed claims that time is money, life a stage, and no man an island, just as we are likely to grant a certain sense to one who sings "I'm a Howlin' Wolf," or "You are the Sunshine of My Life". The point of these trite examples is only to illustrate that metaphoric attribution is normally understood to involve a distinct transfer of meaning, from ordinary, direct reference to a peculiar form of innovative, indirect reference. Indeed, such would seem to be warranted by the etymology of the term, which comes to us from the Greek μετα (usually taken to signify the idea of displacement, or transfer) and φԑρω (to carry, or bear).

3The Anglo-American philosophical tradition is most fundamentally a foundationalist enterprise, which is to say that, despite many great differences over various matters of style and substance, representatives of this tradition have for the most part endeavored to provide definitive criteria for the establishment of our truth-claims. I will return to a more detailed discussion of this project below; presently, I want only to indicate that consideration of metaphor in this tradition has conformed quite consistently to this goal. Thus, even while professing their sincere admiration for the many important aesthetic aspects of metaphor, English-speaking philosophers (with only a few notable exceptions) have insisted that any adequate "theory" of metaphor must begin by addressing its "cognitive" aspect, that is, its role in the process by which we acquire and communicate knowledge.

4As to this role, there has been significant disagreement — in particular, over exactly how the alleged metaphorical "transfer of meaning" is accomplished, and just what it produces for cognition. More precisely, I submit that discussion of metaphor in the Anglo-American philosophical tradition has centered on two specific questions: first, whether metaphors can be accurately "translated" into literal terms (and if so, how); and second, whether metaphorical assertions bear a "cognitive content" that renders them legitimate "truth-value candidates" (and if so, how such content may be discerned). In relation to these two prevailing questions-which, roughly speaking, might be said to reflect the traditional philosophical preoccupation with questions of truth and meaning — it will be useful to identify four distinct positions that may be represented schematically as follows:

Metaphorical assertions have "cognitive content"



Metaphorical assertions are








  • 1 Most notably, in the fear of the seductive, irrational power of figurative language, which was tho (...)
  • 2 In particular, in Thomas Hobbes (Leviathan 1,5) and John Locke (An Essay Concerning Human Understa (...)

5Position (1) is the traditional view derived from classical philosophy1 and made famous by the British empiricists.2 It holds that metaphor is just the nonliteral use of words or expressions as decorative ornaments for stylistic embellishment, a use that is, as such, most inappropriate for precise expression or careful reasoning. Once duly translated — by straight substitution into commensurate literal terms, or into equivalent literal comparisons (i.e., as similes) — metaphorical assertions may acquire cognitive significance. Indeed, such assertions have on occasion been known to enhance our apprehension of the world, by capturing insights and identifying nuances that might normally be overlooked. Poets exemplify this activity, in their use of metaphor as a means of artistic expression. Nevertheless, on this view, it is imperative that serious thinkers avoid this inordinately frivolous use of language at all costs. For in shifting a term or expression from one object to another, and thereby extending the pretense that something is the case when it is not, metaphor can at best provide amusing diversion; otherwise, it merely promotes confusion and deception, and is thus a grave impediment to all sincere attempts to communicate knowledge, which can only take the form of literal truth. Thus, for this approach, metaphorical assertions have "cognitive content" if and only if they can be "translated" into literal terms.

  • 3 Cf. in particular, Ogden and Richards, The Meaning of Meaning, New York: Harcourt-Brace, (8th ed.) (...)

6Position (2) is best represented among logical positivists, who manage to expand this traditional empiricist aversion to figurative language in general, and to metaphor in particular. Positivists intended to rid philosophy of the exaggerated claims of speculative metaphysics, by accepting only literal statements capable of verification as valid knowledge claims. Indeed, they held as their central principle that the meaning of a statement is identical to the set of experiences which are equivalent to its being true; all other assertions amount to meaningless subjective sentiment. Positivism thus sharply distinguished "cognitive" and "emotive" functions of language, and considered metaphor a typical instance of the latter.3 As such, even though they might be effectively reformulated in literal terms, metaphorical assertions nonetheless amount to mere expressions of attitude or feeling-in short, wholly "emotive" utterances incapable of either verification or falsification, and therefore utterly devoid of truth-value, or "cognitive content".

  • 4 This celebrated term was coined by I.A. Richards in his seminal work, The Philosophy of Rhetoric, (...)
  • 5 Cf. Max Black, "Metaphor," in his Models and Metaphors (Cornell, 1962); Monroe Beardsley, "Metapho (...)

7Position (3) marks a significant reorientation, which has come to define the dominant contemporary perspective. Against the traditional view, metaphors are here considered most often incapable of literal paraphrase, as the meaning of a metaphorical assertion is understood to be the product of a complex comparison and "interaction"4 of distinct concepts, categories, and extended systems of associations. On this view, initiated in the work of I.A. Richards and Max Black, and reasserted in various forms over the past twenty years by many prominent American philosophers and linguists (among them, Monroe Beardsley, Paul Henle, George Lakoff, Mark Johnson and Nelson Goodman)5, understanding metaphor is a distinct intellectual operation, involving an intricate procedure of simultaneous selection and suppression, effectively, an inventive reorganization of disparate elements associated with the distinct subjects in question. In such terms, metaphorical assertions are bearers of a unique, irreducible and indispensable "cognitive content". Despite their ample differences of opinion accounting for the particular mechanisms that create this content, advocates of this position share the view that metaphor is pervasive in our everyday speech precisely because the human conceptual process itself is metaphorically structured, since to conceptualize is just to classify experience in terms of familiar antecedent categories.

  • 6 Cf. Thomas Kuhn, The Structure of Scientific Revolutions (Chicago, 1962), Paul Feyerabend, Science (...)

8This "interaction" view is significant for its recognition of the important role of metaphor in the acquisition of knowledge and in the evolution of language. In terms familiar to this position, many acclaimed philosophers of science (such as Thomas Kuhn, Paul Feyerabend and Mary Hesse)6 have argued that scientific models can be understood as extended metaphors. In transposing the anomalous data or particular facts of one theory to the vocabulary of another theory that is more familiar to us, models offer a powerful tool for discovery. Accordingly, we have light waves and nuclear meltdowns, black holes and big bangs — metaphors that have called forth previously unrecognized similarities through a complex "interactive" process. In this function, metaphors are seen to be constitutive of the theories they express, and hence an essential part of scientific progress.

9Position (4) signals a key break, for denying both "cognitive content" and "translatability", and in effect, rejecting the longstanding idea that our reception and use of metaphor is fundamentally a matter of semantics. Instead, this view proposes that we understand the work of metaphor exclusively in terms of pragmatics. Vital to this position is the claim, associated chiefly with Donald Davidson and his followers, that metaphors convey no coded message, nor mean anything other than what they literally say. Rather, metaphors provoke or inspire us to see "as" rather than see "that" — so what distinguishes metaphor is not meaning, but use. To understand this position well, a closer comparison with the preceding viewpoint in terms of the two questions we have been considering will be instructive. Adherents to this "pragmatic" account agree with the "interactionists" of position (3) in holding that metaphorical assertions cannot be translated into equivalent literal assertions — but this agreement is not because they share the idea that metaphors bear an irreducible "cognitive content," rather, it is because they believe that there is nothing nonliteral in a metaphor to be translated. So evidently, these positions differ on the issue of "cognitive content": against the "interactionists," supporters of the "pragmatic" view deny the existence of "metaphorical meaning" — but they deny it not because they share the positivists" reduction of metaphor to trivial, emotive expression, but because they hold that metaphor achieves its wonders with no more than ordinary word meanings, albeit employed in imaginative new ways.

  • 7 This forms a part of Donald Davidson's argument in "What Metaphors Mean," reprinted in Johnson, op (...)

10As one might expect, these differences have instigated a spirited debate. Proponents of the "pragmatic" account of metaphor criticize their adversaries as follows: you "interactionists" claim that metaphor expresses something that literal language cannot, but then you call that something a "cognitive content" — which is just what literal language expresses.7 This predicament can be avoided, they say, if one simply stops trying to explain or decipher a mysterious encoded content, and instead, limits oneself to describing what metaphor evokes. In a noteworthy passage of his controversial essay "What Metaphors Mean," Donald Davidson writes,

  • 8 Ibid., p. 202.

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. These ideas don't explain metaphor, metaphor explains them [...] simply to lodge this meaning in the metaphor is like explaining why a pill puts you to sleep by saying it has dormative power.8

11Most would agree that when we encounter a metaphor, we often find ourselves forming new ideas and beliefs. As Davidson would have it, metaphors are best understood as causes for — rather than expressions of — such changes.

12Advocates of the above position (3), the "interaction" view of metaphor, have responded with criticisms of their own. Without wanting to deny the significance of metaphor as a "speech act", they accuse any strictly "pragmatic" view of disregarding the fact that metaphors are pervasive in our everyday thought processes and communicative practices, where they are used quite commonly to convey messages, in both ordinary and specialized contexts. Thus, any satisfactory account of metaphor must seek to explain, by reference to some form of general rule or mechanism, just how such content is successfully transmitted. Yet the "pragmatic" approach to metaphor evades this vital question entirely. In an essay entitled "How Metaphors Work: A Reply to Donald Davidson," Max Black has put this point quite simply:

  • 9 Max Black, "How Metaphors Work: A Reply to Donald Davidson," in Sheldon Sacks, On Metaphor (Chicag (...)

The gravest objection to Davidson's vigorously argued standpoint then is that, while rejecting current views, it supplies no insight into how metaphors work....9

  • 10 Cf. Donald Davidson, "What Metaphors Mean," in Mark Johnson ed., Philosophical Approaches to Metap (...)

13It would appear that we have arrived at a standoff. So perhaps it will be a good time to pose the reasonable question, "is there really a difference here that makes a difference?" These positions do agree that metaphors cannot be paraphrased, and this, after all, is what marks the decisive break from the traditional views that have always been so consistently dismissive of (when not hostile to) metaphorical expression. True, one side calls that which escapes paraphrase "meaning", the other calls it "effect" — but is this not just a minor quibble over how we use the word "meaning"? In what remains of this essay, I intend to follow Davidson, and then one of his key interpreters, Richard Rorty, in answering yes, what we have here is indeed a quibble over the meaning of "meaning", but it is not at all a minor one, for it forces us to address the prior question of what it means to have a language.10 To do so, I must return briefly to that other term of my title.

14I have described mainstream Anglo-American philosophy as a foundational enterprise. Through a careful and exhaustive analysis of human consciousness, philosophers in this tradition have sought to discover universal and necessary structures of thought and experience that would provide criteria for the establishment of our truth-claims. In this, Anglo-American philosophers place themselves at the center of the modern scientific project, for their effort to secure essential foundations for empirical knowledge, and indeed, for rational progress in all human endeavours.

15It is generally accepted, and perhaps ironic, that the central categories which define this foundational project originate not in England or America, but in the work of a celebrated Frenchman. I refer, of course, to René Descartes, and his "discovery" of the vexing problem of subjective consciousness. By applying a new and fully rigorous method of doubt to all of his previous beliefs, he revealed the bare cogito — all that could be known according to the criterion of strict certainty. Yet his sharp division of subject and object left a worrisome problem: how is it possible to demonstrate the possibility of objective knowledge from the limited perspective of subjective perception? Significant for our discussion of metaphor today is that this search for cognitive foundations, the epistemological project at the heart of modern philosophy, would later take a radical "linguistic turn" in England and America, as philosophers began to question this alleged primacy and incorrigibility of subjective perception. All sides in the great epistemological debates of the seventeenth and eighteenth centuries had shared the assumption that we can experience only subjective impressions, and not external things. The "linguistic turn" involved a simple reversal of this assumption, arguing instead that our ideas of subjective impressions are themselves always described in reference to the public concepts of ordinary language. With this reversal, the search for criteria to establish our truth-claims-the foundational project — turns toward the study of language. Consequently, language replaces subjective consciousness as that which stands opposed to reality; with it, the persistent problem of reconciling subjective perception with objective knowledge is taken over by a new one, that of demonstrating correspondence between subjective utterances and objective facts about the world. To address this problem, philosophers turn from the analysis of consciousness (and the structure of sense-perception) to an analysis of language (and the structure of meaning), in their ongoing effort to discover the general rules that allow us to represent reality accurately.

  • 11 Donald Davidson, "A Nice Derangement of Epitaphs," in Ernest LePore, ed. Truth and Interpretation (...)

16Such is the project that informs the "interactionist" objection to the "pragmatic" account of metaphor. As metaphor is used to convey meaning, an adequate theory must explain this fact, by reference to some general mechanism that makes such successful exchange possible. As we have seen, the "pragmatic" view is thus considered defective because, as Black contends, "it supplies no insight into how metaphor works." To this charge, my representatives of the "pragmatic" view can only plead guilty, as they want to deny the very existence of what they are being asked to explain. For what the "interactionist" criticism assumes is that meaning (metaphorical and otherwise) is something generated by rules, which language-users must adhere to if they are to be sure they are uttering sense. Against this assumption, Donald Davidson has argued that communication by speech is simply a matter of understanding and making oneself understood. As such, we can make sense of "the idea of two speakers "having the same language" by saying that they tend to converge on passing theories"11 for speech transactions — where a "passing theory" refers only to the particular beliefs or expectations each has about how the other will react to a range of utterances. Successful communication takes place to the extent that such reciprocal beliefs or expectations gradually converge. While this may seem intuitively obvious, it is to refuse the traditional picture of language as a medium of representation, and along with it, the search for a general theory of meaning. Once we have done so, writes Davidson:

  • 12 Ibid., p. 444-46.

[...] we have abandoned not only the ordinary notion of a language, but we have erased the boundary between knowing a language and knowing our way around in the world generally. For there are no rules for arriving at passing theories....12

17In these terms, the distinction between the literal and the Figurative use of language can be understood as just the difference between familiar and unfamiliar noises, or more precisely, between noises that (for the particular speech context in which they are uttered) are associated with patterns of behaviour that are generally predictable, and noises that (again, in a given context) are so strikingly new to us that we begin looking for regular associations with other behaviour. Richard Rorty has sought to defend such a view, writing that

  • 13 Richard Rorty, Contingency, irony, and solidarity (Cambridge, 1989), p. 18.

To have a meaning is to have a place in a language game. Metaphors, by definition, do not [....] Tossing a metaphor into a conversation is like suddenly breaking off the conversation long enough to make a face, or pulling a photograph out of your pocket and displaying it, or pointing at a feature of the surroundings, or slapping your interlocuter's face, or kissing him [....] All these are ways of producing effects on your interlocuter or your reader, but not ways of conveying a message. To none of these is it appropriate to respond with "What exactly are you trying to say?" If one had wanted to say something — if one had wanted to utter a sentence with a meaning — one would presumably have done so. But instead, one thought that one's aim could be better carried out by other means.13

  • 14 Cf. Davidson, op. cit., p. 200: "there are no unsuccessful metaphors, just as there are no unfunny (...)

18It is thus noteworthy that for this type of "pragmatic" account, so-called "dead metaphors" — formerly unfamiliar noises that have, over time, been successfully fixed within a generally predictable pattern of behavior — are not metaphors at all, for they have thus found a place in a language game. Conversely, for the "interaction" view of metaphor, we notice that any account of the particular "interaction" mechanism in question seems to require examples of long-dead metaphors (such as the trite examples I mentioned earlier), as paradigm instances of metaphorical assertion and meaning. On behalf of the "pragmatic" view, then, one might say that metaphors are somewhat like jokes and frogs: though initially quite provocative, they tend to die when dissected.14

  • 15 Cf. Richard Rorty, Contingency_irony and solidarity, op. cit., p. 21.
  • 16 Ibid., p. 9

19Needless to say, this idea of linguistic exchange as mere "convergence on passing theories" has met with significant opposition. In particular, adversaries have expressed concern for the fate of truth and the foundational enterprise. To give up the idea of general rules, or criteria for meaning, they claim, is to fall victim to a dangerous relativism, where we lack any means to defend truth and denounce falsity. Rorty has responded by urging that the foundationalist idea of objectivity be replaced with a pragmatic conception of intersubjectivity. Indeed, he sees such a development as an inevitable consequence of his central anti-foundational argument, which holds that since truth is a property of sentences, and sentences depend for their existence upon vocabularies, and vocabularies are constructed by human beings over time in relation to the practices and purposes of their particular communities, then so is truth.15 Recognizing this, we should stop seeking absolutes, and yield to the pragmatic and democratic impulse to let our public standards of reason, truth and justice be determined by no more than the provisional consensus of the community, in free and open speech encounters. This impluse would rely upon "a picture of intellectual and moral progress as a history of increasingly useful metaphors rather than of increasing understanding of how things really are."16 In these terms, to do philosophy would be only

  • 17 Ibid.

to redescribe lots and lots of things in new ways, until you have created a pattern of linguistic behaviour which will tempt the rising generation to adopt it, thereby causing them to look for appropriate new forms of nonlinguistic behaviour, for example, the adoption of new scientific equipment or new social institutions."17

  • 18 Cf. in particular, Cornel West, "The Politics of American Neo-Pragmatism," in John Rajchman and Co (...)

20This program remains quite controversial; significant problems have been well detailed in the work of many able critics.18 It is evident, for instance, that our everyday speech encounters are never so free and open as pragmatic and democratic ideals would have it. This picture of "intellectual and moral progress" seems to suggest that new and original redescriptions for the advancement of our goals and interests are always freely and readily accessible. But this surely overlooks our inescapable relation at all times to some particular social and linguistic community, whose norms and roles constrain our possibilities; indeed, to assume otherwise is to revert back to a version of the wholly-detached cogito of the traditional model that the pragmatist seeks to overcome. A complete account of what it means to have a language would have to consider the various ways in which the goals and interests of individual speakers are themselves constituted by existing linguistic practice.

21Yet herein lies the promise of the "pragmatic" account of metaphor. In placing metaphor beyond the reach of semantics, as an act of language in defiance of the semantic conventions of existing linguistic practice, the "pragmatic" view testifies to the strategic, even subversive value of failed communication. As such, metaphor is a provocation, a selective moment of non-sense that effectively displays and implicitly calls into question (certain of) the constraints imposed by convention, and in turn, the particular goals and interests that convention endorses in any local speech context. With metaphor, we contest the popular meanings of dominant discourse, and move toward more free and open speech encounters.


1 Most notably, in the fear of the seductive, irrational power of figurative language, which was thought to pose a threat to the essential virtues of temperance and moderation. Consider, for instance, Plato's repudiation of poetry and 'mimesis' in the Republic (Book X) and Aristotle's cautions against the inappropriate use of metaphor in his Rhetoric (1405) and Topics (139). This point is well covered hy Mark Johnson in his introduction to the collection of essays he has edited under the title Philosophical Perspectives on Metaphor (Minnesota, 1981).

2 In particular, in Thomas Hobbes (Leviathan 1,5) and John Locke (An Essay Concerning Human Understanding, IIΙ-Χ, 34). Cf. Johnson, Ibid.

3 Cf. in particular, Ogden and Richards, The Meaning of Meaning, New York: Harcourt-Brace, (8th ed.) 1946.

4 This celebrated term was coined by I.A. Richards in his seminal work, The Philosophy of Rhetoric, (Oxford, 1936).

5 Cf. Max Black, "Metaphor," in his Models and Metaphors (Cornell, 1962); Monroe Beardsley, "Metaphorical Senses," in Nous 12, no. l (1978): 3-16, and "The Metaphorical Twist," in Philosophy and Phenomenological Research 22 (1962): 293-307; Paul Henle, "Metaphor," in Johnson, op. cit.; George Lakoff and Mark Johnson, Metaphors We Live By (Chicago, 1980); and Nelson Goodman, Languages of Art (Hackett, 1976).

6 Cf. Thomas Kuhn, The Structure of Scientific Revolutions (Chicago, 1962), Paul Feyerabend, Science in a Free Society (New Left Books, 1978) and Mary Hesse, Models and Analogies in Science (London, 1963).

7 This forms a part of Donald Davidson's argument in "What Metaphors Mean," reprinted in Johnson, op. cit., pp. 200-220 (originally published in Critical Inquiry 5, no. 1 (1978): 31-47).

8 Ibid., p. 202.

9 Max Black, "How Metaphors Work: A Reply to Donald Davidson," in Sheldon Sacks, On Metaphor (Chicago, 1979), p. 189.

10 Cf. Donald Davidson, "What Metaphors Mean," in Mark Johnson ed., Philosophical Approaches to Metaphor (Minnesota, 1981), especially pp. 217ff., and Richard Rorty, "Unfamiliar Noises: Hesse and Davidson on Metaphor," in his Objectivity, Relativism and Truth (Cambridge, 1991), especially pp. 163ff.

11 Donald Davidson, "A Nice Derangement of Epitaphs," in Ernest LePore, ed. Truth and Interpretation (Blackwell, 1986), p. 445.

12 Ibid., p. 444-46.

13 Richard Rorty, Contingency, irony, and solidarity (Cambridge, 1989), p. 18.

14 Cf. Davidson, op. cit., p. 200: "there are no unsuccessful metaphors, just as there are no unfunny jokes," and also Richard Rorty, "Unfamiliar Noises: Hesse and Davidson on Metaphor," p. 167 in his Objectivity, Relativism and Truth (Cambridge, 1991): "you may not have to kill a platypus to get a satisfactory theory of how it works, but you do have to kill off a metaphor to get a satisfactory theory of how it works."

15 Cf. Richard Rorty, Contingency_irony and solidarity, op. cit., p. 21.

16 Ibid., p. 9

17 Ibid.

18 Cf. in particular, Cornel West, "The Politics of American Neo-Pragmatism," in John Rajchman and Cornel West, eds., Post-Analytic Philosophy (Columbia, 1985): 259-275; Nancy Fraser, "Solidarity or Singularity? Richard Rorty between Romanticism and Technocracy," in Alan Malachowski ed., Reading Rorty (Basil Blackwell, 1990): 303-321; Charles Taylor, "Rorty in the Epistemological Tradition," also in Malachowski, pp. 257-275; and Tom McCarthy, "Private Irony and Public Decency: Richard Rorty's New Pragmatism," and "Ironist Theory as a Vocation: A Response to Rorty's Reply," in Critical Inquiry 15-16 (1990): 355-370 and 644-655 respectively.


Université de Tours

© Presses universitaires François-Rabelais, 1993

Conditions d’utilisation :

Cette publication numérique est issue d’un traitement automatique par reconnaissance optique de caractères.


Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search