“…If relevance theory is right, then it offers a solution to the rhetorician’s dilemma, a way of being precise about vagueness, of making literal claims about metaphors and ironies, without abandoning any of the Romantics’ intuitions. However, rhetoricians could not adopt this solution without jeopardizing the very foundations of rhetoric. For what this solution implies is that metaphor and irony are ordinary exploitations of basic processes of verbal communication, rather than devices based on codified departures from the ordinary use of language. Moreover metaphor and irony exploit quite different basic processes and are more closely related, the former to loose talk, the latter to a variety of echoic uses, than to one another. The very notion of a trope is better dispensed with. If so, then rhetoric has no subject matter to study, or to teach…” [PDF version]
RHETORIC AND RELEVANCE
Dan Sperber & Deirdre Wilson
Paradox and dilemma
The student of rhetoric is confronted with a paradox and a dilemma. We will suggest a solution to the dilemma, which, however, will render the paradox even more blatant.
Let us begin with the paradox. Rhetoric took pride of place in formal education for two millenia and a half. Its very rich and complex history deserves being studied in detail, but it could also be compressed in a few sentences. Indeed, the same substance was inculcated by eighty generations of teachers to eighty generations of pupils. If a general tendency can be discerned, it consists in a mere narrowing down of the subject matter of rhetoric: one of its five branches, elocutio, the study of figures of speech, progressively displaced the four other branches, and in some schools, became identified with rhetoric tout court. (We ourselves will be guilty of this and several other simplifications). This narrowing down was not even compensated by a theoretical deepening. Fontanier’s Les Figures du Discours does not radically improve on Quintilian Institutio Oratoria in spite of the work of sixty generations of scholars in between.
The combination of such institutional success with such intellectual barrenness is puzzling. Moreover, the history of rhetoric cuts across major social changes: the eighty generations of pupils concerned have had little in common: Greek politicians, Roman lawyers, medieval clerks, Renaissance aristocrats, and nineteenth century bourgeois were taught the same stuff. This extraordinary institutional resilience of an otherwise rigid rhetoric turns puzzle into paradox.
Then came the Romantics, and, or so it seemed, the end of rhetoric. The most scathing criticism addressed by the Romantics to classical rhetoric concerned the treatment of metaphor, irony, and other figures of speech. In classical rhetoric, figures were seen as ornaments added onto a text, which made it more pleasant and hence more convincing, without however altering its content. Tropes in particular, it was said, achieve this ornamental effect by replacing a dull literal expression of the author’s thought by a more attractive figurative expression, that is, by an expression the literal meaning of which is set aside and replaced by a figurative meaning.
A mother says to her child:
You are a piglet!
A rhetorician would analyze “piglet” in this context as a metaphor figuratively meaning “dirty child”. The figurative meaning of the metaphorical expression is identical to the literal meaning of the plain expression it replaces. Generally speaking, on this view, every figure has a non-figurative paraphrase.
Or the mother might say:
You’re such a clean child!
A rhetorician would analyze “clean child” as an irony figuratively meaning, again, “dirty child”. The figurative meaning of the metaphorical or ironical expression is claimed to be identical with the literal meaning of the ordinary expression it replaces.
Against the notion of a figure as a mere ornament, the Romantics maintained that a felicitous trope cannot be paraphrased. Thus Coleridge argues that the “infallible test of a blameless style” is:
its untranslateableness in words of the same language without injury to the meaning. Be it observed, however, that I include in the meaning of a word not only its correspondent object but likewise all the associations which it recalls. (1)
Even though a modest stylist, the mother calling her child a piglet achieves some unparaphrasable effects: for instance she appears more forgiving than if she had called him a dirty child. Or saying: “You’re such a clean child!”, the mother conveys not only that the child is dirty but also-with a light touch that explicit paraphrase would loose-that he ought to be clean.
The Romantic critics were unquestionably right in pointing out the richness and importance of those effects of figures of speech which are not maintained under paraphrase. These effects were merely mentioned by classical rhetoricians; they were not described, let alone explained. Still, for all their well-taken criticisms and subtle observations, the Romantics remained content to talk about metaphor in metaphorical terms and proposed no explicit theory either; if anything, they have cast doubt on the very possibility of developing a non-metaphorical theory of metaphor, by rejecting altogether the notion of a literal meaning – the “proper meaning superstition” as I.A. Richards calls it.(2)
The Romantic criticisms have been generally accepted by the contemporary academic heirs of the rhetoricians. It has become almost commonplace that, in Jonathan Culler’s words,
one can never construct a position outside tropology from which to view it; one’s own terms are always caught up in the processes they attempt to describe.(3)
This academization of Romanticism allowed – more paradox – the resurgence of classical rhetoric. For if, as Victor Hugo said, scorning rhetorical typologies, “words are equal, free, of age,” (4) then the words of rhetoric itself are inferior to none and can be freely used. And so we find, in modern literary studies, a Romantic use of rhetorical terms: they are not endowed with a “proper meaning” anymore, but they suggest subtle distinctions and evoke scholarly sophistication and historical depth.
Let us state our bias: we see nothing wrong with a free use of all the resources of language – poetic use or rhetorical jargon included – to interpret particular experiences or particular texts. We do not believe, however, that interpretations of particulars generalize into proper theories. We are aware that the post-Romantic, post-structuralist sophisticated believe even less in proper theories than they do in proper meanings. However we are not sophisticated. Developing a theory encompassing the kind of phenomena classical rhetoric tried to describe, with even greater explicitness than it tried to achieve, seems to us a worthwhile pursuit.
And so, the dilemma: it seems we must either hold on to the relative rigor of a rhetorical approach and miss an essential -maybe the essential – dimension of language use, or start from the Romantic intuition that linguistic creativity does not reduce to a mere combinatorics, and forever forsake scientific ambition. More specifically: on the one side we have the view that an utterance or a text has a literal meaning which, in the absence of indications to the contrary, it is presumed to convey; that view allows a neat definition of semantics as the study of literal meanings, and of tropology as the study of departures from literal meanings. On the other side, we have the view of meaning as mishmash in motion, analytically unappealing, but true to life.
Note that both classical rhetoricians and their Romantic critics take as self-evident that, if there is such a thing as literal meaning, then utterances come with a presumption of literalness. We disagree. You can keep a notion of literal meaning and its analytical usefulness, and drop the presumption of literalness and its implausibility, provided you introduce a presumption of relevance. This, we will argue, makes it possible to reconcile theory and intuition.
The rhetorician’s dilemma is a special case of an even more fundamental problem in the study of human communication. From ancient rhetoric to modern semiotics, communication has been explained as a process of encoding carried by the communicator, followed by a process of decoding carried by the audience. The existence of a common code has been taken as a necessary and essentially sufficient condition for communication. The code model of communication has an appealing simplicity. However, it has become more and more manifest that human communication cannot fully be explained in terms of that model.
Given a rich enough code – and human languages are certainly rich enough in the required sense – whatever can be encoded in some way can be encoded in some other way, i.e. can be paraphrased. The fact, pointed out by the Romantics, that communication achieves some unparaphrasable effects strongly suggest that more is communicated than is actually encoded. Moreover, as modern pragmatics has repeatedly shown, communicators manage to convey information they could have explicitly encoded without however encoding it, by making it somehow an implicit part of their communication.
How are unencodable poetic effects, and encodable but unencoded “implicatures” communicated? Modern pragmatics has an answer for implicatures: they are inferred by the audience on the basis of what can be decoded, of contextual information, and of general expectations regarding the communicator’s behavior. Inference is viewed, then, as an effort-saving partial replacement for encoding and decoding. However, the special flavor and uses of implicit communication on the one hand, poetic effects on the other hand, remain as mysterious in modern pragmatics as they were in classical rhetoric.
In Relevance: communication and cognition, we have developed a novel approach to human communication grounded in a general view of cognition, an approach which, we will try to show, helps solve the rhetorician’s dilemma and its modern pragmatic version.
Instead of viewing the fully coded communication of a well-defined paraphrasable meaning as the norm, we treat it as a never-encountered theoretical limit. Instead of treating a mix of explicitness and implicitness, of paraphrasable and unparaphrasable effects as a departure from the norm, we take it as plain normal communication. We define communication not as a process of duplication of meaning from the communicator’s into the addressee head, but as a more or less controlled modification of the mental landscape – the “cognitive environment” as we call it – of the audience by the communicator, achieved in an intentional and overt way.
A person’s cognitive environment can be modified by the addition of a single piece of new information, but equally well by a diffuse increase in saliency, or in plausibility, of a whole range of assumptions, yielding what will subjectively be experienced as an “impression”. Between the communication of specific information and that of an impression, there is, on our approach, a continuum of cases. Instead, then, of contrasting “meaning” and “rhetorical effects”, or “denotation” and “connotation”, we include both under a single unitary notion of “cognitive effects”. The communication of such cognitive effects is essentially inferential. Decoded meaning structures are not directly adopted by the audience as thoughts of their own; they serve rather as very rich evidence which largely unconscious inferential processes can exploit in order to arrive at comprehension proper. Exploit, but how? Under what guidance? This is where considerations of relevance come in.
Human information processing requires some mental effort and has some cognitive effect. The effort is one of attention, of memory, and of reasoning. The effect is to alter the individual’s cognitive environment, by adding new beliefs, cancelling old ones, or merely altering the saliency or strength of beliefs already held. We may characterize a comparative notion of relevance in terms of effect and effort as follows:
(a) Everything else being equal, the greater the cognitive effect achieved by the processing of a given piece of information, the greater its relevance for the individual who processes it.
(b) Everything else being equal, the greater the effort involved in the processing of a given piece of information, the lesser its relevance for the individual who processes it.
We claim that humans automatically aim at maximal relevance, i.e. maximal cognitive effect for minimal processing effort. This is the most general factor which determines the course of human information processing. It determines which information is paid attention to, which background assumptions are retrieved from memory and used as context, which inferences are drawn.
To communicate is, to begin with, to claim someone’s attention. People won’t pay attention to a phenomenon unless they expect it to be of sufficient relevance to them. Hence to communicate is to imply that the phenomenon displayed, for instance the linguistic utterance, is worth the audience’s attention. Any utterance addressed to someone automatically conveys the presumption of its own relevance. This fact, we call the principle of relevance.
A communicator puts a conceptual structure in the head of her audience: by means, for instance, of a mimic which evokes what it resembles, or by means of an utterance which is automatically decoded into a semantic representation. If the presumption of relevance conveyed by such an act of communication is not mistaken, then the effort required of the audience for constructing this conceptual structure is not expended in vain. That is, this structure will determine enough cognitive effects to justify the effort: effort promises effect. (How well is the promise kept is another matter.)
The task of the audience, then, is to identify the effects the communicator could have foreseen and on the basis of which, she could guarantee the relevance of her communication. Those effects which are (or may have seemed to the communicator) sufficient to make the signal adequately relevant to the audience are intended effects. They constitute together an interpretation consistent the fact that a presumption of relevance has been communicated, i.e., in our jargon, an interpretation consistent with the principle of relevance. Consistency with the principle of relevance is the guiding criterion in the comprehension process. (Note incidentally, that the interpretation selected according to this criterion is not the most relevant one, but one sufficiently relevant to confirm the presumption of relevance.)
In Relevance, we work out in detail how the principle of relevance guides inference and allows the identification of the explicit and implicit content of an utterance. Here, we will merely indicate how it gives rise to metaphorical or ironical interpretations.(5)
Literalness, looseness, metaphor
If verbal communication were guided by a presumption of literalness, then every second utterance should be treated as an exception. If it is guided by a presumption of relevance, or, more precisely by a criterion of consistency with the principle of relevance, then there are no exceptions: the interpretation of every successful act of communication, utterances in particular, meets this criterion.
At a party in San Francisco, Marie meets Peter. He asks her where she lives, and she answers:
I live in Paris.
It so happens that Marie lives in Issy-les-Moulineaux, a block away from the city limits of Paris. Her answer is literally false, but not blatantly so. If Peter presumed literalness, he will be misled.
In ordinary circumstances, however, Mary’s answer is quite appropriate, and not misleading. How come? This is easily explained in terms of relevance theory. A speaker wants, by means of her utterance, to make her hearer see as true or probable a certain set of propositions. Suppose these propositions are all quite easily derivable as implications of a proposition Q. Q however has also other implications whose truth the speaker does not believe and does not want to guarantee. Nevertheless, the best way of achieving her aim may be for her to express the single proposition Q, as long as the hearer has some way of selecting those of its logical and contextual implications that the speaker intends to convey and of ignoring the others.
Our claim is that such a selection process is always at work, is part, that is, of the understanding of every utterance. Whenever a proposition is expressed, the hearer takes for granted that some subset of its implications are also implications of the thought being communicated, and aims at identifying this subset. He assumes (or at least assumes that the speaker assumed) that this subset determines sufficient cognitive effects to make the utterance worth his attention. He assumes further (or at least assumes that the speaker assumed) that there was no obvious way in which achieving these effects might have required less effort. He aims at an interpretation consistent with these assumptions, i.e. consistent with the principle of relevance. When this criterion determines a single interpretation (or closely similar interpretations with no important differences between them) communication succeeds.
In our example, Peter will be able to infer from Mary’s answer quite an amount of true or plausible information: that Marie spends most of her time in the Paris area, that Paris is familiar to her, that she lives an urban life, that he might try to meet her on his next trip to Paris, and so on. It it such cognitive effects which make Marie’s utterance sufficiently relevant to be worth his processing effort, in a way Marie manifestly may have anticipated. So, Peter is entitled to assume that Mary intended him to interpret her utterance in this way. Peter would be mislead by Marie’s answer only if he were to conclude from it that she lives within the city limits of Paris. However it is clear that Marie had no reason to assume that Peter would have to derive such a conclusion in order to establish the relevance of her utterance. Therefore her utterance does not warrant it.
Utterances such as Marie’s answer are, typically, loosely understood. This loose understanding does not follow from a strictly literal interpretation having been first considered and then discarded in favor of looseness: in the above example, Peter would have no ground to discard the literal interpretation to begin with. In fact, at no point is literalness presumed.
An utterance may be literally understood, but only at the end rather than at the beginning of the comprehension process, and only when relevance requires it. Suppose Marie is asked where she lives, not at a party in San Francisco, but at an electoral meeting for a Paris local election. If she answers that she lives in Paris, the proposition expressed will itself be crucially relevant, hence the utterance will be understood literally, and Marie will have lied.
The same procedure – derive enough cognitive effects to make up an interpretation consistent with the principle of relevance – yields in some cases a literal interpretation, in others a loose one. In other cases still, it yields a figurative interpretation. A writer, for instance, describes:
Clarissa’s face was a perfect oval.
If there existed a presumption of literalness, the reader would first have to consider the literal interpretation of that utterance, and then reject it, since it is common knowledge that no human face is a perfect oval. The reader would then look for a figurative interpretation, and in this case somehow recognize an hyperbole: what the author presumably means is that Clarissa’s face was remarkably close to being oval. That it should be interpreted as a case of hyperbole, rather than, say, irony, is obvious, but why this is obvious is not obvious at all in the classical approach.
In terms of relevance theory, the reader does not first consider and then reject the hypothesis that the writer meant to assert that Clarissa’s face was a perfect oval. He just uses the idea expressed as a source of cognitive effects: he builds a mental representation of Clarissa’s face which contains enough of the implications of the idea of its being a perfect oval – the general shape, a striking degree of regularity and symmetry – to justify the presumption of relevance. The utterance so understood produces enough effects for a minimum of effort. Had the author spelled out such an interpretation instead of relying on her readers’ abilities, the effect would have been roughly similar, but the processing effort would have been much greater, hence the relevance would have been lesser.
Let us turn back to our example of a mundane metaphor:
Mother to child: You are a piglet.
While calling somebody a pig is quite standard – the metaphor is “lexicalized” -, calling a child a piglet requires of the hearer some extra processing effort which justifies him in searching for added effect. For instance, young animals are endearing, even when the adult of the species are not; so, the child may feel encouraged to derive not only the obvious contextual implication that he is dirty, but also the further implication that he is, nevertheless, endearing.
The wider the range of cognitive effects and the greater the degree of initiative left to the hearer (or reader) in constructing them, the more creative the metaphor: “piglet” is, if only marginally, more creative than “pig”. In the richest and most successful cases, the hearer can go beyond just exploring the immediate context and the directly invoked background knowledge, accessing a wider area of knowledge, entertaining ad hoc assumptions which may themselves be metaphorical, and discovering more and more suggested effects. The result is a quite complex picture, for which the hearer has to take a large part of the responsibility, but the discovery of which has been triggered by the speaker (or writer).
Take Prospero’s words to his daughter Miranda:
The fringed curtains of thine eyes advance
And say what thou see’st yond.
(Shakespeare: The Tempest I ii)
Coleridge argues, against Pope and Arbuthnot, that these words should not be taken as equivalent in meaning to “Look what is coming yonder”. They are uniquely appropriate to the characters and situation:
Prospero sees Ferdinand and wishes to point him out to his daughter not only with great but with scenic solemnity… Something was to appear to Miranda on the sudden, and as unexpectedly as if the hearer of a drama were to be on the stage at the instant when the curtain is elevated… Turning from the sight of Ferdinand to his thoughtful daughter, his attention was first struck by the downcast appearance of her eyes and eyelids…(6)
Coleridge comments are indeed illuminating, but they invite an objection and a question. The objection is that it is possible to appreciate Shakespeare’s metaphor without understanding it exactly as Coleridge does. The question is how such an understanding is arrived at.
Our way of answering the question also takes account of the objection. In order to understand Prospero’s metaphor the hearer must take into account his knowledge of the appearance of eyelids and of curtains, theater curtains in particular. But this is not enough, for merely retaining the most obvious implication that Prospero is telling Miranda to raise her eyelids would result in an interpretation requiring too much effort for too little effect. A more attentive hearer will invest a little more effort still and get much more effect. This extra effort may consist in the hearer creating a metaphor of his own, for instance Coleridge’s metaphor of the hearer of a drama being brought on stage, and adopting some of the joint implications of Prospero’s metaphor and his. In such a process, the hearer is taking a large part of the responsibility in the conclusions he arrives at. Therefore different hearers with different background knowledge and different imaginations will follow somewhat different routes. They are all nevertheless encouraged and guided by the text, and they all proceed by exploring the text’s implications as relevantly as they can.
How does this approach to metaphor compare with the classical and Romantic accounts? In many ways we are on the Romantic side. If we are right, metaphors are based on fundamental and universal psychological mechanisms. They are in no sense departures from a norm, or, as modern pragmatists would have it, breaches of a rule or maxim of communication. We also reject the classical claim that tropes in general, and metaphor in particular, have a purely decorative function. For us, as for the Romantics, tropes have a genuine cognitive content which, particularly with the more creative metaphors, is not paraphrasable without loss. This content we have proposed to analyze in terms of an wide array of weak cognitive effects whose recovery is triggered by the speaker, but whose content the hearer actively helps to determine.
Despite our general sympathy with the Romantic view of metaphor, we differ sharply from the Romantics on the nature of language and meaning. For us, the existence of loose uses does not mean that language is irremediably vague, and the pervasiveness of metaphor does not make it an aspect of word and sentence meaning. Similarly, the fact that hearers approach utterances without set expectations as to their literalness, looseness or metaphorical character does not mean that these cannot be distinguished. The distinction, however is one of degree rather than of nature. Words and sentences have a literal meaning, but that meaning is is an instrument of communication rather than its content. What hearers expect is that the literal meaning of an utterance will help them infer with a minimum of effort the thought that the speaker intends to convey. This expectation itself derives from, and is warranted by, a more basic expectation of relevance, which is automatically encouraged by any act of communication.
Echo and irony
Just as we deny that the literal meaning of an utterance constitutes its preferred interpretation, we challenge the view that the mood of an utterance (declarative, imperative, interrogative, etc.) determines its speech act type (assertion, request, question, etc.). A sentence mood encodes not an illocutionary force, but a more abstract and by itself inconclusive piece of evidence on the speaker’s intentions. Thus, the same imperative sentence might be used to express a request:
Mother to son: Be a good boy!
to express the antecedent of a conditional assertion:
Be a good boy and you will become a good man.
to report another person’s utterance:
Girl: What did Mommy tell you?
Boy: Be a good boy!
to echo a preceding utterance:
Mother to son: Be a good boy!
Son to mother: Be a good boy! Be a good boy! I am being a good boy!
to echo (in an extended sense of the word) another person’s utterance (or thought, or one’s own past thoughts or utterances, or public opinion, etc.), however removed in time:
She: What kind of an upbringing did you have?
He: Oh, You know, be a good boy! and all that sort of things.
Any utterance will be taken to have whatever illocutionary force is required in order to arrive at an interpretation that is relevant as expected.
What makes an echoic utterance relevant? An echoic utterance indicates to the hearer that the speaker is paying attention to a representation (rather than to a state of affairs); it indicates that one of the speaker’s reasons for paying attention to this representation is the fact that it has been entertained (and possibly expressed) by someone; it also indicates the speaker’s attitude to the representation echoed. An echoic utterance achieves relevance by making it possible for the hearer to recognize, and perhaps to emulate, the speaker’s interest in, and attitude to, somebody else’s thought.
The speaker may express any one of an indefinite variety of attitudes to the representation she echoes. The attitude expressed may be one of approval or even of reverence, as when popular wisdom or holy scriptures are echoed by a speaker who hopes thereby to command greater acquiescence than she would if she were merely to speak in her own voice. The attitude may be one of surprise or even disbelief as when a speaker echoes some amazing statement. There is, too, an attitude, or rather a range of attitudes, which may properly be called ironical: the representations echoed with such an ironical attitude are worth paying attention to because of their very inappropriateness, falsity, or even absurdity, and because of the fact that, notwithstanding, they have been or are being held by some as true beliefs or as realistic expectations.
Irony, then, rests on the perception of a discrepancy between a representation and the state of affairs that it purports to represent. Such a characterization encompasses all varieties of irony from Socratic irony (where the discrepancy lies between the self-confidence and the sense of superiority that Socrates allows his interlocutor to indulge in, and the true rapport de force), to Romantic irony (where all representations-and in particular the poet’s own ambitions-are seen as illusory).
When verbal irony is seen as a the use of a linguistic expression in order to convey the opposite of its literal meaning, the usefulness of that rhetorical device and its relationship to irony as an attitude are quite mysterious. The mystery dissolves when verbal irony is seen as the echoing of an utterance or a thought to which an ironical attitude is taken.
In verbal irony, the ironical attitude is implicitly rather than explicitly conveyed. As a result, the hearer who recognizes and shares that ironical attitude will feel that the speaker and himself stand above the victims of the irony: those who accept the representation echoed at face value. In the special case where the representation echoed is a belief or an expectation of the hearer himself, or a norm that the hearer has failed to conform to, the hearer is not given the option of sharing with the speaker a sense of superiority: the hearer is himself the victim of the irony.
Thus the mother who says ironically:
You’re such a clean child!
is evoking the discrepancy between the norm of cleanliness that the child is supposed to meet and his actual appearance. That, by the way, explains why there are much fewer situations where it would be appropriate for the mother to say ironically to a clean child:
You’re such a dirty child!
Unless the child had been expected to be dirty, there would be no antecedent representation to evoke. Irony is moralistic, not because, as Muecke suggests, “all literature is moral”,(7) but because an easy way of achieving relevance by means of irony consists in echoing moral norms right when they are being violated.
Echoic utterances are a well-defined type. Ironical utterances, on the other hand, are a loosely defined sub-class of utterances of the echoic type: ironical attitudes are many; they shade off imperceptibly into other attitudes, anger or aloofness, for instance. Because of that, the same representation can be echoed several times in the same discourse, but with a changing attitude: the utterance type and content remain the same but the disposition evolves and relevance is renewed.
Four times, Shakespeare’s Mark Antony repeats “Brutus is an honorable man.” The first time, all agree, his audience is not intended to take these words ironically. The fourth occurrence, on the other hand, is blatantly sarcastic. What happens in between? Wayne Booth, however subtle an interpreter, is hampered by the classical model of irony, however much enriched:
“For the populace, When Mark Antony says for the first time that ‘Brutus is an honorable man,’ the invitation is simply to agree or disagree. If any of them takes the further step of judging that Mark Antony does not believe what he says, they will probably decide that he is a liar, not an ironist…”(8)
Booth envisages only two alternatives: either Mark Antony is making a literal assertion, or else he is being ironical, and since irony is excluded at that stage, then a literal assertion it must be (and hence a lie). For lack of intermediate forms between literalness and irony, a total reversal of meaning must take place at the second or third occurrence of “Brutus is an honorable man.” In order to give a richer account of the passage than classical rhetorical tools permit, Booth must resort to metaphor: Mark Antony’s hearers, he writes,
“do not just translate into the opposite conclusion: ‘Brutus is really dishonorable.’ They are forced to make the ironical leap in order to stand with Mark Antony on his platform (a good deal higher, one might say, than the literal one on which he stands) and they must feel themselves drawn to his conclusions by the acrobatic skill which they themselves have shown.”(9)
Relevance theory provides a more powerful analytical tool and thus permits a more fine-grained explicit account of the rapidly evolving mood expressed by Mark Antony. When he first says that Brutus is an honorable man, we do not have to describe him as asserting his own opinion, and even less as asking his audience to agree. They are already on Brutus’s side (“‘Twere best he speak no harm of Brutus here,” a citizen cautions). What Mark Antony does is echo their opinion with what they must take, at this stage, to be a conciliatory attitude. Considerations of relevance cause his audience to understand Mark Antony, not as telling them, but as granting them that Brutus is honorable (and granting what you do not believe is not lying and may even be the moral thing to do).
Then, as he gives his audience reasons to renounce the favorable opinion of Brutus which he repeatedly echoes, Mark Antony conveys a more and more scornful attitude to that opinion (and to Brutus himself who would like to be thought of as honorable). The utterance type is the same throughout: it is echoic. Only the attitude changes. The echoic character of the utterance and the speaker’s evolving attitude are not encoded and therefore cannot be decoded; the audience recognizes them by looking for a relevant interpretation.
Again, we side with the Romantics: irony is not an occasional device, it is a fundamental attitude. Unlike the Romantics, however, we believe that the expression of this attitude by linguistic means can be analyzed and indeed explained without recourse to further tropes and by means of an unambiguous and testable model.
The relevance of rhetoric
If relevance theory is right, then it offers a solution to the rhetorician’s dilemma, a way of being precise about vagueness, of making literal claims about metaphors and ironies, without abandoning any of the Romantics’ intuitions. However, rhetoricians could not adopt this solution without jeopardizing the very foundations of rhetoric. For what this solution implies is that metaphor and irony are ordinary exploitations of basic processes of verbal communication, rather than devices based on codified departures from the ordinary use of language. Moreover metaphor and irony exploit quite different basic processes and are more closely related, the former to loose talk, the latter to a variety of echoic uses, than to one another. The very notion of a trope is better dispensed with. If so, then rhetoric has no subject matter to study, or to teach.
1. Samuel Taylor Coleridge, Biographia Literaria, 1907, London: Oxford University Press, edited by J. Shawcross, volume II, p. 115.
2. I.A. Richards, The philosophy of Rhetoric, 1936, London: Oxford University Press, p. 11.
3. Jonathan Culler, 1981, Ithaca: Cornell University Press, The Pursuit of Signs: Semiotics, Literature, Deconstruction, p. 209.
4. Victor Hugo, Oeuvres complètes: Poésie II, 1985, Paris: Robert Laffont, p.265.
5. See Dan Sperber & Deirdre Wilson, Relevance: Communication and Cognition, 1986, Cambridge, Mass.: Harvard University Press and Oxford: Blackwell. For a More detailed account of looseness and metaphor, see our “Loose Talk,” Proceeding of the Aristotelian Society (N.S.) Vol. LXXXVI, 1986, pp. 153-171. For a more detailed account of irony, see our “Irony and the use-mention distinction”, in Peter Cole (ed.) Radical Pragmatics, 1981, New York: Academic Press, pp. 295-318.
6. Samuel Taylor Coleridge, The Collected Works. Vol. 5-II: Lectures 1808-1819 on Literature, 1987, Princeton: Princeton University Press, pp.527-528.
7. D.C. Muecke, Irony 1970 London: Methhuen, p. 63.
8. Wayne C. Booth, A Rhetoric of irony, 1974, Chicago: The University of Chicago Press, p. 42.