The Marriage of Pragmatics and Rhetoric

A recent collection, Rhetorical Hermeneutics: Invention and Interpretation in the Age of Science (Gross and Keith 1996), testifies to a crisis in rhetorical theory and rhetorical criticism. Can an art whose origin is the Greek city-state and whose practice has been almost exclusively devoted to the production of speeches, be transformed into an art of interpretation, a hermeneutics of persuasion? The answers in the volume range from the profound skepticism of Dilip Gaonkar to the robust optimism of Deirdre McCloskey. In some cases, alternatives are offered: William Keith offers a programmatic solution whose origin is in reverse engineering; David Kaufer offers a detailed solution whose origin is in artificial intelligence. That no one calls on the achievements of a century of philosophy of language, beginning with Gottlob Frege, and including Bertrand Russell, J. L. Austin, Peter Strawson, Ludwig Wittgenstein, and Paul Grice, is a tribute perhaps to the insularity of the academy, the difficult-to-negotiate borders between disciplinary formations. The omission of Paul Grice is especially significant. In philosophy of language, no one ranks at this level of influence other than Austin; no one ranks above, with the exception of Wittgenstein. In the pragmatics of language, however, in the interactions of speakers and hearers, which will be our central concern, Grice reigns supreme. Though nearly a half-century old, his early essays have lost none of their hold on the intellect of scholars interested in communicative interchange. Though Grice never generalized his theory, he was never interested in anything less than a general theory: “I have stated my maxims [as if the purpose of talk exchanges] were a maximally effective exchange of information; this specification is, of course, too narrow, and the scheme needs to be generalized to allow for such general purposes as influencing or directing the actions of others” (1989, 28). It is this statement that opens the door to a union of pragmatics and rhetoric.

But it is a union with problems on both sides of the aisle. To reconstruct rhetoric as a cognitive theory, we must place inference at its center. As regards the canon of invention and the logos at its heart, this move seems natural; as regards “proofs” from emotion and character, and the canons of style and arrangement, this move seems counterintuitive. There is another problem as well: Gricean talk exchanges rule out misdirection, and a rhetorical theory that cannot account for misdirection will lack plausibility. Finally, though Grice’s theory concerns only a single speaker and hearer and Aristotle’s only a single speaker and an audience, we mean our reconstruction to apply generally.

Fortunately, we will be assisted in this effort of reconciliation by the labors of philosophers and classicists. A cognitive theory of rhetoric must acknowledge a debt, not only to Aristotle, but also to Chaim Perelman and Lucie Olbrechts-Tyteca; in our view, The New Rhetoric: A Treatise on Argumentation (1969) is the finest modern attempt to reconstruct Aristotle’s Rhetoric. The New Rhetoric integrates into a cognitive, inference-based theory, not only the inventive material of the first two books, but also much that Aristotle has to say about style and arrangement in book 3. But as its subtitle, A Treatise on Argumentation, hints, its treatment of Aristotle’s proofs from character is deficient, and its treatment of emotional proofs is virtually nonexistent. We think that this imbalance must be corrected; thanks to classicists William Fortenbaugh and Martha Nussbaum, and philosopher William Lyons, we think that it can be. All three see Aristotle’s theory of the emotions as cognitive, one in which inference is necessarily involved.

Classical rhetoric and Gricean pragmatics

According to Grice’s Cooperative Principle, “[E]ach participant recognizes in [talk exchanges], to some extent, a common purpose, or set of purposes, or at least a mutually accepted direction” (1989, 26). In such talk exchanges, moreover, all that is said is said according to meaning_NN. Summarizing his views on meaning_NN, Grice says: “[W]e may say that ‘A meant_NN something by x’ is roughly equivalent to ‘A uttered x with the intention of inducing a belief by means of the recognition of this intention’” (219). Grice’s conversational maxims are the guiding principles of such talk exchanges, each of whose utterances is an instance of meaning_NN. In accord with the maxims, Speaker S and Hearer H are expected to make their comments truthful, relevant, as perspicuous as possible, and exactly as informative as required. In this theory of talk exchanges, understanding is a matter of inference from what is said to what is meant.

Because persuasion is a kind of communicative interaction, Gricean theory should easily extend to cover it. But between Gricean theory and persuasion are differences apparently significant enough to make a responsible incorporation impossible. Since it is a condition of persuasion that there may be a discrepancy between the goals of S and H, the Cooperative Principle cannot be necessary except in some minimal form. Meaning_NN cannot be necessary either, since speakers may systematically mislead their audiences as to their intentions; indeed, persuasion may depend on the lack of recognition of such systematic misdirection. In addition, the conversational maxims cannot be said to apply, especially those relating to truthfulness (Quality) and perspicuousness (Manner), neither of which may be in the interest of S, who may easily prefer only the appearance of truthfulness and perspicuousness. Furthermore, Gricean talk exchanges consist of networks of propositions linked by implication, implicature, or presupposition. This has an unfortunate consequence for the description of persuasion, which often has as its goal or interim goal, not a proposition, but an alteration in emotional state or in the disposition to approve or disapprove of conduct or character. Finally, in persuasion, S and H need not alternate roles; indeed, it is a condition of many persuasive exchanges that H is relatively passive and may not even be entitled to a “turn.”

This last difference is only apparently a discrepancy. It is true that persuasive exchanges differ from Gricean exchanges in that they are not generally dialogical. But this is not an argument against continuous interchange between S and H; even in oratory, both S and H have roles that are active and continuous: S conveying meaning, H constructing meaning through inference. Moreover, when S fails to maintain the Cooperative Principle, H will not be slow to indicate to S that the persuasive situation no longer pertains. As in dialogue, S and H are bound together as the co-constructors of the interchange.

Other discrepancies are an artifact of the paradigmatic case of talk exchanges on which Grice bases his system: the exchange of information. Rhetorical exchanges are another special case of talk exchanges; they have a different purpose: persuasion. In persuasion, the initial agreement of H to be open to influence is in itself an instance of the Cooperative Principle. The introductory remarks of S in the typical classical exordium or introduction—remarks designed to obtain the attention, interest, and trust of the audience—are meant to carry this process one step further: to establish a beachhead of the Cooperative Principle as the foundation of continuing influence. In information exchanges, cooperation is presupposed; in persuasive exchanges, it must be established, strengthened, and maintained.

In persuasive exchanges also, Grice’s definition of meaning_NN needs to be altered from a general insistence on transparency of intent to a formulation that includes the possibility of misdirection and deception. To see how this might be managed, let us adapt a well-known example from John Searle’s Speech Acts, in which an American soldier intends to deceive his Italian captors, to convince them that he is German, not English. But he knows only one bit of German, a line of Goethe: Kennst du das Land wo die Zitronen blühn? This is not an example of Grice’s meaning_NN: the success of S’s utterance depends on H’s ignorance of S’s intention. The American soldier wants his captor to perceive that he is trying to communicate in German; in reality, he is trying to create the impression, falsely, that he is a German. In order to make these exchanges Gricean, we must recognize second-order intentions. As Schiffer puts it,“[I]t is worth noting that, should it be desirable, the definitions of ‘S meant that p’ and ‘S meant that A was to ψ, may easily be altered to accommodate both the primary cases and the ‘counter-suggestive’ cases: we simply require only that S utter x intending A to think S uttered X with those intentions specified in the definitions” (1972, 72). In Searle’s example, S wants H to be persuaded of his German nationality by means of H’s not understanding S’s intention to do so. So S utters x with the intention of inducing a false belief as to his true intention.

Although proposals such as Schiffer’s to amend Grice’s definition of meaning_NN are not problem-free, they point the way toward what we need in our reconstruction. What is important from our point of view is that, although the intentions inferred in deceptive communication are erroneous, they are still inferences that S intends H to make as a result of recognizing S’s first-order intention. This means that communicative causality still fully applies, that is, H’s belief is caused by what S says and by his inference from that utterance to its intention.

This altered definition means that the conversational maxim enjoining truthfulness applies to rhetorical exchanges only as an injunction that its appearance must be maintained. Persuasive exchanges also differ from informational exchanges in that the conversational maxim enjoining perspicuousness may not apply. Persuasion may depend on the systematic employment of vagueness (in which it is not clear what is meant) and ambiguity (in which more than one thing is meant). An expression such as “abortion permitted due to medical emergency” is intentionally vague if S has avoided specification of “medical emergency” in the interests of persuasion. If, however, S intends that one H understand medical emergency as life threatening and another as including the emotional state of the mother-to-be, then the expression is intentionally ambiguous. Of course, S must not convey the intent to be unclear; to do so would be fatal to persuasion. To the extent that vague and ambiguous exchanges are misleading, they conform to the revision of Grice’s meaning_NN that includes higher-order intentions.

But there are two discrepancies between informational and persuasive exchanges that are not so easily dealt with. The goal of Gricean information exchanges is always a propositional mental state. But persuasion often has as its goal or interim goal an alteration in an emotional state, or a mobilization of a value. It is not immediately obvious, however, that Grice’s notion of inference can be applied to so-called emotional “proofs” and “proofs” from character; surely, “proof” applied to emotion and character is a metaphoric extension that obscures more than it explains. Nonetheless, we shall show that, while beliefs in the form of propositions are never sufficient to account for our emotional states and our judgments of character, they are necessary. Thus, cognition is an essential part of such states and such judgments.

There is an additional problem. Although we can see how inference can apply to the canon of invention, it is difficult to see how it can apply to the canons of style and arrangement. Style—the systematic variation among syntactic and semantic choices—seems entirely divorced from the operation of reason in two senses: stylistic variations can apparently be apprehended without its aid and stylistic choices are apparently unconnected with the conduct of arguments. While the literature on the cognitive nature of the central trope of metaphor is extensive, no speech communication, argument, or classical scholar has, to our knowledge, incorporated these insights into their understanding of tropes: not Farrell, or Grootendorst, or Garver. Nor have these scholars integrated into their understanding Perelman and Olbrechts-Tyteca’s insight that a trope can be and often is part of the proof structure of rhetorical texts. But we shall show that, with the help of these insights, the bridge between the Gricean approach and Aristotelian notions of style is easier to build than it may seem. In both Aristotle and Grice, stylistic devices involve inference for their comprehension; in Grice, in fact, they generate conversational implicatures by flouting the maxims. For example, a particularly convoluted style can be seen as a violation of Grice’s maxim of Manner, a prompt to H to infer from this choice an intent on the part of S. Moreover, in persuasive discourse, style often has a persuasive, as distinct from an aesthetic, purpose.

Arrangement—the organization of a speech—may seem equally remote from the inferential process, which apparently functions only within arguments themselves. But, again following Perelman and Olbrechts-Tyteca, we shall show that speakers who follow the conventions of well-established genres—those of the classical oration or the experimental paper, for example—encourage audiences legitimately to infer their acceptability as members of a social or scientific community: an argument from êthos. In addition, because all persuasive genres simultaneously embody the two orders of argument that underlie any persuasive communication—the logical/chronological and the psychological—a speech can gain in persuasive strength by moving gradually from the consensual to the contentious and, in public address at least, from emotional neutrality to impassioned advocacy. Finally, since, as a general rule, well-entrenched genres obey Grice’s maxim of Manner to be orderly, it is possible for speakers to flout this maxim and, by so doing, to create implicatures that serve the interest of their larger persuasive purposes.

We believe that an alliance between rhetoric and pragmatics will be possible if these problems are solved: we must be able to reconstruct classical rhetoric as a cognitive theory, one in which inference plays a central role, not only in invention and in logos as we would expect, but also in êthos, pathos, lexis (or style), and taxis (or arrangement).

A Gricean theory of rhetoric: Presuppositions

The Gricean theory we reconstruct has as its center an S intending to persuade, that is, intending to reinforce or change the beliefs, attitudes, or actions of H. In this construal, just telling a joke is not a rhetorical transaction, nor is just saying hello to an acquaintance in the mall. However, if I tell a joke in order to induce friendly feelings, if I refrain from saying hello in order to convey an insult, then I am engaged in a rhetorical transaction, attempting to persuade you to alter beliefs, attitudes, or actions as a consequence of your recognition of my intent to do so.

The theory we reconstruct assumes communicative causality; that is, it assumes that any effect on H is a consequence of what S says or signals with communicative intent. The cut is not between action and words, since words need not be uttered with communicative intent: the aphasic’s word salad is not communication, but a symptom of disease. Actions without words can also have communicative intent: the angry S shaking his fist as an answer to a proposal is signaling with communicative intent. But the communicative intent of words or actions is not enough; S must intend to persuade by virtue of just those words or actions. By shaking his fist, the angry S must intend to persuade H that he is angry about the proposal and H must be persuaded of that anger by virtue of recognizing the meaning of the gesture as an index of that intent.

The theory we reconstruct has the utterance-in-context as its fundamental building block. It emphasizes, on the one hand, the structural elements of the utterance and, on the other, its orientation toward an ‘immediate’ interaction between a speaker and an audience. Persuasive utterances are always utterances-in-context; that is, the specification of their ‘meaning’ always involves the specification of a forum, an exigence, a syntax, a semantics (sense and reference), and an ‘illocutionary force’. Promise, assertion, greeting, and question are terms that describe such illocutionary forces. If I say “The sun is shining,” in normal circumstances I am engaged in the illocutionary act of asserting the truth of the proposition ‘The sun is shining’. I might also have said “Old Sol is shining.” The latter sentence would be referentially synonymous with the first, but would differ from it in sense; that is, I would not know these sentences to be referentially synonymous unless I knew the sun under the description “Old Sol.” If H understands either assertion as a consequence of understanding S’s intent as embodied in S’s utterance, then H has illocutionary uptake; if H believes the assertion to be true (let us say as an answer to a question), then the utterance has achieved its intended perlocutionary force.

An utterance is always embedded in a particular context without which its meaning will be indeterminate. Austin puts this well when he says that the complete understanding of our speech acts depends on our apprehension of their total context: “[W]e must consider the total situation in which the utterance is issued—the total speech act” (1975, 52). Of course, this is an unattainable ideal, and in practice we must always be satisfied with less than full context and therefore less than full determinativeness.

Persuasive utterances have as their indispensable contextual components a forum, an exigence, and a speaker and an audience. A forum is a public space designed to provide opportunities to speak for those who have the right to exercise those opportunities. To Aristotle, there are three forums: the law-courts; the legislature; and the public arenas that provide occasions for celebration, commemoration, or calumniation—the loci, respectively, of forensic, deliberative, and epideictic oratory. These forums constrain and channel inference: forensic oratory deals with the past, deliberative with the future, epideictic with the expression of community values. Each forum has its preferred mode of reasoning. Deliberative oratory prefers induction from past examples to possible futures; in contrast, forensic oratory, with its necessary focus on the past, prefers deduction (2.20.9), and epideictic, with its focus on values, prefers auxesis or amplification—a form of reasoning that moves from a generalization to its examples, from the fact of courage to instances of courage (Rhetoric 1.9.39).

In all communicative interaction, there is a process prior to any inference. Speakers must determine the issues that are at stake, the staseis. These are adumbrated by Aristotle (Rhetoric 1.13.9–10, 3.16.4, 3.17.1), but are most useful to us in the final form they took in later forensic rhetoric: speakers asked whether an act was committed (an sit), whether it was a crime (quid sit), whether the crime was justified in some way (quale sit). Although no persuasive communication is possible without presupposing the issue at stake, the issue at stake can also be the issue at stake. This shift corresponds precisely to Habermas’s distinction between communication and discourse; in discourse, “the subject of discursive examination is not the rightness claim directly connected with the speech act, but the validity claim of the underlying norm” (1979, 64).

Contenders in polemical exchanges know that the definition of “the issue at stake” is a decisive step in argument because such a definition establishes the presumptions or “inference tickets” that allocate the “burden of proof” among the participants. Consequently, polemicists typically engage in a bitter dispute concerning precisely what is at stake, each side making persuasive efforts to bring “preliminary” or “procedural” issues to the foreground, hoping thereby to shift the burden of proof in their favor.

The existence of an appropriate forum and of my right to employ it are not sufficient for public speech. I need not exercise that right; I will not, in fact, exercise it if I perceive no motive for speaking, no exigence. (Equally, I can experience an exigence but lack a forum for its expression, or the right to speak in that forum.) After the initial exigence, every intervention by one of the participants in a conversation or in a polemical exchange sets up a rhetorical exigence that the following participant must satisfy. For example, a question demands a reply, an objection, a rebuttal (or a concession). Failure by participants to satisfy such exigencies may (legitimately) lead to inferences about their “second-order” intentions.

Speaker and audience are two other indispensable contextual components of persuasive utterance: to describe this crucial pair adequately, we must go beyond the atomic S and H of Gricean analysis. Orators need not hearers, but audiences. We must not equate any aggregate of individual hearers with an audience. Although an audience is made up of individuals, each with its own disposition to be persuaded, no speaker can separately and simultaneously address hearers A, B, C, D, E...and n. Necessarily, from the orators’ point of view, the audiences they address are not actual, but ideal: they are creations of the speech. If, at the end of a speech, the audience applauds, it is applause from an audience co-created by the orator and the aggregation before him, an audience and a speaker that did not exist at the beginning of the speech. Even in the case of one-to-one persuasion, this principle applies: to be persuaded, even a single hearer must be imagined and addressed as an audience; even in such cases, audiences are the creations of speakers. For Aristotle, this principle is entailed by the nature of rhetoric as an art: “[R]hetoric [does not] theorize about each opinion—what may seem so to Socrates or Hippias—but about what seems true to people of a certain sort” (Rhetoric 1.2.11). His analysis of audiences into the young, the middle-aged, and the old exemplifies this principle (2.12–14). Of course, the effects on actual audiences must also be taken into consideration; their investigation, however, is an empirical issue. Although the speaker creates the applause that greets him, that applause is a real effect and can be measured.

We are indebted to Perelman and Olbrechts-Tyteca for a distinction between two sorts of audiences: universal and particular (1969, 30–35). This is a distinction, not between the ideal and the real, but between two sorts of ideal. Both the universal audience and the particular audience are regulative ideals that constrain what speakers feel they can say, what arguments they feel they can make, what appeals they feel are appropriate. In the first case, the imagined audience is humankind in general; in the second, a subset of humankind (e.g., Americans or nuclear physicists or Republicans).

Analysts must find a via media between the postmodern view, in which S is merely an instantiation of larger economic, social, and political forces, and the traditional, Aristotelian view, in which S has a large measure of autonomy. They must do so because, in the traditional, Aristotelian view, rhetoric is an art whose components are wholly within S’s control. If this is the case, every persuasive aspect of style, arrangement, and invention can be traced back, in principle, to decision and voluntary action. Challenged by the analyst, S can always respond that a particular feature of the text either was or could have been a consequence of choice. “Could have been” may seem like an unacceptable hedge, but we mean by this only that speakers need not be conscious of all of their intentions at the time that they are speaking according to them. For example, when speakers answer the telephone by saying “Hello,” they do not usually mean to greet the caller; they are merely signaling their presence on the line. They may not be conscious of this intent at the time of utterance, but they would acknowledge that it was their intent if it were pointed out to them.

Analysts need also to recognize that components of a text may be outside the control of any particular S. Sometimes a social group may decide that texts should have one format or another: the format of the scientific or scholarly paper may be the result. Here we may speak legitimately of collective group intent. If the conventions of scientific writing require an abstract, for example, we may legitimately say that the writing of any particular abstract is a consequence of the prior intent of a group to make the search for information more efficient. We may say this, despite the fact that the intent of S in actually writing the abstract may be merely to comply with the dictates of the group in order to be published. In this way, the writing of an abstract is rhetorical from the point of view of the group, though not for the individual member, except insofar as that member is seen as a representative of the group. However, sometimes social, political, or economic pressures influence textual features in such a way that neither the group nor any of its members is conscious of this influence or of its effects. For example, in scientific communication, there has been over the centuries a movement toward the simplification of the syntax of sentences: their length has been reduced slightly and the number of their clauses reduced significantly (Halliday and Martin 1993). Although these characteristics of the text may be seen as adaptations in the interest of communicative efficiency, they are not within the scope of communicative causality and, therefore, cannot be part of a cognitive theory of rhetoric.

So far, we have been speaking as if “intent” were essential to any theory of communicative interaction. If this were so, it would be a problem for anyone who believes that this concept creates more difficulties than it solves. We believe that Robert Brandom’s work points in the direction of a solution of the problem of intention for those who do not believe in it: he eliminates the concept from his theoretical vocabulary at no apparent cost by redescribing “intentional” acts in social terms, in terms of the commitments involved in making certain utterances and the entitlements their utterance gives to hearers. Thus, under the right circumstances, the assertion “It’s a beautiful moon” counts as an openness to romantic involvement and entitles H to a kiss. But, of course, H could have gotten it wrong. In that case, S may claim that hers was one of a series of astronomical/aesthetic observations and that her previous behavior has indicated no preference for H. Thus, S might contend, H’s inference to an entitlement is not justified; it is incompatible with previous assertions and behavior of S. A suit for sexual harassment may be in the offing. In court, however, H may well claim that he has been deceived, that S led him on. In support of the case that S persuaded him to kiss her, H would adduce a pattern of speech and behavior that, he hopes, rivals S’s in coherence and plausibility.

Rhetorical theory has traditionally concerned itself with efforts, rather than effects. This concern stems from Aristotle’s definition of rhetoric as an art, the performance of which cannot be judged solely according to its effects. In medicine, Aristotle’s paradigm art, effects cannot be the sole criterion of good practice: the operation can succeed even if the patient dies. Although the distinction Aristotle draws is valid, a strict demarcation in rhetoric between communicative causes and their effects cannot be defended. The criteria for judging rhetorical performance must be grounded, not in efforts, but in actual effects: the normative oughts that concern best practice must have actual ises as their basis. A medicine that did not concern itself with the effects of treatment is unthinkable; why should a rhetoric be different? Of course, we must acknowledge that, while illocutionary intent may be inferred from individual texts, illocutionary success and perlocutionary success are empirical issues that can be settled only by an analysis of audience reactions as expressed in adjoining texts which form, along with the speaker’s utterance, what is known in conversational analysis as an adjacent pair. An attack and its response constitute such a pair; so does a speaker comment followed by audience applause. Given these considerations, we can extend a theory of cognitive rhetoric to include, not only illocutionary force, but also illocutionary uptake and perlocutionary force.

A cognitive theory of rhetoric: Principles

In a Gricean theory of rhetoric, all audience reactions must be the products of inference. This creates a problem with Aristotle’s classification of the pisteis or proofs into logos or logical, êthos or characterological, and pathos or emotional. In one sense, logos applies only to “logical” matters, those whose conclusions are propositions; in another sense, logos is all-pervasive, applying, not only to logical matters, but also to matters of emotion and character. Although this distinction within logos is not apparent in the Rhetoric, Aristotle elsewhere distinguishes between the two forms of inference by distinguishing between the demonstrative syllogism, whose conclusion is propositional, and the practical syllogism, whose conclusion is an action or emotion (Nicomachean Ethics 7.3). The question is not one of subject matter:

All people are angry at those who insult them. The Spartans have insulted the Athenians. Therefore, the Athenians are angry.

This is a “logical” syllogism if the conclusion is a statement, a practical syllogism if it is a state of mind that may lead to action. To avoid confusion, we shall limit our use of logos to inferential processes whose conclusions are propositional.

But we must not assume that a distinction made for the purposes of analysis is a distinction that also exists in situ. The logos of actual arguments is never separable from their êthos and pathos, nor is the logical aspect of argument separable from its expressive and contextual aspects. In other words, we must adhere to the methodological dictum that, when we talk about argument, we must go beyond the logos; we must talk about emotional proofs and proofs from character; we must talk about style and arrangement.

Logos

Rhetorical reasoning seems easily compatible with the Gricean variety. It may be enthymematic or paradigmatic, that is, proceeding by example. The following is a typical enthymeme: “[To show] that Dorieus has won a contest with a crown it is enough to have said that he has won the Olympic games, and there is no need to add that the Olympic games have a crown as the prize for everybody knows that” (Rhetoric 1.2.13). In this case, the rhetorical syllogism is incomplete: only the minor premise is mentioned. But this means, not that all enthymemata are incomplete, but only that a complete iteration of premises and conclusion is not necessary to an enthymeme. A syllogism will be rhetorical even if it is complete so long as its premises are endoxa or common opinions. This does not even exclude genuine knowledge from the premises of rhetorical syllogisms so long as this knowledge is not genuinely held, but is merely believed: in Aristotle’s terms, orators and their audiences know the that and not the why: that the earth is at the center of the universe can be the premise of a rhetorical syllogism so long as that proposition is simply believed and not held because the principle is known that heavy things are drawn to a center (Smith 1995; Ross 1949, 21–59).

Complementing rhetorical deduction is rhetorical induction or paradigmatic reasoning from examples or signs. Induction from example takes two forms in Aristotle’s rhetoric. Sometimes speakers enumerate two or three examples and then imply that their conclusion follows because it is another example of the same kind. Here is Aristotle’s illustration: “[When someone claims] that Dionysus is plotting tyranny because he is seeking a bodyguard; for Peisistratus also, when plotting earlier, sought a guard and after receiving it made himself a tyrant, and Theagenes [did the same] in Megara, and others, whom the audience knows of all become examples of Dionysius, of whom they do not yet know whether he makes his demand for this reason” (Rhetoric 1.2.19). In such cases, the rule is unmentioned. But it is also possible to argue from examples to a rule, “from part to whole” (1.2.19): “[I]n the case of the woman of Peparethus [it was argued] that women everywhere discern the truth about [who is the father of] children; for when the orator Mantias at Athens was disputing [the parentage of] his son, the boy’s mother declared the truth. Similarly, when Ismenias and Stilbon were in a dispute at Thebes, the woman of Dondona identified the son of Smenias; and for this reason Thettaliscus was recognized as Ismenias’ son” (2.23.11).

Inferences from examples or from necessary and fallible signs are permissible starting points for rhetorical induction: “[I]f someone were to state that since Socrates was wise and just, it is a fallible sign [or semion] that the wise are just,” the sign would be only probable; in contrast, “if someone were to state that there is a sign that someone is sick, for he has a fever, or that a woman has given birth, for she has milk, that is a necessary sign [or tekhmerion]” (Rhetoric 1.2.18). Aristotle makes the connection between rhetorical induction and its deductive counterpart when he says that examples, necessary signs, and fallible signs are three of the four sources of enthymemes (2.25.8). Signs in effect become enthymemes by becoming the minor premises of syllogisms: If there is smoke, there is fire; There is smoke; Therefore there is fire. We may note that the difference between enthymematic inference and paradigmatic inference is a difference, not in the structure, but in the movement of reasoning from the general to the particular or the reverse.

To these forms of reasoning by implication must be added two other forms of inference, which were not noted by Aristotle: implicatures and presuppositions. I can say that the room is warm and mean that you ought to open the window: a reason for opening the window can count as a request that it be opened. My seemingly uninformative statement implicates the request. Gricean implicatures also explain the informativeness of such pseudotautologies as ‘Boys will be Boys’. We assume the speaker is being informative despite appearances to the contrary. It is important to note that Gricean conversational implicatures and enthymematic ones are alike in that they are cancelable without contradiction. In a sense, such implicatures are enthymematic, since the contextual information they make use of is not “explicit.” Inference to presuppositions also counts as an added type: that the room is warm presupposes that there is a room; that boys will be boys presupposes that there are boys.

Pathos and êthos

Aristotle’s theory of the emotions is compatible with Grice in that it is straightforwardly cognitive: inferences to particular beliefs are a necessary condition of particular emotional states. There is thus a clear “relationship of emotion to reasoned argumentation” (Fortenbaugh 1975, 17). This kind of reasoning is fundamentally different from the use of premises about emotional states to reach demonstrative conclusions about those states. In these cases, the conclusions are propositions about those states, propositions such as ‘The belief that one has been insulted is a necessary condition of anger’.

In Aristotle, “emotional response is explained syllogistically and the efficient cause is treated as the middle term” (Fortenbaugh 1975, 14). Nussbaum goes even further: “[I]f beliefs are an essential part of the definition of an emotion, then we have to say that their role is not merely that of external necessary condition. They must be seen as constituent parts of the emotion itself” (1996, 309). While we cannot say “He has been insulted; therefore, he is angry,” we can say “He is angry because he was insulted.” If Nussbaum is right, the belief that one has been insulted is actually a constituent of anger. But Fortenbaugh’s formulation is incomplete; it may be completed by showing that emotional response can also be generated paradigmatically. Because ragged clothes are a sign of “an apparently destructive or painful evil happening to one who does not deserve it and which a person might expect himself or one of his own to suffer, and this when it seems close at hand” (Rhetoric 2.8.1), the sight of those clothes can move a spectator to pity.

There is a related point that Aristotle does not make; it is inference that permits the experience of differing emotions based on the same beliefs:

I can believe that the mortality rate for free-fall parachutists is very high and so infer that the prospect of a free-fall parachute drop ending in death is very probable. My companion might hold the same belief and make the same inference. Now if we both decide to make a free-fall parachute drop, or are ordered to, I might be in a state of fear but my companion in a state of excitement. For my mind might be dominated by the evaluation of the situation as dangerous to me, while my companion’s mind, while allowing that what he is about to do might be dangerous, is taken up by the realization that here is a challenge worthy of him.

(Lyons 1980, 35)

What permits the experience of different emotions in this case is that each “speaker” construes himself differently. Valuing different things, each generates his own pattern of inference and reaches his own conclusion.

In Aristotle, the characters speakers create for themselves or create within their speeches—their êthê—are also the products of audience inference to particular beliefs. In this case, however, the result is not a state but a disposition. In the case of pathos, speakers evoke the emotion of friendship; in the case of êthos, speakers create in the mind of an audience a disposition that they are acting and will act according to the dictates of friendship. Beliefs that speakers instill in audiences can never guarantee their disposition toward friendliness; nevertheless, since holding some of a particular class of beliefs is a necessary condition for the presence of this disposition, speakers can stimulate friendliness by increasing likelihood that beliefs of that class will be held. For example, friendly feelings may be founded on the following inferred beliefs: “[They are friendly to] those that take them seriously in some way, for example, admiring them and regarding them as serious people and finding pleasure in them, and especially those feeling this way about what they wish to be admired in themselves or in regard to what they want to seem serious about or what they find pleasure in. And [they are friendly to] those like themselves and having similar interests” (Rhetoric 2.4.19–20).

In the Gricean theory we are outlining, pathos and êthos are evoked in accord with the Cooperative Principle and meaning_NN, as modified for persuasive exchanges. If I intend to evoke anger, for example, I will do so by attempting to persuade you that you have been belittled or insulted. If you are persuaded, and you become angry, your anger will be in part a consequence of your recognition of my intent. But, of course, I can deceive you as to the facts; you may not have been insulted. Similarly, if I intend to evoke friendly feelings in you, I will do so by attempting to persuade you that I am well disposed toward you. If you are persuaded, and you feel friendly toward me, it will be in part a consequence of your recognition of my intent. But, of course, I can deceive you about the true state of my disposition toward you; I may hold you in contempt. In both cases, the Cooperative Principle is maintained so long as you believe me to be sharing the truth with you and you view us as sharing values and perspectives. Naturally, in neither case am I likely to say, “I intend to make you angry” or “I intend to make you friendly.” I avoid such statements because they thematize, not anger or friendship and their justifications, but my intent and its justification.

Style reconsidered

A Gricean theory of rhetoric will fail to be complete unless style is construed as a matter of inference. Style is both a level at which discourse is pitched (in modern linguistics, a register) and a set of semantic, syntactic, and prosodic variants within that register. In the former sense, a particular style is appropriate if it is proportional to situation and subject matter; in Aristotle’s words, “[T]he lexis will be appropriate if it is . . . proportional [analogon]” (Rhetoric 3.7.1). The mathematical analogy is exactly right; it emphasizes the close fit between a rhetorical situation and its verbal response. When he praises Gorgias’s exclamation to the swallow who let go her droppings on him, Aristotle exemplifies this principle of proportionality: “[Gorgias] said, ‘Shame on you, Philomela’; for if a bird did it there was no shame, but [it would have been] shameful for a maiden. He thus rebuked the bird well by calling it what it once had been rather than what it now was” (3.3.4). The style is perceived as correct because we infer an exact proportion between the situation and Gorgias’s verbal response to it, an exact proportion that Gorgias presumably intended. Its form is a wish, directed at a bird; its force is a judgment, directed at a woman. This episode is easily redescribed in Gricean terms. In this case, Grice’s maxim of Quality is violated—shame is not applicable to actual swallows—and from that violation we are asked to draw an inference.

But style is also a set of semantic and syntactic variants within a register, a set that in a cognitive theory must depend for its effect on inference. In this sense, style is a matter of schemes or syntactic patterns and tropes or semantic transformations.

For Aristotle, metaphor is the paradigm trope. He regards it as an analogical process including, not only what we would call metaphor, but also what we would call simile, metonymy (“All hands on deck”), personification, and hyperbole (Rhetoric 3.10–11; Poetics 20–22). He regards the understanding of all of these tropes as a matter of inference: “Metaphor most brings about learning; for when [Homer] calls old age ‘stubble,’ he creates understanding and knowledge through the genus, since both old age and stubble are [species of the genus of] things that have lost their bloom” (Rhetoric 3.10.2).

In a cognitive theory of style, all tropes must involve inference in their understanding. In Gricean terms, a maxim is flouted and an implicature created. For example, “All hands on deck” is perceived as a mismatch: hands seldom exist apart from bodies, and, if they did, they would certainly lack motility. But a sailor’s hands represent his most important characteristic— his sailoring labor; because this is so, they can form a metonymy that means “All sailors on deck.” Because hands are a salient characteristic of anyone seen as a source of labor or assistance, we have such other metonymies as “Lend a helping hand,” “Give us a hand up.” All involve the flouting of the Gricean maxim of Manner (see Grice 1989, 33–37).

Although all tropes involve inference in their understanding, not all tropes function in arguments. Perelman and Olbrechts-Tyteca carry a cognitive theory of tropes a step further than Aristotle when they differentiate between their aesthetic and argumentative functions: “We consider a [trope] to be argumentative, if it brings about a change of perspective, and its use seems normal in relation to this new situation” (1969, 169). The purpose of a trope may, after all, be merely aesthetic: “[I]f . . . the speech does not bring about the adherence of the hearer to this argumentative form, the figure will be considered an embellishment, a figure of style. It can excite admiration, but this will be on the aesthetic plane, or in recognition of the speaker’s originality” (169). When Byron writes that “[t]he Assyrians came down like wolves on the fold,” we understand this simile, as we understand all figures: by means of inference. In addition, Byron’s trope is functional; it makes the attack appropriately vivid. Nevertheless, it is not argumentative.

Irony—meaning the opposite of what one says—and litotes—understating what one says to intensify its effect—exemplify the Perelmanian link between trope and argument. Both are argumentative in the following case because both are designed to bring about change in hearer perspective:

[L]itotes can be converted into irony by suppressing the negation. Of the same misshapen man of whom one would say, using litotes, “He is no Adonis,” one might say ironically, “He is an Adonis.” In the first case, we have a movement of thought through a scale of values; in the second, there is a confrontation of a qualification with an apparent reality. In the first case, the direction is dominant; in the second, one does not want to bring about a sudden turnabout of the mind, but one wants the mind to take note of the ridicule arising from an incompatibility.

(1969, 292)

Aristotle shows that he understands the connection between schemes and argumentation when he says of antitheses: “Such a lexis is pleasing because opposites are . . . like a syllogism, for refutation is a bringing together of contraries” (Rhetoric 3.9.8). But Perelman and Olbrechts-Tyteca carry the classical theory of figures a step further when they explicitly place schemes as well as tropes under the rubric of inference. Antithesis, insinuation, hesitation, antimetabole, contrarium, and reticence are their examples (1969, 273, 467, 240, 444, 344, 467).

Antithesis and insinuation are familiar in everyday speech. In a case of hesitation, a speaker says: “At the time the republic suffered great wrong because of the consuls, should we say because of their stupidity or because of their perversity or because of both?” (qtd. in Perelman and Olbrechts-Tyteca 1969, 240). Antimetabole is a reversal in the order of words repeated in successive clauses: for example, I do not live to eat, but eat to live (Quintilian 1921, 9.3.85). Contrarium is a scheme in which one of two opposite statements is used to prove the other, for example, “Now how should you expect one who has never been hostile to his own interests to be friendly to another’s” (ad Herrenium 4.18.25). Reticence is a figure in which “we begin to say something and then stop short, and what we have already said leaves enough to arouse suspicion, as follows: “He who is so handsome and so young, recently at a stranger’s house—I am unwilling to say more” (4.54.67).

Understanding each of these schemes involves inference; moreover, each is argumentative, involving a change of perspective: the example of hesitation turns an apparent choice into a real accusation; the example of antimetabole turns a contrast into a moral judgment; the example of contrarium turns self-interest into selfishness; the example of reticence turns apparent virtue into vice. In Gricean terms, these conclusions are implicatures generated by violations of the maxim of Manner, a maxim related, “not . . . to what is said but, rather, to how what is said is to be said” (1989, 27, 33–37).

In the case of schemes like hyperbaton (word order reversal), alliteration, climax, and anaphora (a form of parallelism), a similar analysis is possible. When, in Profiles in Courage, John Kennedy wrote: “Already American vessels have been searched, seized, and sunk,” he used all of these schemes, each of which flouts the maxim of Manner. It was perfectly obvious to Kennedy and his audience that his initial reversal of noun and adverb violates the word order of ordinary English. So, unless Kennedy is making a mistake no native speaker ever makes, he must be trying to get something across by this reversal, a reversal that alliteration highlights: the sense of the immediacy of this crisis is a reasonable inference. Similarly, the alliteration that marks the three predicates draws attention to them and to their climactic order. It seems a reasonable inference from this flouting of the maxim of Manner that we are meant to be startled by the progressive violations of American sovereignty; we are meant to infer that our sovereignty has been violated beyond endurance.

The link of style to proofs is easily established: the flouting of these maxims expresses the character of speakers and their emotions; it creates êthos and pathos. President Kennedy is, plausibly, indignant with just cause, an effect conveyed by the schemes of hyperbaton, alliteration, anaphora, and climax. These stylistic turns, then, are the equivalent of arguments; from them, as from premises, we are meant to infer the conclusion that Kennedy is indignant and that we ought to be. Kennedy says in effect that a nation that acts that way ought to be confronted.

Arrangement reconsidered

Arrangement—the organization of a speech—may seem equally remote from the inferential process, which apparently functions only within arguments. But we shall show that arrangement can also be a form of proof. Aristotle discusses two forms of organization, both fundamental to any extended persuasive event: the first is logical/chronological and follows the order of implication or of events; the second is psychological and is sensitive to shifting states of mind on the part of audiences. Of the first, logical/chronological order, Aristotle says: “There are two parts of a speech: for it is necessary [first] to state the subject and [then] to demonstrate it” (Rhetoric 3.13.1). Concerning the second, psychological order, he says, for example, that introductions should make an audience “well-disposed and . . . attentive” (3.14.7). Both sorts of arrangements are functions of the same Gricean maxim of Manner—be orderly—though the import of the maxim is construed differently in each case.

The logical/chronological and psychological orders underlie all extended forms of persuasive communication, from the classical oration to the scientific paper. The experimental paper in science, for example, follows a logical order—that of induction. At the same time, it exhibits a sensitivity to the psychology of its audience: its organization constitutes a master finding system. In accord with the êthos of its audience, such an organization encourages objective reading, rather than subjective involvement. In Gricean terms, by heeding the maxim of Manner, scientists create an argument from êthos and establish a beachhead for the Cooperative Principle. In fact, when such formal requirements for communication in science are flouted, serious problems of understanding arise, as Hardy and Littlewood discovered when they undertook the evaluation of the unsolicited scribblings of an obscure Indian mathematician who, as it turned out, was in a class with Euler and Gauss. Two distinguished English mathematicians had already returned Ramanujan’s papers unread to their owner. They had not troubled to make a judgment or, rather, one presumes, they had made a judgment on the basis of êthos alone (Hardy 1967, 30–38).

There is a second way in which the admonition—be orderly—is persuasive: the order in which proofs are presented is crucial to persuasion. The Gettysburg Address provides an example (see Lincoln 1946). Lincoln begins it by asserting that, when founded, the United States was “dedicated to the proposition that all men are created equal.” Now we find ourselves “engaged in a great civil war” in which that proposition is chiefly at issue. As a consequence of that war, “we have come to dedicate” a cemetery for soldiers who have died in its course, though our dedication can in no way match their dedication, the sacrifice they made.

These statements participate in a logical/chronological order: they are the uncontestable premises from which Lincoln will draw his conclusions. They also participate in a psychological order: Lincoln can rely on his Northern audience to be well disposed to such statements and to the person who makes them. In the words of Perelman and Olbrechts-Tyteca: “[A]rguments . . . must be advanced in the order which gives them the greatest strength, and this means that ordinarily one will begin with that argument whose strength is independent of the strength of the others” (1969, 500). In our view, strength must be interpreted both logically and psychologically.

For Lincoln and, he hopes, for his Northern audience, the Civil War is about keeping faith with the Founders and the Union dead; to refuse to do so, Lincoln will imply, is to betray their memory. Because he is aware that an audience must be prepared for so radical a conclusion, Lincoln does not merely state that “it is for us the living, rather, to be dedicated here to the unfinished work which they who fought here have thus far so nobly advanced”; he takes care to give reasons for his assertion: “[I]t is rather for us to be dedicated to the great task remaining before us—that from these honored dead we take increased devotion to that cause for which they gave the last full measure of devotion; that we here highly resolve that these dead shall not have died in vain; that a government of the people, by the people and for the people shall not perish from the earth” (1946, 734). At the end of this final sentence, in three clauses arrayed in climactic order, Lincoln links the living, the Union dead, and the Founders in the common and urgent purpose of national survival. His is “a speech does not leave the hearer the same as he was at the beginning”; if Lincoln has done his work, “the order [he has] adopted is crucial precisely because the changes in the audience are both effective and contingent” (Perelman and Olbrechts-Tyteca 1969, 491; see also 499–500).

Presence reconsidered

Presence, the last persuasive vehicle of which we shall speak, the creation solely of style, arrangement, and those aspects of invention that do not pertain to argument, is an effect that depends on inference and so must be part of any Gricean scheme: The pisteis or proofs are specifically excluded from this presence when Perelman and Olbrechts-Tyteca say that presence, “at first a psychological phenomenon, becomes an essential element in argumentation” (1969, 117; emphasis added). The initial purpose of presence is fully to arrest the hearer’s attention, to “[fill] the whole field of consciousness . . . [s]o as to isolate it, as it were, from the hearer’s overall mentality” (118). The devices by which this is accomplished are an open set. They can be verbal: presence can be created by the accumulation of detail (144–45), the use of the present tense to relate past events (160), the use of symbols and tropes (334–35), the repetition of the same arguments in different form (478). They can also be nonverbal, a point Perelman and Olbrechts-Tyteca do not make: presence can be created by the use of tables, figures, photographs, white space, and type size.

Arrangement routinely enhances presence: speakers commonly use introductions to give prominence to matters on which consensus is high while they reserve conclusions for emotion-laden statements that must be motivated by the cumulative proofs of the speech. But there is another way to create presence. Since well-entrenched genres routinely obey Grice’s maxim to be orderly, it is possible for speakers to arrest the audience’s attention by flouting this maxim. Cicero does this brilliantly in his first speech against Catiline, which begins abruptly: “Quo usque tandem abutere, Catilina, patienta nostra? [How long, Catiline, will you persist in exhausting our patience?]” (1910, 75). In his exordium, Cicero has flouted the maxim of Manner so as to implicate the urgency of the Catilinean crisis; his violation has a persuasive point. This effect is also evident, though in a subtler form, when, in the Gettysburg Address, Lincoln moves with barely a transition from his initial uncontestable premises to his radical conclusion that the living owe the dead a debt, not merely of gratitude, but also of labor and risk. In this passage, Lincoln flouts the maxim of Manner to implicate that the importance of the task ahead is entailed by fundamental American values. Although, in these as in all cases, the initial consequence of presence is psychological—the hearer’s attention is momentarily arrested—its final consequence is argumentative because hearers infer that this arrest must have a point, a point arrived at by inference.

The contribution of arrangement to presence is augmented by that of style. Perelman and Olbrechts-Tyteca elucidate the nature of this contribution in the case of symbol and trope:

Not only is the symbol easier to handle, it can impose itself with a presence that the thing symbolized cannot have: the flag which is seen or described can wave, flap in the wind, and unfurl. In spite of its bonds of participation, the symbol maintains a kind of individuality which makes possible a great variety of manipulations. “There are no longer any Pyrenees [Il n’y a plus de Pyrenées]” does not merely express a political idea[, that of the unification of the royal houses of France and Spain in the reign of Louis XIV]: it evokes also the fatigues and dangers of a frontier and the enormous efforts needed to abolish it.

(1969, 334–35)

The Gettysburg Address exemplifies the role of style in the creation of presence. In the Address, Lincoln’s central trope is metaphor. The United States is a living thing “brought forth on this continent”; it is a nation that “shall have a new birth of freedom,” a nation that “shall not perish from this earth.” Crucial to this continued, though metaphorical, national life is the recent death of these soldiers: they “gave their lives that this nation might live” (emphasis added). In this startling combination of a trope, metaphor, and scheme, polyptoton, Lincoln equates the real death of the soldiers with the nation’s metaphorical life. But it is only we, “the living,” who can turn this equation into a fact; it is only we who can insure “that this nation, under God, shall have a new birth of freedom—and that government of the people, by the people, for the people, shall not perish from this earth.” In his closing epistrophe, Lincoln comes full circle—he turns his metaphor back into a fact: the nation is literally its people.

In the Gettysburg Address, patterns of arrangement and style converge to give presence to the nation as a living thing, a precious inheritance now in jeopardy. Not coincidentally, this view is also the conclusion of Lincoln’s argument. The Address exemplifies presence as a synergy of style, arrangement, and argument in the interest of persuasion.

In the matter of presence, a distinction must be drawn, a distinction especially relevant to the sciences. In creating presence, Lincoln is counting on a lack of awareness on the part of his audience; the effect he achieves would be undermined by an awareness of its means: “[E]verything that promotes perception of a device—the mechanical, farfetched, abstract, codified, and formal aspects of speech—will prompt the search for a reality that is dissociated from it” (Perelman and Olbrechts-Tyteca 1969, 453). This use of presence may be contrasted with its employment in experimental reports. The very detailed descriptions and the clusters of tables and figures common to these reports seem to flout the maxim of Quantity and to put the maxim of Relation in immediate jeopardy. In fact, nothing is amiss: it is the goal of the sciences involved that supplies the relevance of these avalanches of information and the presence they create. In the sciences, though not in public address, presence is routinely under the control of common goals designed to guide audience attention in a manner relevant to disciplinary growth. Presence does not have a strategic purpose in the sciences but is, when Grice’s maxims are strictly obeyed, communicative action all the way down.

Conclusion

Our reconstruction attempts to combine the strengths of Gricean pragmatics and Aristotelian rhetoric. It does so by giving one strand of the rhetorical tradition a cognitive reading, one that suggests the compatibility of two otherwise separate understandings of linguistic communication and interaction. We believe that this reconstruction benefits both pragmatics and rhetoric. Rhetoric will benefit from the realization that its practices are not a jumble of techniques, but are, instead, a coherent theory in the cognitive class; rhetorical theory will thus be seen as a genuine hermeneutics, one that, contrary to the received view, includes the subject matter of the whole of Aristotle’s Rhetoric. Rhetorical theory will also benefit because it can continually be strengthened by developments in disciplinary areas previously regarded as alien to its specific purposes, for example, developments in philosophical pragmatics, such as Robert Brandom’s Making It Explicit: Reasoning, Representing, and Discursive Commitment (1994). In its turn, philosophical pragmatics will benefit from the broadening of its explanatory reach to include talk exchanges whose purpose is persuasive rather than just informative and a widening of its explanatory scope routinely to include the explanation of stylistic and organizational aspects of talk exchanges, as well as aspects related to the emotions and characters of speakers. No longer would works like Brandom’s Making It Explicit be able to exclude these matters as a matter of course.