Nine Ways of Looking at Apologies: The Necessity for Interdisciplinary Theory and Method in Discourse Analysis

Nine Ways of Looking at Apologies: The Necessity for Interdisciplinary Theory and Method in Discourse Analysis

0 Introduction: The Problems, Paradoxes, and Pleasures of Interdisciplinary Research

Of all the aspects of language, discourse analysis is singularly interdisciplinary a word with a somewhat speckled past. At the moment, interdisciplinary is a good word. But it was not always so.

Originally all scholarship was implicitly multidisciplinary, in the sense that sharp distinctions were not explicitly recognized among disciplines. It was only in the mid-nineteenth century that disciplines were rigorously segmented into university departments, with all the budgetary and other turf rivalries that departmental structure brought in its train.¹ As knowledge in many fields, particularly in the social and physical sciences, increased exponentially and got more complex in the late twentieth century, departmental and disciplinary boundaries became at once more essential, to preserve order and identity, and more embarrassingly obstructionist to new ways of thought. The physical sciences seem to have solved the problem by creating new formal fields and new departmental structures to house and identify new ways of pursuing knowledge: molecular biology and biochemistry, for instance. But the social sciences more unsure of both their legitimacy and their domains seem to have had more of a problem in deciding what to do when ideas spill out of their original disciplinary receptacles.

Linguistics is a paradigmatic case. If our turf is, as we like to tell introductory classes, the scientific study of language, what does language properly include? Some lin guists interpret language as language alone: they draw the line in the sand at the point where analysis involves interaction or persuasion, or anything we do with words. Others incorporate these territories into linguistics, willingly or grudgingly, but still try to keep them separate. Here, in a central subdivision, we will discuss language-in- isolation; beyond this impregnable fence that guards the province of philosophy, speech acts and implicature; there, further than the eye can see, next to the kingdom of sociology, conversation; and far away, adjoining the duchies of rhetoric and mass communication, public discourse. Each area has developed its own language, as nations will, unintelligible to those within other areas of linguistics, and even those in adjoin ing principalities. These boundaries are guarded jealously and justified zealously.

There are certainly advantages to territoriality, not only political but genuinely intellectual. Within a field's strict confines one can achieve competence and control. No one, surely, can claim to know all of linguistics any more (as was perfectly pos sible a generation or two ago); but at least one can without undue strain claim mastery over an area like pragmatics or conversation analysis. But disadvantages, to the point of paradox, offset these advantages.² In this chapter I want to discuss the necessity of an inter-, cross-, and multidisciplinary approach for discourse analysis, an area that borrows from and contributes to many fields both within linguistics and outside of it. To illustrate my argument I will use as an example the speech act of apology, con sidering what we need to know about it in order to achieve a full and satisfying explanation of its properties and range of use.

Even if a case could be made for the autonomous treatment of some aspects of language (e.g. syntax, or phonetics), discourse cannot be satisfactorily analyzed in a vacuum, whether contextual or methodological. We might say of syntax that though it is located firmly within the boundaries of linguistics proper, sometimes reference to another subfield (suprasegmental phonology, or dialectology) or discipline (neuro logy) enhances the understanding of syntactic processes. But even in such cases the syntactician would be merely borrowing from outside, not obliterating the bound aries between syntax and the other field. But the assumption of autonomy works less well with discourse analysis. To do a thorough job of talking about discourse, or a discourse, the analyst must have recourse to the findings and methods of other (sub)disciplines; there is no discourse analysis otherwise. At the same time, our discovery procedures and methods of analysis, the questions we ask, and what we consider answers are uniquely our own, even as they represent the commingling of many diverse concepts. Our data may range from small units (sentences or turns) to much larger and more abstract entities (courtroom trials; novels; political events). And when we analyze those data, we must often consider them in terms of the smaller and more concrete units of which they are composed, using tools developed for the analysis of turns or sentences to understand the functions, meanings, and structurings of the larger and more abstract units we term discourse. We may be concerned with any of several aspects of an extended utterance: its role in a longer document (a narrative); its interactive function (in creating small groups like couples or families); its role as a maker of institutional affiliation (academic language) and societal influence (journalism). Therefore our statements will reflect the belief systems of other fields: literary analysis; psychology; anthropology and sociology; political science, as well as areas closer to home (syntax, pragmatics, conversation analysis). This perspective is controversial both within linguistics (on the grounds that we are changing the rules or moving from the finite safety of autonomy to the chaos of interconnection) and from outside (on the basis that we are misusing the methods and languages of disciplines in which we are interlopers). But we must tolerate these critiques and learn to answer them if we are going to accomplish anything interest ing, for it is precisely at the interstices of established disciplines and disciplinary thinking that the interesting work of discourse analysis will be done.

1 A Case in Point: Understanding Apology

Let me take as an example of the interdisciplinary nature of discourse analysis a case that at first may seem overly simple, hardly a part of discourse analysis at all, more typically considered as an exercise in pragmatics or conversation analysis: the apology. But we have to understand apologies as contributions to a larger discourse, viewing them from a variety of perspectives, formal and functional, cognitive and interactive, individual and group, intralanguage and societal; to examine the apology from the perspective of phonology, syntax, lexical semantics, speech act pragmatics, conversational analysis, narratology, and sociolinguistics. In some ways any speech act verb might illustrate the point. But apologies are particularly good examples, theoretically rich as well as practically important. They are hard to identify, define, or categorize, a difficulty that arises directly out of the functions they perform. Hence too, they occur in a range of forms from canonically explicit to ambiguously indirect; the functions served by those forms range from abject abasement for wrongdoing, to conventional greasing of the social wheels, to expressions of sympathy, advance mollification for intended bad behavior, and formal public displays of currently appropriate feeling. Thus, in terms of the relation between form and function, apologies are both one-to-many and many-to-one, a fact that only makes the analyst's task more daunting (and more exciting).

Apology, more than most speech acts, places psychological burdens both on its maker and, less seriously, on its recipient. That is the reason for the plethora of indirect forms that, in appropriate contexts, we recognize as apologies. There does exist an unambiguous apology form, seen in:

I apologize for eating your hamster.

But that form is rarely encountered in the most characteristic apologies, informal ones between intimates. In these cases we usually resort to any of a set of forms that involve one or another of the presuppositions or assertions of apologies (cf. section 2.2), either blurring it or explicitly stating it (allowing other aspects of the act of apology to be passed over in silence). For instance, the speaker's responsibility for the act can be downplayed in favor of an explicit statement conveying regret. I'm sorry about your hamster,

or in extreme cases responsibility may be explicitly assigned elsewhere:

Well, someone left the hamster in the refrigerator!

or the utterance may deny that wrongdoing occurred at all:

Well, that's what hamsters are for, right?

The presence of well in extreme cases like this suggests an awareness that, as apolo gies, these utterances are not fully satisfactory, and that the addressee's goodwill is required to make them function appropriately (cf. Lakoff 1973; Schiffrin 1985). Note that well seems much less strongly mandated in the first case above, with sorry. Indeed, in the latter two cases the speech act may arguably have crossed over the line that separates apology from explanation (cf. section 2.2).

But some forms of apologies refer specifically to one of their functions, perhaps as a way to minimize the utterer's responsibility for the others:

I admit I ate the hamster. (Responsibility)

It was wrong of me to eat the hamster/I shouldn't have eaten the hamster. (Wrongdoing)

Can you find it in your heart to forgive me for eating the hamster? (Wish for forgiveness)

I'll never eat a hamster again as long as I live. (Abjuration of bad behavior)

These cases illustrate the many forms available for the performance of the single act of apology. The converse is also true (perhaps to a lesser degree): a single form, I'm sorry, can function variously as an apology, an expression of non-responsible sympathy, and as a denial that an apology is, in fact, in order at all:

I'm sorry that I ate the hamster.

I'm sorry, Mr. Smith isn't available today.

Well, I'm sorry! but you don't know what you're talking about!³

One advantage to having all these choices, for apologizers, is that they are thus enabled to calibrate the self-abasement to the perceived seriousness of the offense. It may seem that a full canonical apology would always be preferable to an offended party. But this is not necessarily true. Suppose you are at the movies. The show is in progress when someone moving past you steps on your foot. The occasion requires an expression of recognition of wrongdoing. But do you want the full canonical treatment? Both those around you, and you yourself, would be inconvenienced by it. A grunted sorry is all you desire; anything more is inappropriate and embarrassing.

On the other hand, some apologies, to be felicitous, require at least the appearance of contrition. In these cases the recipients must have the power and the right to enforce demands for real remorse.

Another advantage of options is that an apologizer with power can, by making use of an ambiguous form, look virtuous while saving face. This is often seen in legally mandated apologies. A particularly notorious case occurred at the University of California at Berkeley some years back, when a freshman woman accused several football players of acquaintance rape. She was persuaded to accept a plea bargain that involved an apology from the team members. Their apology stated that while they apologized, they had not done what they were accused of doing. Some might argue that the second clause renders the first nonsensical or at least infelicitous (cf. section 2.4). Others might argue that this example perfectly illustrates the ability of institutional power to give meaning to otherwise bizarre utterances. If such vapid apologies have any meaning at all, it can only reside in the acknowledgment that the addressee has been hurt and has personhood or stature enough to require redress.

Similar cases occur in civil suits, in which corporate defendants refuse to publicly admit responsibility even though that might save them the expense and possible face- loss of a protracted trial. Their reasoning is that an apology is legally tantamount to a confession of wrongdoing via the presupposition of the speech act.

There are other problematic cases. One currently in vogue is the public-official apology, a statement made by someone in a position of power regretting bad behavior by previous holders of that office, in the name of the governed, against wronged ancestors of the aggrieved group. There are many such examples in recent years: e.g. President Clinton's apology to Africans for slavery, and Tony Blair's to the Irish for the potato famine. The willingness of many public officials to make such statements is striking compared with their reluctance to make apologies for their own, personal past misbehaviors. The reason is simple: the official cases are not true felicitous apologies, while the personal ones are. No one ever wants to make the latter kind, especially a powerful person, who stands to lose face, and therefore possibly power, by making one.

Most analyses of the apology speech act have focused on its felicity from the speaker's perspective, in particular the assessment of the speaker's state of mind (sincerity as manifested by signs of contrition). But this can create problems. For some speech acts (e.g. promises) felicity can be determined by the speaker's future actions alone. Others, though, like bets, require some sort of uptake from the addressee: You're on! or It's a bet! Apologies are normally considered members of the first class. But perhaps under some conditions especially when the recipients have been outspoken in demanding apologies of a particular form it may be appro priate to assign some responsibility to them for the felicity of the speech act. If, for instance, they make it clear that they have no intention of accepting any apology, no matter what, then surely no apology can be felicitous, and it is the demanders who make the entire performance infelicitous.

Even more confusing are forms that look like apologies but are not. Tannen (1994) has discussed the usage, especially common in women, of forms like I'm sorry, Mr. Smith is out of town until Wednesday. As Tannen notes, these are not meant as apologies: the speaker does not mean to accept responsibility, nor is there any acknowledgment of misbehavior. At most in such cases, I'm sorry is a way for the speaker to head off the addressee's annoyance and prevent an unpleasant closure, by expressing sympathy and connectedness. Sometimes it is little more than a way of bringing a polite end to a less than satisfactory interchange. The I'm sorry! but (you're an idiot) type is similar in form, but quite different in function. It seems to be an example of a but-preface (Baker 1975). On a radio talk show recently about women raising children by themselves, the suggestion was made that this is often successful. A man called and, in the course of his comment, said, I'm sorry! But [children need fathers]. This I'm sorry! is an apparent apology in advance for an utterance that is likely to be offensive. As such, it cannot be sincere, since if you know something you say will be offensive, and you care, you will not say it at all. Since these forms constitute challenges (= I'm confronting you and you can't do anything about it!), they are correctly felt to be rude, and so are seldom used by people with less power or something to lose by being offensive, while the former type are most often used by people in those positions.

2 The Function of Apologies

On both formal (forms like I'm sorry, whether true apologies or not) and functional (the performance of apologies via many speech-act types) levels, apologies have a tendency to be ambiguous. That is in itself a good reason to study them, and a good reason why studying them well requires many disciplinary models and approaches. Some of us, especially in the earlier stages of our careers, have dismissed levels other than those we are comfortable working at as simplistic, subjective, or beyond the legitimate reach of linguistics. But each of the nine levels I will now discuss offers insights about what apologies are and, more generally, what discourse is; and to achieve a full analysis, we have to be aware that all these levels exist and contribute to the meaning and function of apologies.

While there are in English no specific sounds associated with canonical or appropri ate apology, there do exist suprasegmental and nonverbal levels that are important, especially for the addressee, in the determination of the acceptability of an apology. These levels are the basis for hearers' judgments about the apologizer's sincerity and sufficiency of remorse, since we see them as beyond a speaker's control and there fore more likely to be truthful than the verbal utterance (cf. Ekman and Friesen 1969). So for instance an apology made too quickly, or in a monotone, will strike a hearer as scripted, nonspontaneous, and so not deeply felt. A breaking voice, on the other hand, bespeaks sincerity, as do certain nonverbal cues. An inability to make eye contact, generally judged negatively by Americans, has positive value (signifying appropriate shame) with apologies; the shuffling of feet and the use of self-adaptors (Ekman and Friesen 1968) like hand-wringing play a similar role. President Clinton is notorious on such occasions for biting his lip. While smiling is usually positively evaluated in American social interactions, its presence (often identified as a smirk) usually detracts from the effectiveness of an apology.

A question for any analysis of this kind is the extent to which these assumptions are universal. It is popularly believed that nonverbal signifiers of emotion, like the emotions they signify, are universal: everyone feels, or should feel, remorse over the same events; the same amount of remorse; and therefore, should express it in the same way. But this is not necessarily true. What occasions embarrassment in one culture may not in another. The way genuine feelings are translated into surface representations (both how and how much), what Hochschild (1983) terms emotion- work, may well differ across cultures, even cultures that are closely related and whose members speak the same (verbal) language.⁴ Viewers of the 1997 Cambridge Nanny case on television, as well as jurors in that case, commented that the English nanny, Louise Woodward, accused of killing a baby in her care, did not show enough remorse on the stand.⁵ Questioned about this later, she said that we, that is, the English, did not wear our hearts on our sleeves. Jurors basing their verdicts in part on witnesses' demeanor, as they are instructed to do, may make wrong decisions in cross-cultural situations like this.

The semantic problem of apology is this: what do we mean when we talk about apologizing? How does apology differ from explanation (the original sense of the word in Greek), excuse, and justification? The utterance I apologize for X involves several presuppositions (in that word's looser sense) and at least one assertion (Fillmore 1971):

• Presuppositions: X is bad for A(ddressee) Sp regrets X Sp undertakes not to do X again Sp (or someone under Sp's control) is responsible for X Sp could have done otherwise
• Assertion: Speech act puts Sp one-down vis-a-vis A

At least one of these conditions is missing in excuses, justifications, and explanations. In an excuse, the speaker denies either his or her own responsibility (the cat made me do it) or ability to do otherwise (I tried to, but your phone was busy). In a justification, the speaker denies that the action was bad, if properly understood (everybody else gets to do it). In an explanation, the speaker takes responsibility for the action, but suggests that the addressee finds it bad because he or she does not understand it (I did it for your own good). So after apologies and excuses, the speaker ends up one-down; after justifications, both parties may be equal; and after an explanation, it is the recipient who ends up losing face as someone who does not get it. Explanations benefit their speakers, apologies their addressees.⁶

Semantic analyses like this can help us understand otherwise inexplicable choices in discourse. In 1983, Congress had passed a bill making the birthday of the Rev. Dr. Martin Luther King a national holiday. Conservatives were unhappy about this, one of them arguing that King was a man of immoral character whose frequent association with leading agents of communism is well-established. President Ronald Reagan, while privately indicating his agreement with that assessment, publicly waffled. Asked at a press conference whether he agreed with Senator Jesse Helms that King had had communist associations, the president said, We'll know in about 35 years, won't we?

With an election coming up, Reagan was urged by Democratic candidate Walter Mondale to apologize to King's widow. At first his spokesman said he would not, but eventually he phoned her. The call itself was not recorded, but asked later about its content, Coretta Scott King replied, He apologized to me. He said it was a flippant response to what he considered a flippant question.

Prudence might dictate that the Reagan forces leave bad enough alone here. But shortly thereafter an assistant press secretary found it necessary to correct Mrs. King's statement: It was an explanation, he said. He didn't mean the remarks the way they sounded.

Now, suppose that the president had uttered precisely the words Mrs. King attrib uted to him (which would be appropriately described by the press secretary's state ment). Why worry about whether It was a flippant response to what I consider a flippant question is an apology or an explanation? It might function as either: an apology for being flippant under inappropriate circumstances; or an explanation that they misunderstood a remark intended merely in jest.

The spokesman's insistence on defining the speech act differently from Mrs. King kept a divisive issue alive. There had to be a really good reason to do so. For presid ents, and especially an imperial president like Reagan, it is crucial not to be onedown, because that constitutes a loss of power and influence. It was obviously considered more important to avoid this consequence than to remain on good terms with the constituency of the late Dr. King. But we can only understand what otherwise looks like pointless and even damaging intransigence in high places if we understand the lexical semantics of apologizing, and the importance of protecting the president of the United States from FTAs (face-threatening acts: Brown and Levinson 1987).

Autonomous syntax does not have much to say about apologies. One might note the tendency of speakers to distance themselves from both the making of the apology itself, and the actions for which it offers redress, through indirect forms either subjunctive equivalents like: I want to apologize
I'd like to apologize
I guess I owe you an apology

or the placement of the speaker/wrongdoer in other than subject position, or out of the sentence altogether: It's too bad that X happened.
Sorry you got Xed.

or the sequestration of the apology in subordinate clauses, backgrounded and there fore less salient and accessible: I feel I owe you an apology.
It looks to me like an apology might be in order.

While strictly speaking these are syntactic choices, only an autonomous syntactician would characterize them as principally artifacts of syntax. Rather, the embedded or subjunctive syntax is the handmaiden of other aspects of the utterance pragmatics and semantics. We decide on the basis of semantics, pragmatics, and discourse con siderations how noticeable a role we want ourselves to play in our reports, and the syntax obligingly provides us with the means to represent ourselves as we would like to be seen (or not seen). Syntactic form must be part of a discussion of apology, but it cannot be considered meaningful in isolation.

Pragmatics occupies a realm intermediate between language-autonomous, decontextualized approaches and more complex theories entailing the consideration of the linguistic context and extralinguistic circumstances in which utterances occur. In his discussion of speech acts Austin (1962) referred to utterances rather than proposi tions or sentences, because he was talking about language use, rather than mere form. His title indicates that we do things with words. Since we alter reality by our utterances, it makes little sense to see language, or linguistics, as autonomous. In other ways, though, Austin's methods are akin to those of transformational syntax and its lineal descendants: the analysis of decontextualized structures constructed by the analyst.

Austinian analysis can help to explain both the numerousness and the specific forms of apologies, among them: I'm sorry I Xed.
I guess I Xed.
I shouldn't have Xed.
You must be pretty mad that I Xed.
I was a real jerk to X.
and I'll never X again.

Each of these forms comments on one of the conditions underlying the successful performance of an apology: a felicity condition in Austin's terminology, or preparat ory or essential condition (according to Searle 1969). The first example expresses the speaker's regret; the second assumes (though it hedges on) the speaker's responsibil ity for the act; the third, that the act was wrong; the fourth, that the addressee was hurt; the fifth puts the speaker clearly one-down; and the sixth promises that such a thing will never happen again. This point was originally made by Gordon and Lakoff in their theory of conversational postulates (1971), though without an explanation for why conversational postulates are used.

In stating explicitly that one of the conditions for a felicitous apology is met, without explicitly acknowledging that an apology is being performed, a speaker necessarily places considerable responsibility for endowing the act with meaning on the addressee. The latter makes use of Gricean (1975) conversational maxims and implicatures to understand why the speaker is saying something the addressee has no demonstrated need to know a flouting of the Gricean maxims of Quantity and Relevance. Ostensibly the addressee has no need to learn about the speaker's internal psychological state of regret but if the first example above can be understood as implicating an apology, with all the interpersonal baggage that that entails, the utter ance is clearly in obedience with the Cooperative Principle.

Although Austin framed his theory in terms of decontextualized utterances and assumed a strongly speaker-based perspective rather than seeing the discourse as created by all participants playing various roles, the interactive situations implied in his theory suggest a more contextualized, interactional model. For instance, Austin speaks of some speech acts as requiring certain forms of participation on the addressee's part to be felicitous. Thus, in a felicitous bet, an addressee has to say it's a bet, or you're on. Are apologies like bets in requiring some response, or some expectation, on the part of the addressee? If an addressee has no intention of accepting anything the speaker says, if no form at all will elicit forgiveness, Austin might say that no apology could be felicitous, but the fault would reside with the addressee rather than the speaker.

The apology battle between President Clinton and the Republican members of Congress in the fall of 1998 can be explained at least in part through this perspective.⁷ Both sides contributed to the impasse. On the one hand, the President refused to apologize until the last possible moment, when the semen-stained dress made its public appearance. Even at that it took three or more attempts before, in the eyes of the public and the pundits, he got it right. In his first attempt, on August 17, he was angry and belligerent rather than contrite. He called his behavior wrong and the relationship with Monica Lewinsky inappropriate, but did not say I'm sorry.

He tried again on a trip to Europe in early September. The physical distance between Sp and A probably made it easier to utter the apology, but made it less effective. In Moscow on September 2, Clinton said, I have acknowledged that I made a mistake, said that I regretted it, asked to be forgiven. The past tense reports of his earlier speech acts sound at first like apologies, but of course are not performative (as apologies must be), but merely reports of apologies, and therefore have no inter active value. On September 5, in Dublin, the President finally said that he was very sorry about the affair. But since he said it to people who were not the original addressees, not the people purportedly hurt by the behavior, again the utterance was not a felicitous apology.

On September 11, at a prayer breakfast, he tried again. With tears in his eyes, the report in the New York Times begins, the President admitted softly that I don't think that there is a fancy way to say that I have sinned. It should be noted that he has still not quite said I have sinned, but merely said that these words could be said. Indeed, though all the correct language is there in the rest of the speech, It is important to me that everybody who has been hurt know that the sorrow I feel is genuine. I have asked all for their forgiveness.

the expressions of contrition are all framed as indirect discourse, as presupposed rather than asserted, blunting their force and mitigating the speaker's responsibility. On the other hand, the nonverbal aspects are right in place, the tears and the soft voice. At this point the President's apology finally passed muster, suggesting that, as Ekman and Friesen point out, nonverbal signs mean more than verbal.

But even though the people, through the pollsters, voiced approval, the Repub licans in Congress continued to withhold it. Asked what it would take to get their forgiveness, several asserted that nothing would serve. If that was their assumption from the start, could any apology by the President have been felicitous?

The assumptions of speech act theory shed light on why the President may have made the choices he made (we can only guess at his, or anyone's, intentions); and why Americans responded to the repeated attempts as they did. Lexical semantics shows why the President was reluctant to use the s-word, even running serious risks by his refusal to do so. Speech act theory helps explain why people were dis satisfied with his attempts, but also suggests that for one intended set of addressees at any rate, nothing the President said could be a felicitous apology.

All participants in a discourse contribute to its meaning and perhaps even the form it takes (as Clinton's ultimate apology was shaped and reshaped by the reviews early versions got in the media). Utterances are situated in larger events, whether purely linguistic an encompassing utterance, a conversation or another human activity a ritual, a job, a performance. Hence, no single canonical apology form will fit with equal appropriateness into any context. From the perspective of the situated dis course event, what is required in an apology is subsumed under several categories, among them:

• register. Even for equally heinous behaviors, an apology made in a close family context is different from one that is made publicly. Between intimates an apology may not be required (love is never having to say you're sorry) for behavior for which one might be required in a more distant relationship. Different kinds of behavior may convey sincerity in intimate and in formal contexts (touching is often appropriate at home, less so in public).
• genre. In informal circumstances, a simple oral OK from the addressee may suffice to denote forgiveness. But in more formal settings (as in the settling of a lawsuit), a written statement exculpating the defendant may be required from the plaintiff to end the matter, with its wording carefully overseen by both sides.
• key (Hymes 1972). Under some conditions, an apology made ironically or other wise humorously may be acceptable. My father once offended me and later sent me a copy of The Portable Curmudgeon, which I took to be an apology (= I'm a curmudgeon all right, but I can't help it) and forgave him.

Conversation analysis (CA) as a research method has this analogy with autonomous syntactic analysis: because in both the analyst is prevented from dealing directly with meaning, intention, function, or understanding, the question What constitutes an apology? cannot be fully explored by either. Formal structures such as adjacency pairs can reveal what sort of second is preferred when the first member of a turn sequence is an apology.

For instance, the tools and methods of conversation analysis can clarify what con stitutes a preferred second in response to an apology. If a concern of linguists is the determination of what can occur grammatically in the context of something else, then if we are going to achieve a unified field and a cross-disciplinary perspective conversation analysis has to be able to address the question: what form does a pre ferred utterance take, and why? Traditional CA cannot do this, or cannot do it very well, because it does not permit introspection or mentalistic analysis. But (as analysts like Gumperz (1982) and Tannen (1984) have pointed out) without the ability to address questions of intention and effect, the analysis of conversation bogs down much the way pretransformational syntactic analysis did. To shed light on apologies from a CA perspective, the analyst must note that, of the various possible seconds available in response to an apology, different ones are more apt to co-occur with differently formed apologies:

But change A to A′:

and the assignment of ?'s shifts abruptly.

Traditional CA, of course, would never utilize constructed examples or mentalistic judgments like these. Yet there must be some way of talking about what speakers believe, find plausible, and use.

Narrative analysis has become fashionable in many fields, from literature to law, psychology, anthropology, history, and political science. All these fields have come to the realization that humans make sense of their lives through the stories they con struct. We develop psychological problems when our stories about our lives lack coherency (e.g. Schafer 1980; Spence 1982); in courtrooms, jurors determine whose story is more plausible, plaintiff or defendant, or whether the prosecutor's story has been successfully undermined by the defense attorney (cf. Delgado 1989). We can look at apologies as plot points in a story: what events led up to their making; how did the utterance of an apology move the story along? What happens when the internal stories of two people are in conflict A sees B as someone who owes A an apology; B either does not believe she or he has done anything wrong, or believes that their social differences are such that no apology is necessary?

We might look at the tale told earlier of Ronald Reagan and Coretta Scott King as involving just such a set of conflicting narratives. King expected an apology, Reagan did not believe one was in order, for both of the reasons suggested above. Reagan (or his people) was (or were) ingenious enough to construct an utterance that could satisfy the plots of two different groups of storytellers, creating (possibly) successful conclusions to two very different stories. (This happy outcome works best, of course, if the duplicity does not come to light as in this instance it did.)

When an apology is duly made and properly accepted, both parties come away satisfied. A good apology convinces both participants that their narratives are rational and permits both to have more or less happy endings. Even the humbled apologizer gets accepted back into the human fold, recognized as recognizing the need for an apology at this juncture, sharing with the addressee a common view of the narrative they have participated in creating. Even as apologizers are distanced momentarily from the fold of the virtuous, they are welcomed back as being, at any rate, competent.

Sociolinguistic analysis directly links the social group memberships of the pair in volved in the apology and their options and expectations in the event. Larger cultural background plays a significant role in the understanding of the need for apologies and the determination of their appropriate form. For instance, in many societies honor is important, and may both keep an apology from being made where an American might readily make one, and make a formal explicit apology requisite where we might do without one. Apology is always face-threatening for the speaker; but not making a necessary apology may occasion more serious face loss in the long run. As Brown and Levinson (1987) would say, the weightiness of a contemplated apology as a face threat must be computed by giving consideration to the intimacy and power relationships of the parties involved, and the seriousness of the misdeed that occasioned it.

Other extralinguistic issues are equally relevant. If, for instance, as Tannen (1994) suggests, women tend to use I'm sorry as a smoother of difficult moments, but men are less likely to do so, the genders will misunderstand each other (and women, as people who traditionally are interpreted by others, will suffer more from the mis understanding). Similarly, apologies raise the important question of when, how much, and in what way you divulge your real self or private persona to the world via language. As in the Cambridge Nanny case, when one culture believes it is shameful to let one's guard down at all in public, and another believes that the sincerity of a public apology is gauged by sobs, tears, and hand-wringing, it will be difficult for a member of one group to produce an apology that will at once gratify members of the other, and leave the apologizer herself or himself with any shred of self-esteem.

Finally, we can use much of the understanding gained at earlier levels to understand political and social events as reported in the media (both the choices of wording and the decisions as to what to discuss: the text and, perhaps, the metatext(s)). For instance, between the beginning of August and the end of September 1998 a large amount of space in the major American print media was dedicated to the analysis of and judgment upon the President's several apologies; polls of the American people, assessing their opinions about the satisfactoriness of each Presidential apology; reflec tions upon what apologies were and how they were appropriately made; and so on. We may deduce from this that apology had assumed a superhot, perhaps symbolic, importance at that moment (a search using Lexis-Nexus would tell the researcher that never before or since had the word apology received so much play in so many media over so long a time). At this level we can examine the subtext: why do we, whoever we are, require a show of contrition at this time? And why are the demanders never satisfied? Answers to these questions require the examination of language at all the levels discussed above. In this way, through concentration on a particular speech act, located in a specific cultural and societal time and place, we can come to understand a great deal about who we are, what we want, and the rules and assumptions that bind us together as a society.

I would like to thank Deborah Tannen for her perceptive comments and suggestions.

And there were many fewer areas of knowledge identified as disciplines or departments. Within the humanities, for instance, modern languages were recognized only in the late nineteenth century as valid subjects for university study. The first chair in English at Harvard was established in 1876; at Oxford, the English honors degree was created (with some sniping from traditionalists) in 1896 (Delbanco 1999). The social sciences are even newer, with anthropology and sociology dating from the first third of the twentieth century; departments of linguistics became commonplace only toward the end of the 1960s.

As an illustration, if the syntactician is permitted to offer analyses that take no cognizance of the fact that sentences are produced in the service of cognition and communication, then surely such analyses can function only as unintentional self-parodies, the ivory tower at its most aloof and irrelevant, social science turned antisocial (and not too scientific, since form divorced from function tends to offer very few useful or lasting generalizations).

Older readers may recall Steve Martin's line on Saturday Night Live, Well, excu-u-use me!, to precisely this effect.

The relation between real feelings and surface ones proves as intriguing as it is vexing for several disciplines. It manifests itself in Ekman and Friesen's (1975) distinction between automatic expressions of emotion that represent universal human instincts (e.g. scowling to express anger) and those that people learn as part of their culture's communicative repertoire (e.g. Japanese giggling, vs. American joking, to cover embarrassment); in the various distinctions made within several versions of transformational generative grammar (deep, abstract, underlying, or logical vs. surface structure); and in psychoanalytic discussion of the latent vs. manifest content of dreams, symptoms, and errors. Here is another point at which disparate fields come together in a common quest, obscured by differences in vocabulary and methodology.

This was a notorious and controversial case shown on Court TV and tirelessly reported in network news and magazine shows nightly. Louise Woodward, a young British national employed in Cambridge MA as a nanny, was accused of shaking the baby in her charge to death. The evidence was ambiguous. Found guilty by the jury, she was placed on probation by the judge and allowed to go free, both decisions provoking controversy among the public and experts of various stripes.

However, the popularity in high places of the adage Never apologize, never explain argues that the two may be closer than the above analysis suggests.

For the historical record: in January of 1998, evidence came to light that President Clinton had engaged in sexual conduct with a White House intern, Monica Lewinsky. Shortly thereafter on a television interview he said, I have never, at any time, had sexual relations with that woman, Miss Lewinsky. The question remained red-hot for several months, with continual denials on one side and insistences on the other. In August Lewinsky's semen-stained dress came to light, and subsequent DNA testing proved the semen to be the President's. Apologies were then demanded for exactly what (the sexual behavior; the untruthfulness; the fact that the statement had been accompanied by wagging/shaking his finger at us/you/the American people) was never precisely clarified.

REFERENCES

Austin, J. L. 1962. How to Do Things With Words Cambridge: Cambridge University Press .

Baker, Charlotte L. 1975. This is just a first approximation, but In T. J. San, L. J. Vance, and R. E. Grossman (eds), Papers from the Ekventh Regional Meeting of the Chicago Linguistic Society, pp. 37 47.

Brown, Penelope and Stephen R. Levinson. 1987. Politeness Cambridge: Cambridge University Press .

Delbanco, Andrew. 1999. The decline and fall of English literature . November 4, 1999:32 8.

Delgado, Richard. 1989. Storytelling for oppositionists and others: a plea for narrative Michigan Law Review

Ekman, Paul and Wallace, Friesen. 1968. The repertoire of nonverbal behavior: categories, origins, usage, and coding Semiotica

Ekman, Paul and Wallace, Friesen. 1969. Nonverbal leakage and cues to deception Psychiatry

Ekman, Paul and Wallace, Friesen. 1975. Unmasking the Face: A Guide to Recognizing Emotions from Facial Cues Englewood Cliffs, NJ: Prentice-Hall .

Fillmore, Charles J. 1971. Verbs of judging . In C. J. Fillmore and D. T. Langendoen (eds), Studies in Linguistic Semantics . New York: Holt, Rinehart, and Winston , pp. 273 90.

Gordon, David and George, Lakoff. 1971. Conversational postulates . In D. Adams, M. A. Campbell, V. Cohen, J. Lovins, E. Maxwell, C. Nygren, and J. Reighard (eds), Papers from the Seventh Regional Meeting of the Chicago Linguistic Society , pp. 63 84.

Grice, H. P. 1975. Logic and conversation . In P. Cole and J. L. Morgan (eds), Syntax and Semantics 3: Speech Acts . New York: Academic Press , pp. 41 58.

Gumperz, John J. 1982. Discourse Strategies London: Cambridge University Press .

Hochschild, Arlie Russell. 1983. The Managed Heart: Commercialization of Human Feeling . Berkeley and Los Angeles: University of California Press ,

Hymes, Dell. 1972. Models of the interaction between language and social life . In J. J. Gumperz and D. Hymes (eds), Directions in Sociolinguistics . London: Blackwell , pp. 35 71.

Lakoff, Robin T. 1973. Questionable answers and answerable questions . In B. Kachru et al. (eds), Issues in Linguistics: Papers in Honor of Henry and Rene Kahane . Urbana: University of Illinois Press , pp. 453 67.

Schafer, Roy. 1980. Narration in the psychoanalytic dialogue Critical Inquiry (Autumn 1980) : 29 53.

Schiffrin, Deborah. 1985. Conversational coherence: the role of well Language

Searle, John. 1969. Speech Acts Cambridge: Cambridge University Press .

Spence, Donald. 1982. Narrative Truth and Historical Truth New York: Norton .

Tannen, Deborah. 1984. Conversational Style Norwood, NJ: Ablex .

Tannen, Deborah. 1994. Talking from 9 to 5 New York: William Morrow .