Shame and Labeling Theory: Micro- and Macro-levels

Thomas J. Scheff

This study shows how a new level can be added to an existing theory. By analyzing the details of the dialogue in a single therapy sessions, I propose that my earlier labeling theory of mental illness can be enriched by including discourse analysis at a micro-level. The session that I use is between a anorexic woman and her therapist. The same session was analyzed in a well known study (Labov and Fanshel 1977). Their analysis, with my extention, shows how discourse analysis complements earlier work on labeling of mental illness. Unacknowledged shame is seen as a cause of both primary and secondary deviance. In family systems, it causes primary deviance, and in the interaction between the family and the community, it causes secondary deviance.

Labov and Fanshel's (1977) case, "Rhoda," provides an extended example. It suggests that the family labeling process is subtle and outside of awareness. In the anorexic patient's family, stigmatization is a two-way street; through innuendo, all the family members-including the patient- surreptitiously attack each other. But only the patient's violence has an overt component. Like the family members, she attacks others through innuendo, but she also starves herself. All the others use only emotional violence. Since the violence of the family is hidden, it is the patient who was formally labeled. The labeling appears to be extremely subtle and complex, inadvertently reaffirming the status quo in the family. To the extent that the mental health professions side with the family in cases like this one, they too reaffirm and maintain the status quo.

Labeling Theory

I consider myself lucky to be able to criticize my own original formulation of labeling theory (1966, 1984), because when I criticize the work of others, sometimes they take it personally. When I criticize my own work, I am free to view the criticism in its most favorable light, perhaps as a sign of intellectual growth, rather than showing how inadequate the original formulation was. Like most other theories of human behavior, the original theory was highly specialized, yet insufficiently detailed. It was specialized since it dealt behaviorally with social structure/process. It omitted most inner events, both those concerning mental illness and those concerning the societal reaction.

Following Lemert (1951), the original theory distinguished between primary deviance, such as the hallucinations and thought disorder that are taken to be symptoms of schizophrenia, and those same behaviors as they occur in a person who is aware of his or her label as being mentally ill. The theory proposed that much primary deviance is of short duration or of little significance in the life of the bearer. But when the person becomes aware of their label, they may come to play the role of the mentally ill, at first inadvertently, but later, perhaps, involuntarily. In other words, a group of symptoms may be stabilized, through self-consciousness and reaffirmation by others, as a "career" of mental illness.

The original theory, however was insufficiently detailed. First, it was formulated in terms of abstract concepts, "black boxes," that were not clearly defined, like most other general theories of human behavior.. Second, the causal links between these concepts were not specified. The theory described the societal reaction as a system without defining the major subsystems or the links between them.

My final criticism is substantive, pointing toward a major deficit at the core of labeling theory: its omission of emotions. Although Goffman (1963) and others discussed stigma, they paid too little attention to emotions, particularly the emotion of shame.

The original labeling theory was also oriented toward the formal labeling process, court hearings, and psychiatric examinations. Labeling (and non-labeling) in these contexts was crude and overt. But in the family, as we argue here, labeling is covert. It depends upon innuendo, manner, unstated implications, and especially emotion. To detect it, one must interpret words and actions in context.

At this stage of theory development, reliable methods may be premature since they strip away context. The next step in developing a theory may be to understand a single case very well. Part/whole analysis allows one to show the relationship between the smallest parts of discourse and the very largest social system (Scheff 1990; 1997). When one can demonstrate an understanding of the relation between parts and wholes in a series of cases, the stage is then set for a research mode oriented toward testing hypotheses. Since mental illness at this point is still a mystery and a labyrinth, we need to generate models that are interesting enough to warrant testing-ones that have face validity.

This chapter outlines a theory and method that specify the role of unacknowledged shame in mental illness and in the societal reaction. The theory involves a model of feeling traps, recursive loops of shame and anger. The method involves the systematic interpretation of sequences of events in discourse. Discourse from a psychotherapy session is used to illustrate the theory, to allow us to envision the hypothesized moment-by-moment causal sequence.

Pride, Shame, and the Social Bond

Cooley (1922) implied that pride and shame serve as intense and automatic bodily signs of the state of a system that is otherwise difficult to observe. Pride is the sign of an intact social bond; shame is the sign of a threatened one. The clearest outer marker of pride is holding up one's head in public, looking others in the eye, but indicating respect by alternately looking and looking away. In overt shame, one shrinks, averting or lowering one's gaze, casting only furtive glances at the other. In bypassed shame, one stares, outfacing the other.

Pride and shame thus serve as instinctive signals, both to the self and to the other, that communicate the state of the bond. By the state of the bond I mean the mix of solidarity and alienation in a particular social relationship. A completely secure bond would involve mutual understanding and mutual identification. A disrupted bond would involve no mutual understanding and no mutual identification. Most actual relationships fall somewhere between the two extremes. We react automatically to affirmations of and threats to bonds. But in early childhood most of us learn to disguise and ignore these signals. The idea of the social bond is repressed in modern societies, masked by the ideology of individualism. The emotions that express the bond-pride and shame-are also deeply repressed (Scheff 1990; and Lewis 1971).

Lewis's (1971) work is particularly relevant to the conjecture under discussion. She found that in contexts high in potential for shame, (as when a patient appears to suspect that the therapist is critical or judgmental), nonverbal indications of shame are plentiful. These include long or filled pauses ("well," "you know," "uh-uh-uh," and the like), repetition or self-interruption, and particularly, a lowering of the voice, often to the point of inaudibility. These markers are all suggestive of hiding behavior .

In these contexts, however, the painful affect of overt shame was virtually never acknowledged by name. Instead, other words were used, which Lewis interpreted to be a code language. Insecure, awkward and uncomfortable are several examples from a long list. This language is analogous to the code language for designating other unmentionables, such as sexual or "toilet" terms. Like babytalk about body functions used with children, the denial of shame is institutionalized in the adult language of modern societies. Lewis's findings, like the approaches of Cooley, Goffman, and especially Elias (1978, 1982), suggest that shame is repressed in our civilization.

Although Goffman's, Elias's, and Lewis's treatments of shame are an advance over Cooley's in one way, in another way they are retrograde. Their treatment is much more specialized and detailed than Cooley's, whose discussion of the "self-regarding sentiments" is casual and brief. But Cooley had a vision of the whole system lacking in the more recent discussions. His treatment construes pride and shame to be polar opposites. It therefore lays the basis for our construct of the social bond; pride and shame are continuous signals of the state of the bond, an instant read-out of the "temperature" of the relationship.

The emotion of pride is absent from Goffman's and Lewis's formulations. Goffman's omission of pride is particularly disastrous. Since Lewis dealt only with psychotherapy discourse, we are free to imagine from her work that in normal conversation there is more pride than shame. But Goffman's treatments of "impression management," "face," and embarrassment concerned normal discourse, leaving the reader with the impression that all human activity is awash in a sea of shame. He nowhere envisioned a secure social bond, much less a well-ordered society built upon secure bonds. Goffman's omission of pride and secure bonds is particularly misleading for the study of deviance; it undercuts a crucial distinction between "normal" persons and labeled persons. Social situations usually generate pride for the former and shame for the latter. This difference has extraordinary consequences for the social system.

Goffman's (1963) treatment of stigma, although perceptive and useful, is not complete. Like other labeling theorists, his discussion is specialized, focusing on the behavioral aspects of stigma. He acknowledged the emotional component of the societal reaction only in passing. In particular, he mentioned shame only twice, once early in the essay: "[for the deviant] shame becomes a central possibility," (7), and again at the end: "Once the dynamics of shameful differences are seen as a general feature of social life..." (140).

Goffman frequently referred to shame or shame-related affects, but only indirectly ("self-hate and self-derogation", [7]). Without a working concept of the relationship between emotion and behavior, Goffman and the other stigma theorists were unable to show its central role in mental illness and the societal reaction. As a step toward this end, it is first necessary to review once more how emotions may cause protracted conflict.

An earlier report (Scheff 1987) described emotional bases of interminable conflicts. Like Watzlawick et al. (1967), I argue that some conflicts are unending; any particular quarrel within such conflicts is only a link in a continuing chain. Both primary and secondary deviance arising out of interminable conflicts. What is the cause of this type of conflict?

Lewis (1971, 1976, 1981, 1983) proposed that when shame is evoked but not acknowledged, an impasse occurs that has both social and psychological components. Here I sketch a model of impasse, a triple spiral of shame and rage between and within interactants. When a person has emotional reactions to their own emotions and to those of the other party, both become caught in a "feeling trap" (Lewis 1971) from which they cannot extricate themselves. The idea that emotions are contagious between individuals is familiar; the concept of spirals subsumes contagion both between and within parties to a conflict.

A New Labeling Theory ·

My model follows from Lewis's (1971) analysis of therapy transcripts: shame is pervasive in clinical interaction, but it is invisible to interactants (and to researchers), unless Lewis's approach is used. (For methods that parallel Lewis's, see Gottschalk and Glaser's [1969] "shame-anxiety scale.")

Lewis (1971) referred to the internal shame-rage process as a "feeling trap," as "anger bound by shame" or "humiliated fury." Kohut's (1971) concept, "narcissistic rage," appears to be the same affect, since he viewed it as a compound of shame and rage. When one is angry that one is ashamed, or ashamed that one is angry, then one might be ashamed to be so upset over something so "trivial." Such anger and shame are rarely acknowledged and are difficult to detect and to dispel. Shame-rage spirals may be brief, lasting a matter of minutes, or they can last for hours, days, or a lifetime, as bitter hatred or resentment.

Brief sequences of shame-rage may be quite common. Escalation is avoided through withdrawal, conciliation, or some other tactic. In this chapter a less common type of conflict is described. Watzlawick et al. (1967, 107-108) called it "symmetrical escalation." Since such conflicts have no limits, they may be lethal. I describe the cognitive and emotional components of symmetrical escalation, as far as they are evidenced in the transcript.

Labeling in the Family: A Case Study (based on Scheff, 1989)

Labov and Fanshel (1977) conducted an exhaustive microanalysis of a large segment of a psychotherapy session. They analyzed not only what was said but also how it was said, interpreting both words and manner (the paralanguage). They based their interpretations upon microscopic details of paralanguage, such as pitch and loudness contours. Words and paralanguage are used to infer inner states: intentions, feelings, and meanings.

With such attention to detail, Labov and Fanshel were able to convey unstated implications. Their report is evocative; one forms vivid pictures of patient and therapist and of their relationship. One can also infer aspects of the relationship between Rhoda and her family, since Rhoda reports family dialogues. Labov and Fanshel showed that the dispute style in Rhoda's family is indirect: conflict is generated by nonverbal means and by implication.

Indirect inferences, from a dialogue that is only reported, are made in order to construct a causal model. Obviously, in future research they will need to be validated by observations of actual family dialogue. It is reassuring, however, to find that many aspects of her own behavior that Rhoda reports as occurring in the dialogues with her family are directly observable in her dialogue with the therapist. For example, the absence of greeting, and Rhoda's covert aggression in the dialogue she reports with her aunt can be observed directly in the session itself (not included in this chapter but discussed in Scheff 1989).

The Feud Between Rhoda and Her Family

Rhoda was a nineteen-year-old college student who had a prior diagnosis of anorexia. She had been hospitalized because of her rapid weight loss, from 140 to 70 pounds. When her therapy began, she weighed 90 pounds. At five feet five inches in height, she was dangerously under-weight.

Her therapy sessions took place in New York City in the 1960s. Rhoda lived with her mother and her aunt, Editha; her married sister also figures in the dialogue. The session that was analyzed by Labov and Fanshel was the twenty-fifth in a longer series, which appeared to end successfully. The therapist reported improvement at termination. At a five-year follow-up, Rhoda was of normal weight, married, and raising her own children.

Labov and Fanshel focused on the web of conflict in Rhoda's life, mainly with her family and to a lesser extent with her therapist. The conflict was not open but hidden. The authors showed that Rhoda's statements (and those she attributed to the members of her family) were packed with innuendo. They inferred that the style of dispute in Rhoda's family was indirect: although the family members were aggressive toward each other and hurt by each other, both their aggression and their hurt were denied.

Labov and Fanshel's method was to state explicitly as verbal propositions what was only implied in the actual dialogue. This method proposed a cognitive structure for the conflict in Rhoda's family: it translated utterances, words, and paralanguage into purely verbal statements. The set of verbal statements served as a compact, clarifying blueprint for a dense tissue of complex maneuvers that were otherwise difficult to detect and understand.

In addition to this type of analysis, Labov and Fanshel also used another. Following the lead of the therapist, they pointed out cues that were indicative of unacknowledged anger. To reveal this emotion, they used verbal and nonverbal signs: words and paralanguage (such as pitch and loudness). Hidden challenges in Rhoda's family were made in anger and resulted in anger. Rhoda's therapist made explicit reference to this matter: "So there's a lot of anger passing back and forth" (5.27[c] [The numbers refer to the Rhoda transcript, pp. 363-371, Labov and Fanshel 1977]). There were also myriad indications of unacknowledged anger and other emotions in the session itself.

Emotions were not central to Labov and Fanshel's study, but they are to mine. Building upon their assessment of cognitive conflict, and their (and the therapist's) analysis of anger, I show shame sequences in the session that were apparently unnoticed by both patient and therapist. Labov and Fanshel frequently noted the presence of embarrassment and of the combined affect they called "helpless anger," but they made little use of these events.

My study leads me to conclude that labeling occurs at two different levels-the informal and the formal. At the informal level, labeling is quite symmetrical: Rhoda labeled and blamed Aunt Editha and her mother just as much as they labeled and blamed her. The family members casually insulted each other almost constantly. In some sentences, several different insults were implied at once. As Labov and Fanshel pointed out, conflict seemed to be endemic in this family.

At the formal level of labeling, however, there was no symmetry whatsoever. Although the mother and the aunt were just as violent with their insults, threats, and rejections as Rhoda, it was only Rhoda who was physically violent; she tried to starve herself. In contrast to the constant verbal violence, Rhoda's overt violence was highly visible; her dangerously low body weight bore ostensible witness to her self-assault. Although the verbal violence seemed to be visible to the therapist and was documented by Labov and Fanshel, it was invisible in Rhoda's community. If labeling theory is going to lead to further understanding of mental illness, it will need to take a new direction, to make visible what has hitherto been invisible; violence in the micro-world of moment-to-moment social interaction.

I use two excerpts (Labov and Fanshel 1977: 364, 365). The first involves Rhoda's relationship with her mother; the second, with her Aunt Editha. The first excerpt occurred early in the session-it deals with a telephone conversation that Rhoda reported. The mother was temporarily staying at the house of Rhoda's sister, Phyllis. (Since pauses were significant in their analysis, Labov and Fanshel signified their length: each period equals .3 second.)

Excerpt 1

1.8 R.: An-nd so-when-I called her t'day, I said, "Well, when

do you plan t'come home?"

1.9 R.: So she said, "Oh, why!"

1.10 R.: An-nd I said, "Well, things are getting just a little too

much! [laugh] This is-i's jis' getting too hard,

and...... I-"

1.11 R.: She s'd t'me, "Well, why don't you tell Phyllis that!"

1.12 R.: So I said, "Well, I haven't talked to her lately."

Rhoda, a full-time student, argues that she can't keep house without help. Her mother puts her off by referring her to Phyllis. The implication-that the mother is there at Phyllis's behest-is not explored by the therapist. Rather, she asks Rhoda about getting help from Aunt Editha. Rhoda's response:

Excerpt 2

2.6 [a] R.: I said t'her (breath) w-one time-I asked her-I said

t'her.

[b] "Wellyouknow, wdy'mind takin' thedustrag an'justdust

around ?"

2.7 R.: Sh's's, "Oh-I-I-it looks clean to me,"...

2.8 [a] R.: An' then I went like this.

[b] an' I said to her, "That looks clean t'you?"

(It appears that at this point, Rhoda had drawn her finger across a dusty surface and thrust her

dusty finger into Editha's face)

2.9 [a] R.: And she sort of....I d'no-sh'sort of gave me a funny

look as if I-hurt her in some way,

[b] and I mean I didn' mean to, I didn' yell and scream.

[c] All I did to her was that "That looks clean to you?"....

The therapist persists that Rhoda may be able to obtain help from Editha. In a later segment (not shown), Rhoda denies this possibility.

Rhoda's Helpless Anger Toward Her Aunt

I will begin analysis with the least complex segment, the dialogue that Rhoda reports between herself and her aunt (2.5-2.9). Labov and Fanshel showed a thread of underlying anger, anger that is denied by both parties.

Rhoda has explained prior to this excerpt that dust "bothers" her-that is, makes her angry. The authors argue that the request that Editha "dust around" (2.6[b]) involves an angry challenge to Editha's authority, a challenge that neither side acknowledges. It assumes that the house is dusty, that Editha knows it, that she has ignored her obligation to do something about it, and that Rhoda has the right to remind her of it. Although Rhoda uses "mitigating" devices, speaking rapidly and casually, she ignores the etiquette that would have avoided challenge.

(Labov and Fanshel wrote, "The making of requests is a delicate business and requires a great deal of supporting ritual to avoid damaging personal relations surrounding it" [96].) To avoid challenge, Rhoda might have begun with an apology and explanation: * "You know, Aunt Editha, this is a busy time for me, I need your help so I can keep up with my schoolwork." Rhoda's actual request is abrupt. (As customary in linguistics, an asterisk (*) is used to denote a counterfactual, a hypothetical statement not made in the actual dialogue.)

Editha's response is also abrupt: "Oh-I-I-it looks clean to me..." She has refused Rhoda's request, intimating inaccuracy in Rhoda's appraisal. The ritual necessary to refuse a request without challenge is at least as elaborate as that of making one. Editha could have shown Rhoda deference: *"I'm sorry Rhoda, but...," followed by an explanation of why she was not going to honor the request.

Rhoda's response to what she appears to have taken as an insult is brief and emphatic: She contemptuously dismisses Editha's contention. She wipes her finger across a dusty surface and thrusts it close to Editha's face: "That looks clean to you?" Labov and Fanshel noted the aggressive manner in Rhoda's rebuttal: she stresses the words that and clean, as if Editha were a child or hard of hearing. They identified the pattern of pitch and loudness as the "Yiddish rise-fall intonation": *"By you that's a monkey wrench?" implying repudiation of the other's point of view. "If you think this is clean, you're crazy" (202). Rhoda's response escalates the level of conflict: she has openly challenged Editha's competence.

Finally, Rhoda describes Editha's response, which is not verbal but gestural: she gives Rhoda a "funny look as if I-hurt her in some way." Rhoda denies any intention of hurting Editha, and that Editha has any grounds for being hurt: "I didn't yell and scream," implying that Editha is unreasonable.

Labov and Fanshel noted the presence of anger not only in the original interchange but in Rhoda's retelling of it. The nonverbal signs, they said-choking, hesitation, glottalization, and whine-are indications of helpless anger: Rhoda "is so choked with emotion at the unreasonableness of Editha's behavior that she can not begin to describe it accurately" (191). Helpless anger, the authors wrote, characterizes Rhoda's statements throughout the whole session: "she finds herself unable to cope with behavior of others that injures her and seems to her unreasonable" (191).

Labov and Fanshel further noted that her expressions of helpless anger were "mitigated":

All of these expressions of emotion are counterbalanced with mitigating expressions indicating that Rhoda's anger is not extreme and that she is actually taking a moderate, adult position on the question of cleanliness. Thus she is not angered by the situation, it only "bothers" her. Even this is too strong; Rhoda further mitigates the expression to "sort of bothers me."

Mitigation in this instance means denial: Rhoda denies her anger by disguising it with euphemisms.

What is the source of all the anger and denial? Let us start with Rhoda's helpless anger during her report of the dialogue. Helpless anger, according to Lewis (1971), is a variant of shame-anger: we are ashamed of our helplessness. In retelling the story, Rhoda is caught up in a shame-anger sequence: shame that she feels rejected by Editha, anger at Editha, shame at her anger, and so on.

Helpless anger has been noted by others besides Lewis and Kohut. Nietzsche (1887) referred to a similar affect ("impotent rage") as the basis for resentment. Scheler (1912) used Nietzsche's idea in his study of ressentiment-pathological resentment. Horowitz (1981), finally, dealt with a facet of helpless anger under the heading "self-righteous anger."

Rhoda and her family are caught in a web of ressentiment, to use Scheler's term. Each side attributes the entire blame to the other; neither side sees their own contribution. As Labov and Fanshel showed, one of Rhoda's premises is that she is reasonable, and the members of her family are unreasonable. The reported dialogues with her family imply that the family holds the opposite premise: that they are reasonable, but she is unreasonable.

My theory suggests that the dialogue between Rhoda and Editha is only a segment of a continuous quarrel. Since it is ongoing, it may not be possible to locate a particular beginning; any event recovered is only a link in a chain. (Watzlawick et al. 1967; 58). Starting at an arbitrary point, suppose that Rhoda is "hurt" by Editha's failure to help. That is, she feels rejected, shamed by Editha's indifference, and angry at Editha for this reason. She is also ashamed of being angry, however. Her anger is bound by shame. For this reason it cannot be acknowledged, let alone discharged.

Editha may be in a similar trap. Rhoda is irritable and disrespectful, which could cause Editha shame and anger. She could experience Rhoda's hostility as rejecting, arousing her own feelings of helpless anger. Reciprocating chains of shame and anger on both sides cause symmetrical escalation.

The Impasse between Rhoda and Her Mother

Excerpt 1, as reported by Rhoda, may point to the core conflict. It is brief-only three complete exchanges-but as Labov and Fanshel showed, it is packed with innuendo. My analysis follows theirs, but expands it to include emotion dynamics.

Rhoda's first line, as she reports the conversation, is seemingly innocuous: "Well, when do you plan t'come home?" To reveal the unstated implications, Labov and Fanshel analyzed understandings about role obligations in Rhoda's family. Rhoda's statement is a demand for action, disguised as a question. They pointed out affective elements: it contains sarcasm (1977: 156), criticism (161), challenge (157, 159), and rudeness (157). The challenge and criticism are inherent in a demand from a child that implies that the mother is neglecting her obligations.

Implicit in their comments is the point I made about Rhoda's approach to her aunt. It was possible for Rhoda to have requested action without insult, by showing deference, reaffirming the mother's status, and providing an explanation and apology. Rhoda's request is rude because it contains none of these elements.

Rhoda's habitual rudeness is also indicated by the absence of two ceremonial forms from all her dialogues, not only with her family, but also with her therapist: any form of greeting, and the use of the other's name and title. Does Rhoda merely forget these elements in her report of the dialogues? Not likely, since they are also missing in the session itself. Labov and Fanshel tell us that the transcript begins "with the very first words spoken in the session; there is no small talk or preliminary settling down.... Instead the patient herself immediately begins the discussion." Rhoda neglects to greet the therapist or call her by name. Since Rhoda is junior to the therapist, her aunt, and her mother, the absence of greeting, name, and title is a mark of inadequate deference toward persons of higher status. Rhoda's casual manner is rude.

The mother's response is just as rude and just as indirect. According to Rhoda's report, her mother also neglects greetings and the use of names. Like Rhoda's aunt, she neither honors the request nor employs the forms necessary to avoid giving offense. Rather than answering Rhoda's question, she asks another question-a delay that is the first step in rejecting the request.

Labov and Fanshel stated that the intonation contour of the mother's response ("Oh, why'") suggests "heavy implication." They inferred: *"I told you so; many, many, times I have told you so." (When Rhoda gives a second account of this dialogue [4.12-4.15], she reports that the mother actually said, "See, I told you so.") What is it that the mother, and presumably others, has told Rhoda many times? The answer to this question may be at the core of the quarrel between Rhoda and her family.

Whether it is only an implication or an actual statement, the mother's I-told-you-so escalates the conflict from the specific issue at hand- whether she is going to come home-to a more general level: Rhoda's status. Rhoda's offensiveness in her opening question involves her mother's status only at this moment. The mother's response involves a general issue. Is Rhoda a responsible and therefore a worthwhile person, or is she sick, mad, or irresponsible?

Labeling, Shame, and Insecure Bonds

At a superficial level, the mother's I-told-you-so statement involves only Rhoda's ability to function on her own. As can be seen from Rhoda's complaints at the end of the session, however, this implication is symbolic of a larger set of accusations that Rhoda sees her mother and sister as leveling at her: she is either willfully or crazily not taking care of herself, starving herself, and she doesn't care about the effect of her behavior on her family. Her family's basic accusation, Rhoda feels, is that she is upsetting them, but she doesn't care. Rhoda formulates this accusation at the end of the transcript.

Excerpt 3

T. What are they feeling?

5.26 R. ...that I'm doing it on purp-like, I w's- like they ...

well-they s-came out an'tol' me in so many words that

they worry and worry an' I seem to take this very lightly.

To Rhoda, the mother's I-told-you-so epitomizes a host of infuriating, shaming charges about her sanity, responsibility, and lack of consideration. Note particularly that the labeling process to which Rhoda refers here is not explicit; it occurs through innuendo.

The labeling of Rhoda by the other family members and its emotional consequences underlie the whole family conflict. Yet it can be detected only by a subtle process of inference, understanding the meaning of words and gestures in context, in actual discourse. Both the theory and the method of the original labeling theory were too abstract to detect this basic process.

Rhoda responds (in 1.10) not to the underlying implication of her mother's evasion but to the surface question, ""Why do you want to know?" Because, she answers, "...things are getting just a little too much..." The key element in Rhoda's response is the affect. Labov and Fanshel stated that the paralanguage (choked laughter, hesitation, glottalization, and long silence [170]) is an indication of embarrassment (171). Rhoda responds to her mother's accusations by becoming ashamed. The shame sequence that is described is a marker for stigmatization that is otherwise hidden behind polite words.

Rhoda's shame may indicate that she feels that her family's charges have some basis, or that the implied rejection leads her to feel worthless, or both. Since no anger is visible at this instant, it is either absent or bypassed. The verbal text, however, suggests that Rhoda is feeling shame and guilt. She is acknowledging that she needs her mother-a need she has repeatedly denied in the past. She may feel that she is at fault for this reason.

Labov and Fanshel contrasted the force of the mother's response with the weakness of Rhoda's comment (at 1.10). The mother says, "Why don't you tell Phyllis that?" Labov and Fanshel stated that the hesitation and embarrassment that characterize 1.10 are absent from this response. It is a forceful rejection of Rhoda's claims and, by implication, a criticism of Rhoda for even making the request. The mother's emotional response to Rhoda's embarrassment is not simply unsympathetic; it is aggressively rejecting. From the emotional standpoint, Rhoda's back is to the wall. She is trapped in the helpless role of the blamed, with her mother as the aggressive blamer.

The analysis of shame in this dialogue points to an otherwise hidden issue. At this moment we can see that in her family, Rhoda has literally no one to whom she can turn. She is at odds with her aunt. We know from her reports of her sister's comments that Rhoda and she are also in a tangle. No father is mentioned. Rhoda and her family are in a perpetual war, a war hidden beneath the surface of conventional discourse. All of Rhoda's bonds are threatened, yet she has no way of understanding her complete alienation.

The stage is set for violent emotion and/or violent behavior: for mental illness (Rhoda appears to be delusional about her eating and body weight), murder, or suicide (in this case, self-starvation). That the potential for suicide arises when individuals have no one to whom they can turn was conjectured by Sacks [1966] on the basis of his analysis of calls to a suicide prevention center. The repression of shame and the bondlessness that is its cause and effect can give rise to primary deviance in the form of mental illness, murder, or suicide.

In Rhoda's response (1.12), she continues in the role of the one at fault: "Well, I haven't talked to her lately." Her mother has defeated her on all counts. She has refused Rhoda's request without the ritual that would protect Rhoda's "face"; she has implied a victory over Rhoda ("I told you so") that undercuts Rhoda's status, and she has criticized her for making an inappropriate request to the wrong person.

Rhoda appears to feel too baffled, upset, and helpless for an angry counterattack. Her anger at her mother may feel too shameful to countenance. It is reserved for lesser targets: her aunt, her sister, and the therapist. Her mother's rejection, with the implied threat of abandonment, could be the basic source of Rhoda's shame.

Even to the casual reader, the mother's tactics are transparent. Why is Rhoda so baffled by them? Why didn't she use a response like the one suggested by the authors: *"Oh, come off it, Ma! You know it's really up to you when you come home, not Phyllis. Get off my case!"

Rhoda's ineptness may be due to her intense shame, evoked beginning with the first question, asking her mother for help. In this instance the massiveness of the unacknowledged shame is befuddling almost to the point of paralysis.

In the overt form of shame, one is so flustered that speech is disrupted, with inaudibility, repetition, stuttering, and fragmentation. Even though she is only reporting the dialogue, Rhoda's speech shows many of these markers. Bypassed shame, on the other hand, may disrupt one's ability to think clearly, forcing one into a holding pattern, repeating set responses not particularly appropriate to the moment (Scheff 1987). This dialogue suggests that Rhoda is overwhelmed with both kinds of shame.

At the heart of the quarrel is a series of threats between Rhoda and her mother. As in all interminable quarrels, it is not possible to identify the first link. I begin with Rhoda's basic threat, without signifying that it came first: *"If you don't stop shaming me, I will starve myself!" Her mother's basic threat: *"If you don't stop shaming me, I'11 abandon you!"

The abandonment threat in this case is literal: the mother has left Rhoda to stay with her other daughter. Normally, the threat of abandonment would be largely symbolic; carrying out a threat of abandonment is probably rare. But whether it is real or symbolic, threats of abandonment may be the key link in the causal chain.

This chain has potentially lethal force because none of it is visible to both participants. There are four links: (1) Rhoda's shame in response to her mother's behavior toward her; (2) her threat to starve; (3) the mother's shame in response to Rhoda's behavior; (4) her threat to abandon Rhoda. Rhoda is aware of none of these links. Nearest to her awareness is the mother's threat to abandon her, and next, the shaming by her mother. Rhoda is unaware that her mother is shamed by Rhoda's aggressive and self-destructive behavior, and she denies that she is starving herself. The mother is aware of only one link: Rhoda's threat to starve herself. Because of this awareness, she talks to and about Rhoda in code, not daring to mention Rhoda's threat. Her shame over Rhoda's behavior, her own shaming of Rhoda, and her threat to abandon Rhoda apparently are not experienced by her.

The driving force in the quarrel is not the anger that was interpreted by the therapist but the shame in the field between Rhoda and her family. The anger in this family is both generated and bound by shame. Rhoda experiences her mother's threat of abandonment and her mother's anger as shaming. The mother experiences Rhoda's threat of self-starvation and Rhoda's anger as shaming. The symmetry is complete: each side is threatened and shamed by the other, and each side can see only the other's threat.

The system of threats and hidden emotions is comparable to that preceding conflict between nations (Scheff 1994). Each side feels its credibility would be diminished by backing down in the face of threat. Each side therefore escalates the level of threat. The resulting emotions have no limit, unless outside mediation occurs or shame is dispelled. "War fever" may be code language for collective shame-rage spirals.

The theory advanced here attempts to explain the emotional sources of mental illness, and the excessive force of the societal reaction to mental illness, the roots of primary and secondary deviance. Rhoda and her family are caught in an interminable conflict that is driven by triple spirals of shame and anger within and between the disputants. For brevity, I have not included my (1989) analysis of the transaction between Rhoda and her therapist, but because of its relevance to the argument, I provide a brief summary.

Although Rhoda attacks the therapist surreptitiously, using the same tactics she uses against the authority figures in her family-her mother and her aunt-the therapist is too wily to become emeshed in them. She gets angry, but she doesn't attack Rhoda back, as Rhoda's mother and aunt do. By avoiding emeshment in the family conflict, the therapist is able to form a secure bond with Rhoda, leading ultimately to a successful course of therapy.

Research in the labeling tradition suggests that therapists like this one are probably rare. Therapists and other agents outside the family often become emeshed in family conflicts, usually aiding with the family against the patient. Bowen's (1978) seminal analysis of family systems implies this course. Several of our earlier case studies illustrate the emeshment of the outside agents on the side of the family (Retzinger 1989; Scheff 1966, 1987).

Labeling by Psychiatrists

Retzinger's study (1989) of a psychiatric interview goes further; she shows how the psychiatrist is emeshed with the family position and how this emeshment leads to renewed psychiatric symptoms, as predicted by Lewis's theory (1981). The theory proposed here explains the extraordinary forces underlying mental illness and the reaction to it, chain reactions of shame and anger, feelings traps both in patients and in those reacting to them.

A recent study of mental illness using a strictly biographical method has produced findings parallel to ours. Porter (1990) provides an even-handed assessment of endogenous and environmental contributions to the mental illness of a large number of well-documented cases. His summary of the findings for one case-the nineteenth-century patient John Perceval-can be taken to represent his conclusions for the majority of his cases:

Perceval believed that religious terror had brought on his insanity, and that the behavior of his family had exacerbated it. But the real cause of the appalling severity and prolongation of his condition was the medico-psychiatric treatment he had received. Perceval unambiguously condemned as intrinsically counter-productive the very philosophy of placing mad people in lunatic asylums. It set the lunatic amongst "strangers" precisely when he needed to be with his fellows in familiar surroundings. It estranged him from his family. It put him in the charge of an unknown doctor, rather than those members of the caring professions he knew well, his regular physician or his clergy man. It set him in the midst of fellow lunatics, who, if truly mad, must surely be those people least capable of sustaining the mind of one who had just been crushed under a terrible blow. Precisely at the moment when a person needed his morale to be boosted, he was thrown into a situation that must "degrade him in his own estimation." (180-81, emphasis added)

This statement clearly supports labeling theory and points particularly to the two elements in labeling that are emphasized by the new theory: the weakening of social bonds and the accompanying unacknowledged shame. That the psychiatric treatments of the composer Robert Schumann and the dancer Nijinsky resulted in the complete severing of their social bonds is particularly shocking (Porter 1990: 65-71, 71-75).

The position that Porter, a historian, takes toward his findings seems equivocal. He cites none of the labeling theory literature; he states that he sides neither with the patients nor with the psychiatrists. Yet his closing message acknowledges some strain. The first line of his conclusion reads, "This book has not pleaded a cause" (231). He goes on to say that his aim has been merely to focus attention on a body of forgotten writings, the memoirs of the mad. Yet at the end of the conclusion, he states, "clearly, no reader will have taken the opening statement of this Conclusion at face value" (232). For reasons that are never stated, Porter is reluctant to acknowledge the implications of his findings. He seems to make the error of equating taking a stand on his own findings with "pleading a cause."

Conclusion

The case in this chapter contains the three elements fundamental to my theory: inadequate bonds, dysfunctional communication, and destructive conflict. Before her contact with the therapist, Rhoda appears to have been alienated from everyone in her family. No father is mentioned, and she seems to have the barest cognitive attunment with her mother and aunt and virtually no understanding at the emotional level.

The dialogues with her mother and aunt that Rhoda reports clearly indicate dysfunctional patterns of communication. She and her mother are extremely indirect, evasive, and withholding with each other, and she and her aunt are violently disrespectful, although in underhanded ways. It is of great interest that although she tries the same tactics she used in her family on the therapist, the therapist is able to sidestep them, giving Rhoda what turns out to be important lessons in how to communicate directly but respectfully.

The theme of violence is present in these dialogues only in the form of Rhoda's attempts at self-starvation. Like virtually all the other important issues in Rhoda's family, these attempts are disguised and denied: Rhoda claims that she is only dieting and that she is not underweight. As in most important issues in social communication, contextual, prospective, and retrospective knowledge beyond the discourse itself is needed to interpret the meaning of statements and events.

Support for the theory is also found in the cues for hidden emotion that both the therapist and Labov and Fanshel point out in their interpretations. Although the therapist only interprets Rhoda's and her family's anger, Labov and Fanshel's careful analysis of microscopic cues in verbal and nonverbal behavior provides support for our theory of shame-rage spirals. Their analysis frequently pointed to instances of "embarrassment" (shame) and "helpless anger" (shame-anger), suggesting the sequences required by the theory. The theory of shame-rage spirals fills in the wiring diagram of the black boxes in labeling theory: unacknowledged alienation and shame drive the labeling machine.

As in its earlier formulation, our extended labeling theory implies a critique of conventional psychiatry, which is individualistic and affirms the status quo. In focusing exclusively on Rhoda's pathology, it denies the pathology in the family system of which she is a part and, by implication, in the larger social system, our current civilization.

The next step in following up the theory and method developed here would be a systematic comparative study of labeling and non-labeling in actual families (rather than conversations reported by the patient in the Rhoda study) and outside the family, in schools, jobs and in psychiatric examinations. Unifying case study and comparative methods, as I am suggesting here, is a return to the morphological method used in the development of the science of biology, rather than the specialization and compartmentalization of theory and method current in the social sciences.

References

Bowen, M. 1978. Family Therapy in Clinical Practice. New York: J. Aronson.

Braithwaite, J. l989 Crime,Shame,and Reintegration. Cambridge: Cambridge University Press.

Cooley, C. H. , l922. Human Nature and the Social Order. New York: Scribners.

Elias, Norbert. l978, l982, l983. The Civilizing Process, V. 1-3. New York: Vintage

Goffman, E. 1963. Stigma. Englewood Cliffs: Prentice-Hall.

l967. Interaction Ritual. New York: Anchor.

Gottschalk, L., Wingert, C. and Gleser, G.l969. Manual of Instruction for Using the Gottschalk-Gleser Content Analysis Scales. Berkeley: U. of California Press.

Horowitz, M. 1981 Self-righteous Rage. Archives of General Psychiatry. 38: (Nov.), 1233-38.

Kohut, H.E. 1971. "Thoughts on narcissism and narcissistic Rage". The Search for the Self. New York: International University Press.

Labov, W. & Fanshel, D. 1977. Therapeutic Discourse. New York: Academic Press.

Lemert, E. 1951. Social Pathology. New York: McGraw-Hill.

Lewis, H. l971. Shame and Guilt in Neurosis. New York: International Universities Press.

1976. Psychic War in Men and Women. New York: New York University Press.

1979. "Using content analysis to explore shame and guilt in neurosis". In Gottschalk (Ed.), The Content Analysis of Verbal behavior, New York: Halstead Press.

1981. Freud and Modern Psychology. V.1. The Emotional Basis of Mental Illness. New York: Plenum.

1983. Freud and Modern Psychology. V.2. The Emotional Basis of Human Behavior. New York: Plenum.

Nietzsche, F. 1887 On the Genealogy of Morals. New York: Vintage. (1967)

Porter, R.l990. A Social History of Madness. New York: E.P. Dutton.

Retzinger, S. 1989. "A theory of mental illness: Integrating social and emotional aspects". Psychiatry, 52 (3):325-35.

l991. Violent Emotions: Shame and Rage in Marital Quarrels. Newbury Park: Sage Publications.

Sacks, H. l966 The Search for Help: No One to Turn To. unpublished Ph.D dissertation, UC, Berkeley. University Microfilms.

Scheff, T. 1966. Being Mentally Ill. Chicago: Aldine. Second Edition, 1984).

l984 The Taboo on Coarse Emotions. Review of Personality and Social Psychology. 5, 146-169.

l987. The Shame-Rage Spiral: Case Study of an Interminable Quarrel. in H.Lewis (Editor), The Role of Shame in Symptom Formation. Hillsdale, NJ: Erlbaum.

l989. Cognitive and Emotional Components in Anorexia: Re-analysis of a Classic Case. Psychiatry 52, l48-l60.

l990. Microsociology: Discourse, Emotion and Social Structure. Chicago: U. of Chicago Press.

1994. Bloody Revenge. Boulder: Westview Press

1997. Emotions, Social Bonds, and Human Reality: Part/Whole Analysis. Cambridge: Cambridge University Press.

Scheff, T. and Retzinger, S. 1991. Emotions and Violence. Lexington, MA.: Lexington Books.

Scheler, M. 1912 Ressentiment. Glencoe: Free Press (1961).

Thoits, P. l985. Self-Labeling Processes in Mental Illness: The Role of Emotional Deviance. American Journal of Sociology 91: 221-248.

Watzlawick, P., Beavin, J.H., and Jackson, D. 1967 The Pragmatics of Human Communication. New York: Norton.