EMOTIONS, THE SOCIAL BOND AND HUMAN REALITY: PART/WHOLE ANALYSIS*
First two introductions and first chapter only.
Thomas J. Scheff
THEORY AND METHOD 11
l. Part/whole morphology: Unifying theory, method and data 14
2. The limits of literary analysis of texts 39
GENERATING THEORY: THE SOCIAL BOND 53
3. Punishment, child development, and the social bond 56
4. Boytalk-girltalk: A theory of social integration 77
5. Origins of the First World War: an interdisciplinary approach 92
GENERATING THEORY: EMOTIONS AND CONFLICT 118
6. Gender wars: love and conflict in Much Ado About Nothing. 121
7. Microanalysis of discourse: the case of Martha Johnson and her therapist 137
8. Conflict in family systems 163
9. Conclusion: integrating the human sciences 179
Appendix: Retzinger's coding scheme for cues to shame and anger. 190
This book has benefitted by comments on earlier drafts by Walter Buckely, Aaron Cicourel, Keith Oatley, and Suzanne Retzinger. Keith Oatley's comments were particularly helpful in organizing and ordering the many strands. It was published last month by Cambridge U. Press.
This book outlines and gives examples of a new approach to research in the human sciences. It puts into practice the recommendation of C. Wright Mills, for what he called the exercise of the sociological imagination. But I would call it instead the interdisciplinary, human imagination. Here I develop and elaborate ideas that were proposed in an early form in my Microsociology (1990) and in Suzanne Retzinger's Violent Emotions (1991). These two books focused on a subtantive topic: emotions and social bonds in their interrelationship. This book continues with that focus, but codifies the methodological dimension.
My goal is to describe an approach to all human research that allows the interpenetration of theory, method, and data in such a way that each equally casts light on the other, generating a theory that is based directly on observations of actual human behavior, both inner experience and outer conduct. This introduction and the first two chapters emphasize methodology, of relating the smallest parts to the largest wholes. The later chapters apply this approach to verbatim human expressions.
When part/whole methods are applied to verbatim texts, the intricate filigree of even the simplist human transactions are revealed. Inevitably, an important component of this filigree are emotions and bond-oriented behavior. One important goal of the substantive chapters is to show that understanding the intricacy of hunan expressions is not a luxury, but an elementary requirement of human science. It is clear that societies (and the human relationships which constitute them) ride upon extraordinarily complex processes. Because emotional transactions are a vital part of human exisitence, and are usually omitted, the substantive chapters emphasize them, and their relation to behavior which is oriented toward maintaining the social bond. The part/whole method provides a path toward relating the microscopic filigree of human relationships to understanding the largest social structures.
At the heart of my method is what I call part/whole morphology. I borrow the idea of morphology from botany, where it has long formed the foundation of that discipline. In botanical research, the study of single cases (single specimens of plants ) is just as important as making comparisons between plants. Morphology is based equally on single cases and comparative study. By microscopic study of the smallest details of a single plant, the botanist learns how it works as a system, even if no other specimens are available. To the extent that other specimens are available, microanalysis and comparative study can be interwoven, each building on the other. Botany is the study of both individual and groups of plants.
Although not discussed as such, botanical morphology employs what I callpart/whole analysis, since it is concerned with relating the least parts, the tiniest details of an individual plant, with the greatest wholes, the structure and process of plant communities, and their relationship to their environment. The phrase "least parts and greatest wholes" is due to the philosopher Spinoza, who proposed that human understanding requires relating the least parts to the greatest wholes. Spinoza's proposition forms the basis of this book. As Spinoza suggested and as recent research demonstrates, human beings and human relationships are so complex as to require part/whole analysis, even in cases of simple, everyday interaction.
One area in which the many-layered complexity of human activities has become quite clear is ordinary language. The failure of automated computer translation of foreign languages, and of paraphrasing meaning within a language, has strong implications for the human sciences. Typically a computer program will offer fifty to a hundred paraphrases of the meaning of a sentence, none of which are correct, and many of which are ludicrous. For example, a computer program failed to provide a correct paraphrase of the aphorism "Time flies like an arrow." One of the many paraphrases produced was "Time flies (insects) as you would an arrow." Understandably, the program mistook a noun (time) for a verb, and a verb (flies) for a noun.
From a logical point of view, an ordinary language is a mess, since its main terms, the most frequently used words, always have a multitude of meanings. How do human beings ever interpret the meaning of a text or an utterance correctly (i.e. consensually)? The implication is that humans have within them computing equipment infinitely more sophisticated than the most sophisticated computers. To correctly understand ordinary language, humans must have access to part/whole algorithms that allow them to understand the particular meanings of words (and when face to face, of gestures) in context: that is, the meaning of an expression produced by a particular person in a particular dialogue, in a particular relationship, in a particular culture, at a particular time in history. All understanding requires a high order of what George Steiner called "interpretive decipherment." Like the Rosetta Stone, ordinary language is always a problem of creative cryptography, needing vastly more intelligence than the solution of intricate puzzles like the Rubik Cube.
The extraordinary intelligence that humans show in deciphering language and other cultural puzzles occurs with lightning-like rapidity and effortlessness, and often, but not always, accuracy. Even a five-year old can do it, but makes more errors than adults. (When she was five, my daughter dragged her feet about visiting my friend Dennis because she though he would examine her teeth. ). Adults also make errors, but much of the time their interpretations are consensual, or else society would be impossible. Loan companies survive when most of the understandings they enter into with their customers are mutually understood.
The rapidity and effortlessness with which people sometimes understand each other poses a profound problem for the human sciences, because it comes to be taken for granted, not only by the participants, but also by researchers. The design of most studies of human beings assumes that the words and sentences used by the researchers and their subjects are largely unambiguous, and also assumes that their subjects' intelligence is not extraordinary. This assumption is particularly marked in the construction of scales, but is also central to all experiments and most interviews. Even qualitative studies make this assumption, if somewhat more cautiously.
The problem is that humans are capable of not understanding or misunderstanding standardized research situations, or of using them to their own ends, concealment, getting the researcher's sympathy, etc. Similarly, it is all to easy for the researcher to misunderstand or not understand their subjects' responses. Understanding the meaning of human expressions is a complex and intricate process, but it can be understood if part/whole methods are applied. The nearer we take as our data VERBATIM records of human expressions, the closer we can come to understanding our subjects. Verbatim records include transcriptions based on mechanical recording of interaction and all written materials.
It is now taken for granted that the "two cultures" of science and humane letters are so separated that there is no way of connecting them. This assumption pervades both cultures. Here is an example from psychology (Maher 1991, p. 72):
[One approach to psychology] is to assert that individual behavior cannot be predicted, but only "understood" after it occurs. This solution puts [psychology] firmly into the arean of hermeneutics, i.e. the humane study of texts... Close examination shows us clearly that this approach is indistinguishable from that of the biographer writing as a contributor to nonfiction literature...(emphasis added)
The purpose of this book is to show that it is possible to integrate hermaneutics with prediction, that they need not be mutually exclusive. Part/whole morphology, as outlined here, combines the interpretation of texts with the use of explicit theory and method. Contrary to what Maher said, this method is quite distinguishable from the current beliefs and practices in both the scientific and humanistic camps.
There is a powerful intimation that science and humanistic interpretation can be combined in an extraordinary work of the literary imagination by Nuttall (1983). He distinguishes between two types of interpretation of texts. The first he calls "opaque," which involves the separation of the interpretor from the characters represented in the text:
In the opeing of King Lear folk-tale elements proper to narrative are infiltrated by a finer-grained dramatic mode (p 80).
In the second mode, which Nuttall calls "transparent,", the interpretor projects life into the characters:
Cordelia cannot bear to have her love for her father made the subject of a partly mercenary game (p. 80).
Nuttall has two complaints about the formalized interpetations in the "opaque mode" which parallel my discussion here. First, no matter how formalized, opaque language must smuggle in transparent interpetations, otherwise it would be meaningless (p. 84-87). Secondly, opaque interpetation is necessarily narrow and partial, since it rules out the other mode. On the other hand, transparent langauge does not rule out formal analysis, but can easily include both. This latter point that Nuttall made is quite parallel to my assertion that science and hermeutics can be combinded.
The intensive study of single cases, when accompanied by comparative study of cases, enables the researcher to understand human behavior in all of its complexity. When Durkheim, the founder of modern sociology, borrowed what he thought was the morphological method from botany, he left out exactly half of it, the systematic study of single cases. Seeking to demonstrate that social processes exert what he thought was an autonomous influence on human actions, he focused entirely on the comparative part of morphology, shearing away a vital part of an organic approach to knowledge and discovery. The division between sociology and psychology that Durkheim proposed in his early work is leading to impasse. This division, which was necessary in order to found sociology, has frozen into rigid separation, with tragic consequences for both disciplines.
One of the signal weakness of Durkheim's comparative approach, which continues to haunt modern social science, is the lack of temporal data showing time order of events, and therefore of causation. In his powerful study comparing suicide rates in different societies and in different social strata, Durkheim was forced to speculate on the causation of suicide, since he had no data which showed the temporal development of behavior. This is the key weakness of all "structural" analyses: the absence of process data.
The combination of single case and comparative study in botanical morphology enabled researchers to understand both structure and process, by observing both the single plant as a system, and also the system of many plants as a functioning community. The most important aspect of this approach is more subtle, however: one understands the single plant in the context of knowing a great deal about the plant community, and the plant community in the context of knowing a great deal about the single specimen. Morphology offers a methodological solution to a the most intractable problem in the human sciences, the relationship between the individual and the group. In this book I use discourse analysis to explore single cases of social interaction, and show how it may be used in conjunction with social system analysis to understand these cases in the context of other similar cases.
I use the term discourse is an extremely broad way, to include not only spoken language, the usual sense of the word, but also any record of communicative expressions. I resort to this unusual definition to avoid ponderous terms like communication or social interaction The reader should keep in mind the broad way that I use discourse, which includes written texts of any kind. My emphasis on verbatim discourse proposes that we allow the actual voices of our subjects to be heard in our studies, voices which have almost disappeared from the human sciences.
In my approach to morphology, I suggest that the basic molecule of social behavior is what I call the exchange: one action (usually discourse) by one person, and the response of another person. In much of social interaction, such exchanges are usually quite brief, as little as one transcripted line for each participant. Exchanges are small systems made up of least parts and larger wholes, at various levels.
We can think of each exchange as involving a part/whole ladder of levels. An exchange between a particular mother and daughter, for example, is made up of still smaller parts: the words and gestures (Level 1) of each of the component expressions. And each exchange (Level 2) is also itself a part of still larger wholes: the conversation (Level 3) of which the exchange is but a part; all conversations between the two participants (their whole relationship, Level 4), all relationships of their social type (the mother-daughter relationship in that particular society, level 5) etc.
By taking into account the parts and wholes of specific episodes of discourse, and the relation of those episodes to the larger social and cultural wholes of which they are a part, many of the most recalcitrant problems that face the human sciences may be confronted directly. My approach combines elements that are usually pursued separately, attempting to present human experience as a whole rather than as separated parts (such as data) and separated wholes (such as theory).
Basing this approach on the exchange does not eliminate levels of subjective experience which underlie outer behavior. In order to understand the meaning of an exchange, a researcher must make inferences about the motives, intentions, and feelings of the participants, as they themselves do, because they too are parts of the whole, even though only inferential. In this respect the researcher has a great advantage over the participants; he or she has the time and inclination to subject the exchange to microscopic examination. Because of many years of such minute examination of verbatim excerpts of dialogue, it seems to me that most human interaction is so complex that its participants understand only a tiny fraction of their own motives, intentions and feelings. In an earlier publication (Scheff 1990, 100) I referred to both inner and outer levels of dialogue as the "message stack":
3. Implicature (Unstated implications of words and gestures).
All interpretations of meaning require analysis of these four levels, but usually leave out explicit references to the lower three levels, depending almost entirely on the verbal components. Even studies which include nonverbal gestures seldom explicitly refer to the lower two levels. Part/whole morphology of discourse integrates innner experience with outer conduct.
The new approach subsumes and clarifies knotty theoretical problems. Many critics have complained that Parsonian theory overemphasizes social control as a determinant of behavior, the hypothetical grid of norms and sanctions which actors take into account in constructing their behavior. In the abbreviated part/whole ladder implied above, Level 5 implicates such a grid, but only as one part of complex structure in which the exchange is embedded. Part/whole analysis assimilates most of the current theoretical proposals in the human science, theories like social control and rational choice, but locates them within a much larger matrix. Of course actors sometimes deliberately or even rationally compare their options in coming to a decision, as rational choice theory would have it. But sometimes they don't. Both alternatives can easily be included in part/whole analysis.
In terms of methodology, the approach outlined here is addressed first to what Denzin and Lincoln (1994) call the "dual crisis" in qualitative methods, a crisis of representation and a crisis of legitimation. The crisis in representation involves the mass of thorny issues that have arisen in the last twenty years about how to portray the Other in our research, another person, race, gender, or class than one's own. This issue has surfaced with intense criticism of conventional descriptions of the Other in anthropological writing, and also in postmodernist trends. The issue is closely related, if not identical, to a classic tradition in philosophy, the problem of Other Minds. How can we be sure that we understand the thoughts and feelings of a person other than ourselves? The first crisis in qualitative methods concerns the formats we use to depict the minds and behavior of our subjects.
Denzin and Lincoln specify the problem of representation in terms of production and ordering of text and context (578):
... (ethnographers) must take steps to ensure that the words they put in subjects' mouths were in fact spoken by those subjects...But more important, the ethnographer must take care when changing contexts and reordering events...
The crisis of legitimation, as Denzin and Lincoln present it, overlaps with the crisis of representation, in that both involve the truth of our representations of the other, but the crisis of representation goes further, by referring to (578):
... the claim any text makes to being accurate, true, and complete. Is a text, that is, faithful to the context and the individuals it is supposed to represent?
Although Denzin and Lincoln do not use these terms, probably because they equate them with positivism, the crisis of representation appears to be closely related to the issue of validity, and the crisis of legitimation to the issue of reliability. What is the most valid method of representing our subjects, and how can we demonstrate that the results of this method are reliable? The present volume outlines an approach to this exact problem, reporting and relating text and context in a way that offers a measure of both validity and reliability.
The crisis of representation can be confronted by reporting verbatim the exact dialogue of the participants in a specific encounter. Where possible the dialogue should be a transcript based on mechanical recording on audiotape or film. In this way, one can be sure that the actual words and gestures are being represented. This method is also applicable to written texts, such as the telegrams (Chapter 5) between the heads of state immediately prior to the onset of the First World War. Written texts omit nonverbal components, but if sufficient attention is paid to both text and context, one can understand both surface and subsurface meanings in dialogue. Since the researcher is presenting the raw data and the method of analysis, the reader is empowered to confirm or criticize the interpretation, giving rise to a measure of reliability.
This method can also be used with dialogue as remembered by a participant or researcher. With this type of data, of course, the warrant of validity and reliability is less certain, since discourse that is filtered through the memory and perceptions of the reporter is subject to many kinds of distortion. Nevertheless, since this approach injects the remembered voices of named persons into an investigation, it offers an approximation of the least parts of a relationship. I have found this method to be of great help in teaching; many of my courses on social relationships begin with role-playing of dialogue as remembered by the students.
The analysis of a specific exchange (for example, the quarrels with parents, in Chapter 8) usually catches the student's attention, since it suggests new features of which the student was unaware. It would not be an exaggeration to say that many students are astounded when they discover their own contributions to the problems they have with parents and lovers. In the context of similar exchanges of other students in quarrels with their parents, each single case takes on a heightened interest. By using both a single case and a comparative approach, the morphological method, each student can better understand her own quarrel in the context of the other students' quarrels, and vice versa (for example, those students who found new awareness through analysis of their own quarrel usually showed more sympathy with other students' parental quarrels).
Remembered dialogue can be useful in teaching and research, but the strongest warrant for validity and reliability can be obtained through the use of verbatim reproductions of social interaction or written materials. Verbatim records catch more of the least parts than field notes or verbal transcriptions of interviews, especially the gestural parts. Giving the reader access to the raw data, the exact voices of the participants, and the theory and method being used by researcher, the reader is empowered to test the validity and reliability of the author's interpretations of the data. Just as important, the social reality under study comes vividly alive for the reader. This method can be applied in such a way as to combine the advantages of the three most important current approaches in social science: eye-witness qualitative methods, quantitative methods, and theoretical analysis.
The introduction of the subjects' own voices as data to be systematically analyzed, directly in the final report, may overcome the relentless march of standardization, what Weber called routineization, into social science procedures. This trend has been noted in feminist scholarship (see particularly Krieger's  forceful comments). Part/whole morphology incorporates the least parts, the clues to personal identity, into the ever larger wholes of sociological analysis.
Although my approach is a way of upgrading qualitative research by making it more objective and more systematic, in response to the crises of representation and legitimacy, it is also relevant to the practice of quantitative research. The success of a quantitative study does not depend entirely on technical questions such as research design, data gathering and analysis. It also depends on how accurate and how important the hypothesis that is being tested. Quantitative research is often of little importance because the hypotheses that are tested are too simple to catch human reality. They lack grounding in actual human behavior, what the French call gout des terres, the taste of the earth, the intricateness, ambiguity, and complexity of human experience. In these instances the skill and talent in conducting surveys or experiments is wasted, the hypotheses are too far afield to be worth testing.
The approach outlined here is one that attempts to generate increasingly accurate and general hypotheses by close examination of the actual reality of social life. By grounding investigation in examination of the "minute particulars" as Blake said, the least parts of single cases, and later in the comparison of these cases with one another in the context of larger wholes, one may generate hypotheses that are general and important. As pointed out, quantitative analysis leads to verification or disconfirmation of a hypothesis. But verification is the third step in part/whole morphology. Before taking the last step, it is usually necessary to tale at least one of the earlier steps: exploration (conventional eyewitness field work using qualitative methods), microanalysis of single specimens and comparison of specimens. (See Figure 1, Chapter 1.) Figure 1 can be used as a guide for beginning or expanding research in the human sciences.
In Elias's (1978) magisterial study of the civilizing process, he first analyzed excerpts from advice manuals from the same historical era in each of four European languages. These excerpts represent specimens from the 13th through the 18th century. He also examined excerpts in the same four languages from the 19th century, showing a decided change in emotional content. His method involves both single cases and comparisons of cases in different languages and historical eras. For these reasons, his results are specific, general and important. But like literary analysts, his theory and method are not made explicit. Probably for that reason, his work has failed to have the impact it should have had.
In the approach advocated here, since verbatim texts are used, the researcher (and the reader) has the advantage of direct eye-witness observation of the behavior under consideration, as in the best qualitative methods. The researcher has access to features of the text which are often ignored by the participants, and to instant replay, which is also seldom available to the participants. If as suggested in the chapters that follow, the researcher provides the reader with a comprehensive description of the methods employed, the study, like quantitative methods, offers the reader exact definitions of concepts and procedures. Finally, as in the studies to be described below, if the study is either built on or generates general theoretical propositions, then it will have the advantage of being embedded in an abstract theoretical framework, which is the strength of current social science work in theory. Drawing on the strengths of these three areas, the studies presented in this book are therefore reminiscent of current research in the human sciences, because they combine some of the strongest elements in what is currently being done.
But in another way, all of the studies presented here will seem quite different than current studies, because each of them carries out the part/whole theme. As required by the part/whole paradigm, each study is multi-level and multidisciplinary. The studies involve microscopic examination of discourse, and understanding the results of microanalysis in the context of larger wholes, social institutions and cultural systems. The analysis of various levels also involves concepts and propositions from many disciplines, with emphasis on connecting social science propositions with concepts and propositions from history, political science, anthropology, linguistics, psychology, psychiatry and psychoanalysis. The method combines, therefore, what is often separated - for example, microlinguistic analysis of discourse, on the one hand, with social system analysis, on the other.
One feature of the studies presented here which may seem particularly strange and unusual is the pervasive focus on emotional elements in social interaction. Certain emotions, such as shame, pride, and anger, will be identified in virtually all of the episodes that are discussed. This focus does not mean that I think that all behavior is determined by emotions. Rather it means that that I believe that emotions, like perception and cognition, are present in most behavior. Emotions stand out a prominent features here because they are usually left out of studies in the human sciences. Even the most qualitative of studies almost always analyze the verbal part of discourse, ignoring the rich nonverbal accompaniment, like a pianist playing only the left hand. Social science studies are seldom scored for nonverbal behavior. The part/whole method includes not only many levels and disciplinary perspectives, but also all of the components of behavior, including emotions.
It has become a tradition in the human sciences to focus on cognition, (as in cognitive science), behavior, (as in behavioral analysis), and/or beliefs, (as in most survey studies). If emotions are mentioned at all, they are given short shrift, treated as irrational elements briefly and casually. But because I analyze emotions systematically in this book doesn't mean than I think they are the whole story, only that they should be taken as seriously as the other components of human behavior. For example, Chapter 5 suggests that unacknowledged shame, particularly the French reaction to their defeat by the Germans in 1871, was an important factor in the origins of World War I. But in suggesting that a chain of humiliated fury ran through French politics from 1871 to 1914, I am not proposing that that it was the only cause of the war. The actions of the other major nations were just as important, as were other motives, such as desire for power and expansion, and perhaps fear of the other countries. I emphasize one motive, humiliated fury, because earlier explanations have left it out.
The first chapter is very dense, covering a broad spectrum of methodological issues necessary for understanding human beings. It makes the point that human understanding is far richer and more complex than had been previously thought, showing the actual processes of effective thought. The remaining chapters are more accessible, since they involve applications of the main ideas in the first chapter to concrete episodes of human behavior. The idea of parts and wholes is very powerful, but it is also quite abstract. It may be the most general framework possible. It needs to be shown how it applies to real human activity in diverse settings, fleshing it out with actual speech and gestures, as interpreted in context.
At the center of the first chapter is Spinoza's insight that human understanding requires knowledge of the "least parts and the greatest wholes" and the relationship between them. This idea leads to the very heart of human reflexiveness: when we are thinking clearly, we carefully relate ourselves to both the microcosm and the macrocosm. This movement clarifies what it is that we are doing, it makes contact with efforts of others, and makes our thoughts maximally useful. It is at the same time a declaration of both dependence and independence. To use Elias's favorite word, it is a declaration of interdependence, of unity and difference.
This idea suggests a way of correcting for the overspecialization of knowledge in the modern world. There is a joke among academics that they either know everything about nothing or nothing about everything. As in most jokes, there is some truth to this one. There are the parts people, who know everything about next to nothing, and the wholes people, who know next to nothing about everything. By struggling to relate parts to larger wholes (the empiricists in psychology and history, for example) and wholes to the parts of which they are to be composed (theorists in the social sciences and the humanities), perhaps some of the alienation and waste motion in our attempts to understand the human world can be overcome.
This book shows the need to include and study all of the parts of human communication: gestures and emotions are just as important as words, thoughts and actions. All of these components are equally necessary in understanding the character of the participants and the nature of their relationships. It is mainly for this reason that many studies of human beings seem thin and airless, since the various approaches typically omit one or more components. Many studies in the social sciences are based on interviews which focus solely on words. Studies in the behavior sciences usually focus only on thoughts and actions or actions alone (as in the psychology of facial expressions). Social scientists usually study minds without bodies, psychologists bodies without minds. Because of these lapses, it is rarely possible to come to valid conclusions because some of the parts of the system are missing. Although recently emotions are beginning to receive notice, they have been ignored for so long that it will take some time to catch up.
Even studies that include emotions usually leave out one particular emotion, shame. I argue, as implied in Goffman's (1967) brilliant essay on embarrassment, that shame is the master emotion, in that it is an actual part of, or more frequently, is anticipated in virtually all human contact. Several of the chapters focus on this emotion in order to show how shame, ordinarily considered at best as only a vanishingly small part of the human drama, may indeed play a major role in most human activity.
Shame is crucial in social interaction because it ties together the individual and social aspects of human activity as part and whole. As an emotion within individuals it plays a central role in consciousness of feeling and morality. But it also functions as signal of distance between persons, allowing us to regulate how close or far we are from others. Observable cues to shame and embarrassment make social relationships visible to participants, and are available in verbatim transcripts for the researcher, materializing an otherwise intangible but vitally important part of human affairs. This idea runs through many of the chapters; it is most elaborated in Chapter 6-8. Chapters 4, 5, 6, and 8 deal both with single case analysis and comparison of cases, more advanced part/whole morphology than the single case, but still at a low level. Advanced morphological methods requires systematic comparison of many cases, as well as single case analysis.
Needless to say, the approach outlined here will not solve all of the manifold problems of the human sciences, or come anywhere close. The basic idea, of part/whole relations as the fundamental building block of the human sciences, is still much too abstract, even given the illustrations of its use in the chapters below, to be yet easily applied to the task of rebuilding research on human beings. It will need to be further fleshed out with a much wider variety of problems and settings than I undertake in this book. Because these examples are only preliminary, I have given far more emphasis to single case analysis than to comparative analysis. Mature applications of my approach must give balanced attention to both single case and comparative analysis. The examples of applications of part/whole morphology below should be taken for what they are, preliminary and incomplete exercises, used to illustrate a new path for the future.
It will also be evident to the reader that I largely proceed from the
work of others. I show how the part/whole paradigm can expand our understanding
of work that is already been recognized as significant. Finally, my emphasis
on the role of emotion in the chapters below is a further instance of the
limitation of range. A more balanced application of the ideas outlined
here would give equal attention to perception, cognition, behavior and
Section 1: Theory and Method
There is a marvelous passage in one of Kundera's (1995) essays on the history of the novel which exactly evokes the problem I attack here, how to access human reality (Kundera 1995, 128-129, emphasis added):
Try to reconstruct a dialogue from your own life, the dialogue of a quarrel or a dialogue of love. The most precious, the most important situations are utterly gone. Their abstract sense remains (I took this point of view, he took that one. I was aggressive, he was defensive), perhaps a detail or two. but the acousticovisual concreteness of the situation in all its continuity is lost.
And not only is it lost but we do not even wonder at this loss. We are resigned to losing the concreteness of the present. We immediately transform the present moment into its abstraction. We need only recount an episode we experienced a few hours ago: the dialogue contracts to a brief summary, the setting to a few general features. This applies to even the strongest memories which affect the mind deeply like a trauma: we are so dazzled by their potency that we don't realize how schematic and meager their content is.
When we study, discuss, analyze a reality, we analyze it as it appears in our mind, in our rnemory. We know reality only in the past tense. We do not know it as it is in the present in the moment when it's happening, when it is. The present moment is unlike the memory of it. Remembering is not the negative of forgetting. Remembering is a form of forgetting.
We can assiduously keep a diary and note every event. Rereading the entries one day we will see that they cannot evoke a single concrete image. And still worse: that the imagination is unable to help our memory along and reconstruct what has been forgotten. The present-the concreteness of the present-as a phenomenon to consider, as a structure, is for us an unknown planet: so we can neither hold on to it in our memory nor reconstruct it through imagination. We die without knowing what we have lived.
How can a scientist or scholar capture the reality of human life, when the people whom we study usually cannot do it themselves? As Kundera suggests, only the greatest of novelists, giants such as Tolstoy and Proust, have even come close, by reporting the evocative details of sight, sound, and context that we usually ignore or immediately forget.
Kundera's comments clarify and extend the Proustian quest, not only for the lost past, but for the lost present. The title of Proust's masterpiece, if translated literally into English, is not The Remembrance of Things Past, but The Search for Past Time. And the title of the triumphant last chapter is The Past Regained. A great artist, he demonstrates, can find the universal moments of childhood living within his or her memory, but hidden behind the conventional facade of everyday life.
Although most of Proust's commentary concerns the recovery of the distant past, a few passages concern a past so immediate that it edges upon the present. For example, in the section called Within a Budding Grove, there is an incident in which the narrator, Marcel, finally gets to meet Albertine, the girl he has been yearning for (and who later becomes the love of his life). At first he is deeply disappointed with the meeting; the whole episode seems banal and empty; he and she both conventional and distant. But that evening, as he reconsiders the meeting, he begins to remember the fine details of her gestures, facial expression, and inflections. She comes to life for him, in his "darkroom," as he says, where he is able to develop the "negatives" of his impressions of her earlier in the day. By focusing on the details, he is able to regain a past so immediate that it points toward the possibility of recovering the present.
Proust is still ridiculed for his seeming preoccupation with minutiae. A favorite jest is that it takes him fifteen pages to describe turning over in bed. But Proust implies that the ability to recover even fleeting moments of the past and present are the sine qua non of the great artist: it is these recovered moments that breath life into art. The main difference between art and kitsch is the abstractness of the latter. It describes not the details that make up human experience, but conventional abstractions of them, as Kundera suggests.
But why do we need the living present in the human sciences? I propose that it is needed to breath life into our enterprise also. I suggest a method, part/whole analysis, for restoring human reality to the social sciences. This approach is a way of filling in the details of Proust's method of "developing our negatives in our darkroom." Using transcripts or verbatim texts as data, one interprets the meaning of the smallest parts (words and gestures) of expressions within the context of the ever greater wholes within which they occur: sentences, paragraphs, the whole dialogue, the whole relationship, the whole culture and social structure. A central theme in the work of Spinoza was the thought that human understanding requires relating the least parts to the greatest wholes. This book shows how this idea may be carried out in disciplined program of inquiry.
The first chapter describes how the morphological method can be applied in the human sciences. It then develops the idea of part/whole analysis. In order to show how important this approach is for the human sciences, the first chapter takes up the complexity of human intelligence. It slows down part/whole processes that occur in our thinking and feeling that are so fast as to be practically invisible. It shows how these processes lie at the heart of capable, despecialized problem-solving, of what is usually called common sense. It provides examples of two thinkers who made progress toward part/whole thinking, Elias and Levi-Strauss, and the consequences for those who made little or no progress in this vital direction.
The second chapter examines the strengths and limitations of literary analysis in two studies. In the first, a study of poetic closure (B. H. Smith, 1968), the author takes several steps toward locating her work on a part/whole ladder, how the issue of poetic closure and non-closure occurs in Shakespeare's sonnets, and in other poetry, literature, and art. The author takes one more step up the part/whole ladder, suggesting, in passing, that closure in poetry is one aspect of closure in language as a whole. However, she does not elaborate this idea, which could have related her study to those in linguistics which deal with the same issue: openings and closings in conversation. Such an elaboration could have enriched her study and those in linguistics, suggesting continuities and differences between poetry and ordinary language, making the study reflexive.
A more powerful literary study concerns the six heroines of Jane Austen's novels (J. Hardy 1984). The author shows that the romantic love relationship that can be inferred in all six cases involve what he calls the sharing of experience between the heroine and her prospective husband. This idea is demonstrated by close analysis of dialogue, which utilizes, by implication, what I am calling part/whole methods. His discussion of the romantic love relationship is much more convincing than anything to be found in the psychological and psychoanalytic literatures, because it is data-driven. Potentially this method could lead to consensus as to the validity of Hardy's analysis, since he presents both concepts and data for the reader's inspection. However, since he offers no exact definition of what he means by sharing, and no method for classifying hits and misses in his analysis of the data, independent agreement among readers of would be unlikely.
Hardy's idea of sharing as an indication of true love comes very close to my concept of attunement as the mark of a secure bond, which I develop in the next three chapters. These chapters generate and elaborate a theory of the social bond. This theory is based in part on earlier theory and findings, but it is also generated and vivified because is it is driven, like Elias's and Hardy's studies, by close attention to verbatim texts. By combining the skills and sensitivities of social scientists and humanists, perhaps some progress can be made toward understanding the human condition.
In this section , I show how social structure lives in the smallest
parts of discourse, when interpreted within the local and extended context.
The manner of expression, particularly, carries clues to emotions and the
social bond. These ideas will be applied to concrete situations in later
sections of this book: The social status of Goodwin's boys and girls in
Chapter 4 is continually signaled and responded to in their discourse.
When one black child derides another for having thick lips, he doesn't
realize he is re-affirming the social structure of the boy's group and
of the larger society at the same time. In chapter 7, by attending to the
smallest parts of discourse, I show how age, gender, occupational and social
class invade a psychotherapy session, all but overwhelming it. Chapter
5, on the origins of the First World War, shows how smallest parts of telegrams
between heads of state signal the kind of alienation that leads to violence.
The social structures that rule our lives, all but invisible to the untutored
eye, are manifest in the smallest parts of discourse, when interpreted
within larger contexts.
Chapter 1 PART/WHOLE MORPHOLOGY:
UNIFYING SINGLE CASE AND COMPARATIVE METHODS
Here I describe the morphological method as a new stage of inquiry, between the first stage, qualitative methods, and the third quantitative methods. The proposed second stage involves microscopic examination of single specimens, and, if more than one specimen is available, the comparisons of specimens with each other. This method is particularly useful for the objective determination of meaning, a crucial problem for the human sciences. Because the determination of meaning is complex, yet taken for granted, I describe its intricacy. The new method also can be used to generate micro-macro theories, perhaps the next stage in the development of the human sciences.
Morphology of Human Conduct
To form a bridge between qualitative and quantitative methods, which are increasingly separated, part/whole morphology can lead to research which is valid, reliable, and cost-efficient. Qualitative methods involve exploration, the first step in inquiry. Quantitative methods involve verification, the last step. Although preliminary exploration is usually necessary and always helpful, exploration requires verification. The weakness of verification alone is that since experiments and other standardized formats (such as the scale and the standardized interview ) are narrow and rigid, one needs to have considerable knowledge before an adequate testing procedure can be designed. Qualitative methods are like wide-angled lenses with little depth; quantitative methods are as narrow as using the wrong end of a telescope.
Furthermore, since verification is costly and time-consuming, only hypotheses and theories should be tested which are not only plausible, but have been shown to be general and important. The procedure outlined here is more laborious than most qualitative studies, but it is also more cost-efficient than those which automatically seek verification.
The approach outlined here can be seen as the next step after what Giddens (1984) has called "instantiation." He asked for actual instances of the behavior described by any theory. His call, in turn, can be seen as a reiteration of Max Weber's (1947) insistence that the task of sociology is to reduce concepts about society to "understandable action, that is, without exception, to the actions of participating individual (persons)."
To summarize my approach: with or without initial exploration (Stage 1), the researcher would examine individual specimen cases (such as verbatim texts) microscopically, Stage #2. This step can lead directly to the development of a theory grounded in intimate knowledge of the specimen cases, but oriented toward placing them in the largest possible context, generating a micro-macro theory. Should such a theory appear promising, the final stage, verification (3), could follow.
The term morphology is meant in the sense that it was originally used in botany, which includes the intense microscopic investigation of single cases (in botany, a specimen plant). This is not the sense in which Durkheim (1903) used the term. In his usage, because of his bias toward comparative studies, he excluded the examination of the minutia of the single case. The same bias is found also in the otherwise admirable development of "grounded theory," by Glaser and Strauss (1967). In my usage, morphology includes intense study of both the single case and comparisons between cases.
The sequence in which these stages occur, except for the last, verification,
need not be chronological. Investigation of a problem might as easily start
with #2 or #1. This point should be kept in mind in reading this chapter,
because, for emphasis, I focus on the bottom-up strategy of starting with
#1 or 2. There is no reason that one could not start with a theory (topdown),
or better yet, juggling research and theory at the same time (See my comment
on abduction below).
Figure 1 here.
The new stage provide the potential for squaring the circle, achieving at least a measure of both validity and reliability. The format for reporting morphological findings requires that the researcher spell out the concepts and methods used explicitly, and make the texts that were analyzed available to the reader. The interested reader will then be able to apply the researcher's methods to the researcher's data. Provided with the researcher's concepts, methods and raw data, this procedure allows judgment of the relevance (validity) of the concepts and methods employed, and the repeatability of the findings (reliability).
This is not to say that this approach could rapidly solve the many problems of research in the human sciences. It could bring together the necessary elements for solutions, but it is still to early to tell whether it would be practical to do so. Much of the specialization that has run riot in the human sciences, specialization by discipline, sub-discipline, levels of analysis, method and theory, has occurred for reasons that are extraneous to science. But some amount of specialization is probably necessary. In the long run it will probably be necessary to work out some compromise between the kind of de-specialization I advocate here and what Oatley (1996) has called the "social distribution of cognition", the specialized thinking (and behavior) that requires division of labor.
However, for the almost continuous determination of the meaning of human expressions that forms a major part of all social science research, it seems impractical to distribute cognition. Part/whole morphology of social interaction or written texts is particularly suited for determining the meaning of human expressions. Because the new approach uses verbatim transcripts or texts, it allows for the patient interpretation of meaning that include the smallest details and the largest contexts. This method points toward the objective (consensual) determination of meaning.
The Problem of Meaning
Can the meaning of human expressions and behavior be determined? In the tradition of verstehen established by Weber, Dilthey, and others, meaning was the most important component of social science. To understand human behavior, we need to understand the subjective orientation of the actors. But the originators of this tradition and its followers did not consider the determination of meaning to be a technical problem, one that requires the close consideration of human actions and expressions in relationship to the context in which they occur. The technical problem of determining meaning is one that extends into the vital core of all activities of the human sciences, theory, method and empirical research.
Meaning as a technical problem is crucial in the current crisis in the human sciences because of the way in which humans in everyday life are able to exact accurate meanings from expressions that are both highly complex and ambiguous. Understanding ordinary language and other kinds of human behavior requires the consideration of the smallest parts of expressions and their relationship to the largest possible wholes (not only grammar and syntax, but biographical structures, as well as the structure of the entire language and culture.) Part/whole analysis of this kind would seem to be the key component of what is called "common sense." We humans have become so adept and quick at understanding expressions by relating their least parts with the largest wholes that we don't realize the extraordinary complexity of what we do. In this chapter I slow down and therefore make explicit the bare outlines of this process.
In my view, the human sciences are becoming ever further removed from reality because they are so specialized that they cannot use part/whole analysis in the way that their subjects do. Understanding ordinary language and other kinds of human behavior, because of its complexity and ambiguity, requires a global, and therefore a de-specialized point of view. Fragmentation into disciplines, sub-disciplines, levels of analysis, types of method, and schools of thought has deprived the human sciences of the ability to understand human behavior even as well as their subjects do, which is not very well. The approach outlined here would enable us to at least compete with our subjects in understanding their behavior. Although ordinary people do poorly with understanding emotions and social bonds, they are probably more cunning than human scientists in using part/whole analysis to determine meaning. The method outline here is oriented toward uncovering both overt and covert meanings. It is particularly suited toward discovering hidden emotions and bond-oriented behavior.
Hypothesis: The meaning of human expressions and behavior can be determined, but objective interpretation requires disciplined investigation of the complex three- way relationships between meaning, text and context, in the way that will be made explicit below.
Theory: Most current discussions assert that meaning is by and large a subjective matter. This position is by now so established that its adherents assume rather than investigate it. Postmodern thought, a recent development, assumes that meanings are not only subjective, but essentially undecidable. Although there is no actual evidence to support this conclusion, postmodernists postulate that poems, novels, and indeed all texts are inherently ambiguous.
The founder of the postmodern critique, Derrida, made a point which is both true and important. He and his fellow deconstructionists have demonstrated that if taken out of context, any text becomes ambiguous. In this light, James Thurber should be considered the first deconstructionist. In his essay on his late night thoughts about the name Perth Amboy, he tells the results of repeating the name many times. After many repetitions, not only did the name begin to lose its meaning, but the very room began to whirl around his bed. Thurber had stumbled into a way of decontexualing an expression. Words and other expressions in ordinary language are only indexical, they are ambiguous when context is removed. This idea suggests that if we are to understand how meaning can be determined, we must consider the relationship between meaning, text and context.
The only sustained consideration of this three-way relationship has been conducted by Cicourel. Compared to him, I am a late arrival in this field. Although our styles of thought and investigation are different, I recognize that he was the first to realize the crucial importance of this problem, and to devote most of his time and effort in an attempt to solve it. As in my approach, Cicourel has shown repeatedly that a text can only be understood in context, necessitating a detailed ethnography of context. This chapter extends his analysis, by outlining an explicit theory and method.
In this chapter I outline a theory and method to deal with the relationship between meaning, text, and context. Certainly the deconstructionists have not seriously considered this issue. In particular, they have not understood that their main point, that all texts are undecidable when removed from context, implies an equal and opposite corollary: in context, the meaning of a text is decidable. Postmodernists have jumped to an unwarranted conclusion that context or no, the meaning of all texts is undecidable. This chapter will argue for the importance and truth of the corollary: given sufficient investigation of a text and its context, it is possible to approach an objective determination of meaning.
To begin with humor, an everyday example of the issue of decideability. A joke can be lengthy and complex; "getting it" may require weaving together various relevant but conflicting threads in the narrative, ignoring extraneous details, and understanding the way the punch line resolves the conflict. Grown-up humor requires a high level of sophistication, one that is absent in young children.
When a joke brings genuine laughter, its meaning was not ambiguous to the audience; they got the point. But a joke may be funny to one audience and not another. "You have to have been there." Even if the joke was told exactly as before, some element(s) could have been different in the new context. The decidability of meaning changes with context.
Empirical research: The problem of meaning is seldom discussed by empirical researchers; they usually assume that the meaning of subjects' responses is unproblematic. Although the occasional qualitative researcher may be sensitive to this problem, few studies have explored the relationship between text and context in the determination of meaning. Both in direct observation and in sophisticated techniques of measurement, researchers take much the same position as their subjects; the construction of meaning is taken for granted: it is "common sense.".
Rather than explore the matter, empirical researchers have instead divided into two camps, both usually avoiding the problem of meaning. Qualitative studies use interpretation in context, quantitative studies, standardized procedures. The two approaches are exact opposites: qualitative work is unsystematic but contextual, quantitative work is systematic but acontextual.
Qualitative studies, oriented toward face validity, neglect reliability. Face validity is important; it means that an analysis of meaning can be related to ordinary language, a vast repository for understanding the complexity and subtlety of human expressions. But ordinary language is also a repository of bias, a bastion of the cultural status quo. Face validity alone, plausibility, can never be a sufficient basis for determining meaning.
Quantitative studies, oriented toward reliability, neglect validity. Reliability is also of great importance; it insures repeatability. But erroneous procedures can be repeated as easily as correct ones. (See my discussion of standardized scales below). Reliability alone can never be a sufficient basis for determining meaning.
Both approaches are rational, but only in part. Reliance on face validity exemplifies substantive rationality. This approach is sensitive to the particularities of situations, just as procedural rationality is attuned to their general features. Weber has warned that substantive rationality is marred by capriciousness, and procedural (formal) rationality leads to bureaucratic deadlock. A marriage seems to be needed, one which would unite substantive and procedural rationality (validity and reliability).
Several examples will suffice to illustrate the need. The capriciousness of direct observation has been demonstrated repeatedly in ethnographic studies. Oscar Lewis saw a different Tepoztlan than the one reported by Redfield, and Freeman (1983) has argued that Margaret Mead's description of adolescent sexual behavior in Samoa is entirely fictitious. Neither Lewis nor Redfield offer hard data. Freeman's critique of Mead's work is well documented, but by no means foolproof. For example, although his two visits to Samoa covered years compared to Mead's months, they took place long after Mead's visit.
Plausibility is both strength and weakness; it soothes and beguiles our judgment by "commonsense" reasoning, even with claims that are false or groundless (Commonsense will be discussed below). Direct observation is invaluable at the beginning of an investigation, but not sufficient for the objective determination of meaning.
The widespread use of scales to measure psychological and social attitudes provides an example of the need for both validity and reliability in the determination of meaning. Although my colleagues and I (Scheff, Retzinger, and Ryan, 1989) reviewed the literature only on self-esteem scales, our conclusions may also apply to all scales currently in use.
At the time we conducted the review (1988), we estimated (with the help of Morris Rosenberg) that there had been more than TEN THOUSAND studies using self-esteem scales. Yet despite this massive effort, no consistent findings had been reported. We found that the six comprehensive reviews of the self-esteem literature in print at the time of our article were in agreement on the absence of significant findings. The chaotic state of the field is suggested by the most recent of the reviews:
What has emerged ... in the self-esteem literature is a confusion of results that defies interpretation. Hypotheses have been tested about the relationships between self-esteem and hundreds of other psychological variables. Many of these variables have been supported, but most observed trends have been weak and insubstantial. There are few replications or systematic extensions, and it is difficult to know which findings are worth pursuing. Moreover, because different investigators begin with different assumptions, their findings stand in obscure relation to one another. (Jackson 1984, 2-3).
Jackson and three of the other reviews are basically critical of the whole field. On the other hand, two of the reviews are strongly positive and optimistic in their orientation. However, all six of the reviews are in agreement on the lack of significant findings. Indeed, the two positive reviews are the most devastating on this point. Now, some eight years and perhaps four or five thousand studies later, the situation does not seem to have changed: deadlock.
Jackson (1984, passim) takes up an additional issue, one not addressed directly in the other five reviews: lack of a theory or even an abstract concept of self-esteem may be the crucial reason for the failure of the field to develop. I will return to the point below, after discussing the stages of inquiry.
Text and Context
The state of the art in the detailed analysis of meaning can be found in Pittenger, Hockett, and Danehy (1967) and Labov and Fanshel (l977). Each study carefully analyzed every sound, both verbal and nonverbal, hearable in audiotaped excerpts from single psychotherapy sessions. Unlike most studies, they also reported the reasoning upon which they based their inferences about meaning ).
Although the authors of these two studies were not familiar with them, both consistently used two methods which have been described abstractly in the phenomenology of meaning. The first has been called the method of prospective-retrospective understanding (Schutz, 1962). This idea breaks down the extremely broad and abstract concept of context into two components, the past and the future context. One understands the meaning of an expression by placing it in context of what has happened before it occurs, and what happened afterwards. One can further subdivide into the local and extended context. That is, one understands a word or sentence in terms of the some of the text that comes before or after it (local context) or, ranging more broadly, events before and after the entire text, the extended context.
As might be expected, both Pittenger et al and Labov and Fanshel based their inferences of the meaning of expressions on references to the local context, passages occurring before and after the expression in question, often immediately before or after. Occasionally, however, they looked further. An example in the former study is the analysis of the meaning of the patient's third utterance. When the therapist asks her why she came, near the very beginning of the session, the patient lists her symptoms in a formal, organized manner. As it turns out in the rest of the session, her report exactly summarizes her main problems. Also she uses several contractions which suggest repeated use of this form. The study's analysis is based in part on knowledge that she is a nurse, which she mentions in her l6th utterance. Using both the prospective and retrospective method of understanding, Pittenger et al reason that she is enacting the role of a nurse reporting to a doctor.
Pittenger et al's interpretation of the patient's misunderstanding of the therapist's manner, one of the central findings of their study, is based in part on their knowledge that she has had only one prior session with a mental health professional, which she mentions only near the end of the session. (A substantial book can hardly be understood in one reading, especially if one begins at the beginning and then proceed to read the pages in the correct order. Readers who peek ahead in order to understand what they are reading are using the method being discussed here.)
Pittenger et al make no explicit use of historical, biographical or follow-up data: they make no references to factual events in the extended context. But Labov and Fanshel had such data; they made references to the extended context. For example, their understanding of the effects of the therapist's tactics is based in part on their knowledge of events before and after the patient's sessions with the therapist. Before psychotherapy began, the patient, who was anorexic, had starved herself to the point that her life was endangered. At follow-up, five years after the last of many sessions, the patient was reported to be symptom-free. Labov and Fanshel used this information to confirm their analysis of the meaning of the therapist's tactics in the excerpt they studied.
Hypotheticals (Counterfactuals): Even though Pittenger et al did not use historical or biographical knowledge not contained in the actual dialogue, their analysis of meaning is based not only on the dialogue itself, but also on events that they imagine had happened before it began. In the language of phenomenology, such events are called counterfactuals. Because the usage of this term varies somewhat in philosophy and linguistics, I will use the term hypotheticals instead.
Much of the analysis in Pittenger et al is based on imagined, hypothetical events. Already mentioned is their explanation of the form of the patient's report of her symptoms; the researchers imagined her making such a report in this form many times before the present instance in her capacity as a nurse reporting symptoms of a patient to the doctor.
Pittenger et al also make more extended use of hypotheticals in understanding the patient's response to the therapist. They imagine that the patient's response to a therapist who seems to her cold and detached is much like her response to her husband, who she refers to in a way that suggests that she also sees him as distant and unsympathetic. Although Pittenger et al never use the term, their analysis involves almost constant use of hypotheticals. They even name the method they use to understand one type of hypothetical (those that could have occurred instead of the actual utterance), the Principle of Reasonable Alternatives.
Similarly, Labov and Fanshel do not limit their search of the extended context to factual data. Much of their analysis of meaning is based on hypotheticals, such as patterns of discourse within the patient's family. On the basis of re-enactments by the patient of conversations between herself, her mother, and her aunt, Labov and Fanshel imagine patterns of highly conflictful discourse in the family. They largely understand the meaning of the patient's responses to the therapist in these terms.
This chapter suggests that the determination of meaning is neither mostly subjective, as assumed by theoretical approaches, nor objective, as assumed in most empirical research, but a varying mixture of subjective and objective. To the extent that researchers locate all relevant context, and to the extent that their hypotheticals are confirmed by factual data, their interpretation of meaning is objective.
Of course in actual practice discovery of the relevant context, and the confirmation of hypotheticals with factual data, is likely to be beyond the range of a single researcher. But it is conceivable that one or more later re-analyses could approach the objective determination of meaning, limited only by the amount of interest and resources. What kind of data could warrant such an expenditure of effort?
The parameters that determine the answer to this question would probably concern the importance of the problem, method or theory under investigation, the extent of research investment, and the promise shown by existing studies. Choosing an exemplary study from a series of important studies for a replication using part/whole morphology would be ideal. For example, the study of Expressed Emotion (EE) represents considerable investment by a wide variety of researchers in a very fundamental problem, the possible origins of mental illness in family processes.
George Brown (1972) and others have shown a promising connection between family criticism and emotional over-involvement with ex-mental patients and relapse. However, the studies are virtually atheoretical, the methods difficult and time-consuming, and the size of the relationship is only moderate. A part/whole morphological study of the audio-tapes generated in one of these studies might be of particular help in generating a theoretical framework. (For a first step in this direction, see Ryan's  microanalysis of the transcription of one of the cases from an earlier EE study [Hooley 1986]. Ryan shows that the discourse between an ex-patient and his wife exactly fits the shame-rage paradigm (See Section 3).
Another example of texts which might deserve subsequent reanalyses are analyzed in my study of the emotional causes of the First World War (See Chapter 5). Immediately prior to the beginning of the war, there were six telegrams exchanged between the heads of state of Russia and Germany, and the Foreign Minister of England. In addition, the Kaiser's comments that he wrote on three of the telegrams he received have been preserved. Although these texts have been the subject of many prior analyses, I found them each to be incomplete, since they all focused on content, without analyzing manner.
Because of the formal and somewhat oblique language of the telegrams, a deep search of the extended context was needed to determine their meaning. The texts being brief, I feel confident of my analysis of their structure. But the 25 pages that I devoted to the context, political, psychological, and social seems a mere beginning. Since the causes of this war are still a mystery and an enigma, the objective determination of the meaning of the these texts might be of great import.
Even though a single researcher is unlikely to have the time, resources or patience for an objective determination of meaning, the more resources put into microanalysis, searching the context and confirming hypotheticals, the more objective the determination of meaning. Once again I turn to Pittenger et al and Labov and Fanshel as examples. In the cases of both studies, their analysis takes up an entire volume, even though the excerpts they studied are brief (5 minutes for Pittenger et al, 15 for Labov and Fanshel). But their analysis is not complete, because of limitations in their methods and data.
In regard to methods, both studies determine only cognitive meanings. Though both frequently mention anger, embarrassment and other emotions, analysis is brief and casual. For that reason, their determination of emotional meanings is incomplete. In subsequent re-analyses, I (1990. Chapter 6) inferred emotional meanings omitted from the Pittenger et al study, and Scheff and Retzinger (1991, Chapter 6.) inferred emotional meanings omitted from the Labov and Fanshel study.
In regard to data, Pittenger et al were unable, as any subsequent analysts would be, to confirm the accuracy of their hypotheticals in the extended context, since they made no explicit use of historical, biographical or follow-up data. (I personally contacted the therapist in the recorded session used by Pittenger et al; he told me that he had no further information about the session or the patient). Labov and Fanshel had only a limited amount of factual data outside of the text they analyzed. For example, they had no actual verbatim dialogues from the patient's family that would confirm their hypotheticals regarding patterns of discourse there. (I also contacted both Labov and Fanshel, but neither responded).
Although Scheff, and Scheff and Retzinger removed a limitation in the methods of the original studies, they could not remove the limitations of data. Because of the lack of sufficient historical and biographical data in these four studies, their determinations of meaning are a mixture of objective and subjective elements.
An additional expenditure of effort in collecting relevant data in the extended context might have been justified. My re-analysis of the Pittenger et al data is one of the bases of a general theory of interminable conflict. The Scheff and Retzinger (1991, Chapter 7) re-analysis of the Labov and Fanshel data proposes a theory of the causation of anorexia. Although anorexia has been studied extensively, there is no successful theory of its causation. Under these conditions, further steps toward testing might have been warranted.
There is one further limitation of these two studies. Exemplary as they are, their purview is only psycholinguistic: they fail to discuss or investigate the institutional context in which their texts are embedded. For example, even a casual hearing of the recorded session will reveal that the patient in the Pittenger et al study had a strong Boston working class accent. (A LP record comes with Gill, Newman and Redlich 1954). The therapist, on the other hand, has no discernible accent, strongly suggesting a difference in social class between the two subjects. The researchers do not comment on this difference, let alone investigate its implications.
Gender differences are also observable. The therapist's tactic of responding only to the informational aspects of the patient's expressions, brusquely ignoring her emotional responses, is gender related, at least in part. Since the session is an initial one, the therapist no doubt had a rationale for this tactic. But his curtness in ignoring the strong emotions expressed, and abruptness in switching back to information issues seems connected with the state of gender relationships, and perhaps class and age, in the U.S. in the l960's, the era in which the session occurred.
Surprisingly absent from the excerpt is even the slightest attempt of the therapist to explain his tactics, even though they seem to confuse and irritate the patient. On the patient's part, no attempt at overt protest was made. At several points she withdraws or sulks, but she doesn't put into her words her feelings about the therapist's behavior toward her, which borders on being abusive. The male therapist, who is undoubtedly older and of a higher social class dominates the younger, lower class female patient. Although she withdrew several times, the patient mostly subordinates herself to the therapist. If sufficiently analyzed, this session might tell much about the role of gender in therapist-patient contacts in that time and place.
Similarly, in the Labov and Fanshel study, although her accent suggests that the patient is Jewish, the researchers did not investigate this issue. They comment on it, but only in passing. Neither study develops the sociological implications of clues to the embededness of their data in age, class, gender, or ethnic structures. This kind of embededness will be a key issue in my discussion of theory, below.
The Balance between Text and Context
Several implications relevant to the problem of meaning follow from my discussion. One is that no matter how exhaustive the analysis of a text, the determination of meaning will be incomplete, and therefore partly subjective, without referring to relevant historical and biographical knowledge. For example, the interpretation of the meaning of a poem or novel may require substantial biographical knowledge of the author and/or of the historical period in which she wrote. Although this proposition in contrary to most literary theory, it follows inevitably from my argument. Postmodern ideas distract attention from this issue.
One example of the need for supplemental factual data is provided by the study of several hundred recorded psychotherapy sessions by Helen Lewis (1971). Using a systematic procedure for coding emotions (Gottschalk and Gleser, 1969), she discovered a consistent pattern of hiding virtually all of the emotion of shame and much of the anger among patients, and that the therapists ignored most of these episodes. However, lacking the methods used by Pittenger et al and Labov and Fanshel, her analysis of cognitive meanings is incomplete.
Even Lewis's analysis of emotional meanings is incomplete, because she had no data outside of the transcripts to confirm her inferences. For example, since she couldn't tell from the text alone whether the patients were aware that they were hiding their shame and anger experiences, she used an ambiguous term for classifying these experiences: unacknowledged emotion. She was unable to distinguish, therefore, between conscious and unconscious emotion. Nor could she tell the extent that the therapists were aware of the patients' emotions, the emotions that both patients and therapists seemed to ignore. Lewis's classification of all of the shame episodes as either overt, undifferentiated or bypassed shame suggest that she assumed the patients' experiences of shame were below the level of consciousness. But without debriefing, she had no way of to confirm this supposition..
The amount of supplemental data needed for the determination of meaning, the balance between text and context, will vary. In the case of Lewis's study, she might have been able to clear up some of the ambiguity over the patient's and therapists' awareness by reviewing a sample of the requisite episodes in their transcripts, a relatively small addition to the original analysis. This expenditure of effort would have been justified, since the existence of vast amounts of unconscious shame has been a sticking point in the acceptance of Lewis's findings.
On the other hand, a comprehensive analysis of cognitive meanings, in the manner of Pittenger et al or Labov and Fanshel, would have made for a huge additional amount of analysis, even with only a sample of the cases. Such an expenditure of effort would have been difficult to justify. Unlike Lewis's findings on shame, her analysis of cognitive meanings did not suggest an important new direction for future research.
Parts and Wholes in Verbatim Texts
The issues concerning the balance between text and context can be considered in a more abstract way by invoking relations between parts and wholes. The parts are words and accompanying gestures (if available), wholes the biographical, linguistic, social, cultural and other structures in which the text is embedded. This very general mode of analysis has been adumbrated in earlier discussions in social science and philosophy.
The idea of part/whole analysis is implied in C. Wright Mills' (1959, p.7) definition of "the sociological imagination:"
...the capacity to shift from one perspective to another--from the political to the psychological; from examination of a single family to comparative assessment of the national budgets of the world; from the theological school to the military establishment; from considerations of an oil industry to studies of contemporary poetry. It is the capacity to range from the most impersonal and remote transformations to the most intimate features of the human self--and to see the relations between the two.
Mills begins ("from the political to the psychological") and ends his definition with the psychological ("the most intimate features of the human self"). The latter passage implies minute psychological aspects. In the remainder of the passage, Mills ranges over social institutions of the largest magnitude. Since in several places in this book he explicitly refers to relations between parts and wholes, Mills seems to have been implying that the human sciences should study the relations between the smallest parts and the largest wholes.
The idea of relations between parts and wholes is a familiar topic in philosophy. It is particularly crucial in the work of Spinoza. Although his treatment of this theme is not free of l7th century theology, the implication is clear: human understanding requires relating "the least parts to the greatest wholes" (Sacksteder, 1991).
The idea that understanding a problem requires knowledge of its least parts, on the one hand, and the greatest wholes, on the other, may be used as a foundation for inquiry. It points to a path which could lead, potentially, to the objective determination of meaning. In following this path, the least parts-greatest wholes idea also suggests a way of connecting meaning and social structure. As discussed below, the objective determination of meaning makes explicit the micro-macro connections which arise in understanding discourse.
Although the least parts-greatest wholes paradigm provides an explicit contrast to specialized approaches, it subsumes rather than rejecting them. Highly specialized knowledge is implied by understanding least parts (e.g., empirical data) and largest wholes (e.g., abstract theory). Part/whole analysis begins where specialized approaches leave off: it relates the findings that specialized work takes as ends in themselves.
Specialized and General Intelligence
The part/whole approach can serve as complement and corrective to specialization, the reigning pattern in our current quest for knowledge. Specialization brings benefits, but also limitations. The most puzzling aspect of this irresistible tide involves narrowing of vision, not only in behavior, but also in thought. The neurosurgeon needs special skills, but need not think only as a neurosurgeon. Narrowing of outlook because of disciplines, theories, methods and schools of thought is creating a crisis of knowledge in our time. Given their intellectual and emotional commitments, the great majority of researchers seem to be entrapped within their specialty. The part/whole approach described here suggests that we need de-specialization as much as we need specialization.
Pathologies seldom occur at the time of founding a new specialty. At this point a combination of system and intuition, procedural and substantive rationality, may be a necessary condition. A clear illustration of this is the birth of computer science, created largely by the mathematician John von Neumann. The computer is an exact embodiment of what Pascal called "the spirit of system." It is based on processes which are linear, reductive, and explicit. The language which drives a computer is so systematic that the slightest ambiguity, even a missing comma, cannot be tolerated. In this respect, program and ordinary language are exact opposites: the symbols in a program must be completely unambiguous, in ordinary language, the symbols are virtually all ambiguous.
Computers are monuments to procedural rationality. Yet von Neumann's own method of mathematical work, which triggered the birth of computers, was completely intuitive. His intimates have described his mode of operation. To begin, von Neumann would list, on separate pages, a large selection of unsolved mathematical problems. He would then turn the pages, giving each problem a glance, one by one. If he did not quickly see a plausible solution, he would go on to the next problem. When he came to the end of the set of problems, he would begin again, continuing until he either found a solution or quit.
Von Neumann seems never to have troubled grinding out analytical solutions. His method of thinking was entirely intuitive, the exact opposite of the machine he helped create. Von Neumann was an embodiment of what Pascal called "the spirit of finesse" (i.e., intuition). The subsequent development of computer science might be viewed as the routinization of Von Neumann's charisma.
It appears that specialties are effective as long as system and finesse are in balance, or near it. But as a specialty becomes institutionalized, the spirit of system (procedural rationality) increasingly prevails. Making this point exactly, the physicist Boltzman noted, with some bitterness (he was a genius unrecognized during his lifetime): when a new method yields "beautiful results," many become unconsciously wedded to it; they come "to believe that the development of science to the end of all time would consist in the systematic and unremitting application of it." Although this statement was written at the turn of the century, it exactly captures the trajectory of current science and scholarship. An imbalance has been created which seems to be leading rapidly to ineffectiveness, if we haven't already arrived. System alone, as Weber pointed out, creates deadlock.
Science and scholarship which overemphasizes system at the expense of finesse corresponds to the stage that Kuhn (1962) called "normal science." He pointed out that this stage is effective in the "mopping up operations" that are needed in the wake of a great discovery. For example, the Human Genome Project represents a vast investment in mopping up the discovery of the structure of DNA. Perhaps Kuhn was too tactful to point out, however, that normal science is completely ineffective in areas where there have been no great discoveries, as in the humanities and social sciences.
The great problem-solvers in science have usually been intuitive types like von Neumann, who corrected the over-emphasis on system with a great jolt of imagination. Kepler broke through the bizarrely complex mathematical systems of centers and epicenters that were impeding the science of astronomy. He was able to discover the orbit of Mars because he placed the sun rather than the earth at the center of the orbit, an intuitive leap.
Kepler's leap was based on a premise that was entirely irrational, both in its source and content: as a young man, he had literally dreamed a fantastic scheme of crystalline solids which were supposed to determine planetary orbits. (He thought that the orbits of planets were constructed by their enclosure in five perfect solids [a sphere, pentahedron, tetrahedron, etc). The orbit of the earth and the other planets were determined by polyhedrons of various shapes. The orbit of Mars was determined by a sphere. (The circularity of the orbit of Mars was an assumption began by Aristotle and continued by Kepler until he finally made his discovery.)
The main features of Kepler's scheme were outrageous, but contained, unconsciously, the last step Kepler needed to break the impasse. After struggling for decades to determine the orbit of Mars on the assumption that the earth was at its center, his realization that the sun was at the center of his structure of crystalline solids was the final step that allowed a solution. He was right, but for the wrong reason.
Being right for the wrong reason is an enormous advantage that intuition provides in solving problems. Another example of this advantage in Kepler's case involves the astonishing inaccuracy of his numerical calculations. He made a multitude of errors, some quite large. Yet he correctly plotted the elliptical orbit of Mars; his errors canceled each other out. His unconscious, one might say, was working overtime.
Similarly, Einstein intuited the solution to the failure of classical physics with a stroke of imagination, a direct intuition of the nature of the physical universe. Virtually illiterate in mathematics, he first proposed the special theory of relativity as a joke. When David Hilbert, the great French mathematician, was asked why Einstein, rather than others (like himself or Poincare') immeasurably better qualified, made the discovery, he responded: "Because he had learned nothing about all the philosophy and mathematics of time and space" (cited in Feuer, 1982, p. 62).
Both in the case of Kepler and Einstein, it is clear not only that system may not solve difficult problems, it can actually stand in the way. Yet it was Brahe's systematic data which allowed Kepler to make his discovery. Einstein called upon friendly mathematicians, those hostages to system, to put his theories in their final form. Although no such credit was given, his general theory of relativity required virtual collaboration with a mathematical colleague.
Perhaps the classic case of the marriage between system and finesse in problem-solving is the discovery of the structure of DNA by Crick and Watson. Like von Neumann, Watson represented pure intuition. He was a graduate student when he and Crick made their discovery, utterly ignorant and untrained in requisite sciences. The model of DNA they developed was an assembly of the discoveries of others, but an assembly touched by imaginative genius. The scientists on whom the discovery of the structure of DNA was based each understood their own findings and fields, but misunderstood or ignored the work of the others. They had tunnel vision ; they were trapped within their specialized outlook.
Like Einstein in his ignorance, Watson was a complete outsider. Being on the outside, he didn't suffer from the biases and limitations of vision of the insiders, and from specialized procedures that had become more hindrance than help. Watson was something of an outlaw: he abandoned normal procedures and channels in a way that is still shocking to established scientists (Watson, 1980). But Watson also needed a conventional insider, Crick, to help him put the partial discoveries together into a workable whole. Discovery of the structure of DNA, like Kepler's and Einstein's, required a partnership between system and intuition.
Intuition: its Structure and Function
Attempts to use computers to linguistic ends suggest that human intelligence is vast because it involves balancing system and intuition. The failure of automated translation of ordinary language posed a puzzling problem: if machines can't understand ordinary language, how is it understood, quickly and without effort, in daily life? The solution to this problem points to the nature of general intelligence, how it requires balance between system and intuition, and the crucial role of ambiguity in human expressions.
Linguists have long known that certain types of expressions have meanings that are entirely contextual. Pronouns are an obvious example. When words like you, he, she or it are used in discourse, their meanings are blank checks, to be filled in by contextual knowledge. But the failure of automated translation has shown that all expressions in ordinary language are ambiguous without context. (Artificial languages, such as math and computer programs are exempt, since their symbols have singular meanings. Unlike a text in ordinary language, a file in Word can be translated quickly and exactly into Word Perfect.) In ordinary language, commonly used words have many meanings; the correct meaning can be determined only in context.
This point suggests that understanding ordinary language requires a mind that is a general problem-solver; it relates the smallest parts (words) and wholes (not only systems of grammar and syntax, but a vast array of cultural practices). Even before the advent of attempts at automated translation, the philosopher Wittgenstein understood this point. He proposed that understanding even a simple rule ("Stop at red lights") involved what he called "mastery of practice" (understanding an entire cultural system).
The relation of text to context in the determination of meaning can be rephrased in part/whole language. The method of prospective/retrospective understanding implies that understanding ordinary language requires search of the local and extended context. Although the local context is strictly finite, the particular text of which an expression is a part, the extended context is not. The prospective context is all that happened after the expression, the retrospective context all that happened before it.
As if two potential infinities were not enough, the necessity of using hypotheticals to infer meanings requires still another file that may be large, if not infinite: all the statements that could reasonably have been used as alternatives to the actual one. As indicated earlier, Pittenger et al discuss this type of hypothetical under the heading "The Principal of Reasonable Alternatives." Their discussion concerns only alternatives to actual statements in the text, the local context. It is not applied to the past and future in which the text is located, as does my discussion of hypotheticals in the extended context.
One component needed for understanding ordinary language is a vast file of information, which amounts to biographical knowledge of the author(s) of the text, the audience, and historical and cultural knowledge available to both authors and audience. A second component involves the pathways used in the search for understanding. Given the vastness of the context, how can one find all the information needed for an objective determination of meaning? What makes this task impossible for computers is not merely the size of the information base, but also the means for searching this base.
Computers are limited to pre-defined pathways for searching a file, and, for the most part, to strictly logical operations. These operations involve concepts, class names built upon similarities and differences. Although the classificatory principle is a powerful method of searching, human thought seems to involve many other kinds of paths in addition. Recent attempts to use "fuzzy logic" in computer programs represent a step away from strictly logical operations, if only a very small step. Compared with the ambiguousness of ordinary language, fuzzy logic still reflects the spirit of system.
In explaining their thought processes, chess-players and creative scientists usually refer to non-logical paths such as the "feel for the positions of the pieces on the board" in chess, and instinct or intuition in science. Ordinary language makes use of a wide variety of associations which are not logical. Puns, for example, depend on sound rather than meaning. Solving complex problems seems to need not only logical connections, but also idiosyncratic ones: emotional or biographical contiguity, for example. (One remembers the date of one's first investment because it happened the day before the death of John Kennedy).
Pathways of creative thought involve what I call total association, of which logical associations are only one part. Associations which are merely contingent, such as punning, emotional or alliterative associations, etc., allow for a diversity of connective paths that may be as vast as the memory files themselves. It is the number of pathways that allows human intelligence to be an open system, a general problem-solver (Scheff, 1990, 59-62).
Human intelligence is an open system because problem-solving operations need not follow pre-defined pathways. To phrase the matter under discussion here in computer language, the kind of mental processes that occur in problem-solving may be massively parallel, rather than serial in nature (Scheff, 1990; 1993). Total association, much of it improvised at moment of need, allows for a vast number of pathways. Computers, no matter how large the memory and how advanced the program, are limited to pre-defined pathways using non-ambiguous information. For this reason, the number of pathways is likely to be small compared to the size of the file. And logical connections are inherently serial, since they requiring matching each item with each other item for similarities and differences.
The exact difference between artificial and human intelligence is that mental processes use ambiguous tokens, and proceed through improvised pathways. The ability of the mind to deal with ambiguous tokens and contingent pathways gives rise to what is called intuition, the rapid and seemingly effortless solution of complex problems.
Since words and gestures in ordinary language are only indexical, context dependent for their meaning, understanding and using ordinary language requires the analysis of smallest parts and the largest wholes, a general problem solver. It may therefore be the acquisition of ordinary language that lays down the template for human intelligence. This proposal seems to contradict the prevailing theory of language acquisition (Chomsky 1957). Chomsky proposes that language ability is an inborn drive. This idea bypasses what I consider to be the supreme importance of language learning in forming adult intelligence.
Minsky's Society of Mind
Minsky's (1985) theory of mind is useful for explaining intuition. He argued that mind is based upon structures he called "agents," procedures that generate rapid problem-solving. In Minsky's scheme, the infant comes equipped with only a few agentic procedures, eating, sleeping, playing, smiling, crying, etc. (See Figure 2.). Very early, however, the infant begins developing learned agents, skill procedures, such as walking erect, tying shoe laces, throwing a ball (and jumping into a complex realm), speaking sentences. Adult competence depends upon developing, in Minsky's estimate (1985, p. 314), as many as a billion such agents.
Minsky is careful never to use the term self in his discussion, but his scheme suggests a theory of selfhood. He predicates a type of agent more complex than a skill sequence, capable of command and control. An infant early experiences conflicting desires, e.g. playing, eating, and sleeping. The presence of an ego or self means that the infant develops an agent to detect conflicting desires and to mediate between them by issuing commands. In this case, the control agent might command the playing and eating agents: "Stop!" , the crying agent: "Cry!" and the sleeping agent "Wait!" Without a command-control agent, the neonate must respond only passively to stimuli. Self-control and thought require the regulation of conflict.
A command-control agent is complex compared to a skill sequence because the latter is fixed; it is a constant, one might say. Although a molecule of DNA in situ is extremely large and complex, it is unchanging. A command/control agent is not constant, but variable. In order to serve as mediator, it must be able not only to detect conflict, but make decisions which resolve it. It seems to follow, therefore, that self and mind arise out of conflict.
Minsky's discussion of mental processes implies a further complexity. Even in infancy, he proposes a hierarchy of control agents.
Figure 2 about here.
Level 1, under "builder" (level 2), building with blocks, is simple and concrete, made up of discrete actions: begin, add, end. But level 2, builder, already implies control, able to choose between the impulses to begin, add, or end. The diagram also implies, but doesn't show, a level 5, perhaps at this age the highest level, the ego, which mediates all conflicting impulses. Given the growth of the number of agents, and the increasing complexity of the problems to be solved, more and more levels of control surely would be required.
For example, as an adult vocabulary and syntax are acquired, the language agent presupposes a structure huge both in width and height. Yet this structure must be under the control of an agent at a still higher level, the ego. Another example concerns dreams. Supposing that dreams are produced by a dream agent, the nature of dreams might give us some idea of the functions of the ego, those I assume to be unavailable during sleep.
The particular combination of coherence and incoherence typical of dreams suggest the possibility of a tripartite level of control immediately below the ego level, as it functions when one is awake: l. Receive Information, 2. Evaluate, and 3. Remember. During the dream state, #2, Evaluation, seems to have entirely closed shop, and #3, Remember, functions only weakly or sporadically.
Whatever the specifics of the structure of intelligence, this discussion implies that the mind grows not only horizontally, with increasing numbers of skills, but also vertically, with an increasing number of levels of control. Although horizontal growth may occur through passive accumulation, as in memorization, vertical growth occurs only through conflict. The area (size) of intelligence, one might say, is as dependent on conflict as on breadth of learning. This proposition has so many implications for learning and for education that brevity forbids further discussion, except to say that it implies a need for conflict and discovery in education at least equal to passive acquisition.
Minsky's theory suggests an explanation of intuition. The image that he uses is that the mind is like a society made up of myriads of agents: intelligence involves effective cooperation between them. In Minsky's scheme, the mind is a large society; the number of agents is of the same order of magnitude as the human population of the earth. Given the importance of the differentiation of levels of agents, this model emphasizes hierarchy, as do most sociological models of society. But this is the only aspect of social organization that Minsky invokes. His discussion treats society, with this one exception, as a mass of unorganized individuals.
Minsky proposed that most of the actions of agents goes on out of awareness: "An idea will seem self-evident -- once you've forgotten learning it" (1985, p. 128). In the predicated society, most of the activity of individual and groups of agents occurs outside of the ruler's (the ego) awareness. Minsky's model, in the context of the discussion above about parallel sequencing, suggests the way in which complex problems can be solved rapidly and with little conscious effort.
Suppose that the challenge of a complex problem triggers the actions of thousands of agentic attempts at solution simultaneously. There are a wide variety of logical paths to be tried. At the same time, however, many more non-logical paths, which is probably what we call guessing, might also be involved. If one of the paths unconsciously followed leads to a solution, we call the process intuition. The vast number of diverse paths, logical and non-logical, gives rise to a great multitude of simultaneous investigations. Intuition is constituted by this covert process.
Although the covert mental processes summarized under the heading intuition can lead to rapid solutions of novel and complex problems, this process is not infallible. Since deductions are built on past experience, they may be inadequate in situations not encountered earlier. Like any other hypotheses, intuitive insights need to be verified.
General intelligence requires both deduction and induction, a rapid movement between imagination and observation. The philosopher C. S. Peirce called this process abduction. He was referring not only to the scientific process, the testing of hypotheses against empirical data, but more generally any kind of movement back and forth between imagination and reality. Although not quite stated explicitly, Peirce seemed to imply that the most efficient kind of problem-solving involved the rapid movement between imagination and observation by the individual thinker. The approach to inquiry outlined in this book seeks to maximize the opportunities for abduction, for rapid interplay between theory and fact.
The morphological approach, when combined with explicit part/whole analysis, can generate investigations which are both theory- and data-driven. The method of botanic morphology requires the close observation and description first of single specimens, then comparison of specimens with each other and with other plants. The study of specimen plants in botany is microscopic, the analysis of minutiae, the smallest parts. These minutiae are used for two different purposes: first, to understand the way the specimen itself works as an organism, and second, to differentiate specimens of a given plant from each other and from other plants, a larger whole.
The greatest triumph of the morphological method occurred not in botany but in Darwin's research. One of the crucial observations that figured in the theory of evolution was the minute differences in finches' beaks on neighboring islands. He reasoned that the birds on separate islands were beginning to differentiate because differences in their environments called forth differences in form.
The most explicit description of the morphological method, however, occurs not in Darwin but in Goethe. Although today known only as a poet and writer, Goethe was a polymath, as active a scientist as a poet (Amrine, et al 1987). In his substantial studies in botany, he proposed, contrary to Linnaeus, that the scientist should study function as well as form. This idea goes to the heart of the morphological method. One needs microscopic understanding of the details of a specimen, not only to describe it, as Linnaeus did, but also to understand how it works as a system (in the case of plants, as organisms.) Goethe championed what would today be called functionalism, but as a method (in modern terms, systems analysis) rather than a theory.
Goethe proposed that specimens should be understood in terms of gestalten, a word with no exact equivalent in English. It roughly means patterns, but in Gothe's usage, he emphasized one connotation, complete patterns. Goethe meant not only the pattern within the specimen, but also in relation to its environment. Collecting a huge number of specimens of plants and of animal skeletons, and closely investigating their gestalten, form and function, Goethe came close to anticipating, by almost a hundred years, some of the features of Darwin's theory of evolution.
Goethe's morphological method implies part/whole analysis. His use of gestalten as patterns of relationship within and between organisms, and with the environment, and his pursuit of the relationship between form and function both involve least part/greatest whole analysis. In order to understand the organism, he implied, although it is a system itself, one also needs to understand the host environment, and the relationship between organism and environment. Although Goethe couldn't know about cells, one can pursue his metaphor within the organism as well: to understand the cell, one needs to understand the host organism, and vice versa. In modern terminology his approach was to show the interrelations between systems and subsystems.
Some passages from Goethe explicitly suggest part/whole analysis (1790):
In every living thing what we call the parts is so inseparable from the whole that the parts can only be understood in the whole, and we can neither make the parts the measure of the whole nor the whole the measure of the parts; and this is why living creatures, even the most restricted, have something about them we cannot quite grasp and have to describe as infinite or partaking of infinity.
Goethe's idea of part/whole relations can be easily applied to Darwin's theory of evolution. He showed how all living organisms are subsystems within a larger system of organism-environment relations. Can similar concepts and methods can also be applied to psychological and social systems?
Time order and causal inference
One advantage of the morphological method applied to discourse is that because the temporal sequence is unmistakable, it generates causal rather than correlational theories and evidence. In mechanical records of interaction (that have not been subject to tampering), there can never be any doubt about the time order of events.
Retzinger's(1991) study of the causes of escalation in marital quarrels was generated by Lewis's (1971) study of emotions in psychotherapy because it was causal. Lewis noted in many instances that a patient's hostility toward the therapist seem to occur shortly after the patient showed cues to unacknowledged shame. From these instances, Lewis reasoned that the patient's hostility was caused by shame that went unacknowledged by the patient and uninterpreted by the therapist. Since Lewis had only commented, in passing, on the causal sequence leading from shame to anger, Retzinger designed her research to provide systematic support for it. Both studies were grounded in the incontrovertible time order in which anger and shame appeared in the recorded data.
Retzinger developed a causal theory of destructive conflict, and supported it with 16 instances, all escalations of conflict in the four marital quarrels she studied. She found that although overt anger was present in all instances, cues to unacknowledged shame invariably preceded the indications of anger. (The coding system which Retzinger (1991;1995) developed for identifying hidden shame and anger can be found in the Appendix to this book. )
The advantage of a clear temporal order occurs in verbatim texts, but is not absolute. Although unusual, deception and mistakes are possible. For example, it is now clear that after the end of the First World War, the Russian and French governments faked documents attempting to obscure the order and completeness of mobilization of the armies of the combatants. The fact that the Russian army mobilized first and that their mobilization was complete rather than partial is a key piece of evidence in understanding the origins of the war (Chapter 5). Nevertheless, the time order of most verbatim texts is establishable, as is the case with all mechanical recordings. For this reason, the generation of causal hypotheses and evidence is facilitated more than with correlational data.
Returning once more to Pittenger et al and Labov and Fanshel, these studies can be seen as a beginning step in the morphological method. Each was a microscopic examination of a single specimen. Although not using this term, each was able to describe gestalten, recurring patterns within the single texts they studied. Lewis went a step further in this method, microanalyzing not only single texts, but comparing them each to the other, the second step in morphology. But she did not take the final step, the systematic comparison of the patterns of dialogue she found with types of dialogue other than psychotherapist-patient.
Perhaps because she dealt with so many cases, Lewis failed to make inferences about larger patterns beyond the subjects she studied. Analyzing only a single text, but intensively, both Pittenger et al and Labov and Fanshel took an inferential step beyond their texts. As already described, both studies inferred relationships within their subjects' families: Pittenger et al compared the subject's relationship with her therapist with a hypothetical relationship with her husband, and Labov and Fanshel inferred extensive hypothetical relations within the subject's family. The comparison of the Pittenger et al and Labov and Fanshel inferences with Lewis's suggests a surprising advantage of intensive study of single cases, as against Lewis's much more extensive study of many cases. The closer the analysis of a single case, the more likely it will generate micro-macro inferences, even if they are inadvertent.
On the other hand, findings in the approach called Conversation Analysis so far seem to contradict this last proposition. CA, as it is called, is a highly developed example of the microanalysis of verbatim texts. Their analysis of the words and gestures in ordinary language is state of the art in terms of precision and rigor. This approach, however, has produced few inferences about the institutional embededness of their texts.
The lack of higher order inferences in CA seems to follow from their concern for objectivity. The central tactic employed to this end has been to avoid the analysis of meaning; the procedures are intended to concern behavior rather than thoughts and feelings. But the extent to which analysis of meaning is avoided is questionable. If my reasoning is correct, because of the indexicality of ordinary language, the objective determination of meaning requires least parts/greatest wholes analysis. The parts are the words and gestures, but the wholes that are needed to understand them include the thoughts, feelings, motives, intentions and institutional and historical embededness of the subjects who create the text. These components of social interaction seem to be implicated in CA analysis, but only covertly, as in quantitative studies, disguised as common sense.
Like most other approaches, CA is highly specialized: it is tied to only the first and second steps in what I am calling morphology, the microanalysis of single specimens. Like computer science, CA was created by a highly intuitive researcher, Harvey Sacks. Like most other specialties, this one has gone the route of increasing emphasis on system at the expense of finesse. The incorporation of CA into the three stages of inquiry advocated here might be stimulating and beneficial both to CA and to human science as a whole.
One last example of the potential of the part/whole approach concerns the construction of psychological scales, such as the self-esteem scales mentioned earlier. The construction of scales bypasses explicit theory by using what has become a conventional "empirical" method. Psychologists construct scales by first generating a set of scale items intuitively (i.e. by using an implicit, unstated and therefore undiscussed theory), give these items to subjects, perform a factor analysis, and derive groups of items that correlate with each other more than with other groups, so an individual item correlates with summed scores of its group. In the case of self-esteem, the concept comes to be whatever the vernacular term means, which is assumed, without evidence or even discussion, to relate to the operational concept derived from the scale.
I will use items from a standard scale, the Coopersmith Self-Esteem Inventory (1967), to illustrate a glaring problem with this approach. My argument seems to apply equally well to all of the items that make up the scale; three were selected for brevity.
4. I can make up my mind without too much trouble.
19. If I have something to say, I usually say it.
24. Things usually don't bother me.
All three of these items are scored as positives; that is, a "yes" indicates high self-esteem. Although this may be true in some cases, it is probably false in others. Lacking an explicit theory of the relation between feelings, thoughts and behavior, the construction of the scale doesn't deal with the possibility that positive responses could also reflect personality defenses.
Although an explicit theory has never been used in the construction of scales, the procedures used imply an implicit one: other things being equal, when responding to a researcher who is a stranger to them, with little or no incentive or time for truthfulness and self-awareness, subjects usually say what they mean and mean what they say. Given the sizable literature on the contradictions between attitudes (as constructed from interviews, paper and pencil tests, etc) and behavior, the implicit use of such an obviously inaccurate theory suggests that an explicit theory is needed for the construction of any scale.
Neither these three items nor any of the others attempt to tap repressed or hidden feelings. For this reason, they could just as well indicate impulsive, repetitive, or inappropriate speech (19), or the repression of feeling (4 and 24). Indeed, all three of these items could be scored negatively on a scale designed to detect freedom from obsessive-compulsive behavior. The positive scoring of items that might equally well indicate defenses against feelings of unworthiness as genuine self-confidence may be one of the reasons for the poor performance of self-esteem scales.
How could one distinguish between self-confidence and defensive maneuvers? The morphological method, in the form of microanalysis of verbatim dialogue, might be a place to start. Suppose in a study using a self-esteem scale, one also interviewed the respondents afterwards, asking them to explain their answers to the paper and pencil test. By probing for examples, one might be able to form an accurate idea of the particular meaning to the subject of the answers to these three or any other scale items for each subject. By making the transcripts of the interviews available to the reader, and by explaining one's methods, the resulting study might be both valid and reliable.
Part-whole analysis is the effective component in what is referred to as common sense. It is this process that enables the construction of scales (and all other reliable procedures) that the creators believe to measure self-esteem or other attribute. Yet it has become customary in the human sciences to disparage this process as "psychologizing." Unknown to the creators of scales, they also indulge in the same process, but in a covert and therefore undisciplined way.
In naming the concept purportedly measured by the scale, self-esteem in this instance, and in coding responses as positive or negative, the scale-maker is psychologizing, but in a way that is hidden and not debatable. For example, judging that a positive answer to item 4 above is an indication of high self-esteem involves inferring the meaning of the item, but in a mechanical way. The morphological method could determine the meaning of scale items for each subject individually, and perhaps in the long run, lead to modifications of scaling procedures that might make them more effective.
Although there are hints of such inferences, neither Pittenger et al nor Labov and Fanshel took the next step beyond inferring patterns in the subject's families, which would be to infer patterns in types of families, or membership in religious or other institutions. Even if they had made such inferences, they would have been unable to confirm or modify them, since their data was limited to texts only.
Inferences concerning the social institutions within which a text is embedded provide a rich resource for generating a micro-macro theory. Such a procedure is enriching because the texts intimate, in great detail, the subjects' connections to larger institutions. Mention has already been made to institutions beyond the family within which the texts studied by Pittenger et al and Labov and Fanshel were embedded: class, gender, ethnicity, and age. Verbatim excerpts from discourse, one might argue, are microcosms, they contain within them, brief as they may be, intimations of the participants' origins in and relationships to the institutions of the host society.
Perhapsthe idea of functionalism could still be useful, not as a theory but as a method. The discussion of the analysis of psychological and social systems and sub-systems involves the method implied in functional ideas. These earlier ideas about functions concerned relationships between parts and whole, but without a method or data. The method described here leads to the investigation of such relationships, rather than using them as explanations. The stages of inquiry proposed here lead from the microanalysis of single specimen texts up to the generation of theories grounded in texts.
The generation of micro-macro theories from linguistic texts is an unusual procedure, but it has been practiced by two masters. In a series of studies, Levi-Strauss (1963) teased out cultural implications of myths that he gathered from small traditional societies. From the myths of each culture studied he deduced what he called the cognitive structure of that culture, the fundamental dimensions of thought.
Although in some ways Levi-Strauss's approach is similar to mine, it is also quite different. Two differences are basic. First, Levi-Strauss did not use verbatim transcriptions of the myths he gathered. For this reason, he was unable to analyze what I would consider the smallest parts of the cultural systems he studied, the words and their nonverbal accompaniment. But emotions and social structure virtually always ride upon gesture. Even adroit questioning will seldom yield much reliable information about emotions. Institutional connections might be somewhat more available, but even there the descriptions are usually threadbare. Without the nonverbal components of discourse, even if he had wanted to, Levi-Strauss would have been unable to infer much about emotions and social structure.
For example, in my re-analysis (Chapter 8) of the text used by Pittenger et al, I used nonverbal elements in the recording to infer that the patient was responding to the class difference between her and the therapist. At several points, after mispronounced words or other errors, the patient not only corrected her speech, but also showed signs of embarrassment. I inferred that she was seeing her responses from the point of view of the therapist, and judging them from the point of view she attributed to him, rather than her own. This finding is a clear indication of dominance-subordination in the relationship.
These instances of the patient's behavior also hint at what I call engulfment on the patient's side, giving up parts of herself out of deference to the other person. The difference between engulfment, a form of alienation, and isolation, the other form of alienation, on the one hand, and solidarity, on the other, invokes fundamental structures of relationship that are just becoming available to empirical research (See Chapter 4 ) The smallest parts, in this case the words and their accompanying emotion cues, allow inferences about the emotional meaning of the relationship to the patient, and also its social structure.
Just as Levi-Strauss's analysis is entirely cognitive, it is also entirely psychological. He makes inferences about the mental functions of individuals; he has no categories of relationship. Except for hierarchy, Minsky's (1985) brilliant analysis of the nature of mind has a similar limitation. Just as Levi-Strauss interprets culture in individual terms, Minsky interprets society. Neither analysis conceptualizes types of relationships, which are as much building blocks of societies as individual persons are.
Elias (1978), a second master who generated micro-macro theories from linguistic texts, conceptualized both individual and relational structures, both cognitive and emotional meanings. For these reasons, his approach is much closer to mine than Levi-Strauss's.
Drawing upon some 500 years of European history, Elias analyzed excerpts from etiquette manuals in four languages. Closely examining advice on table manners, body functions, sexuality and anger, he showed that his excerpts suggested an explosion of shame connected with modernity. His analysis of an excerpt from a l9th century text advising mothers about their daughters is illustrative. The author, a male, advises fables, silence and suppression when daughters ask sexual questions, such as where babies come from.
Although working with a printed text, Elias was able to tease out the emotions implied, principally the author's intense embarrassment about sexuality, and how the behavior advised would shame the daughter into silence. Elias notes both what is stated and what is omitted, and most importantly, the manner of presentation, to infer emotional meanings.
Elias provides a very close analysis of specific texts, but he also infers emotional meanings involving social structure. In a bold and provocative way, Elias linked changes in emotional expression to changes in social structure. For example, he outlined the way in which the narrowing of the control of the means of violence in a society to a small ruling elite might be related to increasing suppression of individual anger throughout a society.
Unlike Levi-Strauss and Minsky, Elias uses relational categories as well as ones involving individuals. He criticized studies based on what he called homo clausus, the closed, solitary individual. In the place of this conceptualization, he proposed an incipient theory of social solidarity/alienation. He argued that social relationships can involve either independence, interdependence, or dependence. This scheme is only implied in the 1978 study, but is stated more explicitly in his later work (for instance, Elias 1987). His scheme is closely related to Durkheim's analysis of the causes of suicide, and other classic work in sociology (See Chapter 4).
Elias used elements in the two stages of inquiry I am advocating, morphology and micro-macro theory, but his approach is still different from the one outlined here. The difference involves explicitness with regard to concepts and methods. Although he analyzed verbatim texts, Elias has no explicit description of his methods, nor of the concepts in the micro-macro theory he seems to generate. For this reason, it is difficult to argue with his findings, or even to state with any precision exactly what they are. This limitation, lack of explicitness about methods and concepts, may be the reason for the widely varying understandings and evaluations of his work.
Elias's study pointed the way to the approach outlined here because it combined elements of two of the three stages of inquiry. In this particular study, all of Elias's propositions about personality, social structure and social change are derived from and grounded in microanalysis of verbatim texts. The clarity and incisiveness of this study, as compared to most of his later work, are probably due in large part to this procedure. If Elias had been more explicit about his methods and concepts, perhaps the study would have been more widely accepted, and generated later attempts at verification.
Being cavalier about methods and concepts may be characteristic of discoverers. The above comments are also applicable, to a slightly lesser extent to the major work by Lewis (1971). Although she used a standard and therefore explicit method as one part of her approach to coding emotions (Gottschalk and Gleser 1969), she did not attempt to explain or even discuss the more intuitive methods that she combined it with. And although she was explicit about defining some of her concepts, there is no formal statement of theory, and no attempt to infer larger social structures.
This chapter has described a part/whole morphology of human behavior. The core of my proposal is the need for a new step of inquiry that will bridge the chasm between exploration and verification. When combined with qualitative methods, part/whole morphology can be used to approach seemingly intractable problems in the human world, generating comprehensive hypotheses to the point that they might be tested. When combined with quantitative methods, the same two steps can lead to the comprehensive testing of the hypotheses they generate.
The basic strength of this approach is that it places the researcher (and the reader) into direct contact with the raw data of human behavior, verbatim texts or mechanical records of interaction. This approach has an intensely inductive quality that is missing from conventional research designs. Quantitative studies shield both researcher and reader from contact with actual events and sequences of events by cross-sectional designs and by layer on layer of paper and pencil tests, coding, scales, and numerical analysis. Qualitative studies may come closer to human reality, or plausibly appear to do so, but only as filtered through the observers' fallible memory, sensitivities, and biases. As in quantitative studies, the human reality on which studies are based is usually unrecoverable.
The approach outlined here allows for the recovery of large parts of the original events -- instant replay -- in a way not permitted by conventional methods. Such recovery not only allows direct falsification, it also means that data are not lost forever. It can become the grounds for subsequent advances, as I have shown with the Pittenger et al and Labov and Fanshel studies in this chapter. Given this approach, the voices of our subjects are not silenced by our methods, but amplified and preserved for future generations.
Being exposed continually to the raw data of human interaction is particularly stimulating to a researcher. Listening to audiotapes and viewing videotapes or film come very near to being able to reproduce the original scenes at will. Exposure of this kind allows us to learn something new about the subjects and about ourselves, expanding the horizons of the study, whatever its original intent. Human voices and faces, so long absent from so much of human science, spring to life repeatedly, as they are needed as a prod to our sleepwalking through our projects.
Since the approach proposed here is synoptic, rather than specialized, it might be a way to begin integrating the contributions of theory and method, and the various methods, disciplines, levels of analysis, and schools of thought in the human sciences. In this way it is possible to envision, instead of the alienation that now prevails, at least a beginning for community among those of us who try to understand the human condition.
One final implication of the approach outlined here is the need for generalists as well as specialists to solve the problems which beset our disciplines and our societies. Perhaps one direction would be to press for centers of general studies, whether programs, institutes, or departments, to deal with the masses of specialized knowledge with which we are now inundated. The fate of the interdisciplinary programs in our era has not been heartening, but our circumstances cry out for new beginnings.
This chapter has argued that part/whole morphology, by successive approximation, can generate valid micro-macro theories. Although this approach is arduous, it may provide a framework for integrating existing work in the human sciences, and goals for the future. As Blake noted, all art and science is based on "minute particulars." The approach outlined here may yield robust theories of individual and collective behavior that are deeply grounded in the minute particulars of human existence. Needless to say, there is many a slip between the cup and the lip. The chapters below show, in a halting, preliminary way, some of the results of applying the new method.