Interaction and Identity [1 ed.] 9781560001911

Citation preview

INTERACTION & IDENTITY Information and Behavior VOLUME 5

Inform ation and Behavior Series Brent D. Ruben, Series Editor Volume 1 Information and Behavior, edited by Brent D. Ruben Volume 2 Information and Behavior, edited by Brent D. Ruben Volume 3 Mediation, Information, and Communication, edited by Brent D. Ruben and Leah A. Lievrouw Volume 4 Between Communication and Information, edited by Jorge R. Schement and Brent D. Ruben Volume 5 Interaction and Identity, edited by Hartmut B. Mokros

INTERACTION & IDENTITY Information and Behavior VOLUME 5

Hartmut B. Mokros EDITOR

First published 1996 by Transaction Publishers Published 2017 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN 711 Third Avenue, New York, NY 10017, USA

Routledge is an imprint of the Taylor & Francis Group, an informa business Copyright © 1996 by Taylor & Francis All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. ISSN: 0740-5502 ISBN 13: 978-1-56000-191-1 (hbk)

For my father with love, admiration, and gratitude

Contents

Preface INTRODUCTION 1. From Information and Behavior to Interaction and Identity Hartmut B. Mokros

vii

1

PART I: Issues of Theory and M ethod in Interactional Study of Identity 2. Part/Whole Discovery: Stages of Inquiry Thomas J. Scheff 3. Communication Gregory Bateson 4. Pseudounilaterality, Simple-Rate Variables, and Other Ills to which Interaction Research is Heir Starkey Duncan, Jr., Barbara G. Kanki, Hartmut B. Mokros, and Donald W. Fiske 5. Subjective Time, Social Interaction, and Personal Identity Tom Bruneau

25 45

71

97

PART II: Identity as Interactive C onstruction 6. Constructing Discourse Identities in the Openings of Academic Counseling Encounters Agnes Weiyun He 7. Constructing Social Identity in the Workplace: Interaction in Bibliographic Database Searches Jenny Mandelbaum 8. Identity, Subjectivity, and Agency in Conversations about Disease Nancy L. Roth

119

145

171

PART III: Statements of Identity as Interactively Constructed 9. Internal Muzak: An Examination of Intrapersonal Relationships Linda Costigan Lederman

197

10. Constructing Research Narratives and Establishing Scholarly Identities: Properties and Propositions Leah A. Lievrouw 11. Razzing: Ritualized Uses of Humor as a Form of Identification among American Indians Steven B. Pratt

215

237

PART IV: B arriers to Interactive Conceptualizations of Identity 12. Ambiguous Bodies/Believable Selves: The Case of Herculine Barbin Tamsin Lorraine 13. The Constitution of Identity as Gendered in Psychoanalytic Therapy: Ideology and Interaction Margaret A. Carr 14. Interpersonal Icons: Remembered Images and the Closure of Discourse from a Lacanian Perspective Mick Presnell and Stanley A. Deetz

259

273

297

PART V: Remaking Identity In teractio n al^ 15. The (Re)construction and Negotiation of Cultural Identities in the Age of Globalization Getinet Belay 16. Identity Development: From Cultural to Intercultural Young Yun Kim 17. Identities and the Assimilation Process in the Modem Organization Gordon L. Forward and Dirk Scheerhorn 18. Work and/or Caring: Exploring the Identification of Women’s Activities Kim Wittenstrom 19. The Index, in Context Ira Kleinberg Contributors Citation Index Subject Index

319 347

371

393 411 427 433 447

Preface Brent D. Ruben

The idea for a series devoted to information and behavior first em erged in 1982. A t the tim e, one could only im agine the significance and popu­ larity that w ould co m e to be associated w ith the term “inform ation” in the ensuring years. In the first volu m e, as w ell as those published sin ce, w e have tried to resist the tem ptation to be directed by fad or fashion. Rather, our goal has been to exam ine the m ost fundam ental aspects o f inform ation in relation to human behavior. C om m unication and the processing o f in­ form ation has been v iew ed as a basic life processes, ones w hich are necessary to the hum an behavior at all le v e ls— individual, relational, group, organizational, and societal. In this fifth volu m e o f the series, the focu s is on interaction and iden­ tity. Through his ow n w ritings and those o f other contributors, Hartmut M okros provides a broad-ranging exam ination into the linkage betw een tw o m ost important concepts, further extending the line o f inquiry into relationship o f inform ation and behavior to w hich this series is devoted.

vii

1 Introduction: From Information and Behavior to Interaction and Identity Hartmut B. Mokros

Scholarly interest in issues of self-identity has exploded across disci­ plines within the humanities and social sciences in recent years (e.g., Chames, 1993; Fitzgerald, 1993; Gergen, 1991; Giddens, 1991; Keith & Pile, 1993; Lorraine, 1990; Shotter & Gergen, 1989; Taylor, 1989; White, 1992). Common to this interest is the assumption that self-iden­ tity is not a priori, not given or fixed, but communicatively constituted. The “transdisciplinarity” (Rowland, 1988) characterizing this interest may itself be seen as reflective of concern and dissatisfaction with tradi­ tional disciplinary identities and the intellectual trade restrictions they have imposed. Hyper-specialization, methodological obsessiveness, and theoretical impotence are no doubt among the more visible disciplinary qualities to have stimulated desire for new looks and lookings. Within this climate the grounding of such traditionally important, and disci­ pline supporting, oppositions as subject/object, body/mind, individual/ society, male/female have been thrown into question. It is a climate within which communication has come to occupy notable centrality. This is particularly evident in contemporary writings, which share in their move away from developmental perspectives (e.g., Erikson, 1980) a view of identity as contingent (White, 1992), as discursively and interactively constituted (e.g., Gergen, 1991). Interaction and Identity; volume five of Information and Behavior, aims to contribute, theoretically and empirically, to contemporary schol­ arly interests in issues of identity. Although the focus of contributions to 1

2

Interaction and Identity

this volume is on identity, its development and scope is embedded within theoretical discussions of communication that have been a concern of this series since its inception. It is within the context of such discussions that this volume was conceived and developed as a contribution to In­ formation and Behavior. From Inform ation to Communication

Information and Behavior was inaugurated in 1983 to provide an in­ terdisciplinary forum for examining the implications and increasing cen­ trality of information in contemporary life and scholarship, in what has come to be called the “Information Age.” The concept of information gained widespread currency and analytic clout with the rapid develop­ ment and proliferation of information technologies during the later half of this century. The impact and opportunities that these revolutionary technological innovations pose for everyday experience—from re­ definitions of the workplace and experiences of leisure, to the types of competencies and skills necessary to succeed in contemporary society— clearly motivated the founding of this series. As the series has evolved, however, the concept of communication has become an increasingly explicit concern, from a focus on Media­ tion, Information and Communication in volume three (Ruben & Lievrouw, 1989) to a look at the relationship Between Communication and Information in the last volume, volume four (Schement & Ruben, 1993). With a stated aim “to map out the research front addressing the relationships between communication and information” (Schement & Ruben, 1993, p. xi), volume four called attention to the conceptual inseparability of communication from discussions of information and behavior. But there is more at issue in discussions of the relationship between communication and information than claims of conceptual inseparability. Specifically, how concepts of communication and in­ formation relate to our understandings of everyday experience is of particular relevance. Reading across contributions to volume four and earlier volumes of Information and Behavior reveals two distinct paths for discussing the linkage between communication and information, with the first empha­ sizing information and the second communication. While communication and information are commonly treated as interchangeable along the first

From Information and Behavior to Interaction and Identity

3

path, there is, nevertheless, a clear distinction made, with information extended priority and communication subordinated to information. This priority would seem to be based in the idea of information as a representa­ tion of reality, as a representation of the object world in which human experience is situated. Information is seen to “stand for” something, that is, to represent reality out there. As a representation of things out there, information itself assumes the stature of thingness, becomes an entity, whose value increases the more so that it economically, parsimoniously, and objectively represents this reality or enables the reorganization of ex­ isting information so as to allow individuals to cognitively experience re­ ality in a new way. Communication, within this framework, is the process by which representations are coined and ratified as information and, more visibly, the process by which information is exchanged between individu­ als. Underscoring this definition of communication in information terms is the common employment of efficiency and fidelity as criteria for the evaluation of this exchange process. Communication is thus treated as a tool for labeling reality and a tool through which access and ownership of information may be realized by individuals. Along this first path information achieves priority through the implicit assumption that information enables the cognitive capture, organization­ ally and representationally, of reality out there. It is this assumption that is contested along the second, a path that regards communication as pri­ mary. To make communication primary for understanding behavior is to view communication as “constitutive” rather than as an exchange process between two information processors. When communication is viewed in this way, “known” properties of reality are assumed to be contingent upon, or only made possible by virtue of communicative action. This then im­ plies that tokens of information are not simple representations of external reality that reduce uncertainty, or qualities of “the organizational work a message enables its receiver to perform” (Krippendorf, 1993, p. 488), but are instead embedded within communicative activity, and thereby sen­ sible not in relation to some essential, objectifiable reality, but in relation to other informational tokens. The relative value of any token of information (including the concept of information as a token) is thereby seen to not be a function of the economy, parsimony, and objectivity in its ability to represent external reality for “knowing” individuals. Instead its value derives from its pref­ erential use, its currency in discourse. This then implies that informa­

4

Interaction and Identity

tion tokens are politically and ideologically laden and dependent, dis­ cursive realizations within the constitutive moments of communicative action. From this perspective, information tokens create the illusion of a singular knowable universe in so far as they are treated as representa­ tions and guides for knowing an assumed independent reality. Obscured thereby is a view of information tokens as effects of communicative action. Seen as effects, information tokens first and foremost reflect the reality of social organization and value that communicative action un­ avoidably reveals. From Behavior to Identity The distinctions between these approaches are not merely matters of intellectual debate or paradigm difference. They entail radically diver­ gent conceptualizations of the nature of human experience and its adap­ tive and transformative course (cf. Benhabib, 1992; Deetz, 1994). The information path focuses on the role of information in relation to the behavior of persons (with institutions, organizations, and societies simi­ larly personified), with persons assumed to be autonomous, rational, self-contained processors and producers of information (viz. behavior) within this framework. Information innovations are envisaged as poten­ tiating greater autonomy and rationality, thereby furthering what has been called the modernist or enlightenment project of the self (e.g., Giddens, 1991). The principal experience of the self is conceived to be cognitive, in the Cartesian sense. This assumption of the primacy of cognition in lived experience makes the acquisition of information a morally progressive good in the service of the knowing self. The communication path contests this basic assumption of the pri­ macy of the knowing self and the privileged status that information de­ rives from this assumption. Rather than being viewed as a tool or a process of message exchange, as is the case within the information approach, communication is viewed as a site within which experience achieves a sense of coherence, structure, and meaning. The communicative mo­ ment occupies the intersection of the psychological and the social, within always present, yet changing and socially modifiable, biological and physical environmental constraints. Within communicative situations and through communication practices, social institutions and values are pro­ duced and reproduced by individuals in interaction (e.g., Bourdieu, 1977; Giddens, 1984). Just as communication practices inevitably produce and

From Information and Behavior to Interaction and Identity

5

reproduce social institutions and values, they also produce and repro­ duce individual identities in terms of institutions and values. Through interaction with others, persons derive a sense of identity, a sense of place. To then capture what is a person, what are the qualities of lived experience, how is change in a person and in social institutions possible, commands attention to the intersection of the social and psychological in communicative moments (of others). Study of persons from the perspective of communication, at the inter­ section of the psychological and social, requires making problematic the linkages between, and concepts of, person and behavior. As a con­ cept, behavior presupposes unilateral agency. By extension, it is then quite straightforward to treat behavior as information that is indexical of a person’s competencies, beliefs, and desires; that is, of a person’s “intemality.” To emphasize instead a view of behavior as communicatively situated and thereby as communicative practice, is to see behavior as an expression of interactional contingency occasioned within a system of social constraints and enablements. This then directs attention away from behavior as indexical of a person, toward a recognition that what is re­ garded to be the behavior of an individual is a collaborative social sense­ making activity, within which statements of identity in relationship to others are continuously realized. The Communicative Constitution of Identity Identities of persons may be said to be communicatively constituted in two senses: discursively and interactively. The discursive constitu­ tion of identity references the impact of sociocultural knowledge or dis­ course on social practice. Discourse, from this perspective, identifies the expressive possibilities and permissibilities—that is to say, the sys­ tems of etiquette—that guide human agency and provide the parameters within which self-identity is constituted and evaluated. This framework has been by far the most active locus of scholarly interest in the study of identity as communicatively constituted, particularly within the context of what has come to be called social constructionism (e.g., Shotter & Gergen, 1989,1994). Within this framework social conditions, as these are reflected in discursive possibilities, are of primary interest, not “lived moments of social interaction” (Pearce, 1994). Thus, although social interaction is acknowledged as the site within which discourses and iden­ tities—as discursive expressions—are constituted as productions and

6

Interaction and Identity

reproductions, specifically how identities are constituted in lived mo­ ments of social interaction and what is entailed in an interactional analysis of identity has been largely underdeveloped. One obstacle that has stood in the way of such interactional analysis is the failure to make problematic the concept of interaction. In its least productive, yet most common usage, the term interaction serves as a gloss for encounters between one or more persons or personified things. This is seen in studies that claim to be studying interaction for no other reason than because their data come from dyadic conversations, for ex­ ample. Interaction in such cases is conceived in terms of an event (namely, social encounter) rather than as a process. A second sense of the term interaction emphasizes process, in par­ ticular a process of mutual influence (e.g., Capella, 1981). McCall and Simmons (1966) provide a familiar statement of this perspective. They write, “Whenever a relationship of deterministic influence between two events cannot be resolved into a simple function of one but must instead be treated as joint function, as a mutual or reciprocal influence, we have a case of interaction” (p. 47). Events, and the agencies that give rise to events, are prior to interaction within this conceptualization. Social in­ teraction is regarded as equivalent to the concept of interaction as an effect within the familiar analysis of variance model, an effect which accounts for some proportion of total variance unaccounted for (e.g., Kenny & Malloy, 1988). Although these two senses of interaction appear quite distinct, they share in common an assumption of the primacy of persons and things. In both formulations interaction is regarded as a discrete entity. In the first sense, interaction is the event constituted by social engagements. In the second, it is a potentially emergent property that arises within social engagements. Conceptualization of interaction in these terms is anti­ thetical to a constitutive perspective. A constitutive perspective rejects the priority of events and objects and instead posits the priority of an interactional system within which events and objects are achievements (of identity), not as stable entities, but as contingent realizations. Contributions to Interaction and Identity While theoretical discussions focusing on the discursive or textual constitution of identity have been many, empirical examinations of iden­

From Information and Behavior to Interaction and Identity

7

tity as interactionally constituted have been few. Contributions to this volume have been selected to address this imbalance and to thereby encourage empirical study of identity as interactionally constituted and theoretical implications of interactional perspectives on identity. These contributions have been divided into five sections. Included are: (1) discussions of issues of theory and method relevant to empirical study of identity as interactionally constituted; (2) empirical studies of identity as interactionally constituted; (3) examination of interactional properties of enduring statements of identity; (4) considerations of the blockages, barriers, and potential consequences that the denial of a con­ stitutive conceptualization of identity creates in interactional engage­ ments; and (5) formulations that argue for the remaking of identity from an interactive stance.

Issues o f Theory and Method in Interactional Study o f Identity Difficulties involved in conceptualizing data that capture the pro­ cess of interaction while incorporating hierarchies of contextual pa­ rameters requires bridging the separation between theory and research, through a grounding of theory in concrete, well-described phenom­ ena, with theory and data then entering into an ongoing dialogue. The chapters by Scheff, Bateson, Duncan and colleagues, and Bruneau each speaks to this issue. “Contemporary scholars and scientists,” writes Thomas Scheff in chapter 2 in this volume, “seem to have difficulty visualizing part/whole relationships. In the current division of labor, the organic connection between part and whole is lost...[as] theorists deal with wholes but not parts [and] researchers deal with parts but not wholes.” This is in part a product, he continues, of artificially defined and rigid disciplinary boundaries which “make it practically impossible to generate and test general theories.” Scheff offers an approach to analysis that he calls part/whole analysis, namely, “proceeding from the smallest possible minutia up to the largest possible theory,” antecedents of which he traces to the writings of Goethe and Wittgenstein. Central to this approach is fine-grained “morphological” study of concrete phenomena. Although Scheff does not specifically concern himself with identity, the relevance of his analytic approach for the development of an interactional conceptualization of identity is obvious.

8

Interaction and Identity

A similar theme is apparent in Gregory Bateson’s chapter (chapter 3). This chapter, written as the introduction to a well-known interdisci­ plinary collaboration entitled, The Natural History o f an Interview,; has previously only been available on microfilm (McQuown, 1971).1 Al­ though completed more than thirty years ago, this chapter, by certainly one of the intellectual giants of the century, remains surprisingly timely and prescient with respect to the theme of interaction and identity. Bateson proposes communication as the logical site and explanatory frame for the investigation of human experience. Theoretical developments, from information theory, psychoanalysis, linguistic anthropology, gestalt psy­ chology, and learning theory to interpersonal interaction theory, all point, according to Bateson, to the centrality of communication. The availabil­ ity of permanent recordings, in the form of film and videotape, makes possible, Bateson proposes, the “microscopic examination of personal interaction” within the stream of communication. It is through mi­ croanalysis that the constitutive nature of communication, “a sequence of contexts both of learning and of learning to learn,” as Bateson puts it, may be empirically studied. While both Bateson and Scheff offer important considerations for the interactional study of identity, they do not directly confront what this entails in the study of specific cases. Such an effort marks the concern of chapter 4 by Starkey Duncan, Barbara Kanki, Hartmut Mokros, and Donald Fiske. The chapter begins with the identification of two distinct directions in research of social interaction: one focusing on the process of interaction, “how interaction is accomplished by participants,” and the other on the actions of participants in social interaction “as indices of the participants’ properties, characteristics, or transient internal states,” that is to say, on the study of “individual differences.” Chapter 4 ’s major claim is that the uniqueness of an individual or of groups of individuals can not be studied through measures that only record the actions of that individual. This insight becomes clear when individuals are studied in interaction process terms. Duncan and colleagues argue that since a person’s actions in social interaction are potentially contingent (that is to say, interactionally embedded), it then follows that inferences based on the rates of activity produced by a person, like the number or extent of time one looks at another person in interaction, are prone to an error they refer to as pseudounilaterality; namely, “the false assumption” that any action “is necessarily unilaterally determined by the actions of the

From Information and Behavior to Interaction and Identity

9

participant.” In this chapter, and elsewhere (Duncan et al., 1985), Duncan and his colleagues grapple with alternative variables and alternative empirical approaches to the study of persons that capture the interactivity of social encounters. One of the clear dangers of a constitutive or constructionist perspective is the knee-jerk dismissal of the potentialities and constraints that biology and physical environment impose. While it is certainly the case that the reduction of social and cultural forms to biological functionalism miscon­ strues the place of culture (Sahlins, 1976), to ignore biology outright is shortsighted. One need only consider how our evolved biological senses influence and structure the process of sense making. The centrality of visual metaphor (“I see what you mean”) in discussions of thinking and knowledge provides one such example. Chapter 5, by Thomas Bruneau, provides another. Bruneau argues that biological foundations of temporal experience are pervasive and yet have been largely ignored in studies of human communication. Because of the pervasiveness of temporal regu­ larities in human experience, Bruneau suggests that study of the structure of interaction and of identity as interactionally constituted will remain incomplete if such regularities are not considered.

Identity as Interactive Construction Although the chapters by Bateson, Scheff, Duncan and colleagues, and Bruneau clearly have relevance for the development of an interac­ tive account of identity, they nevertheless do not directly address the issue. The subsequent chapters by He, Mandelbaum, and Roth do so as they offer empirical efforts to study the interactive constitution of iden­ tity. They do this through the study of talk in dyadic conversations. The chapters by Agnes He and Jenny Mandelbaum (chapters 6 and 7) share a common research orientation in conversation analysis (CA). An offshoot of ethnomethodology (Garfinkel, 1967), CA owes its develop­ m ent to the research and teaching o f H arvey Sacks (1984). Ethnomethodology assumes that the perceived coherence and orderli­ ness in everyday social encounters is a product of methodical social activities. The CA research approach relies on precise transcriptions of natural conversation that aim to capture the spokeness of talk. Paralinguistic qualities and especially accurate capture of the sequential unfolding of talk are key features that the transcripts aim to preserve.

10

Interaction and Identity

CA does not merely rely on these transcripts for analytic purposes but presents them in its publications, thereby allowing the reader to work through the analysis offered within data from which it emerged. The role of the analyst is to uncover methods by which the engagement be­ tween participants in talk may be said to be an achievement; that is, how talk as conversation is done so as to constitute the taken-for-granted notion of conversation, for example. Chapters 6 and 7 examine the constitution of identities in two realworld settings, within which individuals have real-world aims. What they each demonstrate is how “personal” identity is an “occasioned” and “collaborative” production of the participants in the conversations. That is to say, the achievement of what we might claim as an individual’s identity is shown to be relationally contingent. As He puts it in her study of the opening of academic counseling sessions, “[I]t has been difficult to discuss the construction of the student’s identity in separation from that of the counselor’s and vice versa. “Similarly, Mandelbaum con­ cludes that “speech acts to get the job done, but how it is used to get the job done is replete with relational sensitivity. In this sense when interactants propose what to do next, their proposals appear to be shaped to indicate who they take themselves and one another to be.” The final chapter in this section (chapter 8), by Nancy Roth, also examines talk, but does so quite differently from the approach used in CA, making no claims to transcribe or analyze talk as microanalytically or exhaustively as is true in CA studies. Roth examines what she calls three aspects of the self (i.e., identity, subjectivity, and agency), as these are discussed by Goffman, Foucault, and Giddens, in a set of experi­ mentally guided dyadic conversations. Each dyad was asked to read one of five stories that had supposedly appeared in a local college newspa­ per and to then use the story read as the basis of their conversation over the next ten minutes. These stories reported the case of a student af­ flicted with one of five diseases, including colds, cancer, syphilis, hepatitis-b, and AIDS. Of particularly interest in her data is how participants in conversation construct themselves in relation to these stories. Although she presents no effort at systematicity or synthesis in approaching her data from the perspectives of Goffman, Foucault, and Giddens, it is clear that she favors Giddens’s theory of “structuration” because it addresses the interplay of interactive (Goffman) and discursive (Foucault) influ­ ences on the constitution of self.

From Information and Behavior to Interaction and Identity

11

Statements o f Identity as Interactively Constructed While illustrations of identity as interactionally negotiated are cer­ tainly important, to focus on identity as momentary, the product of inter­ active collaboration and negotiation, obscures the fact that individuals construct and may indeed be said to be socially (and personally) obli­ gated to project statements of enduring or fixed identity. The chapters by Lederman, Lievrouw, and Pratt examine how individuals identify themselves to others (and themselves) as a specific type of person, as occupying a specific niche or role within society. What each of these chapters makes apparent is that statements of enduring or fixed proper­ ties of identity are no less interactive in their development than are those of immediate negotiation. Strongly influenced by the conceptualization of the self developed by George Herbert Mead (1934), Linda Lederman (chapter 9, this volume) explores the “communicative” relationship that a self has with itself. She calls communication of this type intrapersonal communication and aims in her chapter to examine its qualities. Clearly the generation of data is a tricky issue for the study of intrapersonal communication. Lederman’s solution is to examine the accounts provided by partici­ pants in focus group interviews that targeted self-talk as their guiding concern. This approach generates a particular type of self-talk, namely, accounts about self-talk.2 These accounts, she finds, characterize the self as a set of multiple relationships, as a conversational engagement among multiple persona. As an ongoing activity, this relational sense of self becomes the enduring background or context within which interper­ sonal encounters are situated, according to Lederman. She concludes by suggesting that “interpersonal communication is more usefully studied when the selves in interaction with one another are recognized as the presenting selves, the tips of the iceberg of self and the interaction of the presenting selves with other individuals is influenced by the relation­ ships that any individual has among its multiple other selves.” This ongoing set of relationships, among multiple selves and present­ ing selves with social others that Lederman describes, suggests a view of an enduring sense of self as a storyline or narrative. Statements of self as narrative have been much discussed of late (e.g., Bruner, 1990; Ricoeur, 1992) and is the theme that guides chapter 10, contributed by Leah Lievrouw. In particular, Lievrouw examines the research narrative as a

12

Interaction and Identity

statement of identity. Research narratives, she writes, are statements of a “scholar’s own ‘story’ of his or her professional life and work. It is a consciously constructed, strategic account that allows the researcher to communicate the value and necessity of his/her work to peers and other important audiences.” The study of documents of this type is exceed­ ingly important as they contain what might be called the “conventions of legitimation” employed by members of the scientific community to argue for the worthiness of their enterprise and claim their identity as scientists. Lievrouw argues that the conventions employed by research­ ers are based in the properties of narrative more generally. From this perspective research narratives are not primarily motivated as descrip­ tive statements of what a researcher does, as fixed statements of a researcher’s identity through his or her activities. Instead, the research narrative is an interactive occasioning—situated first and foremost within the prevailing climate of public discourse—tailored for specific social audiences and purposes and expressive of its creator’s sense of place: developmentally, socially, psychologically, intellectually, generationally, and geographically. Research narratives are but one example of statements of identity. Statements of identity would seem to be pervasively mandated in every­ day life as economical means to “define” who a person is and what might legitimately be expected of them. As such, these statements are about membership and thereby identity through social bonds (Scheff, 1990). Steven Pratt (chapter 11, this volume) discusses one such state­ ment and how it is achieved. Working within the ethnography of speak­ ing tradition (Hymes, 1962), he examines how a specific speech event, which he calls “razzing,” is employed by Native Americans to interactionally test and affirm the “Indian-ness” of participants in interac­ tion. Pratt claims that “Indian-ness, to the ‘real’ Indian, is not something that one can simply be, but is something one becomes and/or is, which requires the participation of other culturally competent members.” Pratt shows how the communicative practice of razzing makes the concept of Indian-ness a matter of continual scrutiny, assertion, and affirmation even though participants in this process regard their Indian-ness as fixed and stable. Although Pratt does not consider the significance of the relationship that specific qualities of razzing as a communicative activity and the cat­ egory Indian-ness hold, his chapter certainly raises this issue. The cat­

From Information and Behavior to Interaction and Identity

13

egory “Indian” is one that has been imposed upon, and according to Pratt, accepted by Native Americans as an anchor around which a sense of self is constituted. It is then clearly not insignificant that this sense of self as Indian is affirmed and reaffirmed through practices, albeit they are re­ garded as humorous, that emphasize shame and humiliation. To be sure, razzing is a practice that affirms one’s membership and identity as an Indian but it weds to this affirmation statements of denigration and ridi­ cule. The consequences of such practices for the broader social bond and the positioning of individuals within the social bond, within which they achieve a sense of self and self-worth, represents an exceedingly impor­ tant nascent area of scholarship (Scheff, 1990; Scheff & Retzinger, 1991).

Barriers to Interactive Conceptualizations o f Identity The relevance of views of identity as contingent, as interactively achieved, is not merely a matter of academic exercise. Quite to the con­ trary, how identity is conceptualized discursively and reproduced inter­ actively has direct bearing on personal rights, responsibilities, and opportunities. Constitutive perspectives on identity make apparent the repressive and restrictive consequences that views of identity as fixed potentially hold. They thereby also offer opportunities for personal lib­ eration and social reorganization around principles of relational respon­ sibilities (cf. Benhabib, 1992; Ricoeur, 1992; Taylor, 1989). As Judith Butler (1990) has put this: Paradoxically, the reconceptualization of identity as an effect, that is, as produced or generated, opens up possibilities of “agency” that are insidiously foreclosed by positions that take identity categories as foundational and fixed. For an identity to be an effect means that it is neither fatally determined nor fully artificial and arbi­ trary. ... Construction is not opposed to agency; it is the necessary scene of agency, the very terms in which agency is articulated and becomes culturally intelligible, (p. 147)

The chapters by Lorraine, Presnell and Deetz, and Carr examine situa­ tions that perpetuate notions of identity as fixed and singular and resist interactive understandings of identity. Each chapter argues for the liber­ ating potential that interactive views of identity offer. Tamsin Lorraine examines how efforts to achieve a coherent social and personal sense of self when presented with an ambiguously sexed body are consistently undermined. She does this through the analysis of

14

Interaction and Identity

the diary of a nineteenth-century hermaphrodite, Herculine Barbin, a person who eventually committed suicide. Influenced by the work of Judith Butler, which she further extends by drawing on the psychoana­ lytic writings of Jacques Lacan, Lorraine examines the conflicting dis­ courses of self and body contained within this diary. Lorraine argues through her analysis that taken-for-granted views, of not only gender but of sexual assignment as foundational to experienced identity, are restricting and indeed enslaving. Her analysis suggests how social ex­ pectations of fixed sex assignment may have led to Barbin’s suicide after years of torment at the inability to fix her/himself socially and psy­ chologically. As she concludes ‘“ pulling off’ the performance of a self may involve rendering the chaotic material of living into some kind of coherence. That such coherence is often made impossible by the catego­ ries to which we are subjected and to which we subject ourselves, makes it all the more important to open up spaces for radical self-remaking.” Recognition that categories of meaning are ideologically valenced is becoming increasingly apparent and accepted. This is particularly so with regard to the category of gender. Peggy Carr examines the rel­ evance of this realization in the “self-remaking” context of psychoana­ lytic theory and practice, which she argues has been want to ignore these constraints. Specifically, her chapter is concerned with the constraints that unacknowledged native gender ideology impose on the therapeutic process and how these constraints might be removed. She approaches this through the analysis of two “psychoanalytic encounters,” examin­ ing in one a continuously reproduced pattern of self-other representa­ tion and in the second a brief therapeutic moment within which appreciation of otherness and the difficult struggle to achieve this are concurrently revealed. Through the analysis of these encounters she il­ lustrates the centrality of the theoretical identity of the psychoanalyst for both patient and therapist. My identity, psychoanalytic and feminist, may be colored by the stereo­ typy of superficial knowledge or internal conflict but it is focal. And this theoreti­ cal identity, constructed within and against ideologically loaded prescriptions of psychoanalytic technique or up to the minute feminist conceptualizations constrain my interventions.

Native and privileged sociocultural categories impose models of co­ herence onto people, as Lorraine and Carr make apparent, and thereby front barriers to interactive conceptualizations of a coherent self. Cat­

From Information and Behavior to Interaction and Identity

15

egories of regard for others that individuals evolve through their rela­ tional histories provide another source of such barriers. This is the focus of the chapter provided by Mick Presnell and Stan Deetz (chapter 14). They examine the potential role played by remembered images, what they refer to as interpersonal icons, in the development of conversation and relationship. These icons once triggered, they argue, preclude the possibilities of productive conversation in that they fix perceptions of relational partners and thereby override or block possibilities of “see­ ing” the otherness of the other. Like Lorraine, Presnell and Deetz draw heavily on Lacan to theoretically ground the process by which such in­ terpersonal icons freeze relational interactivity and how this type of com­ municative blockage might be undone—through the “talking cure” of unbridled conversation. This requires recognition that these interper­ sonal icons are not to be regarded psychologically, but as elements of political struggle. “The difficulty in doing this,” they write, “is not just a problem in theory but is the central problem for the resolution of the actual conflict, especially within a society that routinely provides per­ sonal rather than interpersonal accounts.”

Remaking Identity Interactionally The chapters by Lorraine, Carr, and Presnell and Deetz each offer a perspective on the remediation of conversational blockage. All three chapters strongly suggest that interactive conceptualizations of identity are healthier and more productive than essentialist notions. This theme is further developed in the final set of five chapters by Belay, Kim, For­ ward and Scheerhom, Wittenstrom, and Kleinberg (chapters 15-19, this volume). Each of these chapters explores opportunities and approaches to reconceptualizing or remaking identity within the contemporary glo­ bal order. What this entails in each case is a stepping past assumptions of a self as enduring and fixed to a view of self as defined through the other and of the self as a multiplicity rather than a unity. In certainly one of the more timely and innovative chapters in this volume (chapter 15), Getinet Belay examines “the challenges and op­ portunities which globalization presents to the construction and nego­ tiation of cultural identities” with an aim of developing “an interactionist conceptual framework for the interpretation of identity shifts in the age of globalization.” Toward this end, Belay first identifies a set of consti­

16

Interaction and Identity

tutive features defining cultural identity and follows this with the intro­ duction of a conceptual vocabulary for discussing cultural identities. Through these moves Belay aims to offer coherence and consistency for scholarly discussion of cultural identity, discussions which he argues have largely failed to appreciate the arbitrariness of such concepts as nationality, ethnicity, race, gender, social class, and the like. He next examines and critiques three interpretations of cultural change in re­ sponse to globalization, these being the development of: (a) uniform human consciousness; (b) hegemony and homogenization; and (c) world metaculture. This is followed by an interactionist approach offered as an alternative interpretation within which globalization is regarded as an historically emergent “interactional context.” As Belay puts it, within this interactional context, “the problem that both individuals and com­ munities face today in interactional negotiations of identity is no longer merely how to handle the otherness of culturally Others, but also how to handle the multiplicity o f one's own cultural Self ” Belay’s chapter is followed by that of Yung Kim, who also concerns herself with the issue of cultural identity. She, however, suggests the concept of “intercultural identity” as an alternative to cultural identity because it “conjoins and integrates, rather than separates and divides.” Whereas Belay focuses on the identification of parameters for discuss­ ing personal agency, Kim makes this her primary concern. She views intercultural identity as an adaptive and developmental course whereby individuals extend past the boundaries of identity provided by any given culture. She states this quite clearly when she writes, “as the ‘old’ per­ son breaks up, new cultural knowledge, attitudes, and behavioral ele­ ments are assimilated into an enactment of growth—an emergent ‘new’ person at a higher level of integration.” Kim illustrates and traces the development of “new” persons with the assistance of autobiographical sketches of individuals who have traversed this course of adaptation. The inescapability of multicultural experience and multicultural mem­ bership in this age of globalization makes theoretical and pragmatic con­ cerns with the negotiation and remaking of identity central to contemporary discourse. The membership demands of modem organi­ zations in which people are employed in this age of globalization have similarly brought to the fore issues of identity, as Gordon Forward and Dirk Scheerhom discuss in their chapter (chapter 17). In the modem organization success and mobility, according to Forward and Scheerhom,

From Information and Behavior to Interaction and Identity

17

come with the price that “one conceal real intent and emotion in order to enact a self that reflects the values, priorities, and persona endorsed by a generalized organizational culture.... [T]his results in individual iden­ tities frozen within the various discourses of the corporation in ways that may enable the individual to ‘succeed’ but also stifle competing needs, identities, and interests.” This, they claim, also stifles and de­ prives the organization of the full range of productive potentials of its employees. To move past this state of infelicities, according to Forward and Scheerhom, first requires critical appraisal of the theoretical stance toward and the pragmatic course of “organizational assimilation.” Prob­ lematic in theoretical accounts of the assimilation process is the treat­ ment of individuals and the organization as fixed, stable entities. Individuals are regarded as types that fit into slots within the organiza­ tion. In practice, problematic is the general lack of sophisticated con­ cern with assimilation process and the development of simplistic one-track socialization processes that have as their goal organizational conformity. Denied is the inherent instability and interactivity of indi­ viduals and organizations, and thereby denial of identity as itself a type of work. “Identity work” involves the interaction of the individual and organization such that assimilation is not an entry process but an ongo­ ing process in which identity is not achieved but continuously emergent. Through such a process commitment and responsibility come to anchor the relationship between the employee and the organization as each re­ alizes the mutual contingency of their identity. Work certainly represents one of the primary sites and processes within and by which individuals achieve a sense of identity. This clearly is what motivates Forward and Scheerhom’s attention to “identity work” within the context of employment in modem organizations. Attention to the issue of identity work is, however, incomplete if the very concept of work is not itself problematized. Feminist scholars, in particular, have done precisely this in their challenge of work as a descriptive category of human activity. Central to this challenge is the claim that the “tradi­ tional” concept work is implicitly gendered and inseparable from the distinction between the public and the domestic sphere (Okin, 1989). Men “work” within the public sphere while women “care” within the domestic sphere. Certainly this equation has begun to erode as work in the public sphere has become a normative experience for women. How­ ever, it is clear that what erosion has occurred is largely superficial as

18

Interaction and Identity

women continue to find themselves responsible for the “not-work” ac­ tivities of the domestic sphere, what Hochschild (1989) has called the “second shift.” Additionally, if work represents an important influence on the formation and transformation of identity, and if work is equated with the traditionally male-dominated public sphere, it is then important to recognize that the impact on identity of work for females in the public sphere is defined and valued in terms of traditional male experience. Writing from within this context, Kim Wittenstrom (chapter 18) ex­ amines the theoretical complexities involved in discussions of women’s identification through their activities. In particular she critiques two prominent theoretical efforts to reconceptualize women’s activities of­ fered by feminist scholars. The first argues that women’s activities within the domestic sphere are no less work than are the activities of men in the public sphere. The second advocates looking at women’s own understanding of their activities rather than imposing on their activities the category work. Both approaches, Wittenstrom suggests, are problematic in that they presuppose a fixed, singular identification of women’s activities: in the first as unacknowledged work and in the second as work of a different kind. In their stead, Wittenstrom pro­ poses an alternative, interactive conceptualization of women’s activi­ ties informed by recent developments in the study of cognitive categorization. She develops this by examining family day care as an activity (of primarily women) that traverses the boundaries of the pub­ lic and domestic sphere and thereby makes apparent the dynamic, polysemic, and situated qualities of the identification women achieve through their activities. In this way Wittenstrom offers empirical grounding for the development of an interactional perspective on the remaking of identity in contemporary society. The final chapter in this section and volume (chapter 19), written by Ira Kleinberg, examines back-of-the-book indexes within the con­ text of the interaction and identity theme in general, and the remaking identity interactionally theme of this section in particular. In a very real sense a back-of-the-book index may be regarded as “indexical” of a book’s identity. While readers of this volume no doubt have had experiences of better and worse end-of-the-book indexes, it is unlikely that many have considered the theoretical assumptions involved in their construction (and use), let alone viewed the index as an expression of identity. How people make use of indexes and how this relates to the

From Information and Behavior to Interaction and Identity

19

theoretical underpinnings guiding the construction of an index has re­ mained largely unexplored. It is these types of concerns that Kleinberg addresses through the construction of and reflection upon the index to this volume. Final Thoughts Little has been said in this introduction (or in the contributions to this volume) about new information technologies and the information revo­ lution in relation to the issue of identity. This is not meant to deny that interest in issues of identity has been stimulated by the profound influ­ ence that new information technologies have had on lived experience. To the contrary, it is clearly apparent that the fluidity of movement and the immediacy of diversity that the technologically resultant shrinking of time and space axes has introduced (e.g., Gergen, 1991) is of enor­ mous relevance for issues of identity. However, to have focused on in­ formation and information technologies in relationship to identity would have obscured the centrality of communication as a constitutive process for understanding contemporary interest in identity. It would also have obscured the fact that the “transdisciplinary” in­ terest in identity as communicatively constituted has apparently evolved by and large independently of the “interdisciplinary” interest in infor­ mation. The distinction between transdisciplinarity and interdisciplinarity alluded to here is important. Interdisciplinarity is typically invoked when a phenomenon, such as information, is approached from multiple (disci­ plinary) perspectives. In contrast, the concept of transdisciplinarity is introduced to describe new ways of viewing phenomena that are not collaboratively or disciplinary motivated. To put this another way, interdisciplinarity involves conscious effort, transdisciplinarity reveals emergent consciousness. The emergent consciousness that is at the core of interest in identity as communicatively constituted is based in the recognition of the repressive potential, if not inevitability, of essentialist ways of knowing. I hope that the thoughts contained in this introduction and the pa­ pers included in this volume will stimulate further discussion of the relationship between communication and information and the relevance of this relationship for the lived experience and imagined possibilities of people in interaction.

20

Interaction and Identity

A uthor Note I wish to express my sincerest appreciation to my colleague Brent Ruben, the founder and series editor of Information and Behavior, for inviting me to develop this volume and for the support and guidance he has extended me through the years. I would also like to thank Julie Billingsley, Stan Deetz, Michelle Dillon, Michael Huspek, Robert Kubey, Leah Lievrouw, Jenny Mandelbaum, Valerie Manusov, and Bill Solomon for assisting with the review of manuscripts; Ira Kleinberg, Shylaja Nukala, and Raul Goyo-Shields for assisting in the prepara­ tion of the manuscript; and especially my wife, Peggy Carr, for mak­ ing a climate of intellectual stimulation and desire possible and only sensible in relationship. Notes 1. Translations of some of the chapters of the Natural History o f an Interview, in­ cluding the Bateson chapter, have been published in French (Winken, 1981) and Spanish (McQuown, 1983). However, English publications have not previously appeared. 2. An alternative approach to data generation would be to examine naturally occur­ ring talk for instances where a person commented on and/or made efforts to undo their prior talk treating that prior talk as if it constituted a failing or success of the self (see Goffman, 1981; Mokros & Lievrouw, 1991).

References Benhabib, S. (1992). Situating the self: Gender, community and postmodernism in contemporary ethics. New York: Routledge. Bourdieu, P. (1977). Outline o f a theory o f practice. Cambridge: Cambridge Univer­ sity Press. Bruner, J. (1990). Acts o f meaning. Cambridge: Cambridge University Press. Butler, J. (1990). Gender trouble: Feminism and the subversion o f identity. New York: Routledge. ______. (1993). Bodies that matter: On the discursive limits o f “sex. ” New York: Routledge. Capella, J. N. (1981). Mutual influence in expressive behavior: Adult-adult and in­ fant-adult dyadic interaction, Psychological Bulletin, 89, 101-32. Chames, L. (1993). Notorious identity: Materializing the subject in Shakespeare. Cam­ bridge, Mass.: Harvard University Press. Deetz, S. A. (1994). Future of the discipline: The challenges, the research, and the social contribution. In S. A. Deetz (ed.), Communication yearbook 17. Thousand Oaks, Calif.: Sage Publications, 565-600. Duncan, S. D., Fiske, D. W., Denny, R., Kanki, B., & Mokros, H. B. (1985). Interac­ tion structure and strategy. Cambridge: Cambridge University Press.

From Information and Behavior to Interaction and Identity

21

Erikson, E. (1980). Identity and the life cycle. New York: W. W. Norton Co. Fitzgerald, T. K. (1993). Metaphors o f identity: A culture-communication dialogue. Albany, N.Y.: State University of New York Press. Garfinkel, H. (1967). Studies in ethnomethodology. Englewood Cliffs, N.J.: PrenticeHall. Gergen, K. J. (1991). The saturated self: Dilemmas o f identity in contemporary life. New York: Basic Books. Giddens, A. (1984). The constitution o f society. Cambridge: Polity Press. ______. (1991). Modernity and self-identity: Self and society in the late modern age. Stanford, Calif.: Stanford University Press. Goffman, E. (1981). Forms o f talk. Philadelphia, Pa.: University of Pennsylvania Press. Hochschild, A. (1989). The second shift. New York: Viking. Hymes, D. (1962). The ethnography of speaking. In T. Galdwin & W. C. Sturtevant (eds.), Anthropology and human behavior. Washington, D.C.: American Anthropo­ logical Association, 13-53. Keith, M. & Pile, S. (eds.). (1993). Place and the politics o f identity. New York: Routledge. Kenny, D. A. & Malloy, T. E. (1988). Partner effects in social interaction, Journal o f Nonverbal Behavior, 12, 34-57. Krippendorf, K. (1993). Information, information society, and some Marxian proposi­ tions. In J. R. Schement & B. D. Ruben (eds.), Between communication and infor­ mation: Information and behavior, volume 4. New Brunswick, N.J.: Transactions Press, 487-522. Lorraine, T. E. (1990). Gender, identity, and the production o f meaning. Boulder, Colo.: Westview Press. McCall, G. J. & Simmons, J. L. (1966). Identities and interactions. New York: Free Press. McQuown, N. A. (ed.). (1971). The natural history o f an interview. Microfilm Collec­ tion of Manuscripts on Cultural Anthropology. Chicago: University of Chicago, Joseph Regenstein Library. ______. (ed.). (1983). El microanalisis de entrevistas: Los metodos de la historia natural aplicados a la investigacion de la sociedad, de la cultura y de la Personalidad. Mexico City: Universidad Nacional Autonoma de Mexico. Mead, G. H. (1934). Mind, self and society. Chicago: University of Chicago Press. Mokros, H. B. & Lievrouw, L. (1991). Communication-information relationship in self-representation: Suicide notes and academic research narratives. Knowledge, 12, 389-405. Okin, S. M. (1989). Justice, gender, and the family. New York: Basic Books. Pearce, W. B. (1994). Recovering agency. In S. A. Deetz (ed.), Communication year­ book 17. Thousand Oaks, Calif.: Sage Publications, 34-41. Ricoeur, P. (1992). Oneself as another. Chicago: University of Chicago Press. Rowland, R. (1988). Woman herself: A transdisciplinary perspective on women *s iden­ tity. New York: Oxford University Press. Ruben, B. D. & Lievrouw, L. A. (1989). Mediation, information, and communication: Information and behavior, volume 3. New Brunswick, N.J.: Transaction Publishers. Sacks, H. (1984). Notes on methodology. In J. M. Atkinson & J. Heritage (eds.), Struc­ tures o f social action: Studies in conversation analysis. Cambridge: Cambridge University Press, 21-27. Sahlins, M. (1976). Culture and practical reason. Chicago: University of Chicago Press.

22

Interaction and Identity

Scheff, T. J. (1990). Microsociology: Discourse, emotion and social structure. Chi­ cago: University of Chicago Press. Scheff, T. J. & Retzinger, S. (1991). Emotions and violence. Lexington, Ky.: Lexing­ ton Books. Schement, J. R. & Ruben, B. D. (eds.). Between communication and information: Information and behavior, volume 4. New Brunswick, N.J.: Transaction Publish­ ers. Shotter, J. & Gergen, K. J. (eds.). (1989). Texts o f identity. Newbury Park, Calif.: Sage Publications. ______. (1994). Social construction: Knowledge, self, others, and continuing the con­ versation. In S. A. Deetz (ed.), Communication yearbook 17. Thousand Oaks, Ca­ lif.: Sage Publications, 3-33. Taylor, C. (1989). Sources o f the self: The making o f modern identity. Cambridge, Mass.: Harvard University Press. Winken, Y. (1981). La nouvelle communication. Paris: Editions du Seuil. White, H. C. (1992). Identity and control: A structural theory o f social action. Princeton, N.J.: Princeton University Press.

Parti Issues of Theory and Method in Interactional Study of Identity

2 Part/Whole Discovery: Stages of Inquiry Thomas J. Scheff

This chapter outlines the idea o f part/whole analysis, relating the smallest parts to the largest wholes. This procedure stands in stark contrast to the kind o f special­ ization that characterizes most contemporary science and scholarship. It leads to two new steps in inquiry: morphology the detailed, microscopic study o f single specimens, andpart/whole theories, which are interdisciplinary and interlevel. Such theories require the specification o f abstract concepts and the causal linkages be­ tween them.

In this paper,11 propose two new stages of inquiry, to form a bridge between quantitative and qualitative methods: morphology and part/ whole theory. These stages depend upon and extend the scientific work of Goethe and Wittgenstein’s approach to philosophy. I suggest that the separation between qualitative and quantitative work is a product of excessive specialization, and give examples of part/whole research that unifies disciplines, methods, and levels of analysis. I begin with a review of Goethe’s approach to inquiry, which is little known in mod­ em science.

Goethe’s Approach to Inquiry During his life time, Goethe (1749-1832) considered his scientific work to be as important as his poetry. Surprisingly, given the little atten­ tion that is shown to Goethe’s science today, it was highly esteemed by the most eminent scientists of his day. Although his work is seen as important to many contemporary sciences, I will review only three fields: his contributions to botany and vertebratology, and his theory of color. 25

26

Interaction and Identity

There is considerable controversy in the evaluation of Goethe’s sci­ entific contributions, with opinion polarized, for the most part, between those like Rudolf Steiner (1926), who argued that Goethe was one of the greatest scientists who ever lived, and the more common opinion, espe­ cially among contemporary scientists, which discounts his scientific work. More balanced views can be found in Magnus (1949), Vietor (1950) and Amrine, Zucker, and Wheeler (1987). Most historians of science agree that Goethe made original scientific contributions, both theoretical and empirical, in many fields, but espe­ cially in botany, optics, osteology, mineralogy, philology, translation, geology, and meteorology. He is usually considered the founder of both the comparative morphology of plants and of comparative anatomy in zoology. His work on plants is of extraordinary interest because of the combination of bold ideas and detailed observations. His discoveries in osteology most clearly illustrate this approach. In 1784, at the age of thirty-five, Goethe finished his first scientific paper, an extremely detailed study of the intermaxillary (the section of the upper jaw that carries the incisors) in humans and other vertebrates. His study illustrates the approach that was to characterize all of his sci­ entific work: viewing a problem in a way that combines the largest out­ look with the diligent pursuit of the smallest detail. He started on the track of the intermaxillary in response to what was considered the key finding of the then current work in osteology; that this bone was found in all vertebrates, but with one great exception—human beings. This “finding” offended one of Goethe’s basic beliefs—his sense of the unity of nature. Since this was almost a century before Darwin, Goethe’s idea was shockingly heretical. Instead of writing a specula­ tive essay, however, Goethe began an extensive empirical investiga­ tion. He made extremely precise and detailed drawings of the skulls of animals and humans. His work showed that although the outlines of the human intermaxillary were usually subtle, it had exactly the same form as the more prominent ones in animals, such as the walrus and the elephant. At first his report was uniformly rebuffed, much to Goethe’s chagrin. The work of the osteologists was influenced by the cultural frame, which supposed a special status for humans, separate from the rest of nature. Since the intermaxillary was usually vestigial, they were unable to see it. The generative idea for Goethe’s labor was the opposite; since he

Part/Whole Discovery: Stages of Inquiry

27

presumed that the bone must be present, it enabled him to find it, and to persist even though in some cases the outlines were barely visible. By 1791, most of the osteologists had accepted Goethe’s findings. Goethe’s work in botany was much more wide ranging than any of this other investigations. As in his work in skulls, he made extremely detailed drawings of the form of thousands of plants. However, unlike Linneaus, who only considered form, Goethe’s studies concerned the relationship between form and function. From his observations of the immense variety that plants took, he sought to understand the generic or universal plant; the basic relationship between form and function in all plants. In some ways, his botanic work intimates a theory of evolution, since it considers the relationship between the development of the form of plants and their environment. Like his botany, Goethe’s theory of color is currently little under­ stood. Once again, this research is based on close empirical observa­ tions. In this case, his observations were of the afterimages after viewing colors. Goethe’s science was based on what he called the morphological method: the close observation, in many instances, of the minutia that differentiated one instance from another. These minutia, he thought, pro­ vided the starting point for all science. At the heart of his science, however, was a more complex notion he called gestalten, a word with no exact equivalent in English. It means shape or pattern, but in Goethe’s usage, he emphasized a connotation, that of a complete pattern. What Goethe meant by complete pattern con­ cerned the object under study, its environment, and the relationship be­ tween the object and the environment. This concept is important for the present discussion because it can be expressed in terms of part/whole relationships, which Goethe occasionally discussed explicitly. For ex­ ample, here is one of his formulations, taken from his “Metamorphosis of plants” (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 why living creatures, even the most restricted, have something about them that we can­ not quite grasp and have to describe as infinite or partaking of infinity.

Goethe’s usage is inconsistent, however. He never stated a clear part/ whole formulation and its relationship to morphology and to the discov­ ery of gestalten (complete patterns).

28

Interaction and Identity

The idea of gestalten is further diluted in Kohler’s (1929) formula­ tions. His discussion does not emphasize the idea of complete patterns, as Goethe’s did. Instead he defined gestalten as organized patterns. From my point of view, there was one serious flaw in Goethe’s approach—his prejudice against abstract analysis, mathematics, and the need for veri­ fication. This flaw is apparent in his peremptory dismissal of Newton’s theory of light. Goethe never attempted to understand it, since it was not based on morphology, but was developed as an abstract theory. Goethe reasoned that since Newton’s theory seemed to contradict his own, New­ ton must be in error. Newton and Goethe were not in conflict because their theories refer to different data and, therefore, different aspects of light. Newton’s theo­ ries concern the physical basis of light, Goethe’s its human reception or phenom enology. W hat differentiates Goethe from m ost phenomenologists is that he was not content to point out the limitations of an analytical scheme; he also did a careful empirical study. Goethe’s botany, and all of his other science, has this dual character: it was both theoreti­ cal and empirical. In my judgment, that is the secret of his astounding success as a lone scientist.

Wittgenstein’s Approach to Inquiry The work of Ludwig Wittgenstein is widely known and admired within the discipline of philosophy, but little known outside. Many philoso­ phers think that he was easily the greatest philosopher in this century, and that his work established the foundations of modem philosophy. Wittgenstein’s approach has some similarities with Goethe’s, but also some key differences. I will introduce these ideas by first recounting some aspects of Wittgenstein’s life. My account is based on several bi­ ographies, but particularly on Monk’s (1990). As a young man just out of college in Austria, Wittgenstein came to Cambridge to study with Bertrand Russell, at Russell’s invitation. At the time of his arrival, Russell had just completed (with Alfred Whitehead) what he and Whitehead considered to be a complete foundation for mod­ em mathematics (Principia Mathematical, 1910). In this work Russell thought he and his co-author had formulated, as a series of propositions, a structure that would serve as the basis of all mathematics. Although only a beginning graduate student, Wittgenstein argued with Russell that the book was not what the authors thought it was. He con­

Part/Whole Discovery: Stages of Inquiry

29

vinced Russell that the formulation was not a foundation, but only one more conventional mathematical system. This exchange had the effect of putting Wittgenstein in the role of Russell’s teacher, rather than vice versa. Russell was extraordinarily receptive to Wittgenstein’s teachings, so long as the latter was attacking Russell’s problem, the foundations of mathematics. For may years, with Russell’s encouragement, Wittgenstein sought to correct the flaws in Principia. The great shift in Wittgenstein’s thinking occurred when he realized that the problem should be much broader than the one formulated by Russell: not just the foundation of math­ ematics, but of all inquiry. Most of Wittgenstein’s life was spent wres­ tling with this larger problem. The shift to a problem larger than Russell’s was fatal to their relationship. When Russell realized the implications of Wittgenstein’s shift, he rejected him and his work. In some ways Wittgenstein’s later work is an application and exten­ sion of the approach to inquiry first developed by Goethe in his botanic studies. It is a morphological approach, as against the analytic one that is the basis for abstract theories. It focuses upon minute differences be­ tween phenomena, just as abstract theory glosses such differences. Ab­ stract theory can only be successful when it grows out o f knowledge o f minutia: this knowledge enables the researcher to distinguish between the figure (a dynamic system) and the ground (all the minutia that are irrelevant to the dynamic system). Although Wittgenstein had an instinct toward morphology, he was inhibited by his training as a philosopher to ever consider doing the kinds of studies with sentences that Goethe did with plants. I think that this inhibition is the principal reason that Wittgenstein was unable to formulate the underlying principles of his approach. In fact, he denied the possibility of such a formulation. This denial is particularly clear in his simile comparing invented and natural languages: [Chemistry, calculus, and similar invented languages] are suburbs of our language. (And how many houses or streets does it take before a town begins to be a town?) Our language can be seen as an ancient.city: a maze of little streets and squares, of old and new houses, and of houses with additions from various periods; and this surrounded by a multitude of new boroughs with straight regular streets and uni­ form houses. (1968, p. 8; numbered remark 18)

In the simile of the side roads and the highway, Wittgenstein (1968) takes this image still further: “In the actual usage of expressions we make detours, we go by side roads. We see the straight highway before

30

Interaction and Identity

us, but of course we cannot use it, because it is permanently closed” (p. 127; numbered remark 426). In my view, however, Wittgenstein overreacted to the excesses of the analytic method, somewhat as Goethe did. The trick is to combine the strengths of morphology and analysis. The highway may be usually, but not permanently, closed if we leam to use this trick. Newton, Darwin, Einstein, and Watson and Crick found open highways. How did they do it? They combined morphology and analysis, at best, within a single per­ son (Darwin). Newton’s Law of Universal Gravitation depended upon Kepler’s discovery of the orbit of Mars, which in turn depended entirely on Brahe’s precise sightings of Mars. In a similar way, Darwin’s theory of evolution was generated by his painstaking and voluminous observa­ tions of the minutia that differentiates members of the same species in different locales. Einstein depended on the minute discrepancies between recent physical findings and classical physics to overthrow the latter. Watson and Crick combined the precise observations of researchers like Rosalind Franklin and Erwin Chargaff with the abstract theory of Linus Pauling. Advances in science occur most rapidly when morphology and theory are both practiced by a single person, as in the case of Darwin, and to a lesser extent, Watson. Inquiry develops best if there is rapid movement between theory and data, between deduction and induction (Charles Peirce called this move­ ment abduction). In actual practice, this movement means using topdown and bottom-up strategies, rather than the current division of labor between theorists and researchers. The discussion of a combined method so far has been limited to physi­ cal science. In the move into human science, Wittgenstein’s use of lan­ guage games and metaphor becomes supremely important. I think that Wittgenstein’s fury and scorn for modem philosophical analysis was based on an instinctual feeling that he was never able to articulate: that analysis that is oversimplified is premature and, therefore, virtually useless. De­ tailed observations of minutia are necessary to generate theories suffi­ ciently complex to account for the important differences between instances.

Linking Abstract Concepts to Empirical Data Durkheim (1897) sought to explain the constancy of suicide rates in religious and other groups in terms of the types of social relationships in

Part/Whole Discovery: Stages of Inquiry

31

each group. He showed, for example, that rates of suicide in predomi­ nantly Catholic regions in Europe were consistently and uniformly lower than rates in predominantly Protestant regions. He also considered varia­ tions in rates between nations and between occupations. These observa­ tions led him to formulate a cultural theory of suicide: differences in culture lead to variations in suicide rates, because culture influences the types of social relationships that prevail in a group. At the core of Durkheim’s theory of suicide is the distinction he made between two kinds of cultures. On the one hand, there are cultures that are suicidogenic because they give too much prominence to individuals (as against the group), characteristic of modem industrial societies. In his formulation, such cultures lead to egoistic or anomic suicide. On the other hand, there are cultures that give too much prominence to the group, such as the cultures of small traditional societies, which leads to what he called altruistic or fatalistic suicide. Durkheim sought to explain high suicide rates in traditional groups (too little emphasis on individuals), leading to culturally prescribed forms of suicide (as in hari kari in Japan), and in groups promoting change and innovations (too much emphasis on the individual), like the most ascetic forms of Protestantism. Although Durkheim did not use the terms solidarity and alienation, his analysis is clearly focused on what would today be called social integration, of being either too loosely or too tightly bound to the group. In his analysis, Durkheim took two important steps: he formulated an abstract theory of suicide, and he gathered data that supported his theory. In combining these two steps into a single study, he invented social sci­ ence. It seems to me, however, that social science composed of only these two steps is not the last word. Durkheim’s explanations of the link between his abstract concepts and his data are phrased in terms that are quite vague. The careful description of the process that links abstract concepts to empirical data would seem to be necessary for the formula­ tion of a complete theory. Explicit models that link concepts (names for precisely defined ab­ stract classes) to actual data are characteristic of successful sciences. The periodic table, which classifies the chemical elements, is an ex­ ample: its form is finished because it grows organically out of a micromodel of the atom. On the other hand, Linnaen classification in biology is a continuing embarrassment, because it is still an arbitrary

32

Interaction and Identity

taxonomy: there is no underlying micromodel of the processes that lead to speciation to give it form. Durkheim’s classification of the types of suicide is Linnean since he didn’t specify a precise model of the process through which types of social relationships at the microlevel lead to suicide rates at the macrolevel. His theory gives a bird’s eye view of society, but suicide rates are discrete events. How can the two levels be connected? The route proposed here is that we may be able to link micro- and macrolevels in social science by specifying precise models of causation at the atomic level, the actual events that make up social relationships. This method provides a morphological basis for what Giddens (1984) has called “instantiation,” and Geertz’s (1983) “thick description” of a single case (Scheff, 1986). Glaser and Strauss (1967) have proposed an intermediate step toward grounding general theories. They show that most of these theories were generated from a single setting, which localizes them. Their method is to apply the theory in a variety of settings in order to seek general under­ standing. It is possible however, that theory generation requires an ear­ lier step, the microscopic examination of single cases, the morphological method, the detection of minute differences that both general theory and Glaser and Strauss’s grounded theory are tempted to overlook. I am not arguing, however, that macro theories (such as Durkheim’s) are unnecessary. I propose that theory construction at each level is equally necessary. By moving back and forth rapidly from the top down and from the bottom up, it may be possible to advance understanding much more quickly than isolated efforts in each direction. Elsewhere (Scheff, 1990; 1994) I have called this the method of parts and wholes. My own interpretation of morphology brings what may be a new idea to the procedure: the most successful inquiries have involved pro­ ceeding from the smallest possible minutia up to the largest possible theory. This combination of small and large is particularly exempli­ fied in the work of Darwin. Since he was not a microbiologist, he took the smallest units of observation that were available to him, the tiny external variations in appearance and behavior among members of the same species. His thought, however, made use of the very largest ideas of system that were available to him, especially the ideas of Malthus. Darwin’s constructed his theory by moving from the bottom up and from the top down.

Part/Whole Discovery: Stages of Inquiry

33

Wittgenstein’s use of language games, humor, and metaphor was a way of hinting at the extraordinary complexity of human societies, and availing himself this complexity. Many years before the failure of attempts at automated language translation, he realized that ordinary languages were open systems: all ordinary human expressions, each word, sentence, or paragraph, is profoundly ambiguous unless inter­ preted in context. Since context contains voluminous detail, contextual interpretations call upon an extraordinary file of knowledge, knowledge of the whole culture in which an expression occurs (Wittgenstein’s “mastery of prac­ tice”), and an extraordinary method of interpretation, the part/whole method (Scheff, 1990). All interpretations of natural language expressions require an enormous file of everything that happened before, during, and after the expression (Mannheim called it the prospective-retrospec­ tive method of understanding), as well as what didn’t happen, but could have (counterfactuals). The method of understanding naturally occurring human expressions seems to require the extremely large manifold of paths of association that is generated when there are a very large number of items (such as words and gestures in a natural language) to be interpreted, and each and every one of these items is ambiguous (i.e., it has more than one meaning). I call this method total association, to distinguish it from logical connections, which are based on unambiguous items, and from Freud’s free associa­ tion, which excludes logical connections (Scheff, 1990). The part/whole method requires the inclusion of the smallest and the largest units, and the steps between them. In the human sciences, the small­ est units that have been carefully explored are the words and gestures of naturally occurring discourse. In the various kinds of analysis of natural language in conventional scholarship, however, most researchers do not use the part/whole method; they restrict themselves to the parts.

The Part/Whole Ladder So far I have argued that microanalysis of the smallest parts of a system is as necessary as macroanalysis in terms of abstract concepts for understanding social structure and process. To illustrate this idea, it is necessary to return to the concept of part/whole relationships. In an open system, any part implies a larger whole, which is in turn part of a

34

Interaction and Identity

still larger whole, and so on, up the ladder. Applied to human societies, this idea suggests a movement back and forth between small concrete parts, and even larger abstract wholes. Concrete Level 1. Single words and gestures 2. Sentences 3. Exchanges 4. Conversations 5. Relationship o f the two parties (all their conversations) 6. Life histories o f the two parties Societal Level 7. All relationships of their type: i.e., therapist-patient, man-woman, etc. 8. The structure o f the host society: all relationships 9. The history and future o f the host civilization 10. The history and destiny o f the human species

Practical intelligence in the life world appears to involve abduction, that is, the rapid, effortless shuttling up and down this ladder. The system can be visualized as Chinese boxes, each box containing a smaller one, and nested within a larger one. Indeed the context of nested contexts is crucial for understanding discourse. The distinction between topic and relationship hints at this larger sys­ tem: the topic is at level two, since it involves sentences, the relation­ ship, at level five. However, all levels are implied in the actual understanding and practice of discourse. The process is too awkward to describe in explicit language, but it takes place constantly, effortlessly, and instantaneously in discourse. Contemporary scholars and scientists seem to have difficulty visual­ izing part/whole relationships. In the current division of labor, the or­ ganic connection between part and whole is lost in the division of intellectual labor: theorists deal with wholes, but not parts, and researchers deal with parts, but not wholes. The study of social and cultural systems is fragmented among disciplines, theories, methods, and schools of thought. The contributors to an interdisciplinary colloquium on parts and wholes (Lemer, 1963) seemed to have had difficulty in even ap­ proaching the topic. Only two of the papers convey some sense of the relationship: Roman Jacobson (natural language as an organic system), and I. A. Richards (poems as organic wholes).

Part/Whole Discovery: Stages of Inquiry

35

Even Jacobson and Richards fail to give a minimal sketch, however. Both locate the system in the verbal parts rather than in the social-cul­ tural whole. Richards discusses only the integrity of the poem rather than its relationship to social and psychological process in creator and audience. He sticks to a topic rather than commenting on relationships. This sam e difficulty continues to haunt current discussion. Postmodernists have discovered the ambiguity of expressions: they show that various expressions are “undecidable,” and that translations may be “indeterminate. Obviously when the social context is shorn away, as it is in most structuralist interpretation, all expressions be­ come ambiguous. But in context, interpretive decipherment can result in consensual understanding. A clear evocation of the contextuality of understanding can be found in Levine’s examination of sociological theory (1985). By showing the contradictions that result from the attempt to eliminate ambiguity from expressions, he makes a case for the role of intuition (abduction) in un­ derstanding the relationship between parts and wholes. Goffman’s technique of “frame analysis” (1974) also can be used to focus on part/whole process. The frames that he refers to can be seen as contexts in the nested context structure of thought. Goffman, however, limits himself to the analysis of the social aspects of framing, missing the opportunity to connect the social and the psychological. His analysis of “frame breaking” nevertheless suggests an important issue. The fram­ ing structure itself, the particular set of Chinese boxes that is tacitly assumed in a group, however limited, takes on a sacred character. Under these conditions, frame breaking, the use of a different part-whole struc­ ture, may be taken as an affront. The theorist who comes closest to the core of the part/whole approach is C. Wright Mills, with his idea of the sociological imagination (1959). At first glance at his book, one might think that Mills did not advocate the kind of analysis of systems and subsystems proposed here, one in­ volving equity between microworlds and macroworlds. Mills certainly has no such reputation. He is seen rather as an analyst of the macroworld, like other sociologists. A closer look, however, suggests a strong affinity between his view of the sociological imagination and the perspective advocated here. Mills (1959) suggested that the link between personal troubles and public is­ sues, the intersection between biography and history, should be the core

36

Interaction and Identity

of the sociological imagination (pp. 11-20). He defined that imagina­ tion as 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 estab­ lishment; from considerations of the oil industry to studies of contemporary po­ etry. It is the capacity to range from the most impersonal and remote transformations to the most intimate features o f the human self— and to see the relations between the two. (emphasis added, p. 7)

His reference to the psychological perspective in the first phrase seems not to have been merely pro forma . In the next-to-last phrase, he pro­ poses the importance of “the most intimate features of the human self.” Mills (1959) enthusiastically subscribed to the potential of psychoanalysis (pp. 159-60) for understanding these intimate features. Although Mills himself never carried out any of the part/whole studies he seems to be advocating, it is obvious that he saw the potential. To illustrate these abstract ideas, I call upon a part/whole analysis implied in a study of poetic form. Flexing part/whole muscles in her introductory comment, Smith (1968, p. vii) outlines a structure of nested contexts: The study is concerned with how poems end. It grew out of an earlier one that was concerned [1] with how Shakespeare's sonnets both go and end; and although the child has consumed the parent, it testifies to its lineage throughout these pages, where sonnets, Shakespearean and other, will be rather frequently encountered. In my earlier attempts to describe and to some extent account for the strengths and weaknesses of Shakespeare's sonnet endings, I found myself involved at almost every point with more general considerations of poetic structure and with what I finally recognized as a subject in its own right, [2] poetic closure. I also found that, although literary theorists from Aristotle on have been occupied with beginnings, middles, and ends, there had not been (aside from a brief and somewhat whimsical essay by I. A. Richards) any treatment of this subject as such. The questions and problems that pushed outward from sonnet endings to lyric closure in general con­ tinued to move out toward even broader considerations of closure in all [3] litera­ ture, in all [4] art, and finally in all [5] experience. Having bumped into a continent, however, and even having set a flag upon the shore, I realized that I was equipped to explore and chart only a bit of the coastal area. It seemed wise, then, to hold the line at poetic closure.

Smith locates her work within a larger structure. A part/whole ladder with five levels is implied: 1. 2.

Closure in Shakespeare’s sonnets In poetry

Part/Whole Discovery: Stages of Inquiry

3. 4. 5.

37

In literature In art In experience

She puts her particular study at level two, poetry, in a structure of nested contexts: it is broader than just Shakespeare’s sonnets (level one), but not as broad as literature (level three). She does, however, make a num­ ber of significant comments about levels four and five. I will use one of these comments to show both the strength and limitation of her analysis. The use of nested contexts, Chinese boxes, one within the other, in­ volves recursion, as Hofstadter (1980) suggests. This kind of part/whole structure may be a way of giving concrete meaning to the abstract con­ cept of reflexiveness, the kind of self-referencing that gives rise to selfawareness. A discussion like Smith’s is reflexive to the extent that it locates its own argument within a part/whole structure, a total rather than a special frame of reference. In the main body of her study, Smith seems to take an important step in this direction on the very last page. Her main argument has estab­ lished that traditional poetry depends upon a large variety of techniques for closure, for establishing that the poem is ending. These techniques are mostly conventions of rhyme and syntax. She notes that most mod­ ern poetry involves not only closural devices, but suggestions of rebel­ lion against these devices. That is, many modem poets seem to want their poems to appear open ended or continuing, rather than closing with a snap. She locates this tendency not only in poetry, but also in literature and art, particularly in painting and music (e.g., John Cage, serial or random music) as a rebellion not only against closure, but against all structure. In this way, she locates her description of the conflict between poetic closure and anticlosure in a larger context, the conflict between structure and antistructure within art as a whole. Her analysis ends with a very brief reference to a still larger whole. In trying to make sense of the conflict over structure in art, she argues that the traditional structures of poetry, which press toward closure, have a hidden ally that gives them their staying power (p. 271): Poetry is not without a sustaining vitality continuously fed and renewed by its relation to the rather formidable institution of language itself—and that as long as we continue to speak at all, no matter what new uses are made of language, there will remain revelations and delights to be found in the old uses. Poetry ends in many ways, but poetry, I think, has not ended.

38

Interaction and Identity

Her comment implies conflict between social institutions. In seeking freedom for new forms within the social institution of art, poets pit them­ selves against another institution, “the rather formidable institution of language itself.” The implication is that poetry will always press toward closure because it is based on language itself, and language, that is the natural language used in everyday life, will always press toward clo­ sure. (It should be noted that discussion of linguistic closure could be grounded in interactional studies of turn taking.) Her comment suggests the value of an additional part/whole level, higher than four, art, but before the leap to all experience at five, the level of social institutions. Studies like those of Smith might be used to give new meaning to the abstract concept of reflexivity. Her study is reflexive in the sense that she located her particular study within a broad and deep structure of nested contexts. She could have easily avoided that responsibility. Her study would have met the expectations in her genre had she made only passing reference to levels three, four, and no reference at all to the level of institutions, and to level five. Such a study would probably give the alert reader a feeling of claustrophobia, as do more conventional schol­ arly and scientific studies. Overspecialization means the failure to es­ tablish responsible structures of nested contexts, of not exploring the relevant part/whole ladder. Perhaps it was tactful of Smith not to get involved in what Goffman would call ’’framebreaking.” Had she said much more about the conflict between the institutions of art and language, most of the members of her audience, quite content with the conventions of the genre, might have been puzzled. She would have found herself in a “no-persons land” be­ tween two genres, literary interpretation on the one hand, and sociology on the other. Perhaps the line between intellectual creativity and respon­ sibility, on the one hand, and deviance, on the other, is vanishingly thin. This discussion suggests an exact definition of reflexivity; the tra­ versing of complete part/whole structures. To the extent that thought and creative effort travels the entire part/whole ladder, it is freed of its purely local dependence and becomes optimally useful and stimulating. However, as already indicated, it also runs the risk of antagonizing the conventional structures of context, and in this way becoming puzzling and opaque. To understand the significance of an expression in context, the re­ searcher as well as the participant needs to refer it to other levels. For

Part/Whole Discovery: Stages of Inquiry

39

example, to understand the meaning of her lover’s farewell, a woman might observe not only the exact words and gestures at level one, but also compare them to how other lovers might have responded, a counterfactual (level seven). The interpretive decipherment of natural language, to use Steiner’s phrase, requires part/whole thinking. Stages of Inquiry By establishing two new stages of inquiry, it may be possible to rec­ oncile the differences between qualitative and quantitative methods. Most existing qualitative work emphasizes the construction of theory, but dis­ regards the importance of verification. Most quantitative work empha­ sizes verification, largely disregarding the construction of theory. My comparison of the morphological and the analytical suggests that existing qualitative work can be seen as the first step in a four-stage process. The first stage is the eye-witness description and interpretation of the phenomena, as in current qualitative and field work. The last stage is verification, the testing of a carefully stated hypothesis. In between the first and the final stages, however, two new stages may be required. The second stage requires the very precise description of the smallest parts in the system under study, its morphology. The third stage is to relate the smallest and largest parts of the system, that is, the statement of a micro-macro theory. This is the stage in which an abstract theory is stated explicitly. The last stage is to verify hypotheses generated by this theory. The actual chronological order of the first three stages is not important. But it is clear that these three stages, which may be contemporaneous, must precede the final stage, verification. They must also contribute to the possibility of verification, by delineating observable cues or markers for the phenomena they describe. This strategy integrates morphological and analytical approaches: that is, it uses bottom-up (data-driven) and top-down (theory-driven) strate­ gies. My study (Scheff, 1994) of the roots of violence provides an ex­ ample. Retzinger and I (Scheff & Retzinger, 1991) developed a theory of social integration that proposes that destructive violence has its ori­ gins in alienation. My study begins with second-by-second analyses of social bonds in psychotherapy sessions and marital quarrels. I apply this approach to the origins of the First and Second World Wars, interpreting verbatim discourse (such as exchanges of telegrams between the heads

40

Interaction and Identity

of state immediately prior to the war) in their historical context. This study proceeds through only the first three stages of inquiry, so is still unfinished. The next step would be to verify the hypotheses that link protracted conflict to alienation. Discussion This paper is a review and new interpretation of morphology in its relationship to theory. My interpretation leads to a proposal for method in the human sciences. Effective inquiry may require four stages: sub­ jective observation, precise morphology, part-whole theory, and verifi­ cation. This sequence may resolve many of the differences between qualitative and quantitative approaches. As presently practiced, both qualitative and quantitative methods usually preclude intimate knowledge of the fine details of the phe­ nomena being studied. Since most qualitative studies depend on un­ corroborated observation, they usually exclude the fine details that do not fit into the researcher’s viewpoint. Quantitative studies use precise instruments in order to catch some of the details. But the price they pay is to exclude most of the surround. Without close attention to the surrounding details, it is almost impossible for the researcher to learn anything new. The researcher is entrapped in the point of view that pre-existed the investigation. A comparison between Goethe’s approach and Wittgenstein’s illus­ trates this idea. It would seem that Goethe’s science was successful be­ cause he was not inhibited about making precise empirical studies of the smallest details of plants and the other phenomena he studied. Like Dar­ win, he made thousands of detailed drawings of the specimens he col­ lected. His intimate knowledge of these details enabled him to sketch general theories. Although Wittgenstein was a brilliant observer and thinker like Goethe, and an admirer of Goethe’s studies, it seems never to have occurred to him to follow in Goethe’s footsteps as an empirical scientist. Witt­ genstein’s discussions of language provide an example. Although they were wide ranging and original, he followed the same method of inquiry that other professional philosophers use; instead of detailed descriptions of naturally occurring words and gestures, Wittgenstein used hypotheti­ cal examples. His method was qualitative and deductive. I believe that

Part/Whole Discovery: Stages of Inquiry

41

this method is the reason that Wittgenstein was unable to complete any of his studies. His method glosses the smallest parts of the system, as does Glaser and Strauss’s. In the human sciences, specialization of discipline, method and/or theory has severed part/whole connections, and in doing so, has trivialized research designs and findings. Artificial boundaries that define disci­ plines, and theoretical and methodological research domains, make it practically impossible to generate and test general theories. A clear example of such artificial boundaries comes from an incident in the history of mathematics. After his death, Georg Cantor’s work on transfinite sets became the most important body of work that established the foundation of modem mathematics (Burton, 1985). But there was a long delay. During Cantor’s lifetime, his work was virtually unknown. I will briefly review this episode. The story has three main characters: George Cantor, an innovator; Leopold Kronecker, Cantor’s critic; and Goesta Mittag-Leffler, who ini­ tially sponsored and defended Cantor. In 1870, Cantor, a young profes­ sor of mathematics at a small German university (Halle), wrote a long paper in which he began to develop set theory. He sent the paper for comments to Kronecker, a senior professor at the University of Berlin, who had been one of Cantor’s teachers, and was the most eminent math­ ematician in Germany at that time. Kronecker’s response was brief and dismissive. He replied to the effect that he didn’t know what the paper was, but he knew that it was not mathematics. Perhaps, he said, it was philosophy, or even theol­ ogy. Rather than becoming his sponsor, as Cantor had hoped, Kro­ necker became his persecutor. Since Kronecker was the leading light in German mathematics, he was able to see that Cantor found it diffi­ cult to get his work published, and that he was never promoted. Set theory disappeared from German mathematics for almost a hundred years, until it was recognized as a revolutionary discovery by math­ ematicians in England and elsewhere. The only reason Cantor was able to publish at all was that he found a sponsor in a friend from his student days, Mittag-Leffler. They had both studied with Weierstrass. Because Mittag-Leffler was wealthy, he had established his own journal, Acta Mathematica. However, at a critical juncture in Cantor’s career, Mittag-Leffler with­ drew his support. When Cantor submitted his first general statement of

42

Interaction and Identity

set theory to Acta Mathematica in 1885, Mittag-Leffler asked him to withdraw it, suggesting that it would be “re-discovered in a hundred years.” (This prophecy turned out to be fairly accurate; the paper was published in Acta Mathematica in 1970, eighty-five years late). Why did Mittag-Leffler withdraw his support? He was a supporter of other underdogs in addition to Cantor. He not only published the work of Sonya Kovalevsky (another fellow student of Weierstrauss’s), but found a teaching appointment at the University of Stockholm, after she had been denied jobs in all other universities in Europe. She was the first woman to have such an appointment. In Cantor’s case, however, Mittag-Leffler apparently suffered a fail­ ure of nerve. Although Cantor’s work received no support whatever in Germany, it initially got favorable reactions in France. But Hermite, a leader of French mathematics, and his colleagues condemned the new paper as solving no current mathematical problem. Like all the other influential mathematicians of the time, they could not see the revolu­ tionary potential of set theory. Although Cantor complied with Mittag-Leffler’s suggestion, this event embittered him. He withdrew from mathematical research for five years, seeking employment in other areas. He published little more on math­ ematics, save his summary of his theory in 1895 and 1897, even though he lived until 1918. Recognition came after the turn of the century, when Cantor had stopped working. It came from outside, largely from England and the U.S. He was awarded the Sylvester Medal in 1904 by the Royal Society of London, for “the most innovative work in mathematics in 2500 years.” The first textbook on set theory, by William and Grace Young, was pub­ lished in England in 1906. Even so, Cantor’s work was still largely un­ known in Germany. As late as 1908, Cantor complained to the Youngs that his work was not recognized by Germans, “who do not seem to know me, although I have lived among them for 52 years” (Burton, 1985, p. 597). This incident illustrates many of the features of boundary setting. Kronecker did not find weaknesses or errors in set theory; he dismissed it out of hand. His reaction was doctrinaire, because he saw that Cantor’s scheme differed from established practices in the discipline of math­ ematics. It offended his sense of propriety by suggesting avenues of thought in addition to those that had been established.

Part/Whole Discovery: Stages of Inquiry

43

Although Kronecker was a good mathematician in his own field (al­ gebra), he was extraordinarily naive outside of it. One of his recorded statements testifies to his stupidity about epistemic issues: “God made the natural numbers; all else is made by man.” His response is an ex­ treme example of academic fundamentalism, an irrational faith in the status quo of knowledge. In the beginning of mathematics (and all other fields), before it had become a professional discipline, procedures were adopted because they were immediately useful. After a certain time had passed however, these procedures began to take on a sacred, unarguable status; they were in­ stitutionalized as absolutes. There seems to be a tendency for disciplines to become rigidly closed systems, especially within national boundaries. During Cantor’s lifetime, German mathematics was such a system. Pub­ lication of his work occurred only because of influence from outside the system; first from France, and later from England. It would appear that for knowledge to continue to grow, some balance is needed between boundary setting and boundlessness. Obviously, if all boundaries were rejected, each researcher free to roam the universe, there would be little progress. But it is equally true that the rigid boundaries, which exist in most academic disciplines today, inhibit progress. If part/ whole inquiry is to flourish, the ever-increasing tendency for academic disciplines to become closed systems must be checked.

Summary This chapter discussed the need for two new steps in the process of inquiry: morphology, the study of minute details, and part/whole theory, connecting micro- and macroworlds. These steps take place after or along with qualitative studies, and before verification. Hyperspecialization of modem science, in disciplines, theories, and methods, has disrupted the connection between data and theory in a way that makes it virtually impossible to generate and test general theories. I explain how this state of affairs came about, and offer examples of the kinds of studies which might correct it.

Note 1. Several pages of this paper are based in part on sections from chapters 8 and 10 of an earlier publication (Scheff, 1990).

44

Interaction and Identity

References Amrine, F., Zucker, F., & Wheeler, H. (1987). Goethe and the sciences: A Re-ap­ praisal. Boston: D. Reidel. Burton, D. M. (1985). The history o f mathematics. Newton, Mass.: Allyn and Bacon. Durkheim, E. (1897). Suicide. London: Routledge (1952). Geertz, C. (1983). Local knowledge. New York: Basic Books. Giddens, A. (1984). The constitution o f society. Berkeley: University of California Press. Glaser, B., & Strauss, A. (1967). The discovery o f grounded theory. Chicago: Aldine. Goethe, J. W. (1785). Philosophical Studies. Quoted in B. Fairley (1963, p. 195), A study o f Goethe. Oxford: Clarendon Press. Goffman, E. (1974). Frame analysis. New York: Harper. Hofstadter, D. (1980). Goedel/Escher/Bach. New York: Random House. Kohler, W. (1929). Gestalt psychology. New York: G. Bell. Lemer, D. (1963). Parts and wholes. Cambridge, Mass.: Harvard University Press. Levine, D. (1985). The flight from ambiguity. Chicago: University of Chicago Press. Magnus, R. (1949). Goethe as a scientist. New York: Harry Schuman. Mills, C. W. (1959). The sociological imagination. New York: Oxford University Press. Monk, R. (1990). Ludwig Wittgenstein: The duty o f genius. New York: Free Press. Russell, B., & Whitehead, A. (1910). Principia Mathematica. Cambridge: Cambridge University Press. Scheff, T. J. (1986). Toward resolving the controversy over thick description. Current Anthropology. 27, 408-09. ______. (1990). Microsociology: Discourse, emotion, and social structure. Chicago: University of Chicago Press. ______ . (1994). Bloody revenge: Emotions, nationalism, and war. Boulder, Colo.:Westview. Scheff, T .J., & Retzinger, S. (1991). Emotions and violence. Lexington, Mass.: Lex­ ington Books. Smith, B. H. (1968). Poetic closure. Chicago: University of Chicago Press. Steiner, R. (1926). Goethe's science. Berlin: Goetheanum. Vietor, K. (1950). Goethe the thinker. Cambridge: Cambridge University Press. Wells, G. (1967). Goethe and the intermaxillary bone. British Journal fo r the History o f Science, 3, 348-61. Wittgenstein, L. (1968). Philosophical investigations (third edition). New York: MacMillan.

3 Communication 1 Gregory Bateson

Bateson2 argues that the major advances in the behavioral sciences by the mid­ twentieth century all involved a concern with communication. In this chapter, he reviews the theoretical premises underlying these advances that influenced and guided the thinking o f those initially involved in the pioneering collaborative microanalytic study o f communication process known as The Natural History of an Interview.3 The aim o f this collaboration, according to Bateson, was the develop­ ment o f “a statement o f the mechanisms o f relationships. ” O dieses ist das Her, das es nicht gibt. Sie wusstens nicht und habens jeden Falls —sein Wandeln, seine Haltung, seinen Hals, bis in des stillen Blickes Licht—geliebt. Zwar war es nicht. Doch weil sie’s liebten, ward ein reines Tier. Sie liessen immer Raum. Und in dem Raum, klar und ausgespart, erhob es leicht sein Haupt und brauchte kaum zu sein. Sie nahrten es mit keinem Korn, nur immer mit der Moglichkeit, es sei. Und die gab solche Starke an das Tier, dass es aus sich ein Stimhom trieb. Ein Horn. Zu einer Jungfrau kam es weiss herbei— und zwar im Silber-Spiegel und in ihr. —Rainer Maria Rilke, Sonette an Orpheus, II. Teil, IV. O this is the creature that does not exist. They did not know that and in any case —its motion, and its bearing, and its neck, even to the light of its still gaze—they loved it. 45

46

Interaction and Identity Indeed it never was. Yet because they loved it, a pure creature happened. They always allowed room. And in that room, clear and left open, it easily raised its head and scarcely needed to be. They fed it with no grain, but ever with the possibility that it might be. And this gave the creature such strength, it grew a horn out of its brow. One horn. To a virgin it came hither white— and was in the silver-mirror and in her. —Translated by M. E. Herter Norton

At the time of the outbreak of World War II, the most promising in­ sights in the behavioral sciences were those derived from Freudian analy­ sis, Gestalt psychology, and cultural relativity. Linguistics had begun to take on new life under the leadership of Sapir (1921,1925,1933a, 1933b) and Bloomfield (1933, 1939). Psychiatry was evolving away from the exclusive study of the individual patient toward the study of human re­ lationships, most dramatically under the influence of Sullivan (1940) and already there were moves toward a mathematics of human relation­ ship under Kurt Lewin (1935) and L. F. Richardson (1939). During World War II and immediately following that period of con­ fusion, a series of exceedingly important new approaches were evolv­ ing more or less independently in a number of different places, but the possible relevance to behavioral science of the work of George Boole (1854), and of Whitehead and Russell (1910-13) was still unexplored. All of these scattered advances were precipitated by the war-time devel­ opment of electronic engineering. A partial list of names and locations of the principle advances will give an idea of what is happening. Rosenblueth at Cambridge and Mexico, and Wisner and Bigelow4 at Massachusetts Institute of Technology and Princeton, respectively, were laying foundations of what has come to be called cybernetics, extending what the engineers and mathematicians had learned about self-correcting mechanisms to the fields of biology and social organization. Von Neumann and Morgenstem (1947)5 at Princeton were laying the basis of the theory of games. Craik (1952), in Cambridge, England, before his premature death, wrote “The Nature of Explanation,” raising the whole question of how messages are coded in a reticulate central nervous system.

Communication

47

Attneave (1959), Stroud,6 and others at Stanford happened to see Craik’s little book and by it were inspired to a new approach to the problems of perception and adaptive action. In Vienna, Von Bertalanffy (1952) was building the beginnings of systems theory with a special emphasis upon those systems (e.g., or­ ganisms) that have a continuous source of energy derived from the environment. Shannon, and others working with the Bell Telephone Laboratories (Shannon & Weaver, 1949),7were building the structure called informa­ tion theory. Ashby (1952,1956), in Gloucester, England, was devising new mod­ els for theories of learning and the evolution of the brain. Other names [McCulloch and Pitts,8 Lorente de No,9 Rashevsky,10 Walter,11Tinbergen (1953), Lorenz (1952)] might be mentioned as con­ tributing to this general trend. What has happened has been the introduction into the behavioral sci­ ences of a number of very simple, elegant, and powerful ideas, all of which have to do with the nature of communication in the widest sense of the word. The steps and sequences of logic have been coded into the causal sequences of computing machines and, as a result, the Principia Mathematica has become a cornerstone of science.

The Natural History of an Interview The natural history o f an interview is an attempt at synthesis. It is written by five persons12 who are professionally concerned with com­ munication problems in diverse fields and we attempt a synthesis of a wide and abstract kind, starting from the most concrete data. We start from a particular interview on a particular day between two identified persons in the presence of a child, a camera, and a camera man.13Our primary data are the multitudinous details of vocal and bodily action recorded on this film. We call our treatment of such data a “natu­ ral history” because a minimum of theory guided the collection of the data. The cameraman inevitably made some selection in his shooting. “Doris,” the subject of the interview, was selected for the study not only because she and her husband were willing to be studied in this way, but also because this family suffered from interpersonal difficulties that had led them to seek special psychiatric aid.

48

Interaction and Identity

These materials then, while collected under circumstances unusual in human relationship, nonetheless provide the natural history of two hu­ man beings over a brief span of time, and the data themselves are suffi­ ciently uncorrupted by theory so that the five authors, each with a particular theoretical bias and interest, could simultaneously approach this mass of detail. Moreover, we shared something less tangible than the common data: certain theories or preconceptions about what hap­ pens when two people interact. My major task in this preliminary chapter is to outline those theoreti­ cal premises that were engendered in us by the recent advances in the study of human communication. Theoretical Premises

Freudian From Freudian theory we accept the premise: (1) that only limited aspects of a part of what happens in human communication are acces­ sible to the consciousness of participants. Our position, however, differs from that of many early Freudians in two respects, which are minor as far as theory is concerned, but major in their implications for method. The important corrective that the Freudian applied to man’s thinking about human nature was an insistence upon the unconscious. The error to be corrected was the notion that in human beings the mental process is preponderantly or entirely conscious. This error had roots in eigh­ teenth-century culture and back into the Reformation and into earlier Judaeo-Greek philosophies of free will. But this error today seems al­ most fantastic. It is now a platitude to state that mental process depends upon hierar­ chic organization. Whether we think of mental levels or of a brain evolved by a process of successive telencephalization, we envisage a hierarchy of both anatomy and function. Our knowledge of hierarchic function— in machines, embryology, physiology, and in human social organiza­ tion—indicates as a truism that under no circumstances can the upper echelons of any hierarchic system handle total information about the processes and events that occur at subordinate or peripheral levels. By the same token, the upper echelons can handle only limited reports— can be only partially conscious—of all that happens at their own upper

Communication

49

level. To provide these upper echelons with total reporting would be to add to the system still higher echelons, themselves, in turn, largely un­ conscious. To us, then, the fact that most mental process (including, especially, the process of perception itself) cannot be inspected by con­ sciousness is a matter of course and what is surprising, and therefore needs explanation, is the fact of consciousness. Unconsciousness is a necessity of the economics of hierarchic organization (Sapir, 1927). This does not mean, of course, that economy of effort or the eco­ nomic use of the channels of communication to avoid jamming is the only factor determining what information shall be allowed to reach the upper echelons of consciousness. The analogy of human social organi­ zations would indicate very clearly both that upper echelons are com­ monly “motivated” not to receive information about certain peripheral events and that there are many events that the subordinate echelons are “motivated” not to transmit upwards. There are, therefore, many mat­ ters that remain “in the unconscious” for reasons other than those of economy, and the unconscious becomes a repository for material that is repressed in the Freudian sense. The second difference between our position and the classical Freud­ ian results from our emphasis upon communication. We are interested in such questions as, What signals are emitted and what orders of aware­ ness does the signaler show by emitting other signals about these sig­ nals? Can he plan them? Can he recall them? And we are interested to know what signals reach the receiver and what signals he knows he has received. Our emphasis is thus upon perception and communication rather than on the internal hierarchies of mental process. From where we sit, the distinction between conscious and unconscious becomes significantly comparable to the distinction between foveal and peripheral vision. A second premise related to Freudian theory holds: (2) that every­ thing that occurs is meaningful in the sense of being a part of the inter­ change as well as nonaccidental. The Freudian emphasis was upon psychic determinism—that no word uttered and no detail of a dream experienced can be accidental. A man cannot “just dream.” Our empha­ sis in The natural history of an interview will extend this psychological idea into the realm of interpersonal process. We shall attempt to see every detail of word, vocalization, and bodily movement as playing its part in determining the ongoing stream of words and bodily movements that is the interchange between the persons. We shall endeavor to think

50

Interaction and Identity

not only in terms of psychic determinism, but in terms of a larger inter­ personal determinism. Two people cannot “just agree” or “just quarrel.” Also, from Freudian theory, we accept the idea: (3) that all messages, whether verbal or nonverbal, are mediated in their creation by primary process and therefore contain, either implicitly or explicitly, all the mul­ tiple reference characteristics of dream or fantasy. If it be possible for a man to seem to talk only about the overt subject of conversation, this is achieved only by vigorous ego function, which carefully excludes or conceals the multiple undertones of implicit content. Further, we accept that the minute analysis of speech and movement will disclose that the messages in both these modalities contain a large proportion of uncon­ scious material with primary process characteristics; that, for example, an unconscious fingering of the dress is likely to denote (or to be a re­ sultant of) sexual interest and/or for its puritanical denial. Also from Freudian theory, we accept: (4) a generalized notion of trans­ ference, that any person emitting learned signals does so upon the (usually unconscious) assumption that the receiver of these signals will understand them “correctly,” that is, he assumes that his vis-a-vis at the given mo­ ment will resemble psychologically some former (or fictitious) vis-a-vis from whom he originally acquired his communicational habits. Closely related to the notion of transference is: (5) the notion of projec­ tion. This explanatory principle differs, however, from transference in that it does not invoke some third historical or fictitious person. When A “projects” upon B, he is merely assuming that B ’s signals are to be inter­ preted as A would interpret these signals if he himself had emitted them. That is, A assumes that B operates according to systems of codification similar to his own. Both transference and projection may, of course, be carried into the future. A may expect that B will exhibit meaningful action of a sort that some historical figure in A’s life would have exhibited under similar circumstances (transference); or he may expect that B will behave as he himself would behave in similar circumstances (projection). Identification must also be mentioned. This explanatory principle in­ vokes: (6) the idea “if you can’t lick ‘em, join ‘em”—or, at least imitate them as you see them. A is said to identify with B when he starts to mold his own meaningful action in terms of what he thinks are B ’s principles of codification. Notably, all of these principles—transference, projection, and identifi­ cation—are likely to be unconscious in their operation and to be in some

Communication

51

degree coercive. That is, any errors that A may make in his assumptions about B are likely to cause A to act in such a way that B is put under pressure to validate these errors by acting as if A’s assumptions were true. An especially interesting case arises when A acts in a way that will coerce B into identifying with A’s self-image, which may be false. Moreover, it must not be supposed that these explanatory or descrip­ tive principles are mutually exclusive. In a given instance, A may con­ sciously or unconsciously assume that B is parental (transference). But A’s technique for dealing with his parent may have involved identifica­ tion.14 He will then adopt towards B that role that he formerly adopted toward the parent.

Gestalt From Gestalt psychology, we have accepted a premise of very great importance: that experience is punctuated. We do not experience a con­ tinuum: on the contrary, our experience is broken up into what seems to us to be events and objects. In Gestalt psychology this idea is basic to the figure-ground hypothesis. For us it is related to the premise that nothing never happens that is, that both sender and receiver of signals are so organized that they can and must use, for their understanding of what is going on, the fact that certain possible signals are not present. The first step in building the figure-ground hypothesis is a postulate of this kind. In order to recognize that there are stars in the night sky, we must see the fact that certain retinal end organs are not stimulated by the darkness. In human relations, no silence is insignificant and the absence of tears may speak volumes. More must be said concerning the punctuation of interpersonal events. Our whole procedure, and, indeed, any analysis of communication data, is shaped by premises that define the units into which the stream of data is to be divided. First, in a macroscopic examination of the interview, we assumed that the 400 feet of film on which the interview is recorded15 can be punctuated into incidents or sequences with beginnings and ends psychologically meaningful to the participants. As will be seen, we have chosen certain of these incidents for microscopic study. Our macroscopic study serves to direct our more particular attention. While we narrow our focus from the interview as a whole to an examination of incidents within the interview and then downward to the finer and finer detail of

52

Interaction and Identity

these incidents, we work throughout with similar assumptions about punctuation of the stream of signals. The historical basis of this assumption will make clearer what is meant. Historically, scientific linguistics has progressed most rapidly since the time when certain popular and preponderantly occidental notions about language were adopted, made rigorous, and extrapolated into the study of minute detail. In their popular form, these notions are, for example, that speech is subdivisible into sentences, which in turn are subdivisible into words, which in turn can be subdivided into letters. Profound modi­ fications have been introduced into this hierarchy by the linguists who needed to describe speech rather than written language, but the essential idea that a stream of communicative material must, of necessity, be sus­ ceptible of such multiple resubdivision is fundamental in linguistics and in that part of communications theory that deals with coded communi­ cation—a much wider field than the conventionally linguistic. A major contribution of the linguists is the demonstration that the stream of com­ munication contains positive signals by which its units are delimited. Moreover, Gestalt theory presupposes a hierarchy of subdivisions characteristic of the process of perception. We do not perceive the firing of unit end organs but, from the showers of neural impulse started by that firing, we build images of identifiables and larger meaningful com­ plexes of identifiables. We can argue from perception to communica­ tion. If an organism’s perception is characterized by Gestalten and this organism is capable of emitting complex streams of communication, then these streams must be dissectable into a hierarchy of successive subdivisions. Many such analysis will be possible, and there will be one that will represent correctly the natural history of the organism. We deal, after all, not merely with the fact that a communicational stream can be dissected, but also with the question, In which of the many ways possible should this particular stream be dissected? What we know about language and communication in general indicates that there will always be one or more hierarchies of Gestalten that will be correct in the sense of describing how the message stream is created and/or how it is received and interpreted by the hearer. The Freudian findings also indi­ cate that in any given instance several different interpretations may be correct. A particular message may be simultaneously interpreted in dif­ ferent ways by different levels of the mind: we face problems of mul­ tiple coding.

Communication

53

The linguists are ahead of the other natural historians in their study of the hierarchy of Gestalten in terms of which a particular kind of behav­ ior should be dissected. Their studies are being fortified by cross-cul­ tural comparison, by dialectical (subcultural) comparison, and by statistics of individual variation. By way of contrast, kinesics—the study of body motion, position, and action as a modality of communication—is a rela­ tively recent development which, like linguistics, is achieving a firm scientific base by the rigorous dissection of the kinesic stream into a hierarchy of Gestalten and subdivisions of Gestalten. In a later section of The natural history o f an interview, Birdwhistell outlines the hierarchy of units that he is devising for kinesic descrip­ tion.16 He is proceeding in a manner comparable to but not identical with that methodology of description that has proved valuable in lin­ guistics. The ultimate validation of this approach in kinesics, will, of course, depend upon the results obtained, but there is very strong a priori argument in favor of the correctness of the approach from all that we know about communications theory in general and about human com­ munication and perception in particular. Returning for a moment to linguistics, other types of description that the linguists have achieved must here be mentioned. The very complex question of “meaning” is too large for discussion in this chapter, but this much may be said—that a tape recording of human speech contains a great deal more than the signals correlated with the lexical meaning of what was said. If a tape recording is transcribed into ordinary script, although some of this more-than-lexical content will be lost, some will still survive in the transcription. Indeed, to reduce a speech to its merely lexical content would require very drastic procedures (in the course of which other and probably inappropriate nonlexical overtones will inevitably added). It would first be necessary to strip the speech of all indications of the context in which it was uttered and by whom and to whom it was uttered. But there would still remain cadences and overtones of a nonlexical nature. To get rid of these it would be necessary to translate the speech into some other lan­ guage and to use as a translator some hypothetical person (or machine) totally insensitive to nonlexical content both of the language from which he is translating and of the language into which he is translating. As we climb the hierarchic ladder of Gestalten from the most micro­ scopic particles of vocalization toward the most macroscopic units of

54

Interaction and Identity

speech, each step on this ladder is surmounted by placing the units of the lower level in context. “Meaning,” as this word is vulgarly used, emerges only at a very high level in this hierarchy. We discriminate the initial phoneme of the word “peter” from the initial phoneme of the word “butter,” but these pho­ nemes are in themselves meaningless apart from their setting in a stream of phonemes. Even the syllables “pete” and “but” are in themselves meaningless or multivalent (except insofar as their possible meaning is restricted when we know how they are placed in a stream of syllables). With each step towards a larger unit—the larger unit being always the smaller unit plus its immediate setting—there is a more and more dras­ tic limitation of possible referents. “Meaning,” therefore, is a function of this restriction of possible meanings. Even the words “peter” and “butter” are multivalent. When the word “blue” is added, the hearer may be pretty sure that the referent of “pe­ ter” is a flag. But still, there is room for doubt. The “Blue Peter” may be referred to as an actual object of action or observation in the larger context of a ship about to leave a particular port. Or the reference may only be metaphoric if the term is used on land. Or the usage of the term may be neither metaphoric nor direct but may be part of a lesson in maritime communication. Or—as here upon this page—the words “Blue Peter” may be mentioned only as an ex­ ample of communicational phenomena. Meaning approaches univalence or nonambiguity only when very large units of the communicational stream are admitted to examination. Even then, the approach to nonambiguity will be asymptotic. As larger and larger bodies of data are admitted, the probability of a given interpreta­ tion will be increased, but proof will never be achieved. The situation is essentially the same as that which obtains in science where no theory is ever proved. The natural history o f an interview is concerned with trying to put together those parts of the communicational stream that the profes­ sional linguist studies (phonemes, morphemes, phrases, vocal modifi­ ers, junctures, etc.) with those parts of the stream that are studied in kinesics (kines, kinemorphs, etc.). A central question, therefore, which we shall have to face when we analyze the data, is the extent to which there is a mutual relationship of “context” between kinesic and lin­ guistic elements.

Communication

55

We face phenomena so structured that there is perhaps no definable upper limit to the size—either spatial or temporal—of the Gestalten. In practice, this would mean that no finite collection of data would confer complete nonambiguity upon any item within the collection; that how­ ever widely “context” be defined, there may always be wider contexts a knowledge of which would reverse or modify our understanding of par­ ticular items.

Context These considerations force us to a method of inquiry that will post­ pone the question of “meaning.” When faced with a given sequence of signals, we shall delay the question “What do these signals mean?” for as long as possible. We shall ask, rather, the collateral question, “Would the meaning be changed by a given change in the sequence or by a given change in the context?” This is a question that can be asked and an­ swered without too much difficulty. We shall, for example, not ask whether the word “Peter” refers to an apostle or a flag, but rather, whether its meaning when the word “Peter” follows the word “Blue,” is pecu­ liarly appropriate to the new context. In kinesic analysis, we shall similarly delay the question of what is meant by the rapid closing and opening of an eye which is invisible to the vis-a-vis and shall ask rather, for example, whether the meaning of this signal would be altered (a) if the other eye were blinked simulta­ neously, and (b) if the blinked eye were one which is invisible to the visa-vis. Parenthetically, we may also ask whether the meaning of the word “Peter” is altered by winking one eye. It is, after all, only an historic accident—a past pathway in the evolu­ tion of science—that has lead to the circumstance that linguists study the data that can be heard, while the kinesicist studies data that can be seen. That the scientists have become specialized in this particular way does not indicate a fundamental separateness between these modalities in the stream of communication. It is for this reason that the work of The natural history o f an interview starts from concrete natural history— from the recorded natural interaction between Doris’s speech and move­ ment and the speech and movement of Gregory. This placing of every signal in the context of all other signals is an essential discipline of our work. A great part of the work that Birdwhistell, Hockett, and McQuown

56

Interaction and Identity

have had to do has involved a grueling process of synchronization. The audible stream for which Hockett and McQuown are specialists was recorded on tape and on film with an unsatisfactory sound track. The analysts had to work frame by frame through the film to establish the point in the audible sequence at which, for example, Doris turned her head or let her shoe fall away from her heel. I describe our data loosely above as the aggregate of signals recorded on film. More accurately, I should have said that our data are the individual signals or messages, each in its immediate and extended context. But the context of a signal emitted by Doris is not merely those other signals that she has recently emitted plus those she emits soon after; it is also the room in which she is speaking, the sofa on which she is sitting, and the signals emitted by Gregory with whom she is talking, and by the little boy, Billy, and the interrelationships among all of these.

Interaction At this point, our concept of communication becomes interactional and our intellectual debt is to G. H. Mead (1934) and to Sullivan (1940) rather than to Freud and the Gestalt psychologists. The system that we now study is no longer merely a descriptive synthesis of Doris’s body motion and speech, but the larger aggregate of what goes on between Doris and Gregory. This larger frame determines meaning for what each person does and says. Rilke’s “Unicom” is present in every conversation between persons and this fictitious beast evolves and changes, dissolves and is recrystal­ lized in new shapes with every move and message. Denial of the unicorn will not prevent its existence, but only cause it to become monstrous. This poetic fancy must be made scientifically real to the reader if he is to understand what The natural history o f an interview is all about. For every human being there is an edge of uncertainty about what sort of messages he is emitting, and we all need, in the final analysis, to see how our messages are received in order to discover what they are. For the schizophrenic this is often dramatically and conspicuously true. Let us illustrate by an example. A schizophrenic patient tells me that he built the China Wall, rowed across the Pacific island, and landed in Seattle. He then walked to California where he was “affriended by those people.” This narrative he offers as if it were a statement of fact. But

Communication

57

whether it was a statement of fact for him depends on my response. If I say, “Nonsense, you were bom in California,” I have thereby verified for him the notion that his narrative is to be taken literally. I have denied it as if it were a literal statement and now it exists for him as a literal statement, which must be defended. From there on, we shall get into an argument not about the question “Is this narrative a statement of fact?” but about the red herring question “Is this a true statement of the facts?” The response that we get tells us about the state of the hearer after he has received the signals we emitted. It may be evident that he misunder­ stood the message either grossly or subtly. The status quo, however, which was obtained when we emitted the message, no longer exists and merely to repeat the message will not do. We are now communicating with a person whose relationship to us is different from what it was a moment ago. Within the framework of this new relationship we must now speak. Of all the elements and vicissitudes of formation and re-formation of relationships, perhaps the most interesting is that process whereby people establish common rules for the creation and understanding of messages. Whatever reply I may make to the patient’s delusional narrative pro­ poses a pact to govern us both in our understanding of the message. If I deny the factual truth of the narrative, I implicitly propose that we agree to treat it as literal. If, on the other hand, I ask him whether he thinks his parents had a part in building the “China Wall” that separates him from them, I have proposed that we agree upon a different set of rules for the creation and understanding of messages. The possible systems of rules that two persons may share are many and complex. Among them must be mentioned a system that has been characterized as symbiotic. Such a label refers to, as I understand it, a system of nonverbalized and usually unconscious pacts in which, for example, A and B “agree” to accept each other’s messages in some spirit other than that in which they were coined. By ignoring overtones and implications, or by reading in overtones that were not intended, the per­ sons maintain a strange semblance of understanding.

Code Distortion In the The natural history o f an interview we shall pay but little atten­ tion to those failures of communication that are due to the randomiza­

58

Interaction and Identity

tion of signals occasioned either by background noise or by imperfect resolving processes in the receiving sense organ.17 We are concerned with a more subtle phenomenon—the distortion of messages that occurs when the persons involved differ from each other in their rules or as­ sumptions governing the making and understanding of messages—their explicit and implicit rules of coding. Imagine a machine that has the function of transmitting a half-tone block (a picture formed entirely of rows of dots) over a wire to another machine. The transmitting machine will transmit over the wire a se­ quence of electrical impulses such that each impulse or absence of im­ pulse is a “yes” or “no” answer to the question, Is there a dot in the given space? When the transmitting machine comes to the end of a line of spaces, it will either transmit a special signal, which will cause the re­ ceiving machine to go on to the next line, or the machines will have to be adjusted to each other so that they operate in terms of a common pact governing how many dots there shall be in a line. A discrepancy regard­ ing the terms of this pact will introduce code distortion. In such a case, the receiving machine will create a picture that may be an absolutely correct record of the sequence of signals emitted by the sender but which, considered as a picture, will be distortion of the original. The top half of figure 3.1 shows a picture to be transmitted, the bot­ tom a distorted version that is created when the receiving machine acts upon the premise that there are only sixteen squares in each line instead of seventeen.18 In figure 3.1, the effect of code distortion is shown, and it is worth­ while to stress the basic difference between such distortion and the loss of information due to entropic noise. In the case of entropic noise, the information that is lost is irretrievable, but in the case of code noise what has occurred is a systematic distortion which that could conceiv­ ably be rectified. All that is needed for this correction is that there be some means whereby the transmitter and receiver can communicate about the rules of communication. This presents special difficulties, but it is a funda­ mental thesis of The natural history o f an interview that at the human level such communication about the rules of communication occurs con­ stantly. This, in fact, is the process whereby the “Unicom” is continu­ ally created and recreated. When my patient tells his story of the “China Wall,” whatever reply I make is a communication to him about how I

Communication Figure 3.1

59

60

Interaction and Identity

received his message and therefore indicates, to him (ideally) how he should restate it in order to have me receive that message that he wants me to receive. It tells him how to code his messages so as to elicit an appropriate response from me. It is necessary again to insist upon the unconscious character of most communication. We are almost totally unaware of the processes by which we make our messages and the processes by which we understand and respond to the messages of others. We are commonly unaware also of many characteristics and components of the messages themselves. We do not notice at which moments in a conversation we cross and uncross our legs or at which moments we puff on our cigarettes or blink our eyes or raise our brows. But the fact that we do not notice these things does not imply that all these details of personal interaction are irrelevant to the ongoing relationship. Just as we are in the main unconscious of the fleet­ ing pacts that we enter into as to how messages are to be understood, so also we are unconscious of the continual dialogue about these pacts. This dialogue is not only between persons and about the pacts that they form, it is also and more strangely a dialogue that governs what each person is. When A makes overtures that B brushes aside, this expe­ rience is to A more than a hint about how to code messages when deal­ ing with B. In everyday language we say that a person’s self-esteem is enhanced or reduced by the responses of others. Or we say that “he sees himself differently.” In communicational terms, we may translate this into a statement that the very rules of self-perception, the rules govern­ ing the formation of a self-image, are modified by the way in which others receive our messages.

Learning and Pathogenesis In part, The natural history o f an interview is a study of how commu­ nication works between two persons, but it is also a study of how com­ m unication fails to w ork—that is, of certain pathologies of communication. Our collaborating team includes not only the two lin­ guists and the founder of kinesics, but also two psychiatrists and the writer, whose initial training in anthropology has finally led him to study schizophrenic communication. It is therefore appropriate to examine a little more closely the relation between psychiatric pathology in the in­ dividual patient and the pathologies of communication that may develop

Communication

61

between persons. In order to keep the subject matter simple, I will ex­ clude from consideration those psychiatric abnormalities that have an established base in organic lesion. To build a bridge between psychiatric functional pathology and the pathologies of communication, it is necessary to insist upon the exist­ ence of the facts of learning and conditioning. Two considerations be­ come especially relevant. First, every failure of communication is painful. Second, the learning organism always generalizes from experience. Fur­ ther, the business of communication is a continuous learning to commu­ nicate. Codes and languages are not static systems that can be learned once and for all. They are, rather, shifting systems of pacts and premises that govern how messages are to be made and interpreted. Every signal that establishes a new premise or pact bringing the persons closer to­ gether or giving them greater freedom may be a source of joy. But every signal that falls by the wayside is in some degree a source of pain to both. The ongoing stream of communication is thus, for each individual, a continuous chain of contexts of learning and, specifically, learning about premises of communication. At this point, it is necessary to consider certain aspects of the learning process and to expand conventional learning theory to make it relevant to an analysis of the interchange of signals between persons. A typical learning experiment involves two entities: an experimenter and a sub­ ject, and the theoretical conclusions derived from such experiments are commonly stated as psychological regularities descriptive of the sub­ ject. In contrast, I shall here view the experimental situation as an inter­ action involving two entities in whose relationship I am interested. I shall regard their relationship as formally characterized by an interchange that is repeated in successive “trials” and shall assume that not only the subject, but also the experimenter is undergoing a learning process de­ termined, at least in part, by reinforcements that the subject provides. As a preliminary to this, it is necessary to define a hierarchy of orders of learning. This may be done as follows: 1. It appears that the simplest learning phenomenon is the receipt of information or command. The event of perceiving a whistle may constitute, for a dog, an impor­ tant piece of information or a command. Before it heard a whistle, it was in another state. The change of state I classify as the simplest learning phenomenon. It is important, however, to remark that this phenomenon is excessively difficult to in­ vestigate and has not been an immediate object of experimental study. It has, how­ ever, been a major focus of theory. What seems to have happened is that in order to

62

Interaction and Identity arrive at a theory to describe what I call the second order of learning, the psycholo­ gists have had to provide some description of this first order process—some ver­ balization of what message the dog has received. The “effect” theory proposes that this message is a promise of reward or a threat of punishment, whereas the associational theory proposes a more automatic and less purposive description of the dog’s response. 2. The second order is the learning or conditioning upon which the vast mass of experimental work has been conducted. Here the word “learning” refers to a change in the dog’s ability to act upon percepts or signals received. What the experiment­ ers study is changes in the dog’s behavior resulting from a sequence of trials. The phenomena studied are of a higher order than those discussed in item one above. The question asked is not “What change occurs in a dog when he hears a whistle?” but “What changes have occurred in the change that a dog undergoes when he hears a whistle?” This subtle difference in the question asked by the experimenter makes it formally impossible for the theorists to deduce answers to the first ques­ tion from data collected to answer the second. The behaviorists had logic on their side when they insisted that we never ask about the subjective experience of the dog. To try to deduce the dog’s experience from data that could only throw light upon change in his experience is to attempt the logically impossible. From the characteristics of a class, I can make no deductions about what a member of that class might be. 3. The third order of learning is a familiar laboratory phenomenon but has received only slight attention from the experimentalists. If we describe the second order of learning as “learning to receive signals,” then the third may be described as “learn­ ing to learn to receive signals.”19What happens in the laboratory is that the animal, having been subjected to experiments of the second order, becomes “wise.” That is, when faced with an entirely new experiment of this sort, the animal requires now a smaller number of trials to achieve that learning of the second order that the experimental situation demands. The animal has acquired a knack or skill for sec­ ond-order learning. This phenomenon has been measured by Hull (1940), studying rote learning of nonsense syllables, and by Harlow (1949), studying problem solv­ ing in rhesus monkeys. 4. There is no theoretical reason to deny the possibility of fourth and higher orders of learning, though none of these have been demonstrated. The nature of the hier­ archy that we are discussing is such that there is no upper limit to the series other than that set by the limitations of brain structure. The number of neurons being finite, it is certain that for any organism there is a practical upper limit to the num­ ber of orders of learning of which it is capable.

Inspection of this hierarchy of learning reveals that the difference between any order of learning and the next higher order is essentially a difference in the size of Gestalt. The higher order is always documented by demonstrating change that results from a larger Gestalt, this larger Gestalt being in general built up of a multiplicity of the Gestalten char­ acteristic of the lower order. But while this generally seems to be the case, there is no theoretical premise by which we might estimate the

Communication

63

multiplication factor, and it is necessary to consider as at least theoreti­ cally possible the case in which this factor would be unity. A single increment in what appears to be a context of lower-order learning might conceivably precipitate major changes of some higher order, whereby all experience of the lower order would be reframed and reorganized. We face here an unpredictability of a sort, which I noted earlier when discussing the indeterminacy of meaning. Larger and larger bodies of data will provide greater and greater certainty of interpreta­ tion, but it is never possible to be sure that the next increment of data will not compel us to a totally new interpretation. There is thus an anal­ ogy—perhaps amounting to identity—between those hierarchies of Gestalten that determines meaning and that hierarchy of Gestalten that we here call contexts of learning. These abstract matters become clearer when we state that learning of the third or higher order is, in popular parlance, called “change in charac­ ter.” Let us suppose that an organism becomes “wise” in dealing with contexts of Pavlovian learning. The change we refer to here may be de­ scribed both as a change in the organism’s expectations and as a change in its learning habits. If we speak in terms of expectations, we will say that the organism now preponderantly expects the universe of experience to be punctuated into sequences resembling the Pavlovian context (i.e., se­ quences in which certain percepts can be used as a basis for predicting later events). Or, if we speak in terms of learning habits, we will say that this organism will respond to the predicted certainty of that which is to come, (e.g., by salivating), but will not endeavor to change the course of events. In a word, the organism has become “fatalistic” and examination of the formal characteristics of the learning context has provided us with a formal definition of one particular sort of “fatalism.” The psychiatrist is interested largely in learning of the third or higher order. If a patient tells him that she can use a typewriter, the psychiatrist will pay but little attention. She has reported only a result of secondorder learning. But when the patient goes on to describe the context in which she learned to typewrite and tells him that the teacher punished every error she made but never praised her for progress, the psychiatrist will prick up his ears. He will see in this narrative a statement of what effect the context of learning to type may have had upon the patient’s habits and expectations (i.e., upon the patient’s character). This enlarge­ ment of learning theory to discriminate orders of learning makes this

64

Interaction and Identity

body of experimental knowledge especially relevant to the psychiatrist. Actually the old barrier between experimentalists and clinicians seems to have grown out of this: that the experimentalists have mainly studied learning of the second order, while the psychiatrist is interested chiefly in effects of the third order. These effects he tries to evaluate in his diagnosis or to achieve in his therapy. If this description of learning is substantially correct, that is, if there really is a hierarchy of orders of this phenomenon and the discrimina­ tion of these orders is something more than an artifact of description, then it becomes theoretically probable that there exist complex sequences of experience and action such that learning of one order will in some degree contradict the learning of some other order. We can imagine, for example, that a human subject might experience a long sequence of Pavlovian learnings but might be penalized (Bateson, Jackson, Haley & Weakland, 1956)20 for exhibiting “fatalism.” Or he might be trained to­ ward obedience but be continually penalized for the finer detail of every obedient act. As between adults, this is familiar enough and may make for bad “personnel relations.” As between parents and small children, I believe that it is, under some circumstances, pathogenic.

Pathogenic Contexts It is now clear, however, in an abstract and formal way, what patterns of interchange we should look for in our data. The discussion that pre­ ceded this reexamination of learning theory concerned the establish­ ment of pacts and premises of communication. But evidently a premise of communication, a rule governing how messages are to be constructed or interpreted, bears the same relationship to the given message as oc­ curs between a higher and lower order of learning. The acceptance of what I have called a premise of communication is the same phenomenon as the acceptance of a role—a momentary or enduring shift in habit and expectation. “Role” is only a word for some phase of character change, be it brief or enduring. It is a description of the pattern exhibited by one person in that two-person system that con­ stitutes a context of learning. It follows that what we have to look for in the data is sequences and, at the metalevel, sequences of sequences. The relevant units will be those segments of the stream that constitute contexts of learning. Problems of

Communication

65

pathology within the stream will become recognizable when we see in­ stances so constructed that learning in a given small sequence would be contradicted by learning in some larger sequence of which the smaller is a component. Theoretically, we may expect instances in which part and whole will be identical—where the multiplication factor relating the part to the whole is unity. A single context (seen in two different ways) may propose contradictory learning at different levels. One other peculiar phenomenon must now be mentioned—namely, that the premises of communication are commonly self-validating. By their operation they may create that consensus that will seem to validate them. He who also believes that all the world is his friend—or enemy— will emit messages and act meaningfully in terms of his premise. He will meet the world in a way that puts pressure upon this very world to validate his belief; a belief he acquired in the first place by the cumula­ tive impact of those contexts of learning that were his communication with some earlier person. An inquiry into the functional psychopathologies thus becomes an investigation of the dynamics of past communication. But curiously enough, because of this fact that communicational premises are self­ validating, it is often not necessary to delve into the past in order to investigate their etiology. The premises are self-validating in the present and therefore the disturbed—like the normal—is continually creating around himself that environment that provides the typical etiology for his communicational habits—his symptoms. One has only to examine the present family relations of a patient to find working today the con­ stellation which that is etiologic for his symptoms. Indeed, we may prof­ itably examine the workings of any typical mental hospital for clues as to why its patients are mentally ill. This broad description of the interchange between persons as a se­ quence of contexts of learning contains the possibilities for two kinds of psychopathological result: the learning of particular error and the disruption or distortion of the learning process itself. Historically, the first of these received most attention in the early days of psychoanaly­ sis when emphasis was placed upon the fact that certain neuroses re­ sult from single and extremely painful experiences in childhood. In terms of what has been said above, we might rephrase this theory as a learning of error—the error being an inappropriate generalization from some terrifying, painful, or over-rewarding experience. Today, less

66

Interaction and Identity

theoretical importance is attached to this kind of pathogenesis, but its occurrence is still undoubted. In contrast, modem psychiatric theory insists more upon those psychopathological results that derive from continual and repeated experi­ ence rather than from isolated trauma. Here the probability that simple error will be generated in the learning individual is much less, since, after all, his opinions, stemming from a multitude of instances, are to that extent validated by the repetition of instances. What is rather to be expected from such an etiology is the distortion of the learning process itself—a type of pathological result more abstract and intangible—and more difficult to correct by any therapeutic experience, since whatever the patient learns from this experience will probably be learned by means of that process that is already distorted. It is, however, necessary to give some substance to the phrase “dis­ tortion of learning.” I have to indicate what sorts of interpersonal se­ quences might have this effect on one or on both of the participants. A context of learning is a definitely structured segment of the stream of interchange between persons. We know from experimental data that while the structuring of contexts of learning is extremely variable, some structuring is always present. The events of which the context is com­ posed—conditioned stimulus, response and reinforcement may be vari­ ously related to each other and still constitute a structured whole. That is, we are dealing with Gestalten (units of the interchange) and are there­ fore again face to face with the peculiar nature of all such units. Al­ though they are in large part the creation of the individuals concerned, and are necessarily a product of the ways in which these individuals perceive and punctuate what is happening, their perception is inevitably guided by culture and convention. Such perception may be rigid or it may be flexible. But the essential fact is that the rules for this punctua­ tion are a part of that system of pacts and premises upon which commu­ nication is based, and for their learning communication must be viewed as a sequence of contexts. What I am describing is a strangely retroflexed procedure; a process that is in a way folded back upon itself. This may be said in many ways and perhaps most simply by stating that the communicational stream is a sequence of contexts both of learning and of learning to learn. At this point, the phrase, “distortion of the learning processes” takes on meaning. It would refer to all those cases in which an individual

Communication

67

punctuates a stream of communication in a way different from his vis-avis, but which are reinforced nonetheless by the pain resulting from his idiosyncratic view. From the point of view of the speaker, it will seem to him that he has incurred punishment for what he thought he was com­ municating, whereas he is in fact being punished for what his messages seemed to be, as perceived by the other. It is clear that this line of thought, if substantially correct, will lead to a formal theory of stability and instability in human relations. We might, therefore, inquire into what engineers call criteria of stability. Is it pos­ sible to classify the degrees and orders of misunderstanding in such a way as to separate those conditions that will be corrected by the partici­ pants, so that the system continues in a steady state, from those others that lead to a progressive deterioration? At the present time, such a ques­ tion can only be posed in the most general terms and no meaningful answers can be imagined. One relevant matter must, however be men­ tioned—that we deal with entities whose behavior is by no means describable in terms of linear equations and monotone logic. What actually seems to happen in many instances is that when what seems to be pro­ gressive change sets in, the situation becomes more or less intolerable for one or both persons and some sort of climactic outburst occurs. Fol­ lowing this, the system either returns to a state that existed before the change began or entirely new patterns of communication may be evolved. There are, after all, larger and larger sequences of interchange than any we meet with in the brief spans of data upon which The natural history o f an interview is based. From what little we know of the relationship between the fine details of human interaction and the longer cycles of the career line, there is reason to expect that the longer cycles will always be enlarged repeti­ tions or repeated reflections of pattern contained in the fine detail. In­ deed, this assumption that the microscopic will reflect the macroscopic is a major justification o f most o f our test procedures. A major function of the techniques of microanalysis is, therefore, to obtain from small quantities of data, accurately and completely recorded, insights into hu­ man relationship that could otherwise only be obtained either by long­ term observation or from the notoriously unreliable data of anamnestic reconstruction. In sum, we are concerned in The natural history o f an interview with presenting the techniques for the microscopic examination of personal

68

Interaction and Identity

interaction. While, of course, the words that people say to each other have importance, the question with which we are concerned, the prob­ lem of describing the relationship between persons, is not a question that can be answered by any summary of the dictionary meaning of their messages. There is a vast difference between the mechanical descrip­ tion “A gave B such and such information,” and the description of the interchange, “A answered B ’s question immediately.” The ultimate goal of the procedures outlined in The natural history o f an interview is a statement of the mechanism of relationships. No state­ ment of mechanism without larger context can be of long-term interest; no statement of relationship, unsubstantiated by a statement of mecha­ nism, can warrant confidence. In order to trace the path from mecha­ nism to validated relationship it is first necessary to lay out for the reader some description of how the flow of linguistic and kinesic material can be systematically described. Notes 1. I am grateful to Professor Norman McQuown for introducing me to microanaly­ sis through The natural history o f an interview while I was a student at the Uni­ versity of Chicago and for his encouragement to publish Bateson’s chapter in this volume. Gregory Bateson wrote this chapter in the late 1950s as the first chapter of The natural history o f an interview. This pioneering collaborative study has to date only been available in English on microfilm [McQuown, N. A. (ed.). (1971). The natural history o f an interview. Microfilm Collection of Manuscripts on Cul­ tural Anthropology, 15th series. Chicago: University of Chicago, Joseph Regenstein Library]. As noted (in note two) in Mokros in this volume, transla­ tions have appeared in French and Spanish. 2. The manuscript has been edited in the following ways. Whenever Bateson used phrases like “the present book,” this has been replaced with “the natural history o f an interview ” so as to minimize potential confusion for the reader. The avail­ able manuscript contained no list of references. I reconstructed a reference list, including only those works that I felt confident Bateson had intended. For those citations in the text where I was unable to identify what I perceived as a plausible source, I deleted the year, but kept the author’s name in the text, and have put the year of the citation in editor’s notes. Also, in those few cases where I have changed the citation year from that offered by Bateson, I have also enclosed this in the references. 3. Wendy Leeds-Hurwitz provides a useful history of The natural history o f an in­ terview collaboration [Leeds-Hurwitz, W. (1987). The social history of The Natural History of an Interview: A multidisciplinary investigation of social communica­ tion. Research on Language and Social Interaction, 20, 1-51]. 4. Bateson here referenced a 1943 publication. 5. Bateson referenced this as a 1944 publication. 6. Bateson here referenced a 1949 publication.

Communication 7. 8. 9. 10. 11. 12.

13.

14.

15. 16.

17. 18. 19.

20.

69

Bateson here referenced only Shannon. Bateson here referenced 1943 and 1946 publications. Bateson here referenced 1922 and 1933 publications. Bateson here referenced a 1948 publication. Bateson here referenced a 1953 publication. The five individuals referred to by Bateson included two linguists (Norman McQuown and Charles Hockett), the developer of kinesics (Ray Birdwhistell), and two psychiatrists (Henry Brosin and Frieda Fromm-Reichmann). Duncan and Fiske [Duncan, S. & Fiske, D. W. (1977). Face-to-face interaction. Hillsdale, N.J.: Erlbaum], in their discussion of this collaboration, identify Alfred Kroeber and David Schneider, both anthropologists, as members of the original research team (p. 136). Bateson interviewed “Doris’*with “Billy,” her son, present in the room during the filming. Bateson identifies the camera and the camera man because they clearly are parties to the interaction. However, he ignores a variety of other “parties” in the environment that “participated” in the interview’s natural history. Among these were the sounds of a television set, the periodic rumbling of a train passing by, and a variety of sounds produced by toys played with by Billy. For a transcrip­ tion, analysis, and fuller discussion of these additional “participants,” see Mokros, H. B. (1977). Snap, click, bang, “whango”: A transcription of extraneous sounds from the first protocol tape of the natural history of an interview. Manuscripts in Cultural Anthropology; no. 321, series LXI, Regenstein Library, University of Chicago. Bateson note: The word “identification” was perhaps an unfortunate choice for two reasons. In the phrase “A identifies B as father” is a statement of transfer­ ence. The phrase “A achieves ego-identity” suggests (as an ideal) A’s escape from all the errors of transference, projection, and identification. Bateson note: The film has irrelevant gaps while the camera was being reloaded after each one-hundred feet. The longest consecutive recording is about threeand-a-half minutes. Birdwhistlell’s method, along with sample transcriptions and analyses of scenes from The natural history o f an interview, is to be found in Birdwhistell, R. (1970). Kinesics and context: Essays on body motion communication. Philadelphia: Uni­ versity of Pennsylvania. See citation in note 13 for another perspective on this issue. Thanks to Raul Goyo-Shields for generating this figure. Bateson note: Unfortunately, in an earlier theoretical paper (Bateson, 1942), I have used, for this third-order learning, the term deutero-learning. This was due to my failure to recognize the receipt of a meaningful signal or the receipt of a bit of information as an example of the simplest order of learning. To achieve any analogy between the mechanical computers and the brain, it is unnecessary to insist that any receipt of information is, in a broad sense, learning. Bateson here referenced Bateson and Jackson (1956).

References Ashby, R. (1952). Design fo r a brain. New York: John Wiley & Sons. ______. (1956). Introduction to cybernetics. New York: John Wiley & Sons. Attneave, F. (1959). Applications o f information theory to psychology. New York: Holt, Rinehart & Winston.

70

Interaction and Identity

Bateson, G. (1942). Comment on “The comparative study of culture and the purpo­ sive cultivation of democratic values” by Margaret Mead. In L. Bryson & L. Finkelstein (eds.). Science, philosophy, and religion; Second symposium. (pp. 8197). New York: Harper & Row, Inc. Bateson, G., Jackson D. D., Haley, J., & Weakland, J. (1956). Toward a theory of schizophrenia. Behavioral Science, 1, 251-64. Bertalanffy, L. von. (1952). Theoretical models in biology and psychology. In D. Krech & G. S. Klein (eds.). Theoretical models and personality theory. Durham, N.C.: Duke University Press. Bloomfield, L. (1933). Language. New York: Holt. ______. (1939). Linguistic aspects o f science. Chicago: University of Chicago Press. Boole, G. (1854 [1940]). The laws o f thought. London: Open Court Publishing Company. Craik, K. J. W. (1952). The nature o f explanation. Cambridge: Cambridge University Press. Harlow, H. F. (1949). The formation of learning sets. Psychological Review, 56, 5165. Hull, C. (1940). Mathematico-deductive theory o f rote learning: A study in scientific methodology. New Haven, Conn.: Yale University Press. Lewin, K. (1935). A dynamic theory o f personality. New York: McGraw Hill. Lorenz, K. (1952). King Solomon's ring. New York: Crowell. Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago. Neumann, J. von. & Morgenstem, O. (1947). Theory o f games and economic behav­ ior. Princeton, N.J.: Princeton University. Rashevsky, N. (1948). Mathematical biophysics (Revised ed.). Chicago: University of Chicago Press. Richardson, L. F. (1939). Generalized foreign politics. British Journal o f Psychology (Monograph Supplement). Cambridge: Cambridge University Press. Sapir, E. (1921). Language: An introduction to the study o f speech. New York: Harcourt, Brace. ______. (1925). Sound patterns in language. Language, 1, 37-51. ______ . (1927). Speech as a personality trait. American Journal o f Sociology, 32, 892-905. ______. (1933a). Language. Encyclopaedia o f the Social Sciences, 9, 155-69. ______ . (1933b). La realite psychologique des phonemes. Journal de Psychologie Normale et Pathologique, 30, 247-65. Shannon, C. & Weaver, W. (1949). The mathematical theory o f communication. Urbana: University of Illinois. Sullivan, H. S. (1940). Conceptions o f modern psychiatry. New York: William Alanson White Psychiatric Foundation. Tinbergen, N. (1953). The herring gull's world. London: Collins. Whitehead, A. N. & Russell, B. R. (1910-1913). Principia mathematica. Cambridge: Cambridge University Press.

4 Pseudounilaterality, Simple-Rate Variables, and Other Ills to which Interaction Research is Heir Starkey Duncan, Jr., Barbara G. Kanki, Hartmut B. Mokros, and Donald W. Fiske

Much o f nonverbal-communication research has been concerned with the use o f various actions, such as gaze and gestures, by participants in interaction as indi­ ces o f the participant’s more permanent characteristics or transient internal states. These data have typically been gathered either by counting the frequency o f the action in question or by timing its duration. Accompanying actions by the partner in this interaction are rarely considered. Ignoring relevant actions by the partner may introduce substantial error in the results and consequently in an interpreta­ tion o f the results. We have termed this error pseudounilaterality (PU). An ap­ proach to dealing with this problem is proposed.

As the study of face-to-face interaction develops, investigators con­ tinue to confront problems of extracting information from variables and dealing effectively with the surprising characteristics of interaction pro­ cesses. In general, one may distinguish two broad lines of research us­ ing interaction data. One line describes the process of interaction itself—how interaction is accomplished by participants. In this research investigators are typically concerned with discovering elements of the organization of interaction, similar to linguists’ attempts to discover el­ ements of the organization of language (i.e., grammar). Examples of This chapter first appeared in Journal o f Personality and Social Psychology; 46, 1335-48. Copyright © 1984 by the American Psychological Association. Reprinted by permission of the publisher. 71

72

Interaction and Identity

research of this sort are in Duncan and Fiske (1977, Part III), Kendon (1977), and Kendon, Harris, and Key (1975). The second line of research is concerned with using actions (often termed nonverbal behaviors) by participants in interaction as indices of the participants’ properties, characteristics, or transient internal states. For example, actions such as gaze, smiling, and speech hesitations may be used as indices of traits such as dominance or affiliation. This work on individual differences has been by far the larger portion of the inter­ action literature. Elements of this literature have been reviewed by Argyle and Cook (1976) and Harper, Wiens, and Matarazzo (1978), among others. This article concentrates on methodological issues in individual differences research. For the most part, standard psychological research techniques devel­ oped in earlier psychological work on individual differences have been applied to the study of individual differences in interaction. However, the properties of interaction indicate that unexpected and undesired conse­ quences may result from this seemingly natural extension. We consider several interrelated issues regarding extracting information from variables and interpreting results in studying individual differences in interaction.

Simple-Rate Variables Most studies of individual differences in interaction have used what may be called simple-rate variables. A simple-rate variable is one that is generated by counting or by timing the occurrence of an action during an interaction and dividing that number by some broader count or timing, representing the maximum frequency or total time that the action could have occurred. For example, in a large-scale study of two-person conver­ sations between adults, Duncan and Fiske (1977, Part II) used three types of simple-rate variables: rate, extent, and mean length. These types are illustrated in this article using speaker gaze as the action in question. Rate of gazing while speaking is calculated by dividing a speaker’s total num­ ber of gazes by the total duration of speaking. For extent of gaze, the total duration of speaker gazes is divided by the total duration of speaking. Mean length is derived by dividing the total duration of speaker gazes by the total number of speaker gazes. These, of course, are standard rate variables, entirely familiar in a variety of types of psychological research. One might consider their use in interaction research to be unexceptional.

Pseudounilaterality, Simple-Rate Variables, and Other Ills

73

For the purposes of this discussion, the significant characteristic of simple-rate variables is that, in representing only gross counts or tim­ ings of actions, these variables do not contain information on the se­ quences in which actions occur in interaction. The essential property of interaction, however, is that it is sequentially ordered. It is important not only that certain actions occur, but also where in the stream of interac­ tion they occur. We hold that a description of interaction processes must be based on observed sequences of actions between participants. For this reason, results based on simple-rate variables cannot be used to draw direct inferences concerning interaction processes. An example may illustrate this issue. Imagine that an investigator found a significant correlation between the extent of smiling by one participant and that by the partner, based on observing a number of interactions. The obvious, intuitive interpretation is that a strong tendency toward mutual smiling in the observed dyads exists. But this, of course, is an invalid inference from the simple-rate data because there is no information on smiling sequences in the variables. The same correlation could have been obtained if all of one participant’s smiling was during the first half of the interaction, all of the partner’s smiling was during the second half, and the amounts of these respective smilings were consistently proportional for the two participants in each interaction. Although this interpretation is improbable to say the least, the point is that it cannot be ruled out as a possible source of the correlation, based on the information available in the variable. We therefore conclude that simple-rate variables cannot sup­ port hypotheses concerning interaction processes. In the external-variable study reported in Part II of Duncan and Fiske (1977), this problem of using simple-rate variables to support these hy­ potheses was pandemic in the results. It became clear that none of the many correlations obtained in the study could be directly interpreted in terms of interaction processes. Duncan and Fiske suggested that simplerate variables (or external variables in their terms) be replaced by ac­ tion-sequence variables in research on interaction. We return to this suggestion later.

Pseudounilaterality As mentioned earlier, most nonverbal communication studies are not primarily concerned with interaction processes, but rather with individual

74

Interaction and Identity

differences. Even if simple-rate variables are inadequate for research on interaction processes, it does not immediately follow that they should also be problematic when used in research on individual differences. Yet this unfortunately appears to be the case. In studies of individual differences using nonverbal behaviors, some action (such as speaker gaze toward the auditor) is used as an index of some characteristic or internal state of the participant. A variable based on that action (i.e., speaker-gaze extent) is taken to reflect directly the action in question for the participant. However, because of the charac­ teristics of the interaction process, the variable may reflect instead ei­ ther a pattern of action by the partner or (more likely) a product of the actions of both participants. This indeterminacy of the variable obvi­ ously undermines the chain of inference linking the variable directly either to some internal property or to an interactional characteristic of the participant. In this way, serious error may occur in the interpretation of individual-difference results based on simple-rate variables. We refer to this sort of interpretive error as pseudounilaterality (PU), defined as the false assumption that the variable is necessarily unilaterally deter­ mined by the actions of the participant. The same point may be made in terms more familiar in psychological research. Many studies of individual differences in interaction using simple-rate variables have been conceptualized using a strong analogy to testing. The simple-rate variable, measuring the extent of use of some action such as smiling, is taken as a score on some dimension such as affiliation. Rate o f smiling would be considered similar to rates of cor­ rect or keyed responses to inventories. We believe that the analogy from tests to interactions breaks down be­ cause of both the sequential dependency of actions within interactions and the variability between interactions. In a testing situation, the test items (the stimuli) and the order in which they are presented are the same for all subjects. However, in naturalistic interactions one would expect variation between partners and in the development of each interaction. So in order to obtain meaningful information on a participant’s style in an interaction, it is necessary to consider the actions of the partner—the stimuli to which the participant is in part responding. Otherwise, information on the par­ ticipant’s actions represents some indeterminate combination of both participant’s and partner’s actions. Interpreting this combination exclu­ sively in terms of the participant is the PU error.

Pseudounilaterality, Simple-Rate Variables, and Other Ills

75

A typical use of simple-rate variables would be to study the extent of use of some nonverbal action for different groups of subjects. For ex­ ample, one might observe the extent of gazing while speaking by males and by females in two-person conversations. If extent of gazing is found to be greater for females than for males, then the investigator would straightforwardly interpret these results to mean females in the conver­ sations actually gazed more at the partner than did males. Similarly, one might study some actions by participants in each of two interactions. Change in the simple-rate variable for a participant from one interaction to the next would be interpreted as change in the participant’s actions in the two interactions; and conversely, lack of change in the variable would be interpreted as constancy in the participant’s actions. Regardless of the intuitive appeal of these inferences, we hold that they are not neces­ sarily justified when natural or uncontrolled interactions are involved. Although it seems counterintuitive, we hope to show that the informa­ tion available in the variable is insufficient to support direct inference about the behavior of the participant from the value of the variable. It is entirely possible that more extensive use of all available information on the interaction may reveal a much more complex state of affairs. As mentioned earlier, in a simple-rate variable the numerator is the action on which the variable is focused, such as number or duration of gazes or gestures. The denominator is a broader, more encompassing value, for example, time with the speaking turn or number of turns, a value that reflects the extent of interaction over which the target action might have occurred. Interpretive error may arise because in interaction research both the numerator and the denominator may reflect the actions not only of the participant in question, but also of the partner. A given variable may be deeply influenced by the actions of both participants. Consequently, it becomes extremely difficult to claim that a variable is the property of either participant, even though it is defined in terms of the actions of only one participant. Examples may clarify the potential interactive character of both nu­ merator and denominator in simple-rate variables. In order to achieve a direct illustration of this point, these examples have been drastically simplified. They are not intended to represent a typical process in ob­ served interactions. They may, however, suggest the interactive charac­ ter of supposedly unilateral actions.

76

Interaction and Identity

Numerator Affected Consider a variable concerned with speaker gaze. For the sake of example, the participant has an entirely uniform gaze pattern while speaking: an alternation of three seconds gazing away from the part­ ner, followed by three seconds gazing toward the partner. That is, from the beginning of the speaking turn and continuing throughout its duration, there is a regular six-second gaze cycle, consisting of a three-second gaze-away phase, followed by a three-second gaze-toward phase. The simple-rate variable in this example is extent of gaze while speaking; number of seconds gazing at the partner while speaking, divided by number of seconds speaking, summed over the interaction. This study is concerned with the amount of change in this variable for the participant from one interaction to another. In the first interaction it happens that the partner typically takes the speaking turn at the end of the first gaze cycle, after the third second of gaze toward the partner and just before the speaker would have begun the second gaze cycle. In this interaction the gaze-extent vari­ able would be about .50 (3/6). In the second interaction the partner typically takes the speaking turn during the first cycle no more than one second after gaze toward the partner begins. Here the gaze-ex­ tent variable would be about .25 (1/4). In each case, the partner’s pattern of taking the turn directly affects the length of the participant’s gaze as speaker (the numerator of the variable). (The denominator is also affected by the partner; it is smaller in the second interaction because the partner takes the turn sooner.) In this example, there has been no change in the participant’s gaze pattern, but there is a substantial difference in the behavior patterns of the re­ spective partners. Nevertheless, results of this sort might be interpreted as reflecting a major, unilateral change in the participant’s gazing be­ tween the two interactions. More subtle actions than the partner’s turn taking can affect the dura­ tion of the participant’s actions. Staying with the example of gaze extent while speaking, imagine an interaction pattern in which, during the speaker’s gaze, there is a tendency for the gaze to be turned away after an auditor back channel (e.g., “m-hm,” “yeah,” or a nod). Once again, different rates of auditor back channels in different interactions will af­ fect measures of speaker gaze duration.

Pseudounilaterality, Simple-Rate Variables, and Other Ills

77

Taken together, the two examples using gaze suggest that a given action by the participant may be affected by more than one action by the partner, making the total effect more complex.

Denominator Affected Staying with the case of stability of actions between interactions, let the action of interest be gesticulation. Once again, the simple-rate vari­ able will be extent: duration of gesturing divided by duration of speak­ ing. The participant in question has a completely stable pattern of gesturing. At the beginning of each speaking turn the participant ges­ tures for five seconds and then stops, having no further gestures for the remainder of the turn. The pattern does not change, either within each interaction or between interactions. In the first interaction the auditor always waits for five seconds after the end of the gesture and then takes the turn, resulting in a value for extent of gesturing of .50 (5/10). In the second interaction the auditor is more active, taking the turn one second after the end of each gesture. Gesture extent for the second interaction is .83 (5/6). The data suggest a 66 percent increase in extent of gesturing by the speaker, but the gesturing pattern and the total number of seconds of gesturing have remained constant. The change is entirely due to the difference in the responsiveness of the respective auditors. These examples illustrate false interpretations of change from one interaction to another. Obviously, the opposite false interpretation is also possible: A participant may appear to be relatively stable in some action from one interaction to the next but is actually producing rather different behavioral patterns that are being counterbalanced by the partner’s responses. Despite the artificiality of the examples, the principle should be clear and clearly applicable to more complex natural interactions. Speaking time, number of turns, and an undetermined number of nonverbal ac­ tions by a participant are not unilateral, but interactively determined. A variable involving trims or any affected action cannot be said to belong exclusively to the participant producing the actions; it is, rather, a prod­ uct of the joint action of both participants in a way that cannot be deter­ mined only on the basis of the simple-rate variable. Variables based on such interactively determined elements cannot legitimately be used as reflecting unilateral individual performance.

78

Interaction and Identity

Investigators may be surprised to find total time speaking, one of the most natural and commonly used denominators for simple-rate vari­ ables, a prime culprit in these examples. Total time speaking is, unfor­ tunately, one of the most interactive of measures. In many cases, one may view participant’s speaking time as a property more of the auditor than of the speaker. It is clear that the interpretation of results from many studies of indi­ vidual differences in nonverbal behavior may be subject to the PU error. A simple example is the early and frequently replicated result that fe­ male interactants gaze more at the partner than do male interactants. On the basis of available data from natural interactions, it cannot be deter­ mined whether the females are actually gazing more, the males are gaz­ ing away more, the partners of the males are taking the turn sooner after the onset of the males’ gazing, the partners of the females are taking the turn more during the females’ gazing away, or some combination of these factors is occurring. On the basis of the information contained in simple-rate variables, one cannot dismiss the possibility that the entire effect has nothing to do with actual gaze patterns by males and females, but rather with differential patterns of turn taking in response to gaze or nongaze by partners of males, as opposed to partners of females. Action-Sequence Variables The PU error has been illustrated using simple-rate variables. As pre­ viously mentioned, the PU error is not, however, limited to any single type of variable. The error is one of interpretation: attributing the value of a variable to the unilateral action of a single participant. It is clear that the PU error may be sharply aggravated by ignoring information on in­ teraction process. For this reason, the error may take its extreme form in the case of simple-rate variables because, by definition, these variables contain no information on interaction process; that is, information on the sequences in which actions occur. By the same token, one may seek to minimize the potential for PU error by incorporating as much se­ quential information in the variable as possible. Given the nature of interaction it is probably impossible to devise a variable that can be interpreted unilaterally with impunity, simply be­ cause in most cases it is probably impossible to interpret an action prop­ erly apart from the interactive context in which it occurs. No type of

Pseudounilaterality, Simple-Rate Variables, and Other Ills

79

variable can adequately capture deeply interactive phenomena. How­ ever, it does appear possible to take practical measures that allow the investigator to gain a much more differentiated view of each participant’s relative contribution to the interaction process. One approach to differ­ entiated information on participants is through analysis of a series of related variables, each of which carries complex interactional informa­ tion. Analysis of such a set of variables permits evaluation of a participant’s pattern of action in terms of the interaction matrix within which it occurs. Let us examine the kinds of sequential information that can be included in variables. Interaction variables containing sequential information were termed action-sequence variables in Duncan and Fiske (1977, Part IV, pp. 314 ff.). Action-sequence variables were designed to provide information on the location in the stream of interaction of each occurrence or nonoccurrence of the target action (again, such as speaker gaze). In thensimplest form action-sequence variables contain information on a participant’s action (e.g., speaker gaze) and some other subsequent or concurrent action by the partner (e.g., taking the turn). To this simple two-action sequence may be added further information on subsequent, preceding, or concurrent actions by both participants. Action-sequence variables are described in terms of: (a) the partici­ pants involved; (b) the aspects of actions relevant to the variable (e.g., onset, offset, or duration of an action); (c) the type of linkage between the sequential actions (temporal or event based, as explained later); and (d) the identification of actions as either independent or dependent. An independent action must occur in order for the variable to be observed and is placed in the denominator when rates are calculated. A dependent action may or may not occur, and it forms the numerator. Staying with speaker gaze as an example, we may illustrate these elements of an action-sequence variable. (More detailed description may be found in Duncan & Fiske, 1977, Part IV, pp. 314 ff.) Let the variable be concerned with the two-action sequence—the auditor’s taking the turn during the speaker’s gaze toward the auditor. In this case the inde­ pendent action is speaker gaze; the action sequence cannot be observed if the speaker is not gazing toward the auditor. The dependent action is the auditor’s taking the turn, which does or does not occur during speaker gaze. With respect to aspects of actions, the variable involves duration for speaker gaze (the speaker must be looking toward the auditor) and

80

Interaction and Identity

onset of taking the turn for the auditor. The linkage between gazing and turn taking is event based (rather than temporal); the speaker must be gazing during the turn attempt. To these two sequence variables (auditor’s taking the turn during speaker’s gaze or nongaze), we add information on whether the speaker’s gaze occurred at the very beginning of the speaking turn or during the course of the turn. Thus, we have both turn-initial gazes and turn-inter­ nal gazes, and turn-initial nongazes and turn-internal nongazes. We have by this time defined eight action sequences: initial speaker gazes during which the auditor took the turn, initial speaker gazes dur­ ing which the auditor did not take the turn, initial speaker nongazes during which the auditor took the turn, initial speaker nongazes during which the auditor did not take the turn, and four comparable sequences for internal gazes and nongazes. We term speaker gazes during which the auditor took the turn auditor-ended gazes, and speaker gazes during which the auditor did not take the turn speaker-ended gazes (the speaker gazes away before the auditor takes the turn). Similarly, we refer to auditor-ended and speaker-ended nongazes. It will be seen that we are treating nongazing at auditor as an action: The speaker is gazing, but elsewhere. We also treat not taking the turn as an action: The rationale is that such a nonoccurrence may be found to have meaning simply because the possible action might have occurred but did not.

Run-Sequence Variables To this point we have stayed entirely within the action-sequence frame­ work described by Duncan and Fiske (1977). We wish now to consider a further element: the duration of events in the sequence. This informa­ tion is important because it is useful to know not only how many times an event occurred, but also the duration of each occurrence. It obviously makes a difference whether, for example, a gaze was very brief or occu­ pied most of an interaction. A participant’s total gazing in an interaction may be made up of many short gazes, a few long gazes, or some com­ bination of these. The composition of gaze lengths may have important interactional consequences or be a result of important interactional events. A single occurrence of an action is termed a run of that action. Thus, we may speak of the duration or length of each run.

Pseudounilaterality, Simple-Rate Variables, and Other Ills

81

Duration can be measured either temporally or via some behaviorally defined unit. Consistent with all the work in our laboratory, the duration of runs in the data to be presented was measured in units of analysis, a complex event defined in terms of the speaker’s actions. The procedure for marking units of analysis was essentially the same as that used in Duncan and Fiske (1977, Part HI), except that we used a smaller num­ ber of speaker actions. Very briefly, in the present data a boundary of a unit of analysis was drawn at the onset of the first syllable following the end of a phonemic clause (Trager & Smith, 1957), during or immediately after which at least one of the following actions occurred: (a) a devia­ tion in the intonation from an essentially level pattern (2 2 1in the TragerSmith system); (b) the utterance of a sociocentric sequence (Bernstein, 1962), such as “you know”; (c) the completion of a grammatical clause, involving a subject-predicate combination; (d) the termination of a hand gesticulation or the relaxation of a tensed hand position (e.g., a fist); (e) an audible inhalation; (f) an unfilled pause (Maclay & Osgood, 1959); (g) a false start (Maclay & Osgood, 1959); and (h) the speaker’s head turning toward the auditor. More detailed description of these units of analysis may be found in Duncan and Fiske (1977, Part III). Units of analysis tend to be two to four words in length. Returning to the example of speaker gaze, we obtained a distribution of run lengths for each of the four gaze- and four nongaze-action se­ quences for each participant in each interaction. That is, for each gaze or nongaze sequence, we counted how many runs were one unit long, two units long, and so forth. Analyses of the eight speaker-gaze sequences are based on distributions of run lengths rather than on rates.

Sex Differences in Speaker Gaze Having considered the problem of describing the actions of individu­ als in interaction, and having introduced the notion of run-sequence vari­ ables, we are in a position to illustrate an alternative approach to the study of individual differences, using data on one of the earliest issues in nonverbal communication research: sex differences in speaker gaze. The analyses presented are not intended to constitute a thoroughgoing examination of speaker gaze, but only to provide an example suggesting the possibilities of run-sequence variables, analyzed using appropriate techniques, for pursuing individual differences in interaction.

82

Interaction and Identity

M ethod

Participants The data are based on transcriptions of sixteen two-person conversa­ tions between adults, taken from a larger set of conversations described in Part II of Duncan and Fiske (1977). Participants were recruited in groups of four, each group containing two females and two males. The participants were paired off for a conversation. Following that con­ versation, the partners were switched to form two new pairs. In this manner, each participant had two conversations: one with a partner of the same sex and one with a partner of the opposite sex. The order of these conversations was counterbalanced across groups. All interacting participants were previously unacquainted. The conversations were seven-minute, get-acquainted conversations. All participants were pro­ fessional-school students at the University of Chicago. In transcribing these conversations, speaker head direction was ob­ served instead of speaker gaze because the images on the videotapes did not permit accurate observation of gaze. However, head direction is taken to represent speaker gaze, and gaze is used in labeling the variables in order to facilitate exposition.

Log-Linear Analyses o f Run Sequences We believe analysis of variance (ANOVA) is representative of analy­ ses of simple-rate variables in nonverbal communication research. As mentioned earlier, in much of this research a variable such as speakergaze extent is taken as a kind of score, indexing to some degree a char­ acteristic of the participant, such as aggressiveness or affiliation. The score, summarizing action across an entire interaction, is treated as be­ ing on an interval scale. An alternative to the score approach is to consider the observed actions as discrete. A gesture either occurs or does not occur. A gaze is directed either toward the speaker or not. From this point of view, analysis focuses on counts of the occurrence of actions and the dura­ tions of each occurrence (together with other information on action sequences). Obviously, this approach is behind the development of run-sequence variables.

Pseudounilaterality, Simple-Rate Variables, and Other Ills

83

Although run-sequence data can be represented in at least two forms for a given variable, only one form is used in this analysis: run length, that is, the number of runs of each length. For example, there may be five runs that are two units long, and four that are three units long. To analyze counts of actions (typically cross-tabulated in terms of relevant variables), the log-linear approach seems the approach of choice. Log-linear analysis is based on odds (the frequency of being in a cat­ egory, divided by the frequency of not being in that category), not on percentages (as with simple-rate variables). Conditional odds are simi­ larly calculated, given one or more particular levels of one or more clas­ sifying variables. For example, conditional on the male level of the sex classifying variable, one can calculate the odds of run length equaling one, as opposed to run length equaling two or more (in these data, 335:694, or .48). Recent discussions of log-linear analysis include Bishop, Fienberg, and Holland (1975), Castellan (1979), Goodman (1978), Haberman (1978,1979), Knoke and Burke (1980), and Reynolds (1977). Table 4.1 presents information from our sixteen conversations on the distribution of run lengths for each of the eight run sequences described above. The data have been organized in terms of the illustrative issue of sex differences in speaker gaze, permitting us to ask if there are sex differences for some variables and not others. Table 4.1 permits analysis of eight, two-way frequency tables, where sex and run length are the crossed variables. Inspection of table 4.1 indicates a number of broad effects on the length of gaze runs, which we do not pursue here because they lie out­ side the sex difference focus of this illustration. For example, initial runs of speaker gaze are more likely to be auditor ended than are inter­ nal runs. Initial auditor-ended runs of gaze tend to be shorter than inter­ nal, auditor-ended runs. It is apparent that these data might be fruitfully analyzed for phenomena other than sex differences. Definition o f table size. For a given type of gaze sequence, run length can theoretically range from one unit to any finite number, up to the total number of units a participant had the speaking turn. In these data the maximum run length was twenty-three units. It may be observed in table 4.1, for each run-sequence variable, the modal run length was one unit, and the longer runs were relatively infrequent. It is also true that, as the run lengths become longer, fewer participants contributed to the data. It seems important to identify a point for each run sequence beyond

Table 4.1 Distribution of Run Lengths for Males and Females on Eight Speaker-Gaze Run Sequences R u n len g th Sex

1

2

3

4

5

M ales F e m ale s

55 46

19 35

21 18

20 17

15 7

M a le s F e m ale s

7 13

8 4

3 2

3 1

1 0

M a le s F e m ale s

15 22

12 5

10 9

2 6

3 3

M a le s F e m ale s

42 60

31 41

20 23

8 14

4 6

M a le s F e m ale s

91 81

80 67

33 44

24 29

13 14

M a le s F e m ale s

10 26

5 5

3 4

2 0

0

M a le s F e m ale s

81 84

66 55

41 37

22 27

13 8

M a le s F e m ale s

34 39

39 25

26 16

21 13

24 12

7

6

8

>13

T otal ru n s

10

II

12

13

0 1

1 0

| 0

1 0

0 1

0 0

0 0

0 0

0 0

0 1

3 0

47 51

1 3

2 0

no 161

9

1. In itia l, sp e ak e r-en d e d n o n g a z e 8 4

4 0

2 2

2 0

149 130

2. In itia l, a u d ito r-e n d e d n o n g a z e 1 0

0 0

0 1

1 0

24 22

3. In itia l, s p e ak e r-en d e d gaze 0 2

1 1

| 2

4. In itia l, a u d ito r-e n d e d g aze 0 8

| 4

1 2

5. In te rn a l, sp e ak e r-en d e d n o n g a z e 6 8

5 1

3 1

4 2

2 2

0 0

0 0

0 1

2 0

26 3 250

6. In te rn a l, a u d ito r-e n d e d n o n g a z e 1 0

1

2 0

24 37

1 1

7. I n te rn a l, s p e ak e r-en d e d g aze 3 13

7 3

3 2

2 2

0 4

0 2

1 0

1 0

5 2

2 5

0 0

0 2

0 0

240 237

8. In te rn a l, a u d ito r-e n d e d g aze

Note. Numbers in boldface type represent the cutoff points for analysis.

4 8

15 9

1 7

1 5

172 143

Pseudounilaterality, Simple-Rate Variables, and Other Ills

85

which data are considered “outliers,” so that the analysis is not unduly influenced by a small number of very long runs or by the few partici­ pants responsible for these long runs. For present purposes, the analysis is focused on the more frequent events. Because the distribution of gaze by run length varies substantially among the different types of run se­ quences, it is ill-advised to apply a uniform definition of outlier to all eight types. In line with these considerations, we developed the follow­ ing, arbitrary rule to eliminate outliers from each run sequence: Starting from a run length of one unit, the cutoff followed the run length that included 85 percent of the cumulative total number of runs (males and females combined). When applied to the tables, this rule seemed to pro­ vide reasonable cutoff points. The following eight table sizes were there­ fore used in the analysis: 1. 2. 3. 4. 5. 6. 7. 8.

initial, speaker-ended nongaze: 2 x 5 ; initial, auditor-ended nongaze: 2 x 4 ; initial, speaker-ended gaze: 2 x 5 ; initial, auditor-ended gaze: 2 x 4 ; internal, speaker-ended nongaze: 2 x 4 ; internal, auditor-ended nongaze: 2 x 3 ; internal, speaker-ended gaze: 2 x 4 ; and, internal, auditor-ended gaze, 2 x 7 .

Table 4.2 Possible Models for the Sex x Run Length Contingency Table Model

Parameters/hypothesis

Degrees of freedom

(a) Saturated model (SR)

Sex, Run Length and Sex x Run Length interaction/all possible parameters included; model is untestable Sex and Run Length/no association

0

(b) Independence model (S, R) (c) Single-parameter model (R) (d) Single-parameter model (S) (e) Equiprobability model (0)

(s-D (r-l)

Run Length/equal probability for S margin Sex/equal probability for R margin

s(r - 1)

Grand mean/equal cell probability

sr- 1

Note, s - number of levels of sex; r - number of levels of run length.

r ( s- 1)

86

Interaction and Identity

Results

Model Fitting In one approach to dealing with cross-tabulated data such as our 2 * N tables, models are specified that contain various effect parameters. These parameters include some subset of the classifying variables, for example, sex and run length in our data, and interaction terms, for ex­ ample, Sex x Run Length. Given the model’s parameters, expected fre­ quencies are generated that can then be compared to the observed frequencies. A specified model is evaluated in terms of how well its parameters explain or fit these data. One can evaluate the adequacy of a model by computing a Pearson chi-square (%2) or preferably the likelihood ratio (G2) statistic from the expected and observed odds. G2 = 2 1 f.jlnif.JF..), where f.. is the ob­ served cell frequencies, F.. is the expected frequencies, and the log is to the base e. The likelihood ratio is preferable to the Pearson chi-square because “(1) the expected frequencies are estimated by maximum like­ lihood methods and (2) [the likelihood ratio] can be partitioned uniquely for more powerful tests of conditional independence in multiway tables” (Knoke & Burke, 1980, p. 30). Table 4.2 lists the full set of possible models for a two-way table. A model containing all possible effect parameters is called the Saturated model (line a) and is not a testable model because the expected and observed odds are identical. But once any parameter is omitted from the model, one can evaluate the adequacy of the remaining parameters. For example, if a model omits the Sex x Run Length interaction term and still fits the data well, one can conclude that the (Sex, Run Length) Inde­ pendence model (line b) is adequate. One step down from the Indepen­ dence model would permit the omission of either one of the effect parameters Sex or Run Length, thus leaving a Single-Parameter model (lines c or d). When the expected frequencies are calculated from either of these models, the various levels of the omitted parameter are assumed to be equiprobable. For example, the odds of a dichotomous variable would be 50:50. The final model in table 4.2, the Equiprobability model, omits both effect parameters so that the only remaining parameter is the grand mean. Thus, every cell of the resulting table of expected frequen­ cies will contain the same value: N/ total number of cells.

Pseudounilaterality, Simple-Rate Variables, and Other Ills

87

In model fitting one is not rejecting a null hypothesis such as the hypothesis of no association in the usual %2test of independence. Rather, the purpose is to accept a model, that is, find a low G2 value relative to degrees of freedom (in general, p > .10). When more than one model fits, the best one is not always an obvious choice. Typically a simpler model is preferable on grounds of parsimony, but there are other sources of information that may guide the final choice and conclusions. Generally speaking, the higher level objectives of a study together with aspects of the research design become especially crucial in the pro­ cess of interpreting results. In log-linear analysis these factors also weigh heavily in the process of designating what subset of models is relevant or legitimate to consider, given the research goals and given the con­ straints imposed by the sampling. Because the issue here is one of sex differences, the Equiprobability and (Run-Length) models are not really directly of interest. The (Sex) model, on the other hand, has obvious relevance but fails to fit any of the run sequences (see table 4.3). If the single-parameter model (Sex) had been adequate for any of these cases, it would be very strong support for sex differences because no other variable or variable interaction is necessary to generate expected values that closely approximate the observed frequency distribution. In testing the fit of the Independence model, we are examining sex differences inso­ far as they are interacting with run length; if the model does not fit, we can then look for different run-length patterns conditional on sex. For our data, the Independence model adequately accounts for the observed frequencies in all run sequences except Run Sequence one.

Single-Parameter Effects Model fitting is not the only way to reveal sex differences. One can also directly assess single-parameter effects by taking the G2difference between two models that differ by the single parameter in question. (The right half of table 4.3 summarizes these effects for each run sequence.) In this way, the difference in G2 between the (Sex, Run Length) and (Run Length) models is equivalent to the single-parameter effect of Sex with one degree of freedom. Similarly, the difference in G2 of (Sex, Run Length) and (Sex) is the single-parameter effect of the Run Length with degrees of freedom depending on the number of run-length categories being analyzed. Finally, the single parameter effect of the Sex x Run

88

Interaction and Identity

Length interaction is determined by subtracting the G2 of the Saturated model (the only model that contains the Sex * Run Length interaction) from the Independence model. Given these differences between models, we see that the single-pa­ rameter effect of Sex is significant in Run Sequences four, six, and eight. Table 4 3 Likelihood Ratio Chi-Square (G2) Values for Models and Single-Parameter Effects: Run Sequences 1-8 Model fitting Model6 (0) (S)

(R) (S,R) (SR) (0) (S)

(R) (S,R) (SR) (0) (S)

(R) (S,R) (SR) (0) (S)

(R) (S,R) (SR) (0) (S)

(R) (S, R) (SR)

Single-parameter effect

G2

df

75.58 75.39 9.07 8.87

9 8 5 4 0

G2 P Run Sequence 1