32
Conversation Analysis and Sociology
1 Introduction
Sociologists have long acknowledged the significance of language for social life. For Marx (Marx & Engels, 1964 [1845]: 42), language represented “practical consciousness.” For Durkheim, it was a prime example of a “social fact” (Durkheim, 1982 [1895]), enduring and transcendent of individual members of society. For Mead (1934), language was the vehicle by which shared symbolization and a reflexive sense of self emerges in the life of each individual. For Parsons (1937, 1951), language was part of the cultural system, the means through which norms and values are transmitted across generations. Yet, despite these acknowledgments, the role of language in human affairs has historically been registered more by mention than by investigation.
This inattention may be traced partly to a disciplinary division of labor forged in the early 20th century in which the study of language was treated as the proper preserve of Linguistics which took language rather than, for instance, language use as its primary object. In part, too, it may be traced to a belief that the details of language use should be studied in an idealized fashion because they are too random and disorderly to sustain principled empirical investigation (Chomsky, 1965). A parallel position with regard to action was present in Sociology. Social theorists had a longstanding tradition of concern with the theory of action rather than its actual analysis (Parsons, 1937), reflecting a similar view of the feasibility of such study (Sacks, 1984a). These views systematically positioned the study of language use as unstudiable either within Linguistics or Sociology.
In the 1960s, however, Harvey Sacks and Emanuel Schegloff began a process of dissent from these viewpoints. They did so not by arguing that Sociology should focus on the study of language per se, but rather by taking up the notion that, from a sociological point of view, the primary significance of language was as a vehicle for social action that can be studied in its concrete particulars.
As Sacks saw it, human interaction is a site of massive organization and order. This order is robust enough that young children who encounter its instantiation in different sets of parents, relatives, and acquaintances will nonetheless undergo socialization into a common set of cultural and interactional practices (Sacks, 1992 Vol 1: 485). And it is robust enough to support the analysis of particular, singular instances of interaction by social scientists because, after all, it must support the particular, singular analyses that the participants use as the basis for moving forward in the interaction itself.
In this chapter we begin with a discussion of how Conversation Analysis emerged as a distinctive form of sociology. From there, we discuss what Conversation Analysis contributes to the sociological enterprise. We conclude with a discussion of future directions for CA work in Sociology.
2 The Emergence of CA in Sociology
2.1 The Sociological Background
To understand the ways in which CA represented an innovation within sociological thinking, it is necessary to appreciate the predominant tenor of American sociological analysis in the 1950s—a decade or so before CA’s emergence. At this time, the field was infused with a form of positivist methodology that stressed the need for objective measures of social phenomena and that insisted that social processes could be understood without reference to the meanings of social action (Hempel, 1952; Nagel, 1961). This positivist outlook was reinforced, from Psychology, by the methodology of behaviorism which declared all subjective attributes of persons to be unobservable and therefore unamenable to scientific analysis. Shadowed by this viewpoint, even those who, like Parsons, attempted to follow Weber (Weber, 1968) in insisting on the relevance of social action to sociology, found difficulty in proposing causal efficacy for such “unobservable” subjective characteristics as norms and values (Parsons, 1937). At most, these characteristics were documented in the aggregate in attitude surveys.
With the occlusion of concrete social action from sociological analysis, so too was there an occlusion of the specific reasons and reasoning of individual social actors. To take a model that was treated as exemplary, Durkheim argued that suicide was to be analyzed exclusively in terms of correlations between suicide rates, measures of social isolation, the business cycle and so on, and that any attempt to understand the concrete reasoning of suicidal individuals was irrelevant (Durkheim, 1951 [1897]: 100).
The cumulative significance of this perspective was its delimitation of sociological inquiry to social behavior in the aggregate, thus excluding all micro-level analyses of social behavior and any aspect of the reasoning that might inform it. Yet suicide, and indeed other social conduct, is not done ‘in the aggregate’. It is done by particular persons in particular circumstances with particular underlying reasons. Indeed, subsequent lines of dissent would come to focus on precisely these issues: aggregate analyses of conduct could hardly be offered when the conduct in question was not understood in its concrete particulars.
2.2 Lines of Dissent
While many intellectual developments in the 1950s and early 1960s contributed to the erosion of the positivist paradigm—the linguistic turn in Philosophy (Rorty, 1967), an emerging focus on cognition in Psychology (e.g. Bruner & Postman, 1949; Miller, Galanter & Pribriam, 1960) and, within Sociology, a growing dissatisfaction arising out of difficulties with the attitudes-action link (Deutscher, 1973; Wicker, 1969)—two major lines of dissent were pivotal in the development of CA. The first of these arose in the work of Harold Garfinkel. Garfinkel argued, following Schütz (1954), that social objects and events are not self-subsistent states of affairs but are actively constituted in consciousness. This claim undermined the view that social processes could be understood without reference to the understandings and reasoning of social actors and promoted the idea that the everyday social world is, first and last, a matter of ordinary shared sense making. This sense making is thus fundamental, unavoidable and primordial to the social world.
In examining how this sense making works, Garfinkel began by looking at how everyday objects and events are recognized and acted upon. Drawing on the gestalt phenomenology of Gurwitsch (1964), he argued that the process of recognition is essentially a circular one in which objects and their contexts are construed in a mutually elaborative process that he called, following Mannheim (1952), the documentary method of interpretation. For instance, the remark “that’s a nice one” uttered by a customer to a greengrocer while pointing at a watermelon, may be understood as expressing a request to buy it. The same utterance uttered while looking at a picture of a friend in her photograph album, may be understood as a compliment. In each case, the utterance is contextually elaborated from a statement into a specific social action by virtue of such matters as what is being pointed at, the speaker, the addressee and the social context in which it is said (Heritage, 1984b). This contextualization is part of the active effort to achieve the shared understandings that Schütz had pointed to.
Nonetheless, to truly show that this process is productive of shared understandings, Garfinkel engaged in a wide variety of quasi-experiments designed to document that social actors hold each other accountable for performing the active work involved in achieving common understandings (Garfinkel, 1967b). For instance, in one such quasi-experiment, confederates in interaction with another insisted that subjects clarify everyday statements (e.g. “I had a flat tire.” “What do you mean you had a flat tire?”). These well-known breaching experiments were highly effective in generating responses of incomprehension and anger (e.g. “What do you mean ‘what do you mean’!”) and vividly demonstrated the great extent to which social actors hold one another accountable for implementing the documentary method to make sense of actions in context.
The upshot of these studies was that joint understandings of everyday situations are the product of the application of shared methods of reasoning about objects and events in context. Actions are produced and positioned so as to be understood in context, and actors rely on this both to understand the actions of others, and to produce understandable actions themselves. This notion lies at the core of Garfinkel’s program. Conversational interaction is not exempt from these reflexive methods; indeed Garfinkel drew some of his central illustrations of the process from interactional data.
The second line of dissent stemmed from the work of Erving Goffman. Like Garfinkel, Goffman started from the issue of sense making. Early in his career, Garfinkel (1952: 357) wrote that:
The big question is not whether actors understand each other or not. The fact is that they do understand each other, that they will understand each other, but the catch is that they will understand each other regardless of how they would be understood.
Erving Goffman adopted much the same position. In “On Facework” (1955: 213), after describing the notion of a line, he wrote that
Regardless of whether a person intends to take a line, he will find that he has done so in effect. The other participants will assume that he has more or less willfully taken a stand, so that if he is to deal with their response to him he must take into consideration the impression they have possibly formed of him.
As this quote makes clear, Goffman understood lines as inferences drawn from the choices that actors unavoidably make in interaction. Also central to Goffman’s analysis, however, is the notion that claims about face and identity are implemented in and through these choices, and that self-presentation is an unavoidable feature of social life.
For Goffman, the choices that actors make are meaningful, made so by the existence of dense networks of norms which shape understandings of how actors should conduct themselves in any given setting. In “The Neglected Situation” (1964), Goffman observed that it is difficult if not impossible to connect the characteristics of actors to the nature of their actions without a clear understanding of the normative organization of the setting in which their actions take place. In this sense action is always contextualized and rule oriented.
In contrast to Garfinkel, however, Goffman pursued the idea that the normative order of interaction could be conceived as a social institution in its own right. He termed this institution the interaction order (1983), viewing it as a layer of normative organization undergirding the operation of all other social institutions. In a fundamental sense, we can think of the interaction order as anchoring the conduct of actors as they enact social roles by providing for a fine-grained order of accountability, in terms of which they produce and regulate their own behavior and understand and evaluate the behavior of others. The idea of an interaction order sui generis is a thoroughly Durkheimian one in the sense that the order is conceived, like language itself, as existing prior to persons and their motivations, and indeed as structuring and sustaining them. Goffman conceived this order as providing for fundamental human wants: (i) negative face wants (the desire to act unimpeded by others), and (ii) positive face wants (the desire for the recognition and affirmation of others) (see Brown & Levinson, 1987 for further development of these concepts). Viewed in these terms, the interaction order represents a broad balancing of both types of wants among a range of actors.
The work of Garfinkel and Goffman was key to the development of CA. Garfinkel’s fundamental achievement was to make sense and sense making central to the study of social interaction and social worlds. Within this perspective, Garfinkel placed human experience, reasoning and interaction as clearly within the remit of sociological thinking. This contribution was critical in facilitating the formation of Conversation Analysis. Goffman, on the other hand, created both a framework within which interaction processes could be situated and showed how the interaction order could itself be situated among other social institutions. Both thinkers contributed in a foundational way to the creation of a new sociological space in which the analysis of social behavior could be developed. Indeed, in the work of Garfinkel and Goffman, we arrive at the threshold of Conversation Analysis. Significantly, however, neither offered a clear direction or the tools with which to proceed.
3 Conversation Analysis as Sociology: Toward a Sociology of Interaction
Walking across the threshold was a two-fold process. The first step was an incremental one consisting of an integration of the separate contributions of Goffman and Garfinkel into a single unified theory of social interaction. The second and more radical step involved the creation of a distinctive methodology that would permit a systematic approach to the details of interactional organization and nail down observations about social conduct that had previously only been sketched.
To begin with, it was necessary to counter the general sociological position mentioned earlier that the details of interaction are random and disorderly, constituting background noise from which real underlying dimensions must be extracted prior to analysis (Bales, 1950). In his lectures, Sacks repeatedly rejected this view. In paraphrase, Sacks’ view was that “it is perfectly possible … to suppose … that wherever one happens to attack the phenomenon one is going to find detailed order. That is, one may alternatively take it that there is order at all points” (Jefferson, 1983b: 1; Sacks, 1992 Vol 1: 484). The consequence of this view formed one of four pillars of CA—that all behavior should be examined under the assumption that it is orderly, communicatively meaningful and distinctive in terms of the construction of social action.
The second pillar of CA is every bit as innovative. It is the belief that social actions are produced, in the first instance, by reference to their immediate local interactional context. Thus, every instance of social action is to be analyzed in terms of the particular local environment of previous social actions in which it is produced. Associated with this position is the notion that more macro-sociological influences on conduct work their way into interaction in and through the local processes that inform the construction of action (Schegloff, 1987b, 1991b). This position was, and remains, the first attempt to escape the airless abstraction in which sociologists had tended to treat the nature and influence of action. Focusing on the production of social action ‘in the here and now’ represented a sustained and systematic effort at such an escape.
The third pillar of CA is the idea that the details of social interaction can be understood in terms of structural organization and that this structure is anterior to and shapes the construction of action in interaction. This notion is possibly the most straightforwardly sociological aspect of CA, essentially inherited by CA from Durkheim through Goffman. Much of this structure is normative: interactants are understood to produce and recognize one another’s social actions via a rule-guided system in terms of which they hold one another accountable.
The final pillar of CA concerns intersubjectivity. CA is committed to the notion that understanding in interaction is, in the first instance, produced and owned by the participants in interaction, and generated as an endogenous feature of interaction. For example, accepting an invitation is a second speaker’s way of indicating that s/he understood the prior social action to be an invitation, and accepting that acceptance is a first speaker’s way of confirming that. This focus on the primacy of participants’, rather than analysts’, understandings involves the recognition that no matter what analysts’ understandings of the situation may be, participants will conduct themselves on the basis of their own understandings. Thus, as Garfinkel (1967b) insistently recommended, external understandings of situations that are at variance with those of the participants, cannot function as substitutes for what actors understand to be the case.
3.1 From Theory to Method
These four pillars collectively constituted an innovative approach to social interaction, and in turn dictated a unique methodology for its empirical study. With the assumption that social interaction may be orderly at a high level of detail, only recordings of precisely what, when and how the interaction occurred are acceptable. Recordings allow for the repeated inspection of the data both by the analyst and by any other who wishes to form his/her own independent judgment in a way that observation filtered through the analyst does not.
The view that actions are constructed first and foremost relative to the local context mandates the use of collections of candidate instances of a phenomenon occurring across speakers and situations in order to validate candidate analyses of that phenomenon. The idea that social interaction is inherently rule-guided suggests that, in situations where the norms or rules thought to be in operation for a given phenomenon are departed from, the departure should be oriented to as consequential. Thus, a key component of CA as a research method involves deviant cases which work not as outliers but as instances that ‘prove the rule’. Finally, the fourth pillar which treats participants’ own understandings as having primacy relative to analysts’ understandings, shows through most plainly in the CA ‘proof procedure’ which requires that an analysis of any given phenomenon be grounded in participants’ orientations to the relevant order.
Together, these methodological components form a rigorous empirical method with which to study the underlying structure of the micro-processes of social life whether that be in everyday conversation or in institutional contexts. These theoretical and methodological innovations of CA are distinctly and distinctively sociological.
3.2 Contemporary CA Research
Contemporary CA research is primarily focused at two levels. The first level concerns the discovery and specification of interactional practices for a given local sequential context (see Sidnell, this volume). A practice is any aspect of action that (a) has a distinctive character, (b) has a specific location within a turn or sequence, and (c) is distinctive in its consequences for the nature or meaning of the action in which it is implemented. By design, this concept describes characteristics of action that are independent of participants’ individual, personal or psychological characteristics. Practices involve specific and recurrent selections in the design, construction and implementation of turns at talk. Practices operate at multiple levels in interaction from prosody to word selection to turn organization and action construction.
At the level of word selection, the offer Would you like some dessert? is more strongly built for an affirmative response than the alternative Would you like any dessert? Here, the insertion of some or any into a polar question is a practice of question design that works to tilt the question toward a positive (or negative) response (Heritage, et al., 2007; Horn, 1978).
Similarly, at the level of action construction, there are different practices for answering polar questions. Interactants routinely select between two dominant answer types—interjections (e.g. yes, mm hm, no) and repetitions. The question Is Tom coming?, can be answered, for instance, with Yes (or some variant such as Yeah) or Tom’s coming. The interjection merely affirms that Tom is coming, leaving the agency for the question with the questioner, whereas the repetition actually asserts that Tom is coming, thereby claiming a greater degree of agency over the assertion’s content than the interjection (Heritage & Raymond, 2005, 2012; Raymond, 2003; Schegloff, 1996c).
At a different level of organization, there are practices for negotiating the closure or continuation of sequences. For instance, sequence closing thirds (Schegloff, 2007b) such as Oh, Great, or Okay are recurrently used to propose sequence termination. Relatedly, gaze withdrawal at a sequence boundary is associated with sequence closure whereas gaze continuation is associated with sequence expansion (see Rossano, 2005b, this volume).
The second level of contemporary CA research focuses on how practices form organized clusters. Some clusters of practices address the management of conversational organization itself. Turn-taking is an example. Given the conversation analytic finding that, overwhelmingly, one person speaks at a time, there is an inherent scarcity of opportunities to speak (Sacks, Schegloff & Jefferson, 1974). As a result, some social mechanism is required to allocate those opportunities. An array of conversational practices focus on the constitution of turn-constructional units (the units of talk that the system allocates), and on their allocation among potential speakers. In the aggregate, these practices constitute a turn-taking system (see Clayman, this volume). Similar clusters of practices are associated with such domains as repair (see Kitzinger, this volume) and reference (see Enfield, this volume).
On the other hand, other organizations are concerned with the management of objectives and social relationships transacted through conversation such as the management of particular kinds of actions and their organization in sequences (Schegloff, 2007b; Stivers, this volume), the associated management of affiliation in social relationships (see Pomerantz & Heritage, this volume; Lindström & Sorjonen, this volume), and the related knowledge distribution among participants (see Hayano, this volume; Heritage, this volume; Lee, this volume).
Each of these domains of organization addresses problems that are intrinsically social and for which solutions must be found within the constraints of a larger social order (Schegloff, 2006a). These include problems of who has rights to talk at a given moment; problems of speaking, hearing or understanding; and problems of positioning vis-à-vis another in terms of knowledge and agreement. These problems are consistently resolved on the ground and ‘in the here and now’. The solutions to these problems are not ad hoc but are the systematic product of the kinds of practices detailed above. It is for exactly this reason that Conversation Analysis has both an impetus and a subject matter that are intrinsically sociological in character.
4 Conversation Analysis in Sociology
As we have already observed, one of the founding ideas of CA is that an institutionalized interaction order is the basis not only of social interaction but of social institutions (Drew & Heritage, 1992a; Goffman, 1983; Schegloff, 2006a). It follows from this that the institutionalized practices making up the interaction order are a foundational and held-in-common reference point for members of a society. There is no female turn-taking system nor a low-SES repair system. Quite the contrary: all current evidence suggests that practices which organize this conduct are shared across gender, race, SES, and other differentiating categories in social life. Indeed, part of a conversation analytic theory of social interaction is the conception that there is a level of order in social interaction that transcends class, race, gender and culture and that operates across the species (Schegloff, 2006a; Stivers, et al., 2009).
On the other hand, it is also a fact that any individual occupies many distinct and sociologically relevant categories simultaneously: not only her race, gender, class and sexuality, but also her status as, for instance, a sister, a lawyer, a daughter, a mother, and so on, all at one time. How then are we, at any given moment in an interaction, to say which of these aspects of identity, if any, is relevant to any particular interactional practice? In accordance with standard CA methods, to argue that any one of these categories is relevant at a given moment in interaction, that category membership must be shown to be the object of the participants’ orientation in the interaction (Schegloff, 1987b, 1991b, 1992c). Yet, most of interaction proceeds without these invocations coming to the surface of the conversation despite the fact that we may intuit that race, class gender or sexuality is somehow affecting the interaction. Coupled with the observation that there is no evidence that any of the basic practices discussed vary by such factors as race, gender or class, the question is how can we gain leverage on a hypothesis that there are differences.
Two main responses to this question have emerged within CA. The first has focused on interactions in which transparently institutional tasks are being pursued by interactants who occupy institutional roles such as doctor and patient, teacher and student, lawyer and witness (Drew & Heritage, 1992b; Heritage & Clayman, 2010). The basic assumption here is that, whatever other roles and statuses are occupied by the participants, the task-based social roles are primary. Thus, this research focuses on the fundamental interactional obligations of, for instance, doctors and patients, while paying relatively little attention to the fact that these persons are also male or female, Black or White, and so on. This institutional talk program has examined ways in which the fundamental organization of interaction is inflected in the context of these tasks and the social roles associated with them. Basic aspects of conversational organization such as turn-taking and sequence organization can be inflected, as can practices that are dedicated to more particular conversational contingencies.
For example, in courts (see Komter, this volume) and news interviews (see Clayman, this volume) interaction is conducted through a modified turn-taking organization that restricts witnesses and interviewees to answering questions, and lawyers and interviewers to asking them with significant consequences for initiative and control in those settings. And in terms of sequence organization, although interaction in some institutional contexts, such as medicine, involves the same sorts of sequence closing thirds (e.g. question → answer → “Okay” or “Good”) that are characteristic of ordinary conversation, in other contexts (e.g. news, courts and mediation hearings), commentary after a question-answer sequence is effectively barred by the institutional demands of the setting which require questioners to position themselves as merely eliciting information, not evaluating it.
Just as institutional contexts can be instantiated by variations in turn-taking and sequence organization, so those contexts, once instantiated, can impact how particular interactional practices are understood. For example, the familiar question How are you? asked at the beginning of an ordinary interaction, is not designed to elicit a detailed account of the recipient’s ordinary affairs, though it allows for the pre-emptive topicalization of extraordinary ones (Jefferson, 1980b; Sacks, 1975; Schegloff, 1986). For precisely this reason “Fine.” is a canonical response, allowing the participants to move on to the ‘business’ of the interaction. In a medical setting, by contrast, patients who visit the doctor for an illness are in a conundrum since they are, by definition, not “fine” and may not wish to present themselves as such. To do so might suggest that the patient’s illness is insufficiently serious to warrant the medical visit (Heritage & Robinson, 2006a). Yet, as a way of moving into the business of the interaction in ordinary interaction, the practice mandates just such a response. Moreover, responding with a list of symptoms to How are you?, even while sick, may attract the inference that they are too readily adopting the sick role (Parsons, 1951). Not surprisingly, patients are visibly challenged by this question in the medical visit context (Heritage & Robinson, 2006a; Robinson, 2006a).
If this first ‘institutional’ line of investigation focuses on task-based roles that are argued to be central to particular social institutions, a second line of investigation focuses on roles and statuses that are more diffusely present in social interaction and present, moreover, in both ordinary conversation and in task-based institutional contexts. In research that falls within this second line of investigation, two kinds of evidence have been developed: (i) qualitative demonstrations of participant orientations to a particular social status, and (ii) distributional evidence arrived at through mixed methods research that combines conversation analysis with survey research. We discuss each in turn.
4.1 Social Status and Interaction: Participant Orientation as Evidence
Demonstrations of participant orientations to particular social statuses often arise from the reflexive nature of referring expressions: the fact that referring expressions frequently embody a relationship between the person referred to and the person doing the referring (Whitehead & Lerner, 2009). Early on, Sacks (1979) observed that selecting “cops” (rather than “police”) was a method for teenage boys to index a common rebellious identity. Similarly, a response to an instruction with “Yes ma’am”, indexes both that the instructor is female and that the recipient is positioning him/herself as a subordinate (Hopper & LeBaron, 1998). Moreover, when this acknowledgment is responded to with “You don’t have to call me ma’am” (Hopper & LeBaron, 1998: 64), it is clear that the respondent is resisting one or both of these aspects. Through this we see an orientation by the respondent to the social categories indexed through “ma’am.” A similar reflexive relational consequence can emerge from the use of racial or ethnic categories. For instance, when a speaker refers to one of four marines as “dark” (and later “colored” and still later “Black”), she evokes the relevance of her own racial identity and, not infrequently, the sharing of that racial category with her interlocutor (Kitzinger, 2005b).
In the preceding examples, the orientation of a speaker to a social category is indexed by the selection of a descriptive term where no such term was required or when there were alternative terms available. However, there are circumstances in which the rules of a given language constrain a speaker’s freedom of maneuver. For example, when a speaker is referring to another individual (i.e. not the addressee) for a second or Nth time (locally subsequent position, Schegloff, 1996c), s/he is required to use a pronoun, and in some languages pronouns are unavoidably gendered (around 30% of languages, according to Siewierska, 2011). The potential omnirelevance of gender in these languages is thus sometimes matched by the actual near omnirelevance of gender in referring expressions. Moreover, across languages, initial references to persons follow a norm that if speakers are not making use of a name, they should use an expression that positions the referent as closely as possible to the speaker or recipient, for instance, my wife or your sister-in-law (Brown, 2007; Enfield, this volume; Stivers, Enfield & Levinson, 2007). Many of these expressions refer to kin and are, therefore, particularly in some languages, frequently gendered. This system of referring expressions has profound consequences for persons in gay and lesbian relationships who must either rely on alternative referring expressions such as partner or utilize classically heterosexual kin terms (e.g. wife, mother-in-law), something which can lead to confusion and misunderstanding as to the sex of the partner (Land & Kitzinger, 2005).
A similar argument has been made for the contrast between English and Lao in terms of person reference. English person reference leaves hierarchy virtually entirely unmarked. There is a norm favoring the use of names for initial recognitional references, and even when kin terms are used, there is no marking of status. By contrast, in Lao, references to third parties, although also typically reliant on names, obligatorily mark the age or hierarchical status of the referent vis-à-vis the speaker (Enfield, 2007b). Enfield observes that, as a consequence, speakers unavoidably index a cultural stance toward social relationships—whether that is in the direction of egalitarian relationships (e.g. bare names in English) or hierarchical ones (e.g. prefixed names in Lao).
The point to be taken from these latter studies is that systems of interactional practices encode forms of social and cultural order. The systems may do so ‘invisibly’, as Kitzinger describes in the “heteronormativity” of the person reference system and thus tend, albeit passively, to reproduce these predominant forms (Kitzinger, 2005a).
Racial categories are not always observably oriented to in the course of interaction. However, ethnographic evidence may support an association between racial/ethnic categories and interactional practices. For instance, in Los Angeles during the 1990s, when tensions between Korean shopkeepers and African American customers were particularly high, African Americans complained of a lack of respect by Korean shopkeepers while these shopkeepers complained of rudeness and a lack of self-restraint by African American customers. Bailey conducted studies of these interactions (1997, 2000) and showed that in a majority of Korean-Korean interactions, interactions were terse and unexpanded. By contrast, a majority of African American customers initiated additional sequences and incrementally elaborated their turns in pursuit of uptake by Korean shopkeepers. For African Americans, the fact that Korean shopkeepers neither initiate nonbusiness sequences, nor respond significantly to their initiations, is to be understood as indicating a lack of respect. For Koreans, the interaction is properly to be understood as a service encounter that should not extend into the personal domain. Here, the inter-cultural communication problem is due to a lack of access to and understanding of a systematically different set of norms to which the other is oriented (see Roberts, Davies & Jupp, 1992 for a range of further examples). It is striking that racial tensions, whatever their other causes, can be fueled by relatively basic interactional practices.
4.2 Social Status and Interaction: Distributional Evidence Through Multiple Methods
In looking at the intersection between interactional practices, on the one hand, and, on the other, social categories such as race, gender or sexuality, typically taken from survey data, combining CA with quantitative methods has been an approach that has met with success. Such an approach is driven by the view that although practices for, for instance, selecting next speaker or initiating repair may be the same across populations, these practices may be deployed with more or less frequency in a given population.
For instance, in the context of medical interaction, Stivers & Majid (2007) examined practices such as selection of next speaker through speaker gaze, address terms or knowledge domain, and then coded these behaviors systematically across physician questions in 322 encounters. This allowed the testing of associations between the frequency of speaker selection and socio-demographic variables such as parent race, education level and gender. They found that physicians select next speakers using the same basic practices across adults and children, men and women, and different race and class backgrounds. However, physicians deploy gaze more frequently with children than adults and are less likely to select children than adults to answer their questions, particularly when the parent is Black or when the parent is less educated and Latino (Stivers & Majid, 2007).
Stivers & Majid suggest that the selection of children to answer questions is an important mechanism for socializing the child into the role of independent patient. This observed bias in the selection of children by race has, as a consequence, that Black children and Latino children of parents with less education, are less likely to receive as much socialization as their counterparts from White and better educated families. We can speculate that this could be a contributing factor to those individuals being less proactive patients when they reach adulthood.
In a related study of pediatrician-parent interaction when children had upper respiratory tract infection symptoms (runny noses, sore throats, coughs, etc.), Mangione-Smith and colleagues found that whereas parents from most racial/ethnic backgrounds would sometimes resist treatment recommendations (53 of the 285 non-African American parents), none of the 40 African American parents resisted a physician’s recommendation (Mangione-Smith, et al., 2006). This suggests that African American parents may indeed be less assertive in this sort of institutional context.
Similar methods have been used to examine associations between gender and interactional practices. For instance, Kollock, et al. (1985) looked at relationships between gender, power and interruption using conversational data from same-sex male, same sex female and opposite sex couples. They examined whether, as previous literature had asserted, gender was predictive of interruption (i.e. men being more likely to interrupt than women) or whether dominance might be the underlying predictor. They used interview data to assess each partner’s judgment of dominance within the couple. They found interruptions to be a function of power position, rather than sex, in all of the couples. Interruptions were no more common in men or women when the couples judged there to be no power difference between them. Here the integration of interview data and conversation analysis allowed the researchers to disentangle the relevance of gender and power in a way that would otherwise be difficult if not impossible.
In a different domain of conduct, presidential news conferences, Clayman and colleagues looked at the characteristics of journalists that were associated with a tendency to question the president in an adversarial fashion (Clayman, et al., 2012). Adversarial questions were identified in terms of whether the question was assertive (i.e. ‘tilted’ in favor of a particular answer), whether the question articulated a position that was opposed to the president or his policies, or whether the question asked the president to account for his policies (see Clayman & Heritage, 2002a). The journalists’ characteristics that were studied included whether the journalist worked for a newspaper versus for television, whether s/he worked for an “elite” news organization or not, whether s/he was a frequent questioner of the president, and whether s/he was male or female. The findings were that female questioners asked significantly more adversarial questions than their male counterparts until the time of the Nixon administration when the Watergate crisis led to a wholesale upward adjustment of adversarial questioning among the entire White House press corps (Clayman, et al., 2010).
4.3 Social Interaction and Outcomes
While all of the studies discussed in the last section document the relationship between social characteristics and the distribution of interactional practices, a related set of studies documents the relationships between interactional practices and social outcomes. So far this research has been conducted in the context of medicine. We briefly summarize two of these studies here. One study examines the effects of question design on whether patients’ medical concerns were sufficiently dealt with during a medical visit (Heritage, et al., 2007). Focusing on patients who had multiple reasons for visiting the doctor (identified through a pre-visit survey), the study showed that how the physician asked about these additional concerns mattered for whether or not they were ultimately disclosed. After the patient’s initial problem presentation, if physicians elicited additional patient concerns using the question “Is there anything else you want to address in the visit today?”, patients were nearly twice as likely to respond with No than if they were asked “Is there something else you want to address in the visit today?” Moreover, these same patients were three times more likely to leave the doctor’s office with one or more concerns unaddressed. Here, a simple feature of question design has large-scale consequences for patients, who leave the office with the possibility of a significant medical condition and with continuing anxieties which may necessitate a further medical visit. Both possibilities will tend to increase personal costs (i.e. deductibles and co-payments) and overall medical system utilization.
A second study linking interaction practices and outcomes looked at treatment decision making in the context of children with upper respiratory tract illnesses. Previous research established that physicians are more likely to prescribe antibiotics inappropriately (i.e. for viral infections) when they perceive parents to expect antibiotics. The open question was what drives these perceptions. A number of interaction practices are involved, including how the parent describes his/her child’s problem at the beginning of the visit (Stivers, 2007b). In some cases the parent only identifies the child’s symptoms (e.g. “runny nose,” “sore throat”); in other cases the parent also offers a “candidate diagnosis” of the underlying cause of the symptoms (e.g. “sinus infection” or “strep throat”). Physicians are significantly more likely to perceive a parent to expect antibiotics if she offers a candidate diagnosis, which, in turn, translates into inappropriate prescribing (Mangione-Smith, et al., 2006; Stivers, et al., 2003). In a context where inappropriate prescribing is driving up the incidence of antibiotic resistant bacteria, this study shows that interactional practices play a significant role in this process. Taken together, these studies demonstrate that, through combining CA with quantitative approaches, relationships between outcomes and social interactional practices can be identified.
5 Future Directions
At the outset of this chapter, we distinguished between CA as Sociology and CA in Sociology. The first of these identifies basic sociological processes through which social interaction is organized. At this point CA has accumulated a substantial number of findings which, taken collectively, point to a theory of action. To be sure, the theory is far from completely specified, but it is already possible to see an outline of what a more fleshed out theory would look like. The theory would take the sequential organization of interaction as its starting point and develop analyses of the practices and their organizations through which fundamental issues in human relations are played out. It will undergo revision in the context of cross-cultural studies of interaction and be examined for its implications for a variety of fields including cognitive psychology, child development and human evolution.
Several areas of research seem, at the present time, to be relatively productive of growth. These include how social actions are organized in sequences, especially those which are not produced under the control of adjacency-pair structures (see Heritage, this volume; Stivers, this volume). How speakers design their turns (e.g. word selection, grammar, prosody) has been considerably and productively studied in recent years, but presently lacks a unifying framework that will draw these studies together. Solid work on person reference has yet to be matched by comparable studies of practices for referring to objects, places and time (see Enfield, this volume). Finally, practices for managing knowledge and its ownership have begun to be investigated but much more remains to be discovered (see Heritage, this volume). In sum, CA is developing an empirical theory of action for Sociology—something that has been considered a primary goal of the discipline at least since Weber (Schegloff, 1996a).
With regard to CA in Sociology, solid research exists that links CA methods and findings to research in medicine and mass media. With now established combinations of CA and quantitative methods, there is substantial opportunity for extending research within these areas as well as developing research on the sociology of the family, of childhood and of organizations. The use of CA in a mixed methods approach to sociological analysis makes it possible to build links between socio-demographic variables, on the one hand, and interactional processes, on the other. However, these relationships are presently under-studied, and we see much room for development. Finally, with the growth of interest in Sociology in the culture-action nexus, there is serious scope for CA investigations of the interactional constitution of cultural capital, collective action and religious commitment.
NOTE
We thank Jack Sidnell for his comments on an earlier draft of this chapter.