Navigation – Plan du site

AccueilNuméros22Communication as the Study of Soc...

Communication as the Study of Social Action: on the Study of Language and Social Interaction

An interview with Anita Pomerantz and Robert E. Sanders, by Nicolas Bencherki
Anita Pomerantz, Robert E. Sanders et Nicolas Bencherki
p. 103-118
Traduction(s) :
La communication comme étude de l’action sociale : au sujet de la recherche sur le langage et l’interaction sociale


In this interview, Anita Pomerantz and Robert E. Sanders, professors emeriti at the University at Albany, SUNY’s department of communication, discuss their views on conducting language and social interaction (LSI) research. They share their understanding of the connection between LSI research and the discipline of communication, and explain what we may gain from focusing on social action instead of solely studying messages. The relationship between an interaction and its context, and the way the latter may be relied on to analyze the former, is also discussed. To offer their insights, Pomerantz and Sanders draw from their active engagement, since the 1970s, as prominent voices of the language and social interaction community, and more particularly as figureheads of conversation analysis (CA).

Haut de page

Texte intégral

1Anita Pomerantz and Robert E. Sanders are both Professors Emeriti of Communication at the State University of New York at Albany. They have been colleagues there for many years, and have collaborated on several projects. It was at the University at Albany’s Department of Communication that I interviewed them about some of the ideas and goals of their work on the way people participate in social interactions and what they accomplish in doing so.

2Anita Pomerantz is one of the foremost representatives of conversation analysis (or CA). Over the course of the past forty years she has drawn attention to interactional details of the way specific communicative actions are done in naturally occurring social interactions, such as extreme case formulations, providing candidate answers in seeking information, pursuing a response, and so forth. During the American Sociological Association’s annual meeting in Montreal, in August of 2017, Pomerantz received the Ethnomethodology and Conversation Analysis section’s Garfinkel-Sacks Lifetime Achievement Award, recognizing that she first formulated many of the field’s core ideas.

3Robert E. Sanders’ work addresses both empirical and theoretical topics under the broader umbrella of language and social interaction, mainly in language pragmatics and intercultural communication as well as conversation analysis. His work focuses on social influence and strategy in naturally occurring talk-in-interaction, which he refers to as “neo-rhetorical” participation in social interaction. Sanders served as the editor of the journal Research on Language and Social Interaction from 1988 to 1998, and chaired the Language and Social Interaction divisions of both the National Communication Association (NCA) and the International Communication Association (ICA).

4The work that Pomerantz and Sanders have published separately and jointly spans a variety of topics: seeking information (Pomerantz), enacting institutional roles (Sanders), ways of responding to assessments (Pomerantz), compliance-seeking (Sanders), giving instructional feedback to medical residents on their communicative practices (Pomerantz), cultural variation in directives (Sanders), and engendering and averting conflict in jury deliberations (Pomerantz and Sanders).

Interview by Nicolas Bencherki,
his interventions are in bold type.

5For many researchers outside LSI [Language and Social Interaction], LSI seems to focus on the minutiae of communication, not the ways in which communication matters in social life. I hope in this interview we can clarify what the work of LSI is about and what it aims to contribute to our knowledge about communication.

6Let me start with this question: How does LSI research differ from other approaches in communication?

7Sanders – To get at that, we have to go back about 50 years. LSI grew out of the upheaval that happened in many social sciences in the ’50s and ’60s. In one way or another, there was sudden appreciation that the macro-phenomena that were being studied are all rooted in micro-occurrences and processes. Chomsky [1957] got at this regarding the production of language. Garfinkel, Sacks, and Goffman got at it regarding the production of social order [Garfinkel, 1967; Sacks, 1992; Goffman, 1959]; Hymes and Gumperz got at it regarding the production of culture [Hymes, 1974; Gumperz, 1982]. Take concepts like roles, identity, stereotyping, and such. Yes, people have roles, yes there are stereotypes. But these aren’t “out there,” ready-made and static, like suits of clothes. They materialize and take form in real time as people interact. Or maybe better to say, they’re re-enacted each time from the ground up, and in the process they can change and evolve. Michael Moerman was an LSI-oriented ethnographer who, referring to culture [Moerman 1988], put it so well I memorized it: “Culture’s being exists in interaction’s time.”

8Here’s an example. I analyzed videos of interactions between supervisors and subordinates, and found that those roles don’t look the same from person to person, or situation to situation [Sanders, 1995]. They’re worked out interactively. One subordinate-supervisor pair for example enacted their roles together in textbook fashion—he consulted, she directed. But another supervisor was routinely walked on by subordinates. They gave lip service to his authority. They were rude, controlling, almost insubordinate, and he let that happen. I’d say he made that happen through ways he interacted with subordinates. If upper management only knew how little this guy controlled what went on in his unit! Back to your question, LSI studies the communicative processes that create, transform, maybe corrupt what other approaches study as a finished product.

9Pomerantz Let me build on the point that Bob just made—that LSI researchers show how roles, identities, etc., are constituted and re-constituted through interaction. In a study that I co-authored [Pomerantz and Rintel, 2004], we show that the ways in which doctors report the patient’s blood pressure, temperature, blood sugar, etc. both reflect and propose the physicians’ roles of paternalism or empowering partners. In contrast to the assumption that roles are relatively stable over the course of encounters, we treat role enactments as matters that are negotiated turn by turn in interaction.

10On occasion people make the distinction between approaches that study macro-phenomena and those that study micro-phenomena. When they do that, they regard LSI in general, and CA in particular, as studying micro-phenomena. This characterization misses the nature of LSI and the CA enterprise. CA studies the reasoning and methods that participants use to do things in interaction and understand what others are doing, their actions. Participants engage in conduct and interpret others’ conduct in terms of actions, e.g. inviting a friend to lunch, disagreeing with another participant’s opinion, or accusing a neighbor of a wrongdoing. The concept of actions is fundamental to how participants behave and, consequently, to LSI researchers’ analyses.

11Sanders – Anita’s point about macro-micro is important. I hope I didn’t make it sound like an either-or proposition. It’s more like both-and. You can think of it like the relation between biology and chemistry. Chemistry gets at processes and structures that underlie and produce biological processes and structures. This means that to understand how some of what biologists study comes about, you have to turn to chemistry. Chemists don’t concern themselves with biological phenomena. But biologists sometimes have to go to the chemistry to fully understand what’s happening. What I said happened in the social sciences in the ’50s and ’60s is as if some biologists—not all of them by any means—realized they needed to look at the chemistry of what they were seeing. Voilà, you get microbiology! Or in our case enterprises within LSI like CA, ethnography of communication, language pragmatics, and discourse analysis.

12Anita, you said in LSI the concept of actions is fundamental to how you examine what is going on when people interact. If we focus on social actions instead of messages, what do we gain?

13Pomerantz Research on young children shows that they understand gestures and talk in terms of social action. In other words, from an early age, they understand that a pointing gesture means, for instance, “look at it,” or “go there.” Social action is therefore a basic unit of interaction. It is how we do things and how we understand the things we are doing. The study of messages, as I understand it, came from a very different place. The model consists in having a self-contained unit of meaning, and looking at whether it gets “communicated” or is distorted in one way or another. This may be a valid stream of research, but it does not handle what we are basically interested in, which is how people coordinate action with one another and negotiate on matters of interest.

14Sanders – I don’t think it’s right to say we study actions instead of messages. It’s closer to say we see “communicating a message” as one kind of action. As Anita said, action is the basic unit of meaning when people communicate. I think of a “message” as what a communicator has inside that gets externalized, to communicate it, to make it known. Actions are a matter of why I say something, to what end, including why I externalize this or that as my message, the purpose of it. What gets done by communicating it right then, in that situation, to those persons. Said otherwise, communicating a message is just one possible purpose for saying something, one possible action that gets done. There are others. If I say to you “Your glasses are dirty!” I may simply want you to know that your glasses are dirty. That’s one possible purpose for my having said that, and so it’s both my message and my action. What you come away with is what I wanted you to know. But maybe I said it to insult you. Insulting you is another action, it’s another purpose for my saying that, but I wouldn’t refer to my having insulted you as a message. What you come away with is what I did. If you want to object, “Wait a minute! The message there is I don’t like you,” I’d say fine, but the difference is I didn’t say it, I showed it through my action. The bottom line is: what we gain by focusing on action is to be more comprehensive about what occurs when people communicate than we can be by focusing on message alone.

15Pomerantz In addition, the way communication is taught, using a message exchange model, is very much individual-based. The underlying idea is that one person has some cognition that they are trying to communicate, and it is that cognition or idea that constitutes the message. As a result, research often consists in asking people about what was in their head and what they intended to communicate.

16Sanders – I’d go further and suggest that the concept of message, as a technical term, needs a lot of work. Anita led me to see that years ago, at a conference where I first heard her speak. She was in the process of transitioning from sociology into the communication field, and in one part of her paper she was sharing her puzzlement about the concept of “message.” She asked something like “When you use the term message, what are you referring to?” Her question brought me up short, and I’ve chewed on it. I’ve come to realize how riddled with trouble the term and the concept are. It’s sometimes used to refer to the main, underlying, thought or feeling that someone is externalizing. Sometimes it refers to the result of that, the entire text or artifact they produced, wherein you find “message variables.” We’re trapped in everyday language that makes it hard to see this. The term “communicate” alludes to getting something across, and therefore lends itself to the concept of message, more than it does to action. Maybe that’s the source of puzzlement about how our concern with action fits into a concern with communication.

17Are we still doing communication research, though, if we are focusing on what people are doing rather than on what they are saying? Is LSI providing a different perspective on the same empirical phenomena, or does its focus on action call for a different set of things to be studied?

18Sanders – Again, I don’t see it as an either-or proposition. Our attention to action is value-added, not an alternative. Focusing on what people are doing through communicating gives us additional things to be studied, things we think are more basic. We’re getting at something that attention to messages alone misses. In doing something by speaking to you, this is something I’ve done either to you, or with you, or about you, or for you. No matter which, my action engages us with each other. You’re then faced with having to do something about it, whether to reciprocate, push back, endorse or oppose. Or maybe just thank me. Whichever, it’s inherent in the concept of action that there will be a reaction, and that’s a key aspect of communication that’s missing from the concept of message. Actions are the glue that binds messages to responses once they’re recognized as also, more basically, action and reaction. It explains how communicating is a matter of being engaged with others, not just bouncing thoughts and feelings off them. We get interaction, there’s no such thing as inter-messaging.

19Pomerantz Communication is broad enough to embrace different perspectives, and that is why language and social interaction flourish as an area within the discipline. There are historical traditions within communication research and the study of messages is certainly one of them. However now, LSI has come to be recognized as another important perspective within the discipline.

20One way to see how LSI fits within the communication discipline is to consider the area of Interpersonal Communication as an important subfield of the discipline. When I taught Interpersonal Communication, I renamed the course Interpersonal Interaction. In doing so, many of the same topics were covered but from a different angle. Interpersonal Communication courses often include, among other topics, perception, with the interest of comparing how the relational partners’ perceptions compare to each other, and relationship building, with an interest in how similar or different in attitudes, interests, etc. the partners are. In contrast, an LSI approach examines how people in relationships talk and act with each other as a way of constituting their relationships.

21Sanders – Exactly. To continue on this example, how do two people know that they have similar attitudes? It’s through interaction. For example, it’s been theorized that people experience lots of uncertainty when they meet someone new, and under certain conditions will be motivated to reduce that uncertainty. Therefore, they process the new person’s messages with the goal of finding out about them and reducing their uncertainty. But this is flawed. Uncertainty reduction is essentially an interactive process, in which both people are engaged in trying to learn about the other and at the same time guide what the other person learns. This is what those two people are doing, jointly, through their talk. This tendency to focus on individuals and their messages also spills over to issues that have come up regarding interviews as a research method. Many studies assume that the interviewee is merely reporting information about what is inside their head. Arguably, though, the person being interviewed is trying to figure out what the interviewer wants from them. The interviewer guides that and possibly makes adjustments as the interview proceeds. Said otherwise, interviewer and interviewee are interacting, and that surely makes a difference in what the interviewee chooses to reveal. Besides asking for information and providing it, other actions may be engaged in during an interview in the questions and in the answers, some aimed at doing a presentation of self. This recognizes that communication is not done by message producing and message processing automata, it is done by humans engaged in joint action.

22Speaking of action, how can language and social interaction researchers describe what the actions that are taking place add up to? In particular, given the very minute attention to turn-by-turn details of interaction, how can analysts zoom back out to the broader picture of what is going on in a sequence?

23Pomerantz I don’t think it’s a problem for LSI researchers to understand the activity or activities going on in a sequence. Just as participants understand the activity they’re engaged in, so do analysts have that understanding. But that’s not quite the issue that you’re raising. You’re asking how LSI researchers who examine interaction turn-by-turn can understand the larger activity as the context for the turn-by-turn actions. Generally, CA researchers, when they start to examine a piece of data, want to hear/see as much of the interaction as is available, and identifying the activities the participants are engaged in over the course of the interaction usually is unproblematic.

24Let me try to answer your question another way. Different LSI and CA researchers are interested in different orders of phenomena, and the broader picture of the activity the participants are engaged in can be anywhere from tangential to essential. Therefore, it’s important to appreciate the researcher’s project goals and the kinds of claims he or she is working to make. I often write papers about a particular strategy people use for seeking information, providing information, disagreeing, etc. These strategies may be identifiable in single turns at speaking. But even for analyses of strategies performed within single turns, part of the analyses involves describing when and where they’re done, examining the context in which they’re performed, and checking the consequences of doing them.

25Sanders – It really is a matter of analytic choices. When you say, “How can we zoom out to a broader picture of what’s going on in an interaction?”, I ask: why you would want to do that? What are you talking about when you ask: “what’s going on?” As Anita indicated, we know a lot about what’s going on from listening to the whole interaction before the analysis starts. Also, from the context or occasion, and what’s being talked about. What’s going on might be a news interview, a negotiation, delivering health care, arranging a night out, teaching a lesson in school, etc. Our interest is to find out the details of how people make such goings-on work, and how they participate in them, it isn’t to find out what such goings-on are. For example, Doug Maynard [1997] examined interactions between physicians and patients where what was going on was that the physician had to give bad news for the patient. Maynard’s interest was in regularities in the way that physicians go about doing that. He found that a common way they do it is to lead into the bad news indirectly, setting up the delivery of information the patient is waiting for in a way that prompts the patient to anticipate where this is going, and giving voice to the bad news him—or herself. I think that finding that there’s this collaborative way in which a difficult situation gets handled is knowledge worth getting about communication. Zooming out to what the situation is and what’s difficult about it takes us away from the communication involved unless we’re using that as simply the backdrop to the communication that takes place.

26Another tendency in communication research, as opposed perhaps to the emphasis on cognitive states that you mentioned, is to reduce communication and action to its context—whether the immediate physical context, the historical context, the economic context, and so forth. A common critique addressed to LSI is precisely that it ignores context.

27Sanders – To address that issue, I’ll go back to the conference where I met Anita. It was a conference that brought communication researchers together with prominent conversation analysts and ethnomethodologists from sociology, including Emanuel Schegloff, Harold Garfinkel, and other luminaries. Among the speakers was a conversation analyst who was interested in education, Hugh Mehan. He was studying parent-teacher conferences about the placement of special needs children into one program or another that they offered, such as for the dyslexic or the learning disabled. These placements were supposed to be based on each conference between parents and staff. But Mehan learned that the school district had quotas for how many kids could be in each program, and the distribution of placements was exactly according to quotas. So, he had an apparent anomaly. Placements were supposed to be based on individual conferences, so how did they end up in line with quotas? Mehan concluded parents were being steered to endorse certain placements, and went into the interactions to see how this was being done. Schegloff took issue with this, insisting that unless Mehan first found something in the interactions that did not look kosher, and then started looking around for why this would be going on, and then found the quotas, his analysis was self-fulfilling. By coming at the data with the issue of the quotas in mind, Mehan was almost guaranteed he was going to find something in the talkbut something he possibly read into it. In short, bringing context in may be tricky if that’s where you start.

28Pomerantz For me, it’s a matter of how much you can maintain your integrity when you do the analysis. For instance, Bob and I studied conflict during jury deliberations about what the sentence should be after voting to convict the defendant in a murder trial [Pomerantz and Sanders, 2013; 2014]. Obviously, we knew that our data came from jury deliberation. Even that brings in context. That alone departs from being absolute purists about not using context. Should we have ignored what we knew about the case over and above what came up in the talk? I don’t see how that’s possible. I think there are degrees, and it’s more important to maintain a skeptical attitude when visiting and revisiting the data. You have to ask yourself, “How much evidence do I have? Is there contrary evidence?” You need a good dose of integrity in doing analytical work to save you from the pitfalls of knowing the context.

29Sanders – We must, indeed, keep in mind that there are dangers, as context can lead you to misinterpret, over-interpret, read into, impose on the data something that’s not there. It’s true, you can’t ignore context, but your reliance on it has to be motivated. You have to specify where it affects the talk exactly. For example, Schegloff [Schegloff 1988] gave a lot of theoretical importance to a little piece of an interaction between mom and her sons. It starts with mom telling her sons she wants to talk to them about where she’s going tonight. One of the sons answers “I know where you’re going.” Mom then asks “where?” and her other son indicates he knows there’s a meeting of parents and teachers at their school. Mom then asks him, “Do you know who’s going to be there?”, and the kid says, “Who?”, to which the mom answers “I don’t know.” The kid then names people who would be there. So, he did know! Why did he first respond as if he didn’t? Schegloff’s interest was in the way the kid must have understood the question in order to answer it that way, a way that theories of speech actions at the time had overlooked. Schegloff contended that some questions like Mom’s can be used as a pre-announcement, to set up announcing a piece of news. Understood that way, the kid could assume this was information he didn’t already have that she did, so he asked to be told the answer. However, I believe there is more to why the kid understood it that way, and that to get at it, you have to look back at the context in which the question was asked, in this case the context within their interaction. Mom had started by asking questions as a way to find out what her son knew (“I know the answer. Do you?”). When she found he did know something about this event, she seems to have shifted to asking a question to find out something she didn’t know (“I don’t know the answer. Do you?”). Her mental context for asking her question changed, and that changed the action her question counted as, but she didn’t register that shift outwardly, in the talk. Therefore, her son answered the question based on the context created by their talk up to then, as if she was still probing what he already knew, so maybe he thought on this one she knew something he didn’t. What we learn from this is that context isn’t some objective background for communication, it’s something we bring into play and manage in our talk. In this case, the talk was revealing of their context, and that at one point they had different contexts in mind. I think that that’s what Schegloff was getting at in responding to Mehan. Don’t import context unless there’s evidence of it in the talk.

30Pomerantz An extreme way of guarding against letting context analytically bias would be that your transcript would only identify speakers as A, B, and C, instead of, say, mom, son one and son two. Three speakers with no identities and no other elements of context. Then, it would be only if you can recognize that this one is speaking as a mother that you get to identify that speaker as mom. I don’t take that extreme position. Bob may disagree with me, but I think ethnographic information sometimes helps. Still, some researchers use it too quickly and jump to conclusions. You should be extremely careful, but it’s nice to have it available.

31Sanders – I don’t disagree, actually. I just think it has to be motivated. We can take for instance the study you did, Anita, on candidate answers [Pomerantz, 1988]. The data were about school attendance officers calling students’ home to see if the parents knew why their kid was not in school that day. Anita’s interest was in how the question was formulated, and why it was formulated that way. In that case, it matters that it was an attendance officer asking about an absent child. In that context, she was asking her question in a way that expected a benign explanation, and wasn’t accusatory, yet it could imply that the kid had sneaked away from school without the parent knowing. You had to know that much. Context does matter, and my own view is that people, when they are interacting, are engaged in some kind of activity, like mom and her sons, which is in itself a context. They are striving to get something done that they can only achieve together. As an analyst, you have to know something about what they are doing, what the task or the activity is. The danger is that, if you don’t take the time to properly motivate bringing in your prior knowledge of the task, you may bring too much of that knowledge into your interpretation of what is going on.

32Pomerantz To say the same thing differently, sometimes you need to know something of the context even to understand what is going on in the interaction. However, how much of it is relevant to what you are finding out analytically is a whole different question.

33If a researcher wants to adopt the more careful approach to context and reconstruct it from the interaction itself, how do they go about doing that? Is it even entirely possible?

34Sanders – I did a study of interactions between pairs of children in the 5 and 6-year-old range [Sanders, 2007] that I intended to be a study of the way they handled conflict. I created a context for them to interact in that would produce conflict. I gave each pair a set of Lego building blocks and told them I wanted them to make one thing, working together. Much to my surprise, the context I created didn’t engender conflict. I suspect it’s because an additional aspect of that context was that each pair of children was conspicuously on display as we recorded them, and they were inhibited. When they did have conflicting ideas about what to make, or how to make it, one child or the other in each pair came up with some very clever ways, strategic ways, to push for what they wanted without getting into a fight in front of us. The way I expected the context to show up in the talk didn’t happen, but it gave me a lens through which to recognize the importance of what did show up. That young children can find ways to communicatively offset a fight. In one instance, a brief quarrel did break out when one girl got frustrated after 10 minutes by her partner’s single-mindedness. But it was mainly done in sing-song which took the anger out of it, and it was about who would be sleeping over at a mutual friend’s house, not what was going on in their play just then. It seems that kids at that age are much more adept than we realized at talking strategically when they have to.

35Pomerantz Your question has two separate but related parts. One part is about what it takes to be careful about using context in one’s analysis. The other part is whether it’s useful to infer the participants’ identities, roles, activities, etc. from only interactional materials. Let me answer the parts in reverse order.

36Yes, it’s useful. Bob and I have already said that we find knowing aspects of the context useful for understanding what the participants are doing and what they’re talking about. We’re not trying to pretend we know nothing. Given that it can be useful to know some aspects of context over and above the interactional context, I’ll go back to the first part of your question. How can one work with knowing context without letting it squelch good analytic work. An example is the study I co-authored about ways doctors reported test results to patients and how patients responded [Pomerantz and Rintel, 2004]. We started with knowledge of the context: the participants’ role identities were doctor and patient. Through repeated viewings of these interactions, we found no segments in which they were enacting other role-identities (e.g. friends), so we didn’t modify those basic role identities. However, when we examined the activity of giving/receiving test results, the role-identity of doctor was too gross to capture the differences in the ways different doctors reported test results. Consequently, we came up with finer descriptions of their roles, based on the talk.

37Sanders – There are certain things you have to know about the context because of the data, but you are, exactly, beginning from the data, and then finding out from that what else you need to know to make sense of it. That was Schegloff’s stance in response to Mehan. As we’ve been saying, the danger is to begin from your own ideas about context prior to finding evidence of it in the talk. But let’s also be clear that there isn’t necessarily one objectively correct context of each interaction that the talk will reveal. Pearce and Cronen [1980] posited that there are many concurrent, overlapping and stacked, contexts within which people interact, such as the context produced by the talk itself, the one produced by participants’ relational history or roles, the institutional context, and so forth. The different contexts that may be inferred from the same interaction can each be valid. But whichever such context you find evidence of in the talk and apply in your analysis can affect what actions you identify and what claims you make. For instance, in an interaction we analyzed between boyfriend and girlfriend, you could say there were at least two contexts through which to come at their interaction [Sanders and Pomerantz, fc]. One was a boyfriend-girlfriend relational, context. The other was a cultural-normative context that gave rise to a conflict between them. We focused on the second one, and what each of these people did to defend a practice [hunting] or enforce a norm [not kill animals] against a breach. We learn something from this about how tricky it is to uphold norms when they’re breached. We didn’t get into the relational aspects of this conflict. If we had, we’d have noticed some details as being important that we didn’t attend to. All we’re saying is that the context you bring to bear in an analysis must begin with the data.

38Pomerantz People can lead you to their context. In the case of the boyfriend-girlfriend interaction Bob referred to, the topic of their interaction was on what the boyfriend had killed on the hunting trip he just returned from. We did not have to determine whether the girlfriend was a member of some group, say anti-hunting people, to infer a lot on where she is on hunting. We learn such things from the way she talks about what he reported and reacts to his accounts.

39Sanders – When Anita and I studied a jury’s deliberations, we knew a lot about the context. Some of that was in the talk, some of that was in the public record, some of that was what we know about the jury system in this country by having grown up here. This helped us understand what was being talked about, but we didn’t use what we knew about the context as a basis for searching the talk for something we believed in advance we’d find in it.

40Pomerantz Right. When I said you have to be careful, part of what that means is that while you may know something ethnographically about the interaction’s context, knowing it does not mean it’s relevant.

41If LSI analyzes single interactions or comparable fragments of interaction, how can LSI’s concern with remaining at the level of interaction, and presuming as few things about context as possible, be reconciled with the need to produce scientific knowledge that can be applicable to more than a single setting?

42Sanders – Well, wait! LSI doesn’t generally use single interactions as its data source. Doing that would make it impossible to know if we’ve seen something that occurs between other people in other interactions. It could be unique to those people, in that instance. Yes, sometimes claims are based on a single interaction, but that’s exceptional. In CA there’s a term for it, “single case analysis.” Usually, it’s done when some phenomenon shows up repeatedly in that single interaction, which is what happened in the case of our analysis of a single jury’s deliberations. You still have to make a judgment call that you have enough evidence for your claim. Schegloff needed only the single instance of the interaction between mom and son, and the single question and answer he focused on, to show that there are more ways to understand and respond to a question than a theorist at that time had posited. However, whether you have a single case analysis or multiple instances, your sample can still be biased, like in any other stream of research, and that’s fine. All research is done by a community over time. I can make mistakes, I can read too much into my data, my sample can be biased, and so forth. Someone else will likely find and expose such shortcomings.

43Pomerantz Generally, I write papers about practices used to do certain work in interaction. For instance, the practice of reporting an ‘unhappy event’ with no actor-agent incorporated to a recipient who is likely implicated is a way of giving recipient an opportunity to deal with his or her potential offense. Having identified a possible practice in one instance, I kept a file open to see if there were other instances of this possible practice. As I found more instances and became convinced that it is a practice used in different situations by different people, I was able to refine my analysis and produce a paper on the phenomenon. We assume such papers are contributions to social scientific knowledge.

44Could knowledge of the context also lead researchers to believe that their findings are specifically relevant to that context, and not more broadly applicable?

45Pomerantz When I wrote my dissertation, I discussed how people agree or disagree. A few years later, Marjorie Goodwin did a study of African-American kids in Philadelphia [Goodwin, 1983], and she found that they do not disagree the way I proposed at all. This led me to think about who my interactants were, and I realized that by and large they were middle-class white women. I had not written it that way. Afterwards, there were other studies that showed that in other contexts disagreement was not done the way I suggested. If I were to write it today, I probably would somehow identify the people I observed. I would not go for a broad claim, because in fact there is good reason to believe that what you observe may or may not be applicable to other settings.

46Sanders – Another way of seeing this issue is that it depends on whether you have interest in larger issues. If I observe meetings at a manufacturing organization, I may get more traction on some larger issue by tying what I observed to special circumstances found in manufacturing organizations, and not more broadly by making claims about meetings as such. But framing interaction in these meetings as being specifically about a manufacturing context poses the risk of beginning the investigation in a way that leads to claims that are too narrow, or maybe too broad.

47That being said, the mentality that we’re hoping LSI breeds, is to move away from an initial concern with larger social issues because they blind you to what is actually happening. You can talk about the larger social issues once you understand how things get done between people in the activities and contexts your data came from. But at the end of the day you use your best judgment about issues of scope, and then put your claims and data out there. As Anita described, it’s only with time and the input of other researchers that we can get a better handle on how to contextualize what we observe, or understand its scope. That’s exactly the way scientific progress gets made in any field.

48To play devil’s advocate, I have been asked, including by manuscript reviewers, about the end result, the outcome of the conversation—for instance, who won the argument. Should we worry about the outcomes of interaction?

49Pomerantz It depends on the goals of the project and the claims the researcher chooses to make. If I want to talk about outcomes, what happened at the end, then, of course, I have to look at that. If I want to write a paper about different kinds of information seeking practices, then I need to get a handle on that and definitely see what happens as a result of using one or another of them. It is often important to track what came before and what comes after it. I do not think the answer to the question can be a general one; we have to appreciate that different researchers have different senses of what would be interesting to find out.

50Sanders – Exactly. For instance, your question, Nicolas, presumes that there’s something to be gained by attending to what came about because of an interaction, afterwards. There is for someone who’s interested in how some specific thing can be brought about, like a win-win settlement through negotiation. However, that doesn’t change the value of examining how people get there, no matter what they end up with. Often, that’s more enlightening about communication than a narrower interest in how to reach a specific outcome. For example, the boyfriend-girlfriend interaction Anita and I analyzed was part of a study of how people go about dealing with the breach of a cultural norm once the breach is flagged. Our interest wasn’t in what the cultural norms were, or which breaches tend to be flagged, or how such a breach can be successfully remedied. Our interest was in how the offender and the offended jointly create this situation, and then work interactively to deal with it. Across our cases, the breach was quickly resolved in one, not resolved in another, and only resolved with considerable effort in the third. But again, our interest was mainly in how they dealt with this. We didn’t ignore outcomes because they were relevant. Under the same circumstance, though, a variety of outcomes was brought about, and we considered that that’s worth noticing. But that’s different from focusing on a particular outcome and how it can be brought about.

51In the boyfriend-girlfriend case, the girlfriend asks how his hunting trip went in a loaded way. She asked, “Did you kill anything?” He answers, “Yeah, we got a couple of turkeys, and a deer, and we shot a dog and some cats.” Before he even finishes that, she interjects, “You shot a dog?” There’s clearly something in her uptake that makes what he said wrong, and that burdens him to deal with that, with her. But he doesn’t; he initially avoids acknowledging that there’s an issue, and when that doesn’t put this thing to rest, he offers a series of justifications, but she doesn’t buy that.

52Pomerantz One interesting question with this piece of data was to figure out what about her utterance—what aspects or features of it—recognizably constituted the issue as not being resolved yet.

53Sanders – Right. That’s something that must have been recognizable to the boyfriend, not just us, because he keeps trying. They go back and forth between his justifying what he did, and her insisting that instead, he acknowledge it was wrong. When it evidently became clear to the girlfriend that she was wasting her time, she conceded, but did it in ways that marked it as insincere.

54Also, in line with the idea that different claims can be made from different analytical postures, and that outcomes is but one possible question, LSI would not necessarily want to provide suggestions or prescriptions, for instance for the couple to lead a happier marital life. That would not be the concern guiding the analysis.

55Sanders – Some research with concerns such as helping couples may get so abstract that details get lost, like ones we focused on in that study we did. Yes, someone might want to say that data like the boyfriend-girlfriend case can tell us something about, say, the health of the relationship between these people. Maybe you could generalize about patterns of interaction of couples in stable versus unstable relationships. In fact, the psychologist John Gottman [1979] did exactly that. He claimed for example that it’s predictive of divorce if newlywed couples handle conflict with certain types of response that are critical and disrespectful towards the partner. Yet, even though I admire how careful and scrupulous Gottman’s work is, he didn’t get very detailed and didn’t get at nuances of phrasing, intonation, and sequencing of actions, like we did. He went as far as he needed to for his correlations to be good enough for the purposes he had, and they’re very good: .85 or so. The kind of work we do may tease out exceptions, and complexities in the way conflict is done. I showed this by re-analyzing one of the conflicts Gottman used to make claims about the onset of emotional displays [Gottman, 1993; Sanders, 1995b]. It may detect something Gottman missed. It might account for why his correlations aren’t perfect. It might be helpful to therapists working with specific couples to steer them to what to listen for in their talk. However, if we’re going to capture such details, we can’t worry about what use it might be to know them.

56Pomerantz In several of my co-authored papers, we offered suggestions about good practices based on our analyses [Pomerantz, Mastriano, and Halfond, 1987; Pomerantz, Gill, and Denvir, 2007], but rather than outright suggestions, such as “Do X”, I prefer discussing the potential benefits and dangers of the practices described in the paper. That’s what we did when we studied the correcting practices of medical preceptors with interns and residents [Pomerantz, Ende and Erickson, 1995]. In other words, and to address your question more directly, I wouldn’t presume to draw from our data a lesson in how to lead a happier marital life. Yet, in line with what Bob just said, I would hope our research might eventually, perhaps via a therapist, provide one or both partners with a better understanding of their interactional dynamics and/or expand their repertoire of ways to handle difficult situations.

57What I am hearing you say is that even when we are looking at “effects” or “outcomes” we can use an interactional perspective, even though these words are typically associated with different research approaches.

58Pomerantz Of course, there are people who are committed to statistical methods and to particular understandings of the word “effects”. Depending on who our audience is and what you are going after—if you are writing a grant proposal, for instance—I would not advise using that term, effects, unless it’s associated with a statistical method. In the terms I’m more comfortable with, our interest is in the “relevance” of what came about, or the “consequence” of an action or actions. In the final section of the Pomerantz, Gill, and Denvir paper [2007] we addressed “the relevance and potential use” of our study. In Pomerantz, Ende, and Erickson [1995], we have a section on the consequences of the practices for patient care and training.

59Sanders – I think we need to be more careful about what we mean by “effect.” For instance, Dean Hewes [1996] suggested that there is no evidence that group deliberation has an effect, that it changes members’ opinions. However, that assertion corresponds to a narrow definition of an effect, particularly what the effect of group deliberation is. When we ask people on a jury to deliberate, we do not necessarily expect them to change their minds. The effect we are looking for is that the matter was resolved one way or another through the process of deliberation itself. Frankly, I was surprised at the relative inability of the jurors we studied to identify issues, and then assess evidence accordingly. What they did do, though, is hear each other formulate positions and decide which positions could be defended to others, or lived with personally. In fact, one of the jurors in our data did vote for guilty, even though she later said she was not convinced that the person was actually guilty. She didn’t change her mind, but she changed her vote, because she had no leg to stand on discursively to maintain that position. If we stipulate in advance what the effect of a group discussion should be, say a change in belief to bring about uniformity of thought, then we are simply derailing a close examination of what actually takes place when people deliberate. Not to mention stipulating in this case that people only do or say something if they believe it. Which is clearly not true. As I said, the effect that matters is that the jury deliberation brought about a conclusion—in this case they concluded they could not agree on a verdict. But only after they tried, after the issue was discussed and everyone involved had a chance to test out their thinking with the others. We need to keep an open mind on these things and examine the process.

60Pomerantz This has to do with the fact that the conventional model of human action consists in people having a belief and then acting on it. In that sense, many social scientists try to find ways of changing people’s beliefs and attitudes with the idea that it will change their behavior. In contrast, language and social interaction researchers seek to problematize how social action actually works. For example, Garfinkel [1967] did various experiments that show it can work the other way around. In one, he instructed people to engage in certain behaviors and found that it affected their beliefs and attitudes. He instructed confederates to engage in a casual conversation with someone they knew, and at some point bring their faces close up to the other’s face almost to where their noses were touching. This consistently led both confederate and subject to think about what was going on as having a sexual aspect though nothing else was said or done to suggest it. However, these kinds of studies are overlooked by researchers who are comfortable applying a model that predicts behavior from the values/attitudes/beliefs of the actors.

61Sanders – I think problematize is the key word. Many people studying communication simply take for granted that communication happened, that what was said is what was meant, and that’s how it was understood. They then focus on what came about because of it. To them, making communication happen is not a problem, or not an interesting one. To us, it is, or can be. What they are interested in is how the communication that happened brings about some result. In contrast, we want to say that communication happening is itself problematic, something people work at, and that’s the focus of our attention.

62You seem to be saying that LSI is occupied entirely with basic research, and has no interest in how your findings could be applied.

63Sanders – You’re mainly right, but not entirely, it’s more complicated. Most LSI researchers consider the work we do as revealing of the communicative complexity and intricacy of what goes on between people in the practical affairs of everyday life. That kind of work can be instructive to practitioners. It can raise their awareness of the importance of the details of their talk and it has attracted interest from some, for example for training doctors, therapists, airline cockpit crews, negotiators, educators, workplace teams. Some LSI researchers are overtly concerned with practical applications. For example, Karen Tracy is interested in creating a training program for emergency call-takers based on her analysis of actual calls, and what brought about misfires and failures. John Heritage has done work on how doctors can be sure that patients have not held any issues back that they wanted the doctor to address before ending an examination. These applied claims, however, still come out of a close analysis of the way people interact in such settings. The danger is that if you start off with an interest in making such claims, you may stop paying close enough attention to the details of how the outcome is produced.

64Pomerantz There are two senses of “applied research”. Your question taps into one of those senses. One sense, the one you’re invoking, is that research is applied when the findings are useful to some group. We’ve already talked about that issue. The other, more traditional, sense of applied research, in contrast to basic research, is when the research generates findings about a particular context, e.g. medical interaction, courtroom interaction, etc. Regarding my own inclination, my first love is to discover interactional practices that are used across contexts, although I have done quite a bit of work that would be considered applied. For example, how families talk about organ and tissue donation [Denvir and Pomerantz, 2009; Pomerantz, 2010].

65Sanders – To latch it back to bigger issues, when you look closely enough, how people disagree with each other is a problem in itself. It’s not only a matter of the jury figuring out what the evidence shows and doesn’t show, and debating that. It’s also a matter of how to tell someone they’re wrong without cutting off the possibility of future dialogue, or producing a tangential stream of talk that ends up at a dead end. The ability of juries to deliberate, therefore, isn’t simply about whether they can rationally assess the evidence. It’s also about people coping with each other and interacting with more or less success and harmony. This way of studying the jury goes back to what Garfinkel had to say about juries in the ’50s. He was critical of psychological research that was looking at juries with an interest in the progression of topics, questions, answers and information exchange, as if they were engaged in a rational process from beginning to end. Garfinkel [1967] felt this approach missed something more important and more basic in their interaction. He argued that the first question we should ask is, “How do these people constitute themselves as a jury? How do they become a jury together and take on this role and interact accordingly?” And this is something that happens interactively from the outset, and continues even as the jurors are also deliberating about the case.

66Pomerantz Probably the most quoted statement that he made in his work on juries is that 95% of what they do is what we do ordinarily, in everyday interaction [Garfinkel 1967, p. 110]. Juries, like people in other conversational situation, do not bring a whole new set of resources, they rely on the same interactional rules and practices as anyone else.

67Can LSI tackle “broader” or “social” problems in the more conventional sense of the word?

68Sanders – Let’s think about this in the opposite way. Why should LSI be expected to address larger social issues or problems? Generally, it’s not what we’re about, although there’s some restlessness among some of us about whether it should be. I think the question is, can people whose interest is larger social issues and occurrences find any value in coming at them with an LSI lens. I’d say that they can and should. Let’s go back to the analogy I drew between biology and chemistry. This question you just asked basically asks whether chemists can tell biologists anything about biological phenomena. I think the answer would be no, that’s not what chemistry is about. The question is, do biologists gain by taking chemical processes into account? We know the answer is certainly yes. So back to communication. There’s a lot of attention being given in news reports and politics to whether Russia is using social media to influence elections. Not just in the U.S. but other Western countries. Yet, I haven’t seen any serious ideas about how that could work. LSI could offer a pathway to answering that. For example, I was struck by the rapid deterioration of Soviet hegemony in Eastern Europe in the late ’80s, and I hypothesized an answer [Sanders, 1992]. I drew on LSI concepts about how actions are interpreted based on where they occur in a sequence, and applied that to the way the progression of news reports about events in the USSR after Gorbachev took office may have been interpreted. Also, we’re currently seeing a lot of social upheaval around ethnic and racial divisions. This has bubbled to the surface in politics, in news media, and sometimes in the streets. The question is, how do such divisions get so deeply ingrained and felt? Researchers with that interest could turn to, perhaps join, LSI. If you could show what racism, or some other problem, looks like as people interact, maybe we could work to eliminate it. They may find work on racist talk useful, such as van Dijk’s [van Dijk, 1987], who tackled the question of indirect, deniable, ways racist attitudes can be expressed. Researchers may also find it in ethnographic work that reveals ways cultural differences surface in very small details of interaction [Gumperz, 1982; Scollon and Scollon, 1981]. In short, I don’t think it’s up to LSI to tackle big social problems. We’re chemists, or maybe cell biologists, not oncologists. But people who do want to say something about larger issues may find some value in the work we do, as it is.

69Pomerantz Amen.

Haut de page


Chomsky, N. (1957). Syntactic structures. The Hague: Mouton.

Denvir, P. and Pomerantz, A. (2009). A Qualitative Analysis of a Significant Barrier to Organ and Tissue Donation: Receiving Less-Than-Optimal Medical Care. Health Communication, 24, 597–607.

Garfinkel, H. (1967). Studies in ethnomethodology. Englewood Cliffs, NJ: Prentice-Hall.

Goffman, E. (1959). The presentation of self in everyday life. Garden City, NY: Doubleday.

Goodwin, M.H. (1983). Aggravated correction and disagreement in children’s conversations. Journal of Pragmatics, 7, 657–677.

Gottman, J. M. (1979). Marital interaction: Experimental investigations. New York: Academic Press.

Gottman, J. M. (1993). A theory of marital dissolution and stability. Journal of Family Psychology, 7, 57–75.

Gumperz, J. J. (1982). Discourse strategies. Cambridge: Cambridge University Press.

Hewes, D. E. (1996). Small group communication may not influence decision-making: An amplification of socio-egocentric theory. In R. Y. Hirokawa and M. S. Poole (Eds.), Communication and group decision making, 2nd ed. (pp. 179–211). London: Sage.

Hymes, D. (1974). Foundations of sociolinguistics: An ethnographic approach. Philadelphia: University of Pennsylvania Press.

Maynard, D. W. (1997). The news delivery sequence: Bad news and good news in conversational interaction. Research on Language and Social Interaction, 30, 93–130.

Moerman, M. (1988). Talking culture: Ethnography and conversation analysis. Cambridge: Cambridge University Press.

Pearce, W. B., and Cronen, V. E. (1980). Communication, action, and meaning: The creation of social realities. New York: Praeger.

Pomerantz, A. (1988). Offering a candidate answer: An information seeking strategy. Communication Monographs, 55, 360–373.

Pomerantz, A. (2010). The value of qualitative studies of interpersonal conversations about health topics: A study of family discussions of organ donation as illustration. In E. M. Alvaro and J. T. Siegel (Eds.) Understanding Organ Donation: Applied Behavioral Science Perspectives. (pp. 272–291). Malden, MA: Wiley Blackwell Press.

Pomerantz, A., Ende, J., and Erickson, F. (1995) Precepting in a General Medicine Clinic: How Preceptors Correct. In Morris, G.H. and Chenail, R.J. (Eds.) The Talk of the Clinic (pp. 151–169.). Hillsdale, New Jersey: Lawrence Erlbaum Associates.

Pomerantz, A., Gill, V. T. and Denvir, P. (2007). When patients present serious health conditions as unlikely: Managing potentially conflicting issues and constraints. In A. Hepburn and S. Wiggins (Eds.), Discursive Research in Practice: New approaches to psychology and interaction. Cambridge University Press, 127–146.

Pomerantz, A., Mastriano, B. and Halfond, M. M. (1987). Student clinicians’ difficulties while conducting the summary diagnostic interview. Text,7(1), 19–36.

Pomerantz, A. and Rintel, E.S. (2004). Practices for reporting and responding to test results during medical consultations: enacting the roles of paternalism and independent expertise. Discourse Studies 6(1) 9–26.

Pomerantz, A. and Sanders, R. E. (2013). Conflict in the jury room. Journal of Language Aggression and Conflict, 1(2), 141–164.

Pomerantz, A. and Sanders, R. E. (2014). The use of categories and their vulnerability in jurors’ claims about the defendant’s bameworthiness. In C. M. Jacknick, C. Box and H. Waring (Eds.), Talk in institutions: A LANSI volume (pp. 8–28). Newcastle upon Tyne, UK: Cambridge Scholars Publishing.

Sacks, H. (1992). Lectures on conversation (G. Jefferson, Ed.). (Vol. 1 and 2). Oxford: Basil Blackwell.

Sanders, R. E. (1992). The role of mass communication processes in producing upheavals in the Soviet Union, Eastern Europe, and China. In S. S. King and D. P. Cushman (Eds.), Political communication: Engineering visions of order in the socialist world (pp. 143–162). Albany: SUNY Press.

Sanders, R. E. (1995a). A neo-rhetorical perspective: The enactment of role-identities as interactive and strategic. In S. J. Sigman (Ed.), The consequentiality of communication (pp. 67–120). Hillsdale, NJ: Erlbaum.

Sanders, R. E. (1995b). The sequential inferential theories of Sanders and Gottman. In D. P. Cushman and B. Kovacic (Eds.), Watershed research traditions in human communication theory (pp. 101–136). Albany, NY: SUNY Press.

Sanders, R. E. (2007). The composition and sequencing of communicative acts to solve social problems: Functionality and inventiveness in children’s interactions. Communication Monographs, 74(4), 464–491.

Sanders, R. E. and Pomerantz, A. (fortcoming). Bounded Segments of interaction: The case of redressing the breach of a cultural norm once it is flagged. In E. Weigand and I. Kecskes (Eds.), From Pragmatics to Dialogue. Amsterdam: John Benjamins.

Scollon, R., and Scollon, S. B. K. (1981). Narrative, literacy, and face in interethnic communication. Norwood, NJ: Ablex.

van Dijk, T. A. (1987). Communicating racism: Ethnic prejudice in thought and talk. Newbury Park, CA: Sage.

Haut de page

Pour citer cet article

Référence papier

Anita Pomerantz, Robert E. Sanders et Nicolas Bencherki, « Communication as the Study of Social Action: on the Study of Language and Social Interaction », Communiquer, 22 | 2018, 103-118.

Référence électronique

Anita Pomerantz, Robert E. Sanders et Nicolas Bencherki, « Communication as the Study of Social Action: on the Study of Language and Social Interaction », Communiquer [En ligne], 22 | 2018, mis en ligne le 01 mars 2018, consulté le 25 octobre 2020. URL : ; DOI :

Haut de page


Anita Pomerantz

Professor emeritus in Communication
State University of New York at Albany

Robert E. Sanders

Professor emeritus in Communication
State University of New York at Albany

Nicolas Bencherki

Assistant professor of Organizational Communication
TÉLUQ, Canada

Nicolas Bencherki holds a PhD in communication from the Université de Montréal and in sociology of action from Sciences Po Paris. Previous to his actual position, he was an assistant professor at SUNY Albany following postdocs at the Centre for the sociology of innovation at Mines ParisTech, and NYU Poly’s Department of Technology Management and Innovation. His research focuses on the intersecting roles of organizational communication and materiality in the interactional constitution of membership, strategy and other conventional organizational issues in the setting of the non-profit and community-based organizations, with a special interest for the concept of property. His work has been published, among others, in Human Relations, Management Communication Quarterly, and the Journal of Communication.

Articles du même auteur

Haut de page

Droits d’auteur

© Communiquer

Haut de page
  • Logo UQAM - Université du Québec à Montréal
  • Logo DOAJ - Directory of Open Access Journals
  • OpenEdition Journals
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search