Methodological breaching experiments: Steps toward theorizing the qualitative interview

Introduction

The qualitative research interview is arguably the most widespread method of inquiry across the human and social sciences today. In numerous fields and disciplines, ranging from sociology and anthropology to communication and education, qualitative interviews are now widely used. This is also increasingly the case in psychology, although this discipline has traditionally been reluctant to accept the so-called qualitative methods among its repertoire of investigative practices (Brinkmann, Jacobsen, & Kristiansen, 2014). This in itself is curious, for most of the founders of the discipline employed what we today refer to as qualitative methods: Wilhelm Wundt developed his anthropological Völkerpsychologie; Sigmund Freud revolutionized the science and treatment of the mind with his conversational approach, which was later employed by figures such as Jean Piaget, who interviewed his own children (Kvale, 2003); William James undertook an impressive phenomenology of religious experience (James, 1902); and Frederic Bartlett devised a proper qualitative tradition in experimental psychology (see Wagoner, 2015) to name just a few. Sadly, this rich qualitative tradition was ignored or even repressed for many years in the 20th century by the mainstream of American textbook psychology, which aimed to construct itself on the background of its image of the natural sciences, however faulty this image of “the real sciences” really was (Latour (2000) has referred to it as a comedy of errors, since the natural sciences are much less mechanical, and, yes, more qualitative, than assumed by psychologists and social scientists who can be said to suffer from a sad case of physics envy).

Since around the year 2000, however, it seems fair to say that qualitative methods are increasingly accepted and put to use to study psychological phenomena. Qualitative interviewing, in particular, is gaining in use to the extent that some commentators now argue that this method is overused in qualitative psychology at the expense of other qualitative research practices (Potter & Hepburn, 2005). I agree that there is an unfortunate tendency to choose qualitative interviews, simply because it is conceived as an easy and perhaps even fun way of “collecting data” for both students and senior researchers, rather than because it represents the most adequate means of answering a specific research question. Of course, it is highly relevant to engage in qualitative interviewing when the topic under scrutiny is something that lends itself to being addressed in the context of a face-to-face encounter with a focus on verbalizations. But not all topics in psychology are like this.

In order to be better equipped to choose an appropriate means of knowledge production, it seems necessary to theorize the interview more extensively than is normally the case in the numerous “how-to” books that exist on the academic marketplace. Too many researchers simply go out and interview without careful theoretical reflection on what an interview actually is and what its potentials and limitations are. Too many qualitative psychologists, in particular, still seem to believe that interviewing is a natural relation between people that provides more or less unhindered and unmediated access to the experiences of the interviewee. This view goes back at least to Carl Rogers in the 1940s, who pioneered the use of the “nondirective interview,” which also became a model for his client-centered therapy: “Through the non-directive interview we have an unbiased method by which we may plumb these private thoughts and perceptions of the individual,” Rogers (1945, p. 282) claimed.

That the interview is more “unbiased” and even “natural” (given its conversational format) than an experiment or a survey is a myth, for it is a staged form of interaction with countless cultural and historical specifics. In this article, I will try to put together theoretical discussions of the qualitative interview that have been published in different sources in order to begin to articulate a coherent theoretical account (Brinkmann, 2013, 2014; Brinkmann & Kvale, 2015). The articles by Inger Skjelsbæk (2016) on the use of interpreters in interviews, and especially by Gustafsson Jertfelt, Blanchin, and Li (2016) on obstacles in intercultural interviewing, provide wonderful occasions to step back and realize how contingent and culturally informed our current practices of qualitative interviewing in fact are. In what follows, I shall first consider these discussions of intercultural interviewing as methodological breaching experiments that enable us to understand the intricacies of a practice that is otherwise normally taken for granted. Next, this leads me to propose a denaturalization of the qualitative interview that must be considered not simply as a neutral instrument, allegedly representing a “natural” human relationship, but rather as a social practice with a history that provides a specific context for human interaction and knowledge production. I end by spelling out some of the elements of this context that qualitative interviewers should take into account when reflecting on this practice of knowledge production.

Methodological breaching experiments

Harold Garfinkel was the creator of the approach to social science that we know as ethnomethodology (Garfinkel, 1967). In brief, the goal of ethnomethodology is to study (thus the “-logy”) how people (hence “ethno-”) use methods (“-methodo-”) to create social order. Unlike common sociological theories that posit abstract entities such as “society,” “context,” or “social structure” that are supposed to causally generate social processes and phenomena, Garfinkel maintained that people are knowledgeable agents who act locally in order to create what we refer to with these abstract terms (society, context, social structure, etc.). How is this creation of social order accomplished in the hurly-burly of our everyday lives? This is a key question for ethnomethodologists.

The funny term ‘hurly-burly’ was introduced into philosophy by Ludwig Wittgenstein when he asked:

How could human behavior be described? Surely only by sketching the actions of a variety of humans, as they are all mixed up together. What determines our judgment, our concepts and reactions, is not what one man is doing now, an individual action, but the whole hurly-burly of human actions, the background against which we see any action. (Wittgenstein, 1981, no. 567)

“S” is the subject and “E” is the “experimenter,” deliberately breaching the ordinary production of social order in this sequence. Like deconstructive analyses that unsettle the taken for granted and reveal implicit presuppositions in a text, such breaching experiments highlight the background expectancies that are rarely thematized in everyday life, and the typical result of Garfinkel’s breaches was bewilderment, unease, anxiety, and even anger on behalf of the conversationalists (see Brinkmann & Kvale, 2015).

The discussions of qualitative interviewing by Skjelsbæk (2016) and especially Gustafsson et al. (2016) can be seen as methodological breaching experiments, which bring the hidden expectancies into the open and thus enable us to begin theorizing the interview. One way that expectancies are breached is by using an interpreter, as recounted by Skjelsbæk. She demonstrates that the interpreter is not simply a neutral conveyer of words, but rather someone who influences what is said and recorded, and how perceptions and reflections are made (p. 5). Another way is seen when conducting qualitative interviews in China, as described by Gustafsson et al. (2016). The latter researchers point out in their discussion that “when using an open-ended interview in one context, the meaning of the situation might differ in another and that, in turn, suggests that the meaning, interpretation, quality and structure of the narratives vary” (p. 2). The authors are aware that much of the methodology behind the qualitative interview is developed in the so-called Western part of the world (p. 3), and their Chinese interviewees typically react differently. For example, according to Gustafsson et al. (2016), Chinese interviewees tend to search for the correct answer to the interviewer’s question and also express a wish to move on to the next question once the first one has been answered. This breaks with the qualitative researchers’ preunderstanding that there is no one correct answer in a qualitative interview (but rather multiple stories to be told, and many descriptions to be given, with the interviewee in the “expert role”), and also with the ideal of dwelling on a concrete description in order to understand the many dimensions of the phenomena studied. Also, the Chinese interviewees are reported to talk about general examples rather than describing concrete experiences from their own lives, which is the ideal in phenomenological interviewing. This, I have found, will also often happen in a “Western” context, although it might be more likely elsewhere, and again, it reveals the implicit norm in qualitative interviewing that authentic “lived experience” ought to take priority over scripted knowledge and abstract theory (see Brinkmann (2007) for arguments that challenge this norm).

All in all, we should acknowledge how remarkable it is that interviews normally take place in ways that are very well ordered, and where everyone—interviewers and interviewees alike—know how to play their roles satisfactorily. As interviewers, we too rarely step back and consider how astonishing this is, and how much implicit know-how the conversationalists bring to the encounter in order to establish the smooth social order of a standard interview. People are probably trained to perform their roles because that have been socialized into what some commentators have referred to as “the interview society” (Atkinson & Silverman, 1997), where everyone is acquainted with interviews from mass media. Latour (2000) has argued that exactly this phenomenon—that people play their roles so well in the human and social sciences—is what makes objectivity so difficult to attain in these sciences. He works from the premise that objectivity means “allowing the object to object,” which is an idea that goes back to John Dewey’s pragmatism, where an object was simply defined as “that which objects, that to which frustration is due” (Dewey, 1925, p. 239). From this perspective, objectivity is attained in scientific practices when objects reveal themselves through acts that frustrate the researcher’s preconceived ideas. In the social sciences, however, “nothing is more difficult than to find a way to render objects able to object to the utterances that we make about them,” Latour (2000, p. 115) claims. Unlike the objects of the natural sciences (molecules, planets, nonhuman animals, etc.), human beings are in Latour’s eyes incredibly complacent, “behaving all too easily as if they had been mastered by the scientist’s aims and goals,” as he says (Latour, 1997, p. xv). With this in mind, the examples given by Gustafsson et al. (2016) are interesting, because we see that the interviewees are in fact behaving in ways that do go against the researchers’ immediate aims and goals, which, furthermore, is exactly what enable us to understand better how interviews work in the knowledge-producing process, thereby giving us a chance to become more objective (in Latour’s sense). Naturally occurring methodological breaching experiments allow us to become more objective concerning the forms and functions of interviews as investigative practices.

Qualitative interviewing as a social practice: Denaturalizing the interview

In their discussion of “the interview society,” referred to above, Atkinson and Silverman argue that “in promoting a particular view of narratives of personal experience, researchers too often recapitulate, in an uncritical fashion, features of the contemporary interview society” where “the interview becomes a personal confessional” (Atkinson & Silverman, 1997, p. 305). Although the conversation in a broad sense is a human universal, qualitative interviewers often forget that the social practice of research interviewing in a narrower sense is a historically and culturally specific mode of interacting, and they too often construe “face-to-face interaction” as “the primordial, natural setting for communication,” as anthropologist Charles Briggs (2007) has pointed out (p. 554). Briggs argues that we need to take “the larger set of practices of knowledge production that makes up the research from beginning to end” (p. 566) into account in qualitative analyses and not simply focus on interviewee talk as if it fell from the sky. Just as it is essential in quantitative and experimental research to have an adequate understanding of one’s technologies of experimentation, it is similarly essential in qualitative interviewing to understand the intricacies of this quite specific knowledge producing practice, and interviewers should be particularly careful not to naturalize the form of human relationship that is a qualitative research interview and simply gloss it over as an unproblematic, direct, and universal source of knowledge.

Briggs (2007) has argued that qualitative interviews imply a certain “field of communicability,” referring to a socially situated construction of communicative processes (p. 556). This construction is an artifact of cultural–historical practices and is placed within organized social fields that produce different roles, positions, relations, and forms of agency that are frequently simply taken for granted. It is noteworthy that the otherwise standardized format of “face-to-face interaction” was only named early in the 20th century by the sociologist Charles Horton Cooley, but was since constructed as “primordial, authentic, quintessentially human, and necessary” (p. 553). It is sometimes forgotten that also the face-to-face interview is a kind of interaction that is mediated by this particular social arrangement that has a history.

If we look at how qualitative interviews have been conceptualized in the theoretical literature, there are two main approaches (what follows builds on Brinkmann (2013)). One approach conceives of interviews as research instruments that facilitate the transfer of lived experience from the interviewee to the interviewer. This, I find, is the most common way of approaching interviews, which also lies in the background of the analyses of Gustafsson et al. (2016) and Skjelsbæk (2016), just as it was articulated quite clearly by Rogers as we saw above. This first approach has a broadly phenomenological ambition, seeking to obtain descriptions of the lifeworld of the interviewee and treating the person’s talk about experience as veridical reports. There is thus a focus on the “what” of communication. Another approach puts more emphasis on “how” things are talked about in the situated interaction of the interview. In line with ethnomethodology, this approach can be said to concentrate on interviewing as a social practice, and it treats interview talk as primarily discursive accounts (rather than reports) that are occasioned by the situation in which they occur, serving to uphold a local social order. The two approaches are contrasted schematically below:

Conception of interviewing: Research instrument versus Social practice

Conception of interview data: Reports versus accounts

Standard analytic focus: Lived experience – the “what” versus situated interaction – the “how”

Typical interviewer style: Receptive versus assertive

Main challenge: Validity of interviewee reports versus relevance of interviewee accounts

In contrast to those approaches that see interviewing as a research instrument designed to capture the “what” of what is reported as accurately as possible, others working from more localist and situated perspectives (e.g. in ethnomethodology) have much greater analytic focus on the “how” of interviewing. They view interviewing as a social practice; as a site for a specific kind of situated interaction, which means that interview data primarily reflect “a reality constructed by the interviewee and interviewer” (Rapley, 2001, p. 304). The idea of obtaining valid reports that accurately reflect a reality outside the conversational situation is thus questioned, and the main challenge becomes instead how to explain the relevance of doing interviews: If what is said in an interview is a product of this social practice itself, why is it then relevant to conduct interviews? Some researchers answer this by emphasizing the performative and potentially transformative aspects of interviewing (Denzin, 2001): If interviews do not concern a reality outside themselves, they can instead be used to perform or facilitate social change. It is also possible to concentrate one’s analyses on the interview itself—how is this kind of interaction accomplished through accounts? Unlike reports, which are thought to refer to experiences in the interviewee’s past, accounts are answers that are “normatively oriented to and designed for the questions that occasion them” (Talmy, 2010, p. 136). If interviewee talk is understood as accounts, it must be seen as a kind of social action that has effects and does something in the situation of which it is a part. This perspective on interviewing is shared by some discourse analysts and conversation analysts, who limit themselves to analyzing interview talk as situated interaction (see Roulston, 2010).

I believe that both of these approaches in their pure forms are misleading. It is true that there are huge problems associated with viewing the interview as a site for uncontaminated reports of the past (we know too much about the constructive role of human memory, and of how the social practice of interviewing mediates what is said, to take this seriously). But it is certainly also true that there are obvious problems associated with denying that we can refer more or less accurately to past experiences in interviews. So, taken to extremes, both approaches become absurd, and I believe that it is now time for the two (sometimes opposed) camps to learn from one another and realize that they need not exclude one another. Musaeus and Brinkmann (2011) give an example of how this be done in practice, and it is further discussed elsewhere (Brinkmann, 2013).

The qualitative interview as a context

In this final section of the paper, I shall try to follow Briggs and describe the “field of communicability” of a qualitative interview. Let us bear in mind that the occasion for doing this is the breaching experienced when interviews are conducted in ways that become problematic, either because of the mediational role of interpreters or because the interviewees have a background in a “noninterview society.” Let us also keep in mind that the field of communicability, at least according to the ethnomethodological reflections with which I began the paper, is not simply there in advance—as a room into which the conversationalists may enter—but rather is a context that is created as the outcome of the efforts of the interaction partners (and which conversely can be referred to by them in a process of mutuality). Before proceeding, however, it might be fruitful to dwell a bit on the extremely widespread and elastic term “context.” Although this word is frequently used by cultural and qualitative researchers, it is quite rarely defined (what follows builds on Brinkmann and Kvale (2015)).

Metaphorically, context is often, implicitly or explicitly, taken to be a kind of container. We are in a context (e.g. an interview), like the keys are in the pocket or the apples are in the bowl. However, when taken in a purely spatial sense like this, the alleged powers of context quite appear quite mysterious, for pockets and bowls have little influence on the state and nature of keys and apples. This also leads to a problem of boundlessness concerning context, for bowls are typically situated on a table, in a room, in a house, in a town, etc. How can we say where the context stops? If we take the simple example of someone going to get a haircut, and if we want to understand the interaction in this situation or the meaning of what happens, then where can we draw the line and determine the context? What if the customer is a man and the hairdresser a woman—is this important? What if he is black and she is white? What if he is old and she is young? And what about features of the physical setting itself? Of the culture of which they are a part? Of the financial situation of society at this point in time? We may go on indefinitely and ask about contextual features that affect what these people do and say to each other, but this is not often done, and instead researchers simply conclude that something took place in a given way “because of the context.” The problem is that this bestows causal powers into something—context—which is very fuzzy and opaque. Some social scientists such as Latour therefore conclude that “context stinks!” (Latour, 2005, p. 148). The concept seems to be explanatory, but, according to Latour, it is actually not, and researchers typically invoke it when they are too lazy to continue describing the particulars that make something happen in a specific way.

Dilley (1999) has also pinpointed the central problem of contextualism: As social and cultural researchers we look for meaning, which is context dependent, but context is boundless, which has the consequence that meaning becomes indeterminate. Perhaps the solution is not to leave the idea of context behind because it stinks, but exactly to return to it over and over again, and, as a qualitative researcher, patiently continuing one’s description of whatever one is studying. In relation to interviewing, this should include careful and patient description, analysis, and theorization of this specific conversational practice. As Dilley continues:

we must never lose sight of the fact that a claim about context is precisely that – an articulation concerning a set of connections and disconnections thought to be relevant to a specific agent that is socially and historically situated, and to a particular purpose. (p. 39)

If this is so, we must ask how the context of interviewing is worked out by participants. Following Briggs and others, I believe we must look at (at least) three things in order to answer this question: The position of the interviewer, of the interviewee, and also the material arrangement of the situation. Concerning the first point, it is significant that there is no one and universal way in which to be an interviewer. In Brinkmann and Kvale (2015), we outlined three ideal types of interviewers called the pollster, the prober, and the participant. The pollster is the most widespread on in the methodological literature and is presented as primarily seeking the opinions and attitudes of the interviewees. These are treated as reports to be analyzed, without challenging the interviewee. The prober is also quite common, and is not content to merely record opinions and attitudes, but is trying to get beyond the surface and inquire into deeper layers of the subjects’ experiential world. The prober may present herself as akin to a friend, perhaps working with what has been called a “method of friendship” (Fontana & Frey, 2005), being able to ask questions that strangers would not normally do. Finally, the interviewer conceived as participant does not treat the descriptions and stories of the interviewees as reports to be analyzed, but rather—in line with the view of interviewing as a social practice—as accounts that are coproduced in the situated interaction of the interview, and which might be challenged in the course of the conversation. The interviewer as participant does not think of herself as a passive spectator to the pure experience of the interviewee but is actively participating in creating a conversation. These ways of acting the role of interviewer give rise to different constructions of the context of interviewing.

Moving on to the positions of the interviewees, Foley (2012) has introduced a helpful distinction between different ways that these are constructed relative to different methods of interviewing. “To conduct particular kinds of research,” she says, “researchers must imagine a certain type of respondent” (p. 305). In survey interviews (and also in some qualitative interviews that resemble the survey form), subjects are constructed as reporters, based on the idea that “there is a truth to be told.” Approaching subjects as reporters means that they can be viewed as “honest and helpful; lying, uncooperative, or forgetful; or unable to comprehend the questions” (p. 306). This construction of the interviewee focuses on the veridicality of the reports elicited by the subject. The interviewee as reporter is implicit in much phenomenological interviewing. The next model of the interviewee highlighted by Foley is the teacher. Approached as teachers, subjects are encouraged to “raise and explore issues that they find to be relevant and allow their voices to be heard” (p. 306). This gives the interviewee much more control over the interview, and the interaction may come to resemble an everyday conversation where both parties can legitimately raise questions and concerns. Foley connects this interviewee subject position with the context of in-depth interviewing. Finally, the third model discussed by Foley is the respondent as member or informant (close to the interviewer’s role as participant), which is a function of the context of ethnographic interviewing. When employing this method, interview subjects are often chosen because they have special knowledge about a setting or a specific social practice, and they are thus positioned as experts, and, according to this model, the interview may turn out to be more factual than experiential, more descriptive than emotional.

In addition to understanding the roles of interviewers and interviewees, we must also consider the role of the material arrangement in bringing about the context of the interview. An interview is surely verbal communication, but it is also—especially in its face-to-face formats—embodied communication. Bodies are never neutral, but carry all their signs of gender, race, class, etc. Bodies are also present in a simpler way, related to how people sit and comport themselves, how they smell and move, and how they are dressed, for example. Besides the body, other material aspects may also be important. Very little has been written on the contextual significance of the role of the sound recorder, the arrangement of the furniture, or the physical location of the setting for the interview. Interview researchers have commonly concentrated on talk and have not paid attention to the material context of the talk. However, social scientists are now generally “in the process of rediscovering the crucial and muiltifarious role of the material” (Michael, 2004, p. 6), and we must also consider this in relation to interviewing. Mike Michael is one of the few qualitative researchers to have done this by demonstrating the thoroughgoing significance of the nonhuman in interview research. His article “On making data social” is both a case study of a specific interview episode, which turned out to be heavily affected by nonhuman material factors, and also a reflection on the role that materiality generally plays in constituting sociological data.

Building on the philosophy of Latour and other material-semiotic theories, Michael goes through an interview episode he had with an interviewee at her home. The topic was the public understanding of science, and Michael carefully lays out the array of entities around him in the interview situation (his article even contains several maps depicting the arrangement):

I was seated on the sofa, the respondent was in an armchair to my right, and the tape recorder was placed on the floor between us. During the preliminary conversation, her pit bull terrier ambled up and sat on my feet … While this conversation was going on, her cat came into the room, and after a few moments of clawing at the tape recorder began to pull it along the ground by its strap. (2004, pp. 12–13)

I will not unfold his case further, but Michael’s general lesson for qualitative interviewers is to point out how a number of relations with materials and nonhumans (here furniture, dog, cat, tape recorder) “must normally be disciplined so that sociological data might be ‘gatherable’” (2004, p. 14). The context of the interview must first be created—enacted through the kind of work that ethnomethodologists are interested in—and this means that a qualitative interview is not simply a function of two or more persons who come together to talk, for their coming together is always mediated by a host of nonhuman factors that could (but not always should) be taken into account, since they may matter analytically. Many things must play together, so to speak, in order for qualitative interviewing to be possible, and the interview as a context does not just fall from the sky, but is the result of numerous actors that orchestrate this complex episode, which we have today come to take for granted.

Conclusions

The main point of this article has been to argue that we can learn a lot if we stop taking the context of a qualitative interview for granted. An interview is not a universally determined form that unfolds naturally if people simply come together to talk about lived experience. Rather, this specific way of interacting is the result of numerous actions on behalf of the participants, and where material factors also play a significant role. In this article, I have drawn a number of distinctions: First between different conceptualizations of the interview, and second between different ways of enacting the roles of interviewer and interviewee. The point has been to argue for heterogeneity and sensitivity to how the particulars of interview episodes codetermine what and how things are talked about. These intricacies become visible by taking a step back from what we take for granted, which can be facilitated by the fortunate methodological breaching experiments that may happen by chance or which we may intentionally try to bring forth. How to do this in methodologically helpful ways, however, must be the topic for another article (but see the Socratic suggestions in Brinkmann (2007)).