11.4.1 An Explanation of the Method

To begin with, a reminder of Stainton-Rogers’ useful definition of the term “discourse” from the perspective of constructionist social psychology: “… a discourse is defined as the product of constructing and the means to construct meaning in a particular way” (2003: 81). Stainton-Rogers is not alone when she notes that the field of DA is characterized by numerous methodological types and definitions of discourse. Her version is however ideally suited to the present purposes and is consistent with DP as a methodology for research.

For its core ideas, DP draws on DA, conversation analysis, rhetoric, and ethnomethodology. It takes its theoretical and analytical origins in, for instance, the pioneering works of Michael Gilbert and Nigel Mulkay who were the first sociologists to apply a discourse analytic methodology in their field. They investigated how scientists “do knowledge” in discourse: in an intriguing study, which uses interviews and written texts, they report an asymmetry between scientists’ treatment of “correct belief” as derived unambiguously from experimental evidence (which therefore has no need of explanation), and errors that must be explained away as these are invariably seen as the result of nonscientific influence. Gilbert and his coresearcher’s interest lay particularly in the latter phenomena, finding that scientists used a far more elaborate repertoire in their accounting for error (recall the discussion of interpretive repertoires in Section 6.2.3).

DP is concerned with the action orientation of language (talk and text), specifically the rhetorical construction and organization of versions of affairs, their social organization—how it works—and what it is designed to do. In Derek Edwards’ and his coworker’s own words, DP is “… concerned with the nature of knowledge, cognition and reality: with how events are described and explained, how factual reports are constructed, how cognitive states are attributed” (1992: 2). It is, in other words, a functional approach to the analysis of discourse with a particular interest in epistemology, and its core assumption is that language is constructive/constructed, functional, consequential, and variable.

A simple example will, it is hoped, illustrate these points. The conceptualization of DP, in the most basic sense, draws a distinct difference between the contents of what a speaker utters and what the utterance actually accomplishes as linguistic action:

it is statistically safer to travel by airplane

In this short statement, the speaker, on the “contents level,” is simply reporting a fact. From the analytic perspective, the speaker is using scientific accounting (“statistically”) to persuade the listener of the factuality of their version of affairs. The upshot is to imply that all other forms of travel are risky and that to ignore this version of affairs is to take action (decisions, purchases, intentions, and so forth) which is risk laden—even foolish—with the consequences “on your own head.” Alternatively, as what Jonathan Clifton describes as “competent members of the same community of speakers,” we know that this is a commonly issued statement (Superman said something similar to Lois Lane after a near helicopter disaster in the 1978 film), so it must be true. But—and it is a big “but”—we cannot be sure that either perspective is anything other than the researcher’s own interpretation in the absence of evidence of how the listener formulates their understanding of its contents. This raises two points that are returned to subsequently: the idea of the “next turn proof” and the role of the researcher.

In Wooffitt’s analysis, DP is “… focused on the ways in which cognitive notions can be treated analytically as situated practices which address interactional and inferential concerns in everyday circumstances” (2005: 116). By locating psychology in language, it makes possible the direct study of the processes of thinking. Contrast this with the traditional experimental method that, from this perspective, is reduced to the study of secondary or indirect phenomena in the form of, for instance, reported recollections of past events. This difference is significant: DP studies psychological phenomena as constructed in everyday talk and text and as “noticed” by both speaker and recipient (the idea of “next turn proof” mentioned earlier), while conventional methods treat discourse as the pathway to what a person is “really thinking.” This inevitably treats phenomena as secondhand.

A further significant difference lies in DP’s focus and interest in what is not said. Robin Wooffitt provides the perfect example to illustrate this: he describes a comparison between transcripts of interviews with a mentally ill patient and people talking about their Psi experiences. Both report strange phenomena. But the Psi reporters linguistically work to “display” themselves as normal (“I was just coming into the kitchen when…”: “I was getting out of the car and then I saw …”) as a preface to their account of a weird happening, whereas the mentally ill patient uses no such rhetorical devices: “the god appeared holding a sword and shield” (paraphrased). The omission is suggestive of abnormal behavior.

The kinds of questions that DP focuses on are then: how is the account constructed to appear, for instance, factual and objective, what resources are used, with what function, and how these connect to topics in social psychology. Again, this contrasts with traditional methods that focus on “why” questions.

The DP project draws its data from everyday talk and text that can take the form of audio and/or video recordings, interviews, and any kind of written text. Following Edwards and his colleague, the focus of analysis is on the social and rhetorical organization present in the data, as opposed to linguistic organization, for instance. It is thus an observational science that seeks to describe and document phenomena in order to support broader theoretical claims. But this does not necessarily make it an inductionist approach, as understood in Karl Popper’s interpretation of the term. In summary, DP is a theoretically informed analytical approach that seeks to investigate and understand social psychological phenomena as they are enacted as located (“situated”) in the speakers’ understanding of how talk normatively and progressively unfolds (the “procedural discursive interaction”).

An approach that locates knowledge work in discourse and takes that discourse as the topic of study has the potential to lead to a greater understanding of how knowledge work “works.” As argued at the end of Chapter 9, such an approach simply extends an existing trend to conceptualize knowledge as social action subscribed to by many scholars in the knowledge management (KM) field.

11.4.2 Grounds for Criticism and the Issue of Measuring Quality

It is usual in scientific research reports to discuss matters of bias (both researcher and participant), validity, and reliability (e.g., the extent to which findings can be generalized to wider phenomena in the real world) in the context of the research methodology and its findings. However, the nature of the present research makes such topics invalid—although this status is itself the subject of some debate—with the possible exception of “validity,” which is returned to later in the chapter. Of more relevance is the question over the extent to which the quality of qualitative research methods may be measured, which is what we turn to now, beginning with a brief perspective on the wider grounds for criticism.

In considering what criticism has been made concerning DA, an interesting viewpoint is expressed by Stainton-Rogers: she suggests that studies in critical social psychology have been largely (as of 2003 ) ignored by researchers in the experimental tradition. Consequently, there is a perspective that criticism of DA methodologies are often raised by its own proponents (e.g., see Charles Antaki and coresearchers, 2002, for a discussion of analytic standards: Emanuel Schegloff, 1997, on the issue of context: on the issue of cognition, see Teun van Dijk, 2006; Jonathan Potter and cowriter, 2003; Linda Wood and cowriter, 2000: on the omission of considerations of experience, unconsciousness, subjectivity, etc., see Bethan Benwell and cowriter, 2012: for the subjectivity of analysis, see Maria Stubbe and colleagues, 2003; and see Charles Antaki, 2012, on the subject of using mixed methods). The reader will have already seen how researchers in critical social psychology also frequently direct criticism in the direction of experimental methods research, despite the apparent lack of reciprocation.

Arguably, the most obvious and problematic issue concerning qualitative research in general, and one that can be understood as underlying many other points of criticism, concerns the question of how to measure the quality of qualitative research methodologies. This is the principal focus of the following discussions.

The question of how to measure the quality of qualitative research methods is the topic of a substantial debate among qualitative researchers (see, e.g., an interesting study by Peter Cooper and Alan Branthwaite, published in the 1970s, which argues for the maturity of qualitative methods by comparing the findings of a qualitative and quantitative study of the same phenomena, finding similar overall results). This evidently concerns the issue of how to determine the values of qualitative research. The problem can be condensed into three of its interconnected characteristics: first, the diversity of method; second, the perspective that conventional criteria of measurement such as reliability and validity, so well established in quantitative methodologies, are irrelevant; and third, the profound difference in epistemology between quantitative and qualitative researchers and their methodologies, which some researchers suggest is the source of the problem. In particular, Lucy Yardley, a psychologist at the UK’s University of Southampton, claims that the first two combined have led to a situation in which there is an absence of firm general guidelines relevant to the work of the qualitative researcher. It is this omission that Yardley, and Robert Elliott and his colleagues seek to address with their “evolving” proposed guidelines.

Specific to the field of DA in psychology, there is a further matter that impacts on the quality question, which concerns the use of the term “measure.” While many scholars debate and propose methods for addressing the quality question, the term “measure” does not seem to feature. The term is absent from relevant discussions offered by, for instance, Wood and Kroger, and Potter and Wetherell, in their respective accounts of how to do DA. There is perhaps one simple reason for this: the term “measure” implies a scale or a benchmark, a mark out of ten, which in turn implies “quantification.” Emphasizing the limited role for “quantification” in DA in general, Linda Wood and her colleague propose more appropriate phraseologies (how research claims can be warranted) as does Potter (how research claims can be validated), for instance.

Examples of how one can strategically approach the quality issue in DA methodologies (in general) are detailed by both Wood and Kroger, and Potter and Wetherell, with the former drawing on the latter. Wood and her colleague propose that the issue concerns warranting—how to give justification to and grounds for analytic claims. They question the application of the traditional notion of “validity” on the commonsense grounds that an analytic account can only ever represent one version of many possible versions of affairs, so it can never be considered as either true or false. If validity is traditionally considered to be the measure of research claims’ “fit” with the world as it is, then clearly DA studies cannot be evaluated on this basis. An alternative conceptualization of “validity” is needed.

Wood and her colleague’s solution for “warranting” centers on two principal components: the trustworthiness of the account, which can be addressed through ensuring that a clear and detailed description of all stages of research is included in the account, and the soundness of an account, which principally concerns the analytic section of the research report. A number of factors need to be adhered to include the grounding of analysis in speakers’ orientations (addressing the speaker’s understanding as displayed in discourse vs. the analyst’s interpretation), which refers to the “next turn proof” analytic tool encountered earlier; the coherence of analysis (a claim should satisfactorily account for exceptions and deviants in a discourse); the plausibility of an account in, for instance, how it relates (can be grounded) to other research work in similar areas; and the notion of “fruitfulness” (Potter and Wetherell’s term), which addresses the implications of a study’s findings for other work and what questions it might raise in terms of future research.

Specific to DP, Jonathan Potter, writing in the late 1990s, provides a set of four pragmatic guidelines. Interestingly, Potter describes his guidelines as “validation procedures” and wastes no time on debates around whether this term should be used in this type of research methodology or not: he quite simply reformulates it. Potter’s procedures start with the analyst’s attention to grounding their claims in speakers’ own understandings as displayed in discourse, which also serves as a check for interpretive claims—the “next turn proof’ aspect. This has correspondence to Yardley’s “sensitivity to context” principle. Second, attention to what Potter describes as deviant cases can be useful in assessing the sufficiency of claims—do deviant cases in a discourse, for instance, support an analytic claim or weaken it? This has far less synergy with Yardley’s principles probably because the explicit search for and importance assigned to “deviant cases” is very particular to DA, particularly DP. Potter’s third procedure concerns an account’s coherence with respect to previous studies in similar areas, which can be clearly related to Yardley’s first principle as well as her third, “impact and importance,” but which varies from Wood and her coworker’s understanding of “coherence.”

The last procedure concerns the reader, which Potter describes as the most important of the four. This refers to the inclusion of extracts of data (the “real” data) in research reports so that the reader is able to make their own judgment of the analyst’s interpretation. Although he does not elaborate on this point here, this is an interesting notion: it potentially makes the reader part of the analytic work as an active contributor and implicates the role of the researcher as interpreter. This can be contrasted with Yardley’s emphasis on a study’s impact and importance in the sense of how a study markedly contributes to the knowledge (does it tell us anything new, does it make a difference?). While it is certain that Potter has this as an objective for research, there is a suggestion that this is made more personal to the individual reader rather than the academic research community as a whole.

The present research reported here draws on Potter’s validation procedures, the application of which is discussed in the following section. As a final point, Potter creates a caveat: in his own words, “… none of these procedures guarantee the validity of an analysis. However, work in philosophy and sociology of science over the last 30 years has cast doubt on the possibility of such bottom-line guarantees in science, whether provided by observation, replication or experimentation” (1998a: 241: italics in original). Even “the independent audit” and “interrater reliability approach,” as outlined by Jonathan Smith and Linda Wood and her coworker, for instance, are no guarantee for an account’s warrantability according to this caveat.

11.5.1 Design

This section of the chapter turns attention from discussions around topics in methodology to those around how the present research was actually carried out.

A first point to note about the DP methodology is that there is no straightforward prescription for analyzing discourse. That is, there is no check box list of actions to take that will lead to the perfect analytical outcome. Derek Edwards and his coworker, in their book introducing DP, synthesize various features of discursive action and the relationships between them in a conceptual scheme referred to as the Discursive Action Model. It is organized into three principal themes: action (e.g., a focus on action rather than cognition), fact and interest (e.g., negotiating the dilemma of stake and interest), and accountability (e.g., the speaker’s displayed sense of accountability in reports). It establishes some useful guiding principles as well as some potential areas on which to focus research (such as, e.g., how speakers construct and manage their remembered accounts of past events as factual and authentic).

More practical as a framework for guiding methodology, Jonathan Potter and Margaret Wetherell map out a 10-stage guide to DA: research questions, sample selection, data collection, interviews, transcription, coding, analysis, validation, report writing, and application. This is not meant as a strict order of business for undertaking DA: the order in which these “actions” are engaged is entirely dependent on each specific research case. Both the Discursive Action Model and the 10-stage guide are used to inform our research, with the latter providing points of discussion here. Note that both sample selection and data collection are addressed in the following subsections, “Research Data” and “Participants and Ethical Considerations,” and that the topic of “interviews” is not relevant to the present study. “Validation,” in the following discussions, addresses how the present study approaches matters of validation procedures. “Report writing” is not relevant to our present purposes, and “application” is addressed in the final chapter of the thesis.

11.5.2 Research Data

The size and content of any sample selection are driven by the research question. Potter and his coworker emphasize that, in DA, the size of the research sample is not a determinant in a study’s success. The analyst’s priority is an interest in the language itself, how it is used and what it accomplishes, not the speakers. Most DA research generally samples a corpus of data from different sources or from the same source (e.g., see Robin Wooffitt’s intriguing 2001 study of verbal interaction between mediums and their clients finding “reported speech” to be a commonly used linguistic device, which, he claims, works to invoke “favorable assessments” of the psychics’ authenticity).

When collecting data, many of the same principles of conventional research methods are relevant: a consideration of ethics, for instance, and ensuring the appropriate permissions are gained. Preference is always for using naturally occurring language in interaction (i.e., with the complete absence of the researcher). But the use of surreptitious recordings would, for instance, be ethically questionable. This raises a particular question concerning ethics in respect of some of the data used here (public online discussion forum), which is addressed in the following subsection, “Participants,” as part of a general discussion on the ethical approach of the present study.

Firms were selected for the study based on the researcher’s prior relationship with their senior management and the nature of their business as having an emphasis on sharing and developing knowledge. This prior relationship transpired to be an essential factor in gaining the cooperative participation of both organizations. Several other organizations, where no prior connection or relationship existed, were also approached as potential participants but, while expressing support for the research project and its aims, all declined to become involved as participants. This suggests a sobering lesson and potential limitation for future research: the nature of the research methodology is such that, without a prior relationship of trust, potential participating organizations are unlikely to agree to take part.

Individual participants were not selected by me as the researcher: instead, these represent, in effect, an opportunity sample in that they happened to be present in the meetings that took place at times and dates when I, with the agreement of the organizations’ senior management, happened to be present at their respective premises (but not physically present in the meetings themselves). Nor was there any deliberate selection made of the meetings to be recorded or influence upon their topics of discussion. In this sense, while the organizations themselves are actively selected by me as the researcher, the actual participants are not. Nor was I physically present in any of the meetings themselves, not even in the guise of observer. I can claim, then, that this data can be considered as naturally occurring language in interaction.

The principal empirical basis for the present study comprises digital audio recordings of 13 individual meetings, collectively representing more than 15 hours of recordings, taking place in two different London-based organizations during March and May 2013. These meetings are regular, scheduled meetings in each case.

No instruments were used in the data capturing part of the project apart from a small digital audio recording device, which was positioned in meeting rooms in advance of meetings to be recorded. All participants were briefed in advance, verbally or via written instruction, of the nature and purpose of the study.

11.5.3 Participants and Ethical Considerations

This section of the chapter focuses on the ethical conduct of the study, with particular consideration given to the use of online discussion data. This is followed by a description of the participating firms. A description of the online data and its source is given at the start of the analysis in Chapter 15.

In compliance with ethical standards for research, two documents were prepared: a participant consent form and an information/briefing form. All participants were given both documents prior to any recording being undertaken. In most cases, participants were given a short verbal briefing on the nature of the research, in particular of their rights to withdraw from the study at any point. Participants, comprising organizational employees, contractors, and/or associates of two independent London-based private sector consultancy firms, were required to sign individual consent forms. All participants and their organizations are treated as anonymous in all aspects of the research.

With respect to the online discussion forum data, no overt permission was sought by the researcher from forum participants. Participants made their contributions voluntarily to a publicly available discussion forum, which is part of an international professional networking website: that is, access to this forum is not restricted to registered forum members only, although in order to access the site users must first register with the website itself. Note, though, that there is no restriction, fee, or qualification required in order to register with this website. A further important point to note is that this group, which is a networking group, publishes no explicit rules, guidelines, or other considerations in respect of members’ contributions and their use thereof. To post a contribution, one must register with the group, but access to its contents is available to any website member.

Was it ethical to sample and use this data collected in this way? The review of studies using data from computer-mediated communications (Section 6.4.3) gives an idea of the widespread practice of using such data in research. On the advantages of using this type of data, in their book on Discourse and Identity, Bethan Benwell and her colleague describe this as particularly “authentic and pure” because it requires no transcription and places the researcher in the position of a “lurker” as opposed to the traditional perception of the “scientist as observer” where the presence of the researcher can influence that which they observe. They, in fact, make no reference to any ethical issues with using data from such sources. In reality, the general thrust of research is more focused on the values and advantages of using this data rather than on any ethical issues that this might raise (e.g., see a study by Charles Antaki and colleagues, published in 2006, which compares everyday conversation talk with online forum discursive interactions). Moreover, in their account of the “revolution” of “Big Data,” Viktor Mayer-Schonberger and cowriter recently describe in some detail how large Internet businesses including Google, for instance, routinely “scrape” the Internet for content as data for their algorithms. There is no indication of any permission being sought from individual contributors.

To answer the initial question, is it ethical: it is claimed here that the use of such data—because of its circumstances in terms of accessibility to wider audiences and the implied acceptance by contributors that their data may be used for purposes other than they intended, together with the practical problems in attempting to gain permission from individual contributors, and the impact that such a requirements would have on the growing and valuable contributions from research in computer-mediated communications—is ethically acceptable. As a caveat, however, it should be noted that this is a gray area to say the least.

What follows is a brief description of each participating organization. Note that the context of individual meetings, where they form the basis of subsequent analysis, is described in each case of use: