Integrating conversation analysis and issue framing to illuminate collaborative decision-making activities

Negotiation in meetings

The analytical framework presented here builds on an extensive CA literature on negotiation in meetings (e.g. Asmuss and Oshima, 2012; Barnes, 2007; Boden, 1994; Choi and Schnurr, 2014; Deppermann et al., 2010; Djordjilovic, 2012; Ehlich and Wagner, 1995; Firth, 1995a, 1995b; Ford, 2008; Francis, 1995; Kangasharju, 1996; Markaki and Mondada, 2012; Maynard, 1984; Svennevig, 2012; Wasson, 1996, 2000). In an early work on negotiation in the workplace, Firth (1995b) argued that the communicative processes that constitute negotiation had been understudied up to that point:

Although communication is recognized to be a crucial feature of negotiation, the ways in which communication is accomplished have been overwhelmingly glossed over or taken for granted … how such things as arguments, counter-arguments, proposals and counter proposals … are interactionally accomplished … remains largely unresearched. (p. 4)

Since the publication of that ground-breaking work, many specific aspects of meeting negotiations have been examined in detail, for instance, who is entitled to make or assess a proposal (Asmuss and Oshima, 2012), distributed leadership (Choi and Schnurr, 2014), and transitions between activities (Deppermann et al., 2010).

The approach presented here integrates prior studies of specific kinds of sequences into a more comprehensive framework for the analysis of negotiation in meetings than what CA has previously offered (other than Maynard’s 1984 study of plea bargaining, which is a somewhat different kind of negotiation than the decision-making activities examined here). This approach can be used to analyze longer sequences than CA typically considers. For instance, the case study examined in this article includes 162 turns, spanning 10.5 minutes. Using CA to identify larger meeting patterns may be controversial among purists of the method, but others may recognize that this approach extends the power and usefulness of CA.

The activity of decision-making

As Drew and Heritage have noted, CA focuses on the ‘interactional accomplishment of particular social activities’ (Drew and Heritage, 1992: 17; Levinson, 1992). The unit of analysis for the framework presented here is the activity of decision-making. In my research in business, government, and academic settings, I have found this activity to be culturally recognized by participants; the overt goal of a broad genre of meetings in institutions is to achieve decisions (Schwartzman, 1989; Wasson, 2000). There seem to be widely shared cultural understandings about the activity, for instance, that decisions should be reached through a process of multiparty collaborative discussion and, ideally, the achievement of consensus. The conversations are usually guided by a facilitator. (Other meeting genres that do not revolve around decision-making are also found in institutions, but those will not be considered at present.)

The types of meetings examined here typically consist of a string of decision-making activities, of varying length and complexity, sometimes with transitional phenomena between activities (Deppermann et al., 2010). A long decision-making activity may include several shorter decision-making activities within it.

Decision-making activities begin when a speaker proffers a proposal (Francis, 1995; Maynard, 1984; Stevanovic, 2012; Wasson, 2000). The turn can take various forms; proposals may be constructed as suggestions, requests, or musings. Furthermore, proposals are not defined solely by the intentions of the speaker, but may also be constructed ‘post hoc’ by subsequent actions of other meeting participants (Wasson, 2000: 468).

Decision-making activities end more or less when consensus is reached. The consensus is articulated by someone, often the facilitator. Or the last person to hold a contrasting view may express an altered view that is in alignment with the rest of the meeting participants. The articulation of consensus may be followed by a chorus of agreement tokens, and sometimes by a ‘coda’, a reiteration of the rationale for the decision (Wasson, 2000).

In between the proposal and the achievement of consensus, a wide variety of turns can occur, and they can range from few to many. However, a key discovery of my research is that in the decision-making activities examined, most turns were organized into four types of sequences, all of which have been well described in the CA literature, although not necessarily with reference to decision-making. These four types of sequences are agreement sequences, disagreement sequences, information sequences, and joking sequences. They are described in more detail later in the article. I do not wish to claim that those four types of sequences are the only ones ever used for decision-making in meetings; further research may uncover additional types. One limitation of my research is that it has taken place in the United States; other parts of the world may reveal different cultural practices. Nonetheless, it is striking that in the 42 decision-making activities examined, only these four types of sequences appeared regularly. While these four kinds of interactional sequences have been analyzed separately in the CA literature, they have not previously been brought together in a single framework for the analysis of negotiations in meetings.

Sources of data: Linguistic and ethnographic fieldwork

The data at the heart of this study are transcripts of audio-recorded meetings. A total of 42 decision-making activities were examined, spread across seven meetings that lasted a total of eight hours. The data come from my dissertation research, which was an exploration of team decision-making at a large high-technology corporation based in the United States. I call the fieldsite Sigma Corporation. The recordings analyzed for the current article are a subset of a larger corpus of recordings, which also included events other than team decision-making.

The 42 decision-making activities ranged from 4 to 216 turns, with an average of 50 turns. Occasionally, the activities were broken up into several segments, for instance, when meeting participants were distracted by the arrival or departure of a team member, or by a side issue which arose out of the decision-making process but then emerged as its own topic of discussion before the group returned to the original issue.

The meeting recordings are complemented by extensive ethnographic data. From 1992 to 1994 I conducted 16 months of fieldwork at Sigma, spending full working hours there five days a week, and socializing after hours as appropriate. Over the course of this fieldwork I attended an extremely large number of team meetings, typically several a day. While I wrote detailed field notes about each meeting I attended, I only recorded a limited number due to the practical limitations of what I could transcribe. Recordings were made toward the end of the fieldwork, when I had established rapport with employees at the fieldsite. By that time, I was able to situate the meetings within my extensive knowledge of meeting participants and their relationships, as well as Sigma’s institutional structures and cultural logics.

Steps in methodology

The methodology I developed for integrating CA, IF, and ethnographic insights in the analysis of decision-making activities included the following steps. I present them in linear fashion for the sake of clarity, but in actual practice I went back and forth among different steps during the process of analysis. An example of the type of annotated transcript that resulted from Steps 1–3 can be seen in Appendix 2.

Focus on a single decision-making activity as case study

In planning this article an interesting dilemma emerged. It was necessary to take readers along on the analysis of an entire decision-making activity in order to communicate the methodology I had developed and its payoff. Yet most decision-making activities in my data set were far longer than the transcript examples typically presented in journal articles. My solution to this dilemma was to focus on a single case study. The full transcript of this case study can be found in Appendix 2. In the body of the article, only turn numbers from the transcript are referenced; the reader must flip to Appendix 2 to view the text. While this creates a somewhat unorthodox experience for readers who are used to seeing bits of transcript sprinkled throughout the text, it accommodates the space constraints of a journal article while enabling readers to access the transcript of a complete decision-making activity.

Furthermore, because the case study is 162 turns long, this article does not include the kind of line-by-line analysis that is characteristic of CA publications. In order to combine insights from IF, CA, and ethnography in a single article, each element is described with some brevity. The combination of these elements is precisely the article’s contribution.

The case study was selected from the set of 42 decision-making activities because it most clearly illustrated all four of the interactional sequences found in these activities, and because it formed an unbroken series of interactions with minimal interruptions or distractions.

Explanation of case study: The Extra Recognition and Awards team evaluates an award application

The negotiation examined here occurred during a meeting of the Extra Recognition and Awards (ERA) team. The ERA team was based at the administrative headquarters of one of Sigma’s businesses. About 750 employees worked at this facility. The considerable majority were design engineers; the rest belonged to support functions such as human resources, finance, and information systems management. The ERA team was cross-functional, meaning that members came from diverse functions in the organization. The team organizer, Anne, worked in the Benefits area of Human Resources. (All names of Sigma employees and teams are pseudonyms.)

The purpose of the ERA team was to arrange special rewards for employees who had worked on projects outside of their regular responsibilities. While employees’ regular responsibilities were rewarded through annual performance reviews, their outside projects were not otherwise formally recognized. Any employee could nominate any other employee for an ERA award by filling out an application.

The meetings of the ERA team primarily consisted of evaluations of applications that had been submitted to the team for consideration. Team members decided whether or not the project described in an application deserved a reward, and if so, they negotiated the nature of the reward, which ranged from American Express Gift Cheques in varying amounts to items such as a weekend at a resort. An especially popular reward was the use of the desirably located ERA parking space for one month. Occasionally the team engaged in other tasks, such as the development of a survey to assess Sigma employees’ perceptions of the ERA team’s performance. But for the most part, team meetings consisted of a series of decision-making activities in the form of application reviews and selection of awards. Team members arrived at decisions through a process of multiparty, collaborative discussion, facilitated by Anne. All decisions ended in consensus. Their interactions showed a marked preference for agreement. I observed, recorded, and transcribed five meetings of the ERA team.

The decision-making activity examined here was the ERA team’s evaluation of one application. It was the sixth and last application considered during that day’s one-hour meeting. Anne, the team organizer, started the discussion by reading the application out loud while displaying it on an overhead projector. The application requested an award for an employee because he had developed a new process to procure business cards that would save Sigma Corporation US$25,000 per year.

The decision-making activity consisted of 162 turns, listed in their entirety in Appendix 2. Meeting participants included the following:

Findings from analysis of IF

Findings from CA

While I started the CA of decision-making activities by identifying the interactional move illustrated by each turn at talk, I found greater analytic purchase in grouping turns into interactional sequences. It turned out that most of the turns in the 42 decision-making activities examined could be grouped into four types of interactional sequences, each well documented in the CA literature. There were some turns that fell outside of these four types of sequences, for instance when the conversation went off on a tangent, or when participants engaged in metacommentary about their process. However, the consistency across decision-making activities was remarkable. Of these 42 activities,

A total of 10 activities included all four sequences. The case study was selected, in part, because it provided clear examples of all four sequences.

Integration of IF and CA: Patterns across the decision-making activity

Once IF and CA are each applied to a decision-making activity, it becomes possible to integrate their findings and identify patterns across issues, frames, and interactional sequences. This kind of integration starts to fill in a gap in studies of negotiation and conflict resolution that advocate an interactional IF approach. While many such studies acknowledge the value of examining interactional sequences, hardly any actually include a CA component in their analysis (Dewulf and Bouwen, 2012; Dewulf et al., 2004). None of these studies has displayed a deep understanding of the range of interactional sequences commonly found in decision-making activities.

By integrating IF and CA perspectives on the case study, new insights about the overall trajectory of the decision-making activity emerged which, in turn, revealed a particular culture of negotiation that characterized certain kinds of interactions at Sigma Corporation. The relationships between shifts in issues, shifts in frames, and shifts in interactional sequences revealed by the annotated transcript (Appendix 2) highlighted three distinct phases of negotiation:

About halfway through Phase 2, some participants began to articulate a broader frame that accommodated both sets of criteria, but the frame was not accepted by all. Anne asked for a vote on award amounts, which indicated that she had given up trying to achieve frame consensus; however, the vote was tied, so even this did not lead to a resolution. The tie vote did, however, result in participants’ joking, which appeared to be used as a way of reducing tension and a form of relationship repair.

These three phases of negotiation are revealing. From a CA perspective, they demonstrate a preference for alignment, agreement, and affiliation (Lindström and Sorjonen, 2013; Pomerantz, 1984; Sacks, 1987). This can be seen in the way that disagreement is delayed until turn 40, in participants’ dogged efforts to reach consensus until they finally achieve it, and in the extensive joking in the latter part of the activity, which contributes to the repair of good feeling among team members. In fact, Phase 2, where the main disagreement is articulated, only takes up 66 turns out of the 162 turns that make up the decision-making activity as a whole. In some sense, then, the actual negotiation primarily occurs during only 41% of the turns. Phases 1 and 3 are more about team members’ displays of mutual alignment and affiliation.

The trajectory of the case study was typical of collaborative decision-making activities at Sigma in several ways. First, the delay of disagreement was common. Second, extended disagreement was often followed by joking. All the seven decision-making activities that lasted more than 90 turns concluded with joking sequences. However, not all decision-making activities featured protracted disagreements. In some cases, there was no disagreement at all; an initial proposal was simply accepted unanimously. In other cases, a disagreement was raised but resolved quickly. In such situations, jokes might or might not occur. A total of 17 of the 42 decision-making activities concluded with a joking sequence.

Decision-making patterns in the context of Sigma Corporation

These decision-making patterns can be further illuminated by contextualizing them in Sigma’s cultural logics and institutional practices. I review key points from Wasson (2000) in the following discussion. The habitus of ERA team members was shaped by their lived experience as managers and professionals in an American corporation. Decision-making practices among managers and professionals in corporate America are shaped by the political exigencies of their work life. Organizations ultimately exert power over managers and professionals by controlling their career advancement (Kanter, 1977; Whyte, 1956). While corporations maintain an ideology of objective meritocracy, underpinned by rationalized ‘performance reviews’, employees recognize that promotions are closely linked to their socially constructed reputation (Jackall, 1988; Ourousoff, 1993). Managers and professionals are therefore highly sensitive to being scrutinized by those who might evaluate them. At Sigma Corporation, I found that the level of surveillance to which an employee felt subjected varied depending on the context and relationships among participants.

Employees’ concerns about managing their reputation produced a disposition toward caution, a tendency to display conformism and speak with care in settings where speakers were concerned about the surveillance of their interlocutors. The caution and conformism that corporate settings tend to inculcate were first described in a spate of literature that appeared in the 1950s (Dalton, 1959; Mills, 1951; Riesman, 1950; Whyte, 1956) and continues to surface in more recent corporate ethnographies (Jackall, 1988; Kanter, 1977; Kunda, 1992).

At Sigma Corporation I found that decision-making on cross-functional teams was particularly shaped by team members’ disposition toward caution. ‘Cross-functional’ teams were those that drew their membership from across the organization. Their composition signified that the team represented the facility as a whole; they dealt with matters that potentially affected most employees. At the same time, their tasks were not usually of critical importance. The teams’ goals were not closely related to most of their members’ primary work activities; few team decisions had the potential to threaten their individual careers. This enabled team members to be particularly flexible in their opinions. Furthermore, team members did not know each other well enough to be fully open and relaxed in their conversation.

An ideology of consensus decision-making prevailed in cross-functional team meetings at Sigma. This ideology involved the notion of equal voice for all participants, and the principle that everyone should subscribe to the final decision. It was described in the following way in a team training manual prepared for Sigma:

All members have the opportunity to participate in the discussion, to ask questions and give opinions … the entire group arrives at a decision … no averaging, majority-rule voting, or horse-trading.

The value placed on consensus and equality in these meetings meant that aggressive, domineering behavior was interpreted negatively. There was no way for participants to enhance their reputation by displaying the kind of confrontational style that in other contexts might be positively interpreted as showing ‘leadership’. Consequently, the main source of symbolic profit for participants in these meetings was to develop a reputation as a good team player and to build their alliances with each other. Of course, it was always possible for them to damage their reputation, by strongly voicing an opinion that alienated other meeting participants, or that provided fodder for damaging gossip. Team members were therefore oriented not only toward accumulating social and symbolic capital in these meetings, but perhaps more importantly, toward preventing its loss.