Turning Points and Conflict Transformation in Mediation

Mediation outcomes can include no agreement, an agreement that resolves the immediate problem but leaves issues unexplored, or more transformative outcomes, such as improved understanding and relationships between the parties. Although many scholars and practitioners have identified strategies that mediators can use to encourage movement toward agreement, mediation practice is often described as intuitive and more of an “art” than a science. A recent study of mediator approaches found that even when mediators report they identify with a given mediation style (such as facilitative or transformative), in practice they still use a wide variety of strategies to meet the needs of any given mediation (Charkoudian et al. 2009).

In this study, our goal was to examine role‐played mediation discourse to identify patterns of mediator and disputant communication that seem to be able to change the nature of the interaction from adversarial to cooperative, and to uncover any differences in communication patterns that lead to conflict resolution versus conflict transformation. Scholars who have studied relational orientations to mediation, such as the transformative model, have found that it improves relationships (Bush and Folger 2005; Poitras 2007) and leads to better organizational‐level outcomes (Bingham et al. 2009; Nabatchi, Bingham, and Moon 2010).

Because of our interest in conflict transformation as a potential mediation outcome, we are attracted to theories that examine turning points. In interpersonal communication scholarship, turning point theory (Baxter and Bullis 1986) emphasizes the importance of critical moments that “trigger a reinterpretation of what the relationship means to the participants. These new meanings can influence the perceived importance of and justification for continued investment in the relationship” (Graham 1997: 351). Turning points in the relational development context are units of analysis that span across time, but turning points have also been defined as moments in conflict that change the nature of negotiation or mediation interaction (Druckman 2001, 2004; Leary 2004).

Mediators are especially interested in recognizing turning points so they can capitalize on them (Leary 2004). While previous mediation literature has discussed mediator timing (e.g., when mediators decide it is the appropriate time to take an action in a mediation) (Castrey and Castrey 1987), making transitions between stages (Wildau 1987), the emergence of cooperation (Poitras 2005), and turning points (Blake 1999; Cohen 2009), there has been limited operational definition of the turning point and no direct investigation of the impact of turning points on conflict transformation. This study will, therefore, contribute to existing theoretical and practical knowledge of how mediators and disputants influence mediation interaction in ways that may lead to transformative outcomes.

The following review summarizes the turning point literature in mediation and reviews Druckman's turning point framework as applied to negotiation. Although there are several definitions of turning points, all appear to include the idea that turning points can fundamentally alter an individual's perception of oneself, the other party, or the issue at hand. Several scholars have defined and examined turning points in the context of long‐term negotiations (Putnam 2004; Crump and Druckman 2012), but others have defined them as instants in mediation interaction that appear to change the nature of the communication (Blake 1999; Cohen 2009).

Conflict transformation in this study refers to conflict outcomes that include a change in the relationship between parties. As described by Robert A. Baruch Bush and Joseph P. Folger (2005), transformative outcomes occur when parties feel more in control of their own destiny, as well as when parties are able to openly acknowledge the experience and or emotions of the other party.

Johan Galtung (1996) suggests that transformation is the most optimal conflict outcome. He posits that transformation occurs only when a conflict becomes fully articulated to all parties; that is, in addition to being aware of incompatible behaviors, they also understand the attitudes, beliefs, and emotions underlying those behaviors. And Linda Putnam (2004: 276) defined transformation as a “moment[s] in the conflict process in which parties reach new understandings of their situation, ones that redefine the nature of the conflict, the relationship among the parties, or the problems they face.”

Given this understanding of conflict transformation, this study assumes that turning points are critical moments in mediation that can lead to greater understanding between parties in a dispute and have the potential to alter their relationships. We do not contend, however, that all turning points will lead to conflict transformation. A key question underlying this study is whether there is a relationship between the characteristics of turning point interaction in mediations that lead to agreement only as compared to those that also include conflict transformation.

Orlando Blake (1999) identified three categories of turning points by interviewing twenty divorce mediators and analyzing the transcripts of their mediation sessions to determine the relationship between the turning points and the final agreement. These turning points were deemed to have occurred when disputants displayed the following behaviors: acknowledging (defined as expressing understanding), engaging (defined as active listening and problem solving), and being forthright (self‐disclosure and expressing feelings). Blake's study found that reciprocity led to more agreements, earlier statements of understanding appeared to lead to cooperation, and expressing too much emotion too early in the process was often seen as manipulative.

In another study of turning points in divorce mediation, Orna Cohen (2009) interviewed fourteen disputants to determine which features they believed either facilitated or impeded agreement. Cohen found that it was important for mediators to balance their attention to the substantive details of the dispute and to disputants' emotions. The two turning points Cohen discusses that are most relevant to workplace disputes were the development of insight (awareness of the other party's point of view) and a shift in focus (from the past to the future).

These are the only two studies we could find that directly examine turning points in mediation. These studies rely on interviews or surveys, in which either mediators or disputants identified their perceptions of important factors that influenced the interaction. Blake extended this approach to include analysis of transcripts and agreements, and to make more direct links between the turning point interaction and final outcomes. For our study, we wanted to directly examine mediation interaction, and we therefore needed an operational definition of turning points to help identify them. For this, we turned to Druckman's framework of turning points in negotiation.

Druckman's (2004) definition of the turning point breaks it down into three interrelated components: precipitants, departures, and consequences. A precipitant is an internal or external influence on the interaction that results in a change in the pattern of interaction for one or both parties. Druckman (2001) defined an external precipitant as an event that occurs outside the negotiations, such as a policy or leadership change, or intervention by a third party. While the mediator is a third party, he or she influences interaction during the mediation, and is therefore considered internal.

Because our study used role‐play simulations in which we did not include any external information or changes during the simulation that could qualify as precipitants, we only focus on internal precipitants here. In a negotiation simulation by Daniel Druckman, Mara Olekalns, and Philip Smith (2009), internal precipitants were defined as occurrences within the negotiation that modified disputants' perceptions of each other or the process. In this study, negotiators completed a survey indicating their perception of the other party's cooperativeness (among other variables not germane to this project). They also watched a video recording of the negotiation and identified critical moments in the negotiation that changed the nature of the interaction. The authors found that participants identified more internal precipitants when at least one negotiator was perceived as cooperative, and in this condition they also experienced more positive outcomes.

Druckman (2004) refers to departure as the moment that signals change from one stage to another in a negotiation. He has identified four common features of departures: (1) a change from earlier events or patterns occurs, (2) a change occurs in the relationship between the parties, (3) the changes are self‐evident such that observers are likely to agree that a change occurred, and (4) an action is taken by one of the parties that leads to consequences for both. Druckman also observed that departures may create an instant reciprocal shift in the negotiation, such as movement to problem solving, or the change may be more gradual; he uses the terms abrupt and nonabrupt to describe these departures accordingly. According to Druckman, we know a departure occurred based on the consequences, which may be positive (greater cooperation) or negative (increased competition or adversarial positioning). Because of our focus on conflict transformation, we only examined positive departures in this study.

Druckman (2004) has noted that many questions about the nature of turning point interaction in negotiation remain unanswered. Specifically he asked, “How do negotiators recognize a departure when it occurs and what actions do negotiators take to bring them about?” (2004: 186). He also points out that future research is needed to examine whether turning points occur when disputants' responses are synchronized. Finally, Druckman asked whether emotion plays a role in how disputants interpret each other's conversational acts and how such interpretations influence their next moves. In this study, we have sought to help answer several of these questions by more closely examining interaction that characterizes turning points in mediation, specifically:

This study used a discourse analysis of simulated mediation role plays. Although there has been a long and robust history of the use of role plays and simulations in both negotiation research and training (Ebner and Efron 2005), discourse analysis is typically conducted with naturalistic data (Putnam 2010a; Stewart and Maxwell 2010). Nonetheless, we have found precedent for the use of simulations for the study of natural linguistic phenomena, such as apologies (Demeter 2007) and invitations (Rosendale 1989).

Deborah Rosendale defined role play as “participation in simulated social situations that are intended to throw light upon the role/rule contexts governing ‘real’ life social episodes” (1989: 487, citing Cohen and Manion 1980: 224). As this definition suggests, role play invites participants to interact “as if” they are in a situation that they are likely to have encountered previously and thus use conventional norms to jointly coordinate their interaction. In looking at social cross‐gender invitations, Rosendale argued that it would be difficult, if not impossible, to observe naturalistic invitation interactions because social actors would not behave the same way if they knew they were being observed. Similarly, she suggested that asking participants to describe how they would phrase an invitation is not a reliable indicator of what they would actually do. Role play is, thus, considered a fruitful way to get at difficult‐to‐capture human interactions without violating privacy.

The data used to inform this project were generated for an earlier study designed to empirically test the hypothesis that mediations in which mediators elicit emotion result in greater conflict transformation than mediations in which mediators ignore or minimize emotions (Jameson et al. 2009). Experienced mediators (with a minimum of one year of mediation experience in the local court system and/or in community mediation) were recruited for participation in this study through a local community dispute resolution center; they received six hours of advanced mediation training in exchange for their agreement to mediate two role plays for the study. Undergraduate and graduate student volunteers from one southeastern university played the role of disputants. (In some cases, students received extra course credit for their participation.)

The central conflict issue in this dispute is that one party has been dependent on the other for help in the workplace and the second party is frustrated with the interruptions. This scenario was familiar to students, because even if they had no work experience they had participated in projects that required group cooperation and in which there is often a perception that at least one party is doing more than the others. The role‐play scenario had been designed for a previous study by the first author, and had undergone a manipulation check to ensure that the disputants were perceived as interdependent, that there was a sense of urgency to resolve the conflict, and that the conflict scenario was considered realistic (Jameson 2001).

The procedure for each role play was as follows. Mediators were given a paragraph that briefly described the conflict and were told that a supervisor had recommended that the parties go to mediation. Student participants were given their roles and then sent to separate rooms to read their roles. They were then asked to complete a questionnaire that asked them to state their interests in the conflict and answer questions about their emotions and perceptions of the other party (for another study). They were told to think about the interests and concerns of their character, and after about ten minutes both parties were invited to enter a private office with their mediator. They were told that the mediation could last for as long as two hours if necessary. The researcher or assistant then turned on the video camera and left the office with the door closed.

A total of twenty video‐taped role plays were transferred to DVD for this project. All three investigators viewed the twenty role plays, and after discussion we purposively selected ten for analysis. The selection criteria were the following: (1) face validity, for example, all parties remained in character and the mediation simulation was coherent; and (2) maximum variation, for example, we chose role plays that allowed for greatest comparison and contrast (Geisler 2004).

In five of the ten role plays that we chose not to include in the study, “disputants” were unrealistically collaborative. In three of them, mediators were excessively directive in bringing parties to agreement. One mediator had little experience compared with the others, and one role play lacked internal coherence (the disputants did not consistently stay in character throughout the role play).

The ten role plays that we included in the study were recorded on 327 minutes of video and forty‐one single‐spaced transcript pages. A basic level of transcription was used that included verbatim text as well as brief annotations to capture nonverbal communication where possible (capture of nonverbal interaction was inconsistent due to varying levels of video quality). This level of transcription was appropriate because the analysis involved a holistic approach focused on the emergent nature of the discourse (Stewart and Maxwell 2010).

In their study of adversarial narratives in mediation, Katherine Stewart and Madeline Maxwell (2010) used an iterative process of reading and annotating naturalistic mediation transcripts to identify discursive strategies and the emergence of narrative themes. In describing the use of discourse analysis in negotiation research, Putnam (2010a) argued that discourse can be used to identify how parties co‐construct the negotiation through their interaction. Druckman (2001) used process tracing to locate turning points by locating the moment when the pattern shift in interaction occurred and working backward to locate precipitants. Following these models, our data analysis proceeded in several stages.

In the first iteration, all three investigators read three transcripts and made annotations in accordance with instructions to (1) locate a place in the transcript when the nature of the interaction seemed to move from adversarial to cooperative, (2) identify the nature of the interaction that preceded that turning point, (3) identify the nature of the disputant communication following the proposed turning point, and (4) examine the agreement for evidence of conflict transformation.

We coded the transcripts for evidence of conflict transformation if either one of two conditions was met: the transcript included explicit statements about the relationship, such as “we should get to know each other better,” and/or the written agreement included relational as well as substantive elements, such as “the parties agree to be more respectful of each other's needs.”

After annotating the three role plays in this way, we discussed our interpretations and constructed more specific coding rules for the analysis of all ten transcripts. The unit of analysis was the speaking turn, defined as an entire utterance of each speaker. We defined the departure as the speaking turn after which there was an evident change in interaction between disputants. Because we were only interested in positive departures, we only noted departures after which the disputants cooperated to reach an agreement. While we had intended to code the difference between abrupt and nonabrupt departures, this proved difficult because of the short‐term nature of our role plays. We, therefore, eliminated this coding category from our analysis.

The first author collected all coded transcripts and compared the interpretations, and all three investigators discussed differences until consensus was reached. This iterative process of independent coding followed by collective negotiation adds validity to the research process. Our diverse backgrounds, experiences, and levels of familiarity with mediation and conflict research allowed for rich discussion, and constant interrogation of the data (Stewart and Maxwell 2010).

Once we agreed on the identification of departures in the transcripts, we identified the three preceding speaking turns as internal precipitants, which we coded as substantive (the majority of the discourse in the speaking turn included reference to the conflict itself) or procedural (the majority of the discourse in the speaking turn was related to what the parties should talk about next). While technically internal precipitants could have occurred earlier in the mediation, we chose this method because of our interest in how the interaction led to departures. Previous turning point studies interviewed participants to ask them at what point they decided to change their interaction from adversarial to cooperative, but this was not possible in our study, and our process was consistent with a discourse analysis method. Because all role plays reached agreement, the outcome was coded as either resolution (agreement only) or transformative (including evidence of conflict transformation, as described above).

The final step in the analysis was to capture the flow of each turning point and categorize the discourse of the precipitants and departures. We used the inductive coding strategy of constant comparison to develop categories of mediator and disputant discourse (Glaser and Strauss 1967): through constant comparison, the researcher is not constrained by an a priori set of coding categories, but instead uses an iterative process to compare the data with existing theoretical frameworks while also looking for new themes or categories. For example, based on previous research on mediator tactics, we expected to find discourse categories such as “paraphrasing” and “focus on common ground” (Blake 1999; Charkoudian et al. 2009). At the same time, we wanted to be open to new types of mediator communication that might be characteristic of precipitants and departures.

While all ten mediations led to a written agreement, there were differences in the level to which the discourse and final agreements showed evidence of conflict transformation. Serendipitously, we found five transcripts with evidence of transformation and five that achieved resolution only. Given that all participants had the same description of their roles, mediations were expected to follow similar patterns. We did find a fairly consistent pattern in which the precipitants consisted of mediator–disputant–mediator speaking turns, and the departure was consistently a turn spoken by a disputant. Note that in the role play descriptions that follow, pronouns for Chris and Pat change depending on the sex of the students who portrayed them.

Nearly all ten of the role plays began with the adversarial narrative reported by Stewart and Maxwell (2010): the mediation typically began with an initial accusation, self‐justification for one's own behavior, and explanation for the other's behavior. The typical response was defensive, often accompanied by a counteraccusation. In two of the role plays, the first disputant was more neutral, expressing confusion about the other's behavior as opposed to blame. The nature of the precipitant and departure discourse was different enough to support Stewart and Maxwell's (2010) claim that every conflict interaction is unique.

Research Question One asked: What are the features of discourse that characterize departures in mediation interaction? The departures were most commonly coded as acknowledgment (eight of the ten were in this category). Acknowledgment was a statement of either new understanding, explicit recognition of what the other party was experiencing (to be discussed further below in response to Research Question Three), or appreciation of an apology or recognition previously offered by the other party. Examples of departures coded as acknowledgment include:

One departure was coded as problem solving:

One departure was coded as communicating emotion, in which one disputant described the emotional context of the conflict and the other party nonverbally agreed by nodding her head:

In summary, we found that statements that acknowledged new information, recognition of the other party's experience, receipt of recognition from the other party, or statements that explicitly communicated emotion most often characterized departures that led to cooperation for the remainder of the mediation.

Research Question Two asked: What are the features of discourse that characterize internal precipitants in mediation interaction? With the exception of two role plays, the preceding three turns before the departure consisted of a speaking turn pattern of mediator–disputant–mediator. To describe these results, we will first focus on mediator discourse and then characterize the disputant discourse. Table One summarizes the categories of discourse that characterize precipitants and departures in this study.

The mediator strategies that directly preceded departures were categorized as encouraging problem solving (8/18 precipitants), providing insights and paraphrasing (5/18), focusing on common ground (3/18), and attending to emotion (2/18). Mediators often employed more than one type of discourse in the same turn, such as focusing on common ground and then encouraging problem solving. The following examples illustrate mediator discourse:

The majority of mediator precipitants were coded as substantive (thirteen) rather than procedural (five). But the mediator discourse often included some attention to substance while also subtly directing the process. Mediator precipitants that were coded as procedural include:

These examples do not include reference to the content of the dispute.

In contrast are two examples (below) of precipitants coded as substantive:

In the latter example, the mediator makes a substantive comment, but this can also be seen as procedural in that the mediator is encouraging Chris to problem‐solve.

We also coded twelve disputant speaking turns as precipitants. These included acknowledgment of the other party (with or without explicit recognition) (5/12), communicating emotions (2/12), stating a desire to reconcile (2/12), accepting responsibility (2/12), and problem solving (1/12). As might be expected, all disputant precipitants were coded as substantive. The following are examples of disputant precipitants:

Research Question Three asked: Are there differences between the characteristics of turning point interactions in mediations that result in resolution only and those that include evidence of conflict transformation? As indicated above, five of our role‐played mediations showed evidence of transformation and five did not. An example of a mediation agreement that showed evidence of transformation included the following elements:

1. Chris and Pat agree to be more open in our communication regarding work‐related issues.

2. They further agree that personal issues are not improper in the workplace because it affects their performance.

3. Pat agrees to honor deadlines in order to appreciate the effect on Chris's job performance.

An agreement that we coded as resolution (without transformation) included:

1. Pat can bring important and urgent questions directly to Chris. Items that are less important can wait until the end of the day. We agree to meet each day at 4:30 for thirty minutes.

2. In the event that Chris appears hostile to Pat, it's ok to make Chris aware.

3. Agree to re‐evaluate in thirty days.

In the role plays coded as including evidence of transformation, we found that all departures were coded as acknowledgment. Four out of five of these also included direct statements of recognition, such as in the example below:

Three of the five departures in role plays coded as resolution only also included statements of acknowledgment, one was coded as problem solving, and one was coded as being forthright. Mediator precipitants in the resolution‐only role plays were most frequently characterized as encouraging problem solving (six instances compared with two in transformative role plays) or focusing on common ground (two times compared with one in transformative role plays).

Mediator precipitants in the transformative role plays were more frequently characterized by providing insights and paraphrasing (four compared with one in the resolution‐only role plays) and attending to emotions (two), which occurred only in transformative role plays. We found no difference in the number of substantive versus procedural mediator precipitants between the transformative and resolution‐only role plays. Consistent with the results noted above, however, most of the substantive precipitants in the resolution‐only role plays directed the parties to consider ideas for problem solving.

Looking at patterns of disputant precipitants (all of which were substantive), four out of five transformative mediations included communication of emotion and/or acknowledgment. In the resolution‐only mediations, disputant precipitants were more diverse, including statements of acknowledgment as well as the expression of a desire to find a solution, acceptance of responsibility, and problem solving.

Because our interest is in illustrating how interaction leads to departures and creates turning points, we share the sequence of precipitants and departures in two role plays. The first is an example of a role play coded as resolution only:

This example is typical of the mediation role plays in which mediators took a stronger problem‐solving orientation. While the departure shows evidence that Chris has developed a new understanding of Pat's situation, it is hard to know whether Pat would hear this as recognition. Some of the earlier discourse in this mediation reveals that Pat had also communicated new understanding of Chris, so there is some reciprocity here. After the departure, the parties were focused on problem solving, but the mediator took the parties off that track with evaluative questions about Chris's behavior. The mediator asked Chris how she felt about behaving selfishly and asked Pat how he felt about the way Chris treated him. The evaluative nature of these questions may have backfired by shaming Chris. While the parties came to an agreement, there were no outright signs of recognition or relational transformation.

The second example is from a mediation that included evidence of transformation:

This example is typical of the mediations that show evidence of transformation in that the mediator summarizes some substantive content, one party expresses emotion and accepts some responsibility, the mediator then asks the next party to engage in perspective taking, and the second party reciprocates by offering a statement of new understanding. In this role play, Chris seems to offer recognition by acknowledging his own attribution error when he says “I have misinterpreted her actions as being lazy … but to know that you need more orientation, I didn't know that.” In the subsequent turn, Pat reciprocates by acknowledging Chris's experience and offering recognition by saying “so I do hope that you get the promotion. I know that if I were up for a promotion I probably wouldn't have helped me either.” The consequence of this departure is that parties talk directly to each other for the first time and focus on problem solving, and their agreement includes being more open with each other about both work and their personal lives.

The preceding description of results suggests that it is the mutual influence of mediators' and disputants' communication that leads to turning points in mediation. This confirms what conflict scholars already know: that unfolding interaction is a complex process of social construction negotiated by all participants (Putnam 2010b). This also confirms what experienced mediators know: there is no “magic bullet” tactic that will lead parties to agreement. The skilled mediator learns how to build rapport and trust with the parties (Goldberg 2005; Poitras 2009), and determines where the parties are and how to ask the right question to help improve the parties' mutual understanding and reach agreement (Bush and Folger 2005).

Our findings have applications to both mediation theory and practice. With regard to theory, the findings support Blake's (1999) categories of disputant turning point behaviors. His categories included acknowledgment (expressing understanding), engaging (proposing solutions), and being forthright (expressing feelings and accepting responsibility). Like Blake, we also found that reciprocity encouraged agreement and earlier statements of understanding led to cooperation. In terms of mediator tactics, our results also support Cohen's (2009) findings that the tactics of providing insight (through paraphrasing and attention to emotion) and of shifting focus (often by encouraging problem solving) help produce turning points.

We had not expected to find that the turning point often occurs when parties are encouraged to begin problem solving. It seems that in our role plays this was the point at which parties realized they had something in common and were capable of cooperation. But on the other hand, problem solving and providing solutions did not typically lead to conflict transformation, a result consistent with Bush and Folger's (2005) model of transformative mediation.

Implications for practice involve the role of the mediator as directive versus empowering, the mediator's use of nonevaluative language, and making distinctions between expressing understanding versus giving recognition.

Bush and Folger (2005) have argued that recognition must come from the parties internally or organically — it cannot be forced upon them. In this study, several mediators used more directive strategies, such as directly asking the parties to engage in perspective taking or offering solutions. In one role play, we found such directive strategies to be fruitful, as one departure came directly after the mediator had asked Chris if he could put himself in Pat's shoes. In this role play, the disputants appeared to need a little push, and the mediator's intervention facilitated transformation as Chris moved from an initial position of detachment to proposing that he and Pat could talk about personal issues that might be interfering with their work.

On the other hand, in several role plays, the mediator appeared to be overly controlling. While the parties did reach agreement, this discourse may have prevented conflict transformation. Mediators in most of the resolution‐only role plays made explicit recommendations for solutions. While this discourse helped the parties reach agreement, parties often remained distant and focused primarily on getting their own needs met. In two role plays, we found that the mediator's paraphrasing used evaluative discourse, which distracted the parties and took them in more adversarial directions before they returned to cooperation.

In the role plays with greatest evidence of conflict transformation, mediators were less directive, mostly asking questions, paraphrasing, and identifying common ground. In these role plays, the disputants did most of the talking, they accepted responsibility for their role in the conflict, and they offered explicit recognition of the other party's experiences. These patterns support Bush and Folger's (2005) claim that parties must experience empowerment and recognition on their own, it cannot be forced upon them, in order to achieve transformation.

Mediators differed in their ability to recognize and capitalize on acknowledgments, which influenced whether those acknowledgments became precipitants that led to departures. Two role plays in which apologies were offered are illustrative. In one role play, Chris offered an apology to Pat early in the mediation, but the apology was not acknowledged by the mediator or by Pat. Chris apologized again at the end of this mediation, but the second apology also remained unacknowledged. We coded this role play as resolution only. In contrast, in a different role play, Chris offered an early apology, and during the mediator's next turn she asked Pat if he would accept Chris's apology. This led to an explicit acknowledgment, and this role play's outcomes were more transformative.

In several role plays, the mediator immediately emphasized one party's acknowledgment of the other, suggesting that mediators are able to recognize and act upon critical moments when they occur (Leary 2004). Importantly, paraphrasing moments of acknowledgment using neutral discourse such as “How do you feel about what Chris has said?” was more likely to lead to transformative outcomes. When mediators paraphrased using evaluative discourse such as “Do you think being sarcastic is the best way to respond?” the interaction often went briefly off track and resulted in mediation outcomes of resolution only. Certainly, the idea that mediators should remain nonjudgmental is not new, but capturing mediator discourse reminds us how difficult it can be to maintain neutral language in the moment. Mediators must not only be aware of potential precipitants in interaction, but they must also respond carefully using neutral or open‐ended discourse. This is an important skill that mediators may need to practice, especially if they hope to achieve conflict transformation.

A discourse analysis of mediation role plays highlights the difficulty of discriminating between communication that states new understanding and communication that offers recognition of the other's experience. Consider the following two statements:

In the former statement, Chris expresses new understanding of Pat's situation, while in the second example Chris acknowledges that Pat's friendship is important to her and she is an asset to the company. Both of these speaking turns were coded as acknowledgment in our coding scheme, but according to Bush and Folger (2005) they could have very different impacts on mediation outcomes. Based on our interpretations of the role plays, we suggest that some acknowledgments expressed empathy and connection with the other party more explicitly, and those were more characteristic of role plays with evidence of transformation. We acknowledge here, however, that recognition is most likely “in the mind of the beholder,” meaning that only a disputant could tell us if he/she interpreted the other party's statement as an expression of recognition. This limitation of our study design is discussed further below.

An obvious limitation of this study is the use of role plays. While the simulations captured lifelike conflict dynamics, workplace conflict interaction is complicated by concerns about retribution from coworkers or supervisors, and the fear of being labeled as incompetent or as a troublemaker (Jameson 2001). Although these role‐play interactions suggest those concerns were simulated to some degree, all ten of the cases came to agreement in less than sixty minutes, with five reaching agreement in under thirty minutes. These results are not typical, and clearly there was some social desirability bias that facilitated these participants coming to agreement in the role‐play exercise.

The specific details of this role‐play simulation also likely influenced the nature of the interaction, precipitants, and departures. It is quite likely that there are other characteristics of precipitants and departures that this role play would not have revealed. As described above, reliance on text analysis did not allow for participant identification of the moments they would define as precipitants, which would have provided valuable insight.

In traditional experimental studies of mediation and negotiation, disputant behaviors are carefully controlled and examined, allowing the researcher to make claims about relationships among variables under controlled circumstances. But a limitation of experimental studies in human communication is that the researcher cannot control everything and always confronts the question of confounding variables. This study limited control to the details of the simulation, presenting the opposite problem: there are so many potential variables that it is hard to make claims about relationships among any of them. While our results are interesting, using real mediations would allow us to address many additional variables.

Some of those potential variables include demographic variables, such as sex. Does the sex of the mediator (the same or different from the parties) seem to affect the interaction? Are the patterns of interaction different when the disputants are male–female dyads? Do the patterns differ when the female has a higher status versus when the male has a higher status? Is recognition given differently in same‐sex versus opposite‐sex dyads? What role (if any) do race and ethnicity play in mediation interaction? The practical challenge of studying these variables is that they all operate at the same time, resulting in an infinite number of possibilities for any interaction, as these ten diverse role plays illustrate.

The role of apologies is another important avenue for future research. We observed explicit apologies in five of our ten role plays, two in transformative role plays, and three in resolution‐only role plays. Apologies were offered by both Pat and Chris, suggesting that this may have been related to an individual's communication and temperament rather than the nature of the offense. We observed differences in how the mediators acknowledged apologies (or not), but even this variable did not consistently determine transformative outcomes. Future studies should examine the role of apologies and how mediators and disputants respond to them in naturalistic mediations to tease out these implications.

Another interesting question suggested by our results involves the impact of recognition on transformation. Because of the limitations of discourse analysis and the role‐play method, we often found it difficult to distinguish between expressions of new understanding versus recognition. Parties' statements that they felt better, that they wanted to be friends, or that they were willing to give each other respect and communicate differently suggest some conflict transformation took place. It is impossible to know, however, if the parties believed the relationship had been truly transformed.

Our findings also raise the broader question of what elements mediation agreements should include. Most of the agreements in this study included instrumental arrangements for improving Chris and Pat's communication with each other. A few addressed relational issues, such as “Pat and Chris agree to respect each other.” Some scholars might argue, however, that if transformation really occurred, such respect would be taken for granted and would not need to be stated in the agreement. According to this argument, including relational elements in the agreement might actually indicate that conflict transformation did not occur. Future research on mediation turning points could benefit from following the lead of negotiation researchers who have combined videotapes with interviews to obtain participants' insights into precipitants and departures, as well as to identify statements of recognition and moments of empowerment.

Understanding the dynamics of turning points in mediation can be beneficial to mediators and other conflict management practitioners interested in facilitating productive communication. This study found that mediator discourse was most often characterized by encouraging problem solving, paraphrasing and providing insights, attending to emotion, and focusing on common ground. We further concluded that paraphrasing using nonevaluative language, highlighting disputants' acknowledgments of each other, and attending to disputant emotion were most likely to lead to evidence of conflict transformation. Cooperative discourse of parties in mediation was characterized by acknowledging the other's experience, accepting responsibility, communicating emotion, stating a desire to reconcile, and engaging in problem solving. Mediators who are aware of these moments when they happen and paraphrase them so that disputants do not “miss” them may be more likely to help parties co‐construct turning points that lead to conflict transformation.