The present study examined the combined effects of actor gender and partner gender in relation to the conversational strategies of young adult friends in the United States. Same- and mixed-gender pairs were observed during assigned negotiation and self-disclosure discussions. In addition, speakers’ conversational strategies in these contexts were tested in relation to friends’ ratings of closeness and conflict. Thus, the research explored possible individual and contextual predictors of communication style as well as how communication style might predict friendship qualities.

Most prior studies comparing women and men in conversation have looked at interactions between strangers, which may exaggerate the extent of average gender differences (Leaper and Ayres 2007). Looking at conversations between friends may better reflect the gendered patterns that young undergraduate adults may express in their daily lives. Moreover, the gender composition of the friendship may moderate average differences between women and men in language use (Leaper 2014; Leaper and Ayres 2007). Gender-related variations in language use may partly depend on the conversation activity; negotiation- and disclosure-oriented conversations are among those settings previously associated with average gender differences (Leaper 2014; Leaper and Ayres 2007). Furthermore, communication styles in these contexts were associated with friendship closeness and conflict (Rose et al. 2016; Ruscher et al. 2003).

I review below the background research for the current investigation. First, the model of language and social interaction used in the study is described. Second, the importance of negotiation and self-disclosure in friendships is explained. Third, the study’s hypotheses regarding actor and partner gender effects are presented. Finally, the possible relations of communication style to perceived friendship closeness or conflict are considered. As with research on most psychology topics (Henrich et al. 2010), the relevant studies were conducted primarily in the United States or similar Western industrialized countries.

Affiliation and Assertion in Language and Social Interaction

In the 1970s, psychologists such as Sandra Bem, Jeannie Block, Janet Spence, and others advanced the construct of psychological androgyny, which allowed for the possibility that individuals could express feminine-stereotyped affiliative qualities (e.g., nurturance) as well as masculine-stereotyped assertive qualities (e.g., decisiveness) (see Leaper 2014, for a review). Analogous models of language and social behavior were also proposed (see Leaper 2014, for a review). Rather than frame self-assertion and interpersonal affiliation as opposing forces, researchers recognized the two motives could be coordinated through collaborative behaviors (see Leaper 2014; Selman 1989). Thus, a speaker could assert the self while affirming the other person (e.g., proposing a shared activity; expanding on the other’s topic). Collaborative communication is distinct from either self-emphasizing communication that is high in assertion and relatively low in affiliation (e.g., a command, a task-oriented suggestion) or other-emphasizing communication that is high in affiliation and relatively low in assertion (e.g., agreements, acknowledgements).

In a meta-analytic review of gender variations in adults’ use of language, Leaper and Ayres (2007) examined a variety of different speech acts. In general, language that was primarily self-emphasizing (e.g., task-oriented statements, suggestions) was more likely for men than for women. Language that was more collaborative—affiliative and assertive (e.g., active understanding, giving support)—was more likely in women than in men. And, there was no average difference in speech that was primarily other-emphasizing (e.g., expressing agreement). The moderator analyses further revealed the incidence and the magnitude of gender differences depended on various situational factors. Notably, these included the conversational activity (e.g., deliberation or self-disclosure) and the type of relationship (e.g., friends vs. strangers; same vs. mixed gender). I review these next.

Negotiation and Disclosure as Conversational Contexts

In friendships and other close relationships, the members of the relationship are regularly faced with situations when they need to negotiate differing wishes and viewpoints. In addition, there are times when one person shares an upsetting experience and the listener is challenged to respond in a supportive and constructive manner. Satisfaction with the relationship is affected by how the partners’ coordinate self-assertive and affiliative motives with one another in these contexts (see Leaper 2014, for a review).

Inevitably, two friends’ viewpoints and interests will sometimes conflict when faced with making a decision; negotiation will be necessary to reconcile the difference. In egalitarian relationships such as friendships, individuals seek to attain a balance between pursuing their own wishes versus accommodating to the other’s needs (Jehn and Shah 1997; Leaper 1998). Thus, friendship is considered an important context for the development of interpersonal negotiation (e.g., Pruitt and Rubin 1986; Selman 1989). Studies suggest people may be more apt to care about the other person’s viewpoint and to use cooperative strategies during negotiations with friends (vs. strangers) (Danziger et al. 2017; Kurtzberg and Medvec 1999; Shah and Jehn 1993).

According to various two-dimensional models of social interaction, communication acts vary in degrees of both affiliation and assertion (see Leaper 2014). With respect to interpersonal negotiation, these reflect the extent that a given individual coordinates their personal desires with the perspective of others (see Selman 1989). Collaborative strategies reflect a balance between self-assertion and interpersonal affiliation, such as proposing a solution that jointly affirms the interests of the self and the other. Other strategies may be primarily self-emphasizing, such as when the speaker makes a directive, or they may be primarily other-emphasizing, such as when the person goes along with the other person’s proposal (Selman 1989).

As summarized in prior reviews, average gender differences in negotiation strategies during joint decision-making tasks have been observed (Kray and Babcock 2006; Leaper and Ayres 2007; Walters et al. 1998). In Leaper and Ayres' (2007) meta-analysis, significant differences (with small effect sizes) were indicated across studies, with women more likely to use collaborative and other-oriented speech acts (e.g., indirect suggestions, proposals for joint solutions) and men more likely to use self-emphasizing speech acts (e.g., direct suggestions). These average differences may reflect underlying gendered norms regarding the relative importance of collaborative, other-emphasizing, and self-emphasizing motives (Hall 2011; Rose and Rudolph 2006; Schwartz and Rubel 2005; Strough and Berg 2000).

Self-Disclosure and Listener Support

To establish and maintain intimacy in friendships, individuals share personal experiences and provide support to one another (Tardy and Dindia 2006). Self-disclosure fosters intimacy by allowing friends to gain a deeper understanding of one another (Collins and Miller 1994; Tardy and Dindia 2006). Consistent with the greater emphasis on intimacy and communion in the socialization of girls and women, meta-analyses indicated (with small effect sizes) women were more likely than were men to share intimate information (Dindia 2000) and to expect self-disclosure and intimacy in their friendships (Hall 2011).

In response to someone’s self-disclosure, listeners can offer support to reassure, and possibly to help, the one who is disclosing (Burleson 2003). When communication is experienced as sensitive and helpful, it can improve emotional states, coping strategies, and health. In contrast, insensitive communication can augment emotional pain, undermine coping, and impede one’s health (Feng and MacGeorge 2006; Holmstrom et al. 2005; Jones and Burleson 2003). Thus, not surprisingly, supportive communication is a significant predictor of relationship satisfaction (Burleson and MacGeorge 2002).

Supportive communication occurs when the listener is an active participant in response to a partner’s disclosure (Burleson 1982, 2003). Burleson (1982) differentiated three broad levels of helpfulness in listening responses (also see Leaper et al. 1995). At the highest level, the listener demonstrates active understanding through reflective comments and questions aimed at helping the listener process their experience. Because these responses are proactive, they are considered both assertive and affiliative. Moderately helpful responses occur when the listener recognizes the other person’s feelings, for example, through simple acknowledgements, asking clarification questions, or sharing a similar experience. These responses are considered affiliative but relatively low in self-assertion (e.g., simple acknowledgements) or moderately assertive and affiliative (e.g., sharing a similar experience). At the lowest level, the listener makes distancing or negative responses such as making an irrelevant comment or trivializing the other discloser’s feelings. These responses are low in affiliation.

Average gender differences in levels of listener support have been observed in prior studies. Women were more likely than men were to use highly supportive communication in response to self-disclosures from friends (Basow and Rubenfeld 2003; Leaper et al. 1995; MacGeorge et al. 2004) or strangers (Burleson 1982; Hannah and Murachver 2007; Mickelson et al. 1995). At the same time, it is worth noting there is within-gender variability and overlap between gender groups; that is, many women and men provide comparable levels of listener support (MacGeorge et al. 2004).

Same-Gender and Cross-Gender Friendships in Emerging Adulthood

Research on friendship has mainly focused on same-gender friendships. However, increasing attention has been directed at mixed-gender friendships (see Monsour 2002; Procsal et al. 2015). These relationships increase in frequency during the course of adolescence and emerging adulthood (Monsour 2002). Coeducational college life may be especially conducive to the formation of mixed-gender friendships (Li and Wong 2018). As students together, young women and men commonly interact and share similar lives with equal status as students in dormitories, classrooms, and student groups. Thus, college students are a population for whom mixed-gender friendships are especially common (Mehta and Strough 2009).

Felt pressure for gender conformity may be reduced in mixed-gender friendships (Kluwer et al. 1998; Monsour et al. 1993). Consistent with this supposition, Leaper and Ayres’ (2007) meta-analysis revealed average gender differences in affiliative speech and assertive speech were more likely in same-gender gender than in mixed-gender interactions (in studies of mostly strangers). Also, Carli (1999; Carli & Bukato, 2000) suggested this pattern occurs when the behaviors reflect gender-typed social norms. That is, when two friends share similar gender identities, they may be more apt to enact the norms associated with their gender in-group. Conversely, when interacting with a different-gender friend, these roles may become less salient and partners accommodate to one another’s styles (Pickard and Strough 2003).

The aforementioned reviews of gender composition effects were based primarily on studies of interactions between undergraduate strangers. A few studies have looked at emerging adults’ friendships when comparing behaviors in same-gender and mixed-gender interactions. For example, in one study on negotiation, men were more competitive than women were in same-gender friendships—but not in mixed-gender friendships (Singleton and Vacca 2007). A similar pattern was observed in a study of listener support in friendship pairs during a self-disclosure conversation (Leaper et al. 1995). Women demonstrated more active listening than did men in same-gender friendships whereas there was no difference in mixed-gender friendships. Thus, the present research may help clarify whether the gender composition of a friendship pair moderates the likelihood of any average gender differences in verbal negotiation or listener support.

Communication Style and Friendship Qualities

For the last facet of the study, speakers’ communication styles during the two conversational settings were tested in relation to their friends’ ratings of closeness and conflict in their relationship. How people communicate during negotiation and self-disclosure settings is generally considered an important predictor of relationship satisfaction (Gottman et al. 2014). During negotiation, cooperative and mutually affiliative strategies may foster greater intimacy and less conflict compared to competitive and self-oriented strategies (Le et al. 2018; Pruitt and Rubin 1986; Selman 1989). During self-disclosure settings, expressing personal thoughts and feelings was related to feelings of closeness in undergraduate women’s and men’s friendships (Radmacher and Azmitia 2006). Also, providing active listening to the disclosing speaker provides support that can strengthen closeness in a relationship (Bippus and Rollin 2003), whereas demonstrating little active listening or making negative responses to disclosures may lead to emotional distancing or conflict (Gottman et al. 2014). Furthermore, a recent meta-analysis indicated greater average well-being in interpersonal relationships among those who indicated high communal (i.e., affiliative) motivation that additionally included self-oriented concern (Le et al. 2018).

Thus, speakers’ communication strategies during the negotiation and disclosure conversations were tested as possible correlates of their friends’ ratings of closeness and conflict in their friendships (see Brendgen et al. 2001; Cillessen et al. 2005, for a similar approach). These ratings were collected prior to the conversations, which means they did not reflect immediate reactions to the social interactions in the present study.

Summary and Hypotheses

The present research investigated actor gender and partner gender effects on language used in social interactions between undergraduate friends during separate negotiation- and disclosure-oriented conversation tasks. In addition, uses of communication strategies in these conversations were tested as possible correlates of friends’ ratings of closeness and conflict in the friendship. Undergraduates are a population for whom having same-gender and mixed-gender platonic friends may be most common. Negotiation and self-disclosure are important situations in friendships that call for the coordination of self-assertive and affiliative interpersonal goals. Even though friendships are generally predicated on equality and reciprocity, individuals may express self-assertion and affiliation differently depending on the gender composition of the friendship. Furthermore, variations in these communication styles may be related to how individuals view the quality of their friendship.

Accordingly, three major sets of hypotheses were tested. (a) In the context of negotiating, I hypothesized that average uses of affiliative negotiation strategies would be higher for women than for men (Hypothesis 1a), self-emphasizing negotiation strategies would be higher for men than for women (Hypothesis 1b), and average speaker gender differences in negotiation strategies would be more likely in same-gender than in mixed-gender friendships (Hypothesis 1c). (b) In the context of disclosure, I predicted that average frequencies of self-disclosure will be more likely for women than for men (Hypothesis 2a), average uses of supportive listener responses will be higher for women than for men (Hypothesis 2b), average uses of distancing or negative listener responses will be higher for men than for women (Hypothesis 2c), and average gender differences in self-disclosure and listener support will be more likely in same-gender than in mixed-gender friendship pairs (Hypothesis 2d).

(c) Next, in relation to friendship closeness and conflict, female same-gender pairs, on average, would rate their friendships higher in closeness (Hypothesis 3a) and lower in conflict (Hypothesis 3b) relative to male same-gender pairs or mixed-gender pairs. (d) I further expected that partners’ ratings of friendship closeness would be positively associated with actors’ affiliative strategies (Hypothesis 4a) and negatively associated with actors’ self-emphasizing or distancing strategies (Hypothesis 4b) in either conversational context. (e) Lastly, I hypothesized that partners’ ratings of friendship conflict would be positively related to actors’ self-emphasizing or distancing strategies (Hypothesis 5a) and negatively related to actors’ affiliative strategies (Hypothesis 5b) in their conversational setting.

Method

Participants

The sample comprised 294 U.S. university students in 147 friendship pairs. These included 47 female-female pairs, 47 male-male pairs, and 53 mixed-gender pairs. The participants’ mean age was 18.9 years (SD = 1.0, range = 17–23). Self-identified ethnic backgrounds were White/European (n = 152, 52%), Latin American/Hispanic (n = 57, 19%), Asian/South Asian (n = 51, 17%), African American/Black (n = 16, 5%), mixed or other ethnic groups (n = 14, 5%), or missing (n = 4, 1%). Most participants were raised in two-parent homes with a mother and a father (n = 249, 85%) where both parents were employed outside of the home (mothers: n = 183, 62%; fathers: n = 199, 68%) and had attended at least some college (mothers: n = 215, 73%; fathers: n = 226, 77%).

Procedure

Participants were recruited from the psychology department’s research participant pool of a public U.S. university in 1996 and 1997. If participants were not satisfying the research requirement in a psychology course, they were eligible for a raffle to win a music player. Participants were required to bring a same- or different-gender friend with them whom they had known for at least 3 months (M = 22.5 months, SD = 28.5, range = 3–156). Another criterion for volunteering was that the friendship was platonic (i.e., no sexual or romantic interest in one another). The sexual orientation of participants was not asked.

The friendship pairs met at a research suite at the university. After obtaining the participants’ informed consent, five phases occurred over the course of approximately 2 h (see online supplement Table 1 s). First, the two friends were placed in separate rooms and asked to complete a set of questionnaire measures, which included ratings of their friend (described later) as well as other measures not used in the present analyses (see online supplement Table 2 s for list of all measures). The survey took most participants approximately 45 min to complete. At the end of the survey, they were requested to write down an upsetting event that they would be willing to discuss later with their friend. This was done to prepare participants for this activity occurring in a later phase of the study. In the second phase, each person was individually administered a perspective-perspective interview (not used in the present investigation; described in online supplement Table 1 s).

Joint Decision-Making Task

For the third phase, a female research assistant instructed the friends to talk about whatever they wanted for 10 min. Afterward, the assistant returned to the room and gave the directions for the fourth phase. The participants were assigned a joint decision-making task that required them to negotiate the best solution to a problem. The directions were scripted as follows:

For this session, we are going to ask you to solve two similar problems by discussing them together. First, we would like you to decide together what the 10 most important things a person would need for survival if he/she was going to spend a month in the mountain wilderness. We would like you to decide how to rank these 10 items in order of importance for survival. Afterwards—if you have time—we would like you to consider a second situation. Decide together what the 10 most important things an astronaut would need to survive on the moon for three days. Again, please rank them in order of importance for survival. One of you please write down your decisions on this sheet when you make them. There is a list for the Astronaut survival topic on the other side. Try to stick to the assigned topics as best as you can. I’ll be back in 5 minutes and find out how you have done.

Thus, not only were the participants required to negotiate ten important items, they also needed to decide together their ranked importance. The second decision-making task was assigned in the event that the participants quickly finished the first task; thereby there would comparable amounts of time spent in negotiation across dyads. The mountain wilderness and moon dilemmas (or similar versions) have been commonly used in studies of decision-making and negotiation (Burleson et al. 1984; Hall and Watson 1970; Leaper 1998; Ohtsubo and Masuchi 2004).

Upsetting or Bothersome Experience

After 5 min, the research assistant returned to begin the fifth and final phase for the study. Participants were asked to take turns discussing a personally upsetting or bothersome experience. One of the participants was randomly selected to start. The researcher’s directions were as follows:

Earlier on the questionnaire, it was noted that you each would be asked to talk about an upsetting or bothersome event concerning someone close to you. First, I would like [participant’s name] to talk with [other participant’s name] about your upsetting event. The idea is to have a conversation together about the topic. Also, try to stick to the topic as best you can. I’ll come back in 5 minutes and then ask [other participant’s name] to talk about [her/his] topic.

The friends were left alone for each discussion while their conversations were video-recorded.

After completing the two self-disclosure conversations, the participants were debriefed. They were asked not to discuss the nature of the study with any students who still might be satisfying the psychology research pool requirement.

Measures

Friendship Quality Ratings: Conflict and Closeness

During the preliminary questionnaire phase of the procedure (explained previously), each participant was asked to rate the quality of their friendship with the person with whom they attended the study. They rated their friendship for closeness and conflict based on the Friendship Qualities Scale (Bukowski et al. 1994). Items were rated on 4-point scale from 1 (not at all true of our friendship) to 4 (very true of our friendship). Conflict was measured with two averaged items (“My friend sometimes bugs or annoys me even though I ask him/her not to” and “My friend and I disagree about many things,” r = .99). Closeness was measured with two averaged items (“When I excel at something, my friend is happy for me” and “I think about my friend even when my friend is not around,” r = .99).

Conversational Measures: Coding and Reliability

The friendship pairs’ conversations during the decision-making and the self-disclosure activities were video-recorded. Transcripts of the conversations were created from these recordings. As explained in the following, two coding schemes were devised to classify negotiation strategies in the decision-making task and listener responses in the disclosure task. The unit of analysis was the speech act or thought unit, which corresponds to a sentence or a clause. Inter-coder agreement on the unit of analysis exceeded 90%.

Female research assistants were trained on the verbal coding schemes for several weeks. After training, the research assistants achieved excellent inter-coder reliability based on 20 transcripts that had not been used during training. The overall agreement was excellent (minimum κ = .75) for coding negotiation strategies as well as self-disclosure and listener responses (Fleiss 1981). Each speech act was classified while simultaneously using transcripts and video recordings.

Interpersonal Negotiation Coding Scheme

Verbal negotiation strategies were conceptualized as varying in their degrees with which they address the perspectives of the self and the other (e.g., Selman 1989). The intersection of these two dimensions allows for strategies that are (a) primarily self-assertive (self-emphasizing), (b) primarily affiliative (other-emphasizing), (c) both assertive and affiliative (collaborative), or (d) neither assertive nor affiliative (withdrawing). As described in the following, for the purposes of the present study, we contrasted speakers’ uses of self-emphasizing strategies and affiliative (other-emphasizing or collaborative) strategies. Withdrawal was rare and was not included in the analyses. (A complete listing of the negotiation coding categories appears in the online supplement Table 3 s.)

Self-Emphasizing Strategies

Self-emphasizing strategies primarily function to advance the speaker’s ideas and wishes. Highly self-emphasizing strategies considered in our analyses were directives (a proposal expressed as an order [e.g., “Write down ‘sleeping bag’ on our list”]), direct suggestions (a proposal made as a statement or repeating a prior suggestion [“A sleeping bag would be helpful”]), repetitions (reiterating a prior suggestion), disagreements (overt resistance to the other’s suggestion [e.g., “I don’t think we need that”]), and verbal aggression (hostile comments [e.g., “That’s a ridiculous suggestion”]). Verbal aggression rarely occurred, and therefore it is not included in the present analyses.

Affiliative Strategies

Affiliative speech acts affirm the other speaker’s ideas or suggestions. These vary somewhat depending on the degree they coordinate affiliation with self-assertion. Other-emphasizing strategies downplay the speaker’s own ideas or interests, and they include requests (soliciting other’s suggestion or support [e.g., “What do you think would be good to bring?”]), agreement (sincere agreement with the other’s suggestions [e.g., “Good idea!”]), and obliging (passive acceptance of partner’s suggestion [e.g., “Okay”]). Moderately reflective strategies assert the speaker’s idea while also considering the other person’s viewpoint, and they include justify (defending one’s own opinion or attempting to change the other person’s opinion through persuasion [e.g., “A sleeping bag would keep us warm”]) and indirect suggestion (a proposal is raised as a question or as a tentative statement [“Should we bring a tent?”). Finally, collaborative strategies refer to proposals that support both the self’s and the other’s ideas (e.g., a suggestion that integrates both speaker’s ideas [e.g., “A tent sounds good, and we could get one of those two-person tents”]); however, these infrequently occurred and therefore were not included in the analyses.

There were other speech act categories that were coded but not used in the present analyses (see online supplement Table 3 s for a complete listing). They include factual statements, queries about factual information, orienting statements (reminders to return to task), irrelevant comments, and withdrawal (e.g., silence, hedging, non-responsive comments). They either were not pertinent to the study’s research questions (e.g., statements of factual information) or were rare in occurrence (e.g., withdrawal).

Self-Disclosures and Listener Support Coding Scheme

Self-disclosures were defined as speech acts (i.e., statements) that expressed feelings, opinions, evaluations, or wishes. Based on prior coding schemes (Burleson 1982; Leaper et al. 1995), listener responses were conceptualized as ranging from highly active and supportive to highly distancing and negative. Each of the listener’s speech acts were classified into one code. (A complete listing of the listener coding categories appears in the online supplement Table 4 s.)

Highly active and supportive responses include elaboration (meaningfully adding to the partner’s disclosure [e.g., “I remember when something similar happened to you in high school”]), probing questions (asking partner to consider aspects of their experience [e.g., “Do you think that might be related to the argument you had last week?”]), reframing (helping the partner understand what happened [e.g., “You’ve had similar discussions with her in the past”]), and advice (suggesting ways to solve a problem [e.g., “Perhaps you could tell him how you feel”]).

Moderately supportive responses include showing similarity (expressing solidarity through sharing a similar experience that does not redirect focus to self [e.g., “That also occurred with my parents”]), encouraging (encouraging the partner to continue talking about the disclosure [e.g., “What else happened?”]), simple acknowledgements (brief statements that acknowledge the other’s feelings [e.g., “That sounds upsetting”]), clarification question (a short question to clarify what was said or meant), and backchannel listening responses (interjections that demonstrate listening [e.g., “mm-hum,” “wow”]).

Distancing or negative responses include self-focused disclosures (changing focus of conversation from partner to self with own disclosure), irrelevant response (reference to a topic that is irrelevant to the partner’s disclosure), and negative response (critical or trivializing comments, abrupt change of topic).

Results

Two sets of analyses were conducted. In the first part, the Actor-Partner Interdependence Model was used to test for actor gender and partner gender effects on friends’ speech behavior during the negotiation activity (Hypotheses 1a–c) and the self-disclosure activity (Hypotheses 2a–d). In the next set of analyses testing Hypotheses 3–5, participants’ ratings of closeness and conflict in their friendship were examined in relation to their gender and speech behaviors. Mean frequencies of occurrence for the speech behaviors during (a) the negotiation and (b) the self-disclosure settings are presented in Table 1.

Table 1 Descriptive statistics for frequencies of verbal strategies used in negotiation and self-disclosure contexts
Full size table

Actor Gender x Partner Gender Effects

The Actor-Partner Interdependence Model (APIM; Campbell and Kashy 2002; Kenny et al. 2006) is an analytic approach that allows for simultaneously taking into account the effects of the actor, the partner, and corresponding interaction effects. In this regard, the APIM addresses a limitation of many previous studies of individual variations in dyadic interaction (see West et al. 2008). When two individuals are in a social interaction, their behavior is not independent; therefore, including both persons’ behavior in an analysis violates the assumption of most statistical models (e.g., ANOVA). In hierarchical linear modeling (HLM), the actor and the partner are nested within the dyad as a unit of analysis. Thus, it is possible to test for the effects of the actor, the partner, and the interaction of actor and partner. The HLM analyses were performed using SPSS, which provides standardized beta estimates for each effect as well as t-tests of significance.

The results of the HLM analyses are summarized in Table 2. They are also presented separately in the following for the (a) negotiation activity and the (b) self-disclosure activity. The frequency scores for the speech measures were converted to standardized z-scores before running the HLM analyses. Positive scores reflected higher than average frequencies and negative scores reflected lower than average frequencies. Therefore, any reported means in the text refer to these z-scores. However, for the descriptive statistics presented in Table 1, the frequency scores for negotiation strategies and listener responses were computed.

Table 2 Actor gender and partner gender effects on conversational strategies in negotiation and disclosure conversations
Full size table

Negotiation Activity

Affiliative negotiation strategies were hypothesized to be more likely for women than for men (Hypothesis 1a). Conversely, self-emphasizing strategies were expected to be more likely for men than for women (Hypothesis 1b). Finally, these hypothesized actor gender differences were hypothesized to be more likely in same-gender than in mixed-gender friendships pairs (Hypothesis 1c). As described in the following, significant actor gender effects lent support to the first two hypotheses but not to the third. Also, there were no significant effects with agreement, collaboration, disagreement, or repetition. Verbal negotiation data was missing for one female-female pair and one male-male pair due to technical difficulties during the session. Therefore, these analyses were based on 46 female-female pairs, 46 male-male pairs, and 53 mixed-gender pairs.

Affiliative Strategies

In support of Hypothesis 1a, several affiliative strategies were significantly more likely for women than for men: requests (β = .18, t = 2.96, p = .003) indirect suggestions (β = .13, t = 2.19, p = .029), and justifications (β = .13, t = 2.01, p = .046).

Self-Emphasizing Strategies

As posited in Hypothesis 1b, men were more likely than were women to use direct suggestions, which is one of the self-emphasizing strategies (β = −.12, t = −2.02, p = .045). In addition, an Actor Gender x Partner Gender interaction was found with directives (β = −.12, t = −2.10, p = .037). Follow-up comparisons indicated individuals generally used more directives with same-gender friends (M = .09, SD = 1.04) than with mixed-gender friends (M = −.16, SD = .91; β = .25, t = 2.07, p = .039). The latter finding was not predicted.

Gender Composition as Moderator

Contrary to Hypothesis 1c, partner gender did not moderate any of the actor gender differences.

Self-Disclosure Activity

Self-disclosing statements and supportive listening responses were hypothesized to be more likely for women than for men (Hypotheses 2a and 2b, respectively). Conversely, distancing or negative listening responses were expected to be more likely for men than for women (Hypothesis 2c). Finally, these hypothesized actor gender differences were hypothesized to be more prevalent in same-gender than in mixed-gender friendships pairs (Hypothesis 2d). As summarized in the following, support for these hypotheses was indicated in significant effects with self-disclosure and many of the listening responses. However, there were no significant effects associated with encouragement, probing questions, advice, rationalizations, showing similarity, acknowledgements, or irrelevant comments.

Self-Disclosures

As proposed in Hypothesis 2a, a significant actor gender effect revealed more average self-disclosing statements occurred for women than for men (β = .20, t = 3.49, p = .001).

Supportive Listening Responses

In support of Hypothesis 2b, women provided more elaborations to friends’ disclosures than did men (β = .13, t = 2.35, p = .019). A significant Actor Gender x Partner Gender interaction with elaborations was also seen (β = .15, t = 2.16, p = .032). Follow-up tests revealed that these responses were more likely in same-gender female pairs (M = .32, SD = 1.38) than in other pairs (M = −.15, SD = .71; β = .47, t = 3.85, p < .001), consistent with Hypothesis 2d.

A significant Actor Gender x Partner Gender interaction occurred with backchannel listening responses (β = .16, t = 2.24, p = .026). Follow-up tests indicated backchannel responses were less common among pairs of same-gender male friends (M = −.24, SD = .80) than among other pairs (M = .11, SD = 1.06; β = −.36, t = −2.90, p = .004), supporting Hypothesis 2d.

There was another significant Actor Gender x Partner Gender interaction with clarification questions (β = .17, t = 2.45, p = .016). Follow-up tests indicated these responses were generally more likely among mixed-gender pairs (M = .23, SD = 1.10) than among same-gender pairs (M = −.13, SD = .92; β = .35, t = 2.93, p = .004). This finding was not predicted.

Distancing or Negative Listening Responses

According to Hypothesis 2c, men were expected to be more likely than were women to provide distancing or negative (i.e., unsupportive) responses to friends’ self-disclosures. This pattern was seen with several strategies. On average, men responded with more negative comments than did women (β = −.17, t = −2.80, p = .005). Also, although men tended to use more self-focused statements than did women (see Table 1), the effect was not significant (β = −.12, t = −1.95, p = .053). Furthermore, a significant Actor Gender x Partner Gender interaction was found with irrelevant comments (β = .16, t = 2.47, p = .015). Follow-up tests indicated irrelevant responses were more likely among same-gender male friends (M = .31, SD = 1.38) than among other pairs (M = −.15, SD = .72; β = .46, t = 3.53, p = .001).

Gender Composition as Moderator

Actor gender differences in self-disclosure and listener responses were hypothesized to be more prevalent in same-gender than in mixed-gender friendships pairs (Hypothesis 2d). As described previously, this pattern was indicated for elaborations, backchannel listening responses, and irrelevant comments.

Friendship Closeness and Conflict: Gender and Speech Acts

Preliminary ANOVAs were performed to test for differences across the three gender compositions in the dyads' average ratings of closeness and conflict in their friendship. Pearson correlations tested the hypothesized associations between speakers’ communication acts in the two activities and their friend’s ratings of closeness or conflict in the friendship.

Friendship Group Comparisons

Female friendships pairs were hypothesized to rate their friendships as higher in closeness and lower in conflict than either male pairs (Hypothesis 3a) or mixed pairs (Hypothesis 3b). Support was indicated for both predictions. First, there were group differences based on the gender composition of the friendship pairs in ratings of friendship closeness, F(2,288) = 20.17, p < .001, ηp2 = .12. Tukey comparison tests (p < .05) revealed that female pairs rated their friendships as significantly closer (M = 4.79, SD = 1.14) than did the male pairs (M = 3.53, SD = 1.43) or the mixed pairs (M = 4.11, SD = 1.44).

In addition, there was a significant gender composition effect on ratings of friendship conflict, F(2,288) = 6.76, p = .001, ηp2 = .05. Tukey comparisons (p < .05) indicated that female pairs rated their friendships as significantly lower on conflict (M = 1.36, SD = 1.30) than did the male pairs (M = 2.15, SD = 1.57) or the mixed pairs (M = 1.91, SD = 1.61). Also, the male pairs rated their friendships as significantly higher on conflict than did the others.

Correlations Between Friendship Ratings and Speech Acts

Spearman correlations were performed to test the associations between the actors’ speech behaviors in the two conversational settings and their partners’ ratings of friendship conflict and closeness. Given the large number of correlations performed, a criterion of p < .01 was used for determining statistical significance for the correlations. Conflict and closeness ratings were missing for two female participants in a same-gender pair and one male participant in a mixed-gender pair.

Partners’ ratings of friendship closeness were expected to predict higher rates of affiliative strategies and lower rates of self-emphasizing (Hypothesis 4a) or distancing strategies (Hypothesis 4b) in either setting. Conversely, ratings of conflict were hypothesized to predict higher rates of self-emphasizing or distancing speech acts (Hypothesis 5a) and lower rates of affiliative speech acts (Hypothesis 5b) in either conversational activity. In the event of any significant correlations, follow-up tests were performed to explore if speaker gender or the pair’s gender composition moderated any of the associations.

No significant associations between actors’ speech behaviors and partners’ ratings of closeness were indicated; therefore, Hypotheses 4a and 4b were not affirmed. However, there were several significant correlations between actors’ speech behaviors and partners’ conflict ratings consistent with Hypotheses 5a and 5b. These results are summarized in the following regarding speech behaviors either during the negotiation activity or for the disclosure activity.

Negotiation Activity

Partners’ ratings of conflict were significantly associated with two speech acts during the negotiation activity that supported Hypotheses 5a and 5b, respectively. First, conflict ratings were positively related to uses of disagreements, r(287) = .360, p < .001. Also, conflict ratings were negatively related to actors’ uses of agreements, r(287) = −.170, p = .004. Neither speaker gender nor gender composition moderated these correlations.

Disclosure Activity

In support of Hypothesis 5a, friendship conflict ratings were associated with uses of negative comments, r(291) = .182, p < .001. The gender composition of the pair moderated this correlation. The association was significant among mixed pairs, r (105) = .355, p < .001, but not among female pairs, r(92) = −.118, p = .262, or male pairs, r(94) = .105, p = .316. The correlations for mixed and female pairs were significantly different, Zdifference = −3.38, p < .001; but the difference was not significant between mixed and male pairs, Zdifference = −.183, p = .065. Negative comments occurred infrequently, and the result therefore should be viewed cautiously.

The next set of results lent support to Hypothesis 5b. Partners’ ratings of friendship conflict were negatively associated with self-disclosures, r(291) = −.197, p = .001. Also, conflict ratings were positively correlated with listeners’ encouraging statements, r(291) = .174, p = .003. In addition, a negative correlation was indicated between friendship conflict ratings and listeners’ back-channel responses, r(291) = −.220, p < .001. However, speaker gender moderated this association, Zdifference = 2.55, p = .010. The correlation was more likely when men were the speakers, r(146) = −.329, p < .001, than when women were the speakers, r(145) = −.039, p = .643. Thus, men’s uses of backchannel responses were related to lower perceived conflict in the friendship whereas women’s use of these responses was unrelated to perceived conflict.

Discussion

The study investigated gender-related variations in the conversational strategies of young adult college friends during assigned decision-making and self-disclosure tasks. Although the coding schemes were somewhat different for analyzing communication patterns in the two activities, they were similar in considering the relative emphases on affiliation and self-assertion (or their coordinator via collaboration) in the speaker’s discourse. Average gender differences were seen in how affiliation and self-assertion were expressed in the two conversational activities that were consistent with patterns indicated in prior studies (see Leaper 2014; Leaper and Ayres 2007). Also, consistent with many studies of social behavior (see Hyde 2005), the effect sizes for these differences were small in magnitude.

Several main findings can be summarized. First, during the negotiation activity, women were more likely than were men to use affiliative speech strategies whereas men were more likely to use self-emphasizing strategies. Contrary to expectation, partner gender did not moderate any of these actor gender differences. Second, during the disclosure conversation, women were more likely than men were to make more self-disclosing statements and to offer supportive listening responses whereas men were more likely to demonstrate distancing or negative listening responses. As predicted, some forms of listening support were most common among same-gender female friends and least common among same-gender male friends. Third, prior to the conversational activities, female pairs rated their friendships as higher in closeness and lower in conflict than did either male pairs or mixed pairs. Finally, ratings of conflict (but not closeness) predicted higher rates of distancing or negative speech and lower rates of affiliative speech in each conversational activity.

During the negotiation task, there were average gender differences in the uses of particular kinds of verbal strategies that confirmed my hypotheses. Women were more likely than men to use strategies that were relatively high in affiliation (including collaborative strategies high in both affiliation and assertion). In particular, this pattern was seen in the relative incidences of indirect suggestions, justifications of prior suggestions, and requests for the other person’s ideas or support. Men, in contrast, were more likely to use direct suggestions, which are self-emphasizing speech acts high in self-assertion but relatively low in affiliation. Thus, women as well as men sought to assert their views; however, women were somewhat more likely to do so while affirming the perspective of their friend. For example, a direct suggestion would be “We need to bring water” whereas an indirect suggestion would be “Do you think we should bring water?” In the former, the speaker is stating a course of action in a somewhat declarative and forceful manner. In the latter, the speaker is being more tentative which may make it easier for the other person to propose a counter-suggestion.

The observed average gender difference in justifications further highlights how women tended to be more collaborative during negotiations. By offering justifications for their suggestions (e.g., “It might rain and a tent would keep us dry”), speakers are taking into account how the other person might consider their suggestion (Selman 1989). Thus, justifications are both affiliative (acknowledging another’s perspective) and assertive (arguing for one’s idea).

As I noted, there was a higher average incidence of direct suggestions for men than for women. Although they are self-emphasizing, direct suggestions reflect relatively more consideration of the other person’s perspective than do directives (i.e., commands). With directives, the speaker more forcefully states a course of action (“Put down ‘water’ on our list”). No gender differences were observed in the usage of directives, and this strategy was generally infrequent (see Table 1). Given the norms of equality and reciprocity that are common in most friendships (Hall 2012), it is not surprising that directives were uncommon. Furthermore, because conversations were being video-recorded, participants may have been especially mindful of acting cooperative with one another.

Average gender differences additionally were indicated in speech acts during the self-disclosure conversation in expected ways. On average, women made more self-disclosing statements than did men. This finding (which had a small effect size) is consistent with trends in prior reports (Dindia 2000). Also supporting our hypotheses, women were more likely than were men to provide supportive listening responses.

Also as predicted, men were more likely than were women to use self-emphasizing or distancing responses to self-disclosure. On average, men used more negative comments. Furthermore, in same-gender pairs, men were more likely than were women to make irrelevant comments after their friend’s disclosures. Irrelevant comments suggest that the speaker is not directing attention to the partner’s disclosure. Moreover, these kinds of responses may signal some men’s discomfort in dealing with emotionally personal matters. Actively responding to a friend’s personal disclosure may challenge some men’s traditional masculinity norms regarding the expression of vulnerable feelings (Pollastri et al. 2018).

Besides testing hypothesized average gender differences in actors’ speech strategies, my study was designed to consider the gender composition of the friendship pair as a possible moderator. Carli and Bukatko (2000) proposed different explanations for gender differences in social behavior that occur primarily in either mixed-gender or same-gender interactions. The authors suggested that gender differences primarily seen in mixed-gender pairs may reflect male dominance and underlying gender-based status differences (i.e., a gender-as-status explanation). This kind of pattern might be especially likely during interactions between strangers (Leaper and Ayres 2007; Wood and Karten 1986). Conversely, gender differences seen primarily in same-gender interactions may reflect gender-typed differences in social norms (Carli and Bukatko 2000). With same-gender partners, individuals may share similar expectations regarding social behavior due to prior gender socialization (i.e., a gender-as-norms explanation). This pattern may be especially likely in friendships. In support, Hall’s (2011) meta-analysis testing gender differences in friendship expectations indicated higher expectations among women for communion (e.g., intimacy) and higher concerns with agency (e.g., status) among men. The observed gender composition effects in the present study were most compatible with the gender-as-norms explanation. As hypothesized, women in same-gender pairs tended to use more elaborations and backchannel listening responses than did men in same-gender pairs. However, these responses were similar for women and men in mixed-gender friendships.

Some prior investigators have speculated that gender-based status and power may present a challenge in mixed-gender friendships (Baumgarte 2002; O'Meara 1989). Other researchers, however, have not seen equality as a serious problem in most mixed-gender friendships (Messman et al. 2000; Monsour et al. 1993). Some evidence suggests gender differences may be attenuated in mixed-gender friendships (Baumgarte and Nelson 2009).

One unexpected result in the present study was that clarification questions during self-disclosure occurred more in mixed-gender than in same-gender friendship pairs. Perhaps the mixed-gender friends were less familiar with one another’s issues and required more detail about the disclosure (Iannone et al. 2017). However, this potential explanation requires testing in future research.

The last facet of the present study was to explore the correlates of participants’ ratings of closeness and conflict in their friendship. Overall, higher scores in closeness and lower ratings of conflict were reported in women’s same-gender friendship pairs than in men’s same-gender pairs or in mixed-gender pairs. Similar patterns have been observed in prior studies (Rose and Rudolph 2006). Of particular interest was whether partners’ ratings of conflict or closeness in the friendship were related to actor’s communication strategies in the two conversational contexts. The ratings of conflict and closeness were collected prior to the assigned conversational activities. That is, they do not reflect the participants’ reactions to the conversations examined in the present study. This timing suggests that any observed correlations between participants’ ratings and their friend’s speech may reflect generalizable patterns that affect the relationship.

Although there were no significant associations with closeness ratings, there were notable results with conflict ratings. First, this was seen in the negotiation context. Uses of agreements and disagreements during decision-making were respectively related to lower or higher ratings of friendship conflict. The result is consistent with prior research suggesting that the relative amounts of disagreement and agreement in an interaction may index conflict (McLachlan 1991). In addition, some communication strategies in the disclosure conversation were related to the partner’s conflict ratings. Lower conflict ratings were associated with higher amounts of self-disclosure. This finding is compatible with prior reports highlighting the positive impact of self-disclosure on relationship qualities in friendships (Morry 2005; Radmacher and Azmitia 2006). Also, for men (but not for women), use of backchannel listening responses was negatively correlated with the partner’s ratings of conflict. In prior studies, men were observed to be less likely than were women to demonstrate active listening to partners’ disclosures (e.g., Leaper et al. 1995). If many individuals experience this pattern with their male friends, then perhaps they are more sensitive whether their male (vs. female) friends are showing signs of listening. In turn, this may affect how they appraise the degree of harmony in their friendship.

Finally, the use of negative comments during the self-disclosure conversation was associated with higher ratings of friendship conflict among mixed-gender pairs but not among same-gender pairs. Some investigators have suggested that mixed-gender friendships pose some challenges for heterosexual adults that are less likely than in their same-gender friendships (Baumgarte 2002; O'Meara 1989). For example, there may be ambiguities in the sexual-romantic intentions of one or both partners (O'Meara 1989). Also, there may be differences in norms for the expression of social behaviors based on prior gender socialization (Carli and Bukatko 2000). This may include the meaning that certain negative comments might have for some women and men (Keltner et al. 2001). As a consequence, negative comments during self-disclosure may be more likely to undermine satisfaction in mixed-gender than in same-gender friendships. Given the infrequent occurrence of negative comments (see Table 1), however, this finding and interpretation should be viewed cautiously.

In sum, my findings regarding the communication correlates of friends’ conflict ratings suggest that individuals’ patterns of communication may affect the quality of their relationships. Notably, meaningful correlations occurred between speakers’ communication styles during two conversation settings and their partners’ ratings of conflict made before these conversations. That is, the conflict ratings were not reactions to the observed conversations. Researchers have similarly observed that communication in a lab setting was correlated with preexisting relationship qualities in friendships (Brendgen et al. 2001) and marriages (Gottman et al. 2014). Moreover, as seen in the present study, communication patterns tend to predict relationship satisfaction similarly for women and men even when there may be average gender differences in the uses of particular strategies.

Limitations and Future Research Directions

In this section, some limitations of the present study are noted and directions for future research are proposed. First, our sample comprised college youth from mostly middle-class families in the United States. This is a population for whom the formation of mixed-gender friendships may be especially likely (Li and Wong 2018). Accordingly, it would be helpful to examine friends from other backgrounds (see Henrich et al. 2010, for a discussion). This would include observing friendships across different cultures, ages, races, ethnicities, gender identities, sexual orientations, political ideologies, and socioeconomic backgrounds (Galupo et al. 2014; Mehta and Strough 2009; Rose and Hospital 2017). These various sociocultural backgrounds may partly shape norms for communication during social interactions in friendships.

A second recommendation is to utilize a more comprehensive measure of friendship qualities to assess possible connections with friends’ communication style. Short measures of friendship qualities were used in the present study due to the extensive number of measures and tasks that participants were asked to complete. Perhaps it is all the more remarkable that meaningful associations were observed between conflict ratings and friends’ speech.

A third proposal is to observe longer conversations. In the present study, the conversations in the negotiation and disclosure settings were each 5 min (5 min per partner in the disclosure context). An especially ambitious method would involve recording people’s daily conversations over several days (e.g.,  Pennebaker and Chung 2014). This approach would allow researchers to consider how individuals negotiate and support one another in naturalistic settings. For example, research suggests that self-emphasizing negotiation strategies (e.g., coercion) may be more likely in real-life conflicts than in the hypothetical task used in the present study (Laursen et al. 2001).

Besides examining longer conversations, a fourth recommendation is to consider a wider range of conversational settings. One strength of the present study was to observe conversations regarding two different topics (negotiation and disclosure). Nonetheless, different gender-related patterns in conversational strategies may emerge in other settings. Also, one study suggests that gender-related variations in negotiation may depend on the particular topic (Bear and Babcock 2012): Average gender differences in negotiation behavior were seen during discussion of a masculine-stereotyped topic but not during discussion of a feminine-stereotyped topic.

One last recommendation is to explore some of the possible reasons for average gender differences in friends’ communication styles. This could include examining if individuals’ gender ideologies or relationship goals might be related to communication strategies in friendships. For example, recent work on traditional masculinity suggests that some men’s concerns with dominance may restrict their emotional expression (e.g., Pollastri et al. 2018). Also, studies have found that communal goals may partly mediate some average gender differences in communication (e.g., Strough and Berg 2000). A better understanding of factors that predict variations in effective communication and their origins can be used to inform efforts aimed at improving close relationships.

Practice Implications

Although the present average differences are small in magnitude, there are some reliably documented findings in the research literature that were replicated here. Most notably, women were more likely than were men to use affiliative verbal strategies in the negotiation and disclosure settings. In contrast, men were more likely than were women to use self-emphasizing or distancing strategies. Some of the observed gender differences in speech behavior were mitigated when individuals interacted in mixed-gender friendship pairs. Also, regardless of gender, greater friendship satisfaction (i.e., lower partner ratings of conflict) was implicated when women or men used more affiliative speech and less distancing speech. One implication of these effects is that mixed-gender friendships may be a helpful means by which women and men can transcend the limitations of traditional gender roles and learn to relate to one another as equals (Louis et al. 2013; Monsour 2002). Moreover, as implied in the present results, higher uses of affiliative speech and lower uses of distancing speech were associated with higher friendship satisfaction (i.e., lower conflict) in women and men. By extension, therefore, it may be helpful to encourage mixed-gender friendships in childhood and adolescence as a means to foster greater gender equality (Fabes et al. 2018).

Conclusions

The present research adds to a growing research literature on gender, language, and communication that began to emerge in the 1970s (e.g., Thorne and Henley 1975; see Leaper 2014, for a review). Small yet reliable patterns of gender difference were seen in the two conversational settings, which were consistent with prior studies based primarily on samples of middle-class, Western adults (see Leaper 2014). Whereas the prior discussion mainly considered the observed differences, it is important to reiterate that the effect sizes were small: That is, there was a large degree of similarity and overlap among women and men in the sample. This point is commonly ignored in many popular self-help books that present exaggerated views of women and men as fundamentally different (see MacGeorge et al. 2004, for discussion). These beliefs can help to perpetuate the status quo and power imbalances in gender relations (Glick and Fiske 1996). Moreover, as suggested in the present research and other studies, average gender differences in communication style do not appear to benefit men or women in their close relationships (Gottman et al. 2014; Le et al. 2018).