Anxiety/Uncertainty Management and Communication Accommodation in Women’s Brief Dyadic Conversations With a Stranger: An Idiodynamic Approach

Abstract

This study uses a combination of anxiety/uncertainty management theory and communication accommodation theory perspective to examine differences in the ways women converse with men or with other women. The study uses an innovative approach, the idiodynamic method, to gather detailed data on a per-second timescale. Participants (n = 24) were randomly assigned to one of the two types of dyads: female–female or male–female. They then engaged in a videotaped conversation lasting approximately 2 to 5 min. Immediately afterward, participants watched the video of their conversation in separate rooms, during which each provided continuous, dynamic ratings of their anxiety level, and that of their partner, throughout the conversation. Participants were interviewed about the reasons for changes in their ratings of both themselves and their conversational partner. Following a grounded theory approach to analysis, six themes emerged from the interviews: awareness of the camera and researchers, comparison of self and partners, self-judgment, worry about other’s judgment, disinterest, and reaction to miscommunication. The data show that communication accommodation was done differently in the female-only versus female–male pairs, which might reflect processes involving uncertainty, group identification, and continuously negotiated meaning. There is value in using methods that investigate communication processes in real time because doing so allows instances of proposed theoretical differences between genders to emerge in actual conversation between persons.

Within dyads, communication behavior can take unexpected twists and turns as it unfolds in real time. The uncertainty inherent in dyadic conversation can be influenced in both obvious and subtle ways by the gender of the conversational partners (Athenstaedt, Haas, & Schwab, 2004; Carli & Bukatko, 2000); mixed-gender dyads may show different communication behavior than same-gender dyads (Hannah & Murachver, 1999). There is a considerable amount of research literature on gender differences in communication, for example, showing that females talk less when paired with a male than they do when paired with another female (James & Drakich, 1993) and that mixed-gender dyads act in stereotype-confirming ways more often than same-gender dyads (Athenstaedt et al., 2004). It has been argued that males and females learn and develop different sociolinguistic subcultures that affect ways of both communicating and interpreting communication (Maltz & Borker, 1982; Mulac, Bradac, & Gibbons, 2001; Wood, 2002). However, the “different cultures” view has been challenged strongly by authors who argue that similarities between women and men outweigh differences (Hyde, 2005; MacGeorge, Graves, Feng, Gillihan, & Burleson, 2004). To further the discussion of gender differences and similarities, we suggest that there is a need to examine the dynamics of real people interacting in real time. The methods based on summative evaluations and imagined conversations used in prior research simply do not capture what is happening in the action and reaction of conversation.

The focus of the present study is on an innovative methodology that allows us to examine and describe emotional reactions to communication processes on a per-second basis, with a particular emphasis on women communicating in same-gender dyads versus mixed-gender dyads. The method we use captures a brief conversation (less than 5 min) on video. Using a structured stimulated recall protocol and specially designed software, interactants immediately review the video of their conversation and rate themselves on an affective state. In the present case, anxiety is rated once per second, producing a dense set of data reflecting continuous changes in affective reactions during communication. In the present study, participants also rate their conversation partner by reviewing the video a second time and rating the anxiety they attribute to their conversation partner. This process allows researchers to identify convergence and divergence between interactants. Prior research has not linked the details of a conversation to ongoing changes in anxiety reactions in real time. To conceptualize patterns of affective reactions, two cross-cultural frameworks, anxiety/uncertainty management (AUM) theory (Gudykunst, 1998) and communication accommodation theory (CAT; Gallois & Giles, 2015), are being used to interpret gender differences.

AUM Theory

AUM theory was developed by Gudykunst (1988, 1993, 1995, 1998) as an extension of Berger and Calabrese’s (1975) uncertainty reduction theory, to explain both intragroup (within cultures) and intergroup (across cultures) communication (Gudykunst & Nishida, 2001; Stephan, Stephan, & Gudykunst, 1999). Gudykunst (1988, 1995) argued that individuals typically experience some degree of both anxiety and uncertainty while interacting with members of another group. During intergroup interactions, people experience more uncertainty than would occur when interacting with members of their own group (Gudykunst, 1995; Jaasma, 2002) because cross-cultural encounters tend to lack the predictability of intragroup encounters (Samochowiec & Florack, 2010). On one hand, some people might react negatively to uncertainty, leading to anxiety and avoidance of interaction with members of another group (Samochowiec & Florack, 2010). On the other hand, some people find uncertainty to be exciting and a reason to approach persons from different groups. Given that both anxiety and uncertainty are probable in cross-cultural dialogue, effective communication depends on managing them both simultaneously (Gudykunst & Nishida, 2001; Samochowiec & Florack, 2010; Stephan et al., 1999).

Managing anxiety is an especially relevant issue for dyadic communication because there are only two people involved. The back-and-forth of dyadic conversation means that if anxiety arises for one speaker, there are no other speakers available to take up the reins while one regains her or his composure or train of thought. Previous research has found that compared to males, females tend to experience higher levels of both foreign language anxiety (Dewaele & MacIntyre, 2014; Mahmood & Iqbal, 2010) and native language communication apprehension (Frantz, Marlow, & Wathen, 2005). If young girls often are encouraged by caregivers to express emotions (McLean & Anderson, 2009), perhaps it is not surprising that Aly and Islam (2005) found that females tend to be self-conscious and show more concern with how they are perceived by other people (Frantz et al., 2005). The present study will examine in detail female participants engaging in conversation with a stranger in both intragroup (with another female) and intergroup (with a male) dyads.

Theories of intercultural communication, such as AUM theory, may provide insight into differences in communication processes between genders that occur in dyadic communication; this is one reason to consider the so-called different cultures approach. Adaptation of AUM theory to the study of communication across genders allows us to retain its core elements. Based on AUM theory, we propose that women communicating in a dyad with men can experience more uncertainty and anxiety than women paired with other women, analogous to the increased levels of uncertainty and anxiety when communicating across cultures. Effective communication, defined by shared meaning (Gudykunst & Nishida, 2001; Stephan et al., 1999), would be less likely when women are talking to men than when talking to other women, especially if the individuals are strangers. According to AUM theory (Gudykunst & Nishida, 2001), there are hypothetical, optimal maximum and minimum levels of both uncertainty and anxiety (neither too high nor too low) for a communication interaction. While too much uncertainty can reduce the confidence needed to predict the attitudes, behaviors, values, and feelings of one’s conversational partner, having too much certainty can lead to overconfidence and mistaken assumptions. High levels of anxiety also can lead to reliance on simplistic information processing and to potentially stereotyping strangers. Low levels of anxiety can indicate comfort and familiarity, but in other cases might suggest little interest and motivation to continue the communication encounter (Gudykunst & Nishida, 2001). AUM theory proposes that the manageable range of both uncertainty and anxiety differs from one individual to the next, varies with experience, and should be considered on an individual basis.

Unfortunately, much of the past research on the AUM theory has been done with large groups of participants through the administration of questionnaires in which respondents answer questions about a recalled or hypothetical conversation (Duronto, Nishida, & Nakayama, 2005; Gudykunst & Nishida, 2001; Jaasma, 2002). A dialectic approach to uncertainty and anxiety might examine fluctuations in affect not only from one interaction to another but also as an interaction unfolds (Gudykunst & Nishida, 2001). Typically, anxiety about communicating with a stranger might be expected to decrease gradually as time goes by. However, anxiety levels can increase suddenly at any moment if a misunderstanding occurs (Stephan et al., 1999) or if one partner somehow offends, embarrasses, or annoys the other—what Gudykunst (1995) called a “catastrophe point” in conversation. Previous AUM research has not employed methods that allow for a description of changes in anxiety that can be linked closely to both communication events and affective reactions as they unfold together in time. It is especially important to take a moment-by-moment view of gender differences in communication as they play out in real action and reaction sequences, as speakers either accommodate each other or not. Such a research approach provides data that connect how people actually act and react verbally with anxiety reactions and an introspective rationale. In contrast, prior studies such as MacGeorge et al. (2004) used imagined conversations, recalled conversations from the past month, and multiple-choice responses to pre-determined scenarios. In contrast, the present study directly examines the affective/communication processes in which gender differences might play out as people both act and react to their conversation partner.

CAT

CAT, first developed in the 1970s under the banner of Speech Accommodation Theory (Giles, 1973), provides a robust account of ways in which communication patterns change depending on the receiver or addressee (Giles, Coupland, & Coupland, 1991). CAT holds that accommodation processes are fundamental to interaction, serving to manage social distance between participants and coherence in the conversation (Gallois & Giles, 2015). A recent meta-analysis found consistent moderate-to-strong positive effects for accommodation efforts on variables such as quality of communication, relational solidarity, compliance, and trust, along with a wide range of correlates and contexts in which CAT has been applied (Soliz & Giles, 2014). In some situations, speakers will be motivated toward greater accommodation and convergence to the speech characteristics (such as speech rate, pitch, volume, word choice, syntax, and so on) of their interlocutor for a number of reasons, including making their message more comprehensible and/or a desire to affiliate with their conversation partner. However, there are occasions when the tendency will be toward non-accommodation, either by maintaining default ways of speaking or diverging from the other person to emphasize group-level differences, such as reinforcing one’s ethnic identity (Bourhis, 1984; Bourhis & Giles, 1977). Using strategies of convergence, divergence, and maintenance, speakers modulate social distances and the comprehensibility of their message. Although some examples of non-accommodation might be considered as striving toward a form of mutual complementarity, as when women raise pitch and men lower pitch to emphasize their respective gendered identities, more often non-accommodation signals a desire to emphasize differences or distances between groups (Dragojevic, Gasiorek, & Giles, 2015). The consequences of non-accommodation most often are described as negative, including increased social distance, reduced satisfaction with the conversation, and negative evaluations of the interlocutor (Soliz & Giles, 2014).

Within a conversation, accommodation can take many forms. Symmetrical accommodation occurs when both parties make adjustments toward to the speech patterns of the other; asymmetrical occurs when one of the parties makes more of the adjustments (Gallois & Giles, 1998). Some studies suggest that women are more likely to converge than men, but generalizations can be difficult to make. For example, Mulac, Studley, Wiemann, and Bradac (1987) reported that females’ gaze patterns tended to converge toward their male interlocutor’s pattern more than males converged toward females. Bilous and Krauss (1988) described both convergence and divergence, as women converged toward men’s utterance length, interruptions, and pauses, but diverged on backchannels and laughter. Speakers who wish to affiliate will tend to converge more than speakers who wish to emphasize group differences, putting goals related to individuation versus social identity into conflict (Gallois & Giles, 2015). Motivated tendencies toward or away from accommodations may take into account a number of factors, including the context of the conversation, the history of intergroup relations, and the idiosyncrasies of the participants and their specific relationship with each other. Dragojevic et al. (2015, p. 2-3) summarize the theory as follows:

Essentially, CAT proposes that speakers come to interactions with an initial orientation, which is informed by past interpersonal and intergroup experiences, as well as the prevailing sociohistorical context. In interaction, speakers adjust their communicative behavior based on evaluations of their fellow interactants’ communicative characteristics, as well as their own desire to establish and maintain a positive personal and social identity. Each speaker evaluates and makes attributions about the interaction, as well as about the other speaker, on the basis of their perceptions of that other speaker’s, as well as their own, communication. These attributions and evaluations then affect the quality and nature of both the present interaction between these speakers and speakers’ intent to engage in future interaction with each other.

Perhaps most relevant to the present study is the notion that speakers begin interactions with an initial orientation toward (or away from) accommodating their interlocutor.

Combining AUM with CAT, taken at a dynamic, individual level of analysis, can address in detail the ways in which genders enact their differences in communication. Of particular interest are the ways in which specific speech accommodations interact with the emotions and AUM of both speakers, on the fly, to create the psychological/emotional context of the communication process. That is, we are explicitly avoiding a “one-size-fits-all” approach to gender differences in communication in favor of a highly localized, context-dependent, and dynamic approach to understanding the movement toward or away from accommodations as conversations unfold. From this perspective, we propose that speakers may move among strategies reflecting convergence, divergence, and/or maintenance, in reaction to conversation events, even within the span of a few moments in time. Such an approach is novel in the CAT and AUM literatures. Fluctuating levels of anxiety and uncertainties provide a meaningful way of tapping into the coordination of ongoing cognitive, affective, and communication processes. In pursuit of this goal, we will employ a relatively new mixed method that allows for a process-oriented approach to studying communication by tracking moment-by-moment changes in anxiety during dyadic conversation (generating quantitative data), along with the participants’ rationale for those changes (generating qualitative data). The method is focused on individual-level analysis of changes or “idiodynamic.”

Idiodynamic Method

The idiodynamic method was developed as a way to capture the interacting processes of complex, dynamic systems (MacIntyre, 2012). Dynamic methods are necessarily different from the conventional approach to research reflected in the studies reviewed above (for an overview of dynamic systems applied to interpersonal communication, see Fogel, 2006). In complex dynamic systems theory (CDST), the focus is on variability within stability and describing the processes underlying change. Multiple interacting processes assemble to create emergent states that are meaningful, some of which may last a long time as in Diamond’s (2007) 10-year study of changes in female sexuality, and other states coalesce and then dissipate quickly as in Gregersen, MacIntyre, and Meza’s (2014) study of anxiety reactions to public speaking in a second language. In the case of anxiety reactions, physical, cognitive, emotional, motivational, social, and other processes are co-regulated; their joint operation can and do coalesce (at times) to create anxiety reactions when talking to a stranger. From a dynamic perspective, we can ask about the antecedent conditions that contribute to a state such as anxiety and we can examine where it goes next—does anxiety build upon itself, cascading into a vicious cycle, or does another process intervene to alter the trajectory of the anxiety reaction? It is easy to imagine an interlocutor who is perceived as unpleasant or judgemental exacerbating anxiety at a critical point in conversation, but equally easy to contemplate a friendly smile or reassuring words alleviating anxiety in the moment. Systems continue to interact and converge in meaningful ways as a conversation proceeds in time as part of a continuous, ever-changing stream of interpersonal events. The salient point for the present study is that conventional methods previously used to study AUM and CAT do not capture the ongoing dynamics of conversation in real time. The idiodynamic method is designed to do so by reporting results on an individual level, in large part because the processes most relevant to anxiety reactions can differ from one person to the next. Van Dijk, Verspoor, & Lowie (2011, p. 62) proposed three key criteria for research methods to address research questions from a dynamic perspective:

. . . if we really want to know how an individual (or group) develops over time we need data that is dense (i.e. collected at many regular measurement points), longitudinal (i.e. collected over a longer period of time), and individual (i.e. for one person at a time and not averaged out).

Most studies currently available in the literature reviewed above do not meet these criteria. Test–retest designs do not generate the density of data required to focus on the variability that defines dynamics; to understand change it seems fairly sensible to watch it closely as it unfolds instead of inferring processes in-between two measurement times. Research that assesses change using experimental methods necessarily averages over multiple persons. The focus of experimental design typically is on differences in the means of dependent variables, with variance due to individuals assigned to the error term in a between-group analysis of variance (ANOVA). Individual differences (correlational) research also partitions variance into trend lines and error around those lines, but the interpretation of the correlation or regression focuses on patterns at the group level. Data that best address the research questions from a CDST perspective are dense, longitudinal, and individual and therefore must be presented and interpreted in ways that differ from conventional research.

The idiodynamic method (Gregersen et al., 2014; MacIntyre, 2012; MacIntyre & Legatto, 2011) allows for multiple streams of data to emerge from a single communication event. The method is partly inspired by Gordon Allport’s (1962) famous call for more idiographic research, including an analysis of whether common patterns and generalizations apply to individuals. There is now good reason to believe, as Allport suggested, that nomothetic patterns rarely apply to individuals (Molenaar, 2004; Molenaar & Campbell, 2009). Allport’s contemporary Saul Rosenzweig coined the term “idiodynamic” to reflect a focus on the unique, individual-oriented, dynamic organization of events through time. Rosenzweig (1986) used the term idiodynamic to refer to the consistent patterns of an individual’s personality, but here it will be used to refer to changes experienced by a small number of individuals over a short period of time (minutes).

The method essentially consists of four steps (MacIntyre & Legatto, 2011):

1. A communication task occurs and is videotaped for immediate playback.

2. A research assistant uses software written for the study to replay the video and the participant completes a second-by-second rating of a given variable (in this case anxiety). The software produces a graph of these ratings that is printed immediately after it is completed by the participant. Ratings are captured at the rate of one-per-second (for example, a 3-min conversation produces 180 anxiety ratings for each participant).

3. The participants review the graph with a researcher and provide their reasoning for changes in ratings, using both the graph and video as a reference. The research assistant guides this process by the recommendations for stimulated recall (Gass & Mackey, 2000).

4. The entire session is transcribed.

For the present study, an additional step was implemented. After reviewing their own graph, each participant watched the recording of the conversation a second time to produce an additional set of ratings to capture data on their partner’s perceived anxiety level during the conversation, repeating steps 2 and 3 with their partner as the target of the anxiety ratings. Using this approach, we can gather information about attributions made about the interlocutor and how the anxiety of one person might connect to the anxiety that same person perceives in their interlocutor.

The Present Study

In the present study, the idiodynamic method is used to describe fluctuations in anxiety. The research assistants conducting the testing are both women, allowing for greater gender consistency across the conversations. The focus in data analysis is on women engaging in dyadic conversation with a stranger, either another woman or a man. Each participant also rates their partner’s anxiety level, allowing for a comparison of affective reactions of both conversation partners over time. Space does not permit a full treatment of both qualitative and quantitative results. Given that our emphasis is on the dynamics of communication events in real time, and ways in which gender seems to affect how the interactions unfold, we have chosen to focus the presentation of dyad-level data from excerpts of the conversations, with emphasis on participants’ explanations for fluctuations in anxiety. The selection of excerpts will be guided by qualitative analysis of the conversations; to maintain a focus on individuals we will not combine data across dyads. The main objectives of this study can be stated as follows:

• Describe how females differ in their explanations of dynamic changes in anxiety between same-gender and mixed-gender dyads.

• Interpret the evidence of anxiety and uncertainty in the rationales provided for the affective reactions.

• Describe accommodation strategies enacted and how do they relate to ongoing communication.

Method

Participants

A total of 72 people completed a pre-test questionnaire during regularly scheduled university classes. The questionnaire contained measures of communication traits (noted below) and asked about favorite topics of conversation, as well as respondents’ willingness to volunteer to come to the lab for a videotaped conversation with a stranger. More than half of the sample (41 of 72) who completed the questionnaire agreed to be contacted for the lab portion of the study; scheduling conflicts and no-shows reduced the final number to 24: 19 females and 5 males. Participants ranged in age from 19 to 32 years, and the average age was 22.5 years. All participants were Caucasian, native English speakers, and were raised in North America.

Materials

1. Cameras—Two high-definition webcams were used to record the participants’ conversations and reasons for changes in anxiety. The cameras were attached to personal computers and videos were encoded directly into a digital file in Microsoft Windows Movie “.wmv” format. The recordings were displayed using Microsoft Windows Media Player embedded in specially written software.

2. Software—Software written for idiodynamic studies (“The Anion Variable Tester”) both played the respondents’ video and recorded their ratings. Participants used a computer mouse to change the reported anxiety level by left-click (increase) or right-click (decrease). Ratings ranged from +5 (high anxiety) to −5 (low anxiety) and appeared on screen. The program featured an auto-zero function that returned the rating to zero at the rate of 1 unit per second. This required participants to constantly monitor the accuracy of anxiety levels showing on the screen. Once ratings were completed, the program provided a bitmap graph of the anxiety ratings and a spreadsheet compatible with Excel of the ratings of anxiety per second.

Procedure

In a pre-test conducted at the end of a regularly scheduled lecture period, participants who gave informed consent completed a questionnaire that included scales measuring trait communication apprehension (McCroskey, 1982), perceived competence (McCroskey & McCroskey, 1988), and willingness to communicate (McCroskey & Richmond, 1987). A description of the lab portion of the study was included at the end of the questionnaire where participants were invited to volunteer to have a brief, recorded conversation with a stranger. Four conversation topics were provided which participants rated from 1 (wanted to talk about most) to 4 (wanted to talk about least). Topics included hobbies and interests, favorite foods, favorite television show or movies, and favorite music. Consistent with the research ethics protocol, participants’ contact information was removed from the questionnaire after they had been contacted for part 2 of the study to protect anonymity. Conversations took place within 3 weeks of completing the pre-test questionnaire.

Each participant volunteering for part 2 of the study was randomly paired with a stranger. It was confirmed at the time of scheduling a lab appointment that they did not know their interlocutor. Each participant was assigned to one of the two conditions: female–female (7 pairs) or male–female (5 pairs). Participants signed a second consent form, filled out a demographics sheet, and rated how similar to themselves they perceived their partner to be based on a first impression. Pairs were videotaped using a webcam while having a conversation. No time limit was imposed in advance, and the conversations were approximately 2 to 5 min in length. The suggested topic of conversation was given by the research assistant based on the ratings of willingness to discuss each of the topics, as noted on the pre-test questionnaire. For example, if both rated willingness to talk about “favorite music” highly (1 or 2), then music was suggested as an initial topic. Participants were not necessarily constrained to any topic. Two female research assistants were present during the conversations. Immediately after each conversation was finished, each research assistant took one participant to a separate room to watch the recording of the conversation. Participants rated their anxiety level throughout the conversation, as they watched the video, using the Anion Variable Tester software.

Instructions for using the software asked participants to rate their anxiety level, with anxiety being defined by feelings of “nervousness, worry, and/or tension.” These three specific terms were used to help define anxiety for the participants and to acknowledge the affective, cognitive, and physical components of anxiety (see Spielberger, 1983). Participants then watched the video a second time; the researcher stopped the playback when spikes or dips in ratings were visible and asked participants to explain the reasons for the changes in ratings. Participants watched the video a third time, rating the anxiety level of their conversational partner. Finally, the video was played a fourth time asking participants to explain reasons for changes in ratings of their conversation partner. With the repeated viewing of the videos, the testing sessions lasted approximately 1 hr. The full testing session was transcribed and analyzed to identify themes present in the reasons for changes in anxiety ratings, which will be the focus of the description of results in the following.

Approach to Coding the Conversations

No pre-existing coding scheme was used; rather, we adopted a grounded theory approach to data analysis (Strauss & Corbin, 1990, 1998). Each dyadic conversation is unique, with both expected and unexpected factors that may have influenced the results in one way or another. Given the idiosyncratic nature of communication, none of these conversations could ever be replicated, even with the same participants. Therefore, to help us understand the fluctuations in the ratings, we will focus on a qualitative analysis of the themes identified by the respondents during the interview portion of the study.

Immediately after the conversations, speakers were presented with a graph showing their quantitative idiodynamic anxiety self-ratings (−5 to +5) as well as her or his own ratings of their partner. The quantitative data were completely embedded within a qualitative approach (Cresswell & Plano-Clark, 2011). The combination of ratings and respondents’ explanations are interpreted as reflecting the respondents’ understanding of anxiety, for both self and other, within their conversations. During analysis, the transcripts, videos, and graphs were reviewed together to aide in interpretation, providing more information than is present in either transcripts or audio recordings alone. Using a constant comparison approach, open coding identified themes in each individual’s video. Using an axial coding process, the number of themes was reduced by grouping them. The process was undertaken separately for the female-only and male-female dyads. Finally, comparison of the themes generated by the axial coding process allowed for the possibility of an overarching theme reflecting the difference between female-only and male-female dyads. Member checks (Maxwell, 2005) were performed with five of the participants to test the interpretation of the results; all agreed that the major categories fit their experience.

Results

We have organized the results into six inter-related themes: awareness of the camera and researchers, comparison of self and partners, self-judgment, worry about other’s judgment, disinterest, and reaction to miscommunication.

Awareness of the Camera and Researchers

It is safe to say that the context of the present study reflects a truly unusual situation in which to meet a stranger. Some female participants expressed an uncomfortable awareness of being videotaped in a room that included their conversation partner and two research assistants. In their interviews following the conversation, none of the male participants mentioned either the camera or researchers. However, four of the females, two in male–female dyad and two in female–female dyads, suggested that their own anxiety ratings were affected by the camera and/or researchers in the room. For some, the camera was an issue “. . . knowing I was on video camera made me anxious because a couple of times I noticed I was trying not to laugh because I was on camera . . .” (Female, Participant 18: F, P18). Another female participant (P16) said that both the camera and the researchers being present during the experiment lead her to experience more anxiety than she normally would. Figure 1 presents a 20-s excerpt of the conversation, the corresponding idiodynamic self-ratings of anxiety, and the rationale provided during the interview by both the male and female interlocutors.

Figure 1.

Idiodynamic anxiety ratings for Participants 15 and 16 in the first 20 s of their conversation.

Although awareness of the camera and researchers seemed to increase anxiety for some participants, two other females noted that at some point during the study, the salience of the camera and the researchers diminished. One female participant stated,

. . . I feel like we were actually talking and it was like there was no camera it was like you guys [the female research assistants] weren’t even there. It’s like we would talk just because we were getting to know each other and it wasn’t like scrutinized or anything . . . (F, P3)

From another dyad, a female participant felt,

the conversation was you know we just gelled more kind of thing we . . . we weren’t really aware of like the camera anymore and that kind of thing right like the actual physical environment we were just kind of engaged in a conversation. (F, P8)

These comments suggest that the camera and the researchers could be anxiety-provoking until a genuine interest in the conversation developed. In general, the women described an awareness of their surroundings as a salient part of the conversational dynamic in ways that the men did not. The role of context in shaping affective reactions is highlighted in these descriptions, but even the highly atypical features of a context (e.g., cameras and researchers) can be ignored when conversation flows well.

Comparison of Self and Partner

A second theme that was mentioned only by females was a comparison between self and partner. This theme emerged within five of the 12 dyads: four female–female and one male–female. Female participants sometimes linked anxiety ratings to their appearance on camera (physical appearance, nonverbal communication behaviors, etc.), relative to their conversational partner. This category of social comparisons was only mentioned when discussing the ratings of the other person’s perceived anxiety level but not when discussing one’s own anxiety ratings. For example, one female participant rated her conversational partner as calm throughout the entire session (never on the anxious side of the graph). However, she did notice changes in her conversational partner who was viewed as “. . . still not really nervous but you know she’s just starting to talk like she seems like she’s really tense a little bit and you know like not as bad as me but . . . (giggles) ¹ ” (F, P3). In more extensive example of this theme, Figure 2 presents a 20-s excerpt of a conversation between two females; note that the graph shows one participant’s (P6) ratings of her own anxiety and what she perceived to be her partner’s anxiety. P6 makes note that her partner shifted the burden of first self-disclosure to P6, after P6 had asked the other person to go first. This non-accommodating conversational maneuver, combined with her interlocutor’s nonverbal behavior, appears to both heighten anxiety for P6 and reduce the anxiety she perceived in her conversation partner. In effect, her partner’s verbal and nonverbal behavior put P6 on the spot, creating an imbalance of power and temporary differences in status that are reflected in P6’s feelings of anxiety and her perception that her interlocutor was feeling relaxed.

Figure 2.

Idiodynamic anxiety ratings for Participant 6 and Participant 6’s ratings of her conversational partner (Participant 5) in the first 20 s of their conversation.

Self-judgment

At several points in the transcripts, female participants made spontaneous negative comments about themselves. Judgments of physical appearance “My little gut . . . oh (laughs) are we recording?” (F, P21), nonverbal communication behavior “I’m clicking the pen so much . . .” (F, P2), and their performance within the conversation “I just feel like she . . . I talked a lot during this video . . . no I just feel like she was just listening so she didn’t have to worry about what to say next or something . . .” (F, P17). Figure 3 presents the first 20 s of a conversation that did not begin smoothly. Note that P23 was critical about her appearance without prompting and offered her comments as an explanation for the rating of her partner’s anxiety, not in the context of her own anxiety. For comparison, P23’s partner recorded zero ratings both for herself and for her assessment of P23 throughout this time period; the anxiety reactions appear to be one-sided. P23’s comments reflect the evaluation of herself as she contemplated the opening of the conversation, partially tied to her partner’s more subdued reaction. However, P23 assumed that she was making her partner nervous even though it is not readily apparent from the idiodynamic ratings, transcript, or video. Perhaps this reflects a tendency present during an attribution process, wherein the view of self-related affect is reflected in one’s assessment of others. Ratings of self and other for P23 showed remarkable convergence but were very different from her partner’s sets of ratings.

Figure 3.

Idiodynamic anxiety ratings for Participant 23 and Participant 23’s ratings of Participant 24 in the first 20 s of their conversation.

Worry About Others’ Judgment

Not only were some of the female participants making negative self-judgments, but some commented on how they believed their conversational partner might be judging their self-disclosure. Anxiety levels tended to increase for females when they felt they were disclosing personal information that could lead them to experience embarrassment, especially when they were somewhat uncertain about their conversational partner’s reaction. Figure 4 presents a 15-s excerpt between two females that occurred approximately 3½ min into the conversation, when both reported feeling comfortable. Prior to this, P3 explained that she was a vegetarian and P4 had said she enjoyed fried chicken (from KFC restaurant). P4 was both discussing eating meat and eating food generally perceived to be unhealthy; she expressed concern that her partner might form a negative impression of her. In other conversations, participants reacted to what they thought might be expressions of increased anxiety resulting from conversational partner’s self-disclosure. For example, “when she told me like her favourite singer I don’t know she seemed like she was afraid I was going to judge her or something for it, so she looked more nervous, I guess, to me” (F, P23). Examples in this category show the need to manage uncertainty; participants did not know for sure how they were being judged. Not only were female participants worried about how they were being perceived but they also worried about appearing to be judging their partner. This suggests an awareness of how the respondent is being perceived by others within the female dyads that was not observed in the mixed-gender dyads. There is a considerable degree of uncertainty, complexity, and the potential for rapid arousal of anxiety in this reflected appraisal process. If the process becomes consciously perceived, an interlocutor, exemplified by P23’s account in Figure 4, may decide whether to acknowledge it and/or react to it or let it slide into the background. That is, communication accommodations might or might not be made upon perceiving an interlocutor’s anxiety about being judged.

Figure 4.

Idiodynamic anxiety ratings for Participant 4 and Participant 3’s ratings of Participant 4 in the time range of 210 to 225 s of their conversation.

Disinterest

In all five of the male–female dyads, the theme of disinterest emerged. Participants discussed interests that differed from their conversation partner in television shows, movies, music, food, and hobbies. Interestingly, when miscommunication occurred, female participants were most often the ones to probe for further information or try to explain their interests to the male partner. For example, after her male partner identified a specific television show that he watched, the female made an effort to learn more about her partner’s interests, “Oh, I’ve never even heard of it. Is it just newer now?” (F, P8). The data set contains occasions wherein male participants expressed little desire to learn about their conversational partner’s interests. In Figure 5, a 20-s excerpt from a conversation between a male and female, at approximately the 1-min mark, the male directly indicates that he did not know anything about the television show that his partner was describing. During the rating of the video, the female participant recognizes that she was giving out nonverbal signs of anxiety, but claims that she does not know why she felt that way. The male participant’s comments indicate that he was not at all interested in continuing this line of conversation. He reinforced this opinion in his commentary about the video when he said “. . . and I was like ‘who the hell is Jason’?,” a thought he did not vocalize. In this case, the interview reveals in more depth what was apparent during the conversation—he had no interest in what she was saying. Not only did this effectively terminate the conversational topic, he showed a bump in anxiety lasting only 1 s. This was not an emotion-arousing conversational event, as neither party reacted strongly.

Figure 5.

Idiodynamic anxiety ratings for Participant 9 and Participant 10 in the time range of 60 to 80 s of their conversation.

Three approaches to disinterest within the conversations were present in the data: (a) the males simply ignored the topic offered by the female, (b) males made a negative comment about their partner’s interests, or (c) the male stated he did not know anything about the topic, signaling a desire to end that topic of discussion. Male participants were clear about their disinterest in certain topics:

I don’t really get the whole reality TV, even though it’s not reality TV, like Survivor and stuff like that where it’s this intense reality show and all of a sudden cameras go off and I’m pretty sure they go to like a five star hotel. (M, P15)

Other participants were less forward with their disinterest mentioning it only to the researcher when explaining their changes in their anxiety levels

Umm . . . that was mostly ‘cause I think when she talked about the show that I didn’t know I was like “Oh God what’s this?” and then it’s like oh this is a bike show and it’s like oh okay I thought she was going to go into some elaborate thing and I just be like “Oh my God what are they talking about?” (M, P10)

Participants also picked up on the disinterest of their partners and believed it lead to changes in anxiety

. . . it seemed like we’re having a conversation and then when I said I didn’t understand . . . like I didn’t know the show it just . . . he was just kind of like “OHH . . .” like kind of shocked, like that right, and just kind of increased . . . like “oh now I have to explain it,” then once he got comfortable with actually explaining it, he seemed to reduce the anxiety again “cause he knew what he was talking about . . .”. (F, P8)

The female speaker in this excerpt seems to be suggesting that her male conversation partner was somewhat oblivious to her lack of experience with a particular topic (‘shocked’ that he had to explain it). Her attempts to go with the flow of conversation and maintain harmony were unsustainable, and she expressed the need for an explanation. As might be anticipated, this led to an emotional reaction for both participants as the conversation backtracked. Excerpts expressing self-expression being favored at the expense of interpersonal harmony are not present in the female–female (intragroup) data.

Reaction to Miscommunication

Miscommunication emerged within both male–female and female–female dyads, but different dyads seemed to react to them in different ways. Miscommunication occurred when one participant spoke about a particular topic and the conversational partner understood something different. This was given as a reason for increases in anxiety level. For example,

[y]eah because I felt like I was explaining something and he just didn’t understand what I was saying so I was like Oh God, so I had to keep rambling cause he was just kind of not as responsive I guess he was just kind of like . . . so then it made me more anxious . . . (F, P13)

Figure 6 presents a 20-s excerpt from a male–female conversation. During this conversation, the male participant mixed up the name of an actor with the name of a character, leading to confusion for both participants and peaks in their anxiety ratings.

Figure 6.

Idiodynamic anxiety ratings for Participant 9 and Participant 10 in the time range of 150 to 170 s of their conversation.

The theme of miscommunication emerged within four of the five male–female pairs, but only one of the seven female–female pairs. Potentially more interesting is the reaction to miscommunication in the female dyad—harmony was preserved because the female participant who was misunderstood did not make her partner aware of the misunderstanding, she attempted to smooth it over within the conversation. In the post-conversation interview, P19 said, “I was like ‘oh’ cause I was agreeing with her but I wasn’t. I didn’t want to be mean . . .” Figure 7 presents the 20-s excerpt from this conversation along with the idiodynamic anxiety ratings and explanations for changes in anxiety for both females. P19 suggested it would be “mean” to break into the conversation and reveal the miscommunication. Generally, when miscommunication occurred within a male–female dyad, it led to breaks in the conversation, awkward silences, and participants making efforts to explain themselves before the interaction again began to flow well. Within the female–female dyad in which miscommunication was observed, the conversation never stopped flowing because the misunderstanding was not revealed to the conversation partner.

Figure 7.

Idiodynamic anxiety ratings for Participant 19 and Participant 20 in the time range of 45 to 65 s of their conversation.

General Discussion

This study is unique in using anxiety as a window into gender differences in communication accommodations. By embedding quantitative ratings within a qualitative approach, subtle differences in the dynamics of affective reactions and associated conversational patterns between strangers can be observed. The methodological approach used to study the communication activity has not been used previously in this way, and we argue that it represents an innovation in the study of dyadic communication.

Together, both the AUM theory and CAT provide a framework for understanding the different approaches that women and men took to the conversations they had with strangers and how they understood the dynamics underlying the communication events. Examining the recordings of the conversations led participants to express an interesting set of rationales for their ratings. Consistent with the terminology of CAT (Dragojevic et al., 2015), we observed that in this set of conversations, women more often used an initial orientation toward convergence, while the men used approaches reflecting both divergence and maintenance strategies; men more often stood their ground and waited for their partner to accommodate. In general, enacting divergent and maintenance strategies has the potential to create uncertainty in ascribing underlying interpersonal motives for communication. In the words of P19, drawing attention to communication problems can be interpreted as being “mean” to one’s interlocutor.

The female participants in the present study described how their anxiety level from moment-to-moment was affected by a number of contextual factors such as being videotaped, the presence of the female research assistants, and the behavior of their conversational partner. Individuals described how they were affected by the other person and shaped by the situation (Markus & Kitayama, 1991; Moscovitch, Hofmann, & Litz, 2005; Singelis & Brown, 1995). Results from the present study are consistent with Frantz et al. (2005) who found that females, compared to males, appear to have a greater concern for how other people perceive them and are more likely to make comparisons within their gender. Compared to males, the female participants described their communicative actions and affective reactions as being more strongly shaped by the actions and perceived reactions of their conversational partner; women showed stronger tendencies toward convergence. The female participants demonstrated a willingness to adjust their behavior to accommodate the other person in ways that men in this study did not (Gallois & Giles, 2015; Markus & Kitayama, 1991; Singelis, 1994; Singelis & Brown, 1995). Social comparison is necessary to be able to adjust one’s own behavior to be better able to relate to the person with whom one interacts. But there may be a price to pay for this accommodative approach. Increased concern about being judged by others was associated with anxiety reactions, something that is reinforced by popular media especially among women (Frantz et al., 2005).

Within the present study, we found little indication from the male participants that contextual factors were influencing their anxiety ratings, as was the case with their female counterparts. While males did not describe much social comparison or worry about how they would be perceived by their female conversation partners in this situation, women did describe being influenced by those two factors. Individuals who are not influenced by contextual factors seem to focus on their own abilities and attributes rather than situational factors (Markus & Kitayama, 1991; Singelis, 1994; Singelis & Brown, 1995), and they can be more direct in their communication (Baumeister, 1998; Singelis & Brown, 1995).

Perhaps the clearest examples of gender differences in approaches to communication in the present study can be found in Figures 5 through 7. On one hand, Figures 5 and 6 show a brief rise in anxiety for the male participant only at the moment he explicitly dismisses his partner’s choice of topic or realizes he has made a mistake. On the other hand, when the female partners in Figure 7 choose to “go with the flow” of conversation, both uncertainty and anxiety rise for several seconds, even as interpersonal harmony is maintained. In these cases, miscommunication in a male–female dyad led to awkwardness and breaks within the interaction. However, miscommunication within a female–female dyad was smoothed over until a more accurate understanding was reached. The male participants generally seemed less willing to adapt to their partners, instead preferring relatively autonomous, direct self-expression consistent with maintaining social distance from their female interlocutor. This approach has the dual effect of reducing uncertainty and managing anxiety, albeit at the expense of interpersonal harmony. The effects on anxiety are consistent with AUM theory (Gudykunst & Nishida, 2001; Samochowiec & Florack, 2010; Stephan et al., 1999).

Cross and Madson (1997) suggest that females tend to form smaller, closer personal relationships; women think in terms of dyads (Foels & Tomcho, 2009). Males, however, form relationships that focus more on group membership and categories with a larger number of people, drawing identity from groups (Tajfel & Turner, 1986). To the extent that this pattern generally reflects participants’ approach to the specific conversational situation used in the present study, especially for the initial orientation, different types of communication accommodations might be expected. CAT predicts that, over time, relationships can become more individuated, and shared idiosyncratic history helps shape conversations between friends and acquaintances. However, strangers lack the personal history shared by friends and may be managing the salience of group membership (e.g., female vs. male) as they enact their identity.

There is some risk in research emphasizing gender differences in communication behavior (Hyde, 2005; MacGeorge et al., 2004). The groups are in constant contact and they have extensive, intimate knowledge of each other. However, CAT theory suggests that interlocutors might choose to emphasize differences as an optional conversational strategy that can be enacted moment-to-moment. People are able to choose to accommodate each other or not—a point of emphasis for CAT but not reflected in the gender-as-culture literature so far. In CAT, cultures interact, adapt, and regulate relationships, acting in ways to emphasize similarities or choosing to act in ways that emphasize group-level differences (Gallois & Giles, 2015). One of the reasons this process matters is that there is uncertainty and anxiety about communication motives and appraisals that other people are making, both across groups and within them.

Uncertainty and anxiety are negotiated continuously during interpersonal communication, and the ongoing processes of managing one’s self-construal and self-presentation play a role in the ongoing communication process. Conversational twists and turns may be partly the result of choosing to emphasize interpersonal versus intergroup similarities or differences in reaction to previous events as communication unfolds. Such a process might not be reflected in broad indices of communication ability, but in specific actions, reactions, and the rationales underlying them.

In considering the broader pattern of results and how the categories might fit together as a whole, there is more than a passing consistency between results from the present study and Markus and Kitayama’s (1991) notion of interdependent versus independent self-construal. An interdependent self-construal emphasizes that the self is connected to others and has an obligation to fit in with one’s surroundings, a sense that “we are in this together.” An independent self-construal emphasizes that each person is a separate and unique individual, a person who “will do my own thing.” Although Markus and Kitayama’s theory did not guide the choice of themes, in the end five of the six categories reported above in which female participants feature prominently provided examples of contextual factors, such as the researchers and video camera, influencing anxiety ratings. Furthermore, females showed more awareness and concern for how they were being perceived by others, and they more often compared themselves with others and worried about the judgments of others. In the present sample, only males showed disinterest, arguably expressing independence from his conversation partner. Examples of miscommunication were more evident in the mixed-gender dyads, but that miscommunication was handled in different ways; the female–female pair worked to preserve harmony between persons but in mixed-gender dyads reported here, males more often wanted to explain themselves.

By restricting our sample to dyads including females, we are able to see how individuals can approach conversations in different ways. AUM and CAT theory help to provide a deeper understanding of what is happening moment-to-moment as each interlocutor manages anxiety and uncertainty by making specific conversational choices to accommodate their partner versus expressing independence. In the present dataset, women’s conversation appears to be influenced at least in part by broad expectancies in the form of interdependent self-construals, consistent with the “different cultures” view where women appear more collectivist and men more individualistic (Hofstede, 1980; Triandis, 1995). But there also is an element of choice involved as a specific conversation unfolds and many other factors come into play, such as the decision to accommodate one’s conversation partner. Presumably, all participants had access to the same “conversational toolbox” of accommodation strategies, so the potential willingness to use those tools differently with strangers may reflect only the opening conversational gambit(s) used by each gender. Perhaps this helps to account for the relatively small gender differences previously reported in the literature using other methods (Hyde, 2005; MacGeorge et al., 2004).

Although the concept of interdependent–independent self-construals has been highly influential in conceptualizing culture and communication, the evidence in support of the theory is mixed. In critiquing research into interdependent–independent self-construals, Matsumoto (1999) suggested that “(i)f future research is to keep up with, and push, our thinking in a progressive manner, we need to complement our traditional methods of research with new, and old, methods” (p. 306). The idiodynamic method used here is novel in that it can show the convergences that occur in real time among the informational elements of conversation, the relationship between dyadic partners, communication accommodations that are made or not made, and AUM processes. The per-second ratings made by the participants provide a novel dimension to the research that previously has not been reported. The many layers of meanings within conversations interact in a continuous stream that cannot be reduced to group-level tendencies, as much of the prior research has done, without losing a substantial amount of information.

Limitations and Future Research

There are limitations with the method and with current study to be considered in the interpretation of the results. First, the method requires dense, longitudinal, and individual-level data (van Dijk et al., 2011). This presents a limitation that led to a qualitative/interpretative approach in the present study, one that is focussed on the detailed meaning of the conversation to the participants rather than statistical analysis leading to generalizations about conversations. Prior research has used methods that better allow for generalization, and the present research is complementary to the existing literature. A second limitation is the specificity of the experimental situation, which was limited to female participants and research assistants, varying only the gender of the interlocutor. Given the amount of data produced, the per-second specificity of the conversation ratings reported above, and the detailed descriptions that are available with the idiodynamic method, it was necessary to limit the study to allow for meaningful comparisons across conversations. The idiodynamic method lends itself to small-n studies that address the nuances in the dynamics of communication as they occur in real time. Small-n allows for a specific focus and presentation of individual-level results, but also imposes limitations on the breadth of the study, and the generalizability of the results cannot be assumed.

In addition, given the novelty of the idiodynamic method, it is unknown at this time whether idiodynamic ratings better reflect introspection (a self-report of their affective state during the conversation) and/or a self-perception process (a report of their affective state based on observable cues in the video); there is no way to tease apart these possibilities within the present study. Future research might manipulate the audio/video combinations to separate introspection from self-perception processes or use eye-tracking methods to better understand the focus of participants’ attention when watching themselves on video.

Within the idiodynamic method, specific elements of the procedure also impose limitations. The focus on popular culture topics (movies, TV, food, etc.) clearly affected the direction of conversations and the affective reactions that went with them. Different topics, such as controversies in politics or religion, might reveal different ways of making communication accommodations, creating and managing anxiety and uncertainty, and enacting self-construals. A second limitation of the procedure employed here is the artificiality of the setting. Although the present study allowed for experimental control, the approach sacrificed some features of natural communication. Participants described experimental artifacts, such as the video camera or the researchers being present in the room, as influencing the way they communicated some of the time. A third limitation is the random assignment of conversation partners which might have produced conversational trajectories that differ from those that might arise if partners were self-selected or the persons already were friends or acquaintances. Knowing one’s conversation partner would likely affect initial uncertainty and anxiety conditions, along with accommodation processes.

In future studies, it would be interesting to include male-only dyads and/or small groups (3-5 persons) to examine AUM in a more complex conversation and assess other potential differences in communication accommodations. It is possible that male-only dyads might experience somewhat different processes than were the focus in the current study. It also is important to consider the roles played by the research assistants who are in the room with the conversation participants. The present results show that participants notice the presence of the research assistants—they are not simply part of the furniture. Our approach was to employ two female research assistants because the focus was on how women react differently when conversing with another woman versus a man. If this study were to be expanded with other possible combinations of male–female partners and research assistants (e.g., female versus male researchers interviewing male participants), carefully accounting for the role of the research assistants in the idiodynamic method is necessary.

Future studies might also include individual difference variables such as personality factors (for example: extraversion, neuroticism, emotional intelligence) to examine how traits interact with the moment-to-moment perception and interpersonal negotiation of anxiety. It might be interesting also for researchers to combine a physiological measure, such as heart rate, event-related potentials, or other physiological indicators with the idiodynamic ratings to examine the relationship between physiology and self-reported affect as it changes in real time. Further studies might focus specifically on real-time processes occurring for individuals with diagnosed conditions, such as anxiety disorder, or ways in which specific social or psychological contexts integrate with the dynamics of conversation trajectories. Research using the idiodynamic method is in its infancy, and numerous adaptations and modifications to this method are possible.

Conclusion

The results of the present study show specific ways in which anxiety and uncertainty are managed and integrated with self-construals, from one moment to the next, within a conversation. The rationales underlying female participant’s changes in anxiety are consistent with a desire to maintain harmonious interactions, and communication accommodations tended toward convergence with their conversation partner. The ways in which male participants approached talking to a female interlocutor seem to reflect more strongly a desire to maintain their individuality and reinforce their group membership through maintenance or divergent communication accommodations. Using an innovative method to combine quantitative anxiety ratings of the self and other, along with participants’ rationale for changes in affect, assessed in real time, reveal both broad and subtle processes that underlie communication processes as they are enacted.