This article was downloaded by: [University of Minnesota Libraries, Twin Cities] On: 03 May 2013, At: 06:07 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Communication Monographs Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rcmm20

Responses to nonverbal intimacy change in romantic dyads: Effects of behavioral valence and degree of behavioral change on nonverbal and verbal reactions Laura K. Guerrero Burgoon

a b

c

, Susanne M. Jones & Judee K.

d

a

Associate Professor in the Hugh Downs School of Human Communication b

Dept. of Communication, Arizona State University, Tempe, AZ, 85287–1205 E-mail: c

Assistant Professor of Communication, University of Wisconsin, Milwaukee d

Professor of Communication, University of Arizona Published online: 02 Jun 2009.

To cite this article: Laura K. Guerrero , Susanne M. Jones & Judee K. Burgoon (2000): Responses to nonverbal intimacy change in romantic dyads: Effects of behavioral valence and degree of behavioral change on nonverbal and verbal reactions, Communication Monographs, 67:4, 325-346 To link to this article: http://dx.doi.org/10.1080/03637750009376515

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-andconditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

arising directly or indirectly in connection with or arising out of the use of this material.

Responses to Nonverbal Intimacy Change in Romantic Dyads: Effects of Behavioral Valence and Degree of Behavioral Change on Nonverbal and Verbal Reactions

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

Laura K. Guerrero, Susanne M. Jones, and Judee K. Burgoon Various theories of nonverbal adaptation feature behavioral valence (i.e., positive vs. negative behavior) and degree of behavioral change (e.g., very low vs. low intimacy) as critical elements affecting whether changes in nonverbal intimacy are met with reciprocity or compensation. The present study, which utilizes data from 100 romantic dyads, makes comparisons across five conditions: very low intimacy, low intimacy, very high intimacy, high intimacy, and a no change (control) condition. Repeated measures analyses of variance showed that targets in the two high intimacy conditions reciprocated their partners' intimacy change by appearing more nonverbally involved and pleasant and engaging in more verbal intimacy. Targets in the two low intimacy conditions also reciprocated by becoming less nonverbally pleasant and fluent as well as more verbally hostile. However, these targets also used verbal repair strategies, thereby showing some degree of compensation. Targets in the very low intimacy condition also became more vocally anxious and less composed after their partners decreased intimacy. Targets in the very low intimacy condition were abo particularly likely to engage in verbal repair strategies. These and other findings are discussed in light of their implications for applying theories of nonverbal intimacy exchange to the context of romantic relationships.

S

cholars of nonverbal and interpersonal communication have long acknowledged the importance of studying patterns of behavioral adaptation. Adaptation generally refers to nonrandom patterns of synchronous or asynchronous behavior (Burgoon, Stern, & Dillman, 1996). For example, in romantic relationships, people may respond to partners' increases in intimacy by becoming more intimate themselves, thus creating a synchronous interaction. Conversely, individuals may respond to increases in intimacy by pulling away, creating an asynchronous interaction. The ways that partners respond to one another's attempts to increase or decrease behavioral intimacy help shape how connected (or disconnected) couples feel (Andersen, 1998). Several theories strive to explain processes of nonverbal adaptation within dyadic interaction. These include affiliative conflict theory (Argyle & Dean, 1965; Argyle & Cook, 1976), expectancy violations theory (Burgoon, 1978, 1983; Burgoon & Hale, 1988), discrepancy-arousal theory (Cappella & Greene, 1982, 1984), cognitive valence theory (Andersen, 1983, 1984, 1998), interaction adaptation theory (Burgoon, Stern, & Dillman, 1995) and the sequential-functional model (Patterson, 1982, 1983; see Andersen & Andersen, 1984, Burgoon, Stern, & Dillman, 1995, and Hale & Burgoon, 1984 for reviews of these and other adaptation theories). These theories focus on two specific adaptation processes, reciprocity and compensation, which are contingent upon a change in the partner's behavior. As Burgoon, Dillman, and Stern

Laura Guerrero is an Associate Professor in the Hugh Downs School of Human Communication. Susanne Jones is an Assistant Professor of Communication at the University of Wisconsin, Milwaukee.Judee Burgoon is Professor of Communication at the University of Arizona. An earlier version of this paper was presented at the 1998 Western States Communication Association conference in Denver, CO. Address all correspondence to: Laura Guerrero, Dept. of Communication, Arizona State University, Tempe, AZ 85287-1205. E-Mail: [email protected]. Communication Monograph, Vol. 67, No. 4, December 2000, pp. 325-346 Copyright 2000, National Communication Association

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

326

COMMUNICATION MONOGRAPHS

(1993) described, both reciprocity and compensation occur in response to the partner's behavior change. Reciprocity occurs when a person's response carries a meaning that is similar to the partner's initial behavioral change (i.e., an increase in smiling is met with a warm glance). Compensation, in contrast, occurs when a person's response carries the opposite meaning as the partner's initial behavior change (i.e., an increase in smiling is met with a cold stare). Importantly, reciprocity and compensation can occur via either nonverbal or verbal channels, even though the majority of work on adaptation patterns has focused only on nonverbal responses (Andersen, 1985; Burgoon, Stern, & Dillman, 1995). To obtain a more complete picture of adaptation patterns, this study examines both verbal and nonverbal responses to changes in nonverbal intimacy. Studying adaptation patterns such as reciprocity and compensation is important for at least three reasons (Burgoon, Stern, & Dillman, 1995). First, patterns of adaptation are a defining feature of human interaction. Second, adaptation is a frequent conversational occurrence. Third, patterns of reciprocity and compensation are essential to defining and maintaining relationships. For example, reciprocating increases in intimacy while compensating for decreases in intimacy is one key to developing and maintaining a close relationship. This leads to yet another reason for studying patterns of reciprocity and compensation: Understanding these adaptation patterns may help scholars gain insight into the spirals of positive and negative behavior that can either enhance or destroy close relationships. Because we believe that patterns of reciprocity and compensation have particularly important implications for close relationships, this study focuses on adaptation to changes in nonverbal intimacy by romantic partners. For the purposes of this study, nonverbal intimacy is defined as behavior that is involved and pleasant. In particular, this study investigates how two theoretically-derived factors-behavioral valence and degree of behavioral change-affect responses to changes in the nonverbal intimacy level of an interaction. Behavioral valence refers to how positively or negatively a particular behavioral change is perceived. Studies have shown that individuals see increases in intimacy as generally positive, whereas decreases in intimacy are perceived as generally negative (Burgoon & Hale, 1988; Burgoon & Le Poire, 1993; Burgoon, Newton, Walther, & Baesler, 1989; Floyd & Voloudakis, 1999; Le Poire & Yoshimura, 1999). This pattern should be especially pronounced in romantic relationships, where there is usually an expectation for positive, affectionate behavior. Degree of behavioral change refers to the extent to which one partner initially alters her or his display of nonverbal intimacy. Theories such as discrepancyarousal theory (Cappella & Greene, 1982, 1984) and cognitive valence theory (Andersen, 1985, 1998) suggest that people will react differently depending on whether their partners have increased intimacy moderately or substantially. To examine both behavioral valence and degree of behavior change, the present study examines five levels of nonverbal intimacy change: very high intimacy, high intimacy, very low intimacy, low intimacy, and no change (the control condition). Our goal was to determine how the valence (increased vs. decreased intimacy) and degree of the nonverbal intimacy change influence a person's behavioral response. In other words, how do these factors affect the likelihood that people will reciprocate or compensate their romantic partners' changes in nonverbal intimacy? Several studies have shown that increases in intimacy are generally associated with attraction and positive emotional responses whereas decreases in intimacy are

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

RESPONSES TO NONVERBAL INTIMACY CHANGE

327

generally associated with less attraction and more uncertainty (e.g., Afifi & Burgoon, 1996; Guerrero & Burgoon, 1996). We also sought to determine whether the degreeto which behaviors are either positive (i.e., very high intimacy vs. high intimacy) or negative (i.e., very low intimacy vs. low intimacy) contributes to patterns of reciprocity and compensation of intimacy. The answers to these questions should inform theory on interpersonal adaptation while testing some predictions specific to expectancy violations theory (EVT), discrepancy arousal theory (DAT), and cognitive valence theory (CVT). This present study is unique in two respects. First, it examines four intimacy change conditions (very low, low, very high, high) and a control condition within the context of romantic relationships. Past studies on intimacy exchange in close relationships have tended to compare only two intimacy change conditions (e.g., low versus high or high versus very high) or they have not utilized a control group. Second, this study examines verbal as well as nonverbal responses to changes in intimacy behavior. Most studies on adaptation have focused exclusively on nonverbal involvement. Nonverbal involvement behaviors include immediacy cues such as forward lean and direct body orientation, as well as nonverbal cues that reflect expressiveness, composure, and attentiveness (Coker & Burgoon, 1987). Most studies that have examined verbal responses to intimacy change have focused only on self-disclosure or general expressiveness (Burgoon, Stern, & Dillman, 1995; Guerrero & Burgoon, 1996). The present study examines three classes of verbal behavior-strategies used to help repair the situation and/or relationship (verbal repair), verbal expressions of positivity and closeness (verbal intimacy), and verbal expressions of negativity and distance (verbal hostility). In addition to providing a more complete picture of the adaptation process, the inclusion of verbal responses might help illuminate some inconsistencies in past research. For example, Guerrero and Burgoon (1996) found that coders perceived targets in a low intimacy condition to engage in more compensation than reciprocity when a global measure of adaptation was employed. However, when specific nonverbal behaviors were coded, targets in the low intimacy condition appeared to reciprocate. Guerrero and Burgoon suggested that these targets may have compensated with verbal behaviors and reciprocated with nonverbal behaviors, especially since depth increased over time in the low intimacy conditions. The present study will help determine whether or not verbal behaviors vary based on changes in a partner's nonverbal intimacy. First, however, we provide a theoretical framework for this study by discussing the roles that behavioral valence and the degree of behavioral change play in various theories of adaptation. Theoretical Predictionsfor Increased Intimacy

In rewarding relationships such as those between romantic partners, increased intimacy typically carries a positive behavioral valence, whereas decreased intimacy typically carries a negative valence (Burgoon, Stern, & Dillman, 1995). According to some theories, however, very high increases in intimacy can sometimes be valenced negatively. For example, DAT predicts that if increased intimacy is highly discrepant from expected behavior, it can lead to aversive emotional reactions, avoidance, and compensation (Cappella & Greene, 1982). Similarly, CVT predicts that very high levels of intimacy change will lead to overarousal, defensiveness, and flight (Andersen, 1984, 1998). Thus, according to these theories, some compensation is

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

328

COMMUNICATION MONOGRAPHS

likely to be present when communicators violate expectations by engaging in very high levels of intimacy. EVT makes a different prediction. According to this theory, responses to intimacy change are based largely on two factors, behavioral valence and communicator reward valence. Communicator reward valence is based on assessing the costs and rewards of interacting with a person. People who are regarded favorably (e.g., as attractive, pleasant, etc.) are considered rewarding. Within romantic relationships, most people are likely to perceive their partners as rewarding. Thus, when a rewarding partner increases intimacy, EVT predicts that reciprocity should occur. Unlike DAT and CVT, EVT predicts that even very high levels of increased intimacy can produce positive affect and reciprocity. Empirical evidence generally supports EVT's prediction. Reciprocity appears to be the prevailing response to increased intimacy across a variety of relationships and situations (Burgoon, Stern, & Dillman, 1995), although a few compensatory patterns have been found on a limited number of variables. For example, Burgoon and Hale (1988) found that people generally reciprocated increases in global involvement, nonverbal intimacy, verbal intimacy, arousal/receptivity, and facial pleasantness, even though they did compensate for very high levels of eye contact near the end of the interaction. Andersen, Guerrero, Buller, and Jorgensen (1998) found that opposite-sex friends tended to reciprocate moderate increases in nonverbal intimacy, while showing a mixed pattern of reciprocity and compensation in response to large increases in intimacy. Le Poire and Burgoon (1994), who examined patterns of reciprocity and compensation during supposed interviews with physicians, found reciprocity to prevail across all manipulations, including both a high and very high involvement condition. In a similar study involving practice medical interviews, Le Poire and Yoshimura (1999) found that pleasant behavior was consistently met with reciprocity, regardless of whether people were led to believe that the interaction would be pleasant or unpleasant (see also Burgoon, Le Poire, & Rosenthal, 1995). Two other recent studies that focused on romantic dyads found fairly strong reciprocity effects for increased intimacy (Guerrero & Burgoon, 1996; Manusov, 1995). Of course, advocates of DAT and CVT can argue that these experiments did not produce increases in intimacy that were large enough to lead to overarousal and compensatory responses. They can also point to Andersen et al.'s (1998) mixed findings in the high intimacy condition as supportive of their claim that increased levels of intimacy can produce at least some compensatory reactions. Theoretical Predictionsfor Decreased Intimacy

When it comes to decreases in intimacy, the theories also make different predictions. DAT predicts that moderate changes in behavior lead to moderate arousal change, which is inherently pleasant and leads to reciprocity. Large changes in behavior, in contrast, lead to high arousal change, which is theorized to be inherently unpleasant and to lead to compensation (Cappella & Greene, 1982). The key to understanding how moderate versus large behavioral changes are defined lies in the region of acceptance. When behaviors deviate so much from what is expected that they fall outside the region of acceptance, they are considered to be large behavioral changes. These large changes then lead to substantial arousal change, aversion, and avoidance. Given that romantic partners typically expect a fairly high level of intimacy in their interactions with one another, the region of acceptance is

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

RESPONSES TO NONVERBAL INTIMACY CHANGE

329

likely to exclude both low and (and to a greater extent) very low levels of intimacy change. Thus, targets in the low and very low intimacy conditions should become overaroused and engage in avoidant behavior (i.e., behavior that is unpleasant and uninvolved), leading to a reciprocal response. Romantic partners may also feel freer to engage in reciprocal negative responses when interacting with one another as compared to strangers. Norms of social appropriateness and politeness are more likely to govern interaction between strangers. The logic behind DAT also suggests that very low levels of intimacy will produce stronger patterns of reciprocal avoidance than will low levels of intimacy. This is because the very low intimacy level is likely to be farther outside the acceptance region and, thus, will generate more arousal change, aversion, and avoidance. In contrast, EVT predicts that when rewarding communicators enact unpleasant behavior (by decreasing intimacy), targets compensate by becoming more intimate in an effort to bring their partners back to a comfortable level of intimacy (Burgoon, Stern, & Dillman, 1995). In DAT terms, the target will try to return the interaction to an intimacy level that falls within the acceptance region. Thus, EVT predicts compensation for decreased intimacy from rewarding partners. DAT, in contrast, predicts reciprocity, especially when the partner exhibits dramatic decreases in intimacy. It should be noted that CVT is only concerned with increases in intimacy (Andersen, 1985, 1998), and is therefore irrelevant when making predictions regarding responses to decreased intimacy. These contrasting theoretical predictions are intriguing when applied to romantic relationships. Past research suggests that romantic partners expect their interaction to be involved and pleasant, so how do they react when those expectations are violated? According to both DAT and EVT, they should feel negative affect. DAT predicts that people are then likely to communicate their negative affect to their partner, which leads to a reciprocal effect. Given the substantial work on negative spirals in close relationships (see Canary, Cupach, & Messman, 1995), such a reciprocal pattern is plausible. EVT, however, makes an equally reasonable prediction. Instead of reciprocating negative behavior, EVT predicts that a person might try to return the interaction to a more intimate level through compensation. In a sense, compensation would constitute an attempt at "behavioral repair." Thus far, empirical findings have failed to unequivocally support either EVT or DAT in regard to their predictions for decreased intimacy. Some studies have supported DAT's prediction by showing reciprocal avoidance in response to decreased intimacy. For example, Manusov (1995) found romantic partners to reciprocate rather than compensate decreased involvement. Le Poire and Burgoon (1994) also found that people reciprocated low levels of involvement and that those in the very low condition exhibited the most negative response pattern. However, they found no evidence to support DAT's prediction that very low levels of involvement would produce the most arousal. Still, these studies generally support DAT's prediction that decreased intimacy, which typically falls outside the acceptance region in romantic relationships, will lead to avoidance. A recent study by Floyd and Voloudakis (1999), however, supports EVT's prediction. Their study on friendship dyads showed that when one person decreased intimacy, the other person exhibited asynchronous behavior, which indicated that compensation occurred. Two other studies have found a mixed pattern of reciprocity and compensation in response to decreased intimacy. In a study of involvement and pleasantness,

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

330

COMMUNICATION MONOGRAPHS

Burgoon, Le Poire, and Rosenthal (1995) set up three inductions for an experiment looking at patterns of reciprocity and compensation among strangers. Some targets anticipated that their partner would be pleasant, others anticipated that their partner would be unpleasant, and a control group was not given any information on what to expect. Targets who expected pleasant interaction but received unpleasant communication showed a pattern of compensation when they were rated on overall verbal and nonverbal pleasantness, but exhibited reciprocity when they were rated on specific vocal and kinesic measures. Guerrero and Burgoon (1996) found a similar pattern. When a romantic partner decreased intimacy, the target showed an overall pattern of compensation, especially in terms of forward lean, body orientation, gaze, and overall body congruence, but reciprocated negative affect and inexpressiveness. Another recent study showed a mixed pattern of behavioral responses to decreases in nonverbal intimacy. Specifically, Le Poire and Yoshimura (1999) found that unpleasant communication evoked the reciprocation of kinesic involvement, and the maintenance of kinesic pleasantness and expressiveness, as well as vocalic involvement, pleasantness, and expressiveness. Thus, for five of the six dependent measures, Le Poire and Yoshimura found that behavior did not change in response to decreases in nonverbal intimacy. Summary

Past studies on patterns of reciprocity and compensation of intimacy change have been fairly consistent when looking at responses to increases in intimacy, but fairly inconsistent when looking at responses to decreases in intimacy. It appears that targets generally reciprocate increased intimacy, although there may be some limited compensation in response to very high levels of intimacy. Past findings also suggest that there is likely to be a mix of reciprocity, compensation, and maintenance in response to decreased intimacy in romantic relationships. Whether this mix leans more toward reciprocity, compensation, or maintenance in a low versus very low intimacy condition is still an empirical question. Furthermore, given the inconsistent results of past studies, it is important to continue investigating responses to decreases in intimacy across a number of different relational contexts. Perhaps romantic partners feel more at liberty to reciprocate negativity than do strangers, who may feel constrained by rules of social politeness (Guerrero & Andersen, 2000). Of course, it is equally reasonable to speculate that romantic partners will engage in compensation because they wish to maintain a high level of intimacy in their relationships. Given the large assortment of verbal and nonverbal behaviors that partners have at their disposal, it is also plausible that people engage in a mix of reciprocal and compensatory responses. To date, the vast majority of studies have only looked at nonverbal responses. Yet romantic partners respond to increases or decreases in nonverbal intimacy using both verbal and nonverbal behavior. In fact, the limited number of studies that have included verbal measures suggest that people sometimes compensate for low intimacy through verbal communication even though their nonverbal communication generally shows reciprocity (Burgoon, Le Poire, & Rosenthai, 1995; Guerrero & Burgoon, 1996). Based on Burgoon, Stern, and Dillman's (1995) prediction that decreased intimacy can lead to compensatory moves designed to bring the interaction back to a more intimate state, it is reasonable to predict that people might use verbal repair strategies to try to restore intimacy to interaction. For instance, when one partner

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

RESPONSES TO NONVERBAL INTIMACY CHANGE

331

pulls away, the other partner might use verbal repair strategies such as asking: "Is something wrong?" "Did I do something to upset you?" or "You still love me, don't you?" These types of questions can function to bring issues out in the open and restore intimacy. In addition to using nonverbal cues related to immediacy and positive affect (or nonimmediacy and negative affect) partners might respond to changes in intimacy by becoming more verbally intimate (through the use of in-depth disclosure, personal nicknames, and so forth) or more verbally hostile (by making rude comments, accusations, personal criticism, and so forth). Thus, along with examining a number of nonverbal cues that are commonly investigated in studies on responses to changes in intimacy levels, this study examines verbal communication related to repair, intimacy, and hostility. In addition to focusing on nonverbal rather than verbal responses, past studies on adaptation patterns in romantic dyads have tended to compare targets across a limited number of conditions. The only studies that have compared low, very low, high, and very high intimacy conditions have utilized strangers (e.g., Le Poire & Burgoon, 1994; Le Poire & Yoshimura, 1999). The two studies that have focused on romantic partners (Guerrero & Burgoon, 1996; Manusov, 1995) have compared two general conditions: increased versus decreased intimacy. Although these two studies found targets to reciprocate increases in intimacy, they obtained inconsistent results for the low intimacy condition. Guerrero and Burgoon (1996) found a mix of reciprocity and compensation in response to intimacy decreases (as detailed above), whereas Manusov (1995) found only reciprocity. These inconsistent findings, as well as the lack of research examining different degrees of intimacy exchange in romantic dyads, warrants a more precise test of the effects that nonverbal intimacy change has on behavioral patterns in romantic relationships. Therefore, the following hypotheses and research questions are advanced: HI: Targets in the high and very high intimacy conditions will reciprocate their partners' behavior by displaying increased nonverbal involvement and verbal intimacy. H2: Targets in the low and very low intimacy conditions will show a mixed pattern of reciprocity and compensation when both verbal and nonverbal responses are considered. RQla: Will target responses differ in the two increase intimacy conditions (high versus very high intimacy)? RQlb: Will target responses differ in the two decrease intimacy conditions (low versus very low intimacy)

Method Participants

Students from corporate communication classes at a large southwestern university were recruited to participate in a laboratory study with their romantic partners. Of the 100 dyads who participated, 13 were casually dating, 9 were engaged, 9 were married, and the remainder were in serious dating relationships. The average age of the participants was 22.4 years (range = 18 to 39) and the average length of the relationship was 2.1 years (range = 2 months to 16.4 years). Two couples were comprised of gay males; the remainder were heterosexual couples. Upon completion of the experiment, students in the corporate communication class were awarded extra credit in their communication class.

332

COMMUNICATION MONOGRAPHS

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

Experimental Procedures

The basic design was a 5 (involvement conditions) X 2 (premanipulation and postmanipulation conversations) factorial, with conversation utilized as a repeated factor. Subjects were randomly assigned to one of the five intimacy change conditions: high intimacy change, very high intimacy change, low intimacy change, very low intimacy change, and no change (control). Sex was balanced so that half of the targets (i.e., the person receiving the intimacy change) in each cell were female. Upon arrival at the research site, a comfortable apartment-like setting, dyads were told that they would be engaging in an experiment which was being conducted to determine how those in different types of romantic relationships (e.g., dating, married) discuss relationship issues, as well as completing a questionnaire for a separate study on personality issues. Both participants read and signed consent forms. Next, participants were told that they would engage in two separate videotaped interactions that would have different topical starting points. Each dyad was given a list of four personal topics that are discussed frequently in romantic relationships: future of the relationship, family, friends, and relational feelings (Guerrero, 1997). Couples took a few minutes to examine the list and decide which topic to discuss first. They were told that the topics should be thought of as "starting points" to prompt conversation, and that they should feel free to let the conversation go "off course." Participants attached lavaliere microphones and waited for the researcher to knock on the one-way mirror before beginning their discussion. After the first 3-minute conversation was completed, the researcher re-emerged and told the couples that they would be completing the personality questionnaires associated with the other study before they began their second conversation. Subjects were told that this was necessary because the experimenters wanted each conversation to begin on a distinct topic. Therefore, rather than continuing onto the next conversation, subjects were told they would take a "short break" to completethe questionnaires. Couples were then separated, supposedly to "ensure their privacy in answering the questions." The target was left in the main room with one of the research assistants to complete a personality inventory. The researcher escorted the confederate to another room, ostensibly to complete a similar questionnaire. Actually however, 80 of the 100 confederates were given instructions to either increase or decrease their normal level of nonverbal intimacy during the second interaction. The other 20 "confederates" completed a personality inventory and did not receive any instructions about altering their behavior. Forty of the confederates were asked to increase their intimacy level so that their behavior in the second conversation would be characterized by high (n = 20) or very high (n = 20) intimacy. Consistent with our conceptualization of intimate communication as behavior that is pleasant and involved, we instructed those in the high intimacy condition to act "warmer and more flirtatious toward their partner." We stressed that their behavior should be different in the second conversation than in the first conversation by "moving somewhat closer to their partner, smiling more, using more eye contact, and perhaps touching the partner." Those in the very high intimacy condition were given similar instructions, except that we stressed that their behavior should be dramatically different in the second conversation by engaging in behaviors such as "moving very close to their partners, smiling a lot more, using a lot more eye contact, and touching the partner several times." Other studies have utilized similar manipulations (e.g., Andersen et al., 1995; Burgoon & Hale, 1988; Burgoon

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

RESPONSES TO NONVERBAL INTIMACY CHANGE

333

& Le Poire, 1993; Le Poire & Burgoon, 1994; Guerrero & Burgoon, 1996; Manusov, 1995). Forty subjects were asked to decrease their intimacy so that the second conversation would be characterized by low intimacy (n = 20) or very low intimacy (n = 20). Those in the low intimacy condition were told to act "colder and more disinterested" toward their partners. We stressed that their behavior should be different in the second conversation than in the first conversation, and that they could accomplish this by "moving somewhat farther away from their partners, smiling less, using less eye contact, and acting somewhat bored." Those in the very low intimacy condition were told to make their behavior dramatically different by "moving far away from their partners, smiling a lot less, using a lot less eye contact, and acting extremely bored." The decision to have the confederates engage in the first conversation before receiving instructions regarding the manipulation helped ensure that they behaved naturally in the pre-manipulation conversation (rather than anticipating a behavioral change). Subjects in our experiment were asked to practice their change in intimacy behavior. None appeared to have difficulty acting highly intimate or nonintimate. After receiving instructions or completing the questionnaire (in the case of the control condition), confederates were escorted back to the main room. Couples were asked to find a new topic to use as a starting point for the second conversation. The researcher again knocked on the one-way mirror to signal the start of the interaction. After three minutes elapsed, the researchers stopped the interaction and once again separated the participants. This time, both targets and participants completed brief questionnaires asking about the interaction they had. Targets were then debriefed and were asked to not share the information they received about the experiment with others. Coded Behaviors

Nonverbal behaviors. Several measures of nonverbal involvement and involvementrelated behaviors were coded. Four undergraduate students previously enrolled in an upper-division nonverbal communication course and three graduate students in interpersonal communication were divided into pairs and trained to judge behaviors for each minute of the interaction (with the first 3 minutes representing the premanipulation conversation and the last 3 minutes representing the post manipulation conversation). After receiving several hours of training, the students worked independently. Facial pleasantness was coded from behind the one-way mirror as the actual interaction took place because of the difficulty of capturing these behaviors adequately on videotape within our laboratory setup. Because only one coder viewed all of the interactions from behind the one-way mirror, only her ratings of facial pleasantness are utilized in the analysis. However, a second coder was present to rate facial pleasantness for approximately 65% of the dyads. This second coder's ratings were used to establish interrater reliability although they are not incorporated into the analysis. All other behaviors were coded from videotapes by pairs of coders who judged one to two behaviors at a time. Eleven different indices of the targets' nonverbal involvement and affect behaviors were obtained. All measures were based on 7-point Likert-type or semantic differential scales developed by Coker and Burgoon (1987) and/or Guerrero (1997). Sample items and reliability estimates for these measures are found in Table 1.

334

COMMUNICATION MONOGRAPHS TABLE 1

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

SAMPLE ITEMS AND RELIABILITY ESTIMATES FOR CODED TARGET BEHAVIORS

Behavior

a

n

General Interest/Involvement (2 items) (e.g., bored/interested) Attentiveness (3 items) (e.g., Distracted/focused) Body Orientation (1 item) (e.g., indirect/direct) Facial pleasantness (2 items) (e.g., the target's face conveyed negative/positive affect) Vocal Pleasantness (3 items) (e.g., cold/warm) Kinesic Expressiveness (3 items) (e.g., nonexpressive/expressive) Vocal Expressiveness (4 items) (e.g., monotonous/animated) Forward Lean (2 items) (e.g., forward/backward) Fluency (2 items) (e.g., choppy/smooth) Vocal Relaxation (2 items) (e.g., cold/warm) Composure (3 items) (e.g., anxious/calm) Verbal Repair (1 item) (not at all/very frequently) Verbal Intimacy (1 item) (not at all/very frequently) Verbal Hostility (1 item) (not at all/very frequently)

.94

.86

.82

.83

-

.94

.87

.94

.93

.85

.95

.75

.92

.78

.80

.82

.98

.78

.90

.75

.94

.73

-

.72

-

.70

-

.68

Note. Interitem reliabilities are based on Cronbach's alpha statistic. Interrater reliabilities are based on Ebel's intra-class correlation.

Verbal behaviors. Two undergraduate students enrolled in an upper-division practicum course on research coded the verbal data independently after receiving several hours of training. Training included practice sessions during which coders evaluated verbal statements both from transcripts and videotapes, with exemplars of the different verbal categories purposely embedded in the transcripts. As with the nonverbal coding, these coders rated the two conversations separately. Using single-item 7-point scales that ranged from "not at all" to "very frequently," the coders estimated the extent to which targets engaged in verbal repair, verbal intimacy, and verbal hostility (see Table 1). Verbal repair was defined as statements or questions that sought to discover or repair problems and to otherwise "get the relationship back on track." Commonly occurring instances of verbal repair included asking the partner if something was wrong, apologizing about past behavior, and the trying to cheer up the partner. Verbal intimacy was defined as making positive statements about the relationship or the partner. This category also included positive self-disclosures and using personal nicknames. For example, one man kept referring to his girlfriend as "Honey," and another man told his girlfriend that he really cared for her and "hated it when they fought." Finally, verbal hostility was defined as making negative statements about the relationship or the partner, including accusations, criticisms, insults, threats, name-calling, and rude behavior. For example, one participant told her partner that he could "walk home" if he kept

RESPONSES TO NONVERBAL INTIMACY CHANGE

335

TABLE 2 MANIPULATION CHECK RESULTS OF CODED INVOLVEMENT ACROSS CONDITIONS

Pre-Manipulation

Post-Manipulation

Conditions

Mean

SD

Mean

SD

Very Low Intimacy Low Intimacy Very High Intimacy High Intimacy Control

4.60 4.83 4.80 4.40 4.70

.77 .48 .65 .66 .61

2.25 3.10 6.30 5.86 4.83

.58 .89 .45 .49 .62

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

Note. Higher scores represent more involvement.

acting like a "jerk" and ignoring her. Another participant said he was sick of "playing relationship games." Manipulation check. Five items from Burgoon and Hales's (1987) immediacy/ affection subscale were used to rate the confederates' behavior (e.g., "Person A communicated warmth rather than coldness," "Person A was intensely involved in the conversation"). This measure was utilized as a manipulation check. Interitem reliability was .93; interrater reliability was .87. Results Manipulation Check

To check the success of the intimacy change manipulation, coders rated how involved confederates were during the pre- and post-manipulation conversations. A 5 (involvement condition) X 2 (pre- vs. post-manipulation conversation) mixed model analysis of variance, with condition as a between-subjects variable and conversation as a within-subjects variable, showed that the manipulation was generally successful. The condition by manipulation effect was significant, F(4,95) — 159.12, p < .001, tf = .80, showing that involvement level change was dependent upon condition. Mean scores (see Table 2) indicate that confederates in the two low conditions decreased their involvement substantially during the second conversation, confederates in the two high conditions increased their involvement substantially, and confederates in the control condition showed only a slight increase in involvement. To further probe the significant interaction, we set contrast codes at 1 and — 1 for the two conversations and then set up three orthogonal contrasts for condition. The first contrast, which pitted the two increase intimacy conditions against the two decrease intimacy conditions, was significant, t(95) = 24.90,/» < .001, -n2 = .61. This contrast confirms that those in the increase intimacy conditions showed more involvement in the post-manipulation conversation, whereas those in the decrease intimacy conditions showed less involvement. The second contrast, which compared those in the low intimacy condition against those in the very low intimacy condition, was also significant, t(95) = 3.14, p < .01, TI2 = .02, although it accounted for only a very small portion of variance. Nonetheless, this contrast demonstrates that those in the very low intimacy condition displayed a larger decrease in their involvement levels than did those in the low involvement conditions. Finally, the contrast that compared those in the very high intimacy condition to those in the high

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

336

COMMUNICATION MONOGRAPHS

intimacy condition was nonsignificant, t(95) = .05, p < .05. Although means indicate that those in the very high intimacy condition displayed more involvement during the post-manipulation conversation than did those in the high condition, the increases in involvement were similar across both conditions. Targets in the high intimacy condition started with a lower level of involvement than those in the very high intimacy condition. Therefore, targets in both the very high and high intimacy conditions showed nearly a 1.5 increase in their involvement score in the postmanipulation conversation. From the above results, we can conclude that our manipulation was successful in distinguishing the two low intimacy groups from the two high intimacy groups. It was also successful in distinguishing the very low and low intimacy groups. However, it was not successful in distinguishing the very high and high intimacy groups, possibly because the high intimacy group differed from the other groups at the outset. Indeed, an additional contrast showed that targets in the high intimacy group displayed significantly less involvement during the pre-manipulation conversation than did the other four groups (very high, low, very low, and control), ¿(95) = 2.56, p < .01, ir2 = .06. Thus, we can conclude that although three of the four experimental groups were similar to the control group at the outset of the experiment, the high intimacy group was dissimilar. This caveat to the otherwise successful manipulation will be kept in mind when interpreting results connected to the high intimacy condition. It is noteworthy, however, that despite their relatively low starting points, those in the high intimacy condition were able to increase their involvement levels substantially in the post-manipulation conversation. In fact, this group had more "room" to maneuver in increasing their involvement level than did the very high intimacy group. Tests of Hypotheses and the Research Question

The two hypotheses focused on how targets responded to their partner's change in intimacy. The general prediction was that targets' responses would vary by condition across the two conversations, with nonverbal and verbal intimacy levels showing similarity in the pre-manipulation conversations and then diverging in the post-manipulation conversations. Thus, we expected conversation (pre- vs. postmanipulation) to interact with condition (very low intimacy, low intimacy, very high intimacy, high intimacy, control). This was tested with a series of mixed model ANOVAs and MANOVAs, with condition serving as a between-subjects variable and conversation serving as a within-subjects variable. As discussed in the method section, 11 different indices of nonverbal behavior were rated. In our analyses, nonverbal behaviors that were conceptually related and showed significant intercorrelation were treated as composite variables in multivariate analyses. Those behaviors that were not significantly correlated with other variables were treated as single dependent measures in ANOVAs. The three verbal measures were only slightly correlated. Thus, these three measures were also treated as separate dependent measures in ANOVAs. Power for detecting significant differences among the cell means was approximately .25 for small effects (r\2 = .10), .66 for moderate effects (•n2 = .25), and in excess of .95 for large effects (TI2 = .50). Regardless of whether we analyzed the data with MANOVAs or ANOVAs, we applied orthogonal contrasts to the interaction tests to examine the predictions made in our hypotheses. The first interaction contrast compares the two high intimacy

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

RESPONSES TO NONVERBAL INTIMACY CHANGE

337

conditions to the two low intimacy conditions, with coefficients of 1 and — 1 applied to the increase and decrease intimacy conditions, respectively. The control condition was assigned a coefficient of 0. This comparison is relevant to HI, which predicts that those in the high and very high intimacy conditions will reciprocate their partners' increased intimacy, and to H2, which predicts that targets in the low and very low intimacy conditions will show a mix of reciprocity and compensation when verbal and nonverbal responses are considered. With these contrast codes applied, a conversation by condition interaction would show that targets in the high intimacy conditions increased intimacy after the manipulation, whereas targets in the low intimacy conditions decreased intimacy after the manipulation. This would be evidence for reciprocity across all four conditions. If, instead, a main effect showed that intimacy increased from the pre- to post-manipulations conversations across all four conditions, the data would suggest that targets in the high intimacy conditions reciprocated, whereas those in the low intimacy conditions compensated. Similarly, a main effect showing decreased intimacy across all four conditions would show that targets in the low intimacy conditions reciprocated, whereas those in the high intimacy conditions compensated. Two additional orthogonal contrasts will be used to test the research question. These interaction tests will compare the low and very low intimacy groups (RQla) and the high and very high intimacy groups (RQlb). For RQla, the interaction contrasts will be set at — 1 for the high intimacy condition, and 1 for the very high intimacy condition, with the other three conditions all assigned 0. For RQlb, the interaction contrasts will be set at — 1 for the low intimacy condition, and 1 for the very high intimacy condition. Again, the remaining three conditions will all be assigned 0. Because the analyses are fairly extensive, only significant findings are reported at the multivariate and contrast levels. General interest/involvement, attentiveness, and body orientation. General interest/

involvement, attentiveness, and body orientation were analyzed together (average r = .38, Bartlett's test of sphericity, 2 df = 87.96, p < .001). A MANO VA revealed a significant condition by conversation interaction, Approximate -F(12, 246) = 4.40, p < .001, Wilks A = .60. The accompanying univariate analyses showed that the interaction was significant for general interest/involvement, F(4, 95) = 8.88, p < .001, TI2 = .27, but not for attentiveness, F{4,95) = .29, p > .05, or body orientation, () Focused contrasts on die interaction means showed that targets in the two high intimacy conditions increased their display of general interest/involvement during the post-manipulation conversation, whereas those in the two low intimacy conditions decreased their display of general interest/involvement during the postmanipulation conversation, t[95) = 5.11, p < .001, -n2 = .22 (see Table 3). The control group showed little change from the pre- to post-manipulation conversations. This finding supports HI by showing that those in the high intimacy conditions reciprocated. In answer to H2, targets reciprocated rather than compensated with behavior reflecting interest/involvement. In response to RQlb, the focused contrast comparing the very low intimacy group with the low intimacy group was significant, ¿(95) = 2.71, p < .01, T|2 = .07, showing that those in the low intimacy group decreased their involvement/interest more than did diose in the very low intimacy group.

338

COMMUNICATION MONOGRAPHS TABLE 3 MEANS AND STANDARD DEVIATIONS FOR CODED VARIABLES

Very Low Intimacy Variables

Low Intimacy

Very High Intimacy

High Intimacy

Control

M(SD)

M(SD)

M(SD)

M(SD)

M(SD)

4.91 (.92) 4.38 (1.01)

4.78 (.97) 4.52 (1.03)

4.62 (.95) 5.20 (.82)

4.81 (.87) 5.10 (.89)

4.77(1.12) 4.82(1.10)

5.16 (.98) 5.16(1.01)

5.00 (.93) 4.98 (1.00)

5.08 (.96) 5.19 (.93)

4.95 (1.06) 5.10 (.98)

4.89 (.98) 4.88(1.06)

4.41 (1.89) 4.40 (1.87)

4.93 (2.03) 4.91 (2.14)

3.65 (2.36) 3.55 (2.23)

3.38 (1.92) 3.63 (1.94)

4.88(1.95) 5.02 (2.10)

4.43 (.97) 3.46(1.11)

4.63 (.63) 3.93 (.63)

4.57 (1.02) 4.89(1.06)

4.27 (.71) 4.46 (.96)

4.18 (.68) 4.39 (.74)

4.23 (.98) 3.87 (.98)

4.71 (.74) 4.51 (.86)

4.39 (.84) 4.77 (.95)

4.06 (.87) 4.28 (1.01)

4.35 (.86) 4.43 (.80)

3.03 (.91) 3.04 (1.01)

2.68 (.54) 2.63 (.54)

2.71 (.80) 3.11 (.71)

2.93 (.94) 3.07 (.76)

2.93 (1.15) 2.94 (.99)

4.12 (.82) 4.27 (.84)

4.06 (.70) 3.93 (.72)

4.07 (.97) 4.55 (.87)

4.04 (.85) 4.32 (1.06)

4.50 (.98) 4.61 (.93)

4.13 (.75) 4.39 (.74)

4.20 (.59) 4.31 (.70)

4.14 (.54) 4.34 (.69)

4.25 (.82) 4.32 (.95)

4.22 (.71) 4.27 (.76)

4.44 (.62) 3.50 (.95)

4.76 (.58) 4.10 (.79)

4.61 (.73) 4.85 (.66)

4.61 (.52) 4.57 (.77)

4.91 (.98) 4.91 (.91)

3.65 (.90) 3.02 (.60)

4.05 (1.00) 3.93 (1.08)

3.87 (1.03) 3.75 (.82)

3.83 (1.05) 3.93 (1.21)

4.16 (.68) 4.01 (.78)

4.38 (1.02) 3.88(1.14)

4.36 (.81) 4.51 (.77)

4.34 (.91) 4.19 (.98)

4.48 (.91) 4.52 (1.15)

4.82 (.58) 4.86 (.71)

1.48 (.30) 4.28 (1.23)

1.32 (.34) 3.10 (1.42)

1.30 (.30) 1.42 (.71)

1.50 (.56) 1.58 (.52)

1.38 (.51) 1.30 (.47)

4.90 (.88) 3.93 (1.48)

5.00 (.74) 3.75 (1.22)

4.52 (.62) 5.40 (.90)

4.81 (.92) 5.65 (.88)

4.73 (.79) 4.78 (.69)

1.35 (.40) 3.15 (1.32)

1.35 (.33) 3.24 (1.05)

1.50 (.74) 1.50 (1.06)

1.68 (.83) 1.35 (.52

1.55 (.81) 1.43 (.57)

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

30

Interest/Involvement ' Pre-Manipulation Post-Manipulation Attentiveness Pre-Manipulation Post-Manipulation Body Orientation Pre-Manipulation Post-Manipulation Facial Pleasantness* Pre-Manipulation Post-Manipulation Vocal Pleasantness*1 Pre-Manipulation Post-Manipulation Kinesic Expressiveness Pre-Manipulation Post-Manipulation Vocal Expressiveness Pre-Manipulation Post-Manipulation Forward Lean b Pre-Manipulation Post-Manipulation Fluency11 Pre-Manipulation Post-Manipulation Vocal Relaxation0 Pre-Manipulation Post-Manipulation Composure 0 Pre-Manipulation Post-Manipulation Verbal Repair"1 Pre-Manipulation Post-Manipulation Verbal Intimacy* Pre-Manipulation Post-Manipulation Verbal Hostility"1 Pre-Manipulation Post-Manipulation

Note, a = interaction contrasts showed that targets in the two high intimacy conditions increased involvement after the manipulation, whereas targets in the two low intimacy conditions decreased involvement, b = a main effect showed that targets in all conditions increased forward lean after the manipulation, c = contrasts showed that targets in the low and very low intimacy conditions varied across the pre- and post-manipulation conversations, d = interaction contrasts showed that targets in the two low intimacy conditions increased their use of this verbal strategy, whereas those in the two high intimacy conditions and the control condition used less or about the same amount of this verbal strategy.

Facial and vocal pleasantness. Facial and vocal pleasantness were initially analyzed together in a MANO VA (1) r = .25, Bartlett's test of sphericity, 1 df = 5.91, p < .05 (1). A significant multivariate interaction between condition and conversation emerged, F(8, 186) = 6.39, p < .001, Wilks A = .62. Accompanying univariate

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

RESPONSES TO NONVERBAL INTIMACY CHANGE

339

analyses showed significant interaction effects for both facial pleasantness, F(4,94) = 11.73,/>< .001, ti2 = .33; and vocal pleasantness, F{4,94) = 6.55,p< .001, if = .22. Contrasts demonstrated that targets in the two low intimacy conditions became less facially pleasant after the manipulation, whereas those in the two high intimacy conditions became more facially pleasant, ¿(94) = 6.21, p < .001, T|2 = .29. Interaction means (see Table 3) suggest that those in the low intimacy conditions showed fairly substantial decreases in facial pleasantness. Those in the high intimacy conditions, in contrast, showed smaller increases in facial pleasantness. In fact, the increase shown by those in the high intimacy condition were not much different from the increase shown in the control condition. For vocal pleasantness, the contrast focusing on the interaction between the high versus low intimacy conditions across the pre- vs. post-manipulation conversations was also significant, ¿(94) = 4.81, j& < .001, TI2 = .20. Means suggest that those in the high intimacy conditions became more vocally pleasant compared to those in the low intimacy conditions, who became less vocally pleasant (see Table 3). The means also suggest that targets in the control condition showed an increase in vocal pleasantness in the second conversation, but to a lesser extent than did those in the high intimacy conditions. A post-hoc contrast of interaction means comparing the two high intimacy conditions (coefficients set at 1) with the control condition (coefficient set at —2) confirmed that targets in the intimacy change conditions showed a larger increase in vocal pleasantness than did those in the control condition, i(94) = 2.32, p < .05, -n2 = .05. Thus, the pattern of means generally supports HI by showing that targets in the high intimacy conditions tended to reciprocate facial and vocal pleasantness. In response to H2, these data suggest that targets in the low intimacy conditions also reciprocated by showing less facial and vocal pleasantness in the post-manipulation conversation. In response to the research question, the interaction contrasts did not reveal significant differences across the high versus very high conditions or the low versus very low conditions, respectively. Kinesic and vocal expressiveness. Kinesic and vocal expressiveness were analyzed together (r = .49, Bartlett's test of sphericity, 1 df = 25.99, p < .001). There were no significant findings connected to these variables. Forward lean. An ANO VA obtained a significant main effect for conversation, F(l, 95) = 4.39, p < .05, t]2 = .04. Targets showed more forward lean (M= 4.31, SD = .68) in the pre- versus post-manipulation conversation (M= 4.19, SD = .76). Inspection of the cell means (see Table 3) suggests that targets in all 5 conditions increased their forward lean after the manipulation, with the most change evident for those in the very low and very high conditions, and the least change evident for those in the low and control conditions. However, the interaction contrasts did not find any significant differences between conditions, which suggests that people increased forward lean regardless of condition, and that the main effect for forward lean may be more a function of time than partner behavior. Fluency. An ANOVA showed a significant conversation by condition interaction for fluency, F(4, 95) = 8.70, p< .001, -rf = .27. Focused contrasts on the interaction means showed that those in the two low intimacy conditions exhibited less fluency after the manipulation, whereas those in the two high intimacy conditions exhibited either more or similar levels of fluency after the manipulation (see Table 3), ¿(95) = 5.37, p < .001, -n2 = .23. Thus, the data on fluency provide some support for HI by

340

COMMUNICATION MONOGRAPHS

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

showing that targets in the high intimacy conditions increased fluency after the manipulation, albeit slightly. In response to H2, targets in the low intimacy conditions showed reciprocity (and perhaps increased arousal) by becoming less fluent over time. Vocal relaxation. An ANOVA found a significant conversation by condition interaction for vocal relaxation, F{4, 95) = 2.75, p < .05, t\2 = .10. The interaction contrast comparing the two low intimacy conditions with the two high intimacy conditions was not significant, ¿(95) = .85, p > .05. Thus, HI was not supported, in that reciprocal effects were not found for vocal relaxation. (In fact, vocal relaxation decreased in the very high intimacy condition.) A second, orthogonal interaction contrast revealed that those in the very low condition differed from those in the low condition, ¿(95) = 2.16, p < .05, -n2 = .05. These findings make sense when the means (see Table 3) are inspected. Targets in the very low intimacy condition showed a substantial drop in vocal relaxation (with an average change of .628 less vocal relaxation), whereas the other three intimacy change groups showed little change (with an average change of .113). A post-hoc interaction contrast (with coefficients set at —4 for the very low intimacy condition and 1 for the remaining conditions) confirmed that targets in the very low intimacy condition differed from targets in the other four groups, ¿(95) = 2.91, p < .01, ty2 = .08. Composure. An ANOVA found a significant conversation by condition interaction for composure, F[4, 95) = 2.62, p < .05, T\2 = .10. A contrast showed that this interaction is primarily attributable to the difference between the low and very low conditions across the two conversations. Specifically, targets in the very low intimacy condition displayed less composure after the manipulation, whereas those in the low intimacy condition displayed more composure, ¿(94) = 2.90, p < .01, T|2 = .08 (see Table 3), which could be construed as compensation. Thus, composure was not reciprocated in the high intimacy conditions as predicted in HI (in fact, composure decreased in the very high intimacy condition). However, in response to RQ1, there was a significant difference between the low and very low intimacy conditions. Verbal responses. The ANOVA on verbal repair showed significant main effects for condition, F{4,95) = 28.19, p < .001, and time, F[l, 95) = 98.58, p < .001. However, these main effects were overridden by the interaction between conversation and condition, F{4, 95) = 36.81, p < .001, T\2 = .61. Two of the accompanying contrasts were also significant. The interaction contrast comparing the two low conditions against the two high conditions accounted for the most variance, ¿(95) = 10.33, p < .001, f]2 = .52. As shown in Table 3, the means suggest that targets in the two low intimacy conditions used more verbal repair following the manipulation, whereas those in the two high intimacy conditions (as well as the control condition) showed little change. The contrast comparing the low intimacy and very low intimacy conditions was also significant, ¿(95) = 3.42, p = .001, -n2 = .11, with means suggesting that targets in the very low intimacy condition were especially likely to use verbal repair strategies. Thus, this analysis provides some support for H2's prediction that some compensatory verbal responses follow decreases in nonverbal intimacy. In the ANOVA on verbal intimacy, the conversation by condition interaction obtained significance, F[4, 95) = 21.68, p < .001, t\2 = .38. A contrast showed that the divergent pattern for the two low conditions versus the two high conditions was

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

RESPONSES TO NONVERBAL INTIMACY CHANGE

341

primary responsible for this effect, ¿(95) = 7.05, p < .001, T|2 = .34. Means show that targets in the low and very low conditions decreased their levels of verbal intimacy after the manipulation, whereas those in the high and very high conditions increased verbal intimacy. This analysis supports Hi's prediction that increases in nonverbal intimacy are met with reciprocity. Relevant to H2, this analysis shows that decreases in nonverbal intimacy are also met with reciprocity. Finally, the ANOVA for verbal hostility revealed significant main effects for condition, F{4,95) = 25.02,p < .001, and time, F{\, 95) = 50.21, p < .001. However, these main effects were overridden by the condition by conversation interaction, F{4, 95) = 28.95, p < .001, -n2 = .44. The contrast comparing the cell means for the two low conditions against the cell means for the two high conditions was significant, ¿(95) = 7.82, p < .001, iq2 = .39. Means showed that targets in the very low and low conditions become more verbally hostile after the manipulation, whereas those in the very high and high conditions (as well as the control condition) stayed the same or become less verbally hostile. Thus, this analysis shows that targets in the low intimacy conditions reciprocated by using more verbal hostility. Discussion Two primary goals guided the present investigation. First, we sought to examine behavioral responses to intimacy change across five different conditions-very low intimacy, low intimacy, very high intimacy, high intimacy, and a no change (control) condition-within romantic dyads. Given the inconsistent findings produced by past studies investigating romantic dyads, coupled with the lack of studies examining multiple intimacy conditions along with a control condition, such a focus was a logical next step in this research tradition. Second, we hoped our data would provide a more complete account of the adaptation process by investigating both verbal and nonverbal responses to changes in nonverbal intimacy. Next, we discuss the present experimental data as it relates to these two goals. Verbal and Nonverbal Responses to Intimacy Change

Consistent with our first hypothesis, all of the significant effects associated with the two high intimacy conditions showed reciprocity. This finding is consistent with EVT, which predicts that reciprocity will be the dominant pattern when rewarding communicators increase intimacy. Nonverbally, targets in the two high intimacy conditions exhibited more general interest/involvement, facial pleasantness, and vocal pleasantness after the manipulation, as evidenced by the interaction contrast comparing the two high intimacy groups with the two low intimacy groups across the two conversations. However, targets in the two high intimacy conditions failed to show reciprocity on several other nonverbal behaviors, including attentiveness, kinesic and vocal expressiveness, and composure. As several studies have suggested (e.g., Burgoon, Le Poire, & Rosenthal, 1995; Guerrero & Burgoon, 1996), people may be most likely to reciprocate behavior related to global involvement and affect. In the present study, targets in the high and very high intimacy conditions also showed reciprocity by increasing their level of verbal intimacy, which was operationalized as including positive statements about the partner and the relationship, including positive self-disclosure and the use of personal nicknames. The same interaction effects that showed reciprocity in the high intimacy conditions showed that reciprocity was also the dominant pattern in response to decreased

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

342

COMMUNICATION MONOGRAPHS

intimacy. Specifically, targets in the low intimacy conditions became less nonverbally interested/involved, less facially and vocally pleasant, and less fluent after the manipulation. Verbally, these targets also showed less intimacy and more hostility following the manipulation, which again provides evidence for a reciprocity effect. These findings are similar to past results on responses to decreases in nonverbal intimacy. Guerrero and Burgoon (1996) found that individuals whose romantic partners decreased intimacy reciprocated by exhibiting less interest and affection, less pleasantness, less expressiveness, and less fluency and composure. An additional study that focused on romantic dyads (Manusov, 1996) found targets to reciprocate decreased intimacy by becoming less involved and active, and by exhibiting more negative affect. Similar to Le Poire and Yoshimura's findings (1999), targets in the present study maintained previous levels of kinesic and vocal expressiveness after their partners decreased intimacy. Thus, as with responses to increases in nonverbal intimacy, targets seem particularly likely to reciprocate intimacy decreases by engaging in behavior that shows less general involvement and more negative affect, but levels of expressiveness are unlikely to change. The literature on nonverbal adaptation and conflict suggests at least two reasons why reciprocity may prevail as a response to decreases in intimacy. The first of these explanations is rooted in theories such as DAT, which predict that substantially low levels of intimacy that are discrepant from expectations lead to arousal and anxiety. According to this reasoning, people may spontaneously show their negative affect (through behaviors such as decreased facial and vocal pleasantness as well as increased verbal hostility) and anxiety (through behaviors such as decreased fluency). As DAT suggests, behavioral decreases in intimacy are likely to fall outside of one's region of acceptance, which is likely to cause arousal and anxiety. In this study, targets in the very low intimacy condition also showed a decrease in composure, suggesting that they were uncomfortable with such a low level of intimacy. Targets in the low intimacy condition, however, seemed able to show more confidence and composure, perhaps as a way of compensating for their partner's decrease in intimacy by appearing calm and poised rather than upset. In line with DAT, small to moderate decreases in intimacy may not be very arousing or anxiety-provoking, making it possible to manage impressions and appear self-confident and relatively unconcerned. The second explanation stems from research on conflict interaction, which shows that people often have a hard time engaging in positive, compensatory actions when their partners are behaving negatively. Burgraff and Sillars (1987), for example, found that husbands' and wives' conflict styles were highly reciprocal. If one partner used a confrontational or avoidant strategy, the other person was likely to follow suit (see also, Pike & Sillars, 1985). Similarly, Rusbult and her colleagues (Rusbult, Drigotas, & Verette, 1994; Rusbult, Johnson, & Morrow, 1986) have argued that compensating for negative behavior is a "costly or effortful reaction" in the short term, because compensating "rather than retaliating when a partner behaves poorly is not the individual's immediate response inclination," even though compensation is in the best long-term interest of the relationship (Rusbult et al., 1994, pp. 126-127). However, some studies, including the present study, have found a limited number of compensatory reactions to decreases in nonverbal intimacy. For example, Burgoon, Stern, and Dillman (1995) analyzed data showing that people in an increase intimacy condition tended to show a fairly consistent pattern of reciprocity across a

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

RESPONSES TO NONVERBAL INTIMACY CHANGE

343

number of different nonverbal measures, whereas people in a decrease intimacy condition tended to show a mix of reciprocity and compensation (by showing less positive affect, vocal expressiveness, and vocal relaxation, but becoming more involved and kinesically active). Similarly, Guerrero and Burgoon (1996) found a consistent pattern of reciprocation for intimacy increases across measures of involvement, affect, expressiveness, and composure. Yet they found a mix of reciprocity and compensation in response to intimacy decreases. Specifically, targets in the decrease intimacy condition displayed less smiling, facial pleasantness, vocal pleasantness, and relaxed laughter (reciprocity), but more forward lean, direct body orientation, and attention (compensation). In the present study, targets in the low intimacy conditions displayed compensation through verbal repair and, possibly, through forward lean. In other words, targets in the low intimacy change conditions tended to ask their partners if something was wrong, if they were mad at them, or if they could do something to make them feel better. Targets were particularly likely to engage in verbal repair in the very low intimacy condition. Interestingly, however, targets in the two low intimacy conditions also engaged in more verbal hostility. These findings paint a picture of targets as both accommodating and retaliatory. They may feel competing needs to restore the relationship to a more intimate state and to defend themselves by attacking the partner. Perhaps targets try to accommodate, but find it difficult to refrain from making hostile comments if the partner continues to act unfriendly. Such an explanation fits with Rusbult et al.'s (1986, 1994) work on conflict interaction, which suggests that although people may try to be accommodating in the face of unfriendly behavior, they experience a natural pull toward retaliation. Consistent with this line of reasoning, the findings from the present study suggest that when compensation for decreased intimacy does occur, it is embedded within a larger pattern of reciprocity. More research is needed to determine the circumstances under which decreases in intimacy are met with some compensatory responses in addition to reciprocity. Even more importantly, this study suggests that targets compensate using specific strategies designed to repair their relationships. If scholars only look at traditional verbal and nonverbal immediacy cues, they may miss these potentially important behaviors. The findings from this study also suggest that the degree to which people decrease nonverbal intimacy makes a difference in some cases, albeit a small difference. Four findings relevant to the comparison between the low and very low intimacy conditions emerged. First, targets in the very low intimacy condition showed a greater decrease in nonverbal interest/involvement than did those in the low intimacy group. Second, targets in the very low group also showed a larger drop in vocal relaxation than did those in the low intimacy group. Third, targets in the very low intimacy condition showed the largest decrease in composure of all five groups. Surprisingly, interaction means showed that targets in the low condition displayed slightly more composure after the manipulation. As proponents of DAT have claimed, dramatic decreases in nonverbal intimacy are more likely than moderate decreases to lead to avoidance and negative behavior. Finally, targets in the very low intimacy condition were more likely to engage in verbal repair than those in the low intimacy condition. It is likely that the very low intimacy condition was perceived as the most threatening, and as such, required the most repair. Taken together, these findings suggest that targets in the very low intimacy condition became more

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

344

COMMUNICATION MONOGRAPHS

anxious and attempted to alleviate some of their anxiety by trying to determine what was wrong so they could repair the situation. Unfortunately, we failed to find any differences between the high and very high intimacy conditions. This is likely a function of both the failure of our manipulation and the difficulty that lies in creating strong expectancy violations when increasing intimacy in romantic relationships. Our manipulation check showed that the difference between the high and very high intimacy conditions was nonsignificant. Therefore, we were unlikely to find different responses to the behavioral displays of targets in these two conditions. Although our inability to create distinct manipulations for the high and very high intimacy conditions might be considered a weakness in our study, we believe that our failure is partially symptomatic of the way nonverbal intimacy is displayed in close relationships. Specifically, it may be difficult to increase nonverbal intimacy enough to violate expectations in romantic relationships. Finally, it is important to note that the effect sizes associated with the verbal variables were particularly strong. The condition by conversation interaction accounted for between 38% and 61% of the variance in the verbal responses studied, with the largest effect size associated with verbal repair. Verbal repair appears to be a particularly useful strategy when people are faced with a unfriendly partner. In some ways it is unsurprising that the condition by conversation interaction accounted for a substantial portion of the variance in verbal repair. After all, targets in the increase intimacy conditions would have little need for such a strategy. Nonetheless, the robust effect associated with this finding suggests that verbal repair constitutes an important option for coping with decreases in intimacy-an option that has until now remained unexplored. Similarly, past studies have rarely looked at verbal intimacy or hostility as reactions to changes in nonverbal intimacy, although this study suggests that both of these general categories of verbal behavior are important. Perhaps the robust effects found for the verbal measures are in part due to the global, inclusive nature of these measures. With the exception of the general measure of nonverbal involvement/interest, the nonverbal cues were measured at a fairly micro level. Still, the effect size associated with the more global measure of nonverbal involvement/interest was smaller than those for the verbal measures, which reinforces the important role that verbal communication plays in the adaptation process. Of course, some of the nonverbal measures also produced fairly strong effect sizes. In particular, the condition by conversation interaction accounted for 33% of the variance in facial pleasantness, 27% of the variance in fluency, and 22% of the variance in vocal pleasantness. These results suggest that, nonverbally, romantic partners respond to changes in intimacy primarily through behaviors indicative of affect and fluency, rather than through more traditional immediacy cues such as forward lean and direct body orientation. Conclusion

The present study adds to the considerable literature on patterns of reciprocity and compensation of nonverbal intimacy behavior. By focusing on romantic dyads, utilizing four intimacy conditions plus a control, and including verbal as well as nonverbal responses to intimacy change, this study extends past research. Our data demonstrate that targets generally showed reciprocal response patterns following

RESPONSES TO NONVERBAL INTIMACY CHANGE

345

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

both increases and decreases in intimacy by their romantic partners, although those in the two low intimacy conditions also tended to use verbal repair, which should be considered a compensatory move. There were also important differences between the low and very low intimacy conditions, with targets in the very low condition showing a decrease in composure, a greater decrease in fluency and vocal relaxation, and more verbal repair. These findings affirm that behavioral valence and degree of behavioral change are important components within theories of nonverbal adaptation. Data from this study also suggest that scholars must consider both verbal and nonverbal responses to intimacy change if they are to truly understand how the adaptation process works.

References Afifi, W.A., & Burgoon, J . K (1996, November). Behavioral violations in interactions: The combined consequences of valence and change in uncertainty on interaction outcomes. Paper presented at the annual meeting of the Speech Communication Association, San Diego, CA. Afifi, W.A., & Metts, S. (1998). Characteristics and consequences of expectation violations in close relationships. Journal of Social and Personal Relationships, 15, 365-392. Andersen, P.A. (1983, May). Nonverbal immediacy in interpersonal communication. Paper presented to the annual meeting of the International Communication Association, Dallas, TX. Andersen, P.A. (1984, April). An arousal-valence model of nonverbal immediacy exchange. Paper presented to the annual meeting of the Central States Speech Association, Chicago, IL. Andersen, P.A. (1985). Nonverbal immediacy in interpersonal communication. In A.W. Siegman & S. Feldstein (Eds.), Multichannel integrations of nonverbal behavior (pp. 1-36). Hillsdale, NJ: Lawrence Erlbaum. Andersen, P.A. (1998). The cognitive valence theory of intimate communication. In M. Palmer & G.A. Barnett (Eds.), Progress in communication sciences, Vol 14. Mutual influence in interpersonal communication theory and research in cognition, affect, and behavior. Norwood, NJ: Ablex. Andersen, P.A., & Andersen,J.F. (1984). The exchange models of nonverbal immediacy: A critical review of dyadic models. Journal of Nonverbal Behavior, 8, 327-349. Andersen, P.A., & Guerrero, L.K. (1998). Principles of communication and emotion in social interaction. In P.A. Andersen & L.K. Guerrero (Eds.), Handbook of communication and emotion: Theory, research, contexts, and applications (pp. 49-56). San Diego, CA: Academic Press. Andersen, P.A., Guerrero, L.K., Buller, D.B., &Jorgensen, P.F. (1998). An empirical comparison of three theories of nonverbal immediacy exchange. Human Communication Research, 24, 501-535. Argyle, M., & Cook, M. (1976). Gaze and mutual gaze. Cambridge, UK: Cambridge University Press. Argyle, M., & Dean J . (1965). Eye-contact, distance, and affiliation. Sociometry, 28, 289-304. Burgoon, J.K. (1978). A communication model of personal space violations. Explications and initial test. Human Communication Research, 4, 129-142. Burgoon, J . K (1983). Nonverbal violations of expectations. In J.M. Wiemann & R.P. Harrison (Eds.), Nonverbal interaction (pp. 11-77). Beverly Hills, CA: Sage. Burgoon, J . K (1994). Nonverbal signals. In M.L. Knapp & G.R. Miller (Eds.), Handbook of interpersonal communication. (2nd ed., pp. 229-285). Thousand Oaks, CA: Sage. Burgoon, J . K , & Dillman, L., & Stern, L.A. (1993). Adaptation in dyadic interaction: Defining and operationalizing patterns of reciprocity and compensation. Communication Theory, 4, 293-316. Burgoon, J.K., & Hale, J.L. (1987). Validation and measurement of the fundamental themes of relational communication. Communication Monographs, 54, 19-41. Burgoon, J.K, & Hale, J.L. (1988). Nonverbal expectancy violations: Model elaboration and application to immediacy behaviors. Communication Monographs, 55, 58-79. Burgoon, J.K., & Le Poire, B.A. (1993). Effects of communication expectancies, actual communication, and expectancy disconfirmation on evaluations of communicators and their communication behavior. Human Communication Research, 20, 67-96. Burgoon, J . K , Le Poire, B.A., & Rosenthal, R. (1995). Effects of preinteraction expectancies and target communication on perceiver reciprocity and compensation in dyadic interaction. Journal of Experimental Social Psychology, 31, 287-321. Burgoon, J.K, Newton, D.A., Walther, J.B., & Baesler, EJ. (1989). Nonverbal expectancy violations and conversational involvement. Journal of Nonverbal Behavior, 13, 97-120. Burgoon, J.K., Stern, L.A., & Dillman, L. (1995). Interpersonal adaptation: Dyadic interaction patterns. Cambridge, UK: Cambridge University Press. Burgoon, J.K, & Walther, J.B. (1990). Nonverbal expectancies and the evaluative consequences of violations. Human Communication Research, 17, 232-265.

Downloaded by [University of Minnesota Libraries, Twin Cities] at 06:07 03 May 2013

346

COMMUNICATION MONOGRAPHS

Burgraff, C.S., & Sillars, A.L. (1987). A critical examination of sex differences in marital communication. Communication Monographs, 54, 276-294. Canary, D.J., Cupach, W.R., & Messman, S.J. (1995). Relational conflict. Thousand Oaks, CA: Sage. Cappella, J.N., & Greene, J.O. (1982). A discrepancy-arousal explanation of mutual influence in expressive behavior for adult and infant—adult interaction. Communication Monographs, 49, 89-114. Cappella, J.N., & Greene, J.O. (1984). The effects of distance and individual differences in arousability on nonverbal involvement: A test of discrepancy-arousal theory. Journal of Nonverbal Behavior, 8, 259-286. Coker, D.A., & Burgoon, J.K. (1987). The nature of conversational involvement and nonverbal encoding patterns. Human Communication Research, 13, 463-494. Floyd, K., & Voloudakis, M. (1999). Affectionate behavior in adult platonic friendships: Interpreting and evaluating expectancy violations. Human Communication Research, 25, 341-369. Guerrero, L.K. (1997). Nonverbal involvement across interactions with same-sex friends, opposite-sex friends, and romantic partners: Consistency or change? Journal of Social and Personal Relationships, 14, 31-58. Guerrero, L.K., & Andersen, P.A. (2000). Emotion in close relationships. In C. Hendrick & S.S. Hendrick (Eds.), Close relationships: A sourcebook (pp. 171-183). Thousand Oaks, CA: Sage. Guerrero, L.K., & Burgoon,J.K. (1996). Attachment styles and reactions to nonverbal involvement change in romantic dyads: Patterns of reciprocity and compensation. Human Communication Research, 22, 335-370. Hale, J.L., & Burgoon,J.K. (1984). Models of reactions to changes in nonverbal immediacy. Journal of Nonverbal Behavior, 8, 287-314. Le Poire, B.A., & Burgoon, J.K. (1993). Participant and observer perceptions of relational message interpretations associated with nonverbal involvement and pleasantness. Unpublished manuscript, University of California, Santa Barbara. Le Poire, B.A., & Burgoon, J.K. (1994). Two contrasting explanations of involvement violations: Expectancy violations theory versus discrepancy arousal theory. Human Communication Research, 20, 560-591. Le Poire, B.A., & Yoshimura, S.M. (1999). The effects of expectancies and actual communication on nonverbal adaptation and communication outcomes: A test of interaction adaptation theory. Communication Monographs, 66, 1-30. Manusov, V. (1995). Reacting to changes in nonverbal behaviors: Relational satisfaction and adaptation patterns in romantic dyads. Human Communication Research, 21, 456-477. Patterson, M.L. (1982). A sequential-functional model of nonverbal exchange. Psychological Review, 83, 235-245. Patterson, M.L. (1983). Nonverbal behavior: A functional perspective. New York: Springer Verlag. Pike, G.R., & Sillars, A.L. (1985). Reciprocity of marital communication. Journal of Social and Personal Relationships, 2, 303-324. Rusbult, C.E., Drigotas, S.M., & Verette, J. (1994). The investment model: An interdependence analysis of commitment processes and relationship maintenance phenomena. In DJ. Canary & L. Stafford (Eds.), Communication and relational maintenance (pp. 115-139). San Diego, CA: Academic Press. Rusbult, C.E., Johnson, D.J., & Morrow, G.D. (1986). Impact of couple patterns of problem solving on distress and nondistress in dating relationships. Journal of Personality and Social Psychology, 43, 1230-1242. Received: March 16, 1999 Accepted: September 12, 2000