Language Style Matching in Romantic Partners’ Conflict and Support Interactions

Abstract

This study examined the association between language style matching (LSM), subjective perceptions of interaction quality (perceived responsiveness and affect), and partner behavior in two communication contexts: conflict and social support. Romantic couples (N = 91) engaged in a video-recorded discussion of either a relationship stressor or one partner’s personal stressor (a social support discussion). LSM was associated with unique outcomes in each communication setting. Higher LSM was associated with lower subjective perceptions of responsiveness and less positive emotion for partners discussing relationship stressors but more positive emotion for partners in social support discussions. Furthermore, higher LSM was associated with more critical and negative interpersonal behavior and less responsive and caring behavior during discussions of relationship stressors but was unrelated to behavior in support discussions. Findings suggest that LSM does not uniformly signal interpersonal rapport (as often assumed) and may instead amplify the positive or negative tone of an interaction.

Keywords

conflict couples language style matching LSM mimicry responsiveness romantic relationships social support stress

Romantic partners who communicate effectively with one another have relationships that are more satisfying, committed, and emotionally intimate (e.g., Eldridge & Christiansen, 2002; Reis & Shaver, 1988; Sprecher, Metts, Burleson, Hatfield, & Thompson, 1995). Through both verbal and nonverbal channels, partners can share thoughts, feelings, and goals, providing one another with privileged access to their internal states. While some aspects of communication are intentional and controlled, others occur automatically. For example, communication is full of coordination phenomena, many of which occur outside conscious awareness (see Giles & Coupland, 1991; Giles, Coupland, & Coupland, 1991). Individuals can accommodate to one another across numerous communicative domains, and they do so to achieve not only greater communicative efficiency but also a desired level of social closeness or distance between themselves and their interaction partner(s).

Behavioral (nonverbal) mimicry is one such phenomenon, in which individuals unwittingly copy one another’s actions. Behavioral mimicry has been documented in gaze (e.g., Richardson & Dale, 2005), posture (e.g., LaFrance, 1982), facial expression (e.g., Bavelas, Black, Lemery, & Mullett, 1986), and motor movements (e.g., Bailenson & Yee, 2007; Chartrand & Bargh, 1999; Webb, Eves, & Smith, 2011). The underlying function of behavioral mimicry has been an issue of some debate, but a large body of research suggests it serves as an unconscious affiliation strategy (see Chartrand & Bargh, 1999). Mimicry communicates a desire for acceptance, group inclusion, and interpersonal harmony (Lakin & Chartrand, 2003; Lakin, Chartrand, & Arkin, 2008; Lakin, Jefferis, Cheng, & Chartrand, 2003; Yabar, Johnston, Miles, & Peace, 2006).

In recent years, social psychological research on mimicry has extended beyond nonverbal behavior to examine verbal coordination in interpersonal interaction. Such innovation has been bolstered by the advent of word-level analytical tools such as the Linguistic Inquiry and Word Count software (Pennebaker, Booth, & Francis, 2007; Tausczik & Pennebaker, 2010). One form of verbal mimicry, known as language style matching (LSM), takes place when communicators pattern certain types ofwords in similar ways. LSM is a dyad-level metric of convergence in the rates at which two communicators use “function words” (Ireland & Pennebaker, 2010). Function words are a class of words that are chiefly associated with grammatical structure as opposed to language content, and include auxiliary verbs (e.g., “could,” “has”), impersonal pronouns (e.g., “that,” “which”), and conjunctions (e.g., “then,” “or”). Because language provides a concrete structural framework within which to organize thoughts and feelings, convergence in the use of function words is thought to reflect communicators’ shared cognitive representations of whatever they are discussing (Niederhoffer & Pennebaker, 2002). Interaction partners who are higher in LSM are using function words at similar rates, whether frequently or infrequently.

Function words can be reliably varied without changing the overall meaning of a message, which is why they are characterized as markers of language style. For example, a person can ask another to accompany him or her to dinner by phrasing his or her request as “Would you like to join me for dinner?” He or she could also communicate this request with the phrasing “How about we have dinner later tonight?” The differences in these two requests may seem trivial, but are likely reflective of meaningful psychological variability. In fact, they are due almost entirely to differences in function words. For example, the use of the pronouns “you” and “me” in the former request is substituted by the pronoun “we” in the latter request. This seemingly minor difference may actually signal a great deal about the requestor’s assumptions regarding the pair’s relationship.

Variability in function word use is linked to a number of individual differences, including stress reactivity, gender, age, deceptive tendencies, depression, and cultural background (Chung & Pennebaker, 2007). The coordinated use of function words has been understood as a word-level manifestation of communication accommodation (Giles & Coupland, 1991; Niederhoffer & Pennebaker, 2002), promoting smoother interactions and negotiating social distance between interactants. It has also been described as an instance of structural priming (e.g., Pickering & Ferreira, 2008; Pickering & Garrod, 2004), wherein the syntactic patterns in one interactant’s utterances are automatically and unconsciously taken up in the other interactant’s utterances.

LSM has been particularly revealing in the domain of romantic relationships, highlighting the importance of coordinated verbal communication for relationship outcomes. Ireland and Pennebaker (2010) examined LSM in spouses’ writings over several decades and found that LSM was greatest during times of joy (relative to dissatisfaction) in their marriages. This finding suggests that LSM might track relationship quality over time. In another study, Ireland et al. (2011, Study 1) found that speed daters were more likely to be a “match” (i.e., choose one another for a future date) if there was greater LSM during their brief conversations. This finding suggests that LSM might signal mutual romantic interest or attraction. Finally, in a study of established dating couples, Ireland et al. (2011, Study 2) further found that couples with the lowest levels of LSM during instant messaging exchanges were most likely to separate 3 months later. This suggests that decreases in LSM might forecast relationship difficulties. While these findings do not directly address the causal pathway between LSM and relationship outcomes, they show that LSM varies reliably with the emotional or motivational state of the dyad.

While the aforementioned studies seem to suggest that verbal mimicry, like behavioral mimicry, may foster or reflect interpersonal affiliation or rapport, other evidence suggests that the implications of LSM for interpersonal functioning may be more complicated. Niederhoffer and Pennebaker (2002) suggested that LSM may be better conceptualized in terms of coordinated engagement, rather than in terms of rapport. Specifically, they proposed that communicators should be higher in LSM to the extent that they are actively participating in the interaction at hand and are converging in their mental states. Importantly, this can be true whether the content of an interaction is benign or destructive.

While converging mental states (and the associated matching of language) may at times coincide with positive feelings and prosocial motivations, this may not always be the case. This is especially true for romantic partners, who must have a wide array of discussions, ranging from planning a much-anticipated vacation to a heated discussion about spending money. In positive interactions, LSM may indeed foster or reflect positive feelings, cooperation, and rapport. However, in conflictual interactions, LSM may instead reflect or amplify negative feelings, discordant goals, and hostility.

Consistent with this idea, Gottman and colleagues (Gottman, 1979, 1980; Gottman, Markman, & Notarius, 1977) found that spouses in conflict often exhibited a damaging form of coordination known as negative affect reciprocity, matching one another’s raised voices, combative postures, and angry facial expressions. Research on emotion and mood contagion also suggests that both positive and negative feelings can transfer between communicators through the unconscious perception and mimicry of affective states (Hatfield, Cacioppo, & Rapson, 1993; Neumann & Strack, 2000). In addition, a large body of research rooted in communication accommodation theory (see Giles, 2016) suggests that coordination during interaction is quite sensitive to context and is managed by interactants based on their social goals and interpersonal orientations. It is therefore reasonable to assume that LSM may not always reflect or facilitate interpersonal rapport. Recent work by Ireland and Henderson (2014) provides preliminary evidence for this assumption. Stranger dyads were asked to have a discussion over instant message and to reach agreement on four issues in 20 minutes. Prior to their discussions, participants were given an egoistic, self-serving motive: They were asked to focus solely on individual interests during negotiation. This study revealed that higher levels of LSM were associated with greater reported social engagement but also with lower levels of agreement. Thus, in an explicitly competitive context, higher levels of LSM appeared to reflect a shared unwillingness to compromise, which resulted in a greater likelihood of impasse during negotiations.

LSM appears therefore to be a more nuanced coordination phenomenon, sensitive to communicators’ relationships, goals, and the communication context. Because function word matching signals convergence in communicators’ mental representations of an interaction, LSM during an aversive discussion may co-occur with destructive behavior and drive negative relational outcomes. While earlier LSM research suggested that LSM might simply index affiliation, these studies failed to examine contexts in which interaction partners had incompatible or conflicting aims. The one study that examined competitive interactions (Ireland & Henderson, 2014) found evidence that LSM did not simply foster rapport. To date, no research has systematically manipulated the interaction context to determine whether LSM is associated with different interaction behaviors and outcomes in distinct conversation contexts.

Thus, in the current investigation, we explored LSM in couples in two different conversation contexts: discussions of relationship stressors (i.e., conflicts) and personal stressors (i.e., social support discussions). We video recorded these conversations for linguistic and behavioral analyses and examined the association between LSM and two important markers of interaction quality: partners’ subjective evaluations of the interaction (indexed by self-reported positive and negative affect and perceptions of partner responsiveness), and objective judges’ ratings of partners’ caring and destructive behavior during the interaction. We chose these two markers of interaction quality because a great deal of research has documented that positive outcomes do not simply reflect the absence of negative outcomes, and vice versa (Lyubomirsky, Sheldon, & Schkade, 2005). In addition, positive and negative affect are largely orthogonal in the emotion literature (e.g., Watson, Clark, & Tellegen, 1988), and positive and negative partner behavior independently predict long-term relationship outcomes (e.g., Fincham & Linfield, 1997; Kiecolt-Glaser et al., 1993).

We selected responsiveness as our positive behavioral outcome because it is a major organizing construct in relationship psychology (Reis, Clark, & Holmes, 2004). Perceived partner responsiveness plays a critical role in effective social support transactions (e.g., Collins & Feeney, 2000; Maisel & Gable, 2009) and is a key predictor of intimacy and satisfaction in close relationships (e.g., Laurenceau, Feldman Barrett, & Pietromonaco, 1998). We selected negativity/criticism as our negative behavioral outcome because previous work on marital conflict has identified blame, hostile framing of questions and statements, and implicit and explicit pressures to change a partner’s behavior as important features of destructive communication behavior (e.g., Heavey, Gill, & Christensen, 2002; Pasch & Bradbury, 1998).

To our knowledge, the current study is the first to examine links between LSM and real-time interpersonal behavior patterns. Previous investigations of LSM have focused on distal outcomes such as relationship initiation or dissolution (Ireland et al., 2011) or the ultimate success of negotiations and collaborative work efforts (Gonzales, Hancock, & Pennebaker, 2010; Richardson, Taylor, Snook, Conchie, & Bennell, 2014; Taylor & Thomas, 2008). This investigation, by contrast, sought to uncover the broader patterns of behavior that co-occur alongside LSM during dyadic interaction. In doing so, we drew from the literatures on social support behavior (e.g., Collins & Feeney, 2000; Maisel, Gable, & Strachman, 2008) and relationship conflict negotiation (e.g., Heavey et al., 2002). As discussed, we focused on two global aspects of behavior that would be applicable to both types of interactions: interpersonal responsiveness (expressions of understanding, validation, and caring; Reis & Shaver, 1988) and negativity/criticism.

We predicted that LSM would be associated with more positive (and fewer negative) self-reports and behaviors in social support conversations but with more negative (and fewer positive) self-reports and behaviors in conflicts. Social support exchanges are, typically, cooperative contexts in which partners have an opportunity to share personal worries and express caring and concern. The underlying motivational context is, therefore, relatively benign and altruistic. The compassion and responsiveness that underlie effective support transactions allow partners to attune to each other’s needs and provide care that is matched to those needs (Reis & Shaver, 1988). As such, engagement and convergence of mental states, reflected in LSM, should facilitate effective support provision and positively influence partners’ affective states and perceptions of interpersonal responsiveness. Indeed, recent research suggests that increases in LSM are associated with perceptions of emotional support in computer-mediated contexts (Rains, 2015) and increased use of cognitive reappraisal strategies to improve mood during discussions of personal problems (Cannava & Bodie, 2016). In conflict interactions, by contrast, partners are frequently ideologically or motivationally opposed. In these contexts, engagement and convergence of mental states may instead reflect an alignment in egoistic motivations, discordant interpretations of the interaction, or negative emotional states (see Ireland & Henderson, 2014).

In summary, we predicted that the associations between LSM and conversation behavior and self-reported outcomes would be moderated by context (see Figure 1). We expected that LSM would predict greater responsive and caring behavior in support discussions but more unresponsive and critical behavior in conflict discussions. We further expected that greater LSM would be associated with more positive affect, less negative affect, and greater perceived responsiveness by partners in social support conversations but less positive affect, more negative affect, and lower perceived responsiveness by partners in conflict conversations.

Figure 1.

Theoretical model depicting proposed relationship between language style matching, conversation context, and partners’ subjective experiences and behavior.

Method

Participants

Ninety-one couples were recruited from an undergraduate psychology subject pool. Participants’ ages ranged from 17 to 23 years (M = 19.3, SD = 1.3), and they had been dating from 3 to 63 months (M = 14.3, SD = 13.7). The sample included five same-sex couples.¹ Fifty percent of participants self-identified as White/Anglo, 19% as Latino/Hispanic, 19% as Asian/Pacific Islander, 3% as Black/African American, 1% as American Indian, and 7% as “Other.” Participants received research credit or $5 for participating. The data from one couple were excluded due to a computer malfunction that precluded transcription of their conversation.

Procedure and Measures

Couples were randomly assigned to either the conflict or social support condition. Within condition, partners were randomly and unwittingly assigned to roles of discloser or respondent. Prior to the lab session, partners completed a battery of self-report measures, including the Relationship Assessment Scale (Hendrick, 1988), a seven-item measure of relationship satisfaction (α = .79).

At the lab, partners were informed that they would participate in brief, videotaped conversations. They were then separated to complete a preinteraction measure of that day’s mood, modified from Watson et al.’s (1988) Positive and Negative Affect Schedule Scale. We computed an index of positive affect (the sum of 10 items, α = .88) and negative affect (the sum of 15 items, α = .88). Next, partners completed a form asking them to list three sources of relationship stress and one personal stressor (external to their relationship) that they would feel comfortable discussing in the lab.

In the conflict condition, the partners were asked to discuss one of the relationship stressors nominated by the discloser.² Discussions of relationship stressors served as our conflict condition because, in these instances, partners were asked to converse about “. . . areas of conflict, disagreement, or tension . . .” in their relationships. The purpose of our conflict condition was to create a situation in which partners would experience a conflict of interest with one another—in which one partner’s personal goals or needs were incompatible with those of the other partner or the relationship overall (e.g., Kelley & Thibaut, 1978; Rusbult & Buunk, 1993). Successful navigation of such conflicts of interest is key to the maintenance of stable and satisfying long-term relationships (e.g., Rusbult & Arriaga, 1997). We also selected this operationalization of “conflict” because prior research on marital conflict (e.g., Pasch & Bradbury, 1998) has involved partners selecting from a list of common sources of couple discord and distress for a subsequent discussion (Geiss & O’Leary, 1981), similar to our own relational stressors. In operationalizing our conflict condition in this way, we feel that it is important to note the distinction between overt conflicts, which may occur to varying degrees when communicating about disagreement, and conflicts of interest, the motivational and relational context in which we sought to embed our participants.

In the support condition, partners were asked to discuss the personal stressor nominated by the discloser.³ Discussions of personal stressors served as our social support condition because conversing about sources of individual distress (that are unrelated to the partner or the relationship) provides opportunities for that partner to be supportive. Of course, partners will vary in the degree to which they provide social support in response to their partner’s stress disclosure (e.g., Collins et al., 2014). The support condition was designed to provide an interaction context in which the seeking and giving of support were normative and likely to occur. Furthermore, there is a precedent in the marital conflict literature for operationalizing discussions of personal areas of challenge and needed growth as social support contexts (e.g., Pasch & Bradbury, 1998). Our decision to compare conflict and social support contexts follows from previous research (Pasch & Bradbury, 1998), and as operationalized here, the two conversation types serve as good comparison contexts because in both cases a stressor is discussed.

Following the nomination of discussion topics, partners were reunited, seated in a comfortable room with couches and nonintrusive video cameras, and instructed to discuss their assigned topic for 5 minutes. Those in the conflict condition tended to discuss disclosers’ concerns about lack of intimacy or time spent together, partner behavior of which the discloser did not approve, lack of communication, and challenges to their future together. Some specific topics included “going out (partying, etc.) without the other,” “lack of communication,” “not talking often/not being available,” and “what we’re going to do after he graduates.” Those in the support condition tended to discuss disclosers’ worries about their academic performance, life after graduation, and struggles with family. Some specific topics included “trying to get good grades in school,” “family conflicts,” “homework workload,” and “focusing on what I need to do to get into med school.”

Following the conversation, partners completed postinteraction affect ratings using the same positive (α = .87) and negative (α = .84) emotions as before. Partners were asked to rate each affect item in terms of how they felt during the interaction. Partners also completed two items measuring perceived partner responsiveness. They rated how “supported” they felt and how “cared for” they felt during the interaction (r = .70).⁴

Finally, partners had a 3-minute conversation about a positive topic (e.g., their first date) designed to alleviate any potential distress from the earlier interaction. Partners were then carefully debriefed, thanked, and dismissed.

Linguistic Analyses

LSM is a dyad-level variable that takes input from each partner’s use of function words. Scores range from zero to one, with greater values reflecting greater similarity in the use of function words. Nine function word subcategories are used to compute LSM: auxiliary verbs, conjunctions, personal pronouns, impersonal pronouns, adverbs, negations, prepositions, articles, and quantifiers (Ireland & Pennebaker, 2010). To prepare the data for analysis, conversations were transcribed and split into two files, one containing each partner’s utterances. The Linguistic Inquiry and Word Count software (Pennebaker et al., 2007) was then used to compute use rates of each function word category for each partner (p₁, p₂). Results were then entered into the following well-validated formula (Ireland & Pennebaker, 2010):

1 - \frac{| {preps}_{p 1} - {preps}_{p 2} |}{{preps}_{p 1} + {preps}_{p 2} + . 0001} .

The above formula results in a score for prepositions. The total LSM score is the average of the values produced by this formula across all nine function word subcategories.

Behavior Coding

Three trained raters (two female undergraduates, one male undergraduate) coded each videotaped interaction for each partners’ responsive and caring behavior, and negative or critical behavior. Coding schemes were adapted from Maisel et al.’s (2008) Responsive Behaviors Coding Guide and Heavey et al.’s (2002) Couples Interaction Rating System. All dimensions were rated from 1 (not at all) to 7 (a great deal). Intraclass correlations (ICCs) were computed to assess interrater reliability. Coders were instructed to read and familiarize themselves with the descriptions of each behavioral coding category (see the appendix). In coding the videotapes, coders were instructed to watch several interactions, to first establish an understanding of the range of the relevant behaviors. Then, they watched each video twice, assigning codes to a single partner after the second viewing. Following protocols adapted from Maisel et al. (2008) and Heavey et al. (2002), coders were instructed to consider the extent to which broad, gestalt classes of behavior were present or absent for a given partner, guided by the presence of “microanalytic” cues (e.g., nodding or backchanneling [“uh-huh”] to signal understanding).

To assess responsiveness, raters coded partners’ behavior on three dimensions. Understanding referred to efforts to gather information and accurately comprehend a partner’s thoughts, feelings, and concerns (ICC_supp = .75, ICC_conf = .70). Validation referred to efforts to demonstrate appreciation for a partner’s perspective (ICC_supp = .66, ICC_conf = .79). Caring referred to partners’ expressions of love, affection, compassion, or emotional support (ICC_supp = .65, ICC_conf = .72). An overall Behavioral Responsiveness score was computed as the average of these three dimensions within each rater (ICC_supp = .75, ICC_conf = .81), and then averaged across raters.

Negative Behavior was computed from a single dimension of criticism/negativity (ICC_supp = .82, ICC_conf = .81). This behavior referred to evidence that a partner was demeaning, teasing, or attacking the character of the other. Negative Behavior scores were moderately positively skewed (overt negativity was relatively uncommon), so they were log-transformed for analyses.

Results

Data Analytic Strategy

Because LSM is a dyad-level variable and partners were nonindependent by virtue of interacting with one another, we conducted dyadic analyses using multilevel modeling (Kenny, Kashy, & Cook, 2006) with Hierarchical Linear and Nonlinear Modeling Software Version 7.01 (Raudenbush, Bryk, Cheong, Congdon, & du Toit, 2011). This analytic strategy allowed us to examine both individual-level and dyad-level variables in our analyses. Correlations between our study variables are presented in Table 1.

Table 1.

Correlations Between Study Variables.

Variable	1	2	3	4	5	6	7	8
1. Language style matching	—
2. Condition (conflict = 0)	−.04	—
3. Role (discloser = 0)	.00	.00	—
4. Post positive affect	.08	−.10	.06	—
5. Post negative affect	−.04	−.01	.05	−.16*	—
6. Perceived partner responsiveness	−.02	.05	−.08	.55***	−.29***	—
7. Responsive behavior	−.06	.15^†	.05	.23**	−.15*	.36***	—
8. Negative behavior	.20**	−.24**	−.03	−.10	.14^†	−.34***	−.72***	—

†

p < .12. *p < .05. **p < .01. ***p < .001.

For each dependent variable, we tested three sequential models. In Model 1, we tested the main effects of LSM, conversation condition (0 = conflict, 1 = support), and role (0 = discloser, 1 = respondent). In Model 2, we added the two-way LSM × condition interaction. This model also controlled for the LSM × role and condition × role interactions. In Model 3, we added the three-way LSM × condition × role interaction. The full set of analyses is presented in Table 2. To allow for interpretation of effect sizes, z scores were computed for all variables used in our analyses. These z scores were then used in an identical set of analyses as those reported in Table 2. The results of these analyses are presented in Table 3, in which all regression coefficients have been standardized.

Table 2.

Multilevel Analyses of LSM and Related Variables.

	Post positive affect	Post negative affect	Perceived partner responsiveness	Responsive behavior	Negative behavior
Model 1
Level 1
Intercept	19.44	8.50	2.85	3.97	0.31
Satisfaction	—	—	2.75***	1.72*	−0.10
Pre positive affect	0.62***	—	—	—	—
Pre negative affect	—	0.38***	—	—	—
Role (discloser = 0)	1.20	0.75	−0.23^†	0.06	−0.01
Total word count	0.01*	0.00	0.00	0.00***	0.00
Level 2
LSM	2.78	0.71	−1.64	−4.04	1.29*
Condition (conflict = 0)	−1.29	−0.17	0.14	0.42^†	−0.12**
Model 2
LSM × condition	29.19	10.10	5.26	8.62^†	−2.58*
LSM × role	3.68	−8.84	2.24	−0.28	0.29
Role × condition	3.96*	−4.36*	−0.41^†	0.21	0.01
Model 3
LSM × condition × role	−143.52***	57.41	−13.47*	3.89	−0.98

Note. LSM = language style matching. All coefficients are unstandardized.

†

p < .12. *p < .05. **p < .01. ***p < .001.

Table 3.

Multilevel Analyses of LSM and Related Variables (All Variables Standardized).

	Post positive affect	Post negative affect	Perceived partner responsiveness	Responsive behavior	Negative behavior
Model 1
Level 1
Satisfaction	—	—	0.32***	0.18*	−0.05
Pre positive affect	0.58***	—	—	—	—
Pre negative affect	—	0.34***	—	—	—
Role (discloser = 0)	0.16	0.09	−0.22^†	0.05	−0.49
Total word count	0.15*	−0.09	0.10	0.30***	0.00
Level 2
LSM	0.01	0.00	−0.06	−0.14	0.22*
Condition (conflict = 0)	−0.17	−0.02	0.13	0.34^†	−0.49**
LSM × condition	0.11	0.03	0.14	0.19^†	−0.30*
Model 2
LSM × role	0.02	−0.05	0.09	−0.01	0.05
Role × condition	0.52*	−0.55*	−0.38^†	0.17	0.02
Model 3
LSM × condition × role	−0.52***	0.20	−0.35*	0.09	−0.11

Note. LSM = language style matching. All coefficients are standardized.

†

p < .12. *p < .05. **p < .01. ***p < .001.

All models controlled for conversation length (total word count), which was correlated with LSM (r = .25, p < .01). Models predicting behavioral responsiveness, negative behavior, and perceived responsiveness controlled for overall relationship satisfaction. Satisfaction scores were moderately negatively skewed, so they were square root transformed for these analyses. Models predicting postinteraction positive and negative affect controlled for preinteraction positive and negative affect, respectively. Role, satisfaction, and preinteraction affect were modeled at Level 1, as they varied across partners within couples. LSM, conversation type, and total word count were modeled at Level 2, as they varied between couples. Continuous predictors at all levels were grand-mean centered.

Preliminary Analyses

Mean levels of LSM did not differ between the conflict (M = .87, SD = .04) and support (M = .87, SD = .04) conversations, t(178) = −0.51, p = ns. LSM was also not associated with either partner’s relationship satisfaction (r_respondent = .04, p = ns, r_discloser = .10, p = ns).⁵

Partners’ Affective Experiences and Perceived Partner Responsiveness

We hypothesized that LSM would be associated with more positive and less negative subjective experiences for partners in the support condition but with less positive and more negative subjective experiences for partners in the conflict condition. We conducted multilevel regression analyses predicting partners’ positive affect, negative affect, and perceived partner responsiveness.

Positive Affect

In our main effects model (see Model 1 in Table 2) predicting postinteraction positive affect (controlling for preinteraction positive affect), we found no effect of LSM or conversation condition. Our two-way interaction model also did not reveal the predicted LSM × condition interaction. However, Model 3 revealed a significant three-way LSM × condition × role interaction, b = −143.52, t(85) = −3.87, p < .001. Follow-up simple slopes analyses indicated that among disclosers, LSM was significantly negatively associated with positive affect in conflict conversations, b = −45.88, t(85) = −2.21, p < .05, but significantly positively associated with positive affect in support conversations, b = 55.05, t(85) = 2.52, p < .05.

Among respondents, LSM was not significantly associated with positive affect in either the conflict, b = 22.98, t(85) = 1.10, p = ns, or support, b = −19.60, t(85) = −0.90, p = ns, conversations (see Figure 2). Overall, these findings were in partial support of our hypotheses. Partners in the discloser role (but not those in the respondent role) experienced more positive affect if their support conversations were higher in LSM but less positive affect if their conflict conversations were higher in LSM. This suggests that in a linguistically coordinated interaction, the process of sharing a personal stressor with a loved one may augment positive feeling, whereas the process of sharing a relationship stressor with that person may reduce positive feeling.

Figure 2.

The effect of condition on disclosers’ and respondents’ postinteraction positive affect at low and high levels of language style matching.

Negative Affect

In our main effects model (Model 1, Table 2) predicting postinteraction negative affect (controlling for preinteraction negative affect), we found no effect of LSM or conversation condition. Our subsequent two-way and three-way interaction models (see Models 2 and 3 in Table 2) also revealed no interactions of LSM with condition and/or role, suggesting that LSM by and large was not associated with partners’ negative feelings during the conversations.

Perceived Partner Responsiveness

In our main effects model predicting perceived partner responsiveness (controlling for relationship satisfaction; Model 1, Table 2), we found no main effects of LSM or conversation condition. In Model 2, we found no significant two-way interactions. However, Model 3 revealed a significant LSM × condition × role effect on perceived responsiveness, b = −13.47, t(85) = −2.32, p < .05. Follow-up simple slopes analyses revealed that, for disclosers, LSM was negatively associated with perceived responsiveness in conflict conversations, b = −8.39, t(85) = −2.33, p < .05, but unrelated to perceived responsiveness in the support conversations (though the trend was in the positive direction, b = 3.60, t(85) = 0.96, p = .34. For respondents, LSM was not related to perceptions of responsiveness in either the conflict, b = −0.04, t(85) = −0.01, p = ns, or support, b = −1.52, t(85) = −0.40, p = ns, conversations (see Figure 3).

Figure 3.

The effect of condition on disclosers’ and respondents’ ratings of perceived partner responsiveness (felt care and support) at low and high levels of language style matching.

Here, as with positive affect, we found partial support for our hypotheses. Disclosers felt less cared for and supported in conflict when their conversations were higher in LSM and felt slightly more cared for and supported in support conversations if those conversations were higher in LSM (though again, this pattern was not significant). It again appears that when disclosing a relationship stressor, a linguistically coordinated interaction may be aversive to the discloser; but when sharing a personal stressor, a linguistically coordinated interaction may be beneficial to the discloser.

Partners’ Responsive and Negative Behavior

As with self-reports, we hypothesized that LSM would be associated with more positive and less negative behavior (as coded by trained raters) in the support condition but with less positive and more negative behavior in the conflict condition.

Responsive Behavior

Our main effects model predicting responsive behavior (controlling for relationship satisfaction; Model 1, Table 2) revealed a marginal main effect of condition, b = 0.42, t(85) = 1.97, p = .05, indicating that partners in the support condition behaved more responsively than did those in the conflict condition. In our two-way interaction model (Model 2, Table 2), we found a trending LSM × condition effect on responsive behavior, b = 8.62, t(84) = 1.62, p = .11. Follow-up simple slopes analyses indicated that, as predicted, partners behaved less responsively if their conflict interactions were higher in LSM, b = −8.11, t(84) = −2.03, p < .05. However, there was no relationship between LSM and responsive behavior in support conversations, b = 0.51, t(84) = 0.13, p = ns (see Figure 4). Model 3 revealed no significant three-way interaction, b = 3.89, t(84) = 0.66, p = ns.

Figure 4.

The effect of condition on responsive (understanding, validating, caring) behavior at low and high levels of language style matching.

Negative Behavior

Our main effects model predicting negative behavior (controlling for relationship satisfaction) revealed main effects of both LSM, b = 1.29, t(85) = 2.40, p < .05, and conversation condition, b = −.12, t(85) = −2.71, p < .01; greater LSM was associated with more negative behavior overall, and there was more negative behavior in conflict (vs. support) conversations (Model 1, Table 2). Model 2 revealed that these main effects were qualified by a significant LSM × condition interaction, b = −2.58, t(84) = −2.44, p < .05. Follow-up simple slopes analyses indicated, as predicted, that greater LSM was associated with significantly more negative behavior in conflict interactions, b = 2.40, t(84) = 3.07, p < .01. In support interactions, LSM was not significantly associated with negative behavior, b = −0.17, t(84) = −0.22, p = ns (see Figure 5). Finally, Model 3 revealed no LSM × conversation × role interaction predicting negative behavior, b = −0.98, t(84) = −0.90, p = ns. As a complement to the findings for responsive behavior, these findings suggest that both partners behaved more negatively when engaged in a linguistically coordinated discussion of a relationship stressor.⁶

Figure 5.

The effect of condition on negative (unresponsive, critical) behavior at low and high levels of language style matching.

Discussion

This study provides clear evidence that the association between LSM and conversation behaviors and outcomes in couples depends heavily on the social and emotional context of the interaction. Although LSM did not differ on average between conflict and support conversations, the implications of LSM were substantially distinct in each context. In conflict discussions, couples who were higher in LSM had disclosers who felt less positive emotion during their conversation and felt less supported and cared for. In support discussions, by contrast, couples who were higher in LSM had disclosers who felt more positive emotion and felt slightly more responded to by their partners. Moreover, in conflict, objective coders rated the behavior of both partners to be less responsive and more negative as LSM increased.

Insofar as LSM reflects social engagement and converging mental states, the lack of a mean difference in LSM between the two contexts suggests that they did not differ in these important aspects of communication. However, the implications of social engagement and shared mental states did appear to differ between contexts. In conflict, LSM may reflect a shared combative attitude (e.g., shared motivation to defend or promote one’s own position or point of view). In conflict conversations, greater LSM was associated with lower positive affect and perceptions of responsiveness among disclosers, and with more negative (and somewhat less responsive) partner behavior. It may be that in this context, LSM amplified the already negative tone of these interactions.

In social support conversations, we had predicted that greater engagement and mental state convergence, reflected in higher LSM, would facilitate the provision of responsive support. We found some evidence for this prediction in that higher levels of LSM were associated with more positive emotional experiences for disclosers, the recipients of support in these contexts. However, the links between LSM and behavior and other self-reported outcomes were relatively weak in the social support condition. One reason may be that social support interactions, although intended to be positive (to inspire benevolent and caring motivation and behavior), are not uniformly pleasant. Because couples were discussing one partner’s significant personal stressor, these conversations may still have been stressful, particularly for the discloser.

As the first investigation of LSM to directly manipulate context, our findings provide further support for Niederhoffer and Pennebaker’s (2002; see also Ireland & Henderson, 2014) argument that LSM is not necessarily a signal of affiliation or rapport but rather a context-sensitive marker of interactional engagement that may have positive or negative implications in different social contexts. Our findings also contribute to our understanding of how LSM might uniquely reflect interaction dynamics for interactants who have existing relationships with one another. Much research has examined LSM in interactions between strangers (e.g., Ireland et al., 2011), and strangers who exhibit higher LSM in interaction may be motivated to reach jointly satisfying outcomes (see Ireland & Henderson, 2014, for an exception). Individuals with considerable relationship history, however, may engage with one another during interaction to satisfy a host of motives, many of which are shaped by these people’s proprietary knowledge of one another and past shared experiences. Given that partners produce a great deal of language together, LSM is an extremely valuable metric for assessing how engaged they are in the wide range of interactions that they have. The current study suggests that the nature of these interactions (e.g., emotional or motivational tone, combative vs. cooperative) is key in determining what LSM reflects for these partners.

Our study also offers important contributions to the research literature on interpersonal responsiveness. Previous work has been devoted to understanding the consequences of greater responsiveness (especially perceptions of responsiveness; Collins & Feeney, 2000; Maisel & Gable, 2009; Reis et al., 2004), but relatively less work has explored the factors that drive responsiveness and the ways in which it can manifest in meaningful social situations. Our study identified that (higher or lower) LSM can be a relatively unbiased, context-sensitive, interaction-based predictor of responsiveness. This finding is particularly important because the key role of responsiveness in promoting intimacy and trust between partners (Laurenceau et al., 1998; Reis et al., 2004) is necessarily realized over time across accumulated interactions.

By revealing that LSM may be one mechanism through which responsiveness can be communicated, our study highlighted a critical nuanced relationship between linguistic synchrony and partners’ overall experience in emotionally involving interactions. Furthermore, because LSM is a dyadic predictor of responsiveness, we importantly demonstrated that one’s perceptions of responsiveness from a partner can be influenced not only by one’s partner’s contributions to a conversation but also by one’s own. This suggests that responsiveness is facilitated by how each partner behaves not only independently but also interactively. A responsive partner may be not so much a person who consistently behaves a certain way but rather someone who flexibly calibrates (even unconsciously) his or her interactional activity to our own.

We believe that this study also contributes to our understanding of LSM in relationships by suggesting avenues for future work and intervention. Existing interventions, such as those designed to help interaction partners redistribute their attention in real-time interaction settings (e.g., Tausczik & Pennebaker, 2013), could be tailored for relationship partners, encouraging them to engage differently with one another to promote more mutually beneficial interaction outcomes. Practitioners in couple and marital therapy could also be informed by considering LSM data in interactions they observe in their treatment of partners. Indeed, some emerging evidence suggests that LSM is associated with increased empathy in interactions between therapists and clients (Lord, Sheng, Imel, Baer, & Atkins, 2015). Finally, we believe that by demonstrating the context sensitivity of LSM, this study suggests new avenues for experimental research, perhaps through manipulation of partners’ mindsets or goal orientations prior to interaction, to examine whether these intrapsychic factors interact with LSM to predict outcomes.

Limitations and Future Directions

The current study has several limitations worth noting. First, because we did not manipulate LSM, we cannot determine if LSM is a cause, consequence, or reflection of engagement and interpersonal responsiveness. For example, our findings suggest that it may sometimes be beneficial to disengage linguistically from an interaction. Such a claim is consistent with communication accommodation theory’s findings suggesting that accommodating to an interaction partner can involve diverging from the partner’s communication style (which is sometimes even misperceived as convergence; Thakerar, Giles, & Cheshire, 1982) and also that convergence and divergence can occur simultaneously across different modalities (e.g., Gallois, Ogay, & Giles, 2005).

This finding is also consistent with research on behavioral mimicry demonstrating that social coordination is more efficient insofar as individuals’ expectations for mimicry are met (as opposed to simply having a greater degree of mimicry; Dalton, Chartrand, & Finkel, 2010). However, it is not yet clear from our data whether disengagement in terms of function word use would directly influence partners’ feelings or behaviors. Nevertheless, our results clearly indicate that LSM does not uniformly signal rapport. Additional research will be needed to replicate our findings and to identify the specific mechanisms linking LSM to positive and negative interaction outcomes. The current research offers some initial steps in this direction and points to exciting avenues for future research. We believe that experimental research will be especially valuable. Future research should strive to directly manipulate LSM between communicators in order to draw more confident inferences about causal mechanisms. Although it may be difficult to experimentally control LSM within natural conversations, recent research (discussed previously) suggests that LSM can be made to fluctuate through real-time feedback during conversation (Tausczik & Pennebaker, 2013).

Another limitation of the current study concerns the two conversation contexts. Whereas LSM was associated with more detrimental self-reports and behavior in conflict discussions, it was not consistently associated with self-reports and behavior in support discussions. As mentioned, this was likely due to the juxtaposition of the stressful nature of the event that was disclosed and the compassionate motives underlying support provision. Future studies should examine contexts that are less emotionally and motivationally ambiguous, such as the sharing of successes or other good news with a partner (i.e., capitalization; Gable & Reis, 2010; Gable, Reis, Impett, & Asher, 2004).

It is also worth noting that the romantic couples in our study were younger and likely more educated (as they were drawn from a university sample) than the general population. Future research should test whether the effects found here extend to demographically distinct samples of couples. We believe that some differences may be found in investigating LSM in more established/older or differently educated couples. However, because LSM is concerned with function word similarity between individuals, these differences may not be especially large. While more experienced or differently educated couples may discuss different kinds of issues or use different kinds of language when describing their perspectives, these differences would be more likely to manifest in content words rather than function words.

As function words are a closed class of relatively short and common words, it may not be the case that differences in couples’ education level or age drastically influence their function word patterns. Nevertheless, to the extent that the use of different content words makes available a somewhat different subset of function words, it could be that some function word differences could emerge from differently educated communicators. However, a couple’s LSM score is concerned with how these words covary between individuals. Therefore, the words themselves should not systematically influence the extent of matching. Still, these are important questions that should be empirically examined in future work.

An additional direction for future research would be to explore the relationship between LSM and interpersonal behaviors and perceptions in longer partner conversations. Previous work has demonstrated that in lengthier conflicts, for instance, demand–withdraw patterns can often emerge (e.g., Christensen & Heavey, 1990), in which one partner’s aggression is met with the other’s disengagement(sometimes called “stonewalling"). It may be the case that the association between LSM and negative behavior, for example, would be different in an interaction exhibiting demand–withdraw characteristics compared to our own, because these expressions of conflict reflect uneven levels of engagement between interactants. Previous LSM research has examined how shifts in LSM over time in longer interactions uniquely predict important conversation outcomes, such as criminal confessions (B. H. Richardson et al., 2014). This suggests that conversation duration may be an important factor in shaping the meaning of LSM and its potential beneficial and deleterious effects for partners discussing sensitive issues.

A final future direction would be to explore LSM and nonverbal mimicry side by side. The literature on behavioral mimicry predates and far exceeds that of LSM, although the two forms of mimicry are believed to be analogs of one another (Niederhoffer & Pennebaker, 2002). This is a striking claim, considering that rapport building is a primary purported function of behavioral mimicry. Are verbal and nonverbal mimicry indexing two different processes, then? This is an empirical question, and future research should strive to examine the two types of mimicry together, across interaction contexts that are both positive and negative.

Conclusion

LSM is a ubiquitous phenomenon of interpersonal language use, whatever the communication medium. While there are still some mysteries as to its functions, it appears to be crucially involved in the many dynamic interactions that unfold over the course of human relationships. We may sometimes take for granted the staggering amount of social information we convey unconsciously when we communicate with others in our lives. Given its prevalence and importance, LSM is certainly one such source of that information that we have only recently begun to study and appreciate. LSM is a valuable metric for researchers, amenable to interdisciplinary efforts throughout the social sciences. It is being built upon and combined with other analytical techniques (e.g., Babcock, Ta, & Ickes, 2014; Tausczik & Pennebaker, 2013) and is helping pave the way to untapped sources of linguistic and psychological data. As our communication evolves, so too should our tools for understanding it.

Footnotes

Appendix

Acknowledgements

The authors would like to thank Ryan Baer, Danielle Beck, Sean Brennan, Debbie Cecena, Angel Chan, Alison Dana, Mitchell Fajardo, Jasmine Johnson, Atina Manvelian, Molly Rose Morrissey, Erin Mueth, Kimberly Ngo, Samantha Sallie, Yingli Sieh, Brett Sinclair, Joel Tennyson, Trevor Tsay, Lauren Tukey, Emily Turner, Rebeca Velasco, and Jessica Williams for their assistance administering the experiment, and transcribing and coding data. We are also grateful to our two anonymous referees and Howie Giles for their outstanding feedback and insight which substantially strengthened our work.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The first author was supported by an NSF Graduate Research Fellowship (ID No. 2013135613) during the analysis of study data and composition of the research article.

Notes

Author Biographies

Jeffrey D. Bowen is a PhD candidate in the Department of Psychological and Brain Sciences at the University of California, Santa Barbara. His research explores the intersecting roles of motivation and mental representation in romantic partners’ efforts to maintain their relationships (i.e., through sacrifice, conflict negotiation, and fidelity). He examines these phenomena with interaction-based paradigms, by assessing patterns in language use, and through computerized cognitive tasks.

Lauren A. Winczewski is a doctoral candidate in the Department of Psychological and Brain Sciences at the University of California, Santa Barbara. Her work examines responsive behavior in romantic couples, with a focus on the mechanisms by which partners are willing and able to seek support from and provide support to one another (e.g., empathy, self-disclosure).

Nancy L. Collins, PhD, is a professor of social psychology in the Department of Psychological and Brain Sciences at the University of California, Santa Barbara. Her research lies at the interface of close relationships, social cognition, and health psychology. Her current work focuses on social support as an interpersonal process, attachment dynamics in couples, and psychological and biological mechanisms through which close relationships foster health, resilience, and thriving.

References

Babcock

M. J.

V. P.

Ickes

(2014). Latent semantic similarity and language style matching in initial dyadic interactions. Journal of Language and Social Psychology, 33, 78-88.

Bailenson

J. N.

Yee

(2007). Virtual interpersonal touch and digital chameleons. Journal of Nonverbal Behavior, 31, 225-242.

Bavelas

J. B.

Black

Lemery

C. R.

Mullett

(1986). “I show how you feel”: Motor mimicry as a communicative act. Journal of Personality and Social Psychology, 50, 322-329.

Cannava

Bodie

G. D.

(2016). Language use and style matching in supportive conversations between strangers and friends. Journal of Social and Personal Relationships, 1-19.

Chartrand

T. L.

Bargh

J. A.

(1999). The chameleon effect: The perception-behavior link and social interaction. Journal of Personality and Social Psychology, 76, 893-910.

Christensen

Heavey

C. L.

(1990). Gender and social structure in the demand/withdraw pattern of marital conflict. Journal of Personality and Social Psychology, 59, 73-81.

Chung

Pennebaker

J. W.

(2007). The psychological functions of function words. In Fiedler

(Ed.), Social communication (pp. 343-359). New York, NY: Psychology Press.

Collins

N. L.

Feeney

B. C.

(2000). An attachment theoretical perspective on social support dynamics in couples: Normative processes and individual differences. In Sullivan

K. T.

Davila

(Eds.), Support processes in intimate relationships (pp. 89-120). New York, NY: Oxford University Press.

Collins

N. L.

Kane

H. S.

Metz

M. A.

Cleveland

Khan

Winczewski

. . . Prok

(2014). Psychological, physiological, and behavioral responses to a partner in need: The role of compassionate love. Journal of Social and Personal Relationships, 31, 601-629.

10.

Dalton

A. N.

Chartrand

T. L.

Finkel

E. J.

(2010). The schema-driven chameleon: How mimicry affects executive and self-regulatory resources. Journal of Personality and Social Psychology, 98, 605-617.

11.

Eldridge

K. A.

Christiansen

(2002). Demand-withdraw communication during couple conflict: A review and analysis. In Noller

Feeney

J. A.

(Eds.), Understanding marriage: Developments in the study of couple interaction (pp. 289-322). Cambridge, England: Cambridge University Press.

12.

Fincham

F. D.

Linfield

K. J.

(1997). A new look at marital quality: Can spouses feel positive and negative about their marriage? Journal of Family Psychology, 11, 489-502.

13.

Gable

S. L.

Reis

H. T.

(2010). Good news! Capitalizing on positive events in an interpersonal context. Advances in Experimental Social Psychology, 42, 195-257.

14.

Gable

S. L.

Reis

H. T.

Impett

E. A.

Asher

E. R.

(2004). What do you do when things go right? The intrapersonal and interpersonal benefits of sharing positive events. Journal of Personality and Social Psychology, 87, 228-245.

15.

Gallois

Ogay

Giles

(2005). Communication accommodation theory: A look back and a look ahead. In Gudykunst

W. B.

(Ed.), Theorizing about intercultural communication (pp. 121-148). Thousand Oaks, CA: Sage.

16.

Geiss

S. K.

O’Leary

K. D.

(1981). Therapist ratings of frequency and severity of marital problems: Implications for research. Journal of Marital and Family Therapy, 7, 515-520.

17.

Giles

(Ed.). (2016). Communication accommodation theory: Negotiating personal relationships and social identities across contexts. Cambridge, England: Cambridge University Press.

18.

Giles

Coupland

(Eds.). (1991). Contexts of accommodation: Developments in applied sociolinguistics. New York, NY: Cambridge University Press.

19.

Giles

Coupland

(1991). Language: Contexts and consequences. Pacific Grove, CA: Brooks/Cole.

20.

Gonzales

A. L.

Hancock

J. T.

Pennebaker

J. W.

(2010). Language style matching as a predictor of social dynamics in small groups. Communication Research, 37, 3-19.

21.

Gottman

J. M.

(1979). Marital interaction: Experimental investigations. New York, NY: Academic Press.

22.

Gottman

J. M.

(1980). Consistency of nonverbal affect and affect reciprocity in marital interaction. Journal of Consulting and Clinical Psychology, 48, 711-717.

23.

Gottman

J. M.

Markman

Notarius

(1977). The topography of marital conflict: A sequential analysis of verbal and nonverbal behavior. Journal of Marriage and the Family, 39, 461-477.

24.

Hatfield

Cacioppo

J. T.

Rapson

R. L.

(1993). Emotional contagion. New York, NY: Cambridge University Press.

25.

Heavey

Gill

Christensen

(2002). Couples Interaction Rating System 2 (CIRS2). Unpublished manuscript, University of California, Los Angeles.

26.

Hendrick

S. S.

(1988). A generic measure of relationship satisfaction. Journal of Marriage and the Family, 50, 93-98.

27.

Ickes

Bissonnette

Garcia

Stinson

L. L.

(1990). Implementing and using the dyadic interaction paradigm. In Hendrick

Clark

M. S.

(Eds.), Review of personality and social psychology: Vol. 11. Research methods in personality and social psychology (pp. 16-44). Newbury Park, CA: Sage.

28.

Ireland

M. E.

Henderson

M. D.

(2014). Language style matching, engagement, and impasse in negotiations. Negotiation and Conflict Management Research, 7, 1-16.

29.

Ireland

M. E.

Pennebaker

J. W.

(2010). Language style matching in writing: Synchrony in essays, correspondence, and poetry. Journal of Personality and Social Psychology, 99, 549-571.

30.

Ireland

M. E.

Slatcher

R. B.

Eastwick

P. W.

Scissors

L. E.

Finkel

E. J.

Pennebaker

J. W.

(2011). Language style matching predicts relationship initiation and stability. Psychological Science, 22, 39-44.

31.

Kelley

H. H.

Thibaut

J. W.

(1978). Interpersonal relations: A theory of interdependence. New York, NY: John Wiley.

32.

Kenny

D. A.

Kashy

D. A.

Cook

W. L.

(2006). Dyadic data analysis. New York, NY: Guilford Press.

33.

Kiecolt-Glaser

J. K.

Malarkey

W. B.

Chee

Newton

Cacioppo

J. T.

Maio

H. Y.

Glaser

(1993). Negative behavior during marital conflict is associated with immunological down-regulation. Psychosomatic Medicine, 55, 395-409.

34.

LaFrance

(1982). Posture mirroring and rapport. In Davis

(Ed.), Interaction rhythms: Periodicity in communicative behavior (pp. 279-298). New York, NY: Human Sciences Press.

35.

Lakin

J. L.

Chartrand

T. L.

(2003). Using nonconscious behavioral mimicry to create affiliation and rapport. Psychological Science, 14, 334-339.

36.

Lakin

J. L.

Chartrand

T. L.

Arkin

R. M.

(2008). I am too just like you: Nonconscious mimicry as an automatic behavioral response to social exclusion. Psychological Science, 19, 816-822.

37.

Lakin

J. L.

Jefferis

V. E.

Cheng

C. M.

Chartrand

T. L.

(2003). The chameleon effect as social glue: Evidence for the evolutionary significance of nonconscious mimicry. Journal of Nonverbal Behavior, 27, 145-162.

38.

Laurenceau

J.-P.

Feldman Barrett

L. A.

Pietromonaco

P. R.

(1998). Intimacy as an interpersonal process: The importance of self-disclosure, partner disclosure, and perceived partner responsiveness in interpersonal exchanges. Journal of Personality and Social Psychology, 74, 1238-1251.

39.

Lord

S. P.

Sheng

Imel

Z. E.

Baer

Atkins

D. C.

(2015). More than reflections: Empathy in motivational interviewing includes language style synchrony between therapist and client. Behavioral Therapy, 46, 296-303.

40.

Lyubomirsky

Sheldon

K. M.

Schkade

(2005). Pursuing happiness: The architecture of sustainable change. Review of General Psychology, 9, 111-131.

41.

Maisel

N. C.

Gable

S. L.

(2009). The paradox of received social support: The importance of responsiveness. Psychological Science, 20, 928-932.

42.

Maisel

N. C.

Gable

S. L.

Strachman

(2008). Responsive behaviors in good times and bad. Personal Relationships, 15, 317-338.

43.

Neumann

Strack

(2000). “Mood contagion”: The automatic transfer of mood between persons. Journal of Personality and Social Psychology, 79, 211-233.

44.

Niederhoffer

K. G.

Pennebaker

J. W.

(2002). Linguistic style matching in social interaction. Journal of Language and Social Psychology, 21, 337-360.

45.

Pasch

L. A.

Bradbury

T. N.

(1998). Social support, conflict, and the development of marital dysfunction. Journal of Consulting and Clinical Psychology, 66, 219-230.

46.

Pennebaker

J. W.

Booth

R. E.

Francis

M. E.

(2007). Linguistic Inquiry and Word Count: LIWC 2007: Operator’s manual. Austin, TX: LIWC.net.

47.

Pickering

M. J.

Ferreira

V. S.

(2008). Structural priming: A critical review. Psychological Bulletin, 134, 427-459.

48.

Pickering

M. J.

Garrod

(2004). Toward a mechanistic psychology of dialogue. Behavioral and Brain Sciences, 27, 169-226.

49.

Rains

S. A.

(2015). Language style matching as a predictor of perceived social support in computer-mediated interaction among individuals coping with illness. Communication Research, 42, 1-19.

50.

Raudenbush

S. W.

Bryk

A. S.

Cheong

Y. F.

Congdon

R. T.

du Toit

(2011). HLM 7. Lincolnwood, IL: Scientific Software International.

51.

Reis

H. T.

Clark

M. S.

Holmes

J. G.

(2004). Perceived partner responsiveness as an organizing construct in the study of intimacy and closeness. In Mashek

D. J.

Aron

(Eds.), Handbook of closeness and intimacy (pp. 201-225). Mahwah, NJ: Lawrence Erlbaum.

52.

Reis

H. T.

Shaver

P. R.

(1988). Intimacy as an interpersonal process. In Duck

(Ed.), Handbook of personal relationships (pp. 367-389). Chichester, England: Wiley.

53.

Richardson

B. H.

Taylor

P. J.

Snook

Conchie

S. M.

Bennell

(2014). Language style matching and police interrogation outcomes. Law and Human Behavior, 38, 357-366.

54.

Richardson

D. C.

Dale

(2005). Looking to understand: The coupling between speakers’ and listeners’ eye movements and its relationship to discourse comprehension. Cognitive Science, 29, 1045-1060.

55.

Rusbult

C. E.

Arriaga

X. B.

(1997). Interdependence theory. In Duck

(Ed.), Handbook of personal relationships: Theory, research and intervention (2nd ed., pp. 221-250). Chichester, England: Wiley.

56.

Rusbult

C. E.

Buunk

B. P.

(1993). Commitment processes in close relationships: An interdependence analysis. Journal of Social and Personal Relationships, 10, 175-204.

57.

Sprecher

Metts

Burleson

Hatfield

Thompson

(1995). Domains of expressive interaction in intimate relationships: Associations with satisfaction and commitment. Family Relations, 44, 203-210.

58.

Tausczik

Y. R.

Pennebaker

J. W.

(2010). The psychological meaning of words: LIWC and computerized text analysis methods. Journal of Language and Social Psychology, 29, 24-54.

59.

Tausczik

Y. R.

Pennebaker

J. W.

(2013). Improving teamwork using real-time language feedback. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 459-468). New York, NY: ACM.

60.

Taylor

P. J.

Thomas

(2008). Linguistic style matching and negotiation outcome. Negotiation and Conflict Management Research, 1, 263-281.

61.

Thakerar

J. N.

Giles

Cheshire

(1982). The psychological and linguistic parameters of speech accommodation theory. In Fraser

Scherer

K. R.

(Eds.), Advances in the social psychology of language (pp. 205-255). Cambridge, England: Cambridge University Press.

62.

Watson

Clark

L. A.

Tellegen

(1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063-1070.

63.

Webb

O. J.

Eves

F. F.

Smith

(2011). Investigating behavioral mimicry in the context of stair/escalator choice. British Journal of Health Psychology, 16, 373-385.

64.

Winczewski

L. A.

Bowen

J. D.

Collins

N. L.

(2016). Is empathic accuracy enough to facilitate responsive behavior in dyadic interaction? Distinguishing ability from motivation. Psychological Science, 27, 394-404.

65.

Yabar

Johnston

Miles

Peace

(2006). Implicit behavioral mimicry: Investigating the impact of group membership. Journal of Nonverbal Behavior, 30, 97-113.