I blame you, I hear you: Couples’ pronoun use in conflict and dyadic coping

In dyadic interaction, a verbal focus on one individual (“you-talk,” “I-talk”), rather than on the couple (“we-talk”) has predominantly been linked to dysfunctional relationship processes. However, context differences in these links have not yet been systematically examined. Is it functional to asymmetrically focus on one partner during support interactions but problematic during conflict? Does a high level of couple-focus represent a resource across contexts? In this preregistered study, we investigated dyad-level pronoun use (we-/I-/you-talk) and their link to situational relationship functioning (SRF) across three interaction tasks (one conflict, two dyadic coping tasks) within couples (N = 365). More specifically, we examined associations of couple-means, i.e. pronoun use as a shared resource/vulnerability between partners, and couple-differences, i.e. functional/dysfunctional asymmetric pronoun use with observed interaction positivity and relationship climate. Results revealed both context differences and similarities. Asymmetric partner-focus (i.e. you-talk) was dysfunctional in conflict, whereas asymmetric partner- and self-focus (i.e., you-talk/I-talk; focus on the stressed partner) were functional in dyadic coping. Beyond asymmetry, you-talk (couple-mean) showed consistent negative associations with SRF in all tasks studied. We-talk (couple-mean) was positively linked to SRF, but only in conflict interactions. In conflict, couple-focus thus represented a shared resource that can buffer from dysfunctional conflict interaction characterized by partner-focus. In line with conceptual frameworks, the dyadic coping results emphasize the importance of focusing on the partner in need. The study corroborates the prospect of pronoun use as a context-specific indicator of relationship functioning. Gender differences, implications for future research and possible interventions are discussed.

functional in dyadic coping. Beyond asymmetry, you-talk (couple-mean) showed consistent negative associations with SRF in all tasks studied. We-talk (couple-mean) was positively linked to SRF, but only in conflict interactions. In conflict, couple-focus thus represented a shared resource that can buffer from dysfunctional conflict interaction characterized by partner-focus. In line with conceptual frameworks, the dyadic coping results emphasize the importance of focusing on the partner in need. The study corroborates the prospect of pronoun use as a context-specific indicator of relationship functioning. Gender differences, implications for future research and possible interventions are discussed.

Keywords
Close relationships, conflict, couples, dyadic coping, language use When couples interact with one another, their pronoun use is a reflection of where they see themselves in the relationship. Pronoun use is a linguistic indicator of the changing focus on the self, the partner and/or on the couple as a dyadic unit. Accordingly, pronoun use as a behavioral marker of relationship quality and its underlying processes has been examined by a number of studies (e.g., Karan et al., 2019;Slatcher et al., 2008; see for an overview : Horn & Meier, in press). In this research line, an enhanced focus on the partners as separate units ("you-talk," "I-talk") as opposed to a focus on the couple as one interdependent unit ("we-talk") has predominantly been linked to dysfunctional relationship processes (e.g., Rentscher et al., 2013;Sillars et al., 1997). Although promising, the literature on pronoun use and relationship functioning has two main limitations: First, most laboratory-based studies to date examined pronoun use in one single relational context, i.e. either conflict or social support interactions. Little is known about context differences in the links between pronoun use and relationship functioning. While for related phenomena, such as language style coordination, distinct outcomes have been documented depending on whether coordination occurred during a conflict or support interaction (Bowen et al., 2017;Richardson et al., 2019), context differences regarding the functionality of pronoun use remain understudied. For instance, an enhanced partner-focus should have different implications depending on whether it is expressed in a coping or in a conflict situation as these contexts represent distinct domains of marital interaction with different implications for the course of a relationship (Pasch & Bradbury, 1998). A focus on the partner is crucial in support and dyadic coping interactions in which expressions of stress and need by the support-seeking partner are to be perceived and responded to (Bodenmann, 1995;Kuhn et al., 2018). In conflict interactions, however, partner-focus may indicate dysfunctional behavior such as blaming or criticism (e.g., "you always . . . "; Gottman & Levenson, 2000).
A second limitation of the current literature is that, with a few exceptions (e.g., Rentscher et al., 2013;Timmons et al., 2021), most studies investigated links between individual partners' pronoun use, i.e. actor/partner effects, and relationship functioning, without considering genuinely dyadic patterns of pronoun use. Imagine that one partner expresses high couple-focus, e.g. "our" problem, in a conversation, but the other partner does not or to a very limited degree, e.g. "your" problem-what does this imbalanced, asymmetric pattern on the couple-level tell us about their relationship functioning? Is the amount of "we-ness" expressed by one partner sufficient to benefit the dyad as a whole in the sense of a "shared resource"? Or is the fact that its expression is asymmetric between partners detrimental for situational relationship outcomes in the sense of a "resource imbalance"?
In general, asymmetric patterns have often been observed in the context of dysfunctional marital interaction (Christensen & Heavey, 1990;Rentscher et al., 2013). On the positive side, however, asymmetry may also indicate attention to one partner's needs which is a hallmark of intimate relationships. Accordingly, striving to meet the partner's needs and compensate resources can sometimes override considerations of balance in intimate interactions rendering asymmetry functional when it serves to adequately respond to a partner's needs (Cutrona et al., 2007;Reis & Shaver, 1988). The present study aimed to address these research gaps by systematically examining dyad-level pronoun use patterns and their link to situational relationship functioning (SRF) across varying interaction tasks within couples. Hereby, SRF refers to indicators of positive relationship functioning such as a high ratio of positive to negative interaction behaviors (see Balance Theory; Gottman, 1993) in a given context. In conflict situations, this would be more validation than criticism, in support situations more active listening than domineering.

Couples interactions: Conflict and dyadic coping
Dyadic interactions span many different relational contexts, but one context that has perhaps received the most scientific attention in couple research is conflict. In the past decades, conflict interaction patterns have been linked to detrimental outcomes both on the relational and individual level including relationship distress and dissolution (Gottman & Levenson, 2000).
While the ability to solve conflicts together is certainly an important ingredient of a well-functioning relationship, it is but one of the multitude of facets that define romantic relationships. Shared resources and experiences that accumulate over time in a sense of togetherness or "we-ness" (Aron et al., 2004;Bodenmann, 1995;Fergus, 2015) as well as the provision of support in times of stress are at least equally important for relationship functioning (Pasch & Bradbury, 1998). Stressful demands from workplace and other responsibilities, i.e. relationship-external stress, often spill over to affect both partners and require them to cope together as a couple. This process has been termed dyadic coping (Bodenmann, 2005) or communal coping (Lyons et al., 1998). Within this shared stress experience, close relationships represent a valuable resource that goes beyond their individual coping resources (Coan & Sbarra, 2015). In the Systemic-Transactional Model (STM), dyadic coping has been introduced as a reciprocal and dynamic interplay between one partner's stress signals and the other partner's support (Bodenmann, 1995). A communal perspective on the stressor ("our" stressor) may foster adaptive individual and dyadic outcomes through partners' joint coping efforts (Berg & Upchurch, 2007;Lyons et al., 1998).
Among the studies that have examined couples' conflict and support interactions, a growing number specifically focused on the verbal aspects of communication. Specific mechanisms that contribute to or impede on relationship functioning such as dyadic coping (Rentscher, 2019), or dysfunctional conflict interaction (Neysari et al., 2016;Rentscher et al., 2013) implicitly manifest in couples' pronoun use.

Dyadic pronoun use patterns in conflict and support interactions
In the present study, we adopted a "dyad-as-a-unit-perspective" and focused on dyadlevel patterns of pronoun use, namely, dyadic levels (couple-mean) and dyadic discrepancies (difference within couple). While the dyadic discrepancy indicates the presence of asymmetry between partners, the dyadic level indicates whether both partners' "contribution" to the mean can compensate one for another in the sense of a shared resource or shared vulnerability (Iida et al., 2018;Robinson & Cameron, 2012).
Let us look at a hypothetical example: Partner A shows a very high rate of first-person plural pronouns (we-talk) during a conversation whereas Partner B shows a moderate rate of we-talk. This couple then has both a high dyadic level (couple-mean) and high dyadic discrepancy (couple difference score) of we-talk. When we simultaneously model associations of we-talk (i.e., the dyadic level and dyadic discrepancy) and SRF, and only the dyadic level has a positive association with the outcome, we could think of we-talk as a shared resource between partners (see Robinson & Cameron, 2012). In this scenario, the couple benefits from a high absolute level of we-talk even though only one partner shows a pronounced we-focus. Beyond that, discrepancy indicates whether asymmetry between partners' pronoun use is functional or dysfunctional.
In sum, associations of dyad-level pronoun use (we-talk, I-talk, you-talk), can indicate shared resources or shared vulnerabilities (dyadic level) as well as functional or dysfunctional asymmetry (dyadic discrepancy), depending on the direction of the effects. In the following, we briefly review the literature on how couples' we-, I-, and you-talk reflect or contribute to relationship functioning to lay the ground for our hypotheses about dyad-level pronoun use.
We-talk. Couples' we-talk is generally seen as reflecting interdependence and has been linked to higher relationship satisfaction and interaction quality (Karan et al., 2019). Positive associations between we-talk and relationship outcomes have been observed in conflict (Seider et al., 2009;Simmons et al., 2005;Williams-Baucom et al., 2010) and dyadic coping interactions (Badr et al., 2016;Robbins et al., 2012;Rohrbaugh et al., 2008). Partners' we-talk in dyadic coping interactions may reflect their shared appraisal of distress, communal coping orientation, and high engagement in mutual coping (Karan et al., 2017;Rohrbaugh et al., 2008). Clinical studies with patient-spouse dyads showed the importance of spouse's we-talk for dyadic coping outcomes (Badr et al., 2016;Karan et al., 2019;Robbins et al., 2012;Rohrbaugh et al., 2008), speaking for the particular benefits of asymmetric couple-focus in social support contexts. In contrast, asymmetric we-talk sometimes reflects a problematic imbalance in conflict interactions, such as when one partner keeps demanding change (Rentscher et al., 2013).
In sum, we-talk has generally been linked to desirable relationship processes both in conflict and in dyadic coping contexts, and can thus be thought of as a "shared resource." In case of we-talk asymmetry, however, opposing effects in conflict ("dysfunctional asymmetric couple-focus") versus dyadic coping ("functional asymmetric couple-focus" by the support-provider) can be expected.
I-talk. Albeit in couple therapy I-talk in the sense of emotional self-disclosure is explicitly trained (Epstein & Baucom, 2013), language-based studies suggest that an enhanced self-focus often reflects emotional negativity (Tackman et al., 2019). As shown by couples studies, high rates of I-talk during couple conflict indicate separateness and lower relationship functioning (Seider et al., 2009;Sillars et al., 1997;Williams-Baucom et al., 2010). This seems to be especially the case for satisfied couples, and an opposing pattern has been observed for dissatisfied couples (Simmons et al., 2005;Williams-Baucom et al., 2010).
However, I-talk has also been conceptualized as a proxy for self-disclosure with high levels of immediacy (Pennebaker & King, 1999;Simmons et al., 2005) and can build the cornerstone for psychological intimacy and dyadic coping (Bodenmann, 1995). From this perspective, I-talk should be beneficial in situations characterized by a need for support despite signaling partners' self-focus or separateness (e.g., I feel very bad, this was hurtful). Disclosing one's needs in an immediate, explicit manner may provide opportunities to build a mutual understanding of the stress (Berg & Upchurch, 2007)-a prerequisite for adequate dyadic coping (Bodenmann, 1995;Cutrona et al., 2007) in which I-talk may play an essential role.
In conclusion, we hypothesize that a high dyadic level of I-talk represents a shared vulnerability in conflict (e.g., indicating negative self-focus or separateness), but a shared resource in dyadic coping (e.g., indicating self-disclosure). Moreover, we expect asymmetric self-focus to be dysfunctional in conflict, but functional in dyadic coping by providing room for the stressed partner's disclosure and adequate support provision.
You-talk. You-talk has previously been linked to detrimental relationship processes, such as blaming or psychological distancing (Rentscher et al., 2013;Sillars et al., 1997;Simmons et al., 2005). While you-talk as a proxy for blaming and domineering seems intuitive, it has thus far predominantly been studied in conflict interactions. Recent evidence suggests that you-talk may play a distinct, beneficial role in coping contexts (Badr et al., 2016;Karan et al., 2017). Dyadic coping is a genuinely asymmetric situation, in which the partner in need discloses and the other partner responds-a focus on the stressed partner may then be functional. Empirical evidence supports the idea of partner-focus as an ingredient of positive dyadic coping (Badr et al., 2016;Lau et al., 2018), possibly by signaling the support-provider's empathetic focus on the partner (e.g., how are you feeling?) and active listening  or paraphrasing (Bodenmann, 2005).
In the current study, we assume that you-talk is not an indicator of negative relationship processes in its own right. Rather we think that, depending on the relational context, partner-focus relates to less positive situational relationship outcomes in conflict interactions, and more positive outcomes in dyadic coping interactions. More

3270
Journal of Social and Personal Relationships 38(11) specifically, we assume that a high dyadic level of you-talk represents a shared vulnerability in conflict interactions, but a shared resource in dyadic coping interactions. For asymmetric you-talk, we expect the same direction of associations, namely dysfunctional asymmetric partner-focus in conflict (e.g., asymmetric blaming), and functional asymmetric partner-focus in dyadic coping (e.g., listening, responsiveness).

The present study
The goal of this study was to investigate dyadic patterns (i.e., dyadic levels and dyadic discrepancies) of pronoun use and their associations with relationship outcomes across conflict and dyadic coping interactions. More specifically, we investigated couples' we-, I-, and you-talk, and their links with SRF (i.e., observational data of relationship climate and interaction behavior) in each of these tasks, and expected context differences in terms of the adaptiveness of pronoun use. For an overview of our preregistered 1 hypotheses, see also Figures 1 and 2.

Participants
The present study used data from a research project about the impact of stress on intimate relationships conducted at the University of Zurich. Couples had to be in their relationship for at least 1 year and were recruited through newspaper and radio advertisements. We used data from 368 Swiss German speaking heterosexual couples who participated in the first assessment point of the longitudinal project.
We applied our preregistered exclusion criteria to obtain the final sample. For the conflict task, we had to exclude data of 14 couples who did not discuss an area of disagreement, but rather a more neutral topic (e.g., mundane conversation without tensions), or shared distress that did not provoke a conflict, suggesting the manipulation of the relationship context did not succeed. Since contextual differences lay at the core of our research questions, we excluded these discussions that were not valid to represent the context of interest. Moreover, observational data was missing for five couples in the conflict task, for six couples in the support task (man as support-provider), and for five couples in the other support task (woman as support-provider). Among the reasons for missing observational data were technical issues, personal issues (e.g., feeling unwell), and one couple did not give consent to use their video data. 2 Our final sample consisted of N ¼ 365 couples, out of which 349 completed the conflict discussion task, 362 the Note. X-axis represents standardized estimates of the SEM models; T-bars are 95% confidence intervals.
U ¼ Association was in the direction opposite to what was hypothesized. Pronoun use accounted for 1.6% of variance in relationship climate and 10.7% in interaction positivity when the woman was support-provider, and for 3.8% in relationship climate and 9.6% in interaction positivity when the man was support-provider.
dyadic coping task (man as support-provider) and 363 the dyadic coping task (woman as support-provider). Partners were between 19 and 82 years old; (women: M ¼ 47.24, SD ¼ 18.33; men: M ¼ 49.25, SD ¼ 18.26). Couples had been in their current relationship for M ¼ 21.66 (SD ¼ 18.19; range: 1-60) years, 66% of the couples were married, 85% lived together, and 65% had children. Participants were mostly Caucasian and represented a Swiss middle-class sample in terms of education and income level (see Kuster et al., 2015 for a detailed sample description). The sample was moreover characterized by relatively high relationship satisfaction (women: M ¼ 4.33, SD ¼ 0.50; men: M ¼ 4.38, SD ¼ .47) according to the 5-point scale of the Relationship Assessment Scale (RAS; Hendrick, 1988;Sander & Böcker, 1993).
Some data of the larger research project has already been used in other publications (e.g., Kuster et al., 2015;Leuchtmann et al., 2019;Neysari et al., 2016). The present article is the first to investigate couples' language use across interaction contexts, representing unique research questions, analytic strategies, and models.

Procedure
Eligible participants were invited to a laboratory session where they were informed about the procedure and provided informed consent. After completing a set of questionnaires not relevant to the present study, couples participated in three videotaped interaction tasks (a conflict discussion task, and two dyadic coping tasks) of 8-minute duration each.
First, couples engaged in a conflict interaction task. With the help of a list about common areas of disagreement (PAQ A; Heavey et al., 1995) and a trained research assistant, partners collaboratively identified a source of tension in their relationship they wanted to discuss. Among the most frequently discussed conflict topics were communication with the partner, annoying habits of the partner, and finances (Kuster et al., 2015;Leuchtmann et al., 2019). After deciding on the topic, participants were left alone in the room and videotaped while discussing the issue. Couples then engaged in two dyadic coping discussions. In an analogous procedure, each partner identified a recent stressor originating from outside of their relationship. Again couples were left alone in the room, and the partner then described their stress while their spouse had been instructed to react as they typically would in their daily lives . The order of the two support tasks, that is, which partner first took the role of the support-provider had been determined randomly. Subjects were debriefed and received compensation (equivalent to 100 USD) for their participation in the study. All study procedures were reviewed and approved by the local institutional review board.

Measures
Trained research assistants rated the videos for two key indicators of SRF, i.e. interaction positivity and relationship climate by relying on standardized coding systems. The training of research assistants included a minimum of 60 hours of coding until they achieved interrater agreement of Cohen's kappa ¼ .90 (see Kuhn et al., 2018;Leuchtmann et al., 2019).
Interaction positivity. For the conflict task, positive and negative interaction was rated for each partner using a modified version (Kuster et al., 2015) of the SPAFF coding system (Gottman & Krokoff, 1989). The positive ratings consisted of four subcategories (interest/curiosity, validation, affect/caring, and constructive criticism), and the negative ratings of seven subcategories (criticism, defensiveness, contempt, domineering, belligerence, stonewalling, and formal negative interaction). All categories were rated according to their occurrence in a 10-second window (48 sequences in the 8 min task). If a pertinent behavior was displayed, a "1" was coded in the respective category, with the categories being mutually exclusive, i.e. only one category in a sequence was coded for each partner.
In line with previous research (e.g., Zemp et al., 2017), we created a composite score for each partner's positivity, representing the ratio of positive to negative relationship behaviors corresponding to the given context (Gottman, 1993). In an initial step, we counted the occurrences of positive and negative interactions for each partner across all sequences. Each partner's positivity and negativity measures were subsequently combined into a single score. More specifically, each partner's composite score was computed by dividing the number of positive interactions (possible range: 0-48) by the number of negative interactions (possible range: 0-48). Higher scores thus indicated more positive interaction overall. In case partners did not show any verbal negative interaction (0 negative sequences 3 ), we handled this computational problem by treating these cases as if they showed one sequence of negative interaction.
In the dyadic coping tasks, the support provider's interaction positivity was rated according to the Coding System for Dyadic Coping (SEDC; System for Assessing Observed Dyadic Coping; Bodenmann, 2005). Four categories (three positive, one negative) were used to rate the situational success of dyadic coping: Problem-focused supportive/common dyadic coping, interest/curiosity, emotion-focused, supportive dyadic coping/common verbal emotion-focused dyadic coping, and negative dyadic coping. As for the conflict task, we built a composite score of the support-providing partner's interaction positivity in each of the two dyadic coping tasks (ratio of positive to negative interaction).
Relationship climate. In addition, relationship climate was rated in each interaction task. More specifically, the following five bipolar items were rated for each couple: coupleorientation vs. self-orientation (we-ness), relaxed vs. tense, cheerful vs. depressed, peaceful vs. irritated, and close vs. reserved (intimacy). Each of these dimensions was rated on a 5-point scale: þ2 (very much), þ1 (rather), 0 (neutral) �1 (rather) or �2 (very much) in sequences of 10-seconds (48 sequences in each interaction task). For each interaction task, each dimension was then averaged across all sequences. In the present study, we used the mean across the 3 items relaxed/tense, peaceful/irritated, and close/ reserved. This resulted in a single indicator for relationship climate in each interaction task that was included in the final models (see Figure 3), with higher values indicating more positive relationship climate. Since context differences between conflict and dyadic coping lay at the core of our research questions, the aim here was to have comparable outcomes across both relational contexts. We thus used the 3 items with the strongest intercorrelations across interaction contexts to form a composite score. Average inter-item correlations and Cronbachs alpha were as follows: r ¼ .22, a ¼ .434 (conflict task), r ¼ .55, a ¼ .748 (dyadic coping task, woman as support-provider), and r ¼ .54, a ¼ .749 (dyadic coping task, man as support-provider; see also "Supplemental Materials A").
Pronoun use. Transcripts of the couples' conversations were analyzed with the German version (DE-LIWC2015; Meier et al., 2019Meier et al., , 2021 of the text analysis program LIWC (Pennebaker et al., 2015) to quantify each partner's use of first person plural pronouns (we-talk), first person singular pronouns (I-talk), and second person singular pronouns (you-talk) as percentages of total words they used. We then computed the dyadic level (couple-mean) and dyadic discrepancy (couple difference score; see Table 1) of each pronoun type, which were the independent variables in the models computed. Personal pronouns in German are largely used in similar ways as in English, and the comparability of the DE-LIWC2015 with the English LIWC dictionary has been empirically established (Meier et al., 2019). For more information on the transcription process, please see "Supplemental Materials A."

Statistical analyses
In order to investigate our research questions, we computed three (one for each interaction task) Dyadic Score Models (Iida et al., 2018). Hence, we simultaneously modeled associations between dyadic levels and differences of pronoun use and SRF (i.e., interaction positivity and relationship climate). As outlined in the preregistration, our intent was to model one dependent latent factor SRF indicated by all items measuring relationship climate and interaction positivity. Since this procedure resulted in a poor model fit, some adjustments to our preregistered modeling procedure were required, which ultimately led us to estimate models with manifest dependent variables (i.e., interaction positivity and of relationship climate), rather than one latent outcome. 4 We describe these adjustments in full detail in "Supplemental Materials C." Most importantly, the adjustments all conform to the overarching goal to examine dyad-level patterns of pronoun use across interaction contexts and their associations with interaction positivity and relationship climate as key indicators of SRF. Our final models are depicted in Figure 3. Since the results did not change when controlling for overall word count and relationship satisfaction, we report the most parsimonious models.

Results
Our hypotheses testing results for the conflict and dyadic coping tasks are depicted in Figures 1 and 2. Means and standard deviations of all variables are reported in Table 1, and correlations among all study variables, including intercorrelations among pronouns within individuals and couples, in "Supplemental Materials A." Detailed results with all parameter estimates are provided in "Supplemental Materials B." For additional insights into the conversations, we report correlations between our measures with positive and negative emotion word use as a proxy for the emotional responses during the conversation ("Supplemental Materials A"). As expected, more emotion words were used when the conversation was characterized by higher levels of I-talk. In the conflict task, higher partner-focus correlated with more negative emotion word use, thus possibly indicating more negative blaming. In the dyadic coping tasks, asymmetric partner-and self-focus moreover went along with more emotion word use by the stressed partner, therefore suggesting more emotional disclosure.

Conflict task
In support of our hypothesis about dysfunctional asymmetric partner-focus, you-talk (dyadic discrepancy) was associated with less positive relationship climate (B ¼ �0.02, SE ¼ 0.01, p < .001) and interaction behavior of the male (but not female, p ¼ .511) partner (B ¼ �0.51, SE ¼ 0.21, p ¼ .014). Our assumptions about dysfunctional asymmetric self-focus and couple-focus were not supported, as the dyadic discrepancies of I-talk (ps > .079) and we-talk (ps > .437) were not linked to any of the dependent variables. Above and beyond asymmetric you-talk, the dyadic level of you-talk was associated with less positive relationship climate (B ¼ �0.02, SE ¼ 0.01, p ¼ .033) and interaction behavior of the female (but not male, p ¼ .072) partner (B ¼ �0.99, SE ¼ 0.30, p ¼ .001), supporting our hypothesis of partner-focus as a shared vulnerability in conflict. The results shed light on the various negative correlates of you-talk during conflict: While both the overall rate as well as asymmetry in couples' you-talk uniquely accounted for variance in relationship climate, associations with interaction positivity differed upon gender. I-talk had one significant association in the direction opposite to what was expected: I-talk (dyadic level) was positively associated with the male partner's interaction positivity (B ¼ 0.55, SE ¼ 0.27, p ¼ .041), but not with relationship climate (p ¼ .149), nor female partner's interaction (p ¼ .750). Men's interaction was thus more positive, the higher a couples' overall rate of I-talk, which contradicted our assumption of self-focus as a shared vulnerability in conflict. Moreover, there was a positive association between we-talk (dyadic level) and relationship climate (B ¼ 0.04, SE ¼ 0.01, p ¼ 0.003), as well as men's interaction positivity (B ¼ 1.47, SE ¼ 0.48, p ¼ 0.002), respectively. Thus, our hypothesis about we-talk as a shared resource found support-the higher a couple's overall rate of we-talk, the more positive was their relationship climate and men's interaction behavior, although the effect again did not generalize onto women's interaction behavior (p ¼ .552).
Upon the suggestion of a reviewer, we reanalyzed our main findings with the frequency of positive and negative interaction as separate outcomes (rather than the ratio), thus accounting for the possibility that positive and negative behaviors might reveal distinct associations. This was not preregistered and we report these results in "Supplemental Material D." The main findings held, but some gender differences (i.e., dyadic level of I-/we-/you-talk) disappeared. Additionally, you-talk (dyadic discrepancy) was linked to more positive and more negative interaction of women, but to less positive interaction of men. This implies that asymmetric partner-focus in conflict might indicate intensification of communication behavior also on the positive side and that this might be particularly the case for women. Since it was still related to lower relationship climate and men's less positive interaction, we would take this result with caution.

Dyadic coping tasks
In support of our hypotheses of functional asymmetric partner-focus and self-focus in dyadic coping, the dyadic discrepancy of you-talk (B ¼ 1.45, SE ¼ 0.34, p < .001) as well as of I-talk (B ¼ 0.57, SE ¼ 0.29. p ¼ .046) was positively associated with the supportprovider's interaction positivity (but not with relationship climate; p ¼ .216 and p ¼ .443, respectively), when the female partner was the support-provider. Beyond asymmetry, the dyadic level of you-talk linked to both outcomes studied in the same task, but in the direction opposite to what was hypothesized: Relationship climate (B ¼ �0.06, SE ¼ 0.03, p ¼ .047) and interaction positivity (B ¼ �1.71, SE ¼ 0.67, p ¼ .011) were negatively associated with the level of you-talk. In other words, the more you-pronouns a couple used (while controlling for asymmetry), the less positive were their relationship climate and the support-provider's interaction behavior. This speaks against the assumptions of partner-focus as a shared resource in dyadic coping.
The results for functional asymmetric partner-focus and self-focus fully replicated in the second dyadic coping task, in which the male partner was the support-provider. The dyadic discrepancy of you-talk (B ¼ 0.93, SE ¼ 0.38, p ¼ .015) as well as of I-talk (B ¼ 0.97, SE ¼ 0.31, p ¼ .002) was positively associated with interaction positivity, but not with relationship climate (p ¼ .418 and p ¼ .570, respectively). The unexpected negative associations between the dyadic level of you-talk and relationship climate (B ¼ �0.09, SE ¼ 0.03, p ¼ .001) as well as interaction positivity (B ¼ �1.33, SE ¼ 0.65, p ¼ .042) were also replicated in this task. In addition, there were positive associations between the dyadic level of I-talk and relationship climate (B ¼ 0.04, SE ¼ 0.02, p ¼ .013), but not interaction positivity (p ¼ .615) when the man was the support-provider. This supports our assumption about self-focus as a shared resource in dyadic coping, while suggesting gender differences. Unexpectedly, we-talk was not linked to any of the outcome variables in either of the dyadic coping tasks.
We again reanalyzed the findings using positive and negative interaction as separate outcomes instead of the positivity ratio. All main findings for the dyadic coping tasks replicated (see "Supplemental Materials D"). In general, there were more associations between pronoun use and positive (rather than negative) interaction, possibly due to the low base rates and thus reduced reliability of our negative dyadic coping measures.

Discussion
The present study was among the first to systematically examine pronoun use and relationship functioning across two relational contexts: conflict and dyadic coping. In general, the results suggest both similarities and differences across relational contexts. For most pronoun types, associations moreover differed depending on the pattern (i.e., dyadic level versus discrepancy) studied, thus underlining the importance of asymmetries in partners' language use.
Our context-specific assumptions about functional asymmetric partner-focus in dyadic coping and dysfunctional asymmetric partner-focus in conflict were supported. Asymmetric you-talk during dyadic coping was linked with higher relationship functioning, and the finding was robust, as it replicated across both dyadic coping tasks examined. An enhanced focus on the stressed partner, as manifested in pronoun use, thus seems to reflect functional ways of dyadic coping. This finding of functional asymmetry in support situations was mirrored by the positive outcome associated with asymmetric self-focus-again, the result was robust across both dyadic coping tasks studied. As expected, focusing on the self when receiving support is functional and might represent stress expression that allows the listener to align dyadic coping efforts with the partner's needs. Counter to our expectations, asymmetric couple focus in dyadic coping did not contribute to SRF beyond asymmetric self-and partner-focus.
Contrasting the findings in dyadic coping, asymmetric partner-focus was linked to lower SRF in the conflict task above and beyond absolute levels of you-talk. This extends previous findings on the problematic use of you-talk in conflict (Rentscher et al., 2013;Sillars et al., 1997;Simmons et al., 2005). In the conflict situation, neither asymmetries in self-nor in couple-focus showed significant associations with relationship functioning beyond you-talk. In line with our expectation, the global level of couple-focus, as reflected in more we-talk, was associated with more positive outcomes in conflict.
The dyadic level of you-talk, on the other hand, showed consistent negative associations with SRF across all interaction tasks. An overall high amount of partner-focus thus represented a shared vulnerability in both of the two key relationship contexts studied. Conversations with a high overall partner-focus likely represent dysfunctional ways of disproportionally focusing on the romantic counterpart, as is the case in blaming, or domineering. This may be similarly dysfunctional when attempting to support the partner in a negative manner (Robbins et al., 2012; see negative dyadic coping: Bodenmann, 1995). Moreover, high absolute levels of partner-focus leave less space for selfand couple focus in the interaction, and may therefore render the interaction less supportive.

Dyadic coping: An interplay of functional asymmetric self-and partner-focus
Our study offers some clarification on the role of you-talk in dyadic coping that has been controversially discussed (Badr et al., 2016;Horn & Meier, in press;Robbins et al., 2012). Partner-focus seems to represent a functional component of dyadic coping, as long as it is directed toward the stressed partner. Moreover, the dyadic coping behavior was more positive when the recipient used more I-pronouns than the support-provider. This is in line with the broader literature on the fundamental role of self-disclosure in close relationships (Reis & Shaver, 1988). As a proxy of self-disclosure, functional asymmetric self-focus likely creates opportunities for the spouse to understand and create a shared reappraisal (Berg & Upchurch, 2007), and asymmetric partner-focus may in turn foster more disclosure (Bodenmann, 1995). This conforms to dyadic coping-based intervention programs that highlight the importance of the stressed partner's immersion, and the support-provider's active listening, empathic joining, and paraphrasing (Bodenmann & Shantinath, 2004). Our results add to the systemic-transactional view of stress and coping (Bodenmann, 1995) by suggesting asymmetric partner-focus and self-focus as linguistic correlates of the dynamic interplay between stress expression and responsive behavior. While our study focused on the aggregated level, future language-based research will benefit from micro-longitudinal analyses (e.g., Neysari et al., 2016) to study the temporal dynamics of self-/partner-focus in dyadic coping conversations. Another promising future direction will be to study how self-/partner-, and couple-focus relate to the specific subcategories of positive dyadic coping (e.g., emotion-focused or problem-focused dyadic coping; Bodenmann, 2005). In general, our findings add to research showing that dyadic coping demands attention to the stressed partner, and that this attention on the partner (e.g., active listening) in turn fosters more positive dyadic coping .
While asymmetric you-talk linked to the support-provider's positive interaction behavior in the dyadic coping tasks of our studies, it was not linked with relationship climate. One possible explanation for this lack of association is that although the supportive behavior was positively intended, it may not necessarily have corresponded with the needs of the stressed partner or the dyad (Bodenmann, 2005;Cutrona et al., 2007). Future research should include multiple outcomes (e.g., subjective appraisal) to obtain a more nuanced view on the adaptiveness of asymmetric you-talk in dyadic coping.
We-talk was surprisingly unrelated to relationship functioning in the dyadic coping tasks. Thus, while we-talk is beneficial in communal coping with severe health issues (Badr et al., 2016;Karan et al., 2017;Rentscher et al., 2017;Robbins et al., 2012), it could be less prompted when couples are instructed to discuss minor individual stressors that primarily concern one partner. Results from our study suggest that couple-focus is more important when discussing relationship-internal rather than relationship-external distress. Future research will benefit from taking the dynamics of the situation more into account. For example, dyadic coping interactions can also be characterized by couple conflict if the distress "spills over" (Leuchtmann et al., 2018), which may alter some of the associations with pronoun use.

Conflict: Partner-focus as a shared vulnerability, couple-focus as a shared resource
Our expectations of we-talk as a shared resource was only supported in the conflict task, in which we-talk (dyadic level) was linked with more positive relationship climate and men's interaction positivity. In contrast to high global and asymmetric use of you-talk, high global use of we-talk therefore represented a functional component of conflict interaction. For I-talk, the association was in the other direction as hypothesized: Selffocus did not represent a shared vulnerability, but rather a shared resource in conflict, especially for men's interaction behavior. The link between asymmetric I-talk (i.e., more I-talk by the woman) and men's interaction behavior moreover showed a similar positive trend beyond the dyadic level, although the association did not reach traditional thresholds for statistical significance. In support of earlier findings (Karan et al., 2019), this suggests that couples who manage to replace dysfunctional partner-focus with couple-focus or self-focus during conflict have higher relationship functioning.
The dyadic discrepancy of we-talk did not relate to outcomes, thus, our assumptions about dysfunctional asymmetric couple-focus did not find support in the data. Rather, the lack of association between asymmetric couple-focus and relationship functioning in our study suggests that "we-ness" can be compensated one for another: The expression of high couple-focus benefits the dyad in the sense of a shared resource, regardless of partners' individual levels of "we-ness." High rates of we-talk by one partner might foster de-escalation and constructive discussion of relationship conflict. This result is in line with theoretical frameworks that see "we-ness" as couples' resilience (Fergus, 2015). We, however, note that the range of asymmetric we-talk was generally small compared to those of asymmetric I-and you-talk in all interaction tasks. This suggests that we-talk was mostly balanced between partners in the present study. While this may be due to the rather satisfied sample studied, it could have potentially undermined the effects of asymmetric couple-focus in this study.
Altogether, the results allude to couple-focus and self-focus as valuable means for constructive conflict discussion. The findings encourage to not only replace you-statements with I-statements (Epstein & Baucom, 2013) but also to monitor westatements in discussions of disagreement. Possibly, healthy conflict discussions comprise of a balance between couple-focus and self-focus (see also Rentscher et al., 2013), along with low partner-focus.

Gender differences in pronoun use findings
When looking at language use during the different situations, our results revealed some gender differences, suggesting that the instructions may not have triggered completely parallel responses in women and men. While both hypotheses about partner-focus as a shared vulnerability and dysfunctional asymmetric partner-focus in conflict were confirmed for relationship climate, the associations with interaction behavior differed by gender. Considering the results of the analyses with separate, though less reliable measures, findings of gender differences did, however, not proof to be robust in the conflict task. Further research is warranted that elicits positive and negative relationship behavior to a fuller extent in order to reliably investigate the possibly distinct trajectories of positive and negative behaviors.
In contrast to the results relying on the positivity ratio, which was the focus of the study, female and male partners showed different associations with asymmetric partnerfocus. In a recent study, couples with aggressive tendencies showed less asymmetric partner-focus compared to non-aggressive couples (Timmons et al., 2021). This hints toward possible interactions between couple and situational characteristics that warrant further research.
In dyadic coping, some additional associations only emerged in the task in which the woman disclosed her stress. For example, I-talk (dyadic level) was related with more positive relationship climate in this task, but not when the man disclosed his stress. Thus, above and beyond asymmetric self-focus by the stressed partner, a high absolute level of self-focus benefited the dyad in the sense of a shared resource when the woman was the support-recipient.
High absolute levels of self-focus might represent the maximization of immersion, paraphrasing and empathizing-techniques that are important for dyadic coping (Bodenmann, 2005;Leuchtmann et al., 2018). Gender differences in stress expression and dyadic coping have been reported previously Dindia & Allen, 1992), and women may perhaps have different expectations and standards of the support they receive, as a possible consequence of their higher sensitivity to the emotional support needs of others (Cutrona et al., 2007). Future research should look at the temporal course of conversations to further clarify gender differences concerning selffocus in stress expression and dyadic coping.

Limitations and future directions
While the present findings illuminate the role of pronoun use as objective indicators of relationship functioning, they encourage the examination of interaction context when studying pronoun use in close relationships. A better understanding of context-specific associations is especially critical in light of the rise of studies that examine couples' interactions in real-life (Karan et al., 2017), where the variety of relational contexts captured is immensely broad.
Our findings should be understood in light of the study's limitations. First, the present sample consisted of couples who were on average satisfied with their relationship. Previous work identified relationship satisfaction as a moderator in associations between pronoun use and relationship outcomes (Williams-Baucom et al., 2010), and further research is needed to replicate our results with couples from different backgrounds (socio-economic and cultural background, relationship satisfaction). Second, our initial intent to model SRF as one latent outcome was not possible due to the independence of relationship climate and interaction positivity. Thus, against our theoretical expectations, positive interaction behaviors did not necessarily go along with more positive relationship climate and modeling these two constructs as separate outcomes accounted for this. Further research is required for a better understanding of different facets of SRF and their measurement before generalizing our findings prematurely. And third, given that effect sizes and explained variance were rather small as is common in language-based research (e.g., Karan et al., 2019), future research will benefit from multi-method approaches that combine language with other behavioral indicators in real life.
Ultimately, we acknowledge that while it would be ideal to use identical outcomes across interaction tasks and test context as a moderator, the present work used distinct measures of interaction positivity in each context. We see it as a challenge and important future direction of the field to develop generic measures assessing positivity in a contextinvariant way.

Conclusion
The current study shows the promise of taking a "dyad-as-a-unit"-perspective when studying couples' pronoun use. As one of the first studies that systematically examined pronoun use across conflict and dyadic coping interactions, our results shed light on some important contextual differences. While couple-focus represented a shared resource in conflict, positive dyadic coping interactions were characterized by a verbal asymmetry directed toward the stressed partner. To the recipe for a functioning and satisfying relationship may thus belong adapting the different relational perspectives of "separateness" and "we-ness" to the demands of the situation. A couples' ability to flexibly shift the focus toward the separate individuals in the relationship might ultimately foster their "we-ness" and sense of togetherness in the long-term. These moments of mutual updating might be necessary for tailoring dyadic coping behaviors to the true needs of the partner. This study opens the door for further validation of linguistic markers of relationship functioning, which will eventually not only help to better understand interaction dynamics in close relationships, but also inform future couple interventions.

Authors' note
During her work on this project, Tabea Meier was a pre-doctoral fellow of LIFE (International Max Planck Research School on the Life Course; participating institutions: MPI for Human Development, Humboldt-Universität zu Berlin, Freie Universität Berlin, University of Michigan, University of Virginia, University of Zurich. Part of the data has been presented as a poster at the Society for Personality and Social Psychology's Annual Convention 2021.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/ or publication of this article: This research has been funded by a grant from the Swiss National Science Foundation (SNF: CRSI11_133004/1) awarded to Guy Bodenmann, Veronika Brandstätter, Mike Martin, Fridtjof W. Nussbeck, and Thomas Bradbury. Moreover, the work of Tabea Meier was supported by the Jacobs Foundation. Tabea

Open research statement
As part of IARR's encouragement of open research practices, the author(s) have provided the following information: This research was pre-registered. The aspects of the research that were preregistered were the hypotheses, statistical analyses and exclusion criteria. The registration was submitted to the Open Science Framework: https://osf.io/fwueh. The data and materials used in the research are available upon request by emailing guy.bodenmann@uzh.ch.

Supplemental material
Supplemental material for this article is available online.

Notes
1. See "files" at https://osf.io/fwueh for preregistration. More details on the statistical hypotheses can be found in the preregistration and "Supplemental Materials A." 2. Couples included in the conflict task versus the rest of the sample did not differ in terms of relationship duration (p ¼ .159). However, partners in the conflict task subsample reported a marginally lower relationship satisfaction (women: p ¼ .059; men: p ¼ .044 on the RAS, compared to those for whom conflict task data was not available/excluded. 3. This was the case in n ¼ 73 women and n ¼ 109 men in the conflict task, as well as n ¼ 298 in the dyadic coping task (woman as support-provider) and n ¼ 289 (man as support-provider). 4. For parsimony reasons, we included all three pronouns (I, we, you) in one model (separately for each interaction task), rather than computing separate models for each pronoun type and interaction task. We then reran all models separately for each pronoun type, which was closer to our preregistered analysis plan. The majority of the effects found were replicated in these additional models (see "Supplemental Material C").