Convergence of naïve coders and participant report ratings of attachment to a former partner: When we should be me

Abstract

Using a Brunswik lens framework, this study examined whether lay observers could accurately detect participants’ attachment to a former partner following romantic breakup or divorce. We predicted that the ratings of post-breakup attachment (completed after reading participants’ transcribed stream-of-consciousness discussions of the recent separation) would be associated with participants’ self-reported ratings of attachment to a former partner and that participants’ natural language use in their narratives would act as the behavioral residue explaining these associations. To enhance the generalizability of our findings, we explored our hypotheses in two samples of adults who had recently undergone romantic relationship dissolutions—people experiencing non-marital breakups (N = 161) and divorce (N = 132). Consistent with hypotheses, in both samples, naïve judges generated assessments of participant attachment that were (a) reliable, (b) strongly associated with participants’ own attachment ratings, and (c) associated with participant breakup-related distress. Of the linguistic cues we examined as behavioral residue, only we-talk (first person plural pronoun use) operated indirectly to link rater and participant attachment scores. We discuss the implications of this work for a deeper understanding of attachment to former romantic partners and for using person-perception paradigms to study attachment relationships.

Keywords

Attachment behavioral residue breakup divorce

[One obstacle] in particular can turn into a deal breaker faster than you can say, “Adios, dude.” If your new romantic interest isn’t over his ex, the relationship should end in a way that’s decidedly not happily ever after.

—Zahra Barnes, Women’s Health (2015)

With remarkable accuracy, naïve observers can detect important information regarding others’ internal states and characteristics. Studies by Gosling and colleagues, for instance, document impressive convergence in observer- and self-reported personality ratings when observers rely exclusively on seemingly superficial information, including the interior of dorm rooms, personal websites, social media, and music preferences (Gosling, 2008; Gosling, Augustine, Vazire, Holtsman, & Gaddis, 2011; Gosling, Ko, Mannarelli, & Morris, 2002; Mehl, Gosling, & Pennebaker, 2006; Vazire & Gosling, 2004). Gosling argues that observers detect individual differences in others’ characteristics using the so-called behavioral residue of these qualities (Gosling et al., 2002)—importantly, this detection process is thought to operate outside of conscious awareness while also influencing consciously accessible perceptions of others.

In the study of close relationships, however, relatively few studies use the person perception methodology, despite the fact that the nearly instantaneous judgments we render about others’ states of mind and intentions likely have profound consequences for interpersonal relations. Upon hearing just 30 s of adults’ recounting a recent marital separation experience, naïve observers are reliable in their detection of how well the participants were coping with the divorce, and these ratings were predictive of participants’ self-report distress over 90 days (Mason, Sbarra, & Mehl, 2010). Other research finds that compared to trained judges, naïve observers demonstrate commensurate, and at times superior, predictive correlations with participant-rated marital relationship satisfaction (Baucom, Baucom, & Christensen, 2012). Using the narratives of adults experiencing a recent relationship dissolution (similar to the paradigm introduced by Mason et al., 2010), we evaluate the accuracy of naïve observers’ assessments of people’s attachment to their former partners—an important predictor of psychological well-being following relationship dissolution (Masheter, 1997). We also examine language as a form of behavioral residue.

Relationship dissolution

Close relationships play a vital role for human health and psychological well-being, and the ending of committed non-marital (e.g., Larson & Sbarra, 2015) and marital relationships (e.g., Sbarra & Emery, 2005; Sbarra & Nietert, 2009) can be quite distressing. Much of the literature on breakups focuses on factors associated with adjustment following a separation experience (Sbarra & Borelli, 2013; Sbarra & Emery, 2005; Sbarra, Hasselmo, & Bourassa, 2015; Wang & Amato, 2000). Scholars have long argued that Bowlby’s attachment theory can be usefully applied to the study of adjustment following relationship dissolution (Kitson & Holmes, 1992; see Fraley & Shaver, 2016, for a review). In addition to variability in the quality of attachment to one’s partner (Fraley & Davis, 1997; Hazan & Zeifman, 1999; Sbarra & Hazan, 2008), people can also vary in the extent to which they remain attached to a former partner after their relationship ends, which may constitute an informative individual difference variable in predicting adaptation to romantic relationship dissolution and future partner selection.

Relatively, little is known about people’s ongoing attachments to romantic partners following the end of a relationship. Researchers currently operationalize ongoing attachment to a former partner using a variety of metrics—quality and frequency of contact (Brown, Felton, Whiteman, & Manela, 1980; Kitson & Holmes, 1992), separation anxiety regarding former partner (Weiss, 1976), anger and hostility toward partner (Tschann, Johnston, & Wallertstein, 1989; Weiss, 1976), nonacceptance of relationship termination (Sbarra & Emery, 2005), preoccupation regarding the partner (Masheter, 1997), and mourning of the loss of the former partner (Hazan & Shaver, 1992). Findings generally suggest that stronger attachment to a former partner is associated with poorer long-term psychological adjustment (Brown et al., 1980; Masheter, 1991), but the strength of the effects differs depending on the operationalization of attachment (see Masheter, 1997, for a discussion).

Moreover, little is understood about the course and process of becoming unattached to a former partner. Normatively, the process of becoming unattached should progress as the length of time since separation increases and as people begin to reconstitute an independent sense of self (cf. Sbarra & Hazan, 2008). Self-concept disruption is often an inherent part of the process of adapting to a relationship transition (Slotter, Gardner, & Finkel, 2010), and importantly, recovering an independent sense of self is associated with improved psychological adjustment (Larson & Sbarra, 2015; Mason, Law, Bryan, Portley, & Sbarra, 2012), making it an important construct to understand. Conceptually, however, there seems to be an important distinction between having a clear sense of self (after a breakup) and remaining attached to a former partner, with the former largely about the self and the latter largely about the self in relation to one’s former partner. In and of itself, then, self-concept disturbance appears to be an inadequate indicator or assessment of the extent to which a person remains attached to a former partner.

Colloquially, enduring attachment to a former partner is discussed as an undesirable characteristic in potential mates. Empirically, we know little about others’ perceptions of people’s attachment to a former partner, including whether or not people can accurately detect attachment, whether perceptions of others’ attachment are associated with theoretically relevant constructs, and what cues observers use to form their impressions of others’ attachment to former partners. In this report, we begin with the basic question of whether naïve observers recognize participants’ attachment to an ex-partner through exposure to their relationship narratives.

Linguistic cues of attachment to a former partner

Brunswik’s organism-environment model (1956) describes the process through which people perceive or recognize characteristics of an ecological environment. Navigating a complex perceptual landscape necessitates reliance upon proximal cues in order to attempt to gain an understanding of the more distal characteristics of the environment. These proximal characteristics of the environment have cascading impacts on distal characteristics, which in turn are those that are employed by the perceiver to make judgments of varying accuracy by others. Perceiver accuracy is influenced by the extent to which the perceiver makes use of the available cues (referred to as cue utilization) as well as the extent to which these cues provide accurate information about the environment in question (cue validity; Brunswik, 1956).

When considering Brunswik’s model, it is important to note that cues can come in various perceptual forms (e.g., visual, auditory). As perceivers, humans simultaneously process cues across sensory modalities; this process means that it can be complicated, if not impossible, to experimentally determine the extent to which any given cue is responsible for the perceiver’s judgments. Thus, in conducting psychological research on this topic, it can be useful to reduce the number of perceptual domains available to the perceiver in making judgments about another individual. Here, we restrict our inquiries to linguistic cues—we measure naïve observers’ judgments based solely on transcriptions of relationship narratives (open-ended speech samples). Doing so enables us to focus solely on one cue category, thus simplifying the perceptual process as it unfolds in real-life interactions but enabling us to more precisely quantify how specific language use categories may shape the judgment process. In so doing, we recognize that we reduce the external validity of the findings while maximizing internal validity.

Linguistic cues

We focus here on linguistic cues for a number of reasons. First, language is an important modality of communication (Krauss & Chiu, 1998; Miller, 1951) and thus may exert a profound impact on social interactions. Second, there is a robust research literature linking people’s spoken and written word use to their psychological states and traits (see Pennebaker, Mehl, & Neiderhoffer, 2003, for a review), which increases our ability to offer predictions about the types of word categories that could serve as potential cues. Third, linguistic and narrative analysis figure prominently in the formal assessment of adult attachment states of mind (see Hesse, 2016, for an overview of the Adult Attachment Interview [AAI; George, Kaplan, & Main, 1996] coding system), with a large body of research supporting the validity of narrative analysis in predicting theoretically relevant outcomes (Bakermans-Kranenburg & van IJzendoorn, 2009), leading us to believe that linguistic cues may reveal important characteristics of attachment states of mind. Importantly, however, only a handful of studies explore the frequencies of different types of word categories and their association with attachment narratives (Borelli et al., 2010, 2011; Cassidy, Sherman, & Jones, 2012; Waters, Steele, Roisman, Haydon, & Booth-LaForce, 2016) or self-reports (Borelli et al., 2013, 2014), and all of these have focused on the links between word use and quality of attachment (e.g., insecure versus secure, preoccupied versus dismissing, avoidance versus anxiety). To date, no studies have explored word use as a measure of degree or strength of attachment, thus the current study is largely exploratory in nature.¹

In the absence of an established research literature linking the frequency of certain types of word use to strength of attachment to a former partner, we relied upon theory and extant literature on language in romantic relationships to develop hypotheses regarding which types of word categories might be potential cues in the Brunswik model. Our selection of one of the linguistic cues was strongly motivated by theory and prior research, whereas the others were chosen in an exploratory manner.

We-talk

In relationship research, one behavioral cue used to assess relationship quality is the frequency of first-person plural pronouns (e.g., we, us, our) in natural language use, frequently referred to as we–talk. Researchers conceptualize and often operationalize we-talk as an index of communal coping or interdependence of selves—the extent to which two or more people approach situations as a team rather than individually and feel integrated into one another’s lives (Lyons, Mickelson, Sullivan, & Coyne, 1998; Rentscher, Soriano, Rohrbaugh, Shoham, & Mehl, 2015; Rohrbaugh, Mehl, Shoham, Reilly, & Ewy, 2008; Rohrbaugh, Shoham, Skoyen, Jensen, & Mehl, 2012). Within intact relationships, greater we-talk is positively associated with and intra- and inter-individual psychological and health outcomes, including higher relationship satisfaction and perceived unity (Agnew, Van Lange, Rusbult, & Langston, 1998; Robbins, Mehl, Smith, & Weihs, 2013; Rohrbaugh et al., 2008; Simmons, Gordon, & Chambless, 2005; Williams-Baucom, Atkins, Sevier, Eldridge, & Christensen, 2010).

When it comes to assessing language as an index of psychological responding, context is key (Pennebaker, 2013). In an intact relationship, it is arguable that we-talk reflects greater interdependence or a more a more communal orientation. Thus, generally speaking, higher we-talk is associated with better relationship outcomes. However, recently investigators have become interested in the boundaries of these positive associations. During a temporary separation, such as a war-time deployment, for example, higher we-talk measured during the deployment is associated with greater relationship satisfaction, but this same measure of we-talk during the deployment is not associated with relationship satisfaction before or after the deployment (Borelli et al., 2013). These findings suggest that associations between we-talk and relationship quality may be different before, during, and after a separation, which begs the question, what happens when relationships come to an end? Following relationship dissolution, we-talk during discussions of the ex-partner likely takes on a very different meaning, and, we believe, may reflect attachment to a partner who is not available. If maintained over time, this attachment could be associated with greater distress, presumably reflecting less adaptation to the breakup (i.e., a failure to transition toward an independent self-construal, see Agnew, 2000; Sbarra & Emery, 2005).

We are aware of only two studies that have studied we-talk in the context of romantic breakups, the subject of this report. Larson and Sbarra (2015) found that participants who repeatedly responded to questions about their breakup and spoke about it in a stream-of-consciously manner aloud every 2 weeks evidenced significant reductions in self-concept disturbance over time, and this construct operated indirectly to reduce we-talk. In other words, there was a strong positive association between we-talk and (self-reported) self-concept disturbance over time—people who were struggling with their sense of self independent of their former partner tended to use more we-talk when describing the end of their relationship (Larson & Sbarra, 2015). Second, Bourassa, Hasselmo, and Sbarra (2017) recently found that, above-and-beyond a host of individual differences variables (including attachment anxiety and avoidance), we-talk expressed during a stream-of-consciousness (SOC) discussion of a recent marital separation was associated with significantly worse psychology adjustment over four years later.²

Social/relational orientation, emotion words, linguistic indicators of temporal focus

In an attempt to be more comprehensive in our approach to testing a Brunswik lens model, we examined other word categories we believed might also reflect enduring attachment to a former partner. We anticipated that people who were in general more oriented to social relationships, and therefore more frequently used words related to social relationships, would be and appear to be more attached to their former partners. We also anticipated that participants’ emotion word use could be utilized as cues, supported by the finding that recently separated adults using greater negative emotion words have lower self-concept clarity (Slotter et al., 2010). Finally, we reasoned that participants who exhibited a lesser focus on the past, indicated by more frequent past tense verb use, and a greater focus on the present and future, indicated by more frequent present and future tense verb use, would both be and appear to be more attached to their former partners. When tasked with discussing a prior separation experience, an on-topic narrative would primarily focus on the past (and thus involve more past tense verbs); however, we expected that as compared to participants with low levels of attachment to their former partners, the narratives of highly attached participants would include a greater focus on the present and future (operationalized as more frequent use of present and future tense verbs) and less of a circumscribed focus on the past.

The present study

For this exploratory study, we investigated whether naïve observers can accurately perceive others’ (degree of) attachment to a former partner on the basis of a verbal separation narrative. To reduce the possibility that observers would rely on paraverbal expressions of emotion (cf. Scherer, Feldstein, Bond, & Rosenthal, 1985), we assigned judges to rate transcripts of participants’ spoken descriptions of their recent separation experiences; in this way, judges were challenged to make ratings based on the explicit linguistic content of the narratives. This work was guided by the following hypotheses. First, based on prior person-perception research, we anticipated that we would observe a high degree of correspondence among naïve judges’ assessments of participants’ attachment to ex-partners from their transcribed spoken narratives (H1). In this respect, we believed the judges would agree in their assessments of attachment to a former partner. Second, we expected that naïve judges’ ratings of participants’ attachment to their former partners would be associated with the self-reported ratings of their attachment to their ex-partner (H2). Next we sought to test the construct validity of judges’ ratings of participants’ attachment—specifically we predicted that higher judge-rated levels of participant attachment would be linked with (a) more recent breakups, (b) longer-lasting relationships, and (c) more self-concept disturbance in participants (H3).

Fourth, we anticipated that higher levels of attachment to a former partner, as assessed via both participant-rated and judge-rated reports, would be positively associated with linguistic cues (H4): We held a priori hypotheses regarding first person plural pronoun use (higher we-talk associated with stronger attachment) and, in an exploratory fashion, tested several other word categories for their association with attachment (social relationship words, positive emotion words, negative emotion words, and words indicating a temporal focus on the past [indexed by frequent past tense verb use and infrequent present and future tense verbs]). Finally, as evidence of the potential validity of these cues within a Brunswik lens framework, we tested whether the association between self-reported attachment to an ex-partner and judge-rated attachment to an ex–partner was mediated by their mutual association with these linguistic cues (H5).

Method

Participants

The data reported on here derive from two larger studies—Study 1 is an investigation of young adults’ adjustment to non-marital relationship breakup (Larson & Sbarra, 2015) and Study 2 is a study of adults’ adjustment to divorce.

In both studies, prior to data collection, the protocol was approved by the Institutional Review Board. In Study 1, participants were recruited via an undergraduate participant pool at a 4-year research university with questions about whether they had experienced a recent breakup; in addition, to expand the generalizability of the sample, study flyers and advertisements were also distributed at a 2-year community college. To be eligible to participate in the study, participants had to have recently broken up with their romantic partner. The resulting sample (N = 161; M _age = 19.50; SD _age = 1.83) was mostly (72.7%) female and Caucasian (59.6%, with 24.2% Latino, 9.3% Asian American, 2.5% Native American, 2% African American, and 4% other) and had, on average, been separated from their ex-partner for 3.5 months from a 21.02 month-long relationship (see Table 1 for descriptive and demographic data on the sample).

Table 1.
Sample demographics and descriptive statistics/frequencies of study variables.

Study 1 Study 2

Breakup sample Divorce sample

N = 161 N = 132

Variable Mean SD Mean SD

% Female 72.7 (n = 117) — 68.6 (n = 96) —

Age 19.49 1.83 43.08 10.91

% Latino/a 24.2 — 22.1 —

% Caucasian 59.6 — 62.6 —

Relationship length in months 21.02 14.51 177.71 112.37

Time since breakup 3.51 2.63 3.74 2.07

% Initiated breakup 43 — 53.8 —

% Who felt more responsible for ending relationship 39.8 — 37.1 —

% Who are friends with ex 5.9 — 35.4 —

% With weekly contact with ex 49.7 — 61.4 —

% Who have biological children with ex — — 6.80% —

% Who have had intercourse with ex post-breakup 37.9 — 18.9 —

% Involved in new relationship 13.4 — 22.3 —

Self-reported attachment^a 57.12 24.96 43.68 24.01

Judge-rated attachment^a 60.96 19.36 47.36 14.08

^aTotal scores.

In Study 2, participants were recruited through newspaper advertisements, e-mail distributions to the university listserv, divorce recovery support groups, and the local family and conciliation court, resulting in a sample of N = 132 (39 men), with a mean age of 43.08 years old (SD = 10 years, 10 months). On average, participants reported having been married or in a marriage-like relationship for 12 years and having separated from their partner 3 months and 3 weeks before the study session (SD = 3 months and 3 weeks). Most (75%) of the participants self-described as White (with 13% Hispanic, 2% Asian, 2% African American, 6% other, and the remainder declining to state) and earned US$50,000 or less in annual income (78%). Fifty percent of the sample reported that they initiated the separation.

Overall, in the combined analyses presented below collapsing across the studies, the total sample size was 293 participants (82 men).

Procedure

Participant procedure

In both studies, participants provided informed consent prior to beginning. Next, participants reported on their attachment to their former partner and demographics. Participants then provided a speech sample regarding their separation experience. Participants were given an audio recorder into which they were asked to speak. Experimenters then left the room so that participants could complete the SOC task in private, returning when four minutes had passed. SOCs were then transcribed verbatim.

In Study 1, participants’ SOCs probed their thoughts and feelings regarding their breakup: They responded to four prompts, each displayed on a computer screen for 1 min, related to their breakup experience. The prompts were as follows: (1) When did you first realize you and your partner were headed towards breaking up?; (2) What do you remember about the separation itself, the actual time when you and your former partner separated?; (3) How much contact have you had with your former partner? What kind/s of contact?; And, (4) How has the breakup affected your thoughts and feelings regarding romantic relationships?

In Study 2, participants completed a nearly identical SOC task in which they were asked to speak about their separation experience for four minutes, with no additional prompts provided. Participants were instructed to focus on their strongest thoughts and feelings and to say anything that comes to mind regarding their separation experience but to try to keep talking for the full four minutes and to try to talk about their relationship in some way.

Language use

Research assistants transcribed the SOCs. The Linguistic Inquiry Word Count (LIWC: Pennebaker, Francis, & Booth, 2015) is a computer program that analyzes a transcribed document and compares it to an internal dictionary, assigning each word to particular linguistic categories (e.g., psychological and grammatical), resulting in a report of the percentage of total words in each category. In the current study, we examined word categories that we anticipated would be indicative of ongoing attachment to one’s ex-partner—(1) first-person plural (we, ours, our) language, negative emotion words (sad, angry), positive emotion words (happy, excited), social process words (family, friends, mate), words evoking a focus on the past (past tense verbs, such as ran, loved, spent), words indicated a focus on the present (present tense verbs, such as like, aim, desire), and future-focused words (future tense verbs, such as will, going to). The LIWC program provides a reliable and valid measure of verbal behavior via text (Tausczik & Pennebaker, 2010) and has previously been used to analyze language on the breakup (Larson & Sbarra, 2015) and divorce SOC tasks (Lee, Sbarra, Mason, & Law, 2011).

Attachment to a former partner

In both studies, participants and naïve judges reported their perceptions of participants’ attachment to their ex-partners using a modified version of the WHO-TO scale (Fraley & Davis, 1997; Hazan & Zeifman, 1994), which assesses a range of attachment features and functions (see Tancredy & Fraley, 2006). This 16-item scale asked participants to rate the extent to which they remained attached to their former partner as reflected in their experience of separation distress, as well as their ongoing desire to use their former partner as a secure base and to seek safety with, and proximity to, this person (I miss my ex-partner when he/she is away). Items were rated on a 7-point Likert-type scale (1 = strongly disagree to 7 = strongly agree) and the scale was scored in this study as the average of responses on each item across the entire scale, with higher scores reflecting greater self-reported attachment to a former partner. The self-reported attachment scale items demonstrated strong reliability, α_Study1 = .94, α_Study2 = .94.

Self-concept disturbance

Participants completed the 12-item Loss-of-Self-Recovery-of-Self Scale (Lewandowski & Bizzoco, 2007), designed to measure self-concept following relationship dissolutions. Instructions prompt respondents to indicate, using a one to seven Likert-type scale, the extent to which statements regarding loss (I do not feel like myself anymore) and recovery of self-concept (I have regained my identity) describe them. For the purposes of this study, we reverse-scored the 6 recovery-of-self items and created a mean score of all 12 items, with higher scores indicating greater self-concept disturbance (see Larson & Sbarra, 2015). Internal consistency of this measure’s items in both studies was strong, α_Study1 = .88, α_Study2 = .91.

Rating procedures

Study 1

The lead researcher explained the study protocol to the three naïve judges (one man and two women), consisting of the following: Judges would be presented with a series of transcripts in which the participants described their breakup experiences. Their task was to read each transcript and then complete the modified WHO-TO questionnaire as if they were the individual speaking in the transcript; in other words, their task was to assess the narrators’ degree of current attachment to their former partner. The judges were blind to all information about study participants and did not have a background in attachment theory. Internal consistency in WHO-TO scores was good in all three judges, α_rater1 = .96, α_rater2 = .95, α_rater3 = .96. To conduct analyses, we first computed a total score from each rater’s score for each transcript. Then, we created a mean judge-rated WHO-TO score by taking the average across the three judges.

Study 2

The lead researcher explained the study protocol to the judges (N = 117: 21 men, 95 women, 1 transgender), who were all undergraduate students at a large public university. In this study, each coder was assigned 27 participant transcripts to read and evaluate using the WHO-TO. The transcripts were distributed across the judges such that each transcript was coded by a large number of students (N _judges ranged from 19 to 21). As with the rating protocol for Study 1, in Study 2, before beginning the rating, judges were informed of the following protocol: They would be presented with a series of transcripts in which participants described their divorces. The judges’ task was to read each transcript and then complete the WHO-TO questionnaire as if they were the individual speaking in the transcript. These judges were blind to all information about study participants and did not have a background in attachment theory. Internal consistency in WHO-TO scores was good in the judges’ scores, ranging from α = .75 to α = .92. As in Study 1, to conduct analyses, we first computed a total score from each rater’s score for each transcript. Then, we created a mean judge-rated WHO-TO score by taking the average score for each participant across the group of judges (N _judges ranged from 19 to 21).

Data analytic plan

The data included in this article derives from two independent data collections, which we combined in order to provide a more comprehensive and thus more generalizable test of our research hypotheses. To test H1, in which we examined the inter-rater reliability of the naïve observers, we computed Intraclass Correlation Coefficients (ICCs) separately by sample due to the fact that the number of judges and the relationships narratives used differed across sample. To test H2–4, we created a combined data set including the mean WHO-TO scores (by averaging the total WHO-TO scores across all judges of any given narrative) for each narrative in the two different samples. To account for sample-specific variability (e.g., divorced adults may in general exhibit higher levels of attachment to their ex-partner than young adults undergoing a breakup due to the longevity of the relationship), we took three important steps: First, we standardized all attachment and linguistic variables within each sample. We then used these within-sample standardized variables in all subsequent analyses. Second, we controlled for sample (0 = breakup sample, 1 = divorce sample) in all analyses. We tested H2 and H3 using partial correlations (controlling for sample) and tested H4 using a hierarchical linear regression in which we tested for the presence of indirect effects while also statistically controlling for any differences between the two samples. Third, we tested sample as a moderator of all paths in the mediation analyses (x to y path, x to m path, and m to y path). None of these moderation effects were statistically significant, and as a result, we do not report these analyses here.

Results

In Study 1, independent samples t-tests did not reveal any differences in key study variables as a function of participant sex. Zero-order correlations indicated that higher self-reported WHO-TO scores were found among participants with more recent breakups and who were not in a new relationship (see Table 2).

Table 2.
Zero-order correlations among study variables in Study 1 and Study 2.

Study Variable 1 2 3 4 5 6 7 8

Study 1 1. Female sex — −.2 .16^* −.09 −.02 .03 −.06 −.01

Breakup sample 2. Age — .14 .02 .08 .13 .09 .07

3. Relationship length — .13 .07 .14 −.02 −.08

4. Time since breakup — .19* .21^** −.19^* −.21^

5. Had intercourse with ex — .43^* .12 .08

6. Involved in new relationship — −.16^* −.18^*

7. Self-reported attachment — .68^*

8. Judge-rated attachment —

1 2 3 4 5 6 7 8

Study 2 1. Female sex — −.15 −.08 −.002 .07 −.01 −.26 −.27

Divorce sample 2. Age — .60 .04 .24 −.13 .08 .11

3. Relationship length — .04 .15 .01 −.04 −.02

4. Time since divorce — −.09 .16 −.04 0

5. Had intercourse with ex — .07 −.10 −.17

6. Involved in new relationship — −.04 −.06

7. Self-reported attachment — .64^*

8. Judge-rated attachment —

p < .05. p < .01. p < .001.

In Study 2, independent samples t-tests revealed that male participants characterized themselves as being significantly more attached to their ex-partners than female participants, t(132) = 3.07, p = .003, and judges also perceived them as being significantly more attached than female participants, t(132) = 3.20, p = .002.

Hypothesis testing

H1: Inter-rater reliability on judge-rated attachment

In Study 1, the average inter-rater reliability of three untrained judges in completing modified WHO-TO questionnaires was high, ICC(1,3) = .84, and ICC(1,1) for the three different rater pairs were .74, .77, and .81. In Study 2, inter-rater reliability of the judges across the whole sample was high for each of the five groups of judges, ranging from ICC(1, 23) = .92 to .96, with the ICC(2, 1) ranging from .56 to .75.

H2: Convergence of judge-rated and self-reported attachment

For the remainder of the analyses, we present findings using the combined sample. All analyses controlled for sample and used scores for each variable that were standardized within sample.

Judge-rated and participant-rated WHO-TO scores were strongly positively associated, r = .66, p < .00001, 95% confidence interval (CI) [.59, .73]³. Sample did not moderate these associations. From the transcripts alone, judges were largely accurate in their judgments of participants’ attachment to a former partner.

H3: Construct validity of judge-rated attachment

Partial correlations revealed that controlling for sample, greater judge-rated attachment was associated with less time elapsed since the separation, r = −.14, p = .02, but not length of the relationship itself, r = −.05, p = .35, and with greater self-concept disturbance, r = .21, p < .0001. Sample did not moderate these associations, and none of these associations held when controlling for participants’ self-reported attachment scores. Given the high correlation between self-reported and judge-rated attachment to a former partner, the judge-rated scale was not a significant predictor of self-concept disturbance after accounting for self-reports of attachment.

H4: Examination of candidate linguistic cues

As reported in Table 3, both participant-reported and judge-rated attachments were significantly positively associated with greater we-talk, lesser past tense words, and greater present tense words during the breakup/divorce SOC speech samples. In addition to these linguistic variables that were associated with both participant- and observer-rated attachment, greater positive emotion word use was uniquely associated with higher observer-rated attachment. Sample did not moderate these associations.

Table 3.
Partial correlations among self- and coder-reported attachment to ex-partner with self-concept disturbance and linguistic cue variables, controlling for sample.

Variable 1 2 3 4 5 6 7 8 9 10

Attachment to Ex 1. Self-reported attachment .66* 0.25* .18 −0.07 0.11 0.03 −.23* .15 −0.04

2. Judge-rated attachment — 0.21* .22* −0.05 .16 −0.02 −.25* .18 0.03

Predictive validity 3. SCD — −0.05 −0.09 0.03 0.12 −0.02 −.23 .22

Candidate linguistic cues 4. We-talk — −.22* −0.06 .26* .12* −.18** −0.06

5. Negative emotion — −0.01 −0.01 −0.09 .12* 0.03

6. Positive emotion — −0.05 −.23*** .14* .13*

7. Social — .22^* −0.07 −.23*

8. Past focus — −.67^*** −.12*

9. Present focus — 0.09

10. Future Focus

Note: SCD = self-concept disturbance.

p < .05; p < .01; p < .001.

Given that participants who used more first-plural words, fewer past tense words, and more present tense words when reflecting over their separation experience both reported and were judged to be more attached to their former partner, we decided to examine these linguistic variables as the potential cues judges used when making their evaluations of participants’ attachment to a former partner.

H5: Linguistic cues as behavioral residue

Here, we were interested in identifying whether the linguistic cues that were associated with both participant- and judge-rated WHO-TO scores operated indirectly to explain the association between these two constructs. Given that length of time since breakup and whether or not the participant was in a new relationship were associated with both WHO-TO variables, we controlled for them in an initial regression—however, neither contributed significantly after controlling for the other variables and, together, they did not alter the indirect effect; therefore, we present the more parsimonious model here.

We used PROCESS Model 4 (Hayes, 2012), which employs a linear regression framework, to test for the presence of indirect effects in a multiple mediation framework; we included the three language variables associated with both self- and judge-rated attachment in a multiple mediator model as represented in Figure 1. Note that our intention in testing for indirect effects was not linked to a desire to establish temporal precedence, but rather to identify whether the variance in participant- and judge-rated WHO-TO overlapped with the presence of the potential linguistic cues.

Figure 1.
Visual depiction of the proposed (a) and observed (b) indirect effects model.

After controlling for whether participants were in the breakup or divorce sample, as well as participant age and gender, we observed a non-zero total indirect effect in which, as a set, the linguistic cues partially explained the association between self-reported and judge-rated WHO-TO scores, b = .05, 95% CI [.02, .09]. The c’ path was reduced but still significant when including the indirect effects, b = .60, 95% CI [.51, .69]. Analyses of the individual linguistic cues revealed that we-talk significantly contributed to the indirect effect, b = .02, 95% CI [.005, .06], but not past, b = .02, 95% CI [−.003, .06], or present tense words, b = .01, 95% CI [−.01, .04]. Thus, the positive association between participant-reported and judge-rated attachment to a former partner was partially explained by their mutual association with higher levels of we-talk.

Discussion

Using a cross-sectional, person-perception research framework, the current study combined data from two samples (N = 293) in which participants experienced a recent romantic separation to explore ongoing attachment to a former partner. We found that (a) across both samples, naïve observers can reliably detect participants’ attachment to a former partner; (b) observer ratings of attachment to a former partner were associated with self-reported attachment to a former partner; (c) stronger observer ratings of attachment were predicted by shorter time elapsed since the breakup and observer ratings of attachment were positively associated with participant’s self-concept disturbance (over-and-above self-reported attachment to a former partner); (d) linguistic cues (we-talk, past tense, and present tense words) were significantly associated with both judge-rated and self-reported attachment; and (e) we-talk (i.e., first-person plural word use) operated indirectly to explain the association between self-reported and judge-rated attachment to a former partner.

Interrater and observer-participant correspondence

In two different samples, naïve observers’ codes of attachment to a former partner were internally consistent, reliable with those of other judges, and strongly positively associated with participants’ self-reports of the same construct. The same pattern of findings held when naïve judges evaluated the breakup narratives of young adults (whose demographic backgrounds, including age and life stage were more closely matched those of the judges’) and the divorce narratives of middle aged adults (whose demographic backgrounds differed from the majority of the judges). The effects held when we used only three naïve judges and when we used more than 19. One of the most interesting things about the pattern of effects emerging from this study pertains to anecdotal accounts of the judges, who repeatedly reported (to the lead researcher) that while completing the task, they did not feel they had an accurate sense of the participants’ states of mind. Thus, the high levels of correspondence both across judges and between judges and participants existed despite of the fact that the phenomenological experience of the judges was often one of semi-confusion and doubt. Of course, these are only anecdotal reports of judges; an empirical test of this observation is necessary in future studies. It would also be interesting to explore whether judges’ confidence in their scores moderates the association between self-reported and observer-rated attachment and/or varies as a function of their attachment style or cultural backgrounds.

Importantly, observer ratings of attachment were positively associated with theoretically related constructs, including shorter time since breakup and more self-concept disturbance. These findings also underscore the potential importance of the construct of enduring attachment to a former partner by documenting a theoretically consistent association between attachment and self-concept disturbance, which itself is associated with a variety of negative outcomes following romantic separations (Larson & Sbarra, 2015; Mason et al., 2012). In sum, these preliminary findings hint at the possibility that naïve observers can, with some of accuracy, detect a person’s risk for maladjustment following breakup. Using a similar coding approach, Mason, Sbarra, and Mehl (2010) demonstrated that judges are reliable in their detection of how adults are coping with their separation. The present findings extend this work by illustrating, with an independent sample, that judges are also reliable in detecting the conceptual related construct of attachment to a former partner.

Linguistic cues

Brunswik’s (1955) model suggests that observer judgments of ecological environments are discerned through behavioral cues that provide the lens for perceiving the underlying construct. This theory provides a framework for understanding the complex process of person perception—the means through which people detect the characteristics of others. Brunswik (1943) argued that because we cannot perceive actual characteristics themselves, we rely upon cues to inform us of individual differences across environments (in this case, across people). In the current study, we found that after relationships dissolve, when people discuss their separation experiences, greater we-talk (i.e., first-person plural words) and verb tense are cues that provide reliable information about degree of attachment to former partners. In other words, these cues evidence both cue validity (meaning that they are associated with participant ratings of attachment) and cue utilization (meaning that they are associated with observer ratings of participant attachment). And although greater use of present tense words and lesser use of past tense words were also associated with both participant- and observer-ratings, when considered in a model simultaneously with the other two linguistic categories, the indirect effect through these variables was not reliably different from zero.

We had an a priori (though not preregistered) hypothesis that higher levels of we-talk would be one linguistic cue through which observers’ and participants’ ratings of attachment would be linked; this hypothesis was supported. When participants used greater levels of first person plural pronouns in describing their relationship dissolutions, they also reported higher levels of attachment to their former partners. In addition, naïve observers perceived the participants with higher we-talk usage as being more attached to their former partners. We are aware of no prior studies finding evidence that attachment to a former partner is associated with greater we-talk as people describe their separation experience. Higher we-talk may constitute an implicit (or non-conscious) sense of the self as part of a plural unit (see Agnew, 2000). Thus, the findings provide some evidence of correspondence across we-talk (which we consider to be a non-consciously monitored aspect of behavior) and self-reported attachment to one’s former partner (a consciously monitored aspect of one’s behavior).

The link between we-talk and attachment suggests that greater first person plural word use may reflect and convey a form of difficulty separating from a former partner, in other words, despite the fact that the relationship has ended, these individuals still construct their sense of self as part of a couple unit. This lack of differentiation in self can be indicative of a refusal or inability to adapt one’s self-concept to the new circumstances. Importantly, in exploring these associations only following a separation, our study neglects a consideration of pre-dissolution participant characteristics. For instance, it is possible that a high we-talk participant has decreased more in we-oriented psychological stance since the relationship dissolution than a low we-talk participant, which could suggest greater adaptation to the separation. Perhaps the inclusion of time-related changes in attachment accounts for the (slightly) weaker association between participant-reported attachment versus observer-reported attachment with we-talk. Participants may have a fuller view of their trajectory of attachment from pre- to post-dissolution or across the time since the relationship dissolution, whereas naïve observers rated narratives as isolated snapshots of participants’ psychological states with no opportunity to consider contextual factors (e.g., change, time, gender, culture). Nonetheless, it is the naïve observer perceptions that may have the strongest implications for participants’ future interpersonal experiences with new people (for instance, in new dating situations).

Importantly, the naïve judges detected these cues from written transcripts of participants’ breakup interviews, protecting against the possibility that paraverbal processes created an artificial association (e.g., if we-talk was associated with emotional distress, and this emotion distress was conveyed in language that could be easily heard; see Mason et al., 2010). In line with our prior suggestion, in future studies it would be informative to explicitly direct judges’ attention to these linguistic variables—conducting an experimental study in which these instructions were manipulated (i.e., one group is told to pay attention to these linguistic cues and the other receives no instructions regarding linguistic cues) could reveal whether awareness of the cues heightens or attenuates judges’ accuracy in attachment ratings.

Implications and future directions

The rating procedure used in this research also has promise for extending research on attachment to a former partner as an important individual difference variable that may explain coping in the wake of a breakup (see Masheter, 1997). In future work, it will be important to assess whether naïve observers can accurately detect this quality in narratives regarding topics unrelated to the former relationship, or whether this level of accuracy in perception is apparent only following exposure to relationship narratives.

In addition, we believe that perceptions of others’ attachment have important implications for person perception. Studies suggest that first perceptions guide people’s behavior in close relationships in important ways (e.g., Tidwell, Eastwick, & Finkel, 2013): Naïve observers are remarkably accurate at detecting individual differences in strangers’ personalities using seemingly small and insignificant information (Gosling et al., 2002, 2011). However, we know almost nothing about how this process occurs with respect to relationship relevant constructs. And yet, there are compelling reasons why understanding this process is important. For instance, in both the empirical and the clinical literatures (see Messman-Moore & Long, 2003 for a review), scholars observe that adults tend to repeat the relational patterns they have experienced before—for instance, people who were abused as children are more likely to find themselves in abusive romantic relationships (Fiorillo, Papa, & Follette, 2013). Indeed, this was one of Bowlby’s (1982) core predictions—that there is continuity in attachment relationships across the life span, with the internal working model serving as a bridge between former and future relational experiences. And this continuity can occur in spite of a consciously expressed desire not to choose partners who resemble former attachment figures. Researchers endeavor to explain how this compulsion to repeat relational patterns is enacted, but as of yet, we have no evidence of the means by which this process occurs. If attachment internal working models largely reside in unconscious memories, as theorists contend (e.g., Main, 2000), then it is likely that the forces that attract one individual to another are also largely unconscious, and therefore difficult to interrupt. However, the understanding of these dynamics is important, as it offers the possibility of assisting adults wishing to avoid repeating maladaptive relational patterns. In the current study, we take one small step in this direction by exploring the extent to which naïve observers can detect others’ attachment to a former partner, as well as the potential linguistic variables that may act as a behavioral residue of this construct.

Limitations

The findings reported here should be considered in light of several limitations. First, our analyses were largely exploratory. Although our hypotheses were guided by theory, the findings should be considered tentative until replicated using a pre-registered confirmatory approach. Second, despite the fact that the sample of men was nearly double the sample size reported on in most breakup studies (combined sample n = 82; see Sbarra & Emery, 2005), the samples were largely female, and the extent to which these findings generalize to men remains unclear. Third, in an effort to study the construct validity of attachment to an ex-partner, we only examined the associations between attachment and linguistic cues in the current study, and this analysis was limited to a single channel of information for the observers. In this case, in the mediation analysis that included the cue variables, the direct effect linking the two assessments of attachment remained reliably different from zero, suggesting that our analyses of specific linguistic cues captured only about 10% of the variance between the two constructs. In future studies, it will be informative to discern how much additional variance between the two methods is explained when other channels of information are available to the observers.

Finally, these data derive from a single, cross-sectional assessment; thus, it is unclear from whether the linguistic cues observed in this sample are specific to discussions of the former relationship or indicative of a more stable or generalized narrative discourse style—for instance, we wonder whether participants using different proportions of we-talk in this sample also would have used different proportions of we-talk when describing an ongoing romantic relationship and/or a topic that is not relational. In other words, although we-talk is associated with positive relational qualities and health benefits in intact relationships (e.g., Rohrbaugh et al., 2008), does using more we-talk when describing a relationship confer risk for ongoing attachment and maladjustment should that relationship dissolve? Further, can naïve observers perceive ongoing attachment to former partners from language samples about topics unrelated to the ex-partner, and, if so, how? Are other behavioral indicators (e.g., posture, vocal tone) telltale signs to naïve observers of one’s enduring attachment to an ex-partner? The answers to these questions and many more ahead of them will advance our understanding of romantic attachment across different phases of relationships.

Conclusion

This research used a person-perception paradigm to investigate attachment to a former partner in two samples following relationship dissolution. Observers were reliable, and their ratings were robustly correlated with participants’ self-reported attachment to their former partner, effects that were partially linked through their association with the degree of we-talk words in the narratives. This work provides an intriguing glimpse into the way in which observers perceive people’s ongoing attachment to a former partner, and provides a strong basis for concluding that, after a romantic separation, discussions of the break-up characterized by higher levels of first-person plural word use likely serve as the behavioral residue for this process.

	Study 1	Study 2
% Female	72.7 (n = 117)	—	68.6 (n = 96)	—
Age	19.49	1.83	43.08	10.91
% Latino/a	24.2	—	22.1	—
% Caucasian	59.6	—	62.6	—
Relationship length in months	21.02	14.51	177.71	112.37
Time since breakup	3.51	2.63	3.74	2.07
% Initiated breakup	43	—	53.8	—
% Who felt more responsible for ending relationship	39.8	—	37.1	—
% Who are friends with ex	5.9	—	35.4	—
% With weekly contact with ex	49.7	—	61.4	—
% Who have biological children with ex	—	—	6.80%	—
% Who have had intercourse with ex post-breakup	37.9	—	18.9	—
% Involved in new relationship	13.4	—	22.3	—
Self-reported attachment^a	57.12	24.96	43.68	24.01
Judge-rated attachment^a	60.96	19.36	47.36	14.08

Study	Variable	1	2	3	4	5	6	7	8
Study 1	1. Female sex	—	−.2	.16^*	−.09	−.02	.03	−.06	−.01
Breakup sample	2. Age		—	.14	.02	.08	.13	.09	.07
3. Relationship length			—	.13	.07	.14	−.02	−.08
	4. Time since breakup				—	.19*	.21^**	−.19^*	−.21^**
	5. Had intercourse with ex				—	.43^***	.12	.08
	6. Involved in new relationship					—	−.16^*	−.18^*
	7. Self-reported attachment						—	.68^***
	8. Judge-rated attachment							—
		1	2	3	4	5	6	7	8
Study 2	1. Female sex	—	−.15	−.08	−.002	.07	−.01	−.26**	−.27**
Divorce sample	2. Age		—	.60**	.04	.24**	−.13	.08	.11
3. Relationship length			—	.04	.15	.01	−.04	−.02
	4. Time since divorce				—	−.09	.16	−.04	0
	5. Had intercourse with ex				—	.07	−.10	−.17
	6. Involved in new relationship					—	−.04	−.06
	7. Self-reported attachment						—	.64^***
	8. Judge-rated attachment							—

	Variable	1	2	3	4	5	6	7	8	9	10
Attachment to Ex	1. Self-reported attachment		.66***	0.25***	.18**	−0.07	0.11	0.03	−.23***	.15**	−0.04
2. Judge-rated attachment		—	0.21***	.22***	−0.05	.16**	−0.02	−.25***	.18**	0.03
Predictive validity	3. SCD			—	−0.05	−0.09	0.03	0.12	−0.02	−.23**	.22**
Candidate linguistic cues	4. We-talk				—	−.22***	−0.06	.26***	.12*	−.18**	−0.06
5. Negative emotion					—	−0.01	−0.01	−0.09	.12*	0.03
6. Positive emotion						—	−0.05	−.23***	.14*	.13*
7. Social							—	.22^***	−0.07	−.23***
8. Past focus								—	−.67^***	−.12*
9. Present focus									—	0.09
10. Future Focus

Footnotes

Acknowledgements

The authors wish to thank the naive observers who rated the narratives used in these studies and the participants in the original data collections.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Jessica L. Borelli

Notes

References

Agnew

C. R.

(2000). Cognitive interdependence and the experience of relationship loss. In Harvey

J. H.

Miller

E. D.

(Eds.), Loss and trauma: General and close relationship perspectives (pp. 385–398). New York, NY: Brunner-Routledge.

Agnew

C. R.

Van Lange

P. M.

Rusbult

C. E.

Langston

C. A.

(1998). Cognitive interdependence: Commitment and the mental representation of close relationships. Journal of Personality and Social Psychology, 74, 939–954. doi:1.1037/0022-3514.74.4.939

Bakermans-Kranenburg

M. J.

van IJzendoorn

M. H.

(2009). The first 10,000 adult attachment interviews: Distributions of adult attachment representations in clinical and nonclinical groups. Attachment and Human Development, 11, 223–263.

Baucom

K. J. W.

Baucom

B. R.

Christensen

(2012). Do the naïve know best? The predictive power of naïve ratings of couple interactions. Psychological Assessment, 24, 983–994. doi:1.1037/a0028680

Bourassa

K. J.

Hasselmo

Sbarra

D. A.

(2017). After the end: Linguistic predictors of psychological distress four years after marital separation. Manuscript submitted for publication.

Borelli

J. L.

David

D. H.

Rifkin-Graboi

Sbarra

D. A.

Mehl

Mayes

L. C.

(2013). Language use in the adult attachment interview: Evidence for attachment-specific emotion regulation. Personal Relationships, 20, 23–37. doi:10.1111/j.1475-6811.2012.01394.x

Borelli

J. L.

Sbarra

D. A.

Mehl

David

D. H.

(2011). Experiential connectedness in children’s attachment interviews: An examination of natural word use. Personal Relationships, 18, 341–351. doi:10.1111/j.1475-6811.2010.01294.x

Borelli

J. L.

Sbarra

D. A.

Randall

A. K.

Snavely

J. E.

St. John

H. K.

Ruiz

S. K.

(2013). Linguistic indicators of wives’ attachment security and communal orientation during military deployment. Family Process, 52, 535–554. doi:10.1111/famp.12031

Borelli

J. L.

Sbarra

D. A.

Randall

A. K.

Snavely

J. E.

McMakin

D. L.

Coffey

J. K.

… Chung

S. Y.

(2014). With or without you: Preliminary evidence that attachment avoidance predicts non-deployed spouses’ reactions to relationship challenges during deployment. Professional Psychology: Research & Practice, 45, 478–487. doi:10.1037/a0037780

10.

Bowlby

(1982). Attachment and loss: Vol. 1. attachment (2nd ed.). New York, NY: Basic Books.

11.

Brown

Felton

B. J.

Whiteman

Manela

(1980). Attachment and distress following marital separation. Journal of Divorce, 3, 303–317. doi:10.1300/J279v03n04_01

12.

Brunswik

(1943). Organismic achievement and environmental probability. Psychological Review, 50, 255–272. doi:1.1037/h0060889

13.

Brunswik

(1955). Representative design and probabilistic theory in a functional psychology. Psychological Review, 62, 193–217. doi:1.1037/h0047470

14.

Brunswik

(1956). Perception and the representative design of psychology experiences. Berkeley: University of California Press.

15.

Cassidy

Sherman

L. J.

Jones

J. D.

(2012). What’s in a word? Linguistic characteristics of Adult Attachment interviews. Attachment & Human Development, 14, 11–32. doi:10.1080/14616734.2012.636649

16.

Fiorillo

Papa

Follette

V. M.

(2013). The relationship between child physical abuse and victimization in dating relationships: The role of experiential avoidance. Psychological Trauma: Theory, Research, Practice, and Policy, 5, 562. doi:10.1037/a0030968

17.

Fraley

R. C.

Davis

K. E.

(1997). Attachment formation and transfer in young adults’ close friendships and romantic relationships. Personal Relationships, 4, 131–144. doi:10.1111/j.1475-6811.1997.tb00135.x

18.

Fraley

C. R.

Shaver

P. R.

(2016). Attachment, loss, and grief: Bowlby’s views, new developments, and current controversies. In Cassidy

Shaver

P. R.

(Eds.), Handbook of attachment: Theory, research, and clinical applications (3rd ed., pp. 40–62). New York, NY: Guilford Press.

19.

George

Kaplan

Main

(1984–1996). Adult attachment interview protocol (3rd ed.). Berkeley, CA: Department of Psychology, University of California.

20.

Gosling

S. D.

(2008). Snoop: What your stuff says about you. New York, NY: Basic Books.

21.

Gosling

S. D.

Augustine

A. A.

Vazire

Holtzman

Gaddis

(2011). Manifestations of personality in online social networks: Self-reported Facebook-related behaviors and observable profile information. Cyberpsychology, Behavior, and Social Networking, 14, 483–488. doi:1.1089/cyber.201.0087

22.

Gosling

S. D.

S. J.

Mannarelli

Morris

M. E.

(2002). A room with a cue: Judgments of personality based on offices and bedrooms. Journal of Personality and Social Psychology, 82, 379–398. doi:1.1037/0022-3514.82.3.379

23.

Hayes

A. F.

(2012). PROCESS: A versatile computational tool for observed variable mediation, moderation, and conditional process modeling [White paper]. Retrieved from http://www.afhayes.com/public/process2012.pdf

24.

Hazan

Shaver

P. S.

(1992). Broken attachments: Relationship loss from the perspective of attachment theory. In Orbuch

T. L.

(Ed.), Close relationship loss (pp. 90–110). New York, NY: Springer-Verlag.

25.

Hazan

Zeifman

(1994). Sex and the psychological tether. In Bartholomew

Perlman

(Eds.), Attachment processes in adulthood (pp. 151–178). London: Jessica Kingsley.

26.

Hazan

Zeifman

(1999). Pair bonds as attachments: Reevaluating the evidence. In Cassidy

Shaver

P. R.

(Eds.), Handbook of attachment: Theory, research, and clinical applications (2nd ed., pp. 436–455). New York, NY: The Guilford Press.

27.

Hesse

(2016). The adult attachment interview: Protocol, method of analysis, and selected empirical Studies: 1985-2015. In Cassidy

Shaver

P. R.

(Eds.), Handbook of attachment: Theory, research, and clinical applications (3rd ed., pp. 553–597). New York, NY: Guilford Press.

28.

Howes

Spieker

(2008). Attachment relationships in the context of multiple caregivers. In Cassidy

Shaver

P. R.

(Eds.), Handbook of attachment: Theory, research, and clinical applications (2nd ed., pp. 317–332). New York, NY: The Guilford Press.

29.

Kitson

G. C.

Holmes

W. M.

(1992). Perspectives on marriage and the family. Portrait of divorce: Adjustment to marital breakdown. New York, NY: Guilford Press.

30.

Krauss

R. M.

Chiu

(1998). Language and social behaviour. In Gilbert

D. T.

Fiske

S. T.

Lindzey

(Eds.), Handbook of social psychology (4th ed., pp. 41–88). Boston, MA: McGraw Hill.

31.

Larson

G. M.

Sbarra

D. A.

(2015). Participating in research on romantic breakups promotes emotional recovery via changes in self-concept clarity. Social Psychological and Personality Science, 6, 399–406. doi:1.1177/1948550614563085

32.

Lee

L. A.

Sbarra

D. A.

Mason

A. E.

Law

R. W.

(2011). Attachment anxiety, verbal immediacy, and blood pressure: Results from a laboratory analog study following marital separation. Personal Relationships, 18, 285–301.

33.

Lewandowski

G. W.

Jr Bizzoco

N. M.

(2007). Addition through subtraction: Growth following the dissolution of a low quality relationship. Journal of Positive Psychology, 2, 40–54. doi:10.1080/17439760601069234

34.

Lyons

R. F.

Mickelson

K. D.

Sullivan

M. J.

Coyne

J. C.

(1998). Coping as a communal process. Journal of Social and Personal Relationships, 15, 579–605. doi:1.1177/0265407598155001

35.

Main

(2000). The organized categories of infant, child, and adult attachment: Flexible versus inflexible attention under attachment-related stress. Journal of the American Psychoanalytic Association, 48, 1055–1096.

36.

Masheter

(1991). Postdivorce relationships between ex-spouses: The roles of attachment and interpersonal conflict. Journal of Marriage and the Family, 53, 103–110. doi:10.2307/353136

37.

Masheter

(1997). Healthy and unhealthy friendship and hostility between ex-spouses. Journal of Marriage and the Family, 59, 453–475. doi:10.2307/353483

38.

Mason

A. E.

Sbarra

D. A.

Mehl

M. R.

(2010). Thin-slicing divorce thirty seconds of information predict changes in psychological adjustment over 90 days. Psychological Science, 21, 1420–1422. doi:1.1177/0956797610381507

39.

Mason

A. E.

Law

R. W.

Bryan

A. E. B.

Portley

R. M.

Sbarra

D. A.

(2012). Facing a breakup: Electromyographic responses moderate self-concept recovery following a romantic separation. Personal Relationships. doi:10.1111/j.1475-6811.2011.01378.x

40.

Mehl

M. R.

Gosling

S. D.

Pennebaker

J. W.

(2006). Personality in its natural habitat: Manifestations and implicit folk theories of personality in daily life. Journal of Personality and Social Psychology, 90, 862–877. doi:1.1037/0022-3514.9.5.862

41.

Messman-Moore

T. L.

Long

P. J.

(2003). The role of childhood sexual abuse sequelae in the sexual revictimization of women: An empirical review and theoretical reformulation. Clinical Psychology Review, 23, 537–571. doi:10.1016/S0272-7358(02)00203-9

42.

Miller

G. A.

(1951). Language and communication. New York, NY: McGraw-Hill.

43.

Pennebaker

J. W.

(2013). The secret life of pronouns: What our words say about us. New York, NY: Bloomsbury Press/Bloomsbury Publishing.

44.

Pennebaker

J. W.

Francis

M. E.

Booth

R. J.

(2015). Linguistic inquiry and word count (LIWC): LIWC2015. Mahwah: Lawrence Erlbaum Associates.

45.

Pennebaker

J. W.

Mehl

M. R.

Niederhoffer

K. G.

(2003). Psychological aspects of natural language use: Our words, our selves. Annual Review of Psychology, 54, 547–577.

46.

Rentscher

K. E.

Soriano

E. C.

Rohrbaugh

M. J.

Shoham

Mehl

M. R.

(2015). Partner pronoun use, communal coping, and abstinence during couple-focused intervention for problematic alcohol use. Family Process. Advance online publication. doi:1.1111/famp.12202

47.

Robbins

M. L.

Mehl

M. R.

Smith

H. L.

Weihs

K. L.

(2013). Linguistic indicators of patient, couple, and family adjustment following breast cancer. Psycho-Oncology, 22, 1501–1508. doi:1.1002/pon.316

48.

Rohrbaugh

M. J.

Mehl

M. R.

Shoham

Reilly

E. S.

Ewy

G. A.

(2008). Prognostic significance of spouse we talk in couples coping with heart failure. Journal of Consulting and Clinical Psychology, 76, 781–789. doi:1.1037/a0013238

49.

Rohrbaugh

M. J.

Shoham

Skoyen

J. A.

Jensen

Mehl

M. R.

(2012). We-talk, communal coping, and cessation success in a couple-focused intervention for health-compromised smokers. Family Process, 51, 107–121. doi:1.1111/j.1545-530.2012.01388.x

50.

Sbarra

D. A.

Borelli

J. L.

(2013). Heart rate variability moderates the association between attachment avoidance and self-concept reorganization following marital separation. International Journal of Psychophysiology, 88, 253–260. doi:10.1016/j.ijpsycho.2012.04.004

51.

Sbarra

D. A.

Emery

R. E.

(2005). The emotional sequelae of nonmarital relationship dissolution: Analysis of change and intraindividual variability over time. Personal Relationships, 12, 213–232. doi:1.1111/j.13504126.2005.00112.x

52.

Sbarra

D. A.

Hasselmo

Bourassa

K. J.

(2015). Divorce and health: Beyond individual differences. Current Directions in Psychological Science, 24, 109–113. doi:1.1177/0963721414559125

53.

Sbarra

D. A.

Hazan

(2008). Coregulation, dysregulation, self-regulation: An integrative analysis and empirical agenda for understanding adult attachment, separation, loss, and recovery. Personality and Social Psychology Review, 12, 141–167. doi:1.1177/1088868308315702

54.

Sbarra

D. A.

Nietert

P. J.

(2009). Divorce and death: Forty years of the Charleston heart study. Psychological Science, 20, 107–113. doi:10.1111/j.1467-9280.2008.02252.x

55.

Scherer

K. R.

Feldstein

Bond

R. M.

Rosenthal

(1985). Vocal cues to deception: A comparative channel approach. Journal of Psycholinguistic Research, 14, 409–425.

56.

Slotter

E. B.

Gardner

W. L.

Finkel

E. J.

(2010). Who am I without you? The influence of romantic breakup on the self-concept. Personality and Social Psychology Bulletin, 36, 147–160. doi:10.1177/0146167209352250

57.

Simmons

R. A.

Gordon

P. C.

Chambless

D. L.

(2005). Pronouns in marital interaction what do “you” and “I” say about marital health? Psychological Science, 16, 932–936. doi:1.1111/j.1467-928.2005.01639.x

58.

Tancredy

C. M.

Fraley

R. C.

(2006). The nature of adult twin relationships: An attachment-theoretical perspective. Journal of Personality and Social Psychology, 90, 78–93.

59.

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. doi:1.1177/0261927X09351676

60.

Tidwell

N. D.

Eastwick

P. W.

Finkel

E. J.

(2013). Perceived, not actual, similarity predicts initial attraction in a live romantic context: Evidence from the speed-dating paradigm. Personal Relationships, 20, 199–215. doi:10.1111/j.1475-6811.2012.01405.x

61.

Tschann

J. M.

Johnston

J. R.

Wallerstein

J. S.

(1989). Resources, stressors, and attachment as predictors of adult adjustment after divorce: A longitudinal study. Journal of Marriage and the Family, 51, 1033–1046. doi:10.2307/353214

62.

Vazire

Gosling

S. D.

(2004). E-Perceptions: Personality impressions based on personal websites. Journal of Personality and Social Psychology, 87, 123–132. doi:1.1037/00223514.87.1.123

63.

Wang

Amato

P. R.

(2000). Predictors of divorce adjustment: Stressors, resources, and definitions. Journal of Marriage and the Family, 62, 655–668. doi: https://www-jstor-org.web.bisu.edu.cn/stable/1566787

64.

Waters

T. E.

Steele

R. D.

Roisman

G. I.

Haydon

K. C.

Booth-LaForce

(2016). A linguistic inquiry and word count analysis of the adult attachment interview in two large corpora. Canadian Journal of Behavioural Science, 48, 78–88. doi:10.1037/cbs0000035

65.

Weiss

R. S.

(1976). The emotional impact of marital separation. Journal of Social Issues, 32, 135–145. doi:10.1111/j.1540-4560.1976.tb02484.x

66.

Williams-Baucom

K. J.

Atkins

D. C.

Sevier

Eldridge

K. A.

Christensen

(2010). ‘You’ and ‘I’ need to talk about ‘us’: Linguistic patterns in marital interactions. Personal Relationships, 17, 41–56. doi:1.1111/j.14756811.20 -1.01251.x

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

	Study 1		Study 2
	Breakup sample		Divorce sample
	N = 161		N = 132
Variable	Mean	SD	Mean	SD
% Female	72.7 (n = 117)	—	68.6 (n = 96)	—
Age	19.49	1.83	43.08	10.91
% Latino/a	24.2	—	22.1	—
% Caucasian	59.6	—	62.6	—
Relationship length in months	21.02	14.51	177.71	112.37
Time since breakup	3.51	2.63	3.74	2.07
% Initiated breakup	43	—	53.8	—
% Who felt more responsible for ending relationship	39.8	—	37.1	—
% Who are friends with ex	5.9	—	35.4	—
% With weekly contact with ex	49.7	—	61.4	—
% Who have biological children with ex	—	—	6.80%	—
% Who have had intercourse with ex post-breakup	37.9	—	18.9	—
% Involved in new relationship	13.4	—	22.3	—
Self-reported attachment^a	57.12	24.96	43.68	24.01
Judge-rated attachment^a	60.96	19.36	47.36	14.08