The constructs of we-ness and separateness have long been studied in couples research (Singer et al., 2015). We-ness has been shown to reflect a schema of interdependence, shared responsibility, and partnership, while separateness would reflect one of independence and a focus on the individual spouses rather the couple as a unit. According to the investment model of interdependence, increases in relationship commitment should lead to a sense of we-ness and consequently to a common couple identity in which “mental representations of the couple became more prominent” (Agnew et al., 1998). Researchers have employed natural language usage as an implicit measure of we-ness and separateness. Psycholinguistic research (Pennebaker and Stone, 2003) has shown that the words we speak may reveal components of our psychological state such as emotional feelings, social identity, and cognitive style. Current research that has taken pronoun usage into account in the context of couple relationships was encouraging. First-person plural pronouns (e.g., we, our, us) seemed to reflect shared experience while first-person (e.g., I, my, me), while second-person singular (e.g., you, your) pronouns would reflect individuated experiences. Individuals who preferentially employed we-pronouns rather than you- and me-pronouns perceived their relationships as more intimate and of higher quality (Fitzsimons and Kay, 2004). Seider et al. (2009) found evidence that pronoun use was associated with the emotional quality of couple interactions and to couple satisfaction. Specifically, they reported that we-ness language correlated positively to interactions characterized by low levels of negative emotional behavior and high levels of positive emotional behavior. The opposite pattern of results was found for separateness. Finally, Seider et al. (2009) reported that older couples showed greater levels of we-ness and a greater sense of shared identity because of the greater number of shared experiences.

Until now, the constructs of we-ness and separateness as assessed by natural language and pronoun usage have never been studied in the family context, even though early family interactions, especially, have been seen as an example of a context of interdependence which involves a partial transformation of identity (Agnew et al., 1998). Childbirth implies the arrival of another individual in the family and in self- and relational awareness (Aron et al., 1992). A model of early family interactions has been developed called “Family Alliance” (FA), which concerned “the degree of early family engagement and cooperation in everyday activity involving triadic family interactions (i.e., mother-father-child),” such as playing together or having a meal (Favez et al., 2010). The aim of the current study was to analyze whether we-ness and separateness language were associated with FA.

In the current study, we examined pronoun usage during a minimally structured parental conversation about the parents’ child’s education. This study was thus the first to include a measure of we-ness and separateness arising from conversational text analysis and the observational assessment of early family interactions, all obtained during actual family interactions. According to the investment model of interdependence, and because family is a context of interdependence, we hypothesized that more positive family interactions would be associated with the use of more we-pronouns and fewer I- and you-pronouns by both mothers and fathers. Like Fitzsimons and Kay (2004), we expected that parents who preferentially used we-pronouns as opposed to you- and me-pronouns would show a higher degree of family engagement and coordination. Secondly, like Seider et al. (2009) we expected that, because of shared experiences in couple and family life, long-term couples and multiparous parental couples would use more we-pronouns and fewer separateness-pronouns. Hypotheses were tested in a multilevel modeling framework.

Data were collected within a sample of 47 non-referred French-speaking heterosexual families (n = 31 primiparous families). The children were 30 girls and 17 boys (Age, M = 15.75 months, SD = 2.73). Parents were aged from 23 to 43 years old (M = 28.87, SD = 3.63) and were paired as couples (Relationshipduration, M = 7.63, SD = 3.50). Participants were recruited with the help of seven gynecologists from hospitals who gave information about the current study to their patients verbally and by means of a flyer.

Data were collected using the Lausanne Trilogue Play procedure (LTP; Fivaz-Depeursinge and Corboz-Warnery, 1999) which was a semi-standardized observation play situation including mother, father, and baby together. Experimenters asked parents to sit in front and on each side of their child so that the three were arranged in a triangle. Technical equipment consisted of two cameras, one recording the parents and one the child. Researchers gave the following guidelines: “We’ll ask you to play together as a family in four separate situations. In the first, one of you plays with the child and the other one is simply present. In the second, you reverse the roles. In the third, the three of you will play together. In the last part, you (i.e., parents) will talk a while together about how you raise your child, and it will be the child’s turn to be simply present.” The research team randomly decided who began the game in order to counterbalance any possible order effect between the mother and father. The partners were allowed to interact for as long as they considered necessary, without any time limit. The mean duration of the LTP in this study was 14 min, 34 s (SD = 4.47), instructions included. Before participating in the LTP, parents were asked to separately complete a brief questionnaire on relationship satisfaction. For ethical reasons, this study was registered with the Commission for the Protection of Privacy in Belgium. Parents signed an informed consent and were assured that the collected data would remain confidential.

Data were analyzed using hierarchical linear modeling (HLM 6.08). Our model included two levels of analysis, i.e., a couple-level (level 2 data) and an individual-level (level 1 data) with we-ness and separateness scores separately as the outcomes. Three kinds of predictors were included, i.e., between-dyads, within-dyads, and mixed variables. Between-dyads variables (i.e., variables with the same score for both partners in a couple, but different from couple to couple, i.e., FA functions, couple’s relationship duration, and primiparity/multiparity) were introduced at the level 2 data (i.e., couple-level). Within-dyads variables (i.e., different within a single couple but similar between couples, i.e., gender) and mixed variables (i.e., variation both within the couple and between couples, i.e., relationship satisfaction and age) were introduced at the level 1 data (i.e., individual-level). Age, gender, and relationship satisfaction were used in the statistical model as control variables. Note that because of close correlations between the different FA functions, with 0.35 < r < 0.81, p < 0.001, we used the residuals of the FA functions as between-dyads variables (for more details, see Table 1).

Preliminary analyses indicated that the means, standard deviations, and ranges for the we-ness and separateness measures were 0.04 for both constructs (SD = 0.03 for we-ness and separateness; Range: 0.00–0.15 and 0.00–0.18 for we-ness and separateness, respectively). Significant negative correlations were found between the two types of pronouns both for mothers, r = -0.42, p = 0.003, and fathers, r = -0.57, p = 0.000. Examining the size of the correlation coefficients suggested that there was 18 and 32% shared variance between we-ness and separateness pronouns for mothers and fathers, respectively. The percentage of shared variance led us to treat the constructs of interest separately.

Two conditional models were tested with we-ness and separateness as outcomes, respectively. These models allowed us to test our main hypothesis, which questioned whether more positive family interactions were associated with the use of more we-pronouns and fewer I- and you-pronouns by both mothers and fathers. Table 2 depicts the results of the conditional models of we-ness and separateness with the FA functions as predictors. For we-ness, significant positive effect of affect sharing (β = 0.012, p = 0.009) was found: The more the family shared warmth and positive affect, the more we-ness pronouns were used during parental conversation. For separateness, significant negative effects of participation (β = -0.020, p = 0.007), organization (β = -0.021, p = 0.020), affect sharing (β = -0.028, p = 0.008), and timing (β = -0.022, p = 0.013) were found. A gender-effect (β = -0.010, p = 0.040) was observed: Within the couple, fathers tended to use more separateness pronouns than mothers. Finally, except for the FA functions, the other variables did not play any important role in the prediction.