Men speak more abstractly than women; gender differences were larger for older adults than for teenagers, suggesting that gender differences in communicative abstraction may be reinforced by one’s experiences

Gender differences in communicative abstraction. Joshi, Priyanka D., Wakslak, Cheryl J., Appel, Gil, Huang, Laura. Journal of Personality and Social Psychology, Oct 14 , 2019. https://psycnet.apa.org/doiLanding?doi=10.1037%2Fpspa0000177

Abstract: Drawing on construal level theory, which suggests that experiencing a communicative audience as proximal rather than distal leads speakers to frame messages more concretely, we examine gender differences in linguistic abstraction. In a meta-analysis of prior studies examining the effects of distance on communication, we find that women communicate more concretely than men when an audience is described as being psychologically close. These gender differences in linguistic abstraction are eliminated when speakers consider an audience whose distance has been made salient (Study 1). In studies that follow, we examine gender differences in linguistic abstraction in contexts where the nature of the audience is not specified. Across a written experimental context (Study 2), a large corpus of online blog posts (Study 3), and the real-world speech of congressmen and congresswomen (Study 4), we find that men speak more abstractly than women. These gender differences in speech abstraction continue to emerge when subjective feelings of power are experimentally manipulated (Study 5). We believe that gender differences in linguistic abstraction are the result of several interrelated processes—including but not limited to social network size and homogeneity, communication motives involving seeking proximity or distance, perceptions of audience homogeneity and distance, as well as experience of power. In Study 6, we find preliminary support for mediation of gender differences in linguistic abstraction by women’s tendency to interact in small social networks. We discuss implication of these gender differences in communicative abstraction for existing theory and provide suggestions for future research.

General Discussion

Across a series of six studies, we find that men communicate more abstractly than women. We find that gender differences in communicative abstraction persist across experimental (Studies 1, 2, 5, and 6) and field (Studies 3 and 4) contexts. The effects conceptually replicate across various measures of abstraction, including emphasizing desirability vs. feasibility (Study 2), using more concrete words (Studies 3 and 4), and adopting higher levels of action identification (Studies 5 and 6).

Effects ranged from small to moderate across studies, with larger effect sizes in our controlled laboratory-style studies than our archival data ones. The former studies allowed for greater control, constraining the topic of communication and potential ways in which a participant was able to communicate. They largely also used paradigms that have been specifically designed in prior work to capture variation in communicative abstraction. On the other hand, our archival data studies gave us an opportunity to examine gender differences in contexts that were much less constrained in terms of numerous factors such as the topic of conversation, number of words, intended audience, length of speech, and overall purpose of conversation. Given the size of the corpora involved, they also relied on an automated method for coding abstraction, which gives up some amount of precision compared to hand coding (see Johnson et al., 2019, for a related discussion). Thus, it is not especially surprising to find smaller effects in these contexts.

What Drives these Effects?

We suspect that the gender difference in communicative abstraction we identify emerges from a set of converging reasons, including women’s social interactions in closely knit small groups, their historical occupation of lower status roles compared to men, their desire to establish close interpersonal relationships, their caution in signaling power and judgmentalness, and their desire to establish their competence. While we do not argue for any one particular process, several of our specific findings across studies may speak to the various (potentially interrelated) processes that might underlie this effect.

For example, in Study 1, we find that when an audience’s distance is made salient, gender differences in linguistic abstraction are eliminated. The moderating role of distance is consistent with this factor playing a role in explaining gender differences in communicative abstraction. That is, women may be relatively more inclined to create proximity with others (indeed, we found this gender pattern in Study 6, although it did not negatively correlate with the use of abstraction) or to conceptualize others as proximal; emphasizing that an audience is distant may block the proximating tendency of women and minimize gender effects in communication. Also important to consider in the context of Study 1’s findings are our findings in Study 3, which showed differences in the communicative abstraction of male and female bloggers. On the one hand, this may be surprising given Study 1’s findings (given that bloggers communicate with a sizeable audience). At the same time, unlike the experimental studies on audience size which made salient the size and/or heterogeneity of the audience and likely reduced variation in perceptions of it, the blogging context preserves the opportunity for variation in perception of one’s audience. For example, female bloggers may differ from male bloggers in terms of their perception of their audience’s homogeneity, size, and similarity to themselves; although speculative, such variation may support the emergence of gender differences in communicative abstraction within the blogging context.

We also considered the role of power in explaining gender effects on communicative abstraction. Across samples, we find effects both when respondents have relatively low levels of power (e.g., students, Mturk respondents) and when they have higher levels of power (members of Congress). Indeed, even within our Congress dataset (Study 4) we find no variation in the gender effect based on relative amount of power (House of Representative members vs. Senators). This is consistent as well with results of Study 5, which experimentally manipulated power and found that this did not interact with gender. These findings, however, do not preclude a role of the subjective experience of power. That is, even when in similar positions, men and women may differ in how powerful they feel. Study 5, which found that women reported lower subjective experience of power than men when power was experimentally primed, and that this subjective experience of power mediated the effects of gender on communicative abstraction, is consistent with a role of subjective power in explaining gender differences in speech. In Study 6, however, we did not find any evidence for gender differences in subjective experience of power, and subjective experience of power did not mediate the effects of gender on communicative abstraction. This suggests at a broad level that while subjective power may play a role in some contexts (e.g., most likely ones in which subjective power is salient, as in Study 5), the routine experience of power is unlikely to be the main driver of these gender effects across variable contexts.

Indeed, in Study 6 we considered a broader set of mediators of a gender effect on abstraction. Gender differences on the measures we collected supported many of our earlier arguments based on the gender literature: women reported greater motivation than men to seek closeness in communication contexts, greater likelihood of interacting in small and homogeneous networks, and greater concerns about establishing their competence. Further, we found that the tendency of women to establish and interact in small groups mediated their tendency to use concrete speech. As mentioned earlier, we certainly don’t see these results as ruling out alternative explanations, but they do suggest the plausibility of communication audience size playing an important role.

Also thought-provoking are the findings from Study 3. In a dataset that allowed us to capture writings of adolescents as well as adults, we found that gender differences were larger for older adults than for teenagers, suggesting that gender differences in communicative abstraction may be reinforced by one’s experiences. This is broadly consistent with our argument that women and men are acculturated in a variety of ways over time that are consistent with the development of different communicative abstraction tendencies. We call for future work to continue to explore this divergence in women and men’s speech, and how these are shaped through one’s interpersonal experiences.