Thomson, Murachver, & Green (2001) asked, "Where is the gender in gendered language?" Arguing that previous work on sex differences in language tended to ignore the dynamic nature of language-in-use, the authors present a series of studies in which they demonstrate that the use of gender-preferential speech is highly dependent upon the sociolinguistic context, the nature of the conversational topic, and the gender salience of the recipient.
In particular, the authors utilize the Communication Accommodation Theory (CAT) developed by Coupland et al. (1988) to explore the nature of sex differences in the language displayed in emails (p. 171). The basic idea is that people engaged in discussion tend to converge towards the language style used by their partner(s), increasing the similarity of the language features used among the group. Of course, in some cases, people can elect to diverge from the language style used by their partner(s), thus emphasizing the differences among the group. Furthermore, people may converge to the language style they think their partner(s) utilize, based on ideas about stereotypes or characteristics of certain social groups.
Thomson and his colleagues asked their participants to engage in a 2-week-long series of emails with 1) a partner who utilized language with typical feminine features and 2) a partner who utilized language with typical masculine features. In both cases, the email partner was actually one of the researchers. The linguistic features of the participants' emails were then coded and analyzed for signs of convergence towards the language style of the partner. Table 1 shows that when participants conversed with the "female" partner, they used feminine language features whereas when they conversed with the "male" partner, they used more masculine language features. A second experiment counterbalanced the gendered name of the partner (e.g., Jack vs. Jill) with the language style of the partner (masculine vs. feminine). Once again, the authors found evidence that participants converged towards the language style of the partner, displaying linguistic features consistent with that style. One interesting difference was that several sex differences did occur based on the participants' own gender...the authors argue that this difference could be due to a tendency to uphold one's own language style when the gender and language style of a partner do not match (p. 174).
In a related study, Thomson & Murachver (2001) demonstrated that people are aware of gender-preferential language styles and that it is possible to identify the gender of one's partner based on the linguistic features they utilize. Participants were assigned to same-sex partners because, the authors argue, gender-preferential language is likely to be maximized in such cases (but, see the interactions between sender & recipient in mixed-sex pairs found by Colley and colleagues). Across three experiments, the authors showed that 1) it was possible to reclassify the emails based on the use of a "combination" of linguistic features, and 2) that students were able to successfully classify the emails as having been written by a male or a female, even when the emails were not "natural," but were created using the gender-preferential linguistic features identified in the first experiment.
Based on these results (and those from the previous study), the authors make a strong argument that, "gender-preferential language is not definitional" (p. 194). In other words, male-preferential and female-preferential language can be utilized by both males and females, depending upon the context, the conversational partner(s), and the task at hand. Furthermore, the authors emphasize the implication of the ability to predict the sex of a writer by pointing out that, "although computer-mediated communication has been hailed as the great equalizer, in many instances it is likely to be just the opposite" (p. 203).
Finally, Thomson (2006) expands the examination of contextual effects on language use in CMC to include topic. Specifically, he argues that the stereotypical nature of the topic being discussed will influence the gender-preferential language used in discussion groups on-line. He examined the language used in female-stereotypical discussion lists (fashion, health, and shopping), male-stereotypical discussion lists (sports, cars, and computers), and gender-neutral discussion lists (music, films, and television) and found that participants displayed linguistic features appropriate for the stereotype of the particular topic (see Table 2).
In a follow-up study in which he controlled the sex composition of the discussion groups, results were similar. Regardless of the participants' own sex (male or female), when asked to discuss "masculine" topics, they displayed more masculine linguistic features and when asked to discuss "feminine" topics they displayed more feminine linguistic features. The only surprise was that more emotion was expressed when discussing masculine topics; emotion is one of the most consistent and commonly referred to feminine linguistic feature in previous research. In conclusion, Thompson argues that, "gendered language in CMC discussion is more a function of the context or topic of discussion than the gender of the participants" (p. 175).
Thomson, R. (2006). The effect of topic discussion on gendered language in computer-mediated communication discussion. Journal of Language and Social Psychology, 25(2), 167-178.
Thomson, R., Murachver, T. (2001). Predicting gender from electronic discourse. British Journal of Social Psychology, 40(2), 193-208.
Thomson, R., Murachver, T., & Green, J. (2001). Where is the gender in gendered language? Psychological Science, 12, 171-175.
Coupland, N., Coupland, J., Giles, H., & Henwood, K. (1988). Accommodating the elderly: Invoking and extending a theory. Language in Society, 17, 1-41.