Gender Construction in User Comments on Selected Facebook Pages: Integrating Topic Modeling with Feminist Critical Discourse Analysis

Interactive spaces on social networking sites (SNSs) provide a significant site for the (re)production, negotiation and/or contestation of gender identity. The comment sections on social media platforms, in particular, allow users to engage in the construction of their gender identities, notably via the digitally-mediated textual interactions hosted by these spaces. Against this backdrop, this study aimed to examine gender construction in a sample of user-generated comments drawn from a number of English and Arabic Facebook Pages pertaining to the themes of feminism and women's rights and violence against women. The study combined topic modeling, a computational method for classifying texts into topics by detecting the clusters and frequencies of co-occurring words, and feminist critical discourse analysis (FCDA) with a view to gaining a better understanding of the research problem by integrating insights from both quantitative and qualitative methods. To investigate the workings of gender construction in the data under investigation using the FCDA paradigm, this study specifically drew on the analytical toolkit proposed by Bucholtz and Hall (2004, 2005, 2010) for analyzing identity in interaction. The results showed that users utilized a number of indexical processes, such as references to identity categories and labels, pronoun use, metaphors, and stance-taking, to construct themselves and others as particular types of gendered beings as well as to align themselves with or against particular gender discourses and ideologies. The findings also revealed the multidimensionality of the construction of gender identity in the studied comments, particularly insofar as gender intersects with other identity-based factors such as region, culture, and religion. Thus, the gendered representations that emerge from the data exhibit a range of complex and multifaceted gender-based positions and identifications.