Combining CDA and topic modeling: Analyzing discursive connections between Islamophobia and anti-feminism on an online forum

Artikel i vetenskaplig tidskrift, 2016

In this article we present an analysis of the discursive connections between Islamophobia and anti-feminism on a large Internet forum. We argue that the incipient shift from traditional media toward user-driven social media brings with it new media dynamics, relocating the (re)production of societal discourses and power structures and thus bringing about new ways in which discursive power is exercised. This clearly motivates the need to critically engage this field. Our research is based on the analysis of a corpus consisting of over 50 million posts, collected from the forum using custom web crawlers. In order to approach this vast material of unstructured text, we suggest a novel methodological synergy combining critical discourse analysis (CDA) and topic modeling – a type of statistical model for the automated categorization of large quantities of texts developed in computer science. By rendering an overview or ‘content map’ of the corpus, topic modeling provides an enriching complement to CDA, aiding discovery and adding analytical rigor.