Algorithms, interaction and power: A research agenda for digital discourse analysis

Discourse analysts have worked on digital data for decades now (see Herring, 1999; Thurlow & Mroczek 2011), initially treating digital data as human texts (Blitvich & Bou-Franch, 2018), then increasingly as human, multimodal interaction mediated through digital media (Androutsopoulous, 2015), also attending to digital platforms and their affordances in ethnographically inspired discourse analyses (Varis, 2014; Jones, Chik & Hafner, 2015; Varis & Hou, 2020; Tagg & Sergeant, 2016; Zentz, 2021). But going further in this direction, I argue that to take digital data seriously, discourse analysts also need to account for algorithmic agency in digital discourse, going beyond the mere affordances of these media in meaning-making processes. Digital data scholars, platform scholars, critical algorithm studies and many other disciplines invite us to come to grips with how digital media program sociality and thus make discourse at least partially technical (van Dijck, 2013 ; Bucher, 2018). Accepting this invitation, this paper outlines four different forms of algorithmic entextualisation and three manifestations of algorithmic power, concluding with five ways of advancing the analysis of algorithmic power in digital discourse.