Computer-supported collaborative learning (CSCL) settings for academic writing have become a staple in foreign language classrooms in higher education. These settings allow learners to discuss their output, assist others and dialogically assess their learning progress. To successfully do so, however, learners need to be able to effectively self-regulate their learning process. The multiple contingencies of self-regulated learning (SRL) in online collaborative writing settings have hitherto received limited attention in research. Recent advances in learning analytics and quantitative ethnography, nevertheless, offer new opportunities to analyse learner discourse and reveal previously underexplored aspects of SRL. Through the use of epistemic network analysis (ENA), this study examines structural patterns in students’ use of SRL strategies and meta-strategies, and models their co-occurrence. Data were collected from a Facebook group integrated into an academic writing course for first-year foreign language majors of English (N = 123). The results illustrate how students engage in cognitive and meta-cognitive discourse, and show that other strategies and meta-strategies in the network mainly occur in isolation. The use of ENA, in addition, reveals the different contingencies in the SRL process over time. This study contributes to the fields of quantitative ethnography, learning analytics and SRL by: 1. Showing how ENA can add to our understanding of the SRL process, and 2. by discussing which self-regulatory strategies and meta-strategies are predominantly used in CSCL settings for academic writing, which ones deserve additional attention when integrating CSCL settings in this context, and what educational interventions can be designed as support.