Can you quantify the qualitative? Developing a coding and video annotation method for L2 interaction

Studying interactional discourse involving L2 learners is recognisably an important endeavour in applied linguistics research. Such inquiries, however, have been approached rather differently across sub-disciplines of L2 studies. Quantitative approaches often involve coding interactional discourse using conventional measures (e.g. number of turns) or pre-determined interactional features (e.g. agreeing/disagreeing). Such analyses might miss important, non-frequency-related insights about interactional practices of L2 learners, and risk crude, imprecise categorisations on which distributional patterns are identified. Conversation analysis (CA) offers much more nuanced insights into the co-constructed and locally contingent nature of interactional practices, but is less amenable to empirical generalisation and exploration of relationships between variables that some disciplines or research purposes demand. This article proposes a CA-informed coding framework and video annotation method for analysing L2 spoken interaction. Using an example study involving 56 L2 English learners in a speaking assessment task, this article demonstrates how a more robust coding framework can be developed based on CA principles and findings; how the qualitative data analysis application MAXQDA can facilitate coding and intercoder reliability checks; and how complementary quantitative and qualitative findings can be presented. The proposed method contributes to L2 research by striking a balance between precise, micro-level analysis of interactional phenomena and gaining macro-level insights into distributional patterns and relationships. We conclude by suggesting some potential applications in future L2 interactional competence (IC) studies such as pedagogical interventions, assessing IC in monologic tasks, and learner-AI interactions.