A long-time critique of studies in CALL has been the frequent lack of solid theoretical underpinnings to help explain how computer applications, digital games included, can facilitate language learning in practice. One approach to address this concern is to draw on theoretical models of second-language acquisition (SLA) to clarify processes of game-based learning. The cognitive-interactionist model of SLA, based on Long’s interaction hypothesis, is one of the leading theoretical frameworks used to account for language learning. Methods of empirical discourse analysis developed using this model have been adopted in studies investigating game-based learner interaction. The sociocultural model, on the other hand, is a different theoretical conception of SLA based on Vygotskian psychology and has also been employed in a number of studies to examine the ways in which games may facilitate language learning. This entry provides an overview of both theoretical frameworks and defines their key concepts. It also considers a methodologically rigorous alternative to theory-based discourse analysis. Several practical approaches to discourse analysis in a DGBLLT context are presented, and significant findings from a selection of studies on learner interaction in digital games and virtual worlds are reviewed. Finally, suggestions for fruitful future avenues of research on game-based interaction are discussed.