The advance of digital video technology in the past two decades facilitates empirical investigation of learning in real time. The focus of this paper is the combined use of real-time digital video and a networked linguistic corpus for exploring the ways in which these technologies enhance our capability to investigate the cognitive process of learning. A perennial challenge to research using digital video (e.g., screen recordings) has been the method for interfacing the captured behavior with the learners' cognition. An exploratory proposal in this paper is that with an additional layer of data (i.e., corpus search queries), analyses of real-time data can be extended to provide an explicit representation of learner's cognitive processes. This paper describes the method and applies it to an area of SLA, specifically writing, and presents an in-depth, moment-by-moment analysis of an L2 writer's composing process. The findings show that the writer's composing process is fundamentally developmental, and that it is facilitated in her dialogue-like interaction with an artifact (i.e., the corpus). The analysis illustrates the effectiveness of the method for capturing learners' cognition, suggesting that L2 learning can be more fully explicated by interpreting real-time data in concert with investigation of corpus search queries. (Contains 4 tables and 6 figures.)