This paper discusses a method of collecting and analysing multimodal data during classroom-based digital literacy research. Drawing on reflections from two studies, the authors discuss theoretical and methodological implications encountered in the collection, transcription and presentation of such data. Following an ethnomethodological framework that co-develops theory and methodology, the studies capture digital literacy activities as real-time screen recordings, with embedded video recordings of participants' movements and vocalisations around the tasks during writing. The result is a multimodal rendition of digital literacy events on-and off-screen, allowing linguistic and multimodal transcriptions to capture the complexity of the data in a format amenable to analysis. Acquiring such data allowed for the development of detailed analyses of digital literacy events in the classroom, including interaction that would otherwise have escaped standard ethnography and video analysis, through sensibilities that approach social and material items without a priori hierarchies. This leads us to a 'performative' notion of digital literacies and an analytic methodology that is useful for researchers paying greater attention to the sociomaterial assemblages in which digital literacy events unfold.