Toward a transcription and analysis of live streaming on Twitch

Abstract Online live streaming is a new media genre that combines the broadcast of an activity with cross-modal video-mediated communication. Lacking an analytical entry point, descriptions and micro-level analysis of this type of interaction are rare. Using the case of online live streaming of video games, this paper asks two questions. First, how should a transcript look like in order to systematically account for the activity and the cross-modal communication between broadcaster and audience? Secondly, how does the unfolding of the activity influence the cross-modal discourse during online live streaming? In answering the first question, this paper develops a multi-column transcription scheme, which includes the broadcaster's spoken language & embodied conduct, the audience's written chat messages as well as an annotation of game events. It is argued that the basic principles of this format are applicable to other forms of live streaming and research questions. Afterwards, the transcript is used for exemplary analysis to address the second question. It will be demonstrated that broadcasters are more tightly regulated by the unfolding of the activity than the audience and that this leads to different cross-modal communicative practices. The numerous audience members write quick and brief single-turn messages, whereas the broadcasters are selective and produce fewer but more elaborate responses that span several utterances. Lastly, the paper introduces ‘pivoting’ and argues that it is a novel communicative behavior typical for online live streaming. Based on a second analysis, the paper will show that game events elicit pivoting utterances or messages from the broadcaster and audience, which attribute a highly contextual and locally negotiated meaning to the event.