The discourse of online live streaming on twitch : communication between conversation and commentary

Abstract

Live streaming on Twitch is a new form of online broadcasting of video games. Aside from the visual game, it features computer-mediated communication between a speaking streamer and a large chatting audience. This communicative setting is quickly expanding to other types of broadcasting. So far, there has been a limited number of sociological and ethnographic studies that argue that live streaming is a mixture of conversation and gameplay commentary. This conversation is said to break down with an increase in audience size and devolve into a chaotic waterfall of text. These widely shared assessments and arguments are based on common-sense and ad-hoc definitions of terms such as conversation and commentary, but they have not been tested or studied trough discourse analysis of natural occurring data. Therefore, the study sets out to describe the organization of discourse of live streaming on Twitch. The second chapter is a detailed introduction to the phenomenon of video game live streaming, the existing prior research as well as the communicative environment of Twitch's website. Chapter three is dedicated to the linguistic literature review and will present the analytical framework of the Birmingham school and its current descriptions of spoken conversation. To apply the framework to live streaming of video games, the review will also introduce contemporary research on chat communication and online video chat. This introduction will provide a suitable starting point to discuss the discourse of Twitch. The methodology chapter explains the transformation of the visual gameplay, the streamer's spoken language and the written chat into a coherent transcript, which is open to the discourse analysis. This transcript is studied in the three original research chapters five, six and seven. Each chapter describes the organization of discourse of live streaming from a different perspective, which will allow an overall synthesis of the study in the conclusion. Chapter five begins with the distribution of discursive moves between the participants and describes who is talking to whom, for which purposes and how often. There are different directions of communication such as from viewer to streamer, from streamer to viewer or from viewer to viewer. Often, these directions of talk and chat correspond to different social purposes within the communication. For example, viewers tend to write questions in the chat, when they address the streamer. In turn, streamers tend to give spoken responses to these questions as they address their viewers. It leads to a very consistent pattern in the dialogue. However, the chapter will also show that the largest amount of communication is not strictly dialogical. Streamers and viewers produce monological moves that talk about the game but have no direct recipient. Dialogical communication between the parties and more monological communication about the game co-occur. Therefore, the chapter will conclude that a direct equation of live streaming with conversation or commentary is too limited and needs further refinement to accurately account for the discourse of live streaming. In the sixth chapter, the dissertation shows how dialogical and monological moves combine to larger units of discourse and how they relate to the simultaneously unfolding gameplay. This will include the most common dialogical exchange patterns between streamer and viewers, which will be explained in reference to the streamers'status as players, the chat participants' status as viewers, and their different modes of communication. Their discussion reveals that research on cross-modal discourse must distinguish spoken-to-written communication from written-to-spoken communication. They have a different form and spoken-to-written communication features a new discursive move, the topicalizer, which is absent from written-to-spoken communication. The topicalizer is a repetition of a written chat message in order to topicalize it in the spoken mode. The analysis also shows that the streamer's monological commentary consists of two discursive practices with distinguishable content and a different temporal relationship to the unfolding gameplay. Lastly, the chapter demonstrates that the organization of discourse is highly cyclical and re-occurs in patterns that are aligned to the structure of the game and the organization of the broadcast. Chapter seven studies 'donation alert messages', which are a unique type of cross-modal exchange that receives privileged interactional treatment during live streaming because they are elicited by the payment of a 'donation'. The chapter classifies the most common types of these messages based on their form and content and discusses their role in the overall organization of discourse on Twitch. Chapter eight summarizes the findings of the three original research chapters and synthesises them in a new descriptive model. The model is an adaptation of Birmingham school's original description of discourse applied to live streaming. It is able to explain which discursive practices are likely to occur in different situations of the broadcast. Finally, the chapter also contains critical reflection on the research and suggest potential future research projects that may challenge established notions of casual conversation as talk that is unaffected by instrumental or monetary motivations.