COHERENT CHAOS: STRUCTURAL AND LINGUISTIC ANALYSIS OF COMPUTER-MEDIATED DISCOURSE ON TWITCH.TV

Abstract

Twitch.tv has grown immensely popular as a livestreaming website within the last decade, resulting in many live streamers gaining attention and even celebrity-like statuses. These live streamers draw in audiences of thousands who view their livestreams and participate in the livestream chatroom. Such a large number of participants producing messages every second draw to the assumption that these chatrooms are fast-paced, chaotic, and thus largely incoherent as a whole. Using an adaptation of Sinclair & Coulthard’s (1975) well-known exchange structure model, this thesis aims to disprove that assumption in part by establishing the extent of structural coherence within such chatrooms. Equally, a corpus of chatroom data will be analysed to identify the usage of language and linguistic strategies that support coherence. Chatroom data was extracted from top livestream channels with private streamers that averaged at least 10,000 concurrent live viewers. 5000 chat messages underwent close analysis using the aforementioned exchange structure model. All chatroom data was also run through AntConc (Anthony, 2022a), a software that aids in corpus analysis. Despite the large number of participants, the propensity for non-sequential exchanges similar to other online chatrooms, and the high speed at which chat messages are produced, the chatrooms of such Twitch streams are still largely coherent in terms of exchange structure. Direct addressivity in chat messages to intended recipients, prefabricated exchanges, and usage of emotes and lexis unique to Twitch was also observed to enhance communication and understanding on Twitch. Investigating how large Twitch chatrooms work structurally and behaviourally uncovers a possible avenue as to why viewers still participate in such chatrooms despite its seeming incoherence, and offers insight into the Twitch community, culture, and experience.