This thesis presents a systematic discourse analysis of sustained antagonistic debate—called 'drama'—on the video-sharing website, YouTube. Following a two-year observation of a YouTube community of practice discussing Christianity and atheism, 20 video 'pages' (including talk from videos and text comments) from a drama event were identified and transcribed, producing a 86,859 word corpus comprising 136 minutes of video talk and 1,738 comments. Using metaphor-led discourse analysis (Cameron & Maslen, 2010b) of the total corpus, metaphor vehicles were identified, coded, and grouped by semantic and narrative relationships to identify systematic use and trace the development of discourse activity. Close discourse analysis of a subset of the corpus was then employed to investigate membership categorisation (Housley & Fitzgerald, 2002), impoliteness (Culpeper, 2011), and positioning (Harré & van Langenhove, 1998), providing a systematic description of different factors contributing to the emergence of 'drama'. Analysis shows that 'drama' developed when negative views of one user's impolite words exposed the different expectations of other users about acceptable YouTube interaction. Hyperbolic, metaphorical language derived from the Bible and narratives about tragic historical events often exaggerated, escalated, and extended negative evaluations of others. Categories like 'Christian' were used dynamically to connect impolite words and actions of individuals to social groups, thereby also extending negative evaluations. With implications for understanding 'flaming' and transgression of social norms in web 2.0 environments, this thesis concludes that inflammatory language led to 'drama' because: (1) users had diverse expectations about social interaction and organisation, (2) users drew upon the Bible's moral authority to support opposing actions, and (3) the online platform's technical features afforded immediate reactions to non-present others. The 'drama' then developed when users' responses to one another created both additional topics for antagonistic debate and more disagreement about which words and actions were acceptable.

Notre enquête a été conduite auprès de membres d’un groupe d’anciens élèves hébergé sur le site de réseau social professionnel LinkedIn. Elle a permis de produire deux résultats. Le premier résultat est une typologie des figures du réseautage en ligne. Nous proposons trois figures. Le NetMining, qui relève d’un usage exploratoire du site et oùl’utilisateur cumule de nombreux contacts. Le NetWorking où l’utilisateur sélectionne des relations sur la base de critères précis. Le NetSticking où l’utilisateur reproduit en ligne un réseau personnel basé sur la confiance. Ces trois tendances permettent de souligner deux caractéristiques propres aux pratiques de réseautage en ligne. D’une part, elles reposent surune pluralité de dispositifs de communication où LinkedIn occupe une place tantôt centrale, tantôt marginale. D’autre part, ces pratiques ne reposent pas nécessairement sur la poursuite d’une action planifiée. Le second résultat produit par notre enquête est un dispositif d’étude des pratiques de réseautage. Il associe des techniques de visualisation de graphes et analyses d’entretiens semi-directifs / Our study had been carried on members of a social network hosted on the professional social network site, LinkedIn. It produced two main outcomes. The first is a typology of online Networking on three different classes. The first is the NetMining, which is a exploratory use of the website, where the user accumulated contacts. The second category is the NetWorking, where the users select contacts based on defined characteristics. The NetSticking is the third category, in which the user reproduced online a personal network based on trust. Those three categories highlight two main characteristics, specific to online networking. First, networking uses different means of communication, in whichLinkedIn is sometimes central, but sometimes marginal. Those practices are besides not necessarily based on planed actions. The Second result of our enquiry is a study device of the networking practices which associated graphs visualization technics and semi-directed interviews analyses