Programm 80. StuTS

Live-streaming platforms such as Twitch are central meeting hubs for young people, and their chats are dominated by emotes, small images posted alongside or instead of words. Emotes help build shared repertoires and community identity, but they can also carry or mask toxic content. Whether an emote counts as toxic depends on the norms of the community using it (Mihailova, 2022). However, research has mostly examined single communities, so it remains unclear what happens when communities with different norms meet in one chat.

This study asks whether toxic emote practices travel across community boundaries. I analysed a collaborative stream in which Jynxzi, one of Twitch’s most-followed gaming streamers, visits Maya Higa’s educational Alveus Sanctuary, and compared its chat with one solo stream from each creator (216,086 messages total). Using Emotes-2-Vec (Moosavi et al., 2024), a Word2Vec-based method that infers emote meaning from context, I scored each emote by its similarity to toxic versus non-toxic words. I then compared mean emote toxicity and the share of toxic-leaning uses across streams using Kruskal-Wallis, Mann-Whitney U, and chi-square tests.

The findings were as follows: Jynxzi’s chat used more toxic-leaning emotes (82.1%) than Maya’s (60.0%). Contrary to my hypothesis, the collab stream (79.5%) was nearly indistinguishable from Jynxzi’s solo stream rather than adapting to Maya’s, although channel-specific ridicule emotes largely disappeared. The most toxic-leaning emotes were not slurs but reaction and laughter emotes (e.g., LOL, BRUH) appearing next to insults, consistent with ambiguous negativity. Effect sizes were small, so I interpret the results as tendencies rather than firm differences between communities.

In conclusion, emote norms travel with a community rather than being set by the space it enters. This is of vital importance for creators who host collaborations and may shape what young viewers learn is acceptable interaction online.