Abstract
<jats:p>This study examines the social and emotional dimensions of streamer–viewer interaction on Twitch through thematic analysis of viewer comments. As users shift from passive consumers to active participants, interaction practices in live streaming have transformed. Twitch's real-time communication enables collective and interactive experiences while fostering dynamic parasocial relationships. The study applies qualitative thematic content analysis to 3,000 randomly selected comments from gaming, reaction, and chatting streams of a highly viewed Turkish streamer. Using NLP-assisted pre-coding and manual validation, themes were identified as direct address, humor, positive and negative emotion, toxic language, and content reaction. Findings show that behavior varies by stream type: positive emotions dominate gaming (13%), humor reaction streams (10%), and direct address chatting streams (35%), while toxic language remains low (0.3%). Results position Twitch as a digital public space where users engage emotionally, co-produce meaning, and compete for visibility.</jats:p>