Resumen: Twitter for academic purposes has been analysed from multiple perspectives such as genre analysis, the use of multimodality and hypertextuality, or type of participants; yet interactivity between writers and readers remains under-researched. This study analyses academic-related conversations from the Twitter conference genre, particularly focusing on the discussion session. Its objective is to identify the main interactional patterns, communicative functions, and digital discourse features in tweets. Dialogic turns were classified into comments, questions, responses, follow-up conversations, and automatic comments. Findings reveal that the main reasons behind online interaction correspond with community building and knowledge construction purposes. The digital medium does shape the form of tweets, which shows a high level of evaluative language, conversational style features, hedging, and emojis. All in all, these discursive features help create a welcoming and engaging style needed to engage in online science communication practices on social media. Idioma: Inglés DOI: 10.32714/ricl.13.01.05 Año: 2024 Publicado en: Research in corpus linguistics 13, 1 (2024), 86-112 ISSN: 2243-4712 Factor impacto SCIMAGO: 0.278 - Linguistics and Language (Q2)