A computational analysis of Telegram's narrative affordances

This paper offers an empirical investigation of the narrative profiles afforded by public, one-way messaging channels on Telegram. We define these narrative profiles in terms of the contribution of messages to a thread of narrative continuity, and test the double hypothesis that 1) Telegram channels afford diverse narrative profiles, corresponding with distinct vernacular uses of the platform's features, and that 2) networks of Telegram channels sampled from thematically distinct seed channels lean towards distinct profiles. To this end, we analyse the textual contents of 2,724,187 messages from 492 public messaging channels spanning five thematic networks. Our computational method builds up the narrative profiles by scrolling down channels and classifying each message according to its narrative fit with the surrounding messages. We thus find that Telegram channels afford several distinct storytelling profiles, which tend to defy traditional notions of narrative coherence. We furthermore observe correspondences between the thematic orientations of channels and their narrative profiles, with a preference for disparate profiles in channels pertaining to conspiracy theories and far-right counterculture, a preference for coherent profiles in channels pertaining to cryptocurrencies, and mixed types in channels pertaining to disinformation about the war in Ukraine. These empirical observations thus inform our further theorization on how platform features allow users to construct and shape narratives online.