Abstract
The evolution of language has been a hotly debated subject with contradictinghypotheses and unreliable claims. Drawing from signalling games, dynamicpopulation mechanics, machine learning and algebraic topology, we present amethod for detecting evolutionary patterns in a sociological model of languageevolution. We develop a minimalistic model that provides a rigorous base forany generalized evolutionary model for language based on communication betweenindividuals. We also discuss theoretical guarantees of this model, ranging fromstability of language representations to fast convergence of language bytemporal communication and language drift in an interactive setting. Further wepresent empirical results and their interpretations on a real world datasetfrom \rdt to identify communities and echo chambers for opinions, thus placingobstructions to reliable communication among communities.