Bilingual societies: from data toagent-based modeling

A great part of the cultural diversity contained within the languages of the world is at risk, as many languages have become endangered in the last decades1. To preserve this diversity, one must first understand what drives the death of languages, and what could enable their coexistence. This is why we seek here to uncover the mechanisms at work in language shift through agent-based modeling2. A large scale empirical study of the patterns of spatial mixing of languages in multilingual societies using Twitter data unveils an interesting diversity. It ranges from an almost complete mixing of languages to their complete separation. To understand how these different states can arise, we propose a model in which coexistence of languages may be reached when learning the other language is facilitated and when bilinguals favor the use of the endangered language. Simulations carried out in a meta-population framework highlight the importance of spatial interactions arising from the mobility of people to explain the stability of a border between two linguistic regions.Peer reviewe