Linguistic convergence and divergence of Basque on Twitter: a multilingual computational sociolinguistic analysis

This article explores linguistic accommodation on X – formerly Twitter – with regard to Basque, Spanish, French, and English in the Basque Country. From the perspective of computational sociolinguistics (CSLX), natural language processing (NLP) techniques and semantic network analysis on 1,267 geolocated conversations are employed, examining linguistic convergence and divergence, as well as territorial interactions. The results reveal a predominance of Spanish and lower convergence in Basque, suggesting sociolinguistic pressure in digital environments. French exhibits a high degree of intralinguistic convergence, whereas English is associated with greater variability in responses. Furthermore, thematic and territorial factors influence language choice and communicative adaptation, highlighting the persistence of patterns already present in offline settings. This study underscores the methodological challenges of working with minoritised languages and limited computational resources typical of low-resource languages, and calls for further exploration of the sociolinguistic dimension of computer-mediated communication (CMC), fostering a better understanding of multilingual communicative dynamics in social media.