A Tale of Two Twitters? Identifying Bridges Between Language Based Twitterspheres

Abstract

Global political developments – such as Brexit, climate change, or forced migration – are entangled with communication that transcends national publics. Meanwhile, the EU’s integrity suffers, also due to polarised online discourses, which are sometimes actively manipulated. Therefore, an overview of online communication beyond language barriers is essential. However, whether and how online media create a global space that sustains deliberation of national and global interests by citizens, remains understudied. We approach this problem by exploring relations between the Italian and German Twittersphere, while asking: 1) What is the macrostructure of this bilingual network? 2) Are there bridges between these language communities in the form of single accounts and how can they be described? 3) Are there bridges in the form of groups and what are they tweeting about? We build on an innovative network crawling strategy for language-based Twitter follow networks. We developed it further to combine strengths of rank degree, snowball, and forest fire sampling. Thereby, we collect a network sample of the most central accounts in the Italian-German Twittersphere. Preliminary results suggest a bridging quality of soccer and connections between political clusters of both languages by EU politicians. Furthermore, larger network clusters connect mainly with one linguistic domain while smaller communities show a bridging behaviour. The final paper will present results of months of data collection, focusing on the relation between topics discussed within clusters and their connectivity. While it focuses on the German-Italian Twittersphere, our methods open up new avenues of enquiry regarding multi-language public spheres.

Census of Twitter users: Scraping and describing the national network of South Korea

“…Bruns and others scraped the German Twitter network dataset TrISMA [ 13 ] and later the dataset served as a ground truth for the so-called rank degree sampling method [ 11 ]. Recently, Munch and Rossi compared two national Twitter networks, Italian and German, based on the language detection of user profile [ 14 ]. Our data collection mainly followed the language-based approach.…”