The Swedish immigration discourse in traditional media and social media
Title: The Swedish immigration discourse in traditional media and social media
SNIC Project: SNIC 2020/5-604
Project Type: SNIC Medium Compute
Principal Investigator: Marc Keuschnigg <marc.keuschnigg@liu.se>
Affiliation: Linköpings universitet
Duration: 2020-12-01 – 2021-12-01
Classification: 50401 10208
Homepage: https://liu.se/en/research/computational-text-analysis
Keywords:

Abstract

In the last two decades, social media has become an alternative source for news consumption as well as an arena for discussing current events, facts, and politics. Previously, traditional media such as newspapers, television, and radio were thought to set the agenda (McCombs & Shaw, 1972) and framing (Entman, 1993) of public discourse. Recent discoveries, however, demonstrate that social media audiences are increasingly taking over the role of ‘editorial gatekeeping,’ defining the perceived relevance and offering understandings of news events (e.g., Haim et al 2018, Meraz 2011, Sayre et al 2010). Existing studies suggest that the dynamics of both agenda-setting and framing in traditional media and in social media are highly intertwined (Neuman et al 2014, Guggenheim et al 2015), but they fall short in dissecting the mechanisms of mutual influence, convergence or divergence in shared understandings of ongoing events. We explore the interactive relationship of traditional media and social media focusing on the discourse developing around immigration in Sweden since the early 2000s. Based on large text corpora obtained through our institutional partner The National Library from the main national newspapers and from popular social media platforms (Flashback and Twitter), and using automated text analysis, we study the co-evolution of shared understandings of immigration-related developments and events in these two mediums. We are interested in how socially shared understandings spread, how cross-media interactions accelerate or decelerate their diffusion in offline and online audiences, and how larger “shocks” such as the Sweden Democrats’ entering of parliament and the “refugee crisis” of 2015 play out on the text-analytically measured processes. To capture “shared understandings” we develop topic models of media discourse for both corpora. To explore the differences between discourse constructions and vocabularies, we employ word embedding models. To further study the structure of discussion and tendencies of communities on Flashback towards extremization and polarization, we are aiming to involve network analysis. As the size of the corpora are substantive, we are here applying for computation resources to run the planned models.