The Swedish immigration discourse in traditional media and social media
Title: The Swedish immigration discourse in traditional media and social media
DNr: NAISS 2023/5-561
Project Type: NAISS Medium Compute
Principal Investigator: Marc Keuschnigg <>
Affiliation: Linköpings universitet
Duration: 2024-01-01 – 2025-01-01
Classification: 50401 10208


This application is an extension of the ongoing project SNIC 2022/5-571, preceded by our collaboration with SNIC/NAISS in 2021 (SNIC 2020/05-604) and 2022 (SNIC 2021/5-537). The aim of this work is investigating changes in the Swedish immigration discourse in national newspapers and social media, a highly relevant topic in the context of increasing political polarisation of public debate with respect to refugees, immigration, and Islam in Sweden. In this work, we analyze overlaps and mismatches in the interpretations of immigration in two entirely different but equally important societal segments which actively produce immigration narratives: nationwide traditional media and online social media. We focus on the case of the Swedish immigration discourse over the period of 15 years (2006 - 2020) to gain insights into the parallel evolution of discourses in the national media and on Sweden’s largest online forum. The period of observation captures the rise of the far-right party Sweden Democrats and the European ‘refugee crisis’ of 2015. We complement previous work by introducing another approach to align and compare the immigration discourses produced by very different groups of actors. We model both corpora in a single topic model with constraints imposed on topic-word associations (Magnusson et al., 2018; Hurtado Bodell et al., 2022; Watanabe and Zhou, 2022). In sum, this paper aims to take on a macroscopic view of immigration discourses in traditional and social media and (1) identify overlaps and mismatches between the interpretations of immigration among journalists and the online public, (2) compare reactions of immigration discourses in both spheres to major events, and (3) identify points of change in both domains and compare the timing and the nature of these changes for both corpora. We have developed a new direction in the course of this project. We are now focusing on designing a methodological approach for comprehensive comparison between multiple large-scale text corpora generated by different sources. We, therefore, would like to ask for extension of the project to enable us pursue this new research direction.