(How) Do Far-Right Narratives Get Disseminated into the Mainstream? An Exploratory Study of Three Cases from Germany

The far right not only communicates in their own bubble but also succeeds in influencing mainstream agendas (Saldivia Gonzatti & Völker, 2025). This could e.g. be seen recently in the discussion about nomination of the law professor Frauke Brosius-Gersdorf as a candidate for the Federal Constitutional Court of Germany. After discussions including false claims about both alleged plagiarism in her doctoral thesis and her stance on abortion, the election was postponed and she eventually resigned. A Berlin thinktank analyzed Twitter/X data for this case and found far-right actors as drivers (Sälhoff, 2025). We investigate the influence of far-right actors on narratives in mainstream media by following three topics in the German discourse over one year: The above described campaign against Frauke Brosius-Gersdorf, the narratives around the German unemployment benefits (“Bürgergeld”, citizen’s money) and the narratives around refugees and violence. Our exploratory analysis is based on data of various outlets and platforms from November 1, 2024, until October 31, 2025. We collected around 500,000 online news articles, 1.5 million posts from the biggest German-speaking Telegram channels, automated transcripts from six talkshows in German public broadcast, and descriptions from 74,000 TikTok posts by accounts registered in the database of public actors (Münch et al., 2023). Based on this rich dataset, we exclude non-political articles and posts (e.g. focusing on sports) following the categories of the news outlets before splitting articles, posts and transcripts into paragraphs. We further clean these for redundancies that often appear due to generic text fragments, and filter paragraphs that are not political using a zero-shot model. This step is manually validated on a subset of 100 paragraphs. We then employ sentence embeddings to detect semantic similarities between paragraphs and use network analysis with semantically close paragraphs clustering together. First results for the case of Frauke Brosius-Gersdorf show a promising picture. Clusters in the network are rather separated, fit subtopics such as plagiarism accusations, criticism by an archbishop, general information about the election as well as statements by the candidate herself and evolve at different times in the debate. The debate is quickly driven by media outlets while Telegram discussions pick up and discuss the information from the media rather than starting own trends. TikTok and talk shows are only marginally present. We are aware of a bias towards paragraphs from the media in our data and especially from a smaller number of outlets. We intend to extend the sample for more media outlets and potentially X data (if data access is granted). We then seek to analyze the other topics in a similar manner. References Münch, F. V., Merten, L., & Schmidt, J.-H. (2023). Die „Datenbank Öffentlicher Sprecher“ (DBÖS) [Dataset]. OSF. https://doi.org/10.17605/OSF.IO/SK6T5 Saldivia Gonzatti, D., & Völker, T. (2025). Far-right agenda setting: How the far right influences the political mainstream. European Journal of Political Research, 1–23. https://doi.org/10.1017/S1475676525100066 Sälhoff, P. (2025, August 25). Die Causa Brosius-Gersdorf. polisphere. https://polisphere.eu/aktuelles/die-causa-brosius-gersdorf