Elamur Research⌕
☰
home / papers / 2026 / German parties shifted towards intuition-based rhetoric after the far right's parliamentary breakthrough

German parties shifted towards intuition-based rhetoric after the far right's parliamentary breakthrough

Paper · 2026 · Peer Saleth, Segun T. Aroyehun, Fabio Carrella et al.

Summary

The paper asks whether German political elites changed how they appeal to truth between 2015 and 2025, and whether any change lines up with the far-right AfD entering the Bundestag in 2017. The authors built two corpora: 4.5 million tweets from more than 2,000 politicians, partly obtained under the EU Digital Services Act Article 40, and 59,170 Bundestag plenary speeches. They score each text with a validated Distributed Dictionary Representation (DDR) that compares evidence-oriented and intuition-oriented language, giving an Evidence Minus Intuition (EMI) score. The embedding model behind it was first trained on historical German parliamentary text. Mixed-effects regressions with election-year breakpoints test for trends and shifts.

In both arenas, intuition-based language gains ground over time. Right-leaning actors, and the AfD above all, have the lowest EMI. In parliament the whole chamber drops sharply when the AfD arrives, while Twitter declines more slowly. This is the first cross-arena, multiparty European evidence linking far-right visibility to a change in elite epistemic style, which earlier work had documented mainly for the US.

Key points

  • Data: 4.5M tweets from 8 parties and 59,170 speeches from 7 parties and 1,391 speakers, covering the 18th–21st electoral terms (2015–2025). Party leaning follows the Chapel Hill Expert Survey (CHES).
  • Embedding pre-training: two historical German Parliamentary Corpora (1867–2021), with 329,390 speeches by 3,786 politicians across 22 parties. These were used only to train the embeddings, not in the analysis.
  • Bundestag mean EMI by party:
    • SPD 0.113
    • CDU/CSU 0.108
    • The Left 0.051
    • Greens 0.039
    • FDP −0.110
    • AfD −0.473
    • BSW −1.098, but only n=153, so it is reported descriptively.
  • Twitter: the AfD's mean EMI is −0.268.
  • Models: linear mixed-effects regressions with actor random intercepts and lagged EMI, plus breakpoints at 2017, 2021 and 2025. In the constrained Bundestag model, the 2017 level shift is about −0.42 and highly significant.

Interesting ideas

  • I1. Two arenas with opposite weaknesses: the authors pair a free-form, algorithm-shaped platform with a rule-bound parliament, so a trend found in both is unlikely to be an artifact of either one.
  • I2. Continuous stance score from embedded seed lexicons: DDR places documents by their similarity to two concept dictionaries in embedding space and takes the difference, rather than counting keywords.
  • I3. Separate corpora for pre-training and measurement: the embedding model learns from a large historical in-domain corpus that is kept apart from the analysed data, which avoids leakage.
  • I4. Constrained versus group-specific breakpoints: the authors first fit one shared shift per event and then let each group have its own shift, controlling for lagged values. This separates common changes from group-specific ones.
  • I5. Flagging thin subsamples as descriptive only: tiny groups (BSW, n=153) are reported but not interpreted.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (learn): evaluate parser and reader changes on two deliberately contrasting corpora, such as formal treebank text and informal fiction, and trust an improvement only when it holds in both.
  • I2 (frames): use embedding-expanded lists of evidential versus belief markers only to propose candidate cues, then keep the accepted cues as explicit, inspectable lexicon entries for tagging speech-act or epistemic stance in prag.
  • I3 (learn): keep the corpora used to induce lexicons or embeddings disjoint from the evaluation sets, so that measured gains are not circular.
  • I4 (learn): when tracking metrics across model releases, test whether a regression is shared across all genres or limited to one, using the same shared-versus-per-group comparison.
  • I5 (answer): report test categories with very few items separately and keep them out of headline accuracy claims.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓University of PotsdamPotsdam · Germany44 works · 5,305 citations🎓University of KonstanzKonstanz · Germany17 works · 232 citations🎓University of BristolBristol · United Kingdom14 works · 725 citations🎓Universidade Estadual de Campinas (UNICAMP)Campinas · Brazil2 works · 2 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Social Sciences · 1,772 worksSociology and Political Science · 577 worksPolitical Science and International Relations · 182 worksCommunication · 97 works

Cite

Cite as Peer Saleth, Segun T. Aroyehun, Fabio Carrella, Christoph M. Abels, Stephan Lewandowsky and David Garcia. 2026. German parties shifted towards intuition-based rhetoric after the far right's parliamentary breakthrough. arXiv:2608.05075.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2608.05075 · arXiv 2608.05075

@misc{saleth2026german,
  title = {{German parties shifted towards intuition-based rhetoric after the far right's parliamentary breakthrough}},
  author = {Peer Saleth, Segun T. Aroyehun, Fabio Carrella, Christoph M. Abels, Stephan Lewandowsky and David Garcia},
  year = {2026},
  doi = {10.48550/arxiv.2608.05075},
  eprint = {2608.05075},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2608.05075}
}

References

54 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Freedom House. Freedom in the world 2020: A leaderless struggle for democracy. Tech. Rep., Freedom House, Washington, DC (2020). ↗
  2. Repucci, S. & Slipowitz, A. Democracy under siege. Tech. Rep., Freedom House, Washington, DC (2021). find
  3. Lazer, D. M. J. et al. The science of fake news. Science 359 , 1094–1096 (2018). find
  4. Van Bavel, J. J. et al. Political Psychology in the Digital (mis)Information age: A Model of News Belief and Sharing. Social Issues and Policy Review 15 , 84–113 (2021). find
  5. Simchon, A., Brady, W. J. & Van Bavel, J. J. Troll and divide: the language of online polarization. PNAS Nexus 1 , pgac019 (2022). find
  6. McCright, A. M. & Dunlap, R. E. Combatting misinformation requires recognizing its types and the factors that facilitate its spread and resonance. Journal of Applied Research in Memory and Cognition 6 , 389–396 (2017). find
  7. Loomba, S., de Figueiredo, A., Piatek, S. J., de Graaf, K. & Larson, H. J. Measuring the impact of COVID-19 vaccine misinformation on vaccination intent in the UK and USA. Nature Human Behaviour 5 , 337–348 (2021). find
  8. van der Linden, S., Leiserowitz, A., Rosenthal, S. & Maibach, E. Inoculating the public against misinformation about climate change. Global Challenges 1 , 1600008 (2017). find
  9. Ribeiro, M. H., Calais, P. H., Almeida, V. A. F. & Meira, W. “everything I disagree with is #FakeNews”: Correlating political polarization and spread of misinformation (2017). URL http://arxiv.org/abs/1706.05924. ↗
  10. Jones-Jang, S. M., Kim, D. H. & Kenski, K. Perceptions of mis- or disinformation exposure predict political cynicism: Evidence from a two-wave survey during the 2018 US midterm elections. New Media & Society 23 , 3105–3125 (2021). find
  11. Cantarella, M., Fraccaroli, N. & Volpe, R. Does fake news affect voting behaviour? Research Policy 52 , 104628 (2023). find
  12. Müller, K. & Schwarz, C. Fanning the flames of hate: Social media and hate crime. Journal of the European Economic Association 19 , 2131–2167 (2021). find
  13. Lorenz-Spreen, P., Oswald, L., Lewandowsky, S. & Hertwig, R. A systematic review of worldwide causal and correlational evidence on digital media and democracy. Nature Human Behaviour 7 , 74–101 (2023). find
  14. Lewandowsky, S., Garcia, D., Simchon, A. & Carrella, F. When liars are considered honest. Trends in Cognitive Sciences 28 , 383–385 (2024). find
  15. Cooper, B., Cohen, T. R., Huppert, E., Levine, E. E. & Fleeson, W. Honest behavior: Truth-Seeking, Belief-Speaking, and fostering understanding of the truth in others. Academy of Management Annals 17 , 655–683 (2023). find
  16. Lasser, J. et al. From alternative conceptions of honesty to alternative facts in communications by US politicians. Nature Human Behaviour 7 , 2140–2151 (2023). find
  17. Lasser, J. et al. Social media sharing of low-quality news sources by political elites. PNAS Nexus 1 , pgac186 (2022). find
  18. Carrella, F. et al. Different honesty conceptions align across US politicians’ tweets and public replies. Nature Communications 16 , 1409 (2025). find
  19. Garrett, R. K. & Weeks, B. E. Epistemic beliefs’ role in promoting misperceptions and conspiracist ideation. PLoS ONE 12 , e0184733 (2017). find
  20. Lewandowsky, S. in Willful construction of ignorance: A tale of two ontologies (eds Hertwig, R. & Engel, C.) Deliberate Ignorance: Choosing Not to Know , Vol. 29 of Strüngmann Forum Reports 101–118 (MIT Press, Cambridge, MA, 2021). find
  21. Aroyehun, S., Lewandowsky, S. & Garcia, D. Epistemic orientation in parliamentary discourse is associated with deliberative democracy (2026). URL http://arxiv.org/abs/2604.19699. ↗
  22. Varshizky, A. Alfred rosenberg: The Nazi Weltanschauung as modern gnosis. Politics, Religion & Ideology 13 , 311–331 (2012). find
  23. Voegelin, E. The Collected Works of Eric Voegelin (University of Missouri Press, 1989). find
  24. Lewandowsky, S., Ecker, U. K. H. & Cook, J. Beyond misinformation: Understanding and coping with the “post-truth” era. Journal of Applied Research in Memory and Cognition 6 , 353–369 (2017). find
  25. Aroyehun, S. T. et al. Computational analysis of US congressional speeches reveals a shift from evidence to intuition. Nature Human Behaviour 9 , 1122–1133 (2025). find
  26. van Spanje, J. Contagious parties: Anti-immigration parties and their impact on other parties’ immigration stances in contemporary Western Europe. Party Politics 16 , 563–586 (2010). find
  27. Abou-Chadi, T. & Krause, W. The causal effect of radical right success on mainstream parties’ policy positions: A regression discontinuity approach. British Journal of Political Science 50 , 829–847 (2020). find
  28. Meguid, B. M. Competition between unequals: The role of mainstream party strategy in niche party success. American Political Science Review 99 , 347–359 (2005). find
  29. Gessler, T. & Hunger, S. How the refugee crisis and radical right parties shape party competition on immigration. Political Science Research and Methods 10 , 524–544 (2022). find
  30. Völker, T. & Saldivia Gonzatti, D. Discourse networks of the far right: How far-right actors become mainstream in public debates. Political Communication 41 , 353–372 (2024). find
  31. Renström, E. A., Bäck, H. & Carroll, R. Threats, Emotions, and Affective Polarization. Political Psychology 44 , 1337–1366 (2023). find
  32. Wagner, M. Affective polarization in multiparty systems. Electoral Studies 69 , 102199 (2021). find
  33. Berning, C. C. Alternative für Deutschland (AfD)–Germany’s new radical right-wing populist party. ifo DICE Report 15 , 16–19 (2017). find
  34. Lewandowsky, M., Giebler, H. & Wagner, A. Rechtspopulismus in Deutschland. Eine empirische Einordnung der Parteien zur Bundestagswahl 2013 unter besonderer Berücksichtigung der AfD. Politische Vierteljahresschrift 57 , 247–275 (2016). find
  35. Reuters. German spy agency pauses ‘extremist’ classification for AfD party pending court case (2025). URL https://www.reuters.com/world/europe/german-spy -agency-pauses-extremist-classification-afd-party-local-court-says-2025-05-08/. ↗
  36. Deutsche Welle. DW captures attack on Green Party politician in Dresden (2025). URL https://www.dw.com/en/dw-captures-attack-on-green-party-politician-i n-dresden/a-69029359. ↗
  37. Backes, U. in Extremismus und politisch motivierte gewalt im vereinten deutschland (ed.Enzmann, B.) Handbuch Politische Gewalt 363–395 (Springer VS, Wiesbaden, 2013). find
  38. BKA. BKA - Politisch motivierte Kriminalität - Vorstellung der Fallzahlen zur Politisch motivierten Kriminalität 2023 (2024). URL https://www.bka.de/DE/ UnsereAufgaben/Deliktsbereiche/PMK/PMKZahlen2023/PMKZahlen2023.htm l. ↗
  39. Rydgren, J. Is extreme right-wing populism contagious? explaining the emergence of a new party family. European Journal of Political Research 44 , 413–437 (2005). find
  40. van de Wardt, M., Pilet, J.-B., Schumacher, G. & van Witteloostuijn, A. Contagion from abroad: How party entry in western europe is influenced by party family members abroad, 1961–2016. Comparative Political Studies 57 (2024). find
  41. Terren, L. & Soler-i Martı́, R. “Glocal” and transversal engagement in youth social movements: A Twitter-based case study of Fridays For Future-Barcelona. Frontiers in Political Science 3 , 635822 (2021). find
  42. Barberá, P. et al. Who leads? who follows? measuring issue attention and agenda setting by legislators and the mass public using social media data. American Political Science Review 113 , 883–901 (2019). find
  43. Lewandowsky, S., Jetter, M. & Ecker, U. K. H. Using the president’s tweets to understand political diversion in the age of social media. Nature Communications 11 , 5764 (2020). find
  44. Krijestorac, H., Garg, R. & Mahajan, V. Cross-platform spillover effects in consumption of viral content: A quasi-experimental analysis using synthetic controls. Information Systems Research 31 , 449–472 (2020). find
  45. Sältzer, M. et al. Twitter-Accounts der Kandidierenden zur Bundestagswahl 2021 (GLES) (2023). Version 2.1.0, ZA7721 [Datensatz]. find
  46. Abrami, G., Bagci, M., Hammerla, L. & Mehler, A. German Parliamentary Corpus (GerParCor) , 1900–1906 (European Language Resources Association, 2022). URL https://aclanthology.org/2022.lrec-1.202/. ↗
  47. Abrami, G., Bagci, M. & Mehler, A. German parliamentary corpus (GerParCor) reloaded , 7707–7716 (ELRA and ICCL, 2024). URL https://aclanthology.org/2 024.lrec-main.681/. ↗
  48. Richter, F. et al. Open discourse (2020). URL https://doi.org/10.7910/DVN/FI KIBO. ↗
  49. Jolly, S. et al. Chapel Hill Expert Survey trend file, 1999–2019. Electoral Studies 75 , 102420 (2022). find
  50. Heinze, A.-S. Dealing with the populist radical right in parliament: Mainstream party responses toward the Alternative for Germany. European Political Science Review 14 , 333–350 (2022). find
  51. Decker, O., Kiess, J., Heller, A., Schuler, J. & Brähler, E. Die Leipziger Autori-tarismus Studie 2022: zentrale Ergebnisse. Forschungsjournal Soziale Bewegungen 36 , 6–17 (2023). find
  52. Böhmer, H., Diermeier, M., Mertens, A. & Sommer, J. Strack-Zimmermann schlägt AfD: Der TikTok-Europawahlkampf. IW-Kurzbericht 33/2024, Köln (2024). URL https://hdl.handle.net/10419/299131. ↗
  53. Deutscher Bundestag. Plenarprotokolle des Deutschen Bundestages (2025). URL https://dserver.bundestag.de/btp/. ↗
  54. Grootendorst, M. BERTopic: Neural topic modeling with a class-based TF-IDF procedure. arXiv (Cornell University) (2022). URL https://arxiv.org/abs/2203 .05794. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/saleth-2026-german-parties-shifted-towards-intuition/