1. médialab Sciences Po
  2. Actualités
  3. Séminaire de doctorants

Séminaire de doctorants

Le séminaire du 22 avril 2025 accueillera Katrin Herms, doctorante à l'Université de Lausanne, et Armin Pournaki, doctorant au Max Planck Institute for Mathematics in the Sciences, au Laboratoire Lattice et au médialab de SciencesPo.

Rendez-vous, Séminaire de recherche

⇥

Salle K.011, 1 place Saint Thomas d'Aquin 75007 Paris

"Between numbers and words: mixing discourse analysis and social network analysis to study the reconfiguration of the French-speaking tweetosphere during the Covid19 health crisis", by Katrin Herms

Abstract

Political communities on X (before Twitter) have changed their structure in the run-up to the French Presidential elections in 2022 with radical left- and radical right-leaning profiles sharing hashtags such as #touscontremacron (Chavalarias 2022). In this article, a mixed methods approach is developed to examine the reconfiguration of the French digital public space in the broader context of the Covid19 health crisis and through the lens of discourse analysis. We developed a qualitative and quantitative methodology to examine how content about the political management of the pandemic was received and interpreted by platform users in 2020. More precisely, our study examines collective ad-hoc framing of news in what we call “quote trees” (Roth, St-Onge & Herms 2022): a so far understudied, cascade like conversational phenomenon that occurs when a message is shared and commented on multiple times. We analyzed the 16 largest punctual controversies related to the keywords ‘covid’ and/or ‘corona’ triggered by the media profile @afpfr. We could visualize and assess the digital corpus for manual hand coding of 1220 tweets, and combined this approach with a structural network analysis of around 10 000 French speaking Twitter users. This triangulation of methods enabled us to compare claims with the Twitter users’ estimated political orientation. Our results propose explanatory avenues for the shift from horizontal to vertical polarization on Twitter.

Biography

Katrin Herms is a PhD candidate in sociology at the University of Lausanne, co-supervised by Laurence Kaufmann and Francesca Musiani. Her dissertation project, entitled “Polarization and solidarity in the context of crisis: a mixed-methods study of online and offline mobilization during the Covid19 pandemic”, investigates hybrid politicization processes in France and Germany.

Katrin is currently a lecturer in Digital Culture at the Sciences Po medialab, and a lecturer in Media Sociology at the University of Luzern in Switzerland. She is an associated member of Centre Marc Bloch in Berlin and Centre Internet et Société in Paris, both CNRS satellites. In 2024, she was a research fellow in the “Digital News Dynamics” group at the Weizenbaum-Institut in Berlin and a lecturer in Social Theory at Berlin’s Humboldt University. Katrin is also a member of the junior laboratory “Collectif d’étude des discours sur l’écologisme en France” of LabEx SMS, Toulouse University.

"Computational Approaches to Analyzing Political Narratives in Digital Traces", by Armin Pournaki

Abstract

Narratives are key interpretative devices by which humans make sense of political reality. As the significance of narratives for understanding current societal issues such as polarization and misinformation becomes increasingly evident, there is a growing demand for methods that support their empirical analysis. I shall present results from my thesis that fill this methodological gap by introducing a network-based approach for the inductive extraction of narratives from text. The method allows to empirically surface the role of (conflicting) narratives as key mechanisms for the formation of group identity and ideological polarization using large datasets of political speeches and social media posts (Reddit, Twitter/X).

Biography

Armin Pournaki is a joint PhD candidate at the Max Planck Institute for Mathematics in the Sciences, Leipzig, the Laboratoire Lattice (ENS-PSL), Paris and the SciencesPo médialab. He is currently working on methods for the computational extraction of (political) narratives from raw text and empirically investigating their role in phenomena like polarization and issue alignment. His thesis approaches these questions using a theory-driven combination of computational linguistics and network science.

Practical information

The session will take place on Tuesday, April 22, 2025, from 2:00 PM to 4:30 PM, in person and in English at Sciences Po, Room K011, 1 Place Saint-Thomas-d'Aquin, Paris 75007.

Registration is mandatory via this link.