Comment des #méthodes d’analyse textuelle peuvent-elles nous aider à mieux comprendre le discours de #démission de F. #Legault ?
Je publie dans @ledevoir.com une analyse disponible ici de façon interactive : antoinelemor.github.io/assets/prese...
www.ledevoir.com/opinion/idee...
5/5 Conclusion: a political communication exercise. Little explanation of departure. Heavy justification of mandate. Protecting the CAQ's image.
💻 Open source code: github.com/antoinelemor/nlp-pol
#Legault #Quebec #CAQ #CanadaPolitics #DataScience #NLP
4/5 👥 149 positive actor mentions vs 5 negative — little bitterness, lots of thanks, no self-criticism.
😊 Emotional tone: +0.33 (positive). Pride and gratitude dominate despite stepping down.
3/5 ⚖️ 6 justifications for resigning vs 99 defending his record
🏛️ Dominant theme: party politics (24 mentions), ahead of healthcare (15) and economy (13). He talks more about his party than public policy.
2/5 📈 72% of the speech dedicated to his legacy
This isn't a resignation speech. It's a legacy speech. Legault explains less than he defends.
1/5 🧵 NLP-POL #3: Legault resigns. What does statistical analysis of his speech reveal?
🔍 Interactive analysis: antoinelemor.github.io/assets/presentations/nlp-pol-legault-resignation-en.html
I analyzed 122 sentences from the speech using machine learning methods. Here's what the data shows. ⬇️
Full analysis, slides, code and methodology : antoinelemor.github.io/blog/2026/nl...
#NLP #OpenSource #Macron #Diplomacy #Python #politicalscience #LLM #frenchdiplomacy #computational #socialsciences
9/9
Tools used (all open source):
→ Transcribe-tool (transcription): github.com/antoinelemor/Transcribe-tool
→ LLM_Tool (annotation): github.com/antoinelemor/LLM_Tool
8/9
On tone: 51% pragmatic + confident, only 7% alarmist. Combative peaks (12%) are targeted and reserved for defense and sovereignty issues.
7/9
On policy content: focus on diplomatic method (24%) rather than concrete measures (3%). The "how" of negotiating matters more than the "what".
6/9
On France's positioning: 85% agency score. France presented as active agent (46%), partner (23%), leader (8%). Victim in only 3% of cases.
5/9
On sentiment toward key actors: explicit European anchoring. France (+0.67) and Europe (+0.51) positive. US slightly negative (-0.06), China clearly negative (-0.42). Ukraine and India = preferred partners (+0.75). Russia? Barely mentioned.
4/9
On worldview: Geopolitical Anxiety Index of +0.08. Neither catastrophist nor triumphalist. 61% of geopolitical frames emphasize opportunities, 39% threats.
3/9
On rhetorical structure: deliberate alternation between confident and alarmist registers. Almost symmetrical. Macron systematically balances threats and opportunities. A (very) controlled rhetorical pendulum. 2/9
#NLPPOL - Second analysis of the year: Emmanuel Macron's annual address to French ambassadors (January 6, 2026).
Goal of #NLPPOL : popularize NLP in political science and showcase concrete applications with my open source tools Here's what the data reveals. 🧵
antoinelemor.github.io/blog/2026/nl...
Full analysis, slides, code and methodology:
antoinelemor.github.io/blog/2026/nl...
Anyone can use and critique. That's science. 9/9
Tools used (all open source):
→ Transcribe-tool (audio transcription):
github.com/antoinelemor/Transcribe-tool
→ LLM_Tool (annotation framework):
github.com/antoinelemor/LLM_Tool 8/9
On responses to journalists: 72% partial answers.
Concrete questions (governance, diplomacy, legality) remain unanswered.
An operation without an exit plan? 7/9
antoinelemor.github.io/blog/2026/nl...
On "us vs them" discourse:
American entities → positive score (+0.39)
Foreign entities → negative score (-0.53)
A binary discourse that serves to justify an assumed regional interventionism. 6/9
antoinelemor.github.io/blog/2026/nl...
Revealing detail: only Trump mentions humanitarian concerns. Never his advisors. 5/9
antoinelemor.github.io/blog/2026/nl...
On themes: journalists ask about governance, diplomacy, legal framework. Trump responds about oil and the economy. The operation is justified by resources, not by protecting populations. 4/9
antoinelemor.github.io/blog/2026/nl...
On rhetorical posture: Trump is not the most aggressive. Hegseth and Rubio carry the most offensive discourse, proportionally. 3/9
antoinelemor.github.io/blog/2026/nl...
Analysis of Trump's press conference on Maduro's forced extradition (January 3, 2026). 858 sentences analyzed. Here's what the data reveals. 2/9
antoinelemor.github.io/blog/2026/nl...
For 2026, I'm launching #NLP - POL: regular analyses using natural language processing applied to politics
Goal: popularize these methods, and showcase concrete applications with my open source tools, links available in the🧵
1st case #Trump on #Venezuela 🧵/9
antoinelemor.github.io/blog/2026/nl...
« Une étude parue dans la "Revue française de science politique" décrit une pénétration "continue, progressive et transpartisane" des idées d’extrême droite dans les discours de politique générale des premiers ministres depuis les années 1970 » @lemonde.fr www.lemonde.fr/idees/articl...
5/ 🛠️ L’outil (en constante évolution) est accessible ici :
github.com/antoinelemor... N’hésitez pas à tester, forker, commenter… et à me faire part de vos retours pour les prochains développements.
3/ L’outil a été développé avec le concours de Shannon Dinan, Jérémy Gilbert, le @capp-ul.bsky.social et des membres de la @clessn.bsky.social. Objectif : démocratiser l’usage de l’IA et du machine learning en sciences sociales, tout en respectant des standards élevés de rigueur scientifique.
2/ Dans cette présentation, je reviens sur : la logique générale de l’outil, des exemples concrets d’annotations, quelques pistes d’usage en sciences sociales et en analyse des politiques publiques.
1/ 🧵 Pr ceux qui n’ont pas pu assister à ma présentation de LLM Tool à l’@ulaval.ca et le @capp-ul.bsky.social, voici la rediffusion. LLM Tool est une plateforme simple qui permet d’annoter n’importe quel texte à partir de catégories définies par ses utilisateurs www.youtube.com/watch?v=C0vN...
If you're interested in these research methods, we welcome empirical applications, methodological innovations, and/or theoretically informed or critical contributions. The call for papers is open until 20 January 2026! Looking forward to meeting you there! 3/3
