Published October 20, 2026
| Version v1
Preprint
Open
Why Computational Humanities needs falsifiable theories
Authors/Creators
- Barré, Jean1
- Baumard, Nicolas2
- Bollée, Jeanne1
- Bluet, Alexandre2
- Camps, Jean-Baptiste3
- de Dampierre, Charles2, 4
- Dubourg, Edgar2
- Estebe, Sylvain5
- Fan, Zhaolun2
- Fiorio, Grégory2
- Fu, Ze2
- Geng, Mingmeng1
- Godreau, Ulysse3
- Henke, Konstantin1
- Liu, Ruoting2
- Maunoir, Paco2
- Hall Maudslay, Rowan3
- Morin, Olivier2
- Mpouli, Suzanne3
- Samani, Hossein2
- Seminck, Olga1
- Thouzeau, Valentin2
- Zhong, Ying2
- Amsili, Pascal1
- Landragin, Frédéric1
- Moins, Théo3
- 1. LATTICE, CNRS, École normale supérieure–PSL, Université Sorbonne Nouvelle, Montrouge, France.
- 2. Institut Jean Nicod, Département d'études cognitives, École normale supérieure–PSL, EHESS, CNRS, Paris, France.
- 3. Centre Jean-Mabillon, École nationale des chartes–PSL, Paris, France.
- 4. Bunka AI.
- 5. Center for Humanities Computing, School of Culture and Society, Aarhus University, Aarhus, Denmark.
Description
Computational humanities has powerful tools and abundant data, but too little research tests
falsifiable theories. We argue for a shift: dedicating a share of research effort to theory-driven
hypothesis testing, coexisting with descriptive scholarship and exploration. We review some past
attempts at building theory-driven research programmes in the humanities and outside of them,
some successful, some not. Successful turns towards theory-driven research do not diminish the
relevance or importance of descriptive work but make it more focused. We show how emerging
theories can supply predictive frameworks usable by computational humanities and outline
practical steps for bridging gaps between theory-driven and descriptive research programmes.
Files
collective-theories-paper.pdf
Additional details
Funding
- This work has received support under the Major Research Program of PSL Research University "CultureLab" launched by PSL Research University and implemented by ANR with the reference ANR-10-IDEX-0001.