Thomas Pellard
thomaspellard.bsky.social
Thomas Pellard
@thomaspellard.bsky.social
CNRS Research scientist 🇫🇷 historical-comparative linguistics, archaeolinguistics, phylolinguistics, geolinguistics, endangered languages, languages of Japan
Reposted by Thomas Pellard
En fait une partie du problème réside aussi dans le fait de penser que les textes de SHS devraient être faciles à lire et comprendre sans formation parce que finalement c'est rien de très sérieux (alors que les maths, personne ne considère devoir pouvoir comprendre sans formation).
October 28, 2025 at 3:43 PM
Reposted by Thomas Pellard
Tu connais certainement mon argument: il n'y aurait pas tant d'attirance pour l'IA générative, pour produire ou résumer des documents, s'il n'y avait pas tant de documents d'utilité douteuse.
October 25, 2025 at 11:49 AM
Reposted by Thomas Pellard
Switching to codeberg.org for most recent projects was the solution for me. It's free, they don't use data for AI training, and since they don't (yet) offer integration with Zenodo etc. , I learned to embrace the moment when I upload a Zip-folder to Zenodo to publish some supplement :-)
Codeberg.org
Codeberg is a non-profit, community-led organization that aims to help free and open source projects prosper by giving them a safe and friendly home.
codeberg.org
October 17, 2025 at 6:22 AM
Derived adjectives with ma- are indeed na-adjectives, so you should use da, but da can be dropped in the sooken informal style, especially when there is no final particle. But 顔真っ赤だ is correct. And yes, I triggers irregular alternations like aka -> makka and ao -> massao
October 14, 2025 at 6:43 PM