معرفی
Nicolas Ballier is a Professor at Université Paris Cité (formerly Université Paris Diderot), where he teaches and conducts research in linguistics and digital humanities. Previously, he taught at the Université de Rouen and Paris 13. His work focuses on the intersection of linguistics, computational methods, and language learning technologies, with particular expertise in neural machine translation and speech processing.
His primary research interests include:
- Corpus prosody
- Neural machine translation
- Automatic analysis of learner corpora
- Digital humanities
- Epistemology of linguistics (third revolution of grammatisation)
- Interpretability of neural machine translation systems
- Representation of speech in Whisper audio models
Dr. Ballier's recent work explores how computers transform linguistic data (the 'third revolution of grammatisation'), with a focus on neural machine translation interpretability and speech analysis using large language models like Whisper. His research bridges theoretical linguistics with practical applications in language learning and translation technologies, particularly focusing on how these technologies can be made transparent and useful for translators and language learners.
His 15 most recent publications (2022-2024) demonstrate a strong focus on neural machine translation interpretability, Whisper applications for language assessment, and learner corpus analysis. The publications span multiple prestigious venues including EAMT, ACL, LREC-COLING, and specialized journals in speech technology and computational linguistics, showing consistent productivity and impact in his fields.
Dr. Ballier has been PI or team member on numerous European-funded projects including DOKTORAND (2012-2016), KVARK project (2014-2026), PHC Ulysses (2019), multitraiNMT (2021), and LT-LIder project (2024-Nov 2026). He has developed platforms like PAPTAN for neural machine translation experiments and MAKE-NMT VIZ for investigating machine translation interpretability.
He has supervised PhD students through collaborative projects and has been involved in research initiatives like DLLA (Deep Learning for Language Assessment) exploring CEFR levels with keylog data, Neuroviz (2021-2022), and SPECTRANS (2020-2022) focusing on specialized neural translation and probing information flow in neural networks.
