Marvin Lavechin

Computational Psycholinguistics lab, MIT

picture.jpg

marvinlavechin@gmail.com

I am currently a Simons postdoctoral fellow in the Computational Psycholinguistics lab at the Massachusetts Institute of Technology (Cambridge, US) under the supervision of Roger Levy and Elika Bergelson. Formerly, at Meta AI, LSCP (Paris, France), and GIPSA (Grenoble, France).

My work is at the intersection of Artificial Intelligence and Cognitive Sciences. In my interdisciplinary research program, I study language development in infants and machines. In particular, my research articulates along two major axes: 1) developing and democratizing artificial neural networks to automatically analyze children’s language environment and linguistic production (supervised learning) and 2) running human-inspired computer simulations of language development to build more efficient algorithms and identify mechanisms driving language acquisition in infants (unsupervised/self-supervised learning).

News

Apr, 2025 [🤞 Preprint] Our comparison of the LENA and ACLEW algorithms to analyze children’s language environment across neurodevelopmental profiles has been submitted to Developmental Science! Grateful for this amazing collab. with Lisa Hamrick, Bridgette Kelleher, and Amanda Seidl!
Mar, 2025 [📺 Media] Our research has been featured in a French television documentary “L’homme à la machine: la mécanique du langage” (France 2).
Dec, 2024 [🎉 Funding] Glad to share I obtained funding from the Simons Center for the Social Brain for a 2-year postdoc at MIT! I’ll join the Computational Psycholinguistics lab starting May 1st, 2025 to work with Roger Levy and Elika Bergelson!
Sep, 2024 [📝 Paper] Our paper Decode, move and speak! Self-supervised learning of speech units, gestures, and sound relationships using vocal imitation has been accepted to Computational Linguistics!
Sep, 2023 [🎓 Ph.D. defense] Successfully defended my thesis on September, 2023! Yay!

Selected publications

  1. Simulating Early Phonetic and Word Learning Without Linguistic Categories
    Marvin Lavechin, Maureen Seyssel, Hadrien Titeux, and 4 more authors
    Developmental Science 2025
  2. Modeling the initial state of early phonetic learning in infants
    Maxime Poli, Thomas Schatz, Emmanuel Dupoux, and 1 more author
    Language Development Research 2024
  3. Decode, move and speak! Self-supervised learning of speech units, gestures, and sound relationships using vocal imitation
    Marc-Antoine Georges, Marvin Lavechin, Jean-Luc Schwartz, and 1 more author
    Computational Linguistics 2024
  4. Modeling early phonetic acquisition from child-centered audio data
    Marvin Lavechin, Maureen Seyssel, Marianne Métais, and 5 more authors
    Cognition 2024