In 2026, NeTS organized a series of seminars to bridge the gap between language acquisition in linguistic theory and machine learning, with Roni Katzir and Charles Yang, among others.
The NeTS team has participated in the BabyLM Challenge competitions since 2024. The objective was to train small language models (BabyLMs) under a limited word budget, using specific architectures and morphologically driven tokenization strategies. Some results: Chesi et al. (2024) Different Ways to Forget - Fusco et al. (2025) Linguistic Units as Tokens.
In 2024, the NeTS team began investigating the complex relationship between categorical (grammaticality) judgments and gradient (acceptability) judgments. Check some of the findings from the T-GRA2L project.
In 2013, a group of prominent generative linguists (R. Kayne, R. Frank, G. Longobardi, P. Svenonius, I. Roberts, A. Belletti, G. Cinque, A. M. Di Sciullo, D. Sportiche, L. Rizzi, M. Den Dikken, R. Manzini, N. Richards, D. Delfitto, H. van Riemsdijk, H. Haider) discussed @ NeTS core problems in linguistic theorizing.