Welcome to our homepage! InCLow stands for Interpretable, Cognitively inspired, Low-resource Language Models.
We are a thriving research team led by Arianna Bisazza, working at the intersection of natural language processing and (cognitive) linguistics. Our projects revolve around two key questions:
How to develop reliable language technologies that work well for a large variety of languages?
What can we learn about human language while attempting to do that?
We’re proudly part of the larger Computational Linguistics group (GroNLP) at the University of Groningen, the Netherlands.
Take a look at our team members, publications, and reading group agenda!
Latest News
[Jun 2025] Arianna gave a talk at the COLT Symposium, organized by Gemma Boleda and Marco Baroni at UPF, Barcelona. This was followed by a round-table on the convergence (or lack thereof) between findings coming from the fields of experimental and theoretical linguistics, neuroscience and the study of large language models.
[May 2025] Arianna and Gabriele were interviewed by Groningen’s Oog Radio station. The questions covered (of course) AI, but also the GroNLP group, what makes this research field so exciting (and a bit crazy nowadays!)… and a bit of music :) Recording available here: HappyHourFM.
[May 2025] Jaap gave a talk in Edinburgh on MultiBLiMP: A Multilingual Benchmark of Linguistic Minimal Pairs.
[May 2025] Master student Anais Almendra from the University of Chile joined us for a 3-month visit. She works on morphological analysis of the endangered language Madupungun.
[April 2025] Gabriele presented his research on word-level quality estimation for post-editing and context attribution for language models at the Multilinguality and Language Technology group of DFKI Saarland and the DEEL/FOR teams at IRT Saint-Éxupery.
[April 2025] Jirui gave two talks at KPN and University of Amsterdam on three topics on Retrieval Augmented Generation (MIRAGE, likelihood gauges answer accuracy, and Consistency in Multilingual Context Utilization)
[Mar 2025] Arianna gave a keynote talk at NoDaLiDa/Baltic-HLT 2025, titled “Not all Language Models need to be Large: Studying Language Evolution and Acquisition with Modern Neural Networks”.
[Feb 2025] PhD student Akari Haga from NAIST, Japan, joined us for a 6-month visit. She works on BabyLM-style models for Japanese.