Congratulations to Prof. Joseph Marvin Imperial, Faculty of the NU Manila College of Computing and Information Technologies, for leading the UniversalCEFR Initiative—a global research collaboration across 13 universities and institutions dedicated to democratizing multilingual language proficiency assessment for AI and education. The project brings together experts from Cardiff University, University of Exeter, University of Gothenburg, Bielefeld University, the National Research Council of Canada, and more, to develop the largest open CEFR-aligned dataset for fair and inclusive language evaluation in AI.

The project addresses a critical gap in multilingual AI research—the lack of unified and accessible datasets that align with the Common European Framework of Reference for Languages (CEFR). By curating and standardizing CEFR-labeled datasets across 13 languages, the UniversalCEFR project enables researchers to develop and evaluate AI models that can more fairly and accurately assess language ability across linguistic and cultural contexts. The initiative’s flagship contribution, the UniversalCEFR Dataset, represents the largest and most diverse open compilation of CEFR-aligned corpora to date. It provides researchers and educators with a transparent, standardized resource for benchmarking AI models in language learning, assessment, and educational technology applications.

This groundbreaking research will be presented at EMNLP 2025, an A*-ranked international conference on Natural Language Processing, to be held in China this November.

Learn more: https://universalcefr.github.io/