analiza aiqbench
This research introduces AfriSUD, a critical dataset of syntactically annotated treebanks for nine diverse African languages designed to evaluate the limitations of current NLP models. The study highlights a significant "syntax gap," demonstrating that existing architectures struggle to capture structural features like agglutination and tone prevalent in these underrepresented language families. This work is essential for advancing equitable AI development by providing high-quality, native-speaker verified data necessary for training robust multilingual systems.
AfriSUD: A Dependency Treebank Collection for Evaluating Models on African Languages
arXiv:2606.12708v1 Announce Type: cross Abstract: Despite their linguistic diversity and global significance, African languages remain underrepresented in research and resources to support NLP. We aim to bridge this gap by introducing AfriSUD, the first large-scale collection of syntactically annotated treebanks for nine diverse African languages spanning major language families and regions across Sub-Saharan Africa. Using the Surface-Syntactic Universal Dependencies (SUD) framework, our community-led effort provides high-quality, native-speaker verified data that capture typological key features such as agglutination and tone. We evaluate a range of models on AfriSUD for part-of-speech tagging and dependency parsing including non-transformer baselines, multilingual pretrained encoders, and LLMs. Our results reveal a significant syntax gap, where models still show clear limitations across the nine languages, suggesting that existing architectures may not fully capture the structural diversity of African-language syntax.
Authors:Happy Buzaaba, Cheikh Mouhamadou Bamba Dione, David Ifeoluwa Adelani, Sylvain Kahane, Kim Gerdes, Bruno Guillaume, Kevin Guan, Aremu Anuoluwapo, Naome A. Etori, Shamsuddeen Hassan Muhammad, Utitofon Inyang, Peter Nabende, David Sabiiti Bamutura, Andiswa Bukula, Chinedu Uchechukwu, Rooweither Mabuya, Idris Akinade, Christiane Fellbaum
Abstract:Despite their linguistic diversity and global significance, African languages remain underrepresented in research and resources to support NLP. We aim to bridge this gap by introducing AfriSUD, the first large-scale collection of syntactically annotated treebanks for nine diverse African languages spanning major language families and regions across Sub-Saharan Africa. Using the Surface-Syntactic Universal Dependencies (SUD) framework, our community-led effort provides high-quality, native-speaker verified data that capture typological key features such as agglutination and tone. We evaluate a range of models on AfriSUD for part-of-speech tagging and dependency parsing including non-transformer baselines, multilingual pretrained encoders, and LLMs. Our results reveal a significant syntax gap, where models still show clear limitations across the nine languages, suggesting that existing architectures may not fully capture the structural diversity of African-language syntax.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
arXiv:2606.12708 [cs.CL]
(or arXiv:2606.12708v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.12708 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 191 cuvinte
Subiecte conexe
AfriSUD: A Dependency Treebank Collection for Evaluating Models on African Languages
arXiv · 2 sept. 2026
Game-Theoretic Disempowerment: Loss of Control to Agencyless AI
LessWrong · 27 sept. 2026
Dependency Parsing Across the Resource Spectrum: Evaluating Architectures on High and Low-Resource Languages
arXiv · 15 oct. 2026
The Entangled Dimensions of Decision Theory
LessWrong · 30 iul. 2026