Testing Low-Resource Language Support in LLMs Using Language Proficiency Exams: the Case of Luxembourgish
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909991221329920 |
|---|---|
| author | Lothritz, Cedric Cabot, Jordi Bernardy, Laura |
| author_facet | Lothritz, Cedric Cabot, Jordi Bernardy, Laura |
| contents | Large Language Models (LLMs) have become an increasingly important tool in research and society at large. While LLMs are regularly used all over the world by experts and lay-people alike, they are predominantly developed with English-speaking users in mind, performing well in English and other wide-spread languages while less-resourced languages such as Luxembourgish are seen as a lower priority. This lack of attention is also reflected in the sparsity of available evaluation tools and datasets. In this study, we investigate the viability of language proficiency exams as such evaluation tools for the Luxembourgish language. We find that large models such as Claude and DeepSeek-R1 typically achieve high scores, while smaller models show weak performances. We also find that the performances in such language exams can be used to predict performances in other NLP tasks in Luxembourgish. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_01667 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Testing Low-Resource Language Support in LLMs Using Language Proficiency Exams: the Case of Luxembourgish Lothritz, Cedric Cabot, Jordi Bernardy, Laura Computation and Language Large Language Models (LLMs) have become an increasingly important tool in research and society at large. While LLMs are regularly used all over the world by experts and lay-people alike, they are predominantly developed with English-speaking users in mind, performing well in English and other wide-spread languages while less-resourced languages such as Luxembourgish are seen as a lower priority. This lack of attention is also reflected in the sparsity of available evaluation tools and datasets. In this study, we investigate the viability of language proficiency exams as such evaluation tools for the Luxembourgish language. We find that large models such as Claude and DeepSeek-R1 typically achieve high scores, while smaller models show weak performances. We also find that the performances in such language exams can be used to predict performances in other NLP tasks in Luxembourgish. |
| title | Testing Low-Resource Language Support in LLMs Using Language Proficiency Exams: the Case of Luxembourgish |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2504.01667 |