Assessing Small Language Models as Text Simplification Evaluators

David Carranza Navarrete, Jan Bakker, Jaap Kamps


Abstract
Text simplification requires reliable automatic evaluation, yet existing learnable metrics such as LENS and LENS-SALSA are specialized and costly to develop. Moreover, it remains unclear how these metrics compare to using large language models (LLMs) as evaluators. Exploring this question is important because LLM-based evaluation could make simplification research and deployment more flexible and easier to adapt than training new task-specific metrics for each setting. In this work, we empirically compare several small, open-weight instruction-tuned LLMs with LENS and LENS-SALSA in both reference-based and reference-free evaluation settings. We measure their alignment with human judgments across multiple datasets. Our results provide insight into when small LLMs can serve as effective evaluators and when specialized metrics remain preferable, informing the design of future evaluation pipelines for text simplification and related text generation tasks.
Anthology ID:
2026.determit-1.9
Volume:
Proceedings of the 2nd Workshop on Evaluating Text Difficulty in a Multilingual Context (DeTermIt! 2026)
Month:
May
Year:
2026
Address:
Palma, Mallorca (Spain)
Editors:
Giorgio Maria Di Nunzio, Federica Vezzani, Liana Ermakova, Hosein Azarbonyad, Jaap Kamps
Venues:
DeTermIt | WS
SIG:
Publisher:
ELRA Language Resources Association (ELRA)
Note:
Pages:
83–86
Language:
External URL:
https://lrec.elra.info/lrec2026-ws-determit-09
DOI:
10.63317/54qbtfwgkz9j
Bibkey:
Cite (ACL):
David Carranza Navarrete, Jan Bakker, and Jaap Kamps. 2026. Assessing Small Language Models as Text Simplification Evaluators. In Proceedings of the 2nd Workshop on Evaluating Text Difficulty in a Multilingual Context (DeTermIt! 2026), pages 83–86, Palma, Mallorca (Spain). ELRA Language Resources Association (ELRA).
Cite (Informal):
Assessing Small Language Models as Text Simplification Evaluators (Carranza Navarrete et al., DeTermIt 2026)
Copy Citation: