Automatically Suggesting Diverse Example Sentences for L2 Japanese Learners Using Pre-Trained Language Models * Research conducted during internship at NII, Japan.
Enrico Benedetti, Akiko Aizawa, Florian Boudin · 2025
Read Automatically Suggesting Diverse Example Sentences for L2 Japanese Learners Using Pre-Trained Language Models * Research conducted during internship at NII, Japan. by Enrico Benedetti, Akiko Aizawa, Florian Boudin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-03580.txt.
Description
Enrico Benedetti, Akiko Aizawa, and Florian Boudin test ways to suggest example sentences to learners of Japanese in a 2025 paper. Good examples should be varied and matched to a learner level. The team used pre trained language models in two ways. One picked sentences from a new collection of Japanese sentences, with the model scoring quality. The other wrote new sentences from scratch.
Raters included learners, native speakers, and a large language model. They scored sentences for difficulty, variety, and naturalness. Raters often disagreed, except on difficulty. Still, all groups preferred the retrieval method, above all for beginners and advanced learners, while direct writing scored lower on average.
Catalog record
| CREATOR | Enrico Benedetti, Akiko Aizawa, Florian Boudin |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-03580