Grammar Control in Dialogue Response Generation for Language Learning Chatbots
Dominik Glandorf, Peng Cui, Detmar Meurers, Mrinmaya Sachan · 2025
Read Grammar Control in Dialogue Response Generation for Language Learning Chatbots by Dominik Glandorf, Peng Cui, Detmar Meurers, Mrinmaya Sachan (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-2502-07544.txt.
Description
Grammar can be controlled in chatbot replies for language learners, Dominik Glandorf and co-authors show in a 2025 paper. Large language model chatbots offer cheap conversation practice. Yet their language is hard to steer toward the grammar forms a learner needs right now, which limits their teaching value in a lesson plan.
The team grounds a dialogue response model in a pedagogical repository of grammar skills. They evaluate prompting, fine-tuning, and decoding strategies for grammar-controlled generation in a comprehensive way. Strategic decoding of one model beat a commercial model when small quality losses were allowed. Simulations predict that such replies can support grammar learning matched to a learner's level. Existing learning chatbots stand to gain from that control.
Catalog record
| CREATOR | Dominik Glandorf, Peng Cui, Detmar Meurers, Mrinmaya Sachan |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 59 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-07544