Multilingual Communication in the Asylum Context: Evaluating LLM-Based Machine Translation with Fuzzy Match Augmentation and Adaptive NMT across Resource Conditions under Low-Data Constraints

Thomas Moerman; Arda Tezcan; Lieve Macken

Multilingual Communication in the Asylum Context: Evaluating LLM-Based Machine Translation with Fuzzy Match Augmentation and Adaptive NMT across Resource Conditions under Low-Data Constraints

EAMT 2026 (accepted, best-paper shortlist)

Machine Translation

Low-resource

Fuzzy-match augmentation works with a 358-sentence translation memory across 14 languages. Best-paper shortlist.

Authors

Thomas Moerman

Arda Tezcan

Lieve Macken

Published

May 1, 2026

Retrieval-augmented LLM translation is compared with adaptive NMT across 14 languages under an extremely small translation memory. Fuzzy-match retrieval beats random and zero-shot for 13 of 14 languages, with the largest gains for the lowest-resource languages.

Research theme: Machine translation

Citation

BibTeX citation:

@inproceedings{moerman2026,
  author = {Moerman, Thomas and Tezcan, Arda and Macken, Lieve},
  title = {Multilingual {Communication} in the {Asylum} {Context:}
    {Evaluating} {LLM-Based} {Machine} {Translation} with {Fuzzy}
    {Match} {Augmentation} and {Adaptive} {NMT} Across {Resource}
    {Conditions} Under {Low-Data} {Constraints}},
  booktitle = {Proceedings of the 2026 Conference of the European
    Association for Machine Translation},
  date = {2026-05-01},
  langid = {en}
}

For attribution, please cite this work as:

Moerman, Thomas, Arda Tezcan, and Lieve Macken. 2026. “Multilingual Communication in the Asylum Context: Evaluating LLM-Based Machine Translation with Fuzzy Match Augmentation and Adaptive NMT Across Resource Conditions Under Low-Data Constraints.” Proceedings of the 2026 Conference of the European Association for Machine Translation, accepted, May 1.