Context Volume Drives Performance: Tackling Domain Shift in Extremely Low-Resource Translation via RAG
This addresses domain shift challenges for low-resource language translation, though it is incremental as it builds on existing methods like fine-tuning and RAG.
The paper tackles performance degradation in neural machine translation for low-resource languages under domain shift, using Dhao as a case study, and recovers most of the loss with a hybrid NMT-LLM-RAG framework, achieving a chrF++ score of 35.21 from 27.11.
Neural Machine Translation (NMT) models for low-resource languages suffer significant performance degradation under domain shift. We quantify this challenge using Dhao, an indigenous language of Eastern Indonesia with no digital footprint beyond the New Testament (NT). When applied to the unseen Old Testament (OT), a standard NMT model fine-tuned on the NT drops from an in-domain score of 36.17 chrF++ to 27.11 chrF++. To recover this loss, we introduce a hybrid framework where a fine-tuned NMT model generates an initial draft, which is then refined by a Large Language Model (LLM) using Retrieval-Augmented Generation (RAG). The final system achieves 35.21 chrF++ (+8.10 recovery), effectively matching the original in-domain quality. Our analysis reveals that this performance is driven primarily by the number of retrieved examples rather than the choice of retrieval algorithm. Qualitative analysis confirms the LLM acts as a robust "safety net," repairing severe failures in zero-shot domains.