Subword Segmental Machine Translation: Unifying Segmentation and Target Sentence Generation
This addresses the challenge of improving translation quality, especially for morphologically rich languages in low-resource settings, by integrating segmentation into the translation process, representing a novel paradigm shift rather than an incremental improvement.
The paper tackles the problem of subword segmentation being a separate preprocessing step in machine translation, which can limit translation quality, by proposing subword segmental machine translation (SSMT) that unifies segmentation and translation in a single trainable model. Experiments across 6 translation directions show SSMT improves chrF scores for morphologically rich agglutinative languages, with strongest gains in very low-resource scenarios.
Subword segmenters like BPE operate as a preprocessing step in neural machine translation and other (conditional) language models. They are applied to datasets before training, so translation or text generation quality relies on the quality of segmentations. We propose a departure from this paradigm, called subword segmental machine translation (SSMT). SSMT unifies subword segmentation and MT in a single trainable model. It learns to segment target sentence words while jointly learning to generate target sentences. To use SSMT during inference we propose dynamic decoding, a text generation algorithm that adapts segmentations as it generates translations. Experiments across 6 translation directions show that SSMT improves chrF scores for morphologically rich agglutinative languages. Gains are strongest in the very low-resource scenario. SSMT also learns subwords that are closer to morphemes compared to baselines and proves more robust on a test set constructed for evaluating morphological compositional generalisation.