Interplay of Machine Translation, Diacritics, and Diacritization
This work addresses the problem of optimizing machine translation and diacritization systems for linguists and developers, particularly in low-resource languages, but it is incremental as it builds on existing multi-task learning approaches.
The study tackled how machine translation and diacritization interact in multi-task learning and the effect of diacritics on translation performance, finding that diacritization doubles or triples MT performance in low-resource settings but harms it in high-resource ones, while MT performance is similar with or without diacritics.
We investigate two research questions: (1) how do machine translation (MT) and diacritization influence the performance of each other in a multi-task learning setting (2) the effect of keeping (vs. removing) diacritics on MT performance. We examine these two questions in both high-resource (HR) and low-resource (LR) settings across 55 different languages (36 African languages and 19 European languages). For (1), results show that diacritization significantly benefits MT in the LR scenario, doubling or even tripling performance for some languages, but harms MT in the HR scenario. We find that MT harms diacritization in LR but benefits significantly in HR for some languages. For (2), MT performance is similar regardless of diacritics being kept or removed. In addition, we propose two classes of metrics to measure the complexity of a diacritical system, finding these metrics to correlate positively with the performance of our diacritization models. Overall, our work provides insights for developing MT and diacritization systems under different data size conditions and may have implications that generalize beyond the 55 languages we investigate.