Latest Research in Data Augmentation for Low Resource Language Text Translation: A Review

The translation of low-resource languages remains a significant challenge in Natural Language Processing (NLP) due to the scarcity of high-quality parallel data for training machine translation models. Data augmentation techniques, which artificially expand the size and diversity of datasets, offer a promising solution to this problem. This study comprehensively reviews data augmentation techniques in low-resource language text translation. Sixteen recent articles met the specified eligibility criteria employing the systematic literature review protocol. The review categorized the various data augmentation methods into eight groups: translation-based augmentation, synthetic data generation, sentence manipulation, grammar and error correction, multi-task and advanced transformations, miscellaneous transformations, consistency and self-training, and embedding and contextual methods. The findings highlight the significant impact of data augmentation on improving translation quality, addressing data scarcity, and enhancing model robustness. Despite these advancements, challenges such as grammatical errors, semantic inconsistencies, and the quality of synthetic data remain. This review also provides insights into the effectiveness of different language pairs and dataset sizes, emphasizing the need for further research to refine these methods and address challenges in low-resource languages.