Low-Resource Cross-Lingual Adaptive Training for Nigerian Pidgin
This work addresses the challenge of processing low-resource languages like Nigerian Pidgin, which is incremental as it builds on existing pre-trained models with adaptive techniques.
The paper tackled the problem of developing spoken language processing systems for low-resource Nigerian Pidgin by collecting a parallel English-Pidgin corpus and proposing a cross-lingual adaptive training framework, resulting in up to 2.38 BLEU improvements in translation tasks.
Developing effective spoken language processing systems for low-resource languages poses several challenges due to the lack of parallel data and limited resources for fine-tuning models. In this work, we target on improving upon both text classification and translation of Nigerian Pidgin (Naija) by collecting a large-scale parallel English-Pidgin corpus and further propose a framework of cross-lingual adaptive training that includes both continual and task adaptive training so as to adapt a base pre-trained model to low-resource languages. Our studies show that English pre-trained language models serve as a stronger prior than multilingual language models on English-Pidgin tasks with up to 2.38 BLEU improvements; and demonstrate that augmenting orthographic data and using task adaptive training with back-translation can have a significant impact on model performance.