PidginUNMT: Unsupervised Neural Machine Translation from West African Pidgin to English
This work reduces the barrier for future NLP research on West African Pidgin English, which has over 80 million speakers but no prior NLP work, though it is incremental as it applies existing methods to a new language.
This paper tackled the lack of natural language processing (NLP) work on West African Pidgin English by creating the first NLP resources and models for it, resulting in an unsupervised neural machine translation model that achieved BLEU scores of 7.93 from Pidgin to English and 5.18 from English to Pidgin.
Over 800 languages are spoken across West Africa. Despite the obvious diversity among people who speak these languages, one language significantly unifies them all - West African Pidgin English. There are at least 80 million speakers of West African Pidgin English. However, there is no known natural language processing (NLP) work on this language. In this work, we perform the first NLP work on the most popular variant of the language, providing three major contributions. First, the provision of a Pidgin corpus of over 56000 sentences, which is the largest we know of. Secondly, the training of the first ever cross-lingual embedding between Pidgin and English. This aligned embedding will be helpful in the performance of various downstream tasks between English and Pidgin. Thirdly, the training of an Unsupervised Neural Machine Translation model between Pidgin and English which achieves BLEU scores of 7.93 from Pidgin to English, and 5.18 from English to Pidgin. In all, this work greatly reduces the barrier of entry for future NLP works on West African Pidgin English.