Modeling orthographic variation in Nigerian Pidgin, an English-derived contact language, can improve performance on critical NLP tasks such as sentiment analysis and machine translation.
