Computational Approaches to Hausa Natural Language Processing: Building Resources for an Under-Resourced Language

Hausa is among the most spoken languages in Africa but remains severely under-resourced in the computational linguistics and natural language processing domains, limiting the development of language technology applications that could dramatically expand access to information, education, and digital services for Hausa speakers. This doctoral study addresses the Hausa NLP resource gap through a comprehensive programme of computational resource development and natural language processing system building, making original technical and methodological contributions to the under-resourced language NLP field. Drawing on computational linguistics, machine learning, and language documentation methodology, the research develops a suite of interconnected Hausa language resources and NLP systems. The resource development component builds the largest publicly available Hausa annotated corpus (two million words), a morphological analyser handling Hausa's extensive morphological alternations, a part-of-speech tagged training dataset, a named entity recognition system, and a dependency-parsed treebank for syntactic modelling. Each resource is developed through carefully documented annotation methodologies designed to handle Hausa-specific linguistic challenges including tone marking inconsistency, code-mixing with Arabic and English, and dialectal variation. NLP system evaluation is conducted against held-out test sets with rigorous performance metrics. The research also develops and evaluates a low-resource machine translation system for Hausa-English text, addressing the particular challenges of translating a morphologically complex language with limited parallel data. Theoretical contributions to NLP methodology include novel approaches to handling tonal language annotation and low-resource morphological analysis. Keywords: Hausa NLP, Natural language processing, Computational linguistics, Under-resourced languages, Language technology.