Developing a Computational Lexicon for Igbo: Challenges and Solutions for Natural Language Processing Applications

This study addresses the development of a computational lexicon for the Igbo language, making an original theoretical and applied contribution to the natural language processing infrastructure for a major West African language that remains severely under-resourced in computational linguistics. The absence of adequate computational lexical resources for Igbo impedes the development of machine translation, speech recognition, and digital literacy tools that could significantly expand access to technology for Igbo-speaking populations in Nigeria and the diaspora. The research adopts a computational linguistics methodology drawing on WordNet lexicographic architecture, morphological rule-based computational modelling, and neural lexical acquisition techniques. The study develops a morphological parsing algorithm specifically adapted to Igbo's concatenative morphology and tone-sensitive lexical rules, and applies it to a base corpus of 500,000 tokens assembled from Igbo written sources. The resulting computational lexicon covers 45,000 lemmas with morphological analyses, semantic class assignments, and argument structure information. The system is evaluated against a gold standard test set and achieves 87 percent accuracy in morphological analysis. Findings also document and propose solutions for three language-specific computational challenges: tonal disambiguation in unvocalised text, verb extension morphology complexity, and the handling of borrowing strata in the Igbo lexicon. The study advances the NLP infrastructure for Igbo and provides a replicable methodology for computational lexicon development in other Nigerian languages. Keywords: computational lexicon, Igbo, natural language processing, morphological parsing, language technology.