Developing an Open-source Spell Checker for Dholuo using Hunspell language tool

Abstract

Computational morphology is an important first step in most natural language processing tasks such as spellchecking and machine translation. Computational morphology involves the study of internal structure of words i.e. how verbs and nouns are generated from the root words called the stems. Tnthis work, we carry out a computational analysis of Dholuo morphology. Major morphological components of the prevalent Dholuo words are explored. These include tense, verbs and nouns derivation In this work, we use the Hunspell tool to develop an open source Spellchecker in Dholuo. The Hunspell tool is used to create two files namely the affix and the dictionary files. The affix file enables the creation of all the rules involved in deriving the nouns and the verbs from the root words. All the root words (stems) plus the appended rules are stored in the dictionary file. The developed spellchecker, is the first spellchecker for Dholuo. The system developed achieves an acceptable representation of Dholuo morphology. It can correctly classify Dholuo words with an accuracy rate of 0.814, precision rate of 0.917, recall rate of 0.800 and coverage rate of 0.820.