Open Source Kiswahili Spell Checker (SW-TZ)

Abstract

Different English software products are localized into many native languages spoken around the world, the most popular software products localized so far are word processing software and web browsers. The effort has begun to localize these software products in Kiswahili language which is widely spoken in Tanzania, East and Central Africa. Kiswahili
is an official language of Tanzania, and is a national language in Uganda and Kenya. To make these software products useful to Kiswahili speaking community particularly in Tanzania a new Kiswahili spell checker has been developed and, it uses most of the Kiswahili words spoken in Tanzania and its locale identity is sw-TZ for OpenOffice.org and Mozilla Firefox while Kiswahili spoken in Kenya has a locale identity sw-KE which is common for Microsoft office products. It was developed due to the shortcomings of first ever Kiswahili spell checker developed by Open Kiswahili Localization Project
(KILINUX). A spell-checking tool called MySpell has been used in developing a new spell checker. The dictionary of a new spell checker consists of 4894 stems and its affixes were created based on Kiswahili noun classification system and by selecting different groups of verb stems that follow the same trends in generating other Kiswahili valid words. This
paper addresses the shortcomings of the existing spell checker and the approach carried out to develop a new spell checker.