Isolated-word Error Correction for Partially Phonemic Languages using Phonetic Cues

Partially phonemic languages use writing systems which are in between strictly phonemic and non-phonemic orthography. Therefore, phonetic errors are very frequent in such languages. This paper introduces an approach for development of spellcheckers for partially phonemic languages that use grapheme-to-phoneme mapping for isolated-word error correction. Since, a complete and accurate grapheme-to-phoneme system is overkill for a spellchecker, the framework can deal with incomplete phonological information through the use of metaphonemes. The paper also discusses the implementation of a Bengali spellchecker based on this approach and some other issues specific to the Bengali spell-checking. The framework described here is generic in nature and can be used for any partially phonemic languages by incorporating the language specific parts like phonological rules, the keyboard layout and ranking strategies. This approach is very useful for Indian languages as most of them are partially phonemic in nature.