Indian Journal of Science and Technology

Abstract

In recent times the rate of colloquial and informal Punjabi text has increased dramatically. People are using the web as medium for expressing their ideas, thoughts and emotions, usually in form of colloquial articles and blogs. Punjabi has a wealth of Natural Language Processing (NLP) resources and tools. In comparison, resources for Punjabi dialects, the informal spoken varieties of Punjabi, are still lacking. Punjabi dialects also present many challenges for conversion as oneto-many word mapping, incorrect Punjabi spellings and sub-dialects. In this paper, we present a conversion system for Punjabi text to its equivalent Punjabi dialect text which could be extended easily to be applied to other dialects of Punjabi. The proposed system is developed using the rule based approach that mainly relies on morphological analysis, conversion rules and bilingual dictionaries. Firstly, various Punjabi words, that need conversion, are identified. Then these identified words are replaced using a rule-based component that consists of conversion rules and bilingual dictionaries. A manual error analysis of system’s outcome shows that it produces correct conversion over 95% for Malwai dialect and over 94% for Doabi dialect. The development of these rules is not an easy task as dialect form originates from pronunciation by native speakers.
Keywords: Bilingual Dictionary, Dialect Conversion System, Machine Translation, Punjabi Dialects