A Statistical Approach to English and Sinhala Translation

Abstract

Statistical Machine Translation is not a new term for the natural language process- ing research community. Being a machine translation technique belonging to data driven approach, it outperforms the rule based approach because of its sole reliance on data. SMT is a well established area,ranging from simple model generation to optimization of model quality by parameter tuning. However most researches have been conducted on European languages, extremely few have taken place on Asian Languages. This research tries to overcome this scarcity by building an English- Sinhala Translator using Statistical Machine Translation. Special attention is paid for the impact of parameter tuning on Sinhala Language. This research is an at- tempt to justify to which extent Statistical Machine Translation would perform in translations between English and Sinhala language , and , more generally, between structurally dissimilar languages like Asian and European languages.