ABSTRACT

Developing a machine translation system which can correctly translate any sentence from one language to another is a formidable and challenging task. This suggests that such a system should be implemented incrementally in order to cope with new sentences that have previously not been successfully translated. This paper proposes a new framework to develop Thai to English machine translation incrementally. It is based on the Interlingua approach, which uses syntactic representation of Lexical Functional Grammar (LFG) or an LFG tree structure as Interlingua. The system development is an iterative process that adds more complex grammar rules into the system. Each iteration of the process consists of four phases: syntactic analysis, building Thai LFG tree, mapping a Thai LFG-tree into an English c-structure tree and, lastly, English sentence generation. According to experiment results, this framework has proven to be effective in building and designing a machine translation system.