Korean Language Generation in an Interlingua-Based Speech Translation System

摘要

The Speech Systems Technology Group at Lincoln Laboratory has been developing anautomatic speech-to-speech translation system for English to Korean translation. For the machine translation module, an interlingual system has been adopted. This system analyzes the source language text and represents the results of the analysis in a semantic frame, which is a representation of the meaning of the input text from which the text in the target language is generated. GENESIS, a language generation system developed at the Spoken Language Systems Group at the Laboratory for Computer Science at MIT, has been utilized for the language generation component. GENESIS had previously been used in limited domains for European languages and for Japanese. The work reported here shows that GENESIS is capable of generating a wide range of Korean sentences. It explores the degree to which GENESIS can handle Korean as a target language in a machine translation system and describes extensions to GENESIS that are needed to produce Korean output.