ABSTRACT
This chapter considers machine translation as the representative natural language processing and emphasizes the necessity of natural language understanding as the genuine natural language processing. Three systems, Google Translate, Systran and Excite, were investigated with English as the source language and Japanese as the target language. From the 1950s, when the earliest machine translation systems emerged, to the 1980s, there were two types of approaches to cross-language operation, namely, interlingual and transfer-based approaches. The interlingual machine translation approach has been made operational only in relatively well-defined domains, such as device manuals. The transfer-based machine translation approach has been made more operational, where meaning representation depends partially on the language pair involved in the translation. If the systems are provided with more elaborated natural language generation processes for forward translation, backward translation can be improved to a certain extent.
