The research and development of machine translation in China

MT Summit VII Sept. 1999 Abstract This survey of the situation regarding the R&D of machine translation in China concentrates on how the design and application of MT systems could be on the shadow of the characteristics of Chinese Language. After a brief historical overview and the investigation on R&D environment, some technical features are described, including commercialization, Chinese-English language pair and a multi-level transfer in English-Chinese MT systems. 1 A brief overview on the MT development history Natural language information processing (NLP) in China started from the research on machine translation (MT). As early as in 1956, MT was identified as one of the research subjects in China’s first Science Development Plan. The project was entitled “Machine translation, rules of natural language translation and the mathematics on natural languages”. In 1957 the Chinese Academy of Sciences (CAS) engaged in a study on Russian-Chinese machine translation system and, in 1959, CAS conducted a series of experiments successfully — nine sentences with different pattern and relatively complicated structures were translated. The research then proceeded to English-Chinese language pair. Until the mid of 1960’s, there had been some six research institutions and universities involved in MT research (Liu Y., 1984). After more than 10 years of standstill, China’s MT recovered in mid 1970’s. English-Chinese was the first language pair for the new studies, which were led by the Institute of Linguistics of the Chinese Academy of Social Sciences. Two experimental systems were developed later as the result of the studies. The first was a translation system for titles of technical documentation, followed by the full text translation system for S&T documentation (Liu Z., 1981). On this basis, a number of projects were launched and carried out, focusing on other language pairs, including French-Chinese, Japanese-Chinese, German-Chinese and Russian-Chinese. An experiment for translating Chinese into a cluster of languages including English, French, Japanese, Russian and German was also carried out, using the ARIANE of GETA in Grenoble, France (Feng, 1984). The first commercial English-Chinese MT system, named TranStar was put in the market in 1988 (Dong, etc. 1986). More than 10 MT products became available in market since then. The English-Chinese language pair is targeted in these new systems, and to a less extent to other pairs such as Japanese-Chinese and Chinese-English. Web-oriented translation service has been tried by some producers. During the same period, there was also a fast development in machine translation in Taiwan. Among the existing products and services in market, Behavior Design Co.’s translation service has been regarded as a successful case (Chen, etc. 1991). It should also be mentioned that in Hong Kong, MT systems are also developed for Chinese dialect and Mandarin (Putonghua) (Zhang, 1998). 2 Research and development (R&D) environment In China, the initial support for MT research came from the Government. Funds was provided through the 1950’s National Science Development Plan and the following National Five-Year Economic Development Plans, as well as from National Natural Science Foundation and the 863 High Tech Project started in late …

Content truncated to 500 words