术语与机器翻译——实验结果分析与术语数据库的构建

Terminology and Machine Translation——Experimental Results and Construction of Terminological Databank

术语管理是机器翻译的关键技术之一。通过机译术语与人工翻译术语的实验比较,发现机译效果尚有待于提高。基于平行语料库利用统计原理自动提取双语术语等价词对,有利于改进机器翻译的质量,但依据统计概率或面向人工的翻译词典来测评等值程度,不能全面反映历史典籍文本中特定语境下的术语意义及功能,从而影响术语等价词对的提取范围。基于翻译家英译历史典籍文本的语料提取术语,并从历史语境、语域、文体等译学特征入手进行人工验证,构建描述性术语数据库,有利于提高机器翻译的准确性和适切性。

Terminological management is one of the key ingredients of machine translation. The experimental translation results of terms reveal substantial research is necessary to improve the machine translation quality. The automatic extraction of bilingual term equivalents from parallel corpora is a significant technique,yet evaluating the equivalency between the statistically extracted bilingual terms against the translation ratio or translator-oriented dictionaries can not fully account for the usage and function of a given term in a special historical text. Building a descriptive terminological data- bank out of the versions of historical texts by well-established translators, and assessing the term equivalency by taking into account the influence of historical context, register and style will be essential to improve the accuracy and appropriateness of machine translation.