법률 특화 기계번역을 위한 중국 법률 텍스트의 어휘적 특징 분석 ― 중화인민공화국 민법전을 예로

This paper analyzes the basic statistics and lexical characteristics of the articles and sentences in the Civil Code of the People's Republic of China, as a first step toward the ultimate goal of improving the performance of Chinese legal machine translation through specialized legal machine translation. The statistical analysis shows that, regardless of individual articles or sentences, the clauses in the Civil Code are segmented in different ways due to the pause function of punctuation, which may affect the legal interpretation and translation results in the machine translation process. In terms of lexical frequency, the analysis of the extracted high-frequency vocabulary features shows that the Civil Code has a high frequency of the use of particles (的), modal verbs (应当, 可以, 不得), and ambiguous vocabulary (或者, 等, 其他). In addition, by using the keyword and list functions of the corpus analysis tool AntConc, we extracted candidate legal terms and confirmed their high frequency of use in the overall legal vocabulary, demonstrating the need for parallel data construction of legal terms for specialized legal machine translation. Most research on domain-specific machine translation has focused on optimizing the internal structure of systems from an engineering perspective. However, the fundamental work of analyzing the linguistic features of the text of a specific domain and applying them to machine translation has been challenging. Therefore, this study finds significance in taking a linguistic approach to improving machine translation performance for the specific domain of law.