Term Translation Pair Alignment Based On A Bilingual Parallel Corpus Of Chinese Historical Classics

Translating Chinese historical classics into English is one of the most effective ways to promote the reception of traditional Chinese culture in the world. The major challenges that lie ahead include finding appropriate strategies to handle the numerous historical terms and to strike a balance between transferring the original culture adequately and maintaining an acceptable degree of readability in the translations.This paper presents a systematic survey about the English translations and studies of such monumental works as Shi Ji, Han Shu, Zuo Zhuan, Zhanguo Ce, etc., a basic but essential step for empirical translation studies on Chinese historical classics and bilingual parallel corpus construction in this field. It is also significant to classical Chinese-oriented natural language processing, translation studies on Chinese classics and ancient Chinese history research.Two authoritative translations of Shi Ji in America, Records of the Grand Historian by Burton Watson and The Grand Scribe's Records by William H. Nienhauser, Jr., have been employed as case studies to explore the discrepancies on the textual, linguistic and stylistic levels as influenced by the manipulative factors behind a translation project within the framework of the manipulative school.Burton Watson's translation, Records of the Grand Historian published by Columbia University Press in 1961, was well-known for its outstanding literary significance and achieved such a canonical status in world literary translation that it was accepted into the UNESCO Representative Works:Chinese Translation Series. It has remained popular ever since both as a reliable reference book about Han China studies and as an assigned textbook for undergraduate and graduate students majoring in China Studies at major universities of the English-speaking countries. About thirty years later, William H. Nienhauser, Jr. initiated a literally rendered, heavily annotated full English translation of Shi Ji, The Grand Scribe's Records for China Studies specialists, following the Sinological research traditions and translation principles.6 volumes of such high quality scholarly version have been published.This paper takes Nienhauser's version as the control data and examines the translation patterns of Watson's version and the driving force behind his translation strategies. The results indicate that the translation of classics is also subject to the constraints enforced by a specific social, historical and cultural context, the nature and purpose of the translation project, the translation norms and poetic systems in the target language and culture, the translator's ideologies, etc. The analysis of the evolving process of Watson's translation critierion verifies the descriptive translation studies hypothesis that the acquisition and internalization of linguistic norms by a translator is materialized in the process of his/her socialization. It is proposed that translation of Chinese historical classics is a historical construct. The data analysis of the English translations of Shi Ji demonstrates that the nature of the translations is a significant factor for the data source differences in a parallel corpus and the discrepancies between different translations would have a strong impact on the knowledge acquisition results of natural language processing. An in-depth examination of the English translations of Chinese historical terms reveals that Chinese historical terms have multiple English equivalents. The following aspects have been identified as contributive factors: dynamics of term designations, synonymy of terms, general-noun terms being used as addressing terms, flexibility of terms'part-of-speech, synonymy of term translations, lack of equivalent term concepts in English, translator's cognitive and translation style differences, etc. These findings exert a great challenge for parallel corpus-based automatic term translation pair alignment.The terms to be tackled for alignment in this paper cover official titles, posthumous titles, institutions, etc. in Shi Ji. By integrating the English translation patterns of these terms, such as partial transliteration,2/3/4/5-word phrase translations frequently connected with conjuntions like "of, "of the", "on", "in", replacement of titles with title-holders'names, insertion of title-holders'names, translation ambiguity generated by the intervening words of source terms, this paper proposes co-occurrence frequency and head-word extension approach, MI-based hybrid approach for term translation pair alignment. The experiment results show the first approach is effective in aligning partially transliterated term pairs while the latter performs well in aligning both the partially transliterated term pairs and the general noun term translation pairs.In view of the multiple authoritative English translations of Chinese canonized historical texts, this paper stresses that each version should be aligned with the source text independently so that different translation pairs might have a chance to be aligned automatically. Further validation and classification of the candidate translation pairs in terms of adequacy and acceptability might benefit from philological studies and the corresponding data generated from BNC and COCA. The frequency count data from these corpora might also help to define the translation styles of different term translations, such as everyday English wording or archaic and rare terms. All this is expected to lay a basic foundation for constructing a Chinese historical term translation pair retrieving system and a knowledge base as well as compiling a bilingual machine-readable or human-translator-oriented bilingual dictionary of Chinese historical terms.The research also shows aligning historical term translation pairs based on a parallel corpus of Chinese historical classics and their English translation is still a challenging task for natural language processing. Investigations about the potential influences of the base texts on term translation would also help to carry the research forward.