1 Introduction

In the past 50 years, corpora have played an important role in the field of language research, especially monolingual corpora. However, the construction of bilingual parallel corpora is relatively lagging behind, despite its enormous theoretical and practical value. Most of the parallel corpora established in China are mainly translated from foreign to Chinese texts, and the parallel corpus of academic texts translated from Chinese to foreign is still blank. Since the 21st century, China has actively promoted academic translation activities, but there is a relative lack of research on the translation of Chinese academic works. Parallel corpora of academic literature have unique applications and academic value, such as improving translation efficiency and analyzing language features.

2 Corpus collection

Introduced the corpus collection process of the parallel corpus of Chinese English humanities and social sciences literature. Through methods such as searching the library catalog of the United States Library of Congress, more than 40 literary, historical, and philosophical documents and their translations were identified as the collection objects, with a total word count of approximately 15 million words. The included literature includes monographs and collections of papers from the Republic of China period, after the reform and opening up to the end of the 20th century, and after 2000, some of which were translated by foreign scholars and some were funded by the China Foundation for Translation. The corpus adopts a full-text inclusion method, and the proportion of Chinese and English corpora gradually increases over time. Literature from the 30 years before the founding of the People's Republic of China is temporarily not included.

3 Corpus processing

Elaborated in detail the corpus processing steps in the construction process of the parallel corpus of Chinese English humanities and social sciences literature. Firstly, the literature is digitized through scanning, text recognition, and manual proofreading. Chinese literature is subjected to text recognition using API interfaces from Baidu and Tencent, while English literature is processed using Tesseract OCR software. Secondly, perform text cleaning by removing non body parts and processing hyphens in English. Next, we used the vecalign tool and Google's LaBSE model for sentence alignment, and also developed sentence code. In addition, English corpus is annotated with word prototypes, parts of speech, and syntax, while Chinese corpus is not currently annotated with word segmentation and parts of speech. Finally, code the meta information of the corpus, including relevant information of the original work and the translated work. At the same time, the difficulties in corpus construction were pointed out, such as text recognition errors, heavy workload in text cleaning, and the need for manual proofreading after automatic sentence alignment.

4. Construction of corpus retrieval platform

Introduced the construction of a retrieval platform for parallel corpora of Chinese English humanities and social sciences literature. The search platform has multiple functions, including ordinary word search, using wildcard characters, part of speech search, word prototype search, and regular expression support. The interface provides "search" and "index" functions, which can count word frequency, extract word cluster tables, and output sentences and their translations. The platform also supports dependency relationship retrieval, allowing the retrieval of dependency words, jurisdictional words, and their dependency relationships in sentences. The retrieval function is implemented through SQL, which parses the retrieval expression into SQL query language and converts the results into a user-friendly format. The research case explores the translation differences of paraphrasing verbs in academic translation, and finds that different disciplinary fields and cultural backgrounds affect translation strategies. Translators need to consider language, cultural differences, and disciplinary knowledge construction methods to accurately reproduce the original position.

Application of Parallel Corpus of 5 Chinese English Academic Texts

Wang Kefei (2012) emphasized the importance of parallel corpora in foreign language teaching, translation research, and dictionary compilation. This parallel corpus of Chinese English academic texts has practical value in spreading Chinese culture, improving the quality of academic translation, and providing data support for the study of language features in academic translation. The corpus includes works and translations in the fields of Chinese philosophy, history, sociology, literature, aesthetics, etc., which helps translators to query the translation status of cultural loaded words and make the best translation choices. The parallel corpus of academic texts can meet the professional needs of academic translation. Comparing the academic subject matter corpus of this corpus with the Yiyan General English Chinese Parallel Corpus, it was found that there are differences in the ratio of Chinese to English form symbols, indicating that the language features of Chinese academic original works and translated works may be different. At present, there is still a lack of systematic investigation on the language characteristics of academic translation between English and Chinese. The construction of this corpus will help answer relevant questions.

6 Conclusion

The construction of a parallel corpus of Chinese English humanities and social sciences literature has significant value, providing data support for translation research, practice, language comparison, and foreign language teaching. Multiple parallel corpora have been established in China, but the construction of the Chinese translation foreign language terminology corpus still needs to be strengthened. This study collected over 40 academic works in Chinese and their English translations. Through text recognition, proofreading, text cleaning, sentence alignment, and corpus coding, a parallel corpus of approximately 15 million Chinese English academic texts was established. A supporting retrieval platform was developed, which supports collocation retrieval, wildcard retrieval, and dependency grammar retrieval. This corpus will provide support for academic translation research and practice. Large corpora usually collect literary materials by extracting fragments or chapters to improve representativeness, but this corpus did not adopt this method because the number of English translations and publications of Chinese academic literature was limited before 2000. When expanding the corpus since the 21st century in the future, consideration will be given to using excerpts or chapters to collect the corpus of works.