1. Introduction

This article introduces the translation language as the "third language code" between the source language and the target language, and explores the importance of its regular features in understanding the essence of translation. Identified the bottlenecks currently faced by translation language research, such as the lack of comprehensive and in-depth feature studies, and the difficulty in constructing comparative corpora. This study aims to provide a new analytical path for translation language research by constructing bilingual parallel and comparable corpora, using dependency grammar and quantitative analysis methods, examining its new features and language and cognitive motivations, in order to promote the scientific and theoretical construction of translation research.

2. Research background

Firstly, the development process of research on translation language features was reviewed, from Frawley's definition of translation language as the "third language code" in the 1980s to Baker's proposal of the concept of translation commonality in the 1990s, which became a research hotspot. Translation commonality refers to the language features commonly present in translated texts, such as explicitation, simplification, and standardization, which have been validated in Indo European languages. Since the 21st century, domestic scholars have also begun to pay attention to translation commonalities, verifying them by analyzing language features such as personal pronouns and conjunctions in Chinese texts. However, existing research is controversial, mainly due to the lack of composite corpora and quantitative analysis methods for overall features.

3. Research Design

This chapter elaborates on the research design of the new hypothesis of "compromise hypothesis" regarding translation language features. The research focuses on the overall syntactic features of the translated language, especially the average dependency distance. By comparing translated English with Chinese original text, original English, and translated Chinese with English original text and original Chinese, the regularity of translated language under the influence of source and target languages is explored. The study constructed a corpus that includes translated Chinese, original Chinese, translated English, and original English. The Government Work Report and its English translation, as well as the State of the Union Address and its Chinese translation, were selected as research materials. The research steps include corpus cleaning, word segmentation, sentence alignment, dependency syntax analysis and part of speech tagging, as well as manual inspection and proofreading. The calculation method and formula for dependency distance are introduced, which are used to calculate the average dependency distance of sentences or tree libraries, and are implemented on a large scale through self written programs.

4. Results and Discussion

By comparing and analyzing the MDD values of the Government Work Report and its English translation, the State of the Union Address and its Chinese translation, it was verified that the MDD of the translated language is between the source language and the target language, that is, the MDD of translated English is longer than that of original English but shorter than that of original Chinese, and the MDD of translated Chinese is shorter than that of original Chinese but longer than that of original English. Research has found that this compromise feature in translation language may stem from the mutual restraint between the source and target language systems during the translation process, as well as the translator's coordination between the heterogeneity of the source language and the acceptability of the target language. In addition, the MDD of translated languages is consistent with the phenomenon of "minimizing dependency distance" in human natural language, reflecting a common trend in human language. The compromise feature of translation language can also be explained by cognitive load, as different translation directions bring different cognitive loads to the source language system, which affects the translator's text output. The "compromise hypothesis" proposed in this study helps us better understand the phenomenon and essence of translation, and provides a new conceptual tool for corpus translation studies.

5. Summary

This study summarized the multiple composite comparative analysis of syntactic features in translated languages through the framework of dependency grammar theory and the econometric features of average dependency distance. The study found that the average dependency distance of translated languages in different translation directions is between the source language and the target language, supporting the phenomenon of minimizing dependency distance and revealing the complex psychological process of translators coordinating and balancing between the two language systems. The research value includes: 1) designing a new corpus that includes bidirectional parallel and comparable corpora; 2) Introducing dependency grammar analysis framework and econometric methods to examine translation language; 3) Propose the 'compromise hypothesis' to promote the theoretical construction of the translation commonality hypothesis. The 'compromise hypothesis' brings a new perspective to translation language research, explains existing phenomena, and provides a new theoretical basis for understanding language processing, cognitive load, and translation strategies. Future research can explore the compromise characteristics of translated languages in different language families and deepen our understanding of the characteristics of translated languages. However, the research has limitations such as limited corpus, and the 'compromise hypothesis' needs further verification.