Abstract :
[en] The translation of phrasemes, fixed expressions which are inherently linked to their source language and culture, have long been a major stumbling rock for translators (Sadeghpour, 2012, p. 102) and an even bigger worrying matter for translation teachers, whom, without suitable tools and methodologies, tend to feel as if they are chasing rainbows.
Therefore, one might wonder if technological innovations as AI and CAT tools may constitute one way out of this dilemma, especially in the case of chengyu, which have been considered as the most representative type of Chinese idiomatic expressions (Conti, 2019; 2020, p. 412). Indeed, chengyu, which are fixed non-compositional phrases that usually convey a complex message in a relatively short form (usually four sinograms) and that can take any grammatical function, represent an interesting research topic, so far almost unexplored in the teaching studies field (Conti, 2020, p. 412 & Guo, 2017, p. 101).
In this paper, we will focus on how and if CAT tools – such as Google Translate, DeepL, Bing Microsoft Translator, 百度翻译 Baidu Fanyi, 云译 Cloud Translation, – AI softwares (ChatGPT) and parallel corpora such as the OPUS2 corpus available on SketchEngine, can help translators and translation trainees produce better translations of these fixed expressions, consequently contributing to an improvement of translation teaching techniques.
In this case study, we will use a parallel corpus (ZH-FR) composed of contemporary Chinese novels written by 高行健 Gao Xingjian, 莫言 Mo Yan, 苏童 Su Tong and 余华 Yu Hua (compiled by Kevin Henry in 2016 as part of his PhD thesis dissertation) and their official French translations. Previously extracted chengyu will then be randomly selected and their translations by the aforementioned tools will be analysed and compared to the published French versions translated by human professional translators. This multidisciplinary lapproach, which needs to be refined, will contribute to a better understanding of AI and CAT tools applications to translation training.
