A better or worse communicator? Comparing human and machine translation in source language shining through across registers

• Published in: Lingua Vol. 312; p. 103834
• Main Authors: Li, Jia, Hu, Xianyao
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01-12-2024
• Subjects: Human–machine translation distinction; Machine translationese; Register variation; Shining through; Translation universals; Human–machine translation distinction; Translation universals; Shining through; Machine translationese; Register variation
• ISSN: 0024-3841
• DOI: 10.1016/j.lingua.2024.103834

•A semi-supervised multivariate approach was applied.•Machine translation exhibits a stronger tendency towards shining-through.•The shining-through effect is largely diminished in Chinese-to-English translation.•Human translators are more likely to adhere to the target norms.•Human translation outperforms machine translation in contextual understanding.
This study provides a novel perspective on the distinction between human and machine translation by comparing the shining-through effect in English translations from Chinese across four different registers based on four comparable balanced corpora. By applying two multivariate techniques to multiple lexical, cohesive and syntactic features, this study identifies the following. (i) The shining-through effect is observable in human- and machine-translated English from Chinese; however, it is relatively limited. (ii) Machine translation exhibits a more distinct and persistent tendency towards shining through across different registers. (iii) The shining-through effect is register-dependent, with a stronger presence in general and academic texts than in journalistic and fictional translated texts. These findings confirm shining through as a potential translation universal in both human and machine translated texts. Furthermore, the results suggest that human translators are more sensitive to the social norms in different registers, further supporting the notion that human translators are more alert to contextual and communicative constraints in translation. The observed distinction offers valuable insights for translation educators in restructuring the curriculum for the human–machine collaborative translation model by considering translation norms, register constraints and interpersonal and intercultural factors.