The effects of linguistic proximity on dependency distance and dependency direction of translated English from 35 languages

Abstract

As the “third code”, translated language differs from native language in terms of syntactic features. Translations from different source languages can exhibit syntactic variation due to the source language shining-through effect. This study investigates the impact of linguistic proximity on dependency distance and dependency direction of translated English from various source languages. A corpus of native English and translated English from 35 languages was constructed and annotated for analysis. The findings are: (1) Compared to native English, translated English from 17 languages exhibits shorter mean dependency distances (MDD), while translations from 18 languages exhibit longer MDD. Translations from 13 languages are more head-final, while translations from 22 languages are more head-initial. (2) The dependency distance of translated English correlates positively with the linguistic proximity between source language and English, while the clustering of dependency directions is associated with the linguistic proximity of source languages within language families. (3) For languages spoken in countries where English is an official language alongside the native language, the dependency distance and dependency direction of translated English are influenced more by English than by the native language. These differences arise because dependency distance reflects the cognitive demand involved in processing dependency relations, whereas dependency direction functions as a typological marker distinguishing languages.