본문 바로가기
  • Home

Stylometric Comparative Analysis of Style in Human vs. Machine Literary Translations

  • The Journal of Translation Studies
  • Abbr : JTS
  • 2019, 20(2), pp.111-130
  • DOI : 10.15749/jts.2019.20.2.005
  • Publisher : The Korean Association for Translation Studies
  • Research Area : Humanities > Interpretation and Translation Studies
  • Received : April 30, 2019
  • Accepted : May 28, 2019
  • Published : June 30, 2019

Chang-Soo Lee 1

1한국외국어대학교

Accredited

ABSTRACT

The current research is designed as a pilot study under a project aimed at investigating differences in style between human and machine translators in Korean-English literary translation. The research seeks to address three questions from a stylometric or computational linguistic perspective. (1) Do machine translators have their own unique styles? (2) Are they clearly distinguishable from human translators in style? (3) Are they progressing over time in such a direction that they are becoming more like human translators? These questions are tackled by analyzing the English translations by two human translators and three machine translators of a single Korean short novel. The translations by the machine translators were collected at two points separated by a span of one year, providing us with a total of eight translated texts. Burrows’ delta scores, a popular measure of textual distance, were extracted from the texts and analyzed by two unsupervised statistical methods – multidimensional scaling and hierarchical cluster analysis. The machine translators displayed interdependent styles clearly distanced from one another, while they as a whole were distinctly separated from the human translators. The machine translators showed no evidence of having narrowed the distances between them and the human translators over one year.

Citation status

* References for papers published after 2022 are currently being built.

This paper was written with support from the National Research Foundation of Korea.