Dalarna University's logo and link to the university's website

du.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • chicago-author-date
  • chicago-note-bibliography
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Dazai to Digital: Assessing Translation Accuracy of “Ningen Shikkaku" Across ChatGPT-4, Donald Keene, and Mark Gibeau
Dalarna University, School of Language, Literatures and Learning.
2024 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

This study assesses the translation accuracy of ChatGPT-4 against two human translators, Donald Keene and Mark Gibeau, focusing on the first 50 sentences of Osamu Dazai's Japanese novel "Ningen Shikkaku" translated into English. In the rapidly advancing field of artificial intelligence, where AI increasingly integrates into fields such as translation traditionally occupied by humans, it examines the effectiveness and reliability of AI incapturing both the literal and figurative meaning of a literary text. A significant gap in the field is the scarcity of comparative studies between AI and human translators, and all the more so in Japanese-English translation. Most existing research on AI translation focuses on European languages or evaluates AI against other machine translation tools. The study employs a translation quality assessment framework based on how erroneous the translations are, where either one or two points are deducted for each error depending on severity to evaluate the accuracy of each translation. The identified error types are grounded on the standardized error marking system utilized by the American Translators Association, and endeavors to provide an objective measure of translation quality. The results of the study show that ChatGPT-4's translation incurred the least number of point deductions, roughly half as many as those of Gibeau and Keene. Gibeau's translation rankedsecond in accuracy, with Keene's trailing closely behind. The results also reveal that Keene's translation errors typically stemmed from altered words and phrases, while Gibeau's translation rather added, intensified, or omitted elements. ChatGPT-4's translation had fewer errors overall, except in relation to literalness. It is discussed that the utility of AI in literary translation varies depending on whether accuracy or aesthetic is most valued. Nevertheless, translators can already at present utilize AI to manage routine tasks and accelerate translation processes, enabling them to concentrate on aspects such as flow, rhythm, and readability.

Place, publisher, year, edition, pages
2024.
Keywords [en]
Japanese-English Translation, Artificial Intelligence, ChatGPT-4, Translation Quality Assessment, Literary Translation
National Category
Specific Languages
Identifiers
URN: urn:nbn:se:du-48712OAI: oai:DiVA.org:du-48712DiVA, id: diva2:1867473
Subject / course
Japanese
Available from: 2024-06-10 Created: 2024-06-10Bibliographically approved

Open Access in DiVA

fulltext(1264 kB)213 downloads
File information
File name FULLTEXT01.pdfFile size 1264 kBChecksum SHA-512
d47007d7fcbf06d955789628b983de9e7e768bad283360cfb88c7aa97ba0023630e71ef150004140f349bd67af8b2b25af85f136b4645f9425f100552fe434ec
Type fulltextMimetype application/pdf

By organisation
School of Language, Literatures and Learning
Specific Languages

Search outside of DiVA

GoogleGoogle Scholar
Total: 213 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 409 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • chicago-author-date
  • chicago-note-bibliography
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf