English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Large language models (LLMs) are competitive with the state of the art on awide range of sentence-level translation datasets. However, their ability totranslate paragraphs and documents remains unexplored because evaluation inthese settings is costly and difficult. We show through a rigorous humanevaluation that asking the Gpt-3.5 (text-davinci-003) LLM to translate anentire literary paragraph (e.g., from a novel) at once results inhigher-quality translations than standard sentence-by-sentence translationacross 18 linguistically-diverse language pairs (e.g., translating into and outof Japanese, Polish, and English). Our evaluation, which took approximately 350hours of effort for annotation and analysis, is conducted by hiring translatorsfluent in both the source and target language and asking them to provide bothspan-level error annotations as well as preference judgments of which system'stranslations are better. We observe that discourse-level LLM translators commitfewer mistranslations, grammar errors, and stylistic inconsistencies thansentence-level approaches. With that said, critical errors still abound,including occasional content omissions, and a human translator's interventionremains necessary to ensure that the author's voice remains intact. We publiclyrelease our dataset and error annotations to spur future research on evaluationof document-level literary translation.

Large language models effectively leverage document-level context for literary translation, but critical errors persist