LLMs demonstrate solid source-text comprehension but produce mostly literal or inappropriate literary translations, with only Mistral-Large approaching human creativity scores on a new UCP-based metric across 23 models and 11 books.
- low→routine, literal, unsurprising - medium → some interpretive or stylistic choice - high→clearly inventive, or unexpected
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Beyond Reproduction: A Paired-Task Framework for Assessing LLM Comprehension and Creativity in Literary Translation
LLMs demonstrate solid source-text comprehension but produce mostly literal or inappropriate literary translations, with only Mistral-Large approaching human creativity scores on a new UCP-based metric across 23 models and 11 books.