Optimising ChatGPT for creativity in literary translation: A case study from English into Dutch, Chinese, Catalan and Spanish
This work addresses the problem of enhancing creativity in machine translation for literary translators and researchers, though it is incremental as it builds on existing ChatGPT configurations.
The study investigated how to optimize ChatGPT for creative literary translation across four languages, finding that a minimal prompt at temperature 1.0 yielded the best creative translations, outperforming DeepL in Spanish, Dutch, and Chinese but consistently underperforming compared to human translation.
This study examines the variability of Chat-GPT machine translation (MT) outputs across six different configurations in four languages,with a focus on creativity in a literary text. We evaluate GPT translations in different text granularity levels, temperature settings and prompting strategies with a Creativity Score formula. We found that prompting ChatGPT with a minimal instruction yields the best creative translations, with "Translate the following text into [TG] creatively" at the temperature of 1.0 outperforming other configurations and DeepL in Spanish, Dutch, and Chinese. Nonetheless, ChatGPT consistently underperforms compared to human translation (HT).