Despite the growing accuracy of machine translation (MT) in general-purpose tasks, its performance on culturally embedded, humor-driven text remains largely unexamined, particularly for the English-Indonesian language pair. This study examines the acceptability of humor translation produced by three MT tools, Google Translate, DeepL, and ChatGPT, using 20 humorous utterances drawn from Paul Snyder's stand-up comedy special "I Left the House for This" as the source data. A qualitative-dominant mixed-methods design was employed: the utterances were identified using Ruch, Attardo, and Raskin's General Theory of Verbal Humor, translated by the three tools, and evaluated using an adapted acceptability rubric, followed by a focus group discussion to consolidate the scores. On a 1-3 acceptability scale, Google Translate scored 1.42, DeepL 1.58 (both Unacceptable, though DeepL approached the Less Acceptable range) and ChatGPT 2.35, narrowly crossing into the Acceptable category. The main factors lowering acceptability were the use of formal written register in an informal spoken context, untransferred cultural references, and inconsistent pronoun choice. Despite occasional inconsistencies, ChatGPT produced the most natural and colloquial translations among the three tools. This study concludes that ChatGPT is currently the most suitable MT tool for translating humor from English stand-up comedy into Indonesian, while Google Translate and DeepL still require substantial human post-editing for this type of content. These findings offer novel empirical insight into MT-humor acceptability for an understudied language pair and genre.
Copyrights © 2026