TRANSLATION TECHNIQUES AND ACCURACY OF SWEAR WORDS IN THE END OF THE F***ING WORLD: OFFICIAL PLATFORM VS AI-GENERATED SUBTITLES

This study investigates the translation techniques and accuracy of swear words in the British dark comedy series The End of the F***ing World, comparing official Netflix subtitles with AI-generated translations by ChatGPT. This study aims to classify the pragmatic functions of swear words, analyze the translation techniques applied by both translators, and determine which version is more accurate. A total of 83 swear word data were analyzed by using theory from Pinker's (2007) pragmatic function, Molina and Albir's (2002) translation techniques, and Nababan et al.'s (2012) accuracy assessment instrument. This study adopts a descriptive qualitative approach, using document analysis to collect the data and Focus Group Discussion involving two raters to evaluate translation accuracy. The findings reveal five pragmatic functions, with cathartic being the most dominant (30.11%). Netflix predominantly applied Established Equivalent (29 data, 34.94%) and Reduction (26 data, 31.33%), while ChatGPT relied most heavily on Established Equivalent (36 data, 43.37%). ChatGPT achieved a higher cumulative accuracy score (2.66) compared to Netflix (2.41), suggesting that AI-generated translation demonstrates competitive competence in handling pragmatically sensitive language in subtitle translation.