Evaluating ChatGPT's Translation of Harry Potter: A Qualitative Study of Translation Techniques, Accuracy, and Acceptability

Abstract

This research investigates the translation techniques employed by ChatGPT in translating J.K. Rowling's Harry Potter and the Sorcerer's Stone, focusing on the accuracy and acceptability of the translations produced. Utilizing a qualitative approach, the study analyzes data from the original English text, the Indonesian translation generated by ChatGPT, and feedback from five evaluators based on predetermined criteria. The data analysis process involved comparing the source and target texts to categorize translation strategies and calculating average scores from raters. The findings reveal that ChatGPT effectively utilized various translation strategies, with Translation by Paraphrase Using a Related Word being the most used (62%). Other strategies included Translation by a More Neutral or Less Expressive Word (24%) and Translation by a More General Word (8%). The overall accuracy score was relatively high; however, challenges arose in maintaining acceptability, particularly regarding expressive tones and cultural nuances. This study contributes significantly to understanding AI's role in literary translation, showcasing both its potential and limitations in handling complex literary elements. By highlighting these aspects, it advances the discourse on AI's evolving capabilities in creative fields.