Translation Quality of Artificial Intelligence and Machine Translation Vs. Human Translation Utilizing MTPE Skills (An Empirical Study on Allusion Translation)

Abstract

Allusion is one of the culture-bound expressions that need careful consideration while translating. Machine translation (MT) and human translators (HTs) encounter difficulties in dealing with them. This study compares Translation Quality (TQ) of MT and Artificial Intelligence (AI) to HTs utilizing MTPE focusing on identifying the MTPE skills to keep HT in favor of MT and AI.

A quantitative and qualitative mixed method was adopted using a test of 30-item in-context English-to-Arabic allusions translated by Google Translate and ChatGPT and then given to a random sample of 40 HTs. The TQ of AI, MT and HT target texts were assessed following O'Brien's (2012) model. The participants wrote reports on MTPE skills and were involved in a focus group discussion to determine the MTPE skills used. One-Sample t-Test, One-Way ANOVA and POST HOC Test were used. Results show HTs utilizing MTPE are of Moderate Quality (60%), and MT and AI-based translations are of Low Quality (44.44% & 42.22%). HTs employ some MTPE skills and strategies that resulted in statistically significant differences between HTs of allusions compared to MT and AI in favor of HTs. The study recommends enhancing MTPE skills among translation students and implementing training for further developing translators.