Comparing DeepL and Google Translate in Translating Arabic Expressive Speech Acts in Masameer: al-Film

Abstract

In cross-language communication, machine translation is an important tool that supports the understanding of meaning between speakers from different cultural backgrounds. This study compares the translation quality of expressive speech acts in the Masameer animation using two automatic translation engines: DeepL and Google Translate. Expressive speech acts rely heavily on pragmatic context and emotional expression, which presents a challenge due to the presence of implicit meanings, cultural values, and social nuances that are often difficult to translate automatically. This study employs a qualitative-comparative approach by analyzing 19 representative data points. The translations were assessed by two experts using Nababan’s model and validated through method triangulation, including interviews, content analysis, and theory-based discussions. The results show that Google Translate is superior in terms of accuracy (2.00 and 2.68) and acceptability (2.63), compared to DeepL, which scored 1.84 and 2.68 in accuracy and 2.36 in acceptability. However, both engines remain weak in translating metaphorical expressions and culturally specific contexts. This research fills a gap in comparative studies on Arabic-Indonesian machine translation and contributes to the development of translation technology and Arabic language learning, particularly in the areas of pragmatics and intercultural communication.