Limitations of machine translation of Egyptian dialect in social comedy series into English : A comparative study between AI Subtitling Tools /

Abstract

This comparative study investigates the limitations of Machine Translation (MT) in subtitling Egyptian Arabic comedy series, with a particular focus on the preservation of pragmatic meaning and cultural nuances. The selected dataset is extracted from Egyptian social comedy shows, and subtitled using three AI tools: FreeSubtitles.ai, Veed.io, and Maestra.ai. These machine-generated subtitles are evaluated against the theoretical framework of Mona Baker’s (1992) pragmatic equivalence model and an extended version of Pedersen’s (2017) FAR model., The findings indicate that MT tools frequently rely on literal translation, often failing to capture the implicit cultural nuances embedded in Egyptian colloquial Arabic. This study highlights the need to expand AI training datasets to include dialectal Arabic (DA), not just Modern Standard Arabic (MSA) to improve the quality and contextual accuracy of AI translated subtitles. Ultimately, this study affirms that human translators and posteditors still play an essential role in delivering contextually appropriate and culturally sensitive subtitles in comedic content.