Abstract

This article investigates subtitles generated by machine translation (MT) technology on YouTube. The selected videos for the analysis have been taken from different genres (interview, speech, documentary, poem, and vlog). Such diversity enables the assessment of linguistic complexity and cultural sensitivity. In detail, the article discusses the “mechanisms” MT systems use to overcome cultural and linguistic constraints, which may result in either correct outputs (“True”) or incorrect ones (“False”), depending on how the system handles linguistic and cultural complexities. This article qualitatively identifies the translation mechanisms used by MT when auto-translating the auto-generated subtitles on YouTube from English to Arabic. The analysis shows that MT technology on YouTube uses seven main mechanisms: literal, script mixing, transliteration, deletion, addition, explicitation, and irrelevant translation. Furthermore, the analysis confirms that many translation errors stem from inaccuracies in the Automatic Speech Recognition (ASR) system, which generates the original (source-language) subtitles. The present study concludes that MT has revolutionized the process and the speed of subtitle generation, providing an opportunity for accessibility for readers. It recommends that it is essential to advance our knowledge of how Artificial Intelligence (AI) overcomes the cultural and linguistic gaps between English and Arabic.