TRANSLATION OF CULTURAL UNITS IN MACHINE TRANSLATION SYSTEMS

Abstract

Abstract: The rapid development of artificial intelligence and machine translation technologies has significantly influenced modern translation studies. Machine translation systems such as Google Translate and DeepL Translator are widely used for translating texts between different languages. However, the translation of cultural units remains one of the most difficult challenges for machine translation systems. Cultural units include idioms, proverbs, national traditions, historical references, customs, and culturally specific lexical items that reflect the identity and worldview of a particular nation. This thesis examines the problems of translating cultural units in machine translation systems, especially between English and Uzbek languages. The study analyzes semantic, pragmatic, and cultural difficulties that arise during automatic translation. Comparative and descriptive methods are used to evaluate translation accuracy and cultural adaptation in machine-generated texts. The findings demonstrate that machine translation systems often translate cultural expressions literally, resulting in semantic inaccuracies and loss of cultural meaning. The research also highlights that human translators possess cultural competence and contextual understanding that artificial intelligence systems still cannot fully replicate. The study concludes that machine translation technologies require further improvement in cultural and contextual processing, particularly for low-resource languages such as Uzbek.