The Limits of Artificial Intelligence in Translating Cultural Nuances: a Comparative Study of Human and Machine Translation
Year: 2026
Type: dataset
Abstract: This article investigates the limits of artificial intelligence (AI) in rendering cultural nuances through a comparative study of human and machine translation (MT). Its aim is to identify the categories of culturally loaded meaning in which contemporary neural machine translation and large language models most systematically fall short of human translators, and to explain why. Methodologically, the study adopts a qualitative, illustrative comparative design: a purposive sample of culturally loaded source items — idioms, proverbs, culture-specific realia, honorifics and forms of address, kinship terms and pragmatic markers — is analysed by comparing typical MT output with professional human renderings against a cultural-adequacy framework drawn from Newmark and House. The findings show that MT performs well on denotative, context-independent content but falters wherever meaning depends on shared cultural knowledge, pragmatics, affect and register, while human translators retain a decisive advantage as cultural mediators. The article argues for a complementary, human-in-the-loop model rather than full automation, and draws implications for translator training and for lower-resource languages such as Uzbek.
Language: English
Subfield: Language and Linguistics
Field: Arts and Humanities
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7225321118