THE EFFICIENCY OF AI TOOLS IN DETECTING EUPHEMISMS AND TABOOS IN ENGLISH AND UZBEK LANGUAGES

Abstract

This study evaluates the efficiency of artificial intelligence (AI) tools in detecting euphemisms and taboos in English and Uzbek languages, adopting a linguo-pragmatic perspective. Euphemisms and taboos, as context-sensitive linguistic phenomena, pose challenges for automated detection due to their cultural and pragmatic nuances. Using natural language processing (NLP) models, including transformer-based architectures and multilingual embeddings, this research assesses AI performance in identifying these linguistic constructs across both languages. A mixed-methods approach combines quantitative metrics (precision, recall, F1-score) with qualitative analysis of cultural misinterpretations. Results indicate that AI tools achieve high accuracy in English but face limitations in Uzbek due to data scarcity and cultural complexity. These findings highlight the potential of AI in linguistic analysis while underscoring the need for culturally informed training data.