From AI Output To Intercultural Competence: A Didactic Algorithm For Interpreting Semantic Loss In Proverbs And Idioms

Abstract

The rapid integration of artificial intelligence (AI)–based translation systems into language education has transformed translation practices while raising concerns about the handling of culturally marked units such as proverbs and idioms. Although AI tools often ensure lexical accuracy, they frequently generate semantic loss, particularly in pragmatic, axiological, and conceptual layers. This study analyzes 100 Uzbek and English proverbs and idioms translated by three AI systems using a multi-layer cultural-marker framework to identify patterns of semantic reduction. The findings show that while denotative meaning is generally preserved, figurative imagery, evaluative nuance, and discourse-related functions are often weakened. In response, a five-stage didactic algorithm—unit selection, AI output collection, comparative analysis, semantic loss diagnosis, and adaptive reconstruction—is proposed to transform AI output into a tool for developing intercultural interpretive competence. The model positions AI translation as a diagnostic resource rather than a final authority and introduces measurable criteria for assessing cultural adequacy in translation tasks.