POLITENESS WITHOUT TASK EQUIVALENCE? A CONTRASTIVE PRAGMATIC ANALYSIS OF ENGLISH AND UZBEK GENERATIVE AI PROMPTS

Abstract

Cross-linguistic comparisons of generative artificial intelligence prompts frequently associate English imperatives with communicative efficiency and Uzbek request constructions with politeness. Such comparisons require closer examination because interpersonal wording may vary alongside the task itself. This article investigates the relationship between directive form, politeness marking and task specification in four English– Uzbek prompt pairs reproduced in a dissertation manuscript. A qualitative contrastive analysis distinguishes the requested operation, target material, content scope, output constraints and interpersonal framing. The findings identify several sources of non-equivalence that cannot be attributed to politeness alone: paragraph-level analysis is compared with text-level analysis, summarisation with shortening, and an unqualified word-count target with an approximate target accompanied by an additional demand for detail. The analysis also shows that Uzbek imperative morphology can coexist with respectful address, while interrogative request forms retain a directive function. A procedure for checking task equivalence before interpreting pragmatic differences is proposed. The contribution is methodological: the article establishes criteria for separating linguistic description, sociocultural interpretation and claims about model performance. It does not estimate national communication preferences or experimentally measure the effects of politeness on generated responses.