]THE IMPACT OF MACHINE TRANSLATION ON THE TRANSFER OF PRAGMATIC MEANING: AN EMPIRICAL ENGLISH–UZBEK PERSPECTIVE

The rapid expansion of artificial intelligence technologies and neural machine translation systems has significantly transformed global intercultural communication and multilingual information exchange (Koehn, 2020; O’Brien, 2023). Contemporary machine translation tools such as Google Translate, DeepL, Microsoft Translator, and ChatGPT-based systems increasingly demonstrate high levels of grammatical fluency and semantic accuracy. Nevertheless, despite substantial technological progress, machine translation continues to encounter serious difficulties in transferring pragmatic meaning, particularly between linguistically and culturally distant languages such as English and Uzbek (House, 2015; Kecskes, 2014). The present study investigates the effectiveness of AI-based machine translation in preserving pragmatic equivalence in English–Uzbek translation. The research employs a mixed-method corpus-based design combining qualitative pragmatic analysis and quantitative statistical evaluation. A corpus consisting of 160 English expressions containing pragmatic elements was analyzed. The dataset included idiomatic expressions, indirect speech acts, politeness strategies, conversational implicatures, sarcastic statements, humor, and culturally marked units. The selected materials were translated into Uzbek using Google Translate, DeepL, ChatGPT, and Microsoft Translator. Machine-generated translations were compared with expert human translations based on pragmatic equivalence, contextual appropriateness, communicative naturalness, and sociocultural adaptation (Baker, 2018). The findings demonstrate that while AI systems successfully transfer denotative semantic meaning in informational texts, they frequently fail to preserve implicit communicative intention, politeness hierarchy, cultural nuance, figurative meaning, and discourse-sensitive pragmatics. Quantitative analysis revealed that idiomatic expressions and sarcastic statements produced the highest rates of pragmatic failure, while indirect speech acts demonstrated relatively higher translation accuracy. Among the analyzed platforms, ChatGPT-based translation showed comparatively stronger contextual adaptation, although substantial limitations remained in culturally embedded discourse. The study argues that pragmatic competence remains one of the most challenging dimensions of machine translation because pragmatic meaning depends heavily on sociocultural cognition, contextual inferencing, and intercultural communicative norms (Levinson, 1983; Verschueren, 2012). The article contributes to translation studies, intercultural pragmatics, and computational linguistics by proposing an empirical framework for evaluating pragmatic equivalence in low-resource language pairs such as English and Uzbek.人工智能技术与神经机器翻译系统（Neural Machine Translation Systems）的迅猛发展，极大重塑了全球跨文化传播与多语言信息交流格局（Koehn, 2020; O’Brien, 2023）。当下主流机器翻译工具，如谷歌翻译（Google Translate）、DeepL、微软翻译（Microsoft Translator）以及基于大语言模型（Large Language Model）的ChatGPT类系统，其语法流畅度与语义准确率正持续提升。尽管技术已取得长足进步，但机器翻译在传递语用意义（Pragmatic Meaning）层面仍面临严峻挑战，尤其在英语与乌兹别克语这类语言与文化隔阂显著的语种对之间（House, 2015; Kecskes, 2014）。 本研究旨在探究基于人工智能的机器翻译在英-乌兹别克语翻译中保留语用等效性（Pragmatic Equivalence）的有效性。研究采用基于语料库的混合研究设计，融合定性语用分析与定量统计评估两种方法。分析了由160个含语用元素的英语表达构建的语料库，该数据集涵盖习语表达、间接言语行为（Indirect Speech Acts）、礼貌策略、会话含义（Conversational Implicatures）、讽刺性表述、幽默文本以及文化标记性语料单元。研究人员使用谷歌翻译、DeepL、ChatGPT及微软翻译，将筛选出的语料转译为乌兹别克语，并以语用等效性、语境适配性、交际自然度与社会文化适配性为评价标准，将机器生成译文与专家人工译文进行比对（Baker, 2018）。 研究结果显示，尽管人工智能系统可在信息类文本中准确传递外延语义（Denotative Semantic Meaning），但在保留隐性交际意图（Implicit Communicative Intention）、礼貌层级（Politeness Hierarchy）、文化细微差异（Cultural Nuance）、比喻义（Figurative Meaning）以及话语敏感型语用（Discourse-Sensitive Pragmatics）层面时常失效。定量分析结果表明，习语表达与讽刺性表述的语用失效率最高，而间接言语行为的翻译准确率相对更高。在所分析的翻译平台中，基于ChatGPT的译文在语境适配性上表现相对更优，但在文化内嵌型话语处理上仍存在显著局限。 本研究指出，语用能力仍是机器翻译最具挑战性的维度之一，因为语用意义高度依赖社会文化认知、语境推理与跨文化交际规范（Levinson, 1983; Verschueren, 2012）。本研究通过构建针对英-乌兹别克语这类低资源语言对（Low-Resource Language Pairs）的语用等效性评价实证框架，为翻译学、跨文化语用学（Intercultural Pragmatics）与计算语言学（Computational Linguistics）领域提供了新的研究视角。