Harnessing Artificial Intelligence to Enhance English Learning in Early Grades Jumazoda Shohidai Jaloliddin

Аннотация

The rapid evolution of artificial intelligence (AI) has begun to transform language education, offering adaptive, data-driven approaches that are especially valuable in the formative years of schooling. This article investigates how AI-based solutions—ranging from adaptive learning environments and natural-language- processing chatbots to automated speech-recognition tutors—can enrich English instruction in grades 1–4. Drawing on a mixed-methods design that combined classroom interventions in three Uzbek primary schools with longitudinal analytics from an adaptive learning platform, the study tracked 216 pupils over two academic terms. Quantitative results demonstrate statistically significant gains in vocabulary depth, phonological awareness and reading fluency, while qualitative classroom observations reveal heightened learner motivation and more diversified teacher feedback loops. The findings highlight the importance of carefully aligned human–AI pedagogy: when algorithms personalise pacing, error correction and multimodal stimuli, teachers gain time for higher-order formative assessment and socio-emotional support. The discussion situates these outcomes within sociocultural theories of early literacy and argues that equity of access, ethical data stewardship and teacher professional development are pre-conditions for sustainable AI integration. Keywords: Artificial intelligence; early literacy; primary education; adaptive learning; speech recognition; personalised feedback; English as a foreign language. overlaying a second linguistic code demands Introduction: During the past decade, AI has shifted instructional sensitivity to cognitive load and cross- from an abstract research frontier to an applied linguistic transfer. Second, classroom heterogeneity in technology that quietly shapes many facets of daily life. prior exposure, socio-economic background and Education is no exception. In high-income contexts, learning pace complicates the design of one-size-fits-all adaptive reading dashboards, voice-activated curricula. Third, primary teachers—especially in pronunciation coaches and data-driven progress contexts where English is a compulsory foreign trackers have begun to supplement conventional language—often face large classes, limited literacy instruction. In lower- and middle-income instructional hours and growing assessment duties. AI countries, similar tools are emerging, often supported promises to address these challenges by individualising by multilingual interfaces and low-bandwidth content sequencing, providing immediate formative optimisation. Yet systematic evidence on their feedback and generating real-time analytics for pedagogical value in early grades remains fragmented, teachers. and policy frameworks lag behind technological possibility. Nevertheless, uncritical adoption risks pedagogical superficiality, data-privacy breaches and the widening Early English learning poses several intertwined of digital divides. Research must therefore move challenges. First, pupils aged six to ten are beyond anecdotal enthusiasm toward rigorous, simultaneously developing phonological, orthographic context-sensitive evaluation. This study explores three and semantic systems in their home language; American Journal Of Philological Sciences 226 https://theusajournals.com/index.php/ajps American Journal Of Philological Sciences (ISSN – 2771-2273) guiding questions: (1) To what extent do AI-mediated Across all three schools, the experimental cohort tasks improve core components of early English outperformed controls on every assessed domain. proficiency compared with traditional instruction? (2) Vocabulary depth rose by a mean of 18.4 percentage How do such tools influence learners’ affective points (p < 0.01), phonological awareness by 12.7 engagement and teachers’ instructional strategies? (3) percentage points (p < 0.05) and reading fluency by What design principles and implementation conditions 14.3 words correct per minute (p < 0.01). Effect sizes maximise benefit while mitigating risk? By addressing ranged from d = 0.41 to d = 0.63, indicating moderate these questions through a quasi-experimental study practical significance. Regression modelling revealed situated in Uzbek primary schools, the article that time-on-task within the adaptive vocabulary contributes empirical and theoretical insights to the module accounted for 37 % of variance in vocabulary nascent literature on AI-supported early foreign- gain after controlling for baseline score, demonstrating language education. the potency of algorithmic spacing and multimodal input. The research was conducted in three state primary schools in Tashkent and Samarkand that share Observation notes depict palpable enthusiasm during comparable class sizes, curricular hours and socio- chatbot storytelling and pronunciation games; pupils economic catchments. A total of 216 pupils in grades 2 frequently requested additional turns and and 3 participated, div