Enhancing Bilingual Skills and Cultural Sensitivity in Chinese Preschoolers: Optimizing Deepseek AI Interactive Storytelling with Runge-Kutta 6(5) Simulations

Abstract

In China's expanding preschool bilingual education sector, this research investigates a Deepseek-based interactive storytelling method optimized via RungeKutta 6(5) numerical simulations to boost Mandarin-English skills and cultural sensitivity among children aged 3-6. Deepseek, a localized open-source AI,
generates responsive narratives that merge linguistic exercises with cultural motifs, tackling the shortcomings of conventional rote techniques in engaging young
learners. Drawing on Vygotsky's sociocultural theory and Krashen's comprehensible input, the study addresses preschoolers' brief focus spans and cultural
isolation. A mixed-methods design, dominant in qualitative aspects, involved 120 preschoolers from urban vocational settings in Shanghai, incorporating AI session
logs, educator and parent interviews, and thematic reviews. Runge-Kutta 6(5) simulations modeled learning trajectories under varied strategies, revealing optimal
combinations. Outcomes indicated 20-25% gains in bilingual vocabulary and heightened cultural empathy, with children collaboratively crafting tales blending
Chinese traditions like Lunar New Year with English elements. Issues such as AI biases and access inequalities surfaced. Contributions encompass a culturally
attuned AI-simulation framework for ethical early learning, guiding China's curriculum updates and urban-rural fairness. Echoing Hwang et al. (2025a) on robotic
emotional aids and Mubarok et al. (2024) on AI-supported collaboration, it underscores AI's capacity for low-stress, joint learning. Future efforts may extend to
rural zones, enhancing Deepseek for dialect variety and enduring bilingual progress .