From traditional to robot-assisted learning: a multimodal robot-assisted learning framework for enhancing english acquisition in korean preschoolers

Abstract

Traditional language instruction has been the dominant approach in early childhood education, yet integrating robotic systems presents new opportunities to enhance second language acquisition. This study introduces a multimodal robot-assisted learning framework featuring the OpenManipulator-X, an interactive robotic system designed to support English acquisition among Korean preschoolers. A comparative experiment was conducted in which children first participated in teacher-led English instruction, followed by robot-assisted learning using task-based interactions, including pick-and-place activities and collaborative drawing. The study evaluates engagement, vocabulary acquisition, and learning effectiveness across both methods. Results indicate that robot-assisted learning fosters greater engagement and interaction, while language retention remained comparable to traditional instruction. In addition, the robot's physical behaviors-such as object manipulation and drawing-contributed to increased attention and participation. The novelty of this work lies in its integration of multimodal interaction combining robotic embodiment, AR-based manipulation, and tablet-guided educational tasks within a unified instructional system. These findings contribute to child-robot interaction (CRI) research, offering a scalable hybrid learning model for early childhood education. Future work should explore AI-driven personalization and long-term developmental impact.