Abstract : This paper investigates the enhancement of oral proficiency among seven- and eight-year-old Mother Tongue Language (MTL) learners in Singapore primary schools. Linguistic data with a focus on reading aloud and speaking competencies – including pronunciation accuracy, grammatical usage, and lexical diversity – were collected to inform the design of an AI-powered educational tool. These insights guided the development of LEARN (Language automated Evaluation by generating Answers/questions from caRtooNs), a conversational chatbot codeveloped by the National Institute of Education (NIE) – Nanyang Technological University (NTU), and the Singapore Institute of Technology (SIT). LEARN engages students in short, scaffolded dialogues based on picture description tasks, providing an interactive and supportive environment for practicing oral skills in their respective mother tongues. Grounded in educational scaffolding principles, LEARN facilitates gradual language development through iterative conversational loops. This paper outlines the chatbot’s core features, pedagogical affordances, and its potential to enhance oral communication in early language education.

Keyword : AI-Powered Chatbot, Learning Loops, Malay Language, Mother Tongue Languages, Oral Proficiency