Exploring the Evolution of Reading Literacy Guided by TALPer: A Case Study Based on Ordered Network Analysis (ONA)

Abstract

This study employs Learning Analytics to investigate the dynamic development of Reading Literacy within the Taiwan Adaptive Learning Platform (TALP), specifically focusing on interactions guided by the AI companion, TALPer. Utilizing Ordered Network Analysis (ONA), we analyze the dialogue records of 19 students to map behavioral transitions based on the Progress in International Reading Literacy Study (PIRLS) framework. Unlike traditional frequency-based analysis, ONA provides insights into the directed connections and temporal order between AI scaffolding (E-codes) and student cognitive responses (S-codes). The results reveal a significant structural shift along the X-axis (MR1) between the early phase (First 3 Days) and the late phase (Last 3 Days) (p = 0.04, Cohen's d = 4.91). The comparison plot highlights the specific shifts in response strength. The transition from diverse scaffolding interactions (e.g., E_L1~E_L4, E_COR) and meaningful response (S_MEA) toward a highly focused pattern of test initiation and surface response confirms that student engagement evolved from exploratory reading guidance to convergent, assessment-oriented interactions. This research highlights the efficacy of ONA in visualizing the temporal evolution of reading competencies in AI-supported digital environments.