The efficacy of artificial intelligence - powered scaffolding in individual acquisition efficiency of EFL in tertiary educational context

TL;DR

This study explores how AI-powered tools help university students learn English more efficiently by analyzing cognitive processes and learning outcomes.

Contribution

The study introduces a cross-disciplinary framework combining TAM and CLT to evaluate AI-powered scaffolding in EFL learning.

Findings

Perceived usefulness and ease of use directly predict learning efficiency in EFL.

AI-assisted conversation frequency indirectly affects learning through cognitive processing depth.

Cognitive ability enhances the relationship between AI tool usage and cognitive processing depth.

Abstract

This study investigates the cognitive mechanisms and educational efficacy of AI-powered scaffolding in the acquisition of English as a Foreign Language (EFL) in tertiary education.Integrating the Technology Acceptance Model (TAM) and Cognitive Load Theory (CLT), the cross-disciplinary framework explores multidimensional pathways affecting the acquisition efficiency of EFL, focusing on learning efficiency of individual acquisition of EFL (LEF) of university students at all levels, and highlighting the mediating role of Cognitive Processing Depth (CPD) and moderating effects of Cognitive Ability (COA).

Quantitative data analysis from university students using AI-assisted conversational tools of AI -powered scaffolding were specifically conducted via structural equation modeling (SEM) and necessary condition analysis (NCA).

Results indicate perceived usefulness (PU) and ease of use (PEoU) directly predict LEF, while interaction frequency of AI-assisted conversation (AIC) exerts indirect effects through CPD. Cognitive ability strengthens the relationship between AI-conversational tool usage and CPD, supporting Self-Regulated Learning theory. NCA identifies critical thresholds of AIC and PeoU for achieving effective learning outcomes, offering actionable insights for real-time educational interventions.

The findings emphasize the necessity of cognitive adaptation strategies, platform diversification, and learner-centric AI-conversational tool design. While limited by sample homogeneity and cross-sectional data, this study underscores the value of longitudinal approaches and broader socio-cognitive investigations in future research. Collectively, Such findings based on empirical evidence, advance the optimizing of AI-enhanced, cognitively attuned language learning systems.

Figures