Adaptive Game-Based Learning and Arabic Vocabulary Achievement: A Quasi-Experimental Evaluation of AI-Supported Instruction

Abstract

This study evaluates the effect of a rule-based AI-supported adaptive game-based learning system on Arabic vocabulary learning among secondary-level students in Indonesia. A quantitative quasi-experimental pretest-posttest control group design was employed with 84 students, consisting of 42 students in the experimental group and 42 students in the control group. The experimental group learned Arabic vocabulary through an adaptive game-based system that adjusted task difficulty based on response accuracy, error patterns, and learning progress, while the control group received conventional vocabulary instruction. Data were analyzed using descriptive statistics, paired-sample t-tests, Welch’s t-test for gain-score comparison, effect size estimation, normalized gain analysis, and ANCOVA. The results showed that both groups improved, but the experimental group achieved stronger vocabulary gains than the control group. The mean gain score was 7.07 in the experimental group and 3.64 in the control group. Welch’s t-test confirmed a significant between-group difference in gain scores, and ANCOVA showed that the experimental group retained a significant adjusted advantage after controlling for pretest scores. These findings suggest that adaptive game-based learning can support Arabic vocabulary development, although the low N-Gain values and baseline differences require cautious interpretation.