Usability framework for mobile augmented reality language learning

Abstract

After several decades since its introduction, the existing ISO9241-11 usability framework is still vastly used in Mobile Augmented Reality (MAR) language learning. The existing framework is generic and can be applied to diverse emerging technologies such as electronic and mobile learning. However, technologies like MAR have interaction properties that are significantly unique and require different usability processes. Hence, implementing the existing framework on MAR can lead to non-optimized, inefficient, and ineffective outcomes. Furthermore, state-of-the-art analysis models such as machine learning are not apparent in MAR usability studies, despite evidence of positive outcomes in other learning technologies. In recent MAR learning studies, machine learning benefits such as problem identification and prioritization were non-existent. These setbacks could slow down the advancement of MAR language learning, which mainly aims to improve language proficiency among MAR users, especially in English communication. Therefore, this research proposed the Usability Framework for MAR (UFMAR) that addressed the currently identified research problems and gaps in language learning. UFMAR introduced an improved data collection method called Individual Interaction Clustering-based Usability Measuring Instrument (IICUMI), followed by a machine learning-driven analysis model called Clustering-based Usability Prioritization Analysis (CUPA) and a prioritization quantifier called Usability Clustering Prioritization Model (UCPM). UFMAR showed empirical evidence of significantly improving usability in MAR, capitalizing on its unique interaction properties. UFMAR enhanced the existing framework with new abilities to systematically identify and prioritize MAR usability issues. Through the experimental results of UFMAR, it was found that the IICUMI method was 50% more effective, while CUPA and UCPM were 57% more effective than the existing framework. The outcome through UFMAR also produced 86% accuracy in analysis results and was 79% more efficient in framework implementation. UFMAR was validated through three cycles of the experimental processes, with triangulation through expert reviews, to be proven as a fitting framework for MAR language learning.