FRIEND OR FOE? RE-EVALUATING TECHNOLOGY-ENHANCED ASSESSMENT FOR MANDARIN COMPOSITION IN HIGHER EDUCATION

Abstract

The rapid integration of generative artificial intelligence (AI) in our everyday lives is creating an 'academic double standard' in many higher education institutions where faculty members are being encouraged to utilize AI in their work environment to increase productivity however, students will be restricted when it comes to assessments of their knowledge. This study examines the tensions existing in the context of classroom-based language assessment, particularly on the topic of Mandarin as a foreign language (MFL) composition writing assignments at the Universiti Malaysia Sabah (UMS). A mixed methodology applies both quantitative surveys and qualitative interviews were used to explore the views of two populations: 23 Mandarin language students and 7 MFL instructors regarding the usefulness of AI, ethical issues related to institutional uses of AI, and how AI affects teaching practices. The results of the quantitative survey indicate that students view AI through a practical perspective; 69.6% see AI primarily as an instructional tool allowing them to actively negotiate linguistic complexity prior to submitting written compositions instead of viewing AI as a means to cheat. Both student respondents and instructor respondents agree that AI provides high value, instantaneously assists learners with logographic and tonal complexities inherent to Mandarin Chinese. Nevertheless, each of the seven instructors surveyed agreed that if learners rely too heavily upon AI produced written output they may ultimately mask a learner's true ability to communicate. Additionally, this research identifies a significant gap in the governance of these technologies within higher education: While 43.5% of students clearly identify an institutional double standard as perceived by themselves as unfair, 100% of the instructor respondents indicated that they have utilized AI professionally and never received formal instruction or training on ethical considerations related to AI. Therefore, this paper concludes that blanket prohibitions on technology should not exist; Rather, we advocate that higher education institutions develop consistent and transparent guidelines and policies governing the use of AI within educational settings, and shift towards developing process oriented, multimodal assessment frameworks such as oral vivas and drafting portfolios to assess true language proficiency.