Muslihin

Assessment in Arabic language learning as a foreign language (Arabic as a Foreign Language) is currently undergoing a significant paradigm shift. This transformation is influenced by the development of digital technology, the utilization of Artificial Intelligence (AI), and the implementation of the Merdeka Curriculum in Indonesia, which emphasizes flexibility, differentiation, and learning personalization. This article presents a Systematic Literature Review (SLR) aimed at identifying, examining, and synthesizing developments, innovations, methodological approaches, and various challenges in the assessment of the four Arabic language skills, namely listening, speaking, reading, and writing, during the period from 2020 to early 2026. The review employed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) protocol to identify and analyze relevant literature. The findings indicate a shift from conventional evaluation systems based on discrete tests toward the implementation of Technology-Enhanced Assessment (TEA) and more authentic forms of assessment. The literature synthesis reveals several key findings. First, diagnostic assessment, encompassing both cognitive and non-cognitive aspects, plays an important role in designing differentiated learning within the framework of the Merdeka Curriculum. Second, the development of assessment instruments for Arabic language maharah is increasingly oriented toward measuring higher-order thinking skills (Higher Order Thinking Skills/HOTS). Third, the utilization of Generative AI, including ChatGPT, as well as various gamification-based learning platforms, has introduced new approaches to providing faster, more adaptive, and interactive feedback. Based on the synthesis, the effectiveness of modern assessment is not determined solely by the sophistication of the technology employed, but also by the alignment between technology and pedagogical objectives, the transparency of assessment procedures, and its ability to comprehensively represent learners' communicative competence while maintaining the principles of academic integrity.

The landscape of foreign language education, particularly Arabic language instruction, is currently undergoing a fundamental transformation driven by the emergence of Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs). This Systematic Literature Review evaluates recent developments in the integration of GenAI into the Arabic language learning ecosystem, encompassing trends in the evolution of language models, pedagogical effectiveness, the utilization of automated assessment systems, as well as the linguistic and cultural challenges involved. Through a thematic synthesis of empirical and theoretical literature published between 2019 and 2026, this review confirms that GenAI significantly enhances learner engagement, instructional personalization, and language skill acquisition, particularly in vocabulary mastery and reading comprehension. These improvements are driven by the adoption of general-purpose models such as ChatGPT, as well as leading Arabic-centric models such as JAIS and ALLaM. Nevertheless, this innovative integration encounters the inherent structural complexity of the Arabic language. Generative models frequently experience technical limitations in handling syntactic rules (nahwu), morphology (shorof), diacritic disambiguation, and regional dialect variations, as evidenced by performance gaps in linguistic benchmarks such as AraLingBench and AlGhafa. Furthermore, this review discusses rapid advances in trait-based Automated Essay Scoring (AES) supported by precise prompt engineering. Based on these findings, the study concludes that although GenAI has significant potential to bring positive disruption, its successful implementation depends heavily on pedagogical readiness, digital literacy, the application of Human-in-the-Loop (HITL) collaboration, and the development of ethical frameworks aligned with the epistemology of Islamic higher education.