The Accuracy of “Nahwu AI” in Arabic Syntactic Analysis (Tarkib): An Evaluative Study Based on Ibn Malik’s Perspective

Abstract

Advances in artificial intelligence (AI) for Arabic language learning have encouraged the development of automated syntactic analysis platforms, including Nahwu AI. However, the extent to which its analytical output conforms to the principles of classical Arabic grammar remains insufficiently evaluated. This study aims to assess the conformity of Nahwu AI’s syntactic analysis with the grammatical rules of Alfiyyah Ibn Malik and to identify the platform’s strengths and limitations. Employing a qualitative evaluative research design, the study adopted a critical-comparative analysis by comparing the output generated by Nahwu AI with the grammatical principles presented in Alfiyyah Ibn Malik and Syarh Ibn Aqil as normative references. The dataset comprised 30 Arabic sentences selected through purposive sampling and analyzed using the Nahwu AI platform. The findings indicate that 20 sentences (66.7%) were classified as accurate, 5 (16.7%) as moderately accurate, and 5 (16.7%) as inaccurate. The highest accuracy was achieved in tawabi’ and idhafah constructions (80%), followed by jumlah fi’liyyah and jumlah syarthiyyah (75%), whereas jumlah ismiyyah showed the lowest accuracy (37.5%). Most errors involved identifying verbal sentence structures, khabar muqaddam, syibh al-jumlah predicates, mudhaf–mudhaf ilaih relations, and isim syarth functions. These findings suggest that Nahwu AI effectively recognizes basic syntactic structures but remains limited in handling grammatically complex constructions requiring deeper linguistic reasoning. This study contributes a normative evaluation framework that bridges AI-based syntactic analysis with the principles of classical Nahwu, providing implications for the development of more linguistically reliable Arabic learning technologies.