Arabic Text Simplification and Readability Assessment for Educational Applications: A Systematic Literature Review

Abstract

Arabic text simplification and readability assessment support level-appropriate material selection, language learning, early literacy, teacher adaptation, and accessibility, but the degree of educational validation remains uncertain. This PRISMA 2020 systematic review searched Scopus, Web of Science Core Collection, and IEEE Xplore on 6 August 2026 and updated coverage through a targeted ACL Anthology search, a standalone-readability sensitivity search, and citation-cluster checking on 11 September 2026. Following source-identity verification, 194 records were identified, 129 were screened, 69 full texts were assessed, and 61 studies were included. An eight-criterion appraisal rated 45 studies high, 13 moderate, and 3 low quality. Thirty-four studies primarily modelled readability, while recent work increasingly used Arabic pretrained language models, ordinal objectives, graph models, large corpora, and controlled generation. Forty-two studies relied primarily on automatic or benchmark metrics; only two directly measured learner comprehension or retelling, and none measured longitudinal learning gain. Excluding low-quality studies and interpreting documented resource reuse did not alter the central conclusion. The retrieved evidence demonstrates technical feasibility and growing educationally relevant resources, but educational effectiveness remains unestablished. Priorities are independent learner and teacher evaluation, cross-domain and cross-variety validation, interoperable scales, human calibration of automatic metrics, and closed-loop diagnosis, simplification, and level verification.