مجلة الرسالة للدراسات والبحوث الإنسانية
Volume 11, Numéro 1, Pages 344-359
2026-03-28
Authors : Bougherara Antar . Benaouda Adila .
This comprehensive study examines the comparative analysis of errors that occur in bidirectional Arabic–English legal translation when using machine translation (MT) systems versus computer-assisted translation (CAT) platforms. Building upon recent research in neural machine translation and post-editing practices, this investigation analyses a corpus of 65,000 words comprising legal contracts, legislation, court decisions, and professional conduct codes processed through leading CAT environments and state-of-the-art neural MT engines. The study integrates findings from multiple Arabic legal translation corpora and examines the specific challenges posed by Arabic's morphological richness and the structural differences between Islamic and common law systems. Results demonstrate that while neural MT provides increasingly fluent output, critical accuracy errors persist in specialized legal terminology, register appropriateness, and cross-cultural legal equivalence. CAT workflows significantly reduce such errors through human post-editing intervention, yet persistent discourse-pragmatic issues and formatting inconsistencies reveal the continued impact of human factors and current technological limitations. The research contributes to the growing body of literature on Arabic–English legal translation by providing empirical evidence for training priorities, quality assessment frameworks, and the development of domain-aware MT and CAT solutions tailored to the unique characteristics of Arabic legal discourse.
machine translation, ; computer-assisted translation, ; Arabic-English legal translation. ; machine translation, computer-assistedpost-editing,. ; translation quality assessment, . ; neural machine translation
بوسالم أحلام
.
عابد يوسف
.
ص 117-132.
Yahia Zeghoudi
.
pages 74-88.
Said Houari Amel
.
pages 257-268.
Khaldi Anissa
.
pages 147-153.
Beldjenna Amel
.
pages 121-133.