Teaching Arabic-English legal translation using model texts: a mixed-method study

Legal English has distinctive stylistic and linguistic features that make legal translation complex and demanding specific skills and knowledge. Students, in particular, struggle due to limited familiarity with legal discourse and terminology in the target language. This study aims to develop Saudi university students’ Arabic–English legal translation ability through using model texts, drawing on the genre approach to raise awareness of legal conventions and textual features. The purpose is to address the paucity of experimental research on legal translation pedagogy and to evaluate whether model texts can improve students’ performance and awareness of legal conventions and terminology.

Legal English (legalese) is characterized by archaic and rarely used lexicon, Latin/French terms, frequent repetition for precision, long complex sentences with sparse punctuation, passive and impersonal style (Williams, 2004), as well as doublets/triplets and nominalization (Stanojević, 2011), and domain-specific jargon (Rylance, 1994). These stable features make learning possible yet translation difficult without background knowledge (Veretina-Chiriac, 2012; Galán-Mañas, 2013). Legal Arabic also shows unique features such as highly formal expressions, figurative language, masculine dominance, nominalization, doublets/triplets, complex sentences, dominance of active voice, and less standardized punctuation (El-Farahaty, 2015), highlighting Arabic–English legal translation difficulty. Empirical works on legal translator education are often descriptive (errors/practices) or prescriptive (models), with limited experimental studies. Descriptive studies report frequent grammar, vocabulary, and graphic errors in legal translations (e.g., Atabekova, 2023; Phelan, 2017), and for Arabic–English legal translation, common problems include syntactic, lexical, and layout/punctuation errors (Farghal & Shunnaq, 1992; Alrishan, 2019; Alshehab & Rababah, 2020). Prescriptive models emphasize corpus-based training to raise awareness of conventions and phraseology (Albi, 2013; Biel, 2017; Monzó Nebot, 2008; Prieto Ramos, 2021), course design elements such as legal background, vocabulary, authentic graded materials (Gómez González-Jover, 2011), integrative competence (Prieto Ramos, 2011), and contrastive rhetoric within task-based graded tasks (Galán-Mañas, 2013). Two conclusions emerge: trainees face multiple challenges leading to diverse errors, and few studies have experimentally tested legal translation training techniques. In light of this, the study poses two research questions: (1) To what extent does using model translated texts influence Saudi students’ Arabic–English performance in translating different legal text types (short legal documents—certificates, receipts, notifications, powers of attorney—versus legal contracts)? (2) What are the students’ perceptions of teaching legal translation using model texts?

Design: Mixed-method design using pre/post translation tests (quantitative) and a post-course open-ended questionnaire (qualitative). Participants: 37 Saudi male undergraduate English majors (third year) enrolled in an advanced translation course at a Saudi university (two classes: n=20 and n=17). Participation was voluntary with informed consent. Pedagogical treatment: Focused on Arabic–English legal translation over 8 academic weeks (24 instructional hours). Training used model translated texts selected and graded for difficulty from Al-Morshedy (2007) and Hatem et al. (1995), covering two legal text types: (a) short one-page legal documents (certificates, receipts, notifications, powers of attorney) and (b) contracts (longer texts). Each activity comprised four stages: (1) collaborative translation of the Arabic text into English (groups of 3–4), (2) comparison with an expert-translated model, (3) collaborative revision in light of the model, and (4) noticing and discussing linguistic and punctuation features in additional similar model translations. The instructor guided implementation, promoted noticing of legal English features and contrasts with legal Arabic, checked understanding, provided feedback and support, and monitored progress. Assessments: Pre- and post-tests each contained two sections: short legal texts (certificates and powers of attorney) and a contract excerpt. Different but length-matched texts were used across tests to avoid familiarity effects. Each test lasted 1.5 hours. Students could use printed general dictionaries. A 100-point rubric assessed: lexical accuracy/terminology (40), syntactic accuracy/grammar (40), and punctuation (20), primarily based on error counts. Two raters independently scored; average scores used. Error counts and percentages were computed per area and text type. Questionnaire: Seven open-ended questions (in Arabic) elicited perceptions of the training, perceived improvement, merits/demerits, relative difficulty of text types, remaining challenges, and suggestions. Responses were translated to English; thematic analysis was conducted independently by both authors and reconciled.

Quantitative results:

• Total errors decreased from 1054 (pre) to 486 (post). Short documents: 606→291; contracts: 448→195.
• By type (pre→post totals): lexical 491→167; syntactic 375→224; punctuation 188→95.
• Error type proportions (pre→post): lexical 46.58%→34.36%; syntactic 35.57%→46.09%; punctuation 17.83%→19.54%. All error types decreased in absolute numbers; the largest reduction was lexical.
• Students made more errors on contracts than on short documents in both tests.
• Wilcoxon Signed Ranks tests showed significant improvements across all features and both text types (all p=0.000). Means (SD): • Short documents: Lexical 16.10(1.11)→32.81(1.72); Syntactic 18.90(1.05)→29.70(0.87); Punctuation 12.35(1.27)→17.51(1.07); Total 47.35(2.62)→80.02(3.30). • Contracts: Lexical 15.91(1.08)→31.27(1.16); Syntactic 17.84(1.21)→28.40(1.09); Punctuation 12.82(1.18)→17.79(1.14); Total 46.57(2.44)→77.46(3.79).
• Gains were slightly lower for contracts than for short documents. Qualitative results:
• Students valued models for raising awareness of legal conventions and terminology; many reported initial difficulty that eased as familiarity grew.
• Most perceived notable improvement, especially in legal vocabulary and strategies; they felt better prepared for future professional translation tasks.
• Contracts were perceived as more difficult due to denser and less familiar terminology; short documents were considered easier.
• Remaining needs included more exposure to legal English features, extensive practice, and access to resources (lexical lists, legal corpora). Some suggested longer or additional courses.

Using model translated texts enhanced trainees’ awareness and application of legal English terminology and conventions, reducing errors and improving lexical, syntactic, and punctuation performance. The strongest gains were lexical, aligning with students’ emphasis on terminology as central to legal translation and with the training’s consciousness-raising focus on genre features. Improvements in syntax and punctuation likely stemmed from guided noticing across multiple models. Error distributions and persistent challenges, particularly in contracts, align with prior research on legal translation difficulties and error types. Findings support genre- and corpus-informed pedagogy that emphasizes representative, graded materials and explicit attention to legal conventions, terminology, and phraseology. The short 8-week duration limited the depth of exposure per genre, suggesting greater benefits with extended training.

Model texts are an effective core tool for teaching English legal translation to novice translators by familiarizing them with genre-specific linguistic and textual features. Effective implementation requires graded materials based on text difficulty and sufficient training time. The approach is especially impactful in reducing lexical errors; instructors should deliberately draw attention to syntactic and punctuation features through awareness-raising activities. Future work should extend training duration, broaden participant demographics, and examine effects in additional language pairs.

• Short training duration (8 weeks) constrained depth and scope across legal genres.
• Single-gender sample (male only) due to institutional constraints limits generalizability; female trainees’ responses remain to be studied.
• Focus on Arabic–English; effects on other target languages not examined.
• Data not publicly available due to privacy; limits external reanalysis.