الكفايات اللغوية لدى ثلاثة نماذج لغوية كبيرة، في كتابة مراسلات البريد الإلكتروني الرسمي بالعربية -دراسة وصفية تحليلية.
This study aims to evaluate the ability of three emerging Large Language Models (LLMs) — ChatGPT, Gemini, and Jais — that support the Arabic language and are widely accessible. The focus was on their performance in drafting formal Arabic email correspondence for professional and administrative purpo...
| Publicado en: | Arab Journal for the Humanities Vol. 43; no. 171; pp. 11 - 55 |
|---|---|
| Autor principal: | |
| Formato: | Artículo |
| Publicado: |
Kuwait University
Summer2025
|
| Materias: | |
| Acceso en línea: | Ver este registro en EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=hlh&AN=186968993&site=ehost-live header: @attributes: shortDbName: hlh uiTerm: 186968993 longDbName: Humanities International Complete uiTag: AN controlInfo: bkinfo: jinfo: jid: 10269576 1WRG jtl: Arab Journal for the Humanities issn: 10269576 maglogo: N pubinfo: dt: Summer2025 vid: 43 iid: 171 pid: 19632 pub: Kuwait University artinfo: ui: 186968993 10.34120/ajh.v43i171.3297 ppf: 11 ppct: 44 formats: fmt: @attributes: type: P size: 8.9MB tig: atl: الكفايات اللغوية لدى ثلاثة نماذج لغوية كبيرة، في كتابة مراسلات البريد الإلكتروني الرسمي بالعربية -دراسة وصفية تحليلية. aug: au: محمد عدلي عودة affil: أستاذ مساعد، قسم اللغة العربية وآدابها، كلية العلوم الإنسانية والاجتماعية، جامعة زايد، الإمارات العربية. su: Language models ChatGPT Arabic language Linguistic models Salutations sug: subj: Language models ChatGPT Arabic language Linguistic models Salutations keyword: email correspondences functional writing large language models linguistic performance assessment الكتابة الوظيفية النماذج اللغوية الكبيرة تقويم الأداء اللغوي مراسلات البريد الإلكتروني ab: This study aims to evaluate the ability of three emerging Large Language Models (LLMs) — ChatGPT, Gemini, and Jais — that support the Arabic language and are widely accessible. The focus was on their performance in drafting formal Arabic email correspondence for professional and administrative purposes. The study used a descriptive-analytical methodology. It analyzed the outputs of the models across six specific purposes: leave requests, transfer requests, resignation letters, administrative assignments, instruction delivery, and recommendation requests. A unified prompt was designed to ensure variation in the sender’s and recipient’s gender and job position. The models’ outputs were assessed based on clear structural and linguistic criteria, using both quantitative and qualitative analysis. The results showed that Gemini achieved the highest overall score at 88.3%, characterized by offering multiple response options and detailed explanations. Jais ranked second with 81.3%, demonstrating stronger cultural alignment, especially in opening and closing greetings. ChatGPT scored 80.4% but achieved the highest linguistic accuracy rate among the three models. Despite these strengths, all three models required human review and intervention to ensure full quality, clarity, and cultural appropriateness. The study concluded that while LLMs can assist in drafting Arabic emails, they cannot yet fully replace human oversight. It recommended expanding research efforts on Arabic email writing standards. More descriptive and normative studies are needed. The study also highlighted the importance of incorporating cultural dimensions into Arabic communication frameworks. Developing culturally sensitive and linguistically accurate models will improve their usability, relevance, and overall effectiveness in real-world Arabic contexts. pubtype: Academic Journal doctype: Article src: R language: Arabic refInfo: copyright: @attributes: flag: Y custom: Copyright of Arab Journal for the Humanities is the property of Kuwait University, Academic Publication Council and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. item: Arab Journal for the Humanities holder: Kuwait University, Academic Publication Council dt: @attributes: year: 2025 holdings: @attributes: islocal: N |
|---|