Document Type : Research Paper
Authors
1 Shahid Mahallati University of Islmic Studies
2 Department of English Language and Literature, Fculty of Hunamities, Yasuj University
3 Department of English Language and Literature, Faculty of Humanities, Yasuj University
Abstract
AI-based models of analyzing linguistic data have attracted researchers due to the theoretical and practical spread of these models in recent years. Among them, machine translation and AI-based translation have been a focal point in this development However, many questions regarding their efficiency and accuracy are raised at the same time. The current research is an attempt to employ an evaluative framework from systemic functional linguistics and investigate the accuracy of the transfer of ideational meta-function in the English translation of Quranic verses by GPT-4o. To this aim, 200 verses of the holy Quran were selected purposively and the transfer of six categories of the aforesaid meta-functions including material, verbal, existential, mental, behavioral, and existential were explored based on verb category, participants, and conditions of each process. Results revealed that GPT has been able to transfer material, behavioral, and existential processes while failing to transfer mental and relational processes. It seems that GPT’s success in the first section cannot be attributed to AI’s ability to distinguish the qualities and conditions of the processes but to the limitedness of verbal category and objectivity of their semantic components, participant types, and conditions. Since the characteristics of each part of speech for each process are determined and defined in systemic functional linguistics, they can be attached to the machine translation models in the form of syntactic-semantic tags added to each lexical-grammatical unit to increase the accuracy of translation. It seems that GPT’s success in the first section cannot be attributed to AI’s ability to distinguish the qualities and conditions of the processes but to the limitedness of verbal category and objectivity of their semantic components, participant types, and conditions. Since the characteristics of each part of speech for each process are determined and defined in systemic functional linguistics, they can be attached to the machine translation models in the form of syntactic-semantic tags added to each lexical-grammatical unit to increase the accuracy of translation.
Keywords
Main Subjects
)