Translation and Interpreting Research

Translation and Interpreting Research

The Effect of ChatGPT-Assisted Pre-Editing on the Naturalness and Readability of Google Translate's Persian Output

Document Type : Translation Studies

Authors
Hamideh Nemati Lafmejani 1
Seyed Mehdi Mousavian 2
1 Assistant Professor in Translation Studies, Department of English Language, SR.C., Islamic Azad University, Tehran, Iran
2 Department of English Language, SR.C., Islamic Azad University, Tehran, Iran
10.22054/tir.2026.96338.1100
Abstract
The quality of machine translation depends not only on the capabilities of the translation system but also on the linguistic characteristics and complexity of the source text. Although pre-editing has been widely explored in relation to controlled language and machine translation, the potential of generative artificial intelligence, particularly ChatGPT, as a pre-editing tool remains underexplored. This study investigates the effectiveness of ChatGPT-5 in pre-editing English news texts prior to machine translation into Persian. Thirty-five English news articles were processed using ChatGPT-5 to generate controlled language optimized for machine translation. Both the original and ChatGPT-pre-edited versions were subsequently translated into Persian using Google Translate and evaluated by two professional translators for naturalness and readability on a three-point scale. The results demonstrate notable improvements following ChatGPT-5 pre-editing: mean naturalness increased from 1.45 to 2.34, while mean readability rose from 1.77 to 2.71. These findings suggest that ChatGPT-assisted pre-editing can enhance the naturalness and readability of machine-translated English-to-Persian journalistic texts. The study highlights generative AI as a promising addition to the machine translation workflow, extending the role of pre-editing from a traditionally human-intensive practice toward an AI-assisted process and contributing to the growing body of research on generative AI in translation.
Keywords
AI-assisted translation
machine translation
pre-editing naturalness
readability
Translation and Interpreting Research
Volume 3, Issue 12
Autumn 2026
Pages 41-55