Decoding Idioms with AI: A Linguistic Analysis of ChatGPT and Google Translate in English–Urdu Translation

Syeda Sara  Shah; Moneeba  Habib

doi:10.63056/

Authors

Syeda Sara Shah MS English Linguistics Scholar COMSATS University Islamabad, Pakistan Author
Moneeba Habib PhD Scholar, NUML. Faculty English Linguistics, COMSATS University Islamabad, Pakistan Author

DOI:

https://doi.org/10.63056/

Keywords:

ChatGPT, Google Translate, English idioms, Urdu translation, Machine translation, Meaning accuracy, Cultural appropriateness, AI-based language models, Figurative language

Abstract

The increasing use of online translation tools in Pakistan has highlighted the challenges of translating idiomatic expressions, which often carry cultural and figurative meaning. Idioms frequently pose difficulties for machine translation, leading to loss of meaning or unnatural expressions in the target language. Despite the rise of AI-based translation systems, limited research has explored their effectiveness in handling idiomatic and culturally embedded language. This study investigates how ChatGPT and Google Translate render English idioms into Urdu, focusing on meaning accuracy and cultural appropriateness. Using a comparative-descriptive approach, data was collected by translating 20 commonly used English idioms on both platforms and analysing the outputs. Findings reveal that ChatGPT produces largely accurate and culturally natural translations, employing context-aware paraphrasing, while Google Translate often generates literal and awkward translations. The study underscores the potential of AI-based models in managing figurative language and emphasizes the importance of cultural adaptation in translation.