Human versus AI in Literary Translation: Rendition of Cultural Nuances in Shahnameh

This comparative case study aims to provide a descriptive account of the types and frequencies of culture-specific items (CSIs) based on Newmark’s (1988) five areas of CSIs and to specify the translation strategies used by human translators and an artificial intelligence (AI) platform called Microsoft Copilot according to Davies’s (2003) model. CSIs were randomly selected from every ten verses of the first book of Shahnameh. After categorizing the CSIs, the translation strategies were specified. The findings revealed that the most frequent translation strategy employed by human translators was Transformations (28%), whereas AI predominantly used Localization (38%). It was also found that spelling differences exist between human translation and the AI output. Additionally, the AI translation demonstrated greater consistency in translating CSIs with the same meaning, whereas the human translation was inconsistent. This study contributes to the field by examining the role of AI in translating poetry, especially epic poetry. It also helps assess the feasibility of AI platforms in assisting humans with the translation process and underscores the importance of including AI training in literary translation curricula.