Exploring Translation Strategies in AI-Assisted Rendition of Banyumasan Javanese Wayang Kulit Dialogues into Indonesian and English

Abstract

Wayang Kulit Banyumasan, a traditional shadow puppet theater from Central Java, presents unique linguistic and cultural challenges when translated into Indonesian and English. The Banyumasan dialect of Javanese is characterized by krama (polite register), idiomatic expressions, and cultural depth, which are difficult for machine translation systems to capture. This paper explores translation strategies in AI-assisted translation of Banyumasan Wayang Kulit dialogues, focusing on the script 'Bawor Jadi Ratu' performed by Ki Kabul Idamanto. Dialogues were translated into Indonesian and English using AI tools (Google Translate, ChatGPT), then compared with human-translated versions. Results show that AI often produces literal and redundant outputs, misinterpreting cultural expressions. Human translators apply strategies such as cultural adaptation, modulation, and equivalence to produce culturally acceptable translations. This study proposes a hybrid approach where AI provides drafts while human translators refine cultural and pragmatic nuances, ensuring both efficiency and accuracy.