Translating Classical Poetry into Modern Prose

2026-06-01

AI Summary

This study addresses the challenges of translating classical Telugu poetry from the 13th to 17th centuries into modern Telugu and English prose by introducing Padyam2Gadyam, the first bilingual parallel dataset comprising 600 poems alongside their human-verified translations. Leveraging this resource, the authors evaluate the generative translation capabilities of five prominent large language models (LLMs), combining human validation with qualitative analysis to expose the limitations of current machine translation evaluation metrics in literary contexts. The experiments demonstrate that existing LLMs still exhibit substantial room for improvement in translating poetic texts into both target languages. This work establishes the first benchmark dataset and systematic evaluation framework for the machine translation of classical literary works, particularly within the understudied domain of Telugu literature.

Abstract

We introduce Padyam2Gadyam, a dataset for the task of poem-to-prose translation from 13th-17th Century Telugu Classical Poetry to contemporary Telugu and English prose. The dataset consists of 600 poems and their human-verified Telugu and English prose translations. We evaluated 5 contemporary Large Language Models (LLMs) on their ability to do poem-to-prose translation into Telugu and English. Our results indicate that while there are differences across LLMs, their overall performance leave a large room for improvement in both languages. Through qualitative analysis, we discuss the the capabilities and limitations of contemporary MT evaluation approaches for this task.

Problem

• poem-to-prose translation
• Classical Telugu Poetry
• Machine Translation
• Large Language Models
• MT evaluation

Innovation

• poem-to-prose translation
• classical Telugu poetry
• Padyam2Gadyam dataset
• large language models
• machine translation evaluation