Large Language Models for Classical Chinese Poetry Translation: Benchmarking, Evaluating, and Improving

TL;DR

This paper benchmarks, evaluates, and improves large language models for translating classical Chinese poetry, addressing challenges of cultural fidelity and poetic elegance, and introduces a retrieval-augmented method that outperforms existing approaches.

Contribution

It introduces the PoetMT benchmark, a GPT-4 based evaluation metric, and proposes a Retrieval-Augmented Translation method to enhance LLMs' performance in Chinese poetry translation.

Findings

Existing LLMs underperform on classical Chinese poetry translation.

The RAT method improves translation quality across multiple metrics.

Human evaluation confirms RAT's superior performance.

Abstract

Different from the traditional translation tasks, classical Chinese poetry translation requires both adequacy and fluency in translating culturally and historically significant content and linguistic poetic elegance. Large language models (LLMs) with impressive multilingual capabilities may bring a ray of hope to achieve this extreme translation demand. This paper first introduces a suitable benchmark (PoetMT) where each Chinese poetry has a recognized elegant translation. Meanwhile, we propose a new metric based on GPT-4 to evaluate the extent to which current LLMs can meet these demands. Our empirical evaluation reveals that the existing LLMs fall short in the challenging task. Hence, we propose a Retrieval-Augmented Machine Translation (RAT) method which incorporates knowledge related to classical poetry for advancing the translation of Chinese Poetry in LLMs. Experimental results show that RAT consistently outperforms all comparison methods regarding wildly used BLEU, COMET, BLEURT, our proposed metric, and human evaluation.