An Empirical Study on the Robustness of Massively Multilingual Neural Machine Translation

TL;DR

This paper empirically examines the robustness of massively multilingual neural machine translation for Indonesian-Chinese, introducing a new benchmark dataset to evaluate translation quality under various natural noise conditions.

Contribution

It presents a novel robustness evaluation benchmark dataset for Indonesian-Chinese translation and analyzes the impact of noise and model size on translation robustness.

Findings

Correlation between error types and noise presence

Model size influences robustness and error patterns

Automatic evaluation metrics correlate with human judgments

Abstract

Massively multilingual neural machine translation (MMNMT) has been proven to enhance the translation quality of low-resource languages. In this paper, we empirically investigate the translation robustness of Indonesian-Chinese translation in the face of various naturally occurring noise. To assess this, we create a robustness evaluation benchmark dataset for Indonesian-Chinese translation. This dataset is automatically translated into Chinese using four NLLB-200 models of different sizes. We conduct both automatic and human evaluations. Our in-depth analysis reveal the correlations between translation error types and the types of noise present, how these correlations change across different model sizes, and the relationships between automatic evaluation indicators and human evaluation indicators. The dataset is publicly available at https://github.com/tjunlp-lab/ID-ZH-MTRobustEval.