Exploring the Correlation between Human and Machine Evaluation of Simultaneous Speech Translation

TL;DR

This study evaluates the reliability of automatic metrics, especially GPT models, in assessing the quality of simultaneous speech translation by analyzing their correlation with human judgments of translation faithfulness.

Contribution

It demonstrates that GPT-3.5 with direct prompting correlates strongly with human assessments in evaluating translation accuracy without reference texts.

Findings

GPT-3.5 shows high correlation with human judgment

Context window size affects evaluation accuracy

Semantic similarity metrics are effective for assessment

Abstract

Assessing the performance of interpreting services is a complex task, given the nuanced nature of spoken language translation, the strategies that interpreters apply, and the diverse expectations of users. The complexity of this task become even more pronounced when automated evaluation methods are applied. This is particularly true because interpreted texts exhibit less linearity between the source and target languages due to the strategies employed by the interpreter. This study aims to assess the reliability of automatic metrics in evaluating simultaneous interpretations by analyzing their correlation with human evaluations. We focus on a particular feature of interpretation quality, namely translation accuracy or faithfulness. As a benchmark we use human assessments performed by language experts, and evaluate how well sentence embeddings and Large Language Models correlate with them. We quantify semantic similarity between the source and translated texts without relying on a reference translation. The results suggest GPT models, particularly GPT-3.5 with direct prompting, demonstrate the strongest correlation with human judgment in terms of semantic similarity between source and target texts, even when evaluating short textual segments. Additionally, the study reveals that the size of the context window has a notable impact on this correlation.