Convergences and Divergences between Automatic Assessment and Human Evaluation: Insights from Comparing ChatGPT-Generated Translation and Neural Machine Translation
This addresses the problem of unreliable automated evaluation for researchers and practitioners in machine translation, highlighting the need for human judgment, but it is incremental as it builds on existing comparative studies.
The study compared automated metrics and human evaluation for assessing translation quality from ChatGPT and neural machine translation systems, finding convergence in formal fidelity but divergence in semantic and pragmatic aspects, with automated metrics missing improvements from prompt engineering.
Large language models have demonstrated parallel and even superior translation performance compared to neural machine translation (NMT) systems. However, existing comparative studies between them mainly rely on automated metrics, raising questions into the feasibility of these metrics and their alignment with human judgment. The present study investigates the convergences and divergences between automated metrics and human evaluation in assessing the quality of machine translation from ChatGPT and three NMT systems. To perform automatic assessment, four automated metrics are employed, while human evaluation incorporates the DQF-MQM error typology and six rubrics. Notably, automatic assessment and human evaluation converge in measuring formal fidelity (e.g., error rates), but diverge when evaluating semantic and pragmatic fidelity, with automated metrics failing to capture the improvement of ChatGPT's translation brought by prompt engineering. These results underscore the indispensable role of human judgment in evaluating the performance of advanced translation tools at the current stage.