Abstract
Specialized translation relies on documentary and terminological resources, including corpora, which are particularly useful for terminology. However, their compilation and exploitation have several limitations: they require time, technical skills, and access to data that can be difficult to collect. This study examines the extent to which LLMs can assist specialized translators in finding equivalents from English to French.
We evaluate four proprietary models: GPT-4o, GPT-5.2, Claude Sonnet 4.5, and DeepSeek, across two specialized domains, Earth, Environmental and Planetary Sciences (EEPS) and Natural Language Processing (NLP). The experiment is based on 80 terms per domain and compares two prompting strategies: terminology and translation modes.
The results highlight clear differences between models and prompting strategies, with lesser differences observed across domains. Claude Sonnet 4.5 achieves the best results in the most favorable configuration, while DeepSeek is noted for its greater stability. Analysis of confidence estimates also shows that they are only a partial indicator of terminological accuracy.
Overall, the findings suggest that LLMs can be useful tools for specialized translators, but cannot, at this stage, replace specialized corpora. This research paves the way for future work on the real practical usefulness of LLMs for specialized translators in work and educational contexts.
Blogger's Review: This study showcases the potential of LLMs in specialized terminology translation but also reminds us that despite their impressive performance, traditional corpora remain essential for ensuring accuracy and professionalism in translations. Future research will help better understand the practical value of LLMs in specialized translation.