Tag
CVSS-X is a large-scale synthetic speech-to-speech translation corpus extending CVSS to enable translation from English into 28 languages, with over 16,000 hours of parallel speech pairs for bidirectional research.
This paper investigates how politeness and impoliteness in user prompts affect LLM responses across three languages and five major models, finding that politeness effects are language- and model-dependent rather than universal. The authors release PLUM, a multilingual corpus of 1,500 human-validated prompts with politeness annotations, and assess response quality using eight factors.