Subtask: MT Test Suites. "Mind the Gap: Exposing LLM Translation Blind Spots"
We are submitting our test suites for WMT 2026 shared task where our data is multilingual and includes German, Polish, and Arabic. To give a richer human view of how different MT systems (from WMT 2026 submissions) performed, espacially how they are wrong when they are wrong, we call for native speakers of these three languages.
Please reach out if you are interested. The expected outcome is co-writing the finding papers together on this shared task we are attending.
Kind regards,
Lifeng Han