Abstract
How do machine translation systems—and, more broadly, artificial intelligence—handle untranslatable expressions? The primary benefit of focusing on untranslatable expressions is that it raises awareness of the complexity involved in transitioning to machine translation, as was the case at the European Court of Justice in Luxembourg, for example.
The project to create a European repository—a wiki-style database of untranslatable expressions—could make both machines and human translators smarter together. I will discuss the most recent approaches to establishing such a collaborative and open platform.