Unfaithful beauties and invisible errors

Right and wrong in translation

Abstract

Translation errors – whether committed by humans or machines – can be funny, irritating, or downright dangerous. So how can they be avoided? And, given the way meaning tends to shift across languages and cultures, is there such a thing as a single correct translation? Researchers at Heidelberg University’s Institute for Translation and Interpreting are trying to answer these questions. Their work focuses on Machine Translation technologies that are used in popular systems such as Google Translate or DeepL. Predicting and detecting potential Machine Translation errors is key to improving the technology and, ultimately, striking a balance between fully automated translation processes and those that require human intervention.