machine-translation-quality Machine translation can be fast and cheap. It is a valuable resource when a quick translation is needed. Billions of words have been translated automatically and the pace continues to accelerate. Many professional translators are afraid of losing their jobs to machines. The fear is oversized though. Man made translation with a real person taking responsibility of the quality doesn’t compete in the same league with a raw machine translation with unknown quality.

Raw machine translation is the best guess made by a machine. There’s no guarantee that it will be correct. However, machine translation is rarely totally flawless or completely useless. The quality is mostly good enough for understanding the core message. Although there are a few tricks to make sure you’ll get the best possible machine translation quality, the quality level itself isn’t a huge problem.

Uncertainty is the main reason why machine translation isn’t used efficiently. An ordinary user won’t know how good the automatic translation truly is. One extremely common use case is to rely on machine translation when one needs to understand a piece of foreign language text. The user would have to be able to read both the source and the target language to be able to evaluate the quality – but then he wouldn’t need the automatic translation in the first place.

Without real measurement, machine translation quality is more like a subjective perception than a cold fact, and it’s harming everyone. Here are three reasons why we should know the quality:

  1. It is possible to improve only when you know the current state. Without any facts of the quality level, no one knows if the translator is the best, the worst or something between. Measured quality enables benchmarking the competitors’ performance or monitoring the effects of the last update. Competitive setting creates motivation and incentive to become better.
  2. Users can choose the best option for their needs. Translation quality is not a simple or unambiguous concept. Machine translation quality varies between languages and text types even when using the same service. This is why there isn’t any machine translation service that would be the best for all situations. Depending on the situation the quality may become remarkable higher if one machine translator is chosen over another.
  3. Transparency increases. Machine translation service providers might be scared of people’s reactions to comprehensive quality scores and comparisons. But as long as the quality is unknown there will be a shadow over machine translation. People appreciate honest and open brands, and transparency makes machine translation both interesting and accessible. Automatic translation won’t ever fulfill all translation needs. Thos who need professional translation will buy it also in the future. But those you hesitate to even try machine translation could be encouraged to do so if they knew what they will get.


Machine translation quality scores Do you want to know how machine translators performed last month? Click here to check machine translation quality scores.