Removing the Uncertainty with Qualified Machine Translation

by | Aug 16, 2014

Qualified machine translation is the easiest way to get the best out of machine translation (MT). While automatic translation technologies have been developed with giant leaps in the last decade, the main problem with machine translation is, not unlike the problems with human translation, uncertainty. The user can’t be sure how good the translation really is if he doesn’t speak the language. However, quality is still a focus and a concern, because the message won’t be delivered if the translation is bad. Despite this uncertainty, MT is very popular all around the world for “instant” translations, as well as for professional workflows that save money.

One reason behind the success is that it is easy to use machine translation. There are many free and paid machine translation services available. There are options for everyone, and the available translators differ in many aspects. For example, the language selections are machine translator specific. Roughly speaking, there are three types of MT when considering languages: 1) machine translators that are designed for only one language pair, 2) those that support a few languages, and 3) those that try to translate between as many languages as possible.

All these machine translator types have their pros and cons, and some of them are more suitable in some situations than in others. To make it even more complex, machine translation quality differs a lot also on the sentence level. The MT technology and corpus ( have direct links to both the translation quality and the overall ability to translate every sentence. Machine translators can rarely translate every word or sentence. The problem is that it isn’t easy to see which one is the best option for the current needs for each sentence. With qualified machine translation this issue is solved automatically.

Qualified machine translation contains only the best available automatic translations. Without going into technical details, machine translation qualifiers compare different machine translations, evaluates the quality and selects the best translation separately for each sentence. This whole process is performed automatically to ensure the best possible machine translation quality.

Also raw, unprocessed machine translation can produce good quality, there’s no doubt about it. However, as compared to qualified MT, it contains a much higher risk. Or actually, the risk is on the user. The risk of getting a bad translation, which changes the message of the text, may not be worth it. Overall when using machine translation the user is always responsible for the translation. It is the users’ interests to get as good machine translations as possible.

This is a guest post by Multilizer; a technology company specialized in developing professional localization and translation tools for both individuals and companies. Each month over 120 million words are translated with Multilizer products in more than 150 countries. Lately, the company won the European Innovation Award with its new MT-Qualifier technology.

Want cool localization techniques straight to your email?

Want cool localization techniques straight to your email?

Keep up on the latest in localization management techniques with Wordbee.

You have Successfully Subscribed!