Localization knowledge and technologies

 

Sometimes companies need to cheaply and quickly translate a large volume of text for their internal needs: for example, in order to learn the general content of a text, find out whether it is relevant and select a small part of it for real translation and publication. In this case, machine translation can be used. But one should keep in mind that it only gives acceptable results on some very specific texts and cannot be used whenever accuracy is critical, all the more so where the resulting text must read well. Anyone who has ever used the Google Translate online system knows about the advantages and disadvantages of machine translation. It comes handy when you need to know immediately what the text is about, but there is always a possibility of errors, and the style is often fodder for jokes.

 

 

Usually a company has to handle many similar materials: for example, product catalogs where the same descriptions repeat with small variations. When the materials are localized, it is important that repeated text is always translated in the same way. It is impossible, however, to memorize all the past translations, even if there is only one translator working on the project. Considering that the work on large-scale projects is usually shared among several translators and editors, it is even more difficult to preserve the consistency of translation, terminology and style. Besides, translating again a text that has been already translated means wasting time and making the job cost more than necessary. Special software that aids the translator (Computer Assisted Translation, or CAT) helps resolve this problem. It memorizes the fragments that have been translated and retrieves them automatically.

 

 

When the translation is complete, another problem arises: how can you evaluate its quality and adequacy, especially if the translation was performed into several languages that you don’t know? To resolve this, the methodology of Linguistic Quality Assurance (LQA) was developed. This means that a small fragment of the translation is analyzed and the errors identified. The errors are then categorized by severity and type, and a translation quality score is calculated.

Logrus has an extensive experience in using, researching and developing advanced localization technologies. Experts from Logrus share their knowledge at conferences and workshops, as well as on our website, in the Articles section.
Since January 2007, Logrus has published Professional Translation, a journal on modern localization technologies and trends.