SDL Trados

CAT: Computer Assisted Translation
Consistency thanks to SDL Trados Studio


When working on our professional translations we use software by SDL Trados, which stores each sentence together with its translation as a pair in a database – the so-called translation memory. When translating, all our internal and external translators have access to this database and therefore to all the previous translations. The SDL Trados software detects any time a sentence, a headline or a slogan has already been translated and shows the translator the existing translation. To decide whether this also fits the current professional translation and if it should be used, the context of the previous translation can also be seen.


The translation database can also be searched like a glossary, which enables the translator to check how a specialist term has been translated previously for a certain customer.


In the following we provide some important information on translation databases:


What is a translation memory?

A translation memory is a language database which continually grows and learns from the translator.

All content that has already been translated is collected in the translation memory in the form of language pairs of the source and target text, which are termed translation units. Where appropriate, the collected content is reused. A sentence therefore never needs to be translated twice. The more content is stored in a translation memory, the quicker translators can work. This enables them to process a larger number of projects and their outputs increase.


How does a translation memory work?

Using SDL Trados Studio software, the translator opens the source text and applies the translation memory so that all "100% matches" (identical segments) or "fuzzy matches" (similar but not identical segments) are extracted in the text in a matter of seconds and transferred to the target file.

When working on the text, the translator can either accept the "matches" from the translation memory, or overwrite them with new alternative translations. When manually updating the translation, a translation unit is stored in the translation memory both to be reproduced in future texts and also to translate repetitions within the current text. In a similar way, the segments of the target file which contain no "matches" are manually translated and automatically recorded in the translation memory.


When is a translation memory used?

Every time content is transferred from one language to another, translation memories should be used. Documents that contain numerous repetitions show the advantage of translation memories most clearly.

Translation memories are also of great benefit if content taken out of context needs to be translated. Increasingly more companies rely on content management systems (CMS) to manage their data. Using a CMS individual sections of texts – compared with complete documents – can be produced/edited and then published in a multitude of different formats. Using a translation memory this process can be quicker and more consistent.

Even if no translation memory is used the translators can extract text from the source file and focus on the localisation of the text without having to worry about tags (tags are control codes and are used to format texts). The code of a HTML file, for instance, is completely hidden so that no time is lost looking in irrelevant sections of text for the text that is to be translated.


What advantages does an SDL translation memory hold for businesses?

  • It enables projects to be completed more quickly.
  • It increases the consistency and quality of translations and therefore customer satisfaction.
  • It accelerates the localisation process and reduces the overall costs.

In what way is a translation memory tool different to a terminology tool?

A translation memory tool saves text segments as translation units in the form of language pairs consisting of the source and target text. A segment can consist of a sentence or a section. In comparison, a terminology tool is a database, which contains a list of terms in several languages and rules on their use. Terminology is generally used in conjunction with a translation memory.


In what way is translation memory software different to machine translation?

In machine translation a document is automatically translated without human input.

This type of tool is quick, however produces poor quality translations because a machine can neither read between the lines nor analyse the context of linguistic expressions. Accordingly, the quality and accuracy of machine translations is generally between 50% and 70%. It is not, therefore, advisable to forward the rough draft of a machine translation directly to customers. Furthermore, machine translation is only available for a limited number of supported languages.

By using translation memory software such as SDL Trados Studio on the other hand, the number of supported languages increases ad infinitum, and the actual translation work is undertaken by a professional translator. A translation memory provides 100% matches and fuzzy matches from a database, in which pre-translated texts are stored. The result is a more efficient translation process and more consistent and higher quality results.


Are SDL translation memories compatible with many different file formats?

Yes, SDL Trados Studio is compatible with a multitude of programs for you to create the content. Supported formats include Microsoft Office (2000 – 2003, 2007, 2010, 2013), OpenOffice, RTF, Tab delimited, HTML and XML.

New formats are also supported, for example, Adobe FrameMaker and Adobe InDesign, Adobe InDesign Markup Language (IDML) and InCopy Markup Language (ICML), PDF, XLIFF and XML versions such as DITA, Docbook and W3C ITS.


Further information can be found at