Expensive Memories

As you may already know, since the version 2012 SCHEMA ST4 has turned into more than a mere content management system. With the DocuWeb Translator a Translation Memory System (TMS) is at your disposal. Using it single nodes can be translated directly in the system. In the process earlier translations in the system are searched through and suggested, when a certain amount of matches is found – the main principle of a TMS.

ST4 DocuWeb Translator

Such a feature is a big advantage when it is possible to have translations carried out by foreign subsidiaries or when single nodes need to be retranslated on short notice without having to initiate an external translation process. You save export and import of XML files, having to deal with external translation tools and keeping track of open translations. Up to now one disadvantage had existed: switching from DocuWeb Translator to an external translation tool or vice versa. Sometimes having the translations carried out externally, and internally in alternation had been problematic as external tools did not know about the translations in the other system. Thus the big advantage of translation memories is lost.

SCHEMA ST4 is able to extract the translation segments from nodes and create a translation memory for the DocuWeb Translator based on these segments. In order to carry out an alignment it must only be ensured that the translations in the respective target languages are up-to-date. This is, however, not the case when a node in ST4 needs to be translated on short term that has before been translated in an external TMS.

The Exchange of Translation Memories

In the version 2012 R2 of SCHEMA ST4 there is a new function to help you keep your translation memory in the DocuWeb Translator and that of the service provider synchronized: a TMX export and import. Translation Memory eXchange (TMX) is an open data format for exchanging data between different translation programs. It is based on XML and displays translation memory data (i.e. especially the language pairs) with various additional information [1].

Either from a subtree or from the entire database SCHEMA ST4 creates a TMX file which can be provided to the translator as base for a translation. He can import this bilingual TMX file into his translation tool (e.g. Across or Trados) and thus is offered suggestions for the segments that were translated with the DocuWeb Translator or assigned by alignment. In return you can request the translation memory from the translator as a TMX file and import it into SCHEMA ST4 so that the translations carried out externally are available in the DocuWeb Translator.

Not Perfect but Cost-Effective Nevertheless

TMX is a standardized exchange format but the mapping precision of XML segments is not defined in it. Therefore it might be possible that – depending on the translation tool used and the settings – no 100% matches are achieved for some segments. Especially for inline elements (formatted text, index entries …) the coding of the tools on the market differs greatly. For example inline elements are not represented in the translation memory, the mere existence of an inline element is mentioned without identifying it, or inline elements are partially coded in a proprietary format which leads to information loss during exchange with other systems. Automatically the rate of matches in translations gets worse. Also context relations like the previous or following segment are not represented in the TMX which in case of several matches makes it necessary to select and check manually.

Still you should always demand the translation memory for a translation in the TMX format from your translation service supplier. Only this gives you the opportunity to switch to another tool or service supplier without having to rebuild the valuable translation memory.

[1] http://en.wikipedia.org/wiki/Translation_Memory_eXchange

Leave a Reply

Up ↑

%d bloggers like this: