GALA is a fully representative, non-profit, international industry association for the translation, internationalization, localization, and globalization industry

27 November 2008


QA Check

Hi, I’m trying to get some advice or tips by the GALA community on an issue which we have for some time now, but were not able to find a solution so far. But I’m sure that we are not the only TSP facing this issue, so maybe you have a solution for us you want/can share.

What’s the issue now? We have a client in the medical devices industry, for whom it is absolutely necessary that the same sentence is ALWAYS translated the same way. Or an almost identical sentence keeps the major part of the previous translation and only the actual difference in the source file is adapted, leaving the rest of the sentence unchanged.

Okay, you might say, work with a TM tool and your problem solved.

Unfortunately that’s not that easy. As there are people working on the files and not machines, translators or proofreaders decide from time to time to change an existing sentence, although they are told not to do this.

Is there a tool for automatically checking a translated file if it contains deviations from the TM?

I have not found a tool yet that is capable of doing exactly that. There are QA tools checking inconsistencies of matches within a file or a bunch of files, but none of these tools is back-checking with the TM itself, it the consistency with the TM is maintained.

In a first step it would be nice to have this done for 100% matches, but of course if that can be expanded to high fuzzy matches, the tool would be even more appreciated.
If you have any ideas, please let me (us) know.

Many thanks,
Johannes

Johannes Purer
Puretrans

4 Responses to “QA Check”

  1. Yamagata-Europe said:

    Hi Johannes,

    You can use QA Distiller to check inconsistencies between TTX files and a TMX file.

    There’s a free trial at http://www.qa-distiller.com

    Regards,

    Thomas
    Yamagata Europe

  2. Stefan Gentz said:

    Hi Johannes,

    In SDL TRADOS there’s an easy to make sure (after translation) that all 100 % Matches are taken from the TM:
    Open SDL TRADOS Workbench.
    Open your Translation Memory.
    Open the “Translate Files” dialog (menu > “Tools” > “Translate…”).
    Drag and drop the translated (i.e. bilingual) TTX to the “Files to translate:” list box.
    Specify a “100% or higher match value”.
    In the “Update changed translations” group box, select “Update document”.
    Now click “Translate”.

    “Update document” means that the changes made to the translation units in the document are undone — the translations from the translation memory are used instead. That is, if you have specified a “100% or higher match value” and activated “Update document”, all existing 100% Matches in the TTX will be overwritten with the existing translation from the TM. Please note that this – of course – can only work if you are doing this with a TM version older than the TM. (That is, you need to make sure you are using a TM that does not include possible changes of 100% matches.)

    This is not exactly a smooth verifier and a little bit more brute force. But it’s a quick and reliable solution to make sure that the recycled 100% Match Tus are taken from the TM before translation started.

    Besides, the translation result will give you information about changed translation units. So, if there is e.g. Updated = 1 Segment, you know that here one segement was changed.
    Please also note, that this approach can of course only work properly if you and the translator are not working with a TM server but with file based TMs. A TM server szenario would mean that the existing TU in the TM would be updated with the changed translation once the translators set and closes the TU after changing it. You would have no chance to identify this, because any post translation verifyer would always find the (changed) TU to be identical in the TM (because it was already changed in the TM during translation).

    Kind regards,
    Stefan Gentz

  3. albatros said:

    Hi Johannes,

    we use a couple of QA tools in our translators group (EN>IT). Some of them are built into the TM programs themselves (e.g. in Trados or MemoQ), others are separate applications (e.g. XBench).

    However, I think you are right in saying that these tools do not allow you to check for deviations from the approved TM. So I suggest you adopt a slightly different approach.

    You could pretranslate all the files before handing them out. For that purpose, use a TM program that uses contextual matching, so 100% matches will be very reliable. Both Trados and MemoQ offer this feature and I’m sure that more and more tools on the market today are making it available.

    What’s really useful about this feature in this particular case is that you can “lock” the pretranslated segments, so your translators cannot modify them. I hope this helps.

    Roberto Savelli
    Albatros Soluzioni Linguistiche SRL
    English-Italian Translation and Localization Services
    Forlì, Italy
    http://www.albatrossolutions.com

  4. Chris Phillips said:

    Hi Johannes,

    I only just saw this post so it may be a bit late. I can help you with a script that will turn any list of 100% matches (or all 100% matches) you have into context matches so they will be locked in TagEditor.

    Regards
    Chris Phillips
    indesign4translators.com
    sweL10N.com

Post your opinion

You must be logged in to post a comment.