Pretranslation of Worldserver files not possible

Hi,

I have reported this before but I believe not through this channel and without any outcome so giving it a try here, maybe someone has found a solution by now:-) When we try to pretranslate files in packages downloaded from Worldserver with local TMs, the target segments remain empty. The translations are taken into account during analysis but we just cannot get them into the files. Does anyone know how we can get this done?

thanks,

Annemieke

Parents
  • My colleagues found out that it depends on how you download the files... The Export function produces workable wsxz packages that apparently have some limitations. If you go into the project and choose Translate > In SDL Trados Studio, you can save the wsxz as well but these files do allow pretranslation. Mystery solved:-)
  • I yayed to early, it does not work the same for all file types:-( Although the sdlxliff format pretranslates almost ok (matches are applied and the source is copied to the target segments, except for tagged segments because they remain empty).
    Some Worldservers produce xlf files and they have undesired behaviour. After downloading the files and simply opening the package (not doing anything else), only some 100% matches from Worldserver are applied, the rest of the segments remain empty. Only when putting the cursor in a segment, it is populated. After applying our own template and pretranslating, the 100% and fuzzy matches appear but for none of the remaining segments the source is copied to the target side. Does this have to do with limited support for xlf files and is there something we can do to improve this behaviour?
  • Hi Annemieke,

    I'm not sure I follow you here. Is this a Studio pretranslation issue ir a WorldServer pretranslation issue? Can you share something with me perhaps that helps me step through the issue you have reported?

    Thanks

    Paul

    Paul Filkin | RWS Group

    ________________________
    Design your own training!

    You've done the courses and still need to go a little further, or still not clear? 
    Tell us what you need in our Community Solutions Hub

  • Sorry Paul, I'll try to clarify:-)

    I think it is both a Studio and Worldserver issue but the problem I would like to address is the Studio one. All downloaded packages contain Worldserver project TMs so basically I do not care if Worldserver does not pretranslate correctly, I can do it myself in Studio. When referring to 'the Worldserver' by the way I mean a Worldserver implementation.

    What we do is that we download wsxz packages from Worldserver. In our case the files in those packages can be sdlxliff or xlf files and they usually come in partially pretranslated (Matches are seemingly randomly put in the segments or not, even within a file. It might depend on the Worldserver or the native file type but I have not found a pattern yet). When using packages that have 'native' sdlxliff files, pretranslation in Studio generally works fine. Also, the source is copied to the target side when there is no TM match BUT if the source segment contains tags, the source is not copied to the target side. Packages with xlf files are a different story though....

    I had a package with xlf files yesterday which I simply opened in Studio, not doing any file prepping. Most segments were empty when I opened the files, but once I put my cursor in a segment a 100% match or fuzzy match popped up. Obviously Worldserver did not pretranslate correctly but when I would run a Pretranslate batch task in Studio, the fuzzy and 100% matches would be populated from the attached Worldserver TMs. The segments for which there was no TM match remained empty, no matter what setting I chose. I then started prepping the files in Studio: I applied my project template with the plugin to add my own settings, TMs and termbases (the Worldserver TMs were also kept for leverage). Next, I ran a custom batch task that analysed and pretranslated the files. The 100% and fuzzy matches were pretranslated but still the remaining segments were empty. It seems impossible to copy the source text to the target side.

    In an attempt to fix this before sending out the files, I tried to use the Segment Actions plugin and chose the action 'Target same as source' for 'Not translated' segments. The result was not really what I wanted as some (not all) fuzzy and 100% matches were also reverted to English, even though they had the Draft status. I thought I probably chose the wrong setting and tried again (after repretranslating) but now chose 'Target same as source' for 'Empty segments'. Unfortunately the same thing happened. It all doesn't make any sense to me:-)

    I will send you a package with xlf files by email and hopefully you can figure out what is going wrong and how we can tackle this!
  • Hello Paul and Annemieke,

    Did you find what caused this problem with pretranslation?

    If so, I would be very interested in reading it, because we have the same issue with WorldServer packages in my team. Sometimes, pre-translation just won't apply to some segments, be it TM or Machine Translation. We did not find any recurring pattern other than the fact that most of the time they include tags.

    I can also provide you with other files to test for this very problem.

    Can you please keep me posted?

    Many thanks,
    Anne-Laure
Reply Children