Using OmegaT to translate in the format of Trados (TTX)

From B-Ob8ungen
Jump to: navigation, search

In November 2011 I was asked to do an English-German translation, but the input and output files had to be in the format used by the market leader Trados known as TTX (*.ttx). At that time I failed. In July 2012 the same people in the Netherlands once again asked me to do such a job and a quick search in the Internet made me believe that it could be done.

Converting files

My first research took me to a page called Translating with Rainbow and OmegaT. I remembered something of that kind, but also that this type of solution only worked under Windows (WinXP being installed on my notebook). I found a folder with an *.exe file in it that opened "Okapi-Rainbow". It did not look like the screenshots of the tutorial, but the sentence "Note also that this documentation may not be always synchronized with OmegaT releases, so you may find differences between the descriptions given here and the latest version of the application made me believe that I just had to follow the logic.

Indeed files were created in the OmegaT format *.xlf and I started to translate them in the usual way. There was quite some formatting of the texts (meaning a large number of "tags"), but I could leave them untouched since English and German have more a less the same structure of sentences.

However, once I finished the translation I did not find a way to convert the files that were created back from '.xlf to '.ttx. The files in the target folder that I created did not have any German text but only English text. The TM (translation memory seemed to be correct), but the people in the Netherlands (just like me) had no clue how to make use of it.

Merging files

The next step was to install the Okapi-Rainbow program thoroughly and follow the instructions of the tutorial. This time the file for later merging the text was created. In this new project I could use the TM. There was almost no problem for the first 100 our 300 segments. The segments without any tags were included as 100% matches and even the ones with two or three tags could be copied from the window with "fuzzy matches" to be included.

The task became more difficult in the remaining segments that showed some heavy formatting. The matches that OmegaT offered were unrelated in many cases, because the formatting (like fonts etc.) was often the criteria for the match and not the text. Indeed, it became a difficult task to find the part of text that had to be translated. Short sentences such as "Use the Exit key to ignore changes" were split in many parts making it very difficult to follow and it was certainly easier to "translate" single words, rather than copying them. If longer phrases were hidden behind the formatting I had to use the search function to find my previous translation in the memory.

I made the mistake to make no copy of the project on a USB stick. The main work was done on the computer with Ubuntu 11.10 and OmegaT 2.5... being installed. Without having checked whether the tags were correct or not and without having created the target document (which should also create a batch file for merging the text and create a TTX file) I opened the project with OmegaT 2.3... being installed on the notebook that is running WinXP. I got an error message that did not prevent the project being opened with the result that all 314 segments showed up as not having been translated.

So I had to make the whole exercise again. This time I checked (under Ubuntu) if the tags were correct or not and opened the project on the notebook with OmegaT 2.6... now being installed. The batch file did not merge the file and resulted only in showing the file to be merged in the window of Okapi-Rainbow as being ready for merging. I chose the filter that I believed to be the correct one and created a TTX file. The size of this file was smaller than the *.xlf file (210 instead of 400kB). I sent the file to the Netherlands hoping that the people there could handle it in Trados, but I got the answer back that the file was empty, a statement that was partly contradicted by saying that a machine translation had produced more matches than my file. If empty it should not have produced any matches and, just like the many tags of formatting that had made my life difficult before (no proper match showing and the necessity to search for the previous translation) it may well have been that the heavy formatting prevented the matches from being shown.

The easier way

On the page on compatibility of OmegaT two other ways are described on how to manage Trados files:

  1. A script (for Windows only) was developed enabling OmegaT users to produce Trados uncleaned RTF files for delivery at the end of their translation stage. For details, see the "Exporting from OmegaT to uncleaned RTF" HOWTO.
  2. Trados TTX format is an XML-based format. An OmegaT plugin is now available by means of which this format can be handled in OmegaT.

The first alternative looks much too complicated for me. As far as the second alternative is concerned I first tried to "install" the plugins (being identified as the ones for Linux), but did not succeed with OmegaT 2.5... or OmegaT 2.6... under Ubuntu 11.10. I found a message that said I should be using a single file plugin (for all platforms). I succeeded to have this plugin work in OmegaT 2.6... under WinXP, but not under Ubuntu.

After all, I did not go through the whole project again a) because the people in the Netherlands had no time left, b) were not interested in the experiment (no help at all) and c) I had already wasted much time. There are more steps to consider, if you directly open TTX files in OmegaT. These steps are described at Okapi Filters Plugin for OmegaT.

In conclusion

  • If you are working on a Windows platform and have time and money install Trados
  • You could also try to make it work in a virtual box under Ubuntu
  • But unless you are working with "friends" who have some technical knowledge on a project that is not very urgent, keep your hands off projects in OmegaT trying to handle Trados files.