Comments on Translation Tribulations: ODT files in translation environment tools

A couple of comments: Tagging in .odt files varie...

2012-01-08T13:59:38.538+01:00

A couple of comments:

Tagging in .odt files varies widely depending upon the application used to create or convert the .odt file (or .odp or .ods file, which are essentially the same format). There are wide differences between OpenOffice.org vs. LibreOffice, and even between different versions of LibreOffice. I opened your test file and closed it again in Abiword and LibreOffice and compared the resulting tag counts as displayed in OmegaT on my system. The results were:

File as supplied by you: 24 tags
After saving in Abiword: 16 tags
After saving in LibreOffice v. 3.3.2: 34 tags

That's both a very large number of tags, and a very wide variation, but neither is attributable to OmegaT as such.

One consequence of this is that for comparison of tag display in different CAT tools, great care must be taken to use the same file. Even opening the file in another application and closing it again may be sufficient to reduce or increase the tag count in the CAT tool.

Segmentation: in OmegaT, segmentation is determined by the segmentation rules, which in OmegaT are highly customizable. If you wanted to, you could probably adjust the segmentation rules to produce what you describe here as the "correct" segmentation, and possibly even to keep leading/trailing segment tags out of the displayed segment, thus reducing the tag count (though probably with a trade-off in format control).

User customization like this is very consistent with the OmegaT philosophy. To say that OmegaT segments in a certain way is therefore a little like saying "Microsoft Word uses Ariel" because that happens to be the default font in Word. The ease of such customization (requiring knowledge of regex) is another matter of course, but that's OmegaT for you: a powerful tool in skilled hands.

Keep up the good work,

Marc P.

Hi Kevin, Nice post as usual. Thought I would ad...

2012-01-03T13:31:54.104+01:00

Hi Kevin,

Nice post as usual. Thought I would add Studio 2011 here too, just to complete the picture. We have support for *.ODP and *.ODS as well.

http://youtu.be/1RurEF2fEXs

I think this handles the file correctly as you wanted.

Regards

Paul

I created an ODT file based on the formatting you ...

2012-01-03T10:33:10.625+01:00

I created an ODT file based on the formatting you show, then I imported it into DVX2. The result looks pretty clean. Of course this is only one scenario which may not apply to DOC files converted to ODT. I haven't tried it on DVX1.

Milos, sentence level segmentation is implemented....

2012-01-03T09:52:40.327+01:00

Milos, sentence level segmentation is implemented. The problem here is the use of a tag to represent a space as you can see from the various screen shots. Also, I believe there are other potential problems if you use the "native" ODF support because of how ODF often comes into play in the translation environment. I got onto this particular test in the first place because Open Office was used to convert MS Word DOC files, and this often causes format difficulties. This led me to look at the differences between various tools in handling ODT. If you follow the workflow DOC > ODT > DOC because your tool can't support a DOC file, then you are best advised to take the path which will damage the structure of the original DOC the least. (Or better yet, skip Open Office entirely and just work with DOCX if you have access to an appropriate version of MS Office.)

"Nine segments where there should have been t...

2012-01-02T19:09:50.442+01:00

"Nine segments where there should have been ten."
You have to enable sentence level segmenting in the project properties dialog (best to do it while creating a project). I've got 10 segments in OmegaT with the same document.

Also I don't have such a tag-mess, just 6 paired tags... In case you have created your document using MS Word, it has still issues regarding the implementation of ODF, so it's better to use an application with native ODF support.