Camille Bertrand-Sittler

If you translate these files every year do you keep the previous SDLXLIFF files from the translations created at the time?  If you do then the best solution is to simply Perfect Match and don't align at all.  You also have your TM from the previous translations, why don't you use that?  There is nothing wrong with your approach, I just wondered.

On the alignment... a few questions:

1. what's the source file format?
2. have you got customised segmentation rules?
3. do the source and target have different tag structures?
4. did you use the same translation tool to handle this as the previous times?
5. can we see the source and target files?

Hello Paul,

Thanks for your quick answer.

I'm sorry if I wasn't clear. Our department starts using Studio two years ago. Before that, we would translate directly in FrameMaker. So we've aligned the documents to start somewhere. Last year, we've confirmed all segments manually to have a memory. But this year, the same issue with the tags appears and I have to confirm half of my document manually only because of the tags.

1. The source file format is .mif.

2. No, I haven't customised segmentation rules.

3. Yes, the source and target have the same tag structure.

4. Yes, I used Studio last year for the previous user guide.

5. I must check if I can show the files because of confidentiality issues. 

Does Perfect Match will bypass the tag issue?

To start from behind: no, PerfectMatch will not resolve the tag issue. Have you updated Studio between the last translation and the current one? Has the MIF version maybe changed?

If you updated Studio or the MIF version has changed, please—if possible—try to use the same versions as last year.

No, we haven't updated Studio nor FrameMaker (MIF version). We have been using Studio 2019 since the beginning.

This is then indeed strange, as the translation memory from the manually confirmed segments in last run should have same tags. Or were the font settings/layout of FM files adapted/changed?

If you set the tag view to full tag text instead of tag ID, what do the tags say? Can you post a screenshot of the same segment as above, but with full tag text?

This is then indeed strange, as the translation memory from the manually confirmed segments in last run should have same tags. Or were the font settings/layout of FM files adapted/changed?

If you set the tag view to full tag text instead of tag ID, what do the tags say? Can you post a screenshot of the same segment as above, but with full tag text?

No, the font settings/layout of FM files weren't changed/adapted.

The tags don't say much: 

Well, the tags would say much more, if you would create a screenshot from the actual segment in the editor and not in the Translation memory view. Here you are showing tag id on the left and partial tag text on the right. This is not enough to find out, why the change occurred.

Full tag text view:

You can clearly read, what the tags do.

Looking at your screenshot I see, that on the left (previously translated segment) the complete segment has been manually formatted, while the current segment has just some internal formatting, while the whole segment has a style assigned.

Which tool has been used for the alignment?

In any case: I would confirm all the segments manually again. Then I would create a completely NEW TM and update it ONLY with the newly translated files to make sure, that no tags are enclosing the whole segment, as you see on the left. The view on the right is the view I would expect from any properly formatted file.

Also I would change the view setting in File -> Options -> Editor from "Show formating" to "Show only tags, do not show formatting". This way you have the best possible control over tags. Viewing formatting makes many people think, that they can use the formatting buttons in Studio, what leads to many formatting problems. In FM like in InDesign text will most probably not be made simply "BOLD", but a different font will be used here. However, when you just press "BOLD" from the ribbon or use the shortcut for bold, no font change will be applied in Studio. Studio will just apply bold to the main font in the paragraph, but such a font variation will most probably not exist. Such tags will also never be applied automatically next time, as the bold tags on the target side will not correspond with the proper tags in the source.

This is a screen shot of a segment in the editor with the Translation memory view with full tag text and the same with partial tag text. The segment comes from the memory and not from an aligned document. (I've made the alignment with Studio.) So it's supposed to be a properly formatted file. As you can see, the only difference is the last tag.

Thank you. This screens do not show the same text part as your previous screens.

Here you have tags 10555 and 10561 around text parts, in your new screenshots such tags do not exist.

Tag 10558 can represent footnote or line break. It's really hard to tell without seeing the files.

The screenshot above clearly shows, that the segment has not been properly formatted. In a properly formatted segment tags 10555 and 10561 should not appear at all. These show extra formatting, applied on top of the paragraph style here.

Sorry, I wanted to show segments that come from the memory and not from the alignment. Here is a screen shot of the segment in the editor and one of the sentence in FM, if it can help.

No need to be sorry, I am trying to find out, what's happening and this is why I have so many questions.

So, it looks like within the previous translation, showing in your TM view, segments with missing tags have been confirmed. On the left, in source, you see tags surrounding the text, while on the target side these tags are missing.

There are multiple reasons for that. First: Studio does NOT show tags by default, just formatting. So when the translator seen the segments last time, there was no visible difference between the segment in source and target. Bold was bold and the font looked identical or at least similar. Second: Studio does NOT report missing formatting tags by default. So there were no errors in the process and no one noticed missing tags.

Solution: confirm all the segments again, create a new TM and import only the last translation there. Next time the process should work as expected.

Possible solution to make it faster: remove penalty for missing formatting and different formatting within your current project. Use pretranslate, this should insert all segments as 100%, even with different or missing tags. When done, revert these settings to default. Go to verification settings, tag verifier and change the settings to not ignore formatting tags and not to report spacing around tags. Run the verification for tags, repair any possible errors. When done, create a new TM, attach it to this project, remove the old one. Update the new TM with the new project and keep the project for future reference.

I've removed the penalty for missing or different formatting penalty, as you can see in the screen shot :

So, why do I still have some 99% segments after the pretranslation with missing tags? Is there another setting that I must adjust?

Unfortunately not. There is a tag missing, which is not formatting. I'm afraid you will need to manually confirm all segments.