Hello Caroline PALLEGOIX ,

From the example you have provided-

- you have saved three different versions (source) translations for the original source text- As these are near identical (accent/s/en endings) these are treated as three different valid translation segments and will appear therefore as the 99% matches.

If you change the settings to Overwrite existing translation units 

...these will be overwritten if you go back and change the translation.

The (duplicate) segments already in the TM can be removed in the TM maintenance view 

Does this help you?

Hi Caroline,

The options you're showing are for batch tasks, but if I understand your question correctly this happens during interactive translation, i.e., when you confirm a segment manually, right? In that case, these settings have no effect.

How is your TM added to your project? As a regular TM or as AnyTM? I only see this behavior when I'm using AnyTM.

I forgot to add that when this happens, a quick solution is to run an "Update Main TM" batch task with "Keep most recent translation units" or "Overwrite existing translation units". That removes the duplicates and keeps only the TU that is currently in the document.

Hello Steven,

Thank you very much for your reply!

I understand of course that changing the option doesn't make the "duplicate" segments disappear from the TM.

Unfortunately, your advice doesn't solve my problem: as I said, with both "Merge" and "Overwrite", when I translate a new segment (after checking one of the two options) and then correct it (to be sure it merged/replaced it), it still creates a new segment in the TM, so both right+wrong versions are saved.

As Nora just asked, it's only with AnyTM. Would it mean this function can't work with AnyTM?

Thanks for your help

Caroline PALLEGOIX said:

Would it mean this function can't work with AnyTM?

It does, but these settings only affect batch tasks, not interactive translation.

Hi Nora,

Thank you very much for your suggestion!

Indeed, it might only be with AnyTM... I didn't pay attention to that. But in my example, it's AnyTM indeed. Does it mean this Merge function doesn't work with AnyTM?

I don't really master the Update main TM function, so I'll try this but it's not something I'm used to do, so maybe that would be the solution! I guess I have to select the "problematic" TM as main TM first then.

Yes, if the language pair is not an issue, then definitely add the TM as a regular TM, not as AnyTM, and that will take care of the problem. Just to clarify: a TM added via AnyTM can also be a Main TM. 

To run an Update Main TM task, simply go to Batch Tasks and select Update Main Translation Memories. It takes a few seconds and it can "clean up" those duplicates.

Nora,

Thanks a lot. How do I select a TM as Main?

I have a few enabled and used, the first one being the client one which I don't want to update. What should I do to be sure the software updates the right one?

Hi Caroline,

Several TMs can be added as a "Main TM" to a project. The other type of TM is a "Project TM", which is basically an extract from a TM that contains only segments relevant for the project at hand. I can't tell for sure because your screenshot is cut-off on the left, but it looks like the first TM has been added as a Project TM, while the other four are Main TMs.

To make sure a TM is updated with every new segment you confirm, just make sure the "Update" checkbox is checked.

Perfect Nora,

It looks like it deleted all the duplicates. It still creates them during translation/correction, but at least I have this solution... I often use AnyTM as my TM is for Germany to France and I sometimes get project from Austria to France, then Austria to Belgium, then Switzerland to Switzerland... it would be crazy to have so many combinations and the respective TMs wouldn't be that helpful.

Thank you so much for your time and efficiency.