I found this new cool feature that allows to translate multilingual files with TMs with different source languages.

- I create my TMs with the different source languages (en-jp, de-jp, zh-jp)

- I add a general EN-JP TM from previous translations to the project

- all TMs are enabled for updating

- I select the source language for the segments in the editor (if Studio does not recognize them automatically)

The result is that the segments for EN-JP get written to the en-jp TMs. But the DE-JP segments also get written to the general EN-JP TM, as are the zh-jp segments.

So they all go to their respective TM, created with AnyTM, but ALL of them also go to the general TM with the language pair of the how the project was set up. If the segments were marked correctly in this TM (with source languages DE, EN or ZH) this would not be so bad. But they all get marked as source language EN.

I would have expected that first, the segment pairs only go to the TMs with the correct language pair and secondly, if they all go to one TM, the language IDs should be correct.

Or am I missing something here?

The screenshot shows a TM I have used in a project EN-JP in addition to the AnyTMs for the respective languages.

My biggest concern here would be what kind of matches I get when I use this TM in the opposite language direction....

Did anyone ever test this kind of scenario?

thanks,

Angelika