As I said before, I had already read the article, but I have to admit I focused on finding how I could do the merging, more than its consequences. Now that I know how to do it, I read the article again more thoroughly and even watched the video this time. A very illustrative, useful video, I should say. So now I know exactly how merging works and why it should be used carefully. Thank you very much for this very useful piece of information.

Now, I see it is not possible to actually remove segments in Trados Studio. I wonder if an update of the software could include such a function and whether removing segments in Studio could be a solution to the problem of empty lines in Word and more importantly, empty trans units or whatever code in XML or Excel files. I'd like to know what you think.

I think the solution is either enabling Studio to handle hard breaks in the bilingual files or to support a kind of virtual merging...

This is indeed what I (and many others afair) originally suggested quite a time ago. The current way of merging segments is very similar to a workaround I was using since source editing is there: copy the content of badly segmented source segments into the first segment, then translate this segment and insert a double space for all other segments, where I finally have always left the bad text parts, but always changed the status to sign-off rejected and locked them.

The big difference between the merging of the segments as it is now and my workaround is, that when segments are merged empty segments are created on both sides and this can very well destroy the file structure, while when editing source and leaving the segments with a double space keeps the file structure intact. So maybe a simple change in the method to just insert a space on both sides would result in a much more safe merging.