Armin Mutscheller 

This is the first time I have ever heard anyone ask this question in around a decade!  There was a feature in the old Trados Workbench for this but it was never implemented in Trados Studio as it was actually rarely used even in Trados. One of the reasons is terminology is rarely a 1:2:1 replacement so if you used synonyms in your termbase the result would be pot luck and even then if you had any homonyms the result could be completely wrong.

The fact I have now heard this twice since Trados Studio 2009 seems to vindicate this decision by the product teams.

I think if you consider it worthwhile then you should raise it as an idea here:  Trados Studio Ideas 

Yes, obviously the pre-translated terminology fragments would have to be checked to ensure plausability and correct grammar. Simple word equations rarely work as intended. But with a well-managed termbase that does not contain too many synonyms, and in which preferred/forbidden terms have been flagged, the effort of translating "around" correct, defined terminology and editing incorrect terms would likely be lower than post-editing a machine translation riddled with "best guess" terminology that would have to be validated by research. On a lexical level, it is MT that melts down to pot luck, unless the MT engine processes user-defined glossaries. That is why, in my opinion, terminology management has never been more important than it is today. 

Thank you for pointing me at the Ideas section! I will check it out. Although I doubt that RWS will want to be revive deprecated, "old-schoolish" product features. 

Yes, obviously the pre-translated terminology fragments would have to be checked to ensure plausability and correct grammar. Simple word equations rarely work as intended. But with a well-managed termbase that does not contain too many synonyms, and in which preferred/forbidden terms have been flagged, the effort of translating "around" correct, defined terminology and editing incorrect terms would likely be lower than post-editing a machine translation riddled with "best guess" terminology that would have to be validated by research. On a lexical level, it is MT that melts down to pot luck, unless the MT engine processes user-defined glossaries. That is why, in my opinion, terminology management has never been more important than it is today. 

Thank you for pointing me at the Ideas section! I will check it out. Although I doubt that RWS will want to be revive deprecated, "old-schoolish" product features.