This is something that was raised years ago and I believe the whole TextBase system was being looked into, so this might already be considered. But basically, my suggestion is to allow for dynamic updates of TextBases and not to be limited to the languages that are set.

Current process is to create a TextBase and specify all languages (you have to 'import' to create all languages you will need). Then, if an update job runs for a language pair that was not defined in the TextBase the job fails. This isn't managable for standard LSPs that often deal with a number of customers across a wide range of language pairs - quite often, we don't know what language pairs might land at any given time.

This should be handled like most other CAT tools and TMSs. TM updates should NOT be limited to the language pair, if the language does already exist in the TextBase then it should add it. Consider separate TMs per language pair - currently TextBases are multilingual.