> And I've also been wondering if you could construct the opposite, only using the romanic parts of the language. You would probably get really corrupted medival french?

I reckon you'd basically get Interlingua (https://en.wikipedia.org/wiki/Interlingua), which saw some initial use in scientific communication.