I wonder has anyone ever tried to reconstruct Latin language via data on modern Romance languages as if we know nothing about what Latin actually was.

Both as a fun exercise and as a method to test improve our reconstruction methodologies. If not, “reconstructing” other known languages is also something worth knowing about.

  • 1
    This answer talks about that, but it is not a 100% match: linguistics.stackexchange.com/questions/20777/… Feb 4, 2020 at 11:30
  • 1
    I think this is going to be tainted by prior exposure to the language. You'd have to find someone who knows nothing about Latin, yet is very skilled at language reconstruction, to make sure they aren't making sound change inferences that nobody would make without already knowing Latin... I think this intersection of things could be very tricky to find in an individual.
    – LjL
    Feb 4, 2020 at 18:21
  • 5
    I have no references, but I suspect this is frequently done in computational linguistics. At least, a computer isn't tainted by prior knowledge, and you need test data that is actually known, so I think it's a natural thing to test an algorithm that way. (the helsinki PIE guys come to mind)
    – vectory
    Feb 4, 2020 at 21:49


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.