>Another practical step could be to identify a specific minimal set of language units required for any given popular nonconstructed language. Is this already a thing?

You mean something like this question?

https://news.ycombinator.com/item?id=19331307