In “Names of Large Exotic Animals and the Urheimat of Japonic” by Alexander Vovin, he attempts to connect the words for elephant, tiger, and crocodile with those found in Central or Southern Asia, often from Austroasiatic languages. Though the possibility of loanwords complicates this, it would have interesting consequences for the origin of the modern Japanese people and their language, perhaps requiring a modification of some parts of the Altaic Theory, if it could survive at all. Many of his conclusions could depend on evidence he didn’t analyze there.
​
As for the specifics, if \*wani ‘saltwater crocodile’ was actually related to \*mangi, it might explain apparently irregular n > y before i (then yi > i) in Middle Okinawan waisaba from \*wani-samba. This would mean only ng became y before i, never n. However, unless the sound ng was very rare, I’d expect to see more apparent examples of ng > n and y in that environment. If irregular, or due to causes no longer visible, simple n > l > y could explain variation like the many words ending with \*-y but without \*-y- in compounds. Since these compounds could change t- to -d-, etc., in the next component, previous reconstructions would suggest they ended in a nasal. If this nasal consonant was n, then final -n > -l > -y but -n-t- > -d- would be explained.