whale / whale > crocs and sharks

[Ross Clark (FOA LING)]

Waruno Mahdi wrote:

Virtual *Zambuha[Dj]a has a narrower distribution (Tag. _dambuhala_,
   Ngaju _jambuara_ "whale", Merina Malagasy _lambuara_ "k.o. fish"),
   but Old Javanese _lembwara_ "k.o. large fish" with irregular initial
   _l_ for expected _d_. The problem that this creates is aggravated by
   the circumstance that the Malagasy form coincides with the expected
   if it were a loan from the Javanese, and that would even suggest
   possible etymologies in the direction of _lembu_ "cow" + ???.

This seems like a very long shot, but Proto-Polynesian *tafuraqa "whale",
with no presently recognized external cognates, seems tantalizingly similar
to this one. I realize that a certain amount of tweaking would be required
to get the forms to match exactly, but if *sakulayaR "swordfish" can span
the MP realm, why not?

Ross Clark