🚀 go-pugleaf

Perhaps in Austronesian. I say "perhaps" because there are
no ancient written records and I fear that much of the
reconstructions might be nothing but artifacts of the
methodology.

At any rate, for whatever this is worth:

When I attempted to reconstruct Proto-New-Hebridean
10 years ago, I ended up with labio-velars (some 
modern dialects also have labio-velars). In most 
dialects they went to labials. In some dialects
they went to labials, while labials preceding
front vowels went to dental (via apico-labials)
or to interlabials. In some languages outside the
New-Hebrides, labiovelars went to velars. However,
Otto Dahl, Isidore Dyen, Robert Blust et al. do not
reconstruct labiovelars, which makes me suspect that
my labiovelars are artifacts, and might have originated
from labials and velars followed by back rounded vowels
or semivowels.

One fairly good shred of evidence is the cognates 
for "man" in Sakao (New-Hebrides), Kia (Solomon Islands),
and Bau Fijian.

Sakao has aTngo/n < *at(aeo)ng(iu)n(aeo) (T= apico-dental fricative, ng = velar nasal,
	o/ = IPA o-slash))
Kia   has mane
Bau   has ngone "child"

All this seems to point to an original *ngoane or *nguane
rather than *ngwane (*atangwane would have yielded Sakao 
aTman). There's much more side evidence for those reconstructions,
but I won't bore you with that lengthy stuff.

The evidence of New-Hebridean languages are a bit of a nuisance,
as they force eccentric reconstructions. Thus *mwata "snake"
vs *mata "eye" (for which, incidentally, Bau has ngata and mata
respectively). I'd say: case not proven. But I never liked my
reconstruction of labio-velars. It smacks of an artifact of
the comparative method.