Re: /w/ vs /B/

Lars Finsen wrote:

> In Old Urianian I have to deal with an unrounded /B/ (from IE bh)
> existing alongside the rounded /w/ (unchanged from IE). So far I have
> assumed they have independent evolution in later stages of the
> language. But since they must sound pretty much alike I wonder if
> they will have a tendency to merge. Could anyone with a large
> repertoire of natlangs please confirm this, or mention any examples
> of the opposite in known natlangs?
Austronesian: in languages where *b lenits to a fricative, it may  or may
not merge with the reflex of *w. *W is usually assumed to have been
labio-velar (though AFAIK only reflected as such in Chamorro of Guam as
/gw/); along with *j/y it was rather rare and had distributional oddities.

South Sulawesi, Buginese: *b/*w have merged in the surface phonetics (both
[w]) but not underlyingly-- l. N+w(*b) > mp-, w(*w) does not occur in that
env. 2. w(*w) can have an alternate vocalized pronunciation ['waI] 'water'
~[u'(w)aI], w(*b) cannot. There are also dialects where w(*w) has > h, but
not w(*b).

OTOH! there are closely related languages in the SSul family where the w>h
change has affected both w's, and then moves on > 0.

Malay/Indonesian: *w has been lost in most positions except the env.
/-a_a-/, and *b has also shifted to /w/ in that env. Both are pronounced as
labio-dental approximants. Most words with /w/ are loans from other
dialects, Javanese, Skt., Arabic, Dutch etc.

Eastern Indonesia: In Kei, both are reflected as written /w/; I don't know
the phonetics; but in closely related Fordata (Tanimbar Isl.) *w > w and *b
> v (written).
Leti (SE area) *b > /B/, *w > /w/

Atoni (Timor) both > /f/

Oceanic, Fiji: general OC merger of *b/*p, > Fij./B/ written "v"; *w is
retained as /w/ "w"

So you can see there are lots of possibilities.