Funny W

[Rankin, Robert L]

Thai has all three, l, r, n, in different contexts.  One of our grad students who kept a pretty hefty accent in her English used to regale us with her rendition of "phonological rules", which was "phonorogican lunes" [fonorajikan lu:nz].  (n>n,  l>r,  r>n, but not in that order).
 
On a more serious note, Siouan already has complementarity among l, r, n, d in languages like Crow, so I think any such distribution of W/R would have run afoul of preexisting alternations.  A couple of weeks ago  I provisionally expressed a willingness to go with /mb/ and /nd/ as single phonemes.  These would have to be single units, not clusters, and such segments are not terribly common, especially in the eastern 2/3 of North America.  They're found in the South Pacific, West Africa and perhaps South America mostly.  I'm having strong second thoughts about positing them in Siouan at all.  Siouan has real nasal vowels but no real nasal syllable codas -- all such codas are just offglides of nasal vowels preceding voiced stops.  These generated (merely phonetic) nasal codas don't cause behavior like R or W.  So I'd have to say that it's unlikely W and R were like them.  Maybe relative chronology could help, but I'm still not very convinced.
 
I think, at the moment, I prefer to stick to what we know plus a suspicion and ? or h are involved with funny R and W.  I also generally feel that solutions that rely crucially on the "invisible superstructure" of phonology, like "tiers", "minus-alpha greek letter variables", etc. are not likely to prove useful, but that's just me.  I used to infuriate some of my colleagues on the dictionary project with this attitude.  
 
Bob
-------------------------------------------------------------------------
2. In modern languages like Korean that show this allophonic distribution, exactly what is the phonological nature of these sounds? I would imagine that either the "n" or the "r" would probably be a nasalized tapped r.