<html><body>
<p><tt>> Guys,<br>
> <br>
> I'd like to bring up the extreme distributional skewing of */mb/, */nd/ if you consider them distinct phonological segments. <br>
> <br>
> I explain the skewing as the interaction of prefixes with root/stem initials, i.e., morphological and phonological clusters.<br>
> <br>
> If we ignore the morphology and go with the unit phonemes (which I admit we're sort of doing using the symbols W and R), how do we > explain why these units only occur in very narrowly defined contexts? Don't we already have enough trouble with /glottal stop/?<br>
> <br>
> Bob</tt><br>
<br>
<tt>Exactly what narrowly defined contexts are we talking about here? I understand that *W and *R are virtually all word-initial before oral vowels. Any other restrictions?</tt><br>
<br>
<tt>I'd be very open to the idea that *W and *R result from an interaction of prefixes with root/stem initials. That seems like a very reasonable explanation of their origin. I don't think that should affect the question of their phonological configuration though, unless we can also estimate just what the hypothetical prefixes and initials were.</tt><br>
<br>
<tt>Rory</tt><br>
</body></html>