Funny W

[Koontz John E]

Rory:
> Just to be sure we're clear here, by "nasally-released stop", I mean a
> full stop that is released as the corresponding nasal consonant, not as
> one that is preceded by one.  Thus, for *W I propose *pm/*bm, not *mb,
> and for *R I propose *tn/*dn, not *nd.  Of course, these might easily
> have reflexes mb and nd by metathesis, but that's not what I'm proposing
> for the originals.

I'm afraid I think that nasal-onset is nuch more likely (and common) than
a nasal release, and seems more consistant with the distribution of *R in
words.  Perhaps the difference between *r, *R, and *CR is that *r was
nasal throughout [n] in initial position, and probably oral or nasal
medially, depending on whether or not the following vowel was nasal, while
*R has an oral release, e.g., [nd], and *CR has just the oral part, e.g.,
[Cr] where r is a tap.  In a few cases *CR comes to behave as *Nr [md,
nd], when *R has becomes complete oral.  In essenve, I'm positing an extra
nasalization initially, which affects *r, *w, *R, *W and *pr clusters.
The last presumably have a tendency to become *pR, or actually *WR [md,
mnd].

How would you make pm and tn uniquely likely given the distrition of these
sets within word forms and the outcome of *W and *R in the various
branches?

> Since we know that some cases of *W and *R arose from clustering of *w
> or *r with an obstruentizing consonant, we can suppose that they all did:
> therefore laryngeals.  This is a very reasonable hypothesis for research,
> but it is not solid as an argument.

I don't know.  It tends to combine the distributional facts with the usual
suspects to produce something attested in American languages.   As it can
be seen, I think we don't need to assume that *R is is *?r or *hr, but
such things are found.

In fact, I think Bob is considering that *?-stems with A1 m- and A2 n- in
Dakotan may represent *w-?VNand *r-?VN in a very real way.  I tend to
explain these as *w-V and a form imported from the *r-stems, with OP A2
z^- from *y-V being the original form.  We'd expect A2 *c^h-VN in Dakotan.
In anye vent, in *h-stems we have A1 *p-h... and A2 s^-..., maybe from A2
**y-h...  perhaps via *z^-h...