<html><body>
<p><tt>Thanks, John! That was very informative, and encouraging.</tt><br>
<br>
<tt>The thought that has been floating around in my head for a while is that the *W and *R make up a series that could be defined as full stops that are initially released through the nose rather than through the mouth. These could be considered either single phonemes or clusters. Considered as clusters, we would write them as *W = pm and *R = tn. (Voicing would not be an issue. Actual pronunciation might be anywhere from bm/dn to pm/tn to pm^/tn^, where m^ and n^ are defined as the voiceless equivalents of m and n.) Considered as single phonemes, the effect would be rather like a person with a cold trying to hit m or n, and taking just a moment before sufficient pressure builds up to force a passage through the nasal sinusses.</tt><br>
<br>
<tt>I think this model explains the daughter language reflexes we find more naturally than any other. Postulating any kind of w for the Dhegihan *Wa/*aWa' sets seems a bit out in left field, because all the reflexes involve full labial closure. For other single phonemes, as you say, the territory is crowded. The mb and nd clusters you suggest make the most sense, but if that were the case I should think we would get those epenthetically everywhere a nasal vowel precedes a stop.</tt><br>
<br>
<tt>A nasally released stop set would not conflict with anything else. Phonotactically, it would be very easy to get either ama' or apa'/aba' from *apma'. In the former case, the speakers would simply open the nasal passage a little bit earlier, which would prevent the stop from occurring, yielding ama'. In the latter case, they would open the oral passage earlier, before the nasal passage was opened, to produce an orally released stop preceding an oral vowel, which would eliminate the reason for opening the nasal passage at all. In either case, the resulting consonant would be more quickly and easily rendered than the heavily-marked original, and would immediately merge with a pre-existing consonant phoneme (m or p/b).</tt><br>
<br>
<tt>This would also mesh fairly comfortably with the *pr cluster you mention as being sometimes ancestral to *R. The key here is that we have a stop+sonant cluster, just as pre-existing nasally released stops are stop+sonant clusters. The latter are "clusters" at a single location (labial or alveolar), while *pr is a definite cluster in which the stop is labial and the sonant is alveolar. I would picture the Dhegihan development of this to be: *pr => *pn (r takes on full oral closure along with p, phonotactically equivalent to the double stop sequence pt, but maintains sonant quality by allowing voice to exit through the nose during alveolar closure). Next, double stop sequences are lost. The second stop takes on the stop function of the first stop, as the first stop is reduced to h. Thus, *pt => *htt (both preaspirated and tense), while *pn => *htn (p => h, but forces the following alveolar closure to take on its initial stop function). The *tn part of that is identical to the pre-existing *R, a nasally released alveolar stop, and *htn/*hR eventually merges into *tn/*R in all daughter languages.</tt><br>
<br>
<tt>Finally, if we picture *W and *R as the clusters *pm and *tn, and if the original language had *m and *n (not sure about this) but not eng, then the picture of *W and *R being originally merged phonemes, oral stop + nasal consonant in the same location, would explain why we don't find "funny" consonants in the velar area.</tt><br>
<br>
<tt>I don't know how well this model fits across Siouan, but I think it works pretty well with the Dhegihan and Dakotan cases that you've mentioned. Any thoughts?</tt><br>
<br>
<tt>Rory</tt><br>
</body></html>