Well *there's* a well-thought phonology! I'll say it's disturbingly regular
with respect to POA, tho ... if only you could add dental vowels, it'd
collapse into a regular tesseract I guess. ;) But meanwhile the ejective ->
nasal+approximant rule is quite nifty. Is it ANADEW or a wild innovation?
>There are no semivowels or non-syllabic vowels. (There are a few
>glides, but they're listed below under "Consonants".)
So what IS the difference between a semivowel and a glide? The way they are
used in the phonology?
>2.2.2. Consonant gradation
>(d) A plosive-lateral or ejective-lateral sequence at the onset of an
> open syllable corresponds to a voiced plosive with lateral release
> at the onset of a closed syllable (e.g. [p_hl_w] : [b_l],
> [p_>l_w] : [b_l], [bl_w] : [b_l]). (That is, if I can actually
> learn to pronounce those lateral-release stops. ;)
I'm fairly sure [b_l] is impossible... unless you can twist your lower lip
to a "W" shape? But given that you use [l_w] for a labial lateral in the
first place, maybe [bd_l)] or perhaps even [d_l_w] would be a good phone
here.
>I hope I managed to weed out all the
>inconsistencies that had crept in over the weeks...
>
>Regards,
>Julia
Hate to bring the bad news, but I think I spotted one:
>Some more assimilation in consonant clusters:
>
>(a) A sequence consisting of an ejective and an aspirated plosive (in
> either order) changes into a voiced plosive followed by a voiced
> fricative (e.g. /p_>k_h/ and /p_hk_>/ -> [bG];
> /p_hp_>/ -> [bv] or [bB]).
>(c) An aspirated plosive changes into a fricative if followed by
> another plosive or by an ejective (e.g.
> /p_hk_h/ -> [fk_h] or [p\k_h], /p_hg/ -> [vg] or [Bg]).
>
>(d) An ejective changes into a voiced plosive if followed by another
> ejective or by a plosive (e.g. /p_>k_h/ -> [bk_h]). (I'm not sure
> yet how to handle voicedness assimilation here...)
Note how you list *two* realizations [bG bk_h] for the same cluster
/p_>k_h/.
Of course, if you reversed rule (c) to apply to aspirates *preceded* by
another stop, it'd work again. But currently, rule (a) seems to contradict
rules (b) and (c).
John Vertical