On Sun, 24 Apr 2011 10:53:13 -0400, Alex Fink <[log in to unmask]> 
wrote:

>On Sun, 24 Apr 2011 03:21:52 -0400, neo gu 
<[log in to unmask]> wrote:
>
>> k' and t' are ejectives
>> _H = high tone, low tone unmarked
>
> Well, it's easy to find _some_ phonotactics that fits your words, by
> making it highly permissive.  It looks, for instance, that all of your
> words can be syllabified CCVC, where V may be a long vowel but
> long consonants count as CC.

Thanks. What I'm now wondering is whether VCCV should be 
syllabified as V.CCV or as VC.CV -- I suppose the latter since codas are 
moraic (not evident from the transcription)?

> My approach to the phonotactics would be to start there, and impose
> further restrictions one at a time as you find possible ones.  For
> instance, the maximal codas and onsets in your data under this
> analysis are [X.St k.:l  n.t_>m   #Sk #m: #Ng   ?# n# s# k_># h#
>  X#].  There are various generalisations you might make from the
> sets that appear there.

such as N.k_>m  #st #sk #St #mb #nd etc?

> For instance: every internal coda there also appears as a final coda,
> except that there's internal [k] and the closest final analogue is
> probably [k_>].
> Might these be the same?  Perhaps coda stops are neutralised to
> glottalised (this can nat-happen) unless they form a geminate.  Or,
> strongerly, could plain stops and ejectives be allomorphic?  This is
> almost possible with your data: the distribution plain after a
> consonant or when geminated, ejective otherwise only fails in two
> instances, the [t_>] in [hE_Hhe?:i_Hnt_>mo] and the [k] in
> [?it:S)a_Hkambes].  (So you could accept it if you're willing to change
> the data there, or fudge these cases with other exceptions.)

Possibly the ejectives occur before nasals and non-homorganic 
voiceless stops?

> Another interesting near-pattern I see is that [E] and [e] are almost
> in complementary distribution by tone, [E] high, [e] low.  Here the
> exceptions are [e_H] in [?e_Has:oa?li], so maybe you always get [e]
> before another vowel; and [E] in [m:Ek_>na], which I have no clever
> ideas for at the moment.

That might be [m:E_Hk_>na_H]. Unfortunately there were a couple of 
typoes in the original list; they should have been

hE_HhE?:i_Hnt_>mo
Ngadaho_HstEk_>

Here's a couple additional words:

s:eo_Hant_>ke
ni_HntoX:a_HlEt_>

--
neogu