On Thu, 19 Nov 2009 12:08:22 -0500, Jesse Bangs <[log in to unmask]> wrote:

[On the distribution of onsets and codas]

> This is actually *very* unusual: almost all languages have more onsets
> than codas, and if they have segments that only occur in a coda (like
> English /N/) they tend to have restricted distribution and/or fuzzy
> phonemic status. You have a voice contrast and an entire nasal series
> that only occurs in codas! This is far more surprising to me than
> anything you mention below.

Well, if you can come up with more onset phonemes that can be readily  
glottalized (and that are reasonably common), I'll gladly consider adding  
them. I'm seriously considering adding the liquids, glottalized by adding  
a preceding glottal stop and aspirated by devoicing, giving these onsets  
(under a more traditional analysis)...

p p_> p_h t  t_>  t_h               k k_> k_h ∅
m ?m  m_0 n  ?n   n_0               N ?N  N_0
           l  ?l   l_0               5 ?5  5_0
           ts ts_> ts_h tS tS_> tS_h
           s  s_>  s_h  S  S_>  S_h

Of course, this introduces the trigraph "llh", among other  
virtually-Lovecraftian horrors, to the romanization.

With the codas: it seems that I could (and perhaps should) drop the  
voicing contrast in stops, but it seems like it might be overkill given  
the normalized analysis of the onsets...

p t     k ∅
m n     N
   l     5
   ts tS
   s  S

On Thu, 19 Nov 2009 12:08:22 -0500, Jesse Bangs <[log in to unmask]> wrote:

[On the analysis of glottalization as an "onset" or "mid" feature of the  
peak, rather than as a distinct segment]

> I like this, though I question what would make you analyze it this
> way. Are there morphological or phonological processes that use
> glottalization? If not, why wouldn't you just analyze /?/ as a segment
> of its own? I also see why you want to avoid voiced stops in onset
> positions--how would they handle glottalization? Of course, if /p'/
> and /b/ are in complimentary position, I would be tempted to analyze
> them as allophones.

Why analyze it that way? Because it's a convenient model to describe the  
system in layman's terms without having to introduce phonological  
constraints and dependencies dealing with harder-to-model-simply things  
like vowel harmony, and allowing roots to be disyllablic without thereby  
allowing them to be multisyllabic (on the other hand, I literally just had  
a quick play with multisyllabic roots where the medial consonants are  
limited to /h/ and /?/, and it might actually turn out nicely). Plus, my  
initial thoughts about the native writing system sort of rely on that  
being the "native intuition" regarding the way words are structured.

> Adding aspiration to the mix sounds good. Adding uvularization (is
> that what you meant?) seems like overkill, but it's up to you of
> course.

Some second kind of back fricative alongside /h/. I have no solid idea  
what exactly, but I think an uvular one might be what I've got in mind --  
I could go further back in the vocal tract, if it'd seem more  
naturalistic. Having three just *feels* more "balanced", for some fuzzy,  
ephemeral reasonoid. Falls strictly in "BEEP BEEP BEEP DOES NOT COMPUTE"  
territory, but that does not stop it being my aesthetic instinct for this

On Thu, 19 Nov 2009 12:59:52 -0500, Roger Mills <[log in to unmask]> wrote:

[On the romanization]

> What do you have against the letter "k"???

Nothing. I was aiming for a romanization con-devised by a 16th or 17th  
century Spanish speaker with knowledge of French, but that's an utterly  
minor concern. I also have a semi-native alphabet devised by a 19th  
century English speaker on the back burner. It goes a little something  
like this...

A ∀ C E ∃ H I K L Γ M N И O P S Ƨ T Σ Y '

Can y'all figure out the phone*ic values of the alphabet symbols?

> That could free up "c" for /S/?? (Except apparently you want to show the  
> relationship between x=/S/ and tx=/tS/....-- it could still be shown  
> with "c ~ tc"). Just a nitpick :-)

Nothing whatsoever wrong with nitpicks. I got exactly what I asked for! :-)

> 2. In view of the (I assume) non-contrast between "p'ayg" vs. "pa'ig",  
> this suggests to me that the symbol "y" is unnecessary. Why not just  
> write it with "i" in both cases? (since apparently bisyll. "paig" is not  
> a possible base form).

I could, and maybe I should. Good food for thought, especially if I'm  
going to take on board this notion of normalizing the phonemic analysis.

More replies to more people later. I've got to wrap up for now, and get to  
cogitatin', deliberatin' & masticatin'.