On Mon, 5 Sep 2005 19:13:58 +0200, Benct Philip Jonsson <[log in to unmask]> 
wrote:
>
>Julia "Schnecki" Simon skrev:
>
>> 1.2. Consonants
>>
>> There are four basic points of articulation for consonants: labial,
>> dental, palatal, velar.
>>
>> There are the following consonant phonemes:
>>
>> voiceless aspirated plosives p_h, t_h, c_h, k_h
>> voiceless ejectives p_>, t_>, c_>, k_>
>> voiced unaspirated plosives b, d, J\, g
>> approximants/glides w, r\, j, M\
>>
>> Furthermore, there are two archiphonemes (nasal, /N/, and lateral,
>> /L/) that are realized as [m], [n], [J], [N] resp. [l_w], [l], [L],
>> [L\] depending on their surroundings.
>>
>> (No decision reached on consonant graphemes yet. Sorry.)
>>
>
>What about:
>
>/p_h/ p         /p_>/ p'        /b/  b          /w/  w
>/t_h/ t         /t_>/ t'        /d/  d          /r\/ r
>/c_h/ c         /c_>/ c'        /J\/ z          /j/  j
>/k_h/ k         /k_>/ k'        /g/  g          /M\/ h
>
>/L/ l, /N/ n
>
>?
>
>Yes, I admit I enjoy constructing romanizations...

I'm afraid I can't do much more than endorse BPJ's romanization. I'll point 
out an alternate one, though: Consonantletter + |h| for aspirated an plain 
Consonantletter for ejective. Use of |y| for /@/ pretty much limits /j/ to 
|j|, which leaves a choice of |z|, |q|, or |x| for /J\/.

But I don't completely understand the sandhi rules yet -- they're quite 
different than the simple ones I tend to use, except for some of the glide 
insertions.

Oh, and let me say something about this part of the original msg:

>> Clusters of more than two consonants are avoided. Such clusters can
>> never occur within a morpheme, but sometimes they appear at morpheme
>> boundaries (for example, when a consonant-initial suffix is attached
>> to a stem that ends in two consonants). In such cases, a predetermined
>> "buffer" vowel will appear. ("Problematic", i.e. consonant-initial
>> resp. -final, morphemes come with their own inherent "buffer" vowels.)
>> (There are of course exceptions to this rule; namely, a
>> stop+nasal+stop sequence will drop the nasal, as described above,
>> instead of sprouting a vowel.)

I find working out the details of this kind of thing to be real trouble; 
even with 2 consonants, I never see the end of it. With MNCL-2B I took a 
different approach: I arranged the morphemes so that all intraword clusters 
are also intramorpheme. Then I pick the nicest ones for the roots and if I 
run out, add new types of combinations, such as those with buffer vowels. 
Tha way I don't have to decide everything at once. But then, I haven't 
tried to make MNCL-2B naturalistic. I _do_ allow some 3-consonant clusters: 
of the form Cpl, Ckl, Cbl, Cgl.

Jeff


>--
>
>/BP 8^)>
>--
>Benct Philip Jonsson -- melroch at melroch dot se
>
>         Solitudinem faciunt pacem appellant!
>                                             (Tacitus)
>=========================================================================