NEW: `Ngwei-koo`lei (phonology)

I'm here to present you my latest language, which I've
told you about. This material is all in its own webpage
(http://www.geocities.com/pablo-david/ngwei.html)
but I'd like to get some input if y'all don't want to
go through it.

The language is called `Ngwei-koo`lei. It's a tonal language
derived from the ancient language that the Dráselhadh scholars
named Nolt Lethris ('The Old Speech') in Draseléq. NL was not
tonal; NgKL's tonogenesis is extremely complicated so I'm not
going to explain it here (but rest assured I'm not making it
all up).

Tone is marked before the syllable, and is either

    - low
    ' rising
    ` falling
    ^ high

(low tone is unmarked in monosyllables). In phonemic transcriptions,
I will use /1/, /12/, /21/, and /2/, respectively, for the tones.
The name of the language is /Nwej21 ko:1 lej21/, which breaks up
(etymology in NL, cognate in D[raseléq]):

    `ngwei 'one thousand' < NL /N_gwes/, D /gwes/
    -koo   'village' < NL /koho/, D /ko:/ or /koh/-
    `lei   'speech' < NL /leSa/, D /leT/

thus meaning "Thousand Villages' Speech" -- where the 1000 Villages
was the old name for their country (consisting of a loose federation
of small towns and larger city-states). How they became separated
from the then-Dráselhadh and the Biyuron is something we don't know
so far, but the split of NgKL from NL is about 1400 years before the
current time frame -- in comparison, Draseléq and Biyuron were clearly
split some 600 years afterwards.

The language is extremely analytic, mainly due to phonetic change --
words have been extremely shortened, to the point that only stressed
syllables, and syllables with long vowels, have survived the process
of syncope. Due to these processes, at the same time, the phonemic
inventory has become rather large, while the phonological constraints
have tightened.

Syllables are all of the form (C)(w, j, wj)V(F), where C is any
consonant, <w, j, wj> represent labialization, palatalization
and labiopalatalization (~ C + /j<rnd>/), V is a vowel (short or
long) and F is a nasal (one of /m n N/) or a semivowel (one of
/w j j<rnd>/).

Not all consonants can be labialized, palatalized or both. In
many cases, such a change surfaces as another phoneme. For example,
*/fwj/ > /Pj/, */Tj/ > /C/, etc.

The whole inventory is rather large, 65 consonants (plus 6 vowels,
/i y e a o u/) and the phonological rules are complicated too, so
I'll not mention them here.

Speaking of tones, we see two processes of sandhi, one purely
phonemic, one morphological. Phonetic tone sandhi rules say that
two identical non-level tones cannot occur in a row; if it happens,
the second one becomes level:

    `mei       'day'
    `mei-mei   'days' (instead of *`mei`mei)

    'fain      'head'
    'fain^fain 'heads' (instead of *'fain'fain)

Morphological tone sandhi states that certain affixes that have an
intrinsic non-level tone must change it to a level tone when appending
to a root with the same tone. This produces the same result as phonetic
sandhi for suffixes, but reverse result for prefixes:

      `dong     (distal deictic, 'that')
    + `mei      'day'
    =  dong`mei 'that day'

(phonetic sandhi would've produced *`dong-mei).

Well, this is all for now. Any comments (especially about
prospective failures) will be appreciated, as well as any
note on terminology -- I'm kinda lost over here regarding
tones.


--Pablo Flores
  http://www.geocities.com/pablo-david/index.html
  "... When all men on earth think, day and night, about the
   Zahir, which one will be a dream and which one a reality?"
                              Jorge Luis Borges, _The Zahir_