Artlang: Kekpem

Now there's this artlang I've had on my mind the past two days... I must
write down for you what I've thought of so far :) I've mostly just fleshed
out the phonology along with some speculations regarding the origin of the
various phonological phenomena. All that I have decided in the
morphological domain is that it will be mostly fusional, with some dashes
of agglutination and a bit of introflection here and there (Germanic umlaut
and Semitic vowel alteration being an example of introflection). The syntax
has not been conceived yet.

"Kekpem" is just the codename so far; I've considered adding tones, in
which case I'd have to define the tonemic values in the name. But that will
wait.

PHONEMIC INVENTORY

Vowels:

/i/  /u/
  /@/
  /a/

/@/ is latinized as {e}

Consonants:

/p t k k_w kp/
/b d g g_w gb/
/m n N N_w Nm/
  /l/
  /s h/
/w j/

The labio-velar clusters /kp gb Nm/ are simultaneous articulations and act
as single phonemes. /t d n/ are strictly dental.

PHONOTACTICS

The most striking feature is a front vs back harmony in the vowels coupled
with a coronal vs dorsal harmony in the consonants. The latter means
that /k g N/ and /t d n/ cannot co-exist within the same word/utterance,
while the labials and the glottal /h/ may co-exist with both; /s/ and /l/
have velar allophones for the dorsal environment. /h s l w j/ and all
labials are considered "neutral" phonemes for purposes of front-back
harmony.

As I said, the vowels are harmonized as well. The front/coronal allophone
of /u/ is [y], while the back/dorsal allophone of /i/ is [1], not the
projected [M]; the back unrounded vowel existed in earlier stages but was
later centralized. /a/ is likewise realized as [&] in the front
environment, and as [A] in the back environment. /@/ has no allophonic
variation.

/w/ is labio-velar [w] in the back environment, and a bilabial approximant
(no SAMPA available) in the front environment. /j/ is palatal [j] in the
front environment, and a velar approximant [M\] in the back environment.

Front or back environment is defined by the first non-neutral consonant in
the stem; the vowels are harmonized in accord with the consonants.

The syllable structure of lexical stems is CVC or CVCC; only clusters of
nasal + stop are allowed. As stated, coronals and dorsals do not appear
together at the surface level; Some CV(C) stems exist as well, mostly
grammatical items, where the coda can be /m/, /l/ or /h/.

Ending morphemes are V(C), where V is either /a/ or /@/, and C can be
only /m/.

Some sample words, and how they are realized:

/tuma/   [tym&]
/pik_wa/ [p1k_wA]
/gbas@m/ [gbAx@m]
/laNm@/  [5ANm@]
/dah/    [d&h]

[5] is SAMPA for velarized l

Essentially, the situation is that the tongue will only articulate in one
part of the mouth within the same utterance, while the lips and glottis may
co-articulate freely with the tongue.


HOW DID _THAT_ HAPPEN?

I've spent most of the time looking for plausible reasons as to how this
type of phonology might arise. This is my humblest attempt:

In Proto-Kepkem, where no harmony was yet present, there existed a length
distinction in the vowels. In all morphemes with short vowels, there
occurred a consonant duplication, whereby a sequence like kVC or tVC would
become kVkC or tVtC; only plosives and nasals were duplicated, a duplicated
plosive assimilated the voicing of the postceding consonant, and would lose
any labialization marking; e.g. /tik@/ > /titk@/,  /mutam/
> /mumtam/, /g_wipa/ > /g_wikpa/. This resulted, of course, in simple
gemination in some cases: /p@p@/ > /p@pp@/.

From the resulting clusters, any nasal + plosive clusters simply stabilized
with assimilation of the nasal (so /mumtam/ > /muntam/). Resulting
geminates remained unchanged. Clusters of plosive + plosive/nasal then came
to be pronounced more or less simultaneously, which helped to merge them in
various ways, eventually. To begin with, any labial + non-labial, for
instance, merged with the equivalent non-labial + labial clusters, e.g. /pk
pt/ > /kp tp/. Between Proto- and Old Kekpem, the nasals developed in more
complicated ways, which I'll just briefly summarize here:

NN Nn Nm  >  NN Nm Nm
nN nn nm  >  nn nn mm
mN mn mm  >  Nm mm mm

After a brief period of stability with this system, Old Kekpem starts with
the merge of velar-coronals with labio-velars, and with coronal-velars
becoming coronal-labials; e.g. /k@kt@m/ > /k@kp@m/, /titka/ > /titpa/. In
late Old Kekpem, the coronal + labial clusters further merged with the
labial geminates (as seen also in the nasals above), so that /titpa/
> /tippa/.
After the loss of phonemic vowel length in Old Kekpem, the previously long-
vowel morphemes had their consonants harmonized by a concert of
phonological conditions and analogy within paradigms; e.g /gata/ < /ga:ta/
became /gakpa/ by analogy with the reflex /gakpa/ of an original
alternation /gata/ from the same paradigm (confusing, I know). In the
absence of analogy, offending coronals would become labials, while
offending velars would become coronals; e.g. /ti:k@/ > /tik@/ > /tit@/
and /g_wa:na/ > /g_wana/ > /g_wama/.

Simultaneous with these developments was harmonizing of vowels and the
other phonemes, which in Middle Kekpem became pegged with the steadily
increasing polarization of the coronal vs dorsal environments. By early
Modern Kekpem, the fully active front vs back harmonizing condition had
developed.

~~~~~

This is of course quite rough; the phonetic changes detailed are quite one-
dimensional in that I'm only trying to account for one phenomenon. There
should be much more things going on than just the harmony thing. I'll be
working on this...

Regards,
Óskar