Telek dialects

Over the past week Telek has gained in complexity.  There are now two separate
dialects: Kaldilak, the largest with approx. 600 speakers, and Saskumik, which
has approx. 400 speakers.  They are spoken in the towns of Kaldila and Saskum
respectively.  (No, the final -k in the language and dialect names is not a
coincidence.)  The differences in grammar are not worked out much at this
point, but the phonology is.

Proto-Telek, like both daughter dialects, has three high vowels: i, u, and y
(=high, central, unrounded; IPA barred i) The initial vowels of these stems
serve as demonstration. (Doubled means long).

-iggasi 'unexpectantly short; stunted'
-ydla'as 'walk'
-usmilix 'faint'

Following an alveolar obstruent, the vowels shift forward: u > q (=barred u,
central, high, rounded), y > i, i > i (doesn't change).  Just for the
record, I
do not represent 'q' in the orthography - it is here for demonstrative
purposes
only.

itigassi 'We are stunted'
itidla'as 'We walk'
itqsmilix 'We breathe'

When a syllable does not carry high pitch, they "lower" (not the best term,
but
it'll do):  i > I, u > U.  (This actually does not occur in hyper-careful
speech).  Because the difference between [Y] and [I] are so slight, y > I.
What about [q]?  Well, [Q] is so similar to [U] that one *could* expect a
merger (q > U).  However, since [q] only arises in a fronting process, it is
odd to move further back in the mouth.  The treatment of [q] is where Kaldilak
and Saskumik divide.  In Saskumik, the merger is precisely what happens: they
reverse the fronting process, so q > U.  In Kaldilak, the expect [Q] derounds,
merging with the expected [Y], and fronts further to [I].  Thus Kaldilak
has no
distinction between the high vowels in post-alveolar, low-pitch position; they
all change to [I].  (Note that [U] can never occur after an alveolar due to
fronting).

Saskumik                        Kaldilak
etIgassi                        etIgassi                'We are stunted'
itIdla'as                       itIdla'as               'We walk'
etUsmilix                       etIsmilix               'We faint'

This lead to another development in Kaldilak.  Since the expected [Q] merged
with the front vowels, [q] merged with [y].  So now, u > y, while in Saskumik,
we still find u > q.  We asked to enunciate the above words extremely
carefully, the "lowering" does not occur, and we hear the following.

Saskumik                        Kaldilak
etigassi                        etigassi                'We are stunted'
itidla'as                       itidla'as               'We walk'
etqsmilix                       etysmilix       'We faint'


The other major difference between them lies in the alveolar obstruents.  In
Saskumik, an alveolar obstruent preceding a front vowel palatalizes: t > ch, d
> j, s > sh.  (Fricatives often voice intervocalicly, so descriptively z > zh;
but I prefer to think of this as merely the voicing of [s] and [sh].)
Applying
these changes to the examples above, we get:

Saskumik                        Kaldilak
echigassi                       etigassi                'We are stunted'
ichidla'as                      itidla'as               'We walk'
etqsmilix                       etysmilix       'We faint'

Geminates don't palatalize, which is why the first word is echigassi
instead of
*echigasshi.

Examples of [d] and [s] are:

Saskumik                        Kaldilak
lushi'to                        lusi'to         'my son'
sammijin                        sammidin        'shelter'

The final difference that I'm so far aware of involves geminates.  Saskumik
seems to be losing gemination within morphemes -- never across morpheme
boundaries.  This is a very recent thing, and has not spread to all words
yet.
So they still have -ss- above; but they say _xaashi_ 'leaf' instead of
_xaassi_
'leaf' like the Kaldilak speakers.

When I first discovered this feature, I thought it would force a different
pitch contour to many Saskumik words, because the pitch-accent system targets
heavy syllables.  Thus, where Kaldilak has a heavy syllable due to gemination,
Saskumik would have a light syllable due to degemination.  As it turns out, in
every instance where this would affect the pitch-accent, the Saskumik vowel
must lengthen, making the syllable heavy again.  So it turns out that there is
not difference in the pitch contours in the two dialects.  Quite the surprise
for me: I'm both pleased at the unexpected result (shows the language has a
life of its own), and irritated that the affects weren't as far reaching as I
had hoped.  I'm leaving it in place though.

Marcus