Palatal harmony ( was: Transcription exercise)

Isaac Penzev skrev:
 > Benct Philip Jonsson jazdy:
 >
 >
 > | I thought of the change in some Turkic languages whereby
 > | palatal vowel harmony is replaced by palatal consonant
 > | harmony -- i.e. rounded vowels in 'front' vords become
 > | back, but the consonants of these 'front' words remain
 > | palatalized before the formerly front vowels.
 >
 > Exactly. This phenomenon is especially peculiar to Karaim,
 > but the tendency is notable in Kumyk, Karachay-Balkar,
 > Nogay. ObConlang: That reminds me to try to resume work on
 > Project 20 (formerly known as Kumanzha). This LLL-styled
 > Turkic conlang is going to have the same kind of harmony
 > due to the influence of the East Slavic phonology.
 >
 > | I don't know if it works the other way too so that
 > | palatal(ized) consonants in 'back' words lose their
 > | palatality.
 >
 > There are no phonemicly palatalized conss in Turkic langs,
 > so we cannot be sure if it could work or no.

So Turkic "ch" and "sh"  are [ts`] and [s`] rather
than [ts\] and [s\] then, I take it.

Methinks it could go four possible ways in Heleb:

1) Palatalized and unpalatalized consonants remain distinct
    in all words (the boring solution).

2) Palatalized and non-palatalized consonants remain
    distinct in back harmony words, but merge as palatalized
    in front harmony words.

3) Unpalatalized consonants palatalize in front harmony
    words and palatalized consonants depalatalize in back
    harmony words.

4) Back vowels front in words that contain any palatalized
    consonants.

There are problems with all of them:

1) There would be only palatalized before unrounded long
    front vowels and only unpalatalized consonants before
    long back vowels, while the consonant pairs would be
    distinct before short vowels, as it is only long vowels
    that can be front rounded underlyingly, since front
    rounded vowels originally derived from

         * VjV sequences where at least one V was rounded,

         * VwV sequences where at least one v was front, and

         * VhV/VqV sequences where one vowel was front and
           one was back rounded. (NB _q_ is /G/ or /R/!)

    and I think such a system would anyway tend towards (2)
    -- after all even Finnish has some consonant synharmony
    with /s/ and /l/.

2)   Again the palatalization distinction would only exist
       before short vowels, so (2) would tend towards (3).

3) The unlikelihood of consonants dispalatalizing
    before back vowels, but that may be a minor problem
    if the occurrence of such sequences are restricted
    to begin with.

4) It is unlikely that the presence of palatalized
    consonants would sway a long back vowel to become front
    -- but I guess that in all applicable cases there would
    be a long front vowel in the word as well. Now the
    likelihood of the necessary contexts for the arising of
    two long vowels in the same word is minimal, especially
    in root words, and in derived words vowel harmony and
    analogy would likely level things out in the end.

So all in all the balance is towards (3) with some possible
admixture of (2).

BTW I think labials (except w) will be exempt from
palatalization, so that I get the following
unpalatalized/palatalized pairs:

         t           c [ts\]

         d           j [(d)z\]

         s           ç [s\]

         z           zy [z\], or = j, or = y

         n           ny [n\] (alveopalatal nasal)

         r [r\]      y [j]

         ll [5]      l [L]

         çr [s`]     ç

         hll [K]     hl [C_l]

         k           ky  [c]

         g           gy [J\]

         h           hy [C]

         0           y       (0 = zero < */G/)

         w           wy [H]

         ng          ngy [J]

IMO it is highly questionable if [n\] and [J] would remain
phonemically distinct, and the same goes for [z\] and
[dz\], seing that [z\] is an allophone of /dz\/ in the
other Sohlob dialects, so the likely fate of [z\] < /z/ is
merger with /dz\/ or /j/. The phoneme /x/ exists no longer
in Heleb, having merged with /h/ after non-initial /h/
disappeared, hence /h/ corresponds to what is actually the
palatalization of [x].

Now if i pick scenario (3) none of the sound pairs above
will be phonemically distinct, so perhaps only those not
transliterated with |C+y| will actually be distinguished in
spelling. However spelling is hardly logical, since firstly
Heleb spelling is under the tyranny of Classical Sohlob
spelling, since these languages are in a diglossic
relationship, and secondly the spelling of front rounded
vowels is defiscient -- they lack separate letters, so it
may well be that /k2/ is actually spelled |kyo| and so on.

--
/BP 8^)>
--
Benct Philip Jonsson -- melroch at melroch dot se

    "Maybe" is a strange word.  When mum or dad says it
    it means "yes", but when my big brothers say it it
    means "no"!

                            (Philip Jonsson jr, age 7)