Re: vowel harmony

  --- In conlang@yahoogroups.com, caotope <johnvertical@H...>
  > wrote:
> tomhchappell wrote:
>> Close vs. Open, Front vs. Back, Round vs. Unround, ATR vs.
  >> notATR, Nasal vs. notNasal, are essentially all the features there
  >> are to vowels;
>
> What about phonation? Or is breathyness/creakyness etc. of
  > vowels always considered a question of tone?

  I completely forgot about phonation, John!
  Thanks for reminding me.

  I am referring to "Principles of Phonetics" by John Laver, Professor
  of Phonetics at University of Edinburgh, one of the "Cambridge
  Textbooks in Linguistics", copyrighted and published 1994 by
  Cambridge University Press.  Its ISBN is 0-521-45644-X and its
  Library of Congress number is P221.L293 1994.

 Laver says pulmonic egressive sounds can have three kinds of phonation;
"Voiceless", "Whisper", and "Voiced". He further divides "Voiceless" into two
kinds, "Nil Phonation" and "Breath Phonation", and also divides "Voiced" into
three kinds, "Normal Voice", "Falsetto", and "Creak".

  So there are six (6) simple kinds of phonation.
  Nil, Breath, Whisper, Normal, Creak, Falsetto.

 Some of them can be combined. "Nil" phonation can't be combined with anything;
and "Breathy" phonation can't be combined with anything except "(normal)
Voice", to yield "Breathy Voice" or "murmur". Also, "Normal Voice" and
"Falsetto" cannot be combined.

  So there are six (6) combinations of two kinds of phonation.
  Breathy Voice,
  Creaky Falsetto or Falsetto Creak,
  Creaky Voice,
  Creaky Whisper or Whispery Creak,
  Falsetto Whisper or Whispery Falsetto,
  Whispery Voice.

  There are also two (2) combinations of three kinds;
  Creaky Falsetto Whisper
  or Creaky Whispery Falsetto
  or Falsetto Whispery Creak;
  and Creaky Whispery Voice.

 Laver says Falsetto is never applied one-segment-at-a-time linguistically, but
rather, to whole utterances.
  He also says there is no linguistic use made of the three-way-combinations.

  That leaves the following nine phonation types as possible distinctive
  features of segments;
    Breath,
  Creak,
  Modal or Normal Voice,
Nil,
  Whisper,
    Breathy Voice,
  Creaky Voice,
  Creaky Whisper,
  Whispery Voice.


  But I can't imagine all nine of them occuring simultaneously and
  being distinguishable in allegro speech.

  However, on pp. 295-297, in Part IV "Linear Segmental Analysis"
  Chapter 10 "Resonant Articulations" Section 10.10 "Voiceless and
  Whispered Vocoids", Laver mentions minimal pairs of words in
  Comanche that differ only in that a certain vowel is voiced in one and
  voiceless in the other -- otherwise that vowel is identical in the two
  words.  In the same place he mentions other languages with voicless
  vocoids; Ik (Eastern Sudanic); Dafla (Nisi), Sino-Tibetan Mirish
  language; and Nyangumarda or Nyangumarta, an Australian
  language.
  In those languages, the question of whether the vowel is voiced or
  unvoiced makes a difference in its meaning, so it couldn't be a matter
  for harmony; but in Tlingit (Alaska), Enga (New Guinea), and
  Machiguenga (Peru), if I understand what Laver is saying correctly,
  perhaps it could be.

  > [snip]
>
>> Round vs. Unround sometimes has more than two values in
  >> conlangs, although I am not personally aware of any natlang in
  >> which it has more than two values.
>
> Don't some dialects of Swedish shift /u\/ to /y_c/ - hence
  > contrasting three degrees of roundedness in high front vowels?

  On pp. 278-280 (section 10.4 "Labial Elements of Vocoid Segments),
  Laver says there are actually four levels of roundedness available:
  stretched, neutral, open-rounded, and close-rounded.

  The lip-opening may be expanded horizontally, or contracted
  horizontally, or left alone horizontally.
  Independently of what happens horizontally, the lip-opening may be
  expanded vertically, or contracted vertically, or left alone vertically.

  If the lips are expanded horizontally, then, no matter what happens
  vertically, the sound will be perceived as "stretched".

  If the lips are contracted horizontally and expanded vertically, then
  the sound will be perceived as "open-rounded".

  If the lips are contracted horizontally and not expanded vertically --
  either contracted vertically, or left in their neutral position vertically --
  the sound will be perceived as "close-rounded".

  If the lips are neither expanded horizontally nor contracted
  horizontally, then, regardless of what happens vertically, the sound
  will be perceived as "neutral" in regard to rounding.

  He doesn't give examples of languages making these distinctions;
  if you say Swedish does so, I'll believe you.

  >> I don't think anyone has even proposed that Nasal vs. notNasal
  >> can be given a third value.
>
> How about oral vs. nasal approximant vs. nareal fricative? ... Hell, I
> can even pronounce nareal *trills*! :)
>
> (Of course, I can only make these work if the oral component is a
> stop. But nevertheless, it's certainly possible to have more than two
> values of nasality...)
  Well, that's true, but fricatives and trills and stops aren't vowels.

  I was saying that nasality or the lack of it _for vowels_ was an
  all-or-nothing thing.

  And I was wrong.

  On pp. 291-295, section 10.9 "Nasal Vocoid Articulation", Laver says
  there are two degrees of nasality of vocoids in the Applecross dialect
 of Scottish Gaelic. But his best evidence is in Palantla Chinantec an
Otomanguean Mesoamerican language -- he quotes a reference whith a minimal
triplet, ?e 'leach', ?e~ 'count', ?e~~ 'chase', all identical in tone. He also
mentions Breton and Bengali as well.

  >> "consonant harmony" if it occurs is likely to apply just to syllable
  >> onsets or just to codas; or, even, just to onsets of
  >> stressed syllables or just to codas of stressed syllables.
>
> Couldn't the frequent POA assimilation of nasal+plosive clusters
  > (and maybe some other sorts of clusters too) be considered a sort
  > of consonant harmony? I have a phonology sketch around which
  > extends this to almost all consonant clusters and also prohibits
  > certain kinds of POA combinations in successive syllables.

  "Harmony" refers to the situation where all of a certain kind of
  phoneme in a word (or part of a compound word) must come from a
  single one of two or more disjoint (or nearly-disjoint) lists.

  The kind of assimilation where two consecutive segments must come
  to resemble each other, or the earlier must take on some
  characteristic of the later, or the later must take on some
  characteristic of the earlier, or some epenthetic sound intermediate
  between them must be inserted between them -- that is not enough to
  call it "harmony".

  Nothing that happens just to clusters is really "harmony", by the
  definition I just quoted; it has to happen to separated phonemes as
  well somehow, to count as "harmony".

  > Sibilant harmony (that is, /s/ may not mix with /S/ etc) is,
  > however, the only obvious natlang case of consonant harmony I've
  > read about.

  Your memory appears to be better than mine.

  I believe I read that early stages of some major "dialects" of ancient
  Indo-European had "voicing" harmony for consonants.

  Other than that I can't be specific; I just have a vague notion I've read
  of other kinds.

  Thanks for writing,

  Tom H.C. in MI



		
---------------------------------
 Yahoo! Music Unlimited - Access over 1 million songs. Try it free.

Replies