Re: Non-Finno-Ugric/Turkish vowel harmony systems and the evolution thereof.

At 9:34 AM -0700 10/22/02, Steven Williams wrote:
>I understand the hows and whys of vowel harmony, and
>like the idea a lot. I had a system going, a rather
>nice system, for the earlier versions of my current
>conlang, but I had abandoned it, citing my magic
>'historical reasons' wand and to give me an excuse to
>introduce a lot of irregularity with the rubble of the
>collapsed system.
>
>Can anyone give me examples of vowel harmony systems
>that are neither Finno-Ugric nor Turkish?
Nez Perce has a vowel harmony system which looks very different from both
Finno-Ugric or Turkic systems. I've posted summaries of the system before, but
here it is again in a nutshell.

Nez Perce harmony is of the "dominant-recessive" type. This means that if a
morpheme in a form has a vowel belonging to the dominant set, all vowels in the
form will belong to the dominant set. Otherwise, all vowels will be drawn from
the recessive set. The dominant set in Nez Perce is /i,a,o/, and the recessive
set is /i,e,u/ (/e/ is actually low front, and /u/ can be unrounded). Here are
some forms which show the alternations:

wu:lelikepese
wu:le- lik  -epe        -s              -e
ride-  move -into:brush -stem:formative -singular
(I) am riding into bushes.'

wo:lalikapasaqa
wu:le- lik  -epe        -s              -e        -qa
ride-  move -into:brush -stem:formative -singular -recent:past
'(I) rode into bushes recently'

In this pair, the forms are identical with the exception of the recent past
suffix in the second. This suffix has a vowel drawn from the dominant set, so
this conditions the change from recessive to dominant in the remaining
morphemes.

The curious feature of Nez Perce is that the two sets of vowels don't seem to form
natural classes. In particular, the presence of /i/ in both sets is rather
perplexing. Stems which contain /i/ can trigger dominant harmony or fail to
trigger it. Compare the forms for 'paternal aunt' with those for 'mother',
which is almost idential to it in form:

ci:c  'paternal aunt'
na?ci:c  'my paternal aunt'
ci:ca?  'paternal aunt!'

?i:c  'mother'
ne?i:c  'my mother'
?i:ce?  'mother!'

In the first set, the stem _ci:c_ triggers dominant harmony, while the second set
the stem _?i:c_ does not. In a paper I wrote in grad school, I analyzed this
harmony as being of the ATR/RTR type, with [-ATR] being the active feature. The
presence of /i/ in both sets is due to a constraint which prohibits [-ATR] from
being realized on a high front vowel.

>How would
>vowel harmony evolve in a language previously without
>such a system?
I think John's answer is essentially correct. Consonant gestures can overlap
vowel gestures, so in a real sense, vowels are always adjacent. Minimizing the
number of distinctive gestures will thus drive harmony.

Dirk
--
Dirk Elzinga                                               Dirk_Elzinga@byu.edu

"It is important not to let one's aesthetics interfere with the appreciation of
fact." - Stephen Anderson

Reply