vowel harmony extension? and voiced velar fricative

On Thu, 26 Oct 2000, Yoon Ha Lee wrote:

> Vowel harmony in Turkish is way cool.
>
> But I got to wondering, is there any such thing as *consonant* harmony in
> any language?  I dunno, maybe a word would only have all palatals, all
> labials, etc.  Or go the other way and words would only have all
> fricatives, nasals, etc.
>
> Does this exist anywhere?  Or are there phonetic/speech-production/???
> reasons it probably wouldn't exist?  Because if not, I think I'll try it
> out sometime.  :-)
Claims of consonant harmony have been advanced for languages like
Chumash and Athabaskan. Stems containing sibilants ([s,S,ts,tS])
compel sibilants in affixes to agree in place of articulation. So if
a stem has [...S...] any affixes with sibilants will have [S] or [tS]
as well.

Some hesitate to call this harmony, though. One way of thinking about
how harmony works is to visualize sequences of syllables as sequences
of vowel articulations punctuated by consonants. Under such a view,
the vowels are adjacent to each other, but the consonants never are.
So the Spanish word 'casa' [kasa] might be represented as follows:

        |-a-|-a-|
        |k|V|s|V| (the "V" is a place holder for vowel features)

        'casa'

From the diagram, you can see that the successive vowels are actually
temporally adjacent, but that the consonants are not. This view of the
syllable makes a lot of sense when you consider phenomena like
coarticulation--why the /k/ in 'key' is further forward than the /k/
in 'caw'. In the first, the /k/ is superimposed upon an /i/, while in
the second, it is superimposed upon an /A/ or /O/:

        |-i-|     |-O-|
        |k|V|     |k|V|

        'key'     'caw'

Harmony occurs when vowels within a given prosodic domain agree with
respect to some aspect of articulation (tongue root advancement /
retraction, rounding, fronting/backing, etc). Under the view of
syllables given above, what is happening in harmony is the sharing of
features among *adjacent elements*. Since consonants cannot be
adjacent under such a view, there can be no consonantal "harmony".
So the explanation for consonantal "harmony" has to be found
elsewhere. Now I'm not sure what that explanation would be, but it is
true that consonantal harmony is far more rare than vowel harmony, and
in the languages where it does seem to occur, it is far more
restricted in its operation.

The nasal harmony mentioned by Kristian and others can be seen as a
property of consonants which can be shared by vowels. Likewise, Arabic
Emphasis (usually pharygealization or velarization of consonants) can
also be construed as a property shared among vowels and consonants.
The odd thing about sibilant harmony is that it involves a feature not
realizable on vowels.

This brings up the /h/ ~ [ATR] connection wondered about by And. In
Jakobson, Fant, and Halle (1962), /h/ is assigned the feature
[+tense], which for many people is equivalent to [+ATR]. So the
presence of [+ATR] seems to enhance the tense quality of an /h/.
However, I'm inclined to view the advancement of /a/ to /&/ as being
brought about by a coronal or fronting gesture left over from the now
deleted /s/. The consonant articulation is reduced to the point of
non-frication, but the tongue fronting gesture remains.

But to be perfectly frank, I don't know what's going on in those
dialects of Spanish.

Dirk

--
Dirk Elzinga
dirk.elzinga@m.cc.utah.edu