Matt wrote:

>Nik Taylor wrote:
>>Don't forget nasalization.  There are languages wherein all vowels =
>>be either nasal or oral.
>Really?  I know of languages with nasal *spread* (Mixtec, for example),
>but I've never heard of a language with nasal *harmony*.  Examples?

Below is what is written in "An Introduction to Phonology" by Francis=20

-----<start quote>-----
Desano, a Columbian language of the Amazon basin, is a standard=20
example of a language which treats nasalization as a suprasegmental=20
element. Native Desanao words consist of morphemes which are either=20
all oral or all nasal:

[10.55] Desano (Kaye 1971)
        _nasal_                      _oral_
        [w~a~i~]     'name'          [wai]      'fish'
        [no~hso~]    'kind of bird'  [johso]    'kind of lizard'
        [mi~ni~nu~n] 'small round    [wyarirul] 'large round
                     thing'                     thing'

The voiceless segments /p t k s h/ are NEUTRAL. They neither capable=20
of bearing the property of nasalization, nor of inhibiting its=20
spread. They are 'invisible' to the nasalization rule. But voiced=20
segments are all capable of bearing nasalization: they have both oral=20
and nasalized versions:

[10.56] _Oral_     _Nasal_
          v          v~
          b          m
          d/r        n
          g          N
          j          J
          w          w~

In Desano, nasaliztion is not simply a property of certain=20
consonants. It is extracted from the segmental tier and placed on a=20
separate suprasegmental tier. To represent the Desano situation rules=20
like [10.57] are needed:

        [+nasal]             [-nasal]
         / | \                / | \
         w a i -> [w~a~i~]    w a i -> [wai]

Where there are neutral voiceless consonants, the mapping proceeds in=20
the same way, only this time being blind to the presence of voiceless=20

         /|  \
         johso -> [Jo~hso~]

-----<end quote>-----

There you have it Matt! Isn't that nasal harmony?

-kristian- 8)