--- In phoNet@y..., Piotr Gasiorowski <piotr.gasiorowski@i...> wrote:
> Nasal harmony in the Eastern Tukanoan languages (where Desano
belongs) operates in a rather complex way, since suffixes can
themselves be [+nasal] or [-nasal], as well as neutral "[0nasal]"
with nasal and nonnasal allomorphs, in which case nasality or non-
nasality spreads across morpheme boundaries. See the following
article by Michael Kenstowicz, who describes the situation in
> http://web.mit.edu/linguistics/www/kenstowicz/ijal.doc

Thank you for the reference. I misanalysed my 'corpus'. It seems
that Desano doesn't (redundantly) mark nasal vowels after m, n and
n~. Grrh! (unnasalised!)

Nasal harmony doesn't seem complex in Barasana. All morphemes seem
to end in vowels. As a rule with six exceptions, one mora morphemes
are [+nasal] or [0nasal] - too short to stop the progressive
nasalisation - and longer morphemes, including all noun and verb
roots, are [+nasal] or [-nasal] - long enough to stop nasalisation
spreading. The six exceptions are:

a) two similar one mora 'classifiers' ([-nasal]);
b) two VCV suffixes which start by duplicating the preceding vowel
c) two other VCV suffixes ([0nasal]).

(I assume the paper's 'morpho-phonemic transcription' uses the same
notation for [0nasal] and [-nasal] - otherwise the diminutive -áka
isn't actually an exception but the (infinitive?) suffix -ré is!
Mind you, I'm grateful that the paper is available on-line. My only
text on the 'well-known exogamic and multilingual Vaupés system'
makes no mention of nasal harmony.)

It was interesting to see that the names Francisco and Maria are
converted to [-nasal], but borrowing is not a diachronic process.

I think the morphological complexity lies elsewhere, e.g. the tone.
Negation doesn't look simple, either!