--- In phoNet@y..., Piotr Gasiorowski <piotr.gasiorowski@i...> wrote:
> ----- Original Message -----
> From: richardwordingham
> > 3. Rhinoglottophilia, in which glottal or pharyngeal 'activity'
becomes associated with nasality. This can yield nasalised
consonants, or even plain nasal consonants.
> If I recall correctly, Arabic pharyngeals also tend to make
adjacent vowels nasalised; another nice example is the nasalisation
of *a > a{ng} before glottal h < *s in Avestan.

I suppose the Avestan /h/ could have been voiced, but it seems
unlikely. It seems then that glottality or pharyngeality is enough;
voicing is not needed.

A lot of back of mouth consonant features are best heard in
neighbouring vowels. I can recall that comment being made about
other vowel modifications next to Arabic emphatics and Arabic
dark /l/. Your discussion of PIE *k^e > Latin <ce> but PIE *ke >
Latin <ca> also comes to mind. Returning to nasality, I recall
reading that the nasality of Gaelic <mh> (/v~/) is best discerned in
an adjacent vowel.

> > Of these classes, only rhinoglottophilia could create nasals in a
language without nasals.
> Some languages use nasality as a prosody rather than a segmental
feature. Desano (spoken in Amazonia) has been described (Kaye 1971
[Linguistic Inquiry 2: 37-56]) as a language in which whole
_morphemes_ are [+/- nasal]. Each voiced segment (vowel or consonant)
has a nasal counterpart (e.g. v -- v~, d/r -- n, i -- i~), while /p,
t, s, k, h/ are completely transparent to nasal harmony. One could
imagine a historical trajectory like this: no nasals > suprasegmental
nasality > phonologised nasal segments. I don't know if this is known
to have ever happened, but it's at least a possibility.

Isn't nasal harmony similar to vowel harmony in principle? The basic
question would be the origin of contrastive nasality.

It seems, from http://www.language-museum.com/d/desano.htm, that most
affixes are quite short. The orthography seems to accord to the
scheme used in the phonolgy at
ethnocode=DES&doctype=detail&scale=six&version=1&allpages=1, but note
that the nasals stops are written m, n, n~. I could not see
nasalised e or w. I presume u- is for i-. I suspect 'qu' and 'gu'
are used for consistency with Spanish spelling. Do you know if any
Desano morphemes have nasal and non-nasal allomorphs? The internet
description at the Rosetta Project tantalisingly mentions nasal
spreading, so it could just be that nasal assimilation has been
bounded by morpheme junctures. (It would be interesting to know what
happened to mixed nasal/non-nasal roots; I predict that they would
have been discarded, reanalysed or completely nasalised.) I don't
know how significant it is that most speakers are trilingual.

What brought the whole question of the origins of nasals up to the
surface was some of the changes to nasals being blithely cited by the
Nostraticists on cybalist. I was considering crying foul, but I
though I had best do some checking first.