Re: Sound changes - whither retroflex sounds and glottal stop?

On Mon, 24 Jul 2006 00:59:12 +0300, John Vertical <johnvertical@...>
wrote:

>Paul Bennett wrote:
>>
>>Eric Christopherson wrote:
>> > As for glottal stop, I know it can drop out completely, and combine with
>>  > other consonants to form glottalized ones, and I think in modern
>>Nahuatl  > at least it comes out as /h/.  I have an intuition that it might
>>become  > /N/, but that might be a stretch.
>>
>>The change /?/ -> /N/ I think is documented in Nenets (or a related
>>natlang?)
>>
>>Paul
>
>To my kno'lij, this change occured word-initially in at least Nenets and
>Nganasan (& quite possibly in a few other Samoyedic languages, too.) I
>thought it was /0/ > /N/ however, but checking sum' resorces there indeed
>seems to exist a /?/ phoneme in the languages.
I had heard that this change was /0/ (not /?/) > /N/ as well.  Tapani
Salminen's grammatical sketch
  http://www.helsinki.fi/~tasalmin/sketch.html
says that the change went the other direction, i.e. was loss of initial /N/
(but this is from a synchronic perspective; perhaps the /N/ was added and
then lost again).

The same sketch says there are two different phonemes realized as [?].  One,
|h|, occurs in word-final position and alternates with nasals (and seems to
be a development of word-final /n N/); the other, |q|, occurs
post-vocalically and alternates with obstruents (and seems to be a
development of them).  There's also a non-phonemic [?] after final /b m l
r/.  Anyway, none of these occur in the relevant word-initial environment
where /0/ > /N/ happened.


Maybe another chain shift would be fitting in Eric's situation?  If you've
got /0 ?/ contrasting initially, then you could do /0/ > /N/ (or whatever)
initially followed by /?/ > /0/.

Alex