Print

Print


On Thu, 16 Dec 2010 13:54:55 -0500, John Vertical <[log in to unmask]>
wrote:

>Also typically, phonation changes begin at one end of the mouth, not at the
>center. In Mongolian, *b *d > p t is more common than *g > k; in Uralic, *k
>> g is more common than *p *t > b d; and cf. Western Romance, where voicing
>is universal medially but only k > g occurs initially, eg _gato_). 

Well, huh.  I've heard of this, but usually alongside the assertions that
voicing is easier in forward POAs (since there's more space to fill with
air), and so [g] is the voiced stop that one expects to be a gap, if any is.
 Yet here we have three processes where velars seem to be most
voicing-friendly.  What's going on?  For the k > g stuff, is it that an
original /g/ is less voiced than /b d/ in these languages, so in contexts
that partially voice the phonemically unvoiced stops, /k/ is more likely to
fall into the /g/ percept than /p t/ are into /b d/?  Even if that's the
case the Mongolian is still strange.  

>Mongolian
>also has *p_h > f (in Proto-Mongolian), *k_h > x (frequent) but retains /t_h/,
>AFAIK, everywhere.

Old Chinese had something a little like this too, *[m_0 n_0 N_0] > early
middle [x(_w) t_h x] (alongside an aspirated stop series which was stable).

Alex