On Fri, 3 Sep 2010 21:57:25 -0700, Garth Wallace <[log in to unmask]> wrote:

>I'd say, though I'm hardly an expert, that affricates pattern as a
>third type of release (no release, aspirated, affricate). The palatal
>affricates are a partial collapse of the system, the palatal aspirated
>stops having merged with the affricates and the palatal unaspirated
>stops having merged with another POA (e.g. /c/ -> /k/). How's that

Hm.  It's something to consider.  The cases I can think of where "affricate"
is sensibly treated as a release, it's mostly developed from noisy
aspiration; so maybe whatever process created the current aspirates pushed
old aspirates to affricates.  That said, AIUI it's only voiceless aspiration
(i.e. genuine aspiration, not breathiness) that does this, so I don't know
where to get the voiced affricates then.  

Moreover, if one goes with an origin in a release contrast, then I'd really
want the fricatives to show that release contrast too!  The fricatives match
up very nicely with the stops and affricates taken together; it seems quite
tidy to assume developments with  p : pf) :: p\ : f, t : ts) :: T : s, etc.,
and if we don't do something such we're left with a huge surfeit of
protofricative PoAs.  So what in the world could do that?  This is more or
less the [+-strident] of some feature systems, but I don't know of any real
evidence that [strident] is actually a homogeneous thing across different
places (parsimony of the model is not evidence!).