Re: [tied] *pYerkW+

--- Piotr Gasiorowski <gpiotr@...> wrote:

> On 2007-05-22 22:19, Sean Whalen wrote:

> > I don't
> > know of anything else exactly like tu > pi.
> However,
> > since tw > p does exist I think it's possible.

> So why doesn't it seem to happen? tw > p (more
> commonly dw > b)
> represents assimilation with coalescence, not
> partial metathesis of the
> type you have proposed.

But when described in terms of features coalescence
is simply the removal of a phoneme (with at least one
feature remaining that then moves to another phoneme).

That is, both metathesis of a feature and
coalescence involve movement of a feature. Met. can
move it to a non-touching phoneme, but this is not a
requirement.

> > That
> > is, w = +round but u = i+round.

> Not quite. In terms of distictive features, /i/ is
> [+front, +high], /u/
> is [-front, +high, +round]. Take away [+round], and
> you still have to
> add [+front] to get /i/.

I know that w can be described by more features than
+round. I'm advocating the theory that phonemes are
minimally specified in the minds of speakers, with
additional features automatically added if the
categories remain blank. So a typical language might
have:

V >-round -front -high > a
V+round > -front +high > u
V+front > -round +high > i

C > +[etc.] > t
C+voice > +[etc.] > d
C+contin. > +[etc.] > s

> > When the feature
> > +round moves to t (only specified by C) C+round >
> p.
> > If the round feature moves from u it leaves i;
> from w
> > it leaves nothing.

> Why not *j, to use the same logic?

Because w is specified with only one feature.

> Sorry if what I'm
> saying sounds like
> nitpicking, but metathesis normally has some kind of
> motivation:

The motivation in any given language is arbitrary
and may not be detectable from looking at (remaining)
evidence.

For this I did give a possible mot. (that rsn is a
long uncommon chain of dentals). A dissimilation of
rsn > rsm would seem fine to me; since it's followed
by u there is a switch instead of straight
dissimilation.

> it may
> increase the perceptual saliency of the output,
> repair a phonotactically
> flawed structure, result from a simple slip of the
> tongue, etc. It
> doesn't happen just because an abstract feature
> wants to move.

Abstract changes based on mental classification are
at least as important as those based on perceptual
phonology.

> /nu/ and
> /mi/ are neither confusible nor even
> impressionistically similar. I find
> it hard to imagine a natural mechanism transforming
> the one into the other.

It is the specific environment in this word that
makes nu>mi more likely.





____________________________________________________________________________________Ready for the edge of your seat?
Check out tonight's top picks on Yahoo! TV.
http://tv.yahoo.com/