Re: Phaleran number
|From:||SMITH,MARCUS ANTHONY <smithma@...>|
|Date:||Saturday, August 4, 2001, 19:59|
On Sat, 4 Aug 2001, Thomas R. Wier wrote:
> > You mention a bimoraic minimum in the segment of the email I snipped. This
> > makes me wonder about your forms for 'domesticated owl': fû ~ fufû. Is the
> > vowel underlyingly long here? Or does the lengthening requirement disregard
> > the reduplicant?
> You bring up an interesting point. Obviously, if one were just given the
> singular data, these would be impossible to answer. But the plural makes
> it clear.
So you could conceivablly have a minimal pair that shows up only under
reduplication (or other affixation); one word has an underlying long
vowel, the other an underlying short vowel. In the singular, both have a
long vowel, but when reduplicated, one has a long vowel and the other a
UR sg pl
/fu/ [fu:] -> [fufu]
/fu:/ [fu:] -> [fufu:]
If the underlying form were /fu/, then although the reduplicant
> would still be [fu], the base would also be [fu], making it *[fufu]. So, yes,
> the 'emergence of the unmarked' effect applies also to vowels, as can
> also be seen in the 'black bear' example.
Is there any reason (other than the omniscience of the creator) to think
that vowel shortening is due to markedness considerations rather than just
that the reduplicant template is specified for a short vowel?
> Yeah. That just means that the constraint governing contiguity needs to
> be fairly lowly ranked.
There is also the locality issue: why do you copy the consonants that you
do rather than some other one. Obviously, it is because they are the
segments of the base that are closest to the reduplicant. The definition
of locality I used in my analysis of Pima reduplication dealt with strings
rather than segments, and that derived the fact that the segments were
contiguous, and that the reduplicant is inserted in the middle of
consonant clusters rather than one side or the other.
Anyways, nice work.