syncretism w/o paradigms

[Heidi Harley]

hey jdb et al --

(below are some remarks of jonathan's that were meant to go to the list,
so don't be surprised if you haven't seen them before) --

You wrote:

 Hi Heidi:

 >But are the kind of meta-syncretism facts that you & Williams discuss
 >facts only about indo-european lgs like latin & russian, or does the
 >meta-syncretism effect show up in other lgs too?

 Meta-syncretisms as an effect are not constrained to I-E. But some of
 the hairier patterns that motivate extremely intricate analyses, and
 feature decompositions, are, apparently. The $64K question is where
 to draw the lines between accidental and deep patterns. Sorry, I
 didn't mean to give the idea that I (or the BBC paper) was suggesting
 relegating syncretism all to (synchronically) accidental homohpony.

no, no, I knew you weren't suggesting that; I just was interested in
whether the bigger phenomenon could possibly also be a historical kind
of fact.

 Does anyone know of acquisition studies of any kind about syncretism
 and features? My (very limited) understanding is that children
 overgeneralize regular to irregular, but that it is claimed that they
 do not overgeneralize agreement morphemes in the manner that
 underspecification would suggest. For example, they supposedly do not
 overgeneralize 3sg forms to other parts of the paradigm (I guess the
 tests exclude speech varieties where  "I's the b[o]y what catches the
 fish..." is acceptable), ditto for case. My naive impression is that
 most of this is based on production data, though, and a limited range
 of languages, where for some (English, French) zero morphology might
 confuse the issue.

 Comments?

i don't know the literature on this, though it would surely be
interesting and surely there must be data on this from e.g. the German
or Dutch CHILDES kids?

speaking of i'se the b'y dialects (like Nfld. Engl., spoken where I'se
from), it has always seemed to me that the spreading of -s throughout
the paradigm is symptomatic of the unmarkedness of 3sg in English...
but of course it really doesn't look in standard English like 3sg is
unmarked. However, without negative values, the feature geometric
approach of mine and betsy's really *can't* have a marked 3 person sg
form -- any VI that realizes a bare RE. or an RE with just an Indiv
node dependent (for sg.) would be eligible for insertion in geometries
for every person. (Of course, if we allow negative values, which maybe
we have to anyway, there is no problem). Without negative values, I see
two ways out of this:

English geometries other than 3sg are subject to a radical
Impoverishment rule reducing them to a bare RE node (Bonet-style
retreat-to-the-unmarked Impoverishment)

  (this is for a marked treatment of -s)

-s is unmarked; there is one or more zero-morphs realizing the other
present tense agreement features.

   (this would be an unmarked treatment of -s)

of course the latter would predict the i'se the b'y dialects well, but
it doesn't predict the 2nd person/pl elsewhere-looking forms in the 'to
be' paradigm (confirmed by the 'aren't I' inversion.

anyone remotely sympathetic to the latter?

:) hh



 -Jonathan


 --
 _______________________
 Jonathan David Bobaljik
 University of Connecticut
 Department of Linguistics, Unit 1145
 337 Mansfield Road
 Storrs, CT 06269-1145
 USA

 tel: (860) 486-0153
 fax: (860) 486-0197

 http://bobaljik.uconn.edu/