<!doctype html public "-//W3C//DTD W3 HTML//EN">
<html><head><style type="text/css"><!--
blockquote, dl, ul, ol, li { padding-top: 0 ; padding-bottom: 0 }
--></style><title>Re: paradigms</title></head><body>
<div><br></div>
<div>Hi Dan,</div>
<div><br></div>
<div>Since my name came up in an earlier posting on this, I'll
make a couple of remarks, though I'm limited by (i) other pressing
obligations this week, (ii) an imperfect undertanding (on my part) as
to what the questions are, and (iii) a hazy recollection of the paper
you mention, with no time to look it up (see (i)). If the discussion
continues, I'll try to jump in again later.</div>
<div><br></div>
<div>Question 1:</div>
<div>(Apparent) non-compositional meaning is discussed in DM work on,
e.g., idioms. Is there anything here that is specific to paradigms? In
particular, is there any reason to suspect that existing approaches to
(apparent) non-compositionality won't extend to cover the cases that
are apparently "in paradgims"? My recollection was that one
of the cases you discussed had a structure [A B] for something filling
the role of a pronominal of some sort, while A's use elsewhere was as
a determiner. Is this qualitatively different from (other) idioms,
which may contain determiners without the (full) determiner
sematnics... "on the run", etc.?</div>
<div><br></div>
<div>Question 2:</div>
<div>Relatedly, I am not sure I understand what is meant by:</div>
<blockquote type="cite" cite>meaning corresponds exactly to what would
be expected for</blockquote>
<blockquote type="cite" cite>individual paradigm
cells.</blockquote>
<div><br></div>
<div>See the brief discussion of paradigms in Dave Embick's LI paper
on Latin, where he responds directly to the "periphrastic cells"
approach to Latin analytic passives in Börjars, Vincent &
Chapman. One point Embick makes is that paradigms are a convenient
device for "looking up" the output that corresponds to some
set of input, defined, say in terms of admissible feature combinations
and a base lexeme, but surely these are the observations we seek to
have a theory of, not the theory itself.</div>
<div><br></div>
<div>If the observation is that the feature combination [leave, F1,
F2] yields a word, while [leave, F1, F3] yields a phrase, simply
listing these as such in a list somewhere seems (to me) to add little
to our understanding. For F1 = past, F3 = neg, we get<i> left</i>
versus<i> did not leave</i> in English. Compare this with a language
that has negative and positive forms of the verb, and one would be
tempted to say that 'did not leave' is non-compositional (<i>do</i>
does not have its regular meaning) and has a meaning corresponding
exactly to the neg, past paradigm 'cell'. But rather than simply
listing the periphrastic form in a cell, these observations formed the
starting point for a syntactic analysis which seeks to explain the
distribution of the forms, why the latter is (observationally)
periphrastic, and why the particular pieces that are used have the
forms they do. From this perspective, the argument "for
paradigms" from meaning looks mis-constituted.</div>
<div><br></div>
<div>Since you asked earlier, in the Syncretism w/o Paradigms paper
mentioned by Heidi and others, I engaged Edwin Williams's theory of
paradigm structure (rather than the periphrastic arguments) because it
was one of the few 'paradigm theories' that I've seen (see also Andrew
Carstairs-McCarthy's work, and in phonology, work by John McCarthy and
Michael Kenstowicz; this list is not exhaustive) where paradigms are
more than a listing of input-output correspondences, and the actual
structure of the paradigm plays some non-trivial role. I happen not to
be convinced by these theories, for reasons I've put forward
elsewhere, but I'm less convinced (perhaps out of igorance) that the
arguments from periphrasis actually require paradigms in a deeper
sense than that (perhaps unfairly) sketched above.</div>
<div><br></div>
<div>Open to pointers to the arguments out there...</div>
<div><br></div>
<div>Best,</div>
<div><br></div>
<div>-Jonathan</div>
<div><br></div>
<x-sigsep><pre>--
</pre></x-sigsep>
<div>_______________________<br>
Jonathan David Bobaljik<br>
University of Connecticut<br>
Department of Linguistics, Unit 1145<br>
337 Mansfield Road<br>
Storrs, CT 06269-1145<br>
USA<br>
<br>
tel: (860) 486-0153<br>
fax: (860) 486-0197<br>
<br>
http://bobaljik.uconn.edu/</div>
</body>
</html>