Re: wffs vs graphs
Message-id: <>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Fri, 25 Feb 1994 17:06:02 +0000
Subject: Re: wffs vs graphs

I also would like to see grammatical formalisms which can handle more
complicated structures. Local folk suggested the following:
'Dynamic Predicate Logic', Jeroen Groenendijk and Martin Stokhof,
LInguistics and Phliosophy 14,39-100, 1991  Here, existential quantifiers
can bind free variables outside their syntactic scope, the idea being to
model pronoun linkages in 'donkey sentences' such as 'Someone was walking
in the park. He was whistling.' where the syntactic scope of the 'someone'
is taken to be the first sentence, but the later 'he' is clearly
coreferential with it. DPL is very NL-inspired, but an intersting aspect of
its semantics is that an expression is taken to refer to a relation on
environments, rather than a function from environments to truthvalues. Then
for example  M( Exist x Px )(g,h) iff  there is a k differing from g only
at x & M(P)(k,h). This enables a semantics to be defined which allows
existential quantifiers to 'collect' any free variables to their right in
the expression. DPL treats universal quantifdiers differently for
linguistic reasons, but it seems clear that you could do this for all the
quantifiers if you wanted to.

Although this is in the semantics, it seems similar to the kind of
structure that you are hypothesising for program fragments, in that instead
of defining a constraint on the entire parse structure (like the
'harnesses' Ive been talking about), you put more into the syntactic (or in
the above case the semantic) recursion rules so that these constraints can
be encoded by local conditions where these rules are applied. This still
seems perverse to me :-)

Best wishes.


Beckman Institute                                    (217)244 1616 office
405 North Mathews Avenue        	   (217)328 3947 or (415)855 9043 home
Urbana, IL. 61801                                    (217)244 8371 fax  or