Date sent: Thu, 14 Oct 1999 16:50:00 +0100
Subject: binding & referential dependency
From: A Rosta <[log in to unmask]>
To: Word Grammar List <[log in to unmask]>
Send reply to: [log in to unmask]
> I noticed me saying "me being me" yesterday, & this
> got me thinking about apparent Principle B&C violations
> like
>
> Me being me, it's only to be expected I'm always miserable.
> Bill_1 being Bill_1, it's only to be expected he's always miserable.
> Him_1 being him_1, it's only to be expected he's always miserable.
> Him_1 being Bill_1, it's only to be expected he's always miserable.
>
> but
>
> *Bill_1 being him_1, it's only to be expected he's always miserable.
>
> [all HIMs are nondeictic in these exx]
>
> Below I describe what seems to me to be the correct analysis, and then ask
> a formal/analytical question that's been bugging me.
>
> If the binding principles are understood merely as constraints on
> coreference, then these exx are problematic; or rather, the principles are
> incorrect. But it seems to my that the principles are correct, but that
> you have to understand them in terms not of coreference but of referential
> dependency, i.e. an asymmetrical "X gets X's referent from Y"
> relationship. Then Principle C would say that "R-expressions" are not
> referentially dependent,
R-EXP: hasn't got a sense; has got a referent (but we don't know
what it is yet).
B would say that nondeictic pronouns are
> referentially dependent but not on anything "within its governing
> category" [replace that with formulation of your choice],
some R-EXP's (call them PRONOUN) are _b_ (for 'bound', a kind of
dependent) of some word, whose referent they share (some of these
also take a complement, whose _b_ they are).
> and A that
> "anaphors" are ref dep on something within its gov cat.
some PRONOUNs (HIMSELF, HERSELF, etc) must be _b_ of a codependent.
Notice I'm not having a class: ANAPHOR. It doesn't seem
necessary, since HIMSELF etc are only diff. from other pronouns
by their form and the fact that they are bound of a codependent.
Since the form is so hard to generalise, the putative category
ANAPHOR would have only one idiosyncratic property, which isn't
enough to justify using a category.
I hope you can understand this (I could always try explaining it
again). I think it works for the data, though its possible I've missed
something. What I'm saying is:
Some words have no sense. Instead of getting their referent in the
usual way, it is determined by the context.
Some of these rather special senseless words can get their
referent through the linguistic context: by being the 'bound' of a
word and sharing its referent.
Eight (or nine? I haven't thought about each other yet) of these very
special words can be the 'bound' of a codependent.
Jasper
"Him_1 being
> him_1" would be OK because the second "him" is ref dependent not on the
> first "him" but on the same antecedent as the first "him" is ref dependent
> on. On the other hand, "*Bill_1 being him_1" is bad because "him" is
> ref-dependent on "Bill".
>
> Now here is the snag. How can one formally define referential dependency
> to ensure that it is asymmetric and not merely the same as sharing the
> same referent? I'm looking for a declarative formulation, not one that
> brings in processing.
>
> Maybe such a formal definition is not possible, and what we instead
> have is a syntactic relation "antecedent of"/"ref-dependent of" that
> encodes procedural constraints on the reference assignment portion of
> utterance interpretation. This obviously would contrast with analyses
> based on coreference, where the constraint would be on semantic content.
>
> Views & opinions sought...
>
> I'll be back at email sometime next month.
>
> --And
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|