| We
should like to respond to this criticism as follows. To
place this work in the correct perspective, we may note
that it is a revised form of a 1990 work. We were skeptical
about three fundamental ideas of the existing theory of
the LGB-KOL kind of theory - call it GB theory. First,
we were skeptical about the goals of the existing theory
of binding. It is directly and explicitly concerned with
what can be assigned as antecedent to an anaphor and what
cannot be assigned as antecedent to a pronoun. This concern
is different from the traditional one concerning the anaphor
and the pronoun. Traditionally both anaphors and pronouns
have been viewed as elements without the necessary semantic
substance needed for independent reference, which is why
they must be associated with some other element in the
sentences in which they occur. That the pronouns and the
anaphors are not semantically empty in exactly the same
way is inconsequential; what indeed is not is that as
far as the need for reference is concerned, there is not
strong reason to distinguish between the two. We found
this view more persuasive than the contemporary one which
leads to identification of antecedent for an anaphor and
of disjoint reference for a pronoun, i.e., what cannot
be the antecedent of a pronoun. This view certainly does
not suggest that pronouns are referentially independent
entities; by identifying what cannot be the pronoun's
antecedent in a sentence, it indirectly indicates what
can indeed be. But then since both pronouns and anaphors
are referentially dependent elements, we find no strong
justification for a theory to have explicitly different
goals, but arguably implicitly the same one, with respect
to their interpretation in a sentence. For us the objective
of the theory concerned must be explicitly the same for
both anaphors and pronouns. Thus our skepticism regarding
the existing theory is at a basic, conceptual level; it
is one of principle, not one of dealing with counter examples
- whether these are real or merely apparent is not even
a relevant matter. |
| The
second reason of our skepticism about the existing theory
was its approach to the antecedent-PRO relation. The theories
of binding and control have different concerns: the first
with antecedent and disjoint reference of anaphors and
pronouns respectively, and the second with antecedent
of PRO. PRO has both anaphoric and pronominal features;
and anaphoric and pronominal features are features of
relevance to the assignment of antecedent or disjoint
reference. Therefore it is to be expected that antecedent
-PRO relation, that is, the interpretation of PRO be dealt
with by the binding theory itself. Instead, what the existing
theory has is a taxonomy of theories of binding and control. |
| The
third reason of our uneasiness relates to the characterization
of the binding domain, which is the "governing
category" (GC) in terms of the LGB framework and
"Complete Functional Complex" (CFC) in terms
of the KOL framework. The notion of "government"
figures in both. Delineation of domain is not specific
to the issue of antecedent for anaphors and disjoint
reference for pronouns; elsewhere domain is specified
in more direct and more natural terms; "clause",
"phrase" (such as S, S, NP0 etc.) . This is
not true of either GC or CFC. GC and CFC are clearly
not categories of the same type as clause and phrase;
these are special syntactic domains and are specific
to the binding theory. This calls for an explanation,
but the theory provides none. |
Chomsky
has observed that the anaphor's antecedent is intuitively
the most prominent nominal element in a certain local
domain. This observation prompted us to explore whether
there could be other ways of executing this idea, apart
from c-command. Hierarchy among nominal elements can
be captured in non-structural terms as it is in the
Paninian system, where karta can be viewed as superior
to other karakas (semantic role or theta role bearing
arguments) in this system since it is swatantra (independent)
with respect to them. We thought of considering this
approach seriously under the well- known assumption
that articulation and evaluation of plausible alternatives
irrespective of whether they eventually turn out to
be inadequate and incorrect, is desirable, often a necessity
in rational enquiry. |
| |
|
|
|
