| Section
VII |
| We turn to R-expressions now. We can categorize
arguments into two as (a) dependent arguments and (b)
independent arguments. By dependent arguments we mean
arguments which need reference from other arguments. Independent
arguments do not need reference from other arguments.
Anaphoric arguments and pronouns are conceived of as dependent
elements here and expressions like ram, kitab
etc. are viewed as independent ones. |
| (171) |
|
 |
|
|
| Although pronouns and anaphors both need
reference, they must be assigned reference in different
ways. That expressions like ram , kitab
etc. are referentially independent is to be treated as
definition of such arguments. |
| We do not think there is any need to postulate
a condition to yield the result that expressions like
ram, kitab etc. are free. From our point of view,
this would account to saying that we need a postulation
to ensure that an independent element remains independent.
We would rather view the issue in the following manner: |
| If an independent element is assigned an
antecedent, then this would conflict with its basic property
that it is a referentially indpendent element. That is,
there would be a violation at the definitional level.
Hence we do not feel the necessity of postulating something
like "an R-expression is always free". To us
the question that seems to be of interest regarding these
expressions is the following: |
| How, at all, are R-expressions associated
with other expressions in the manner of co-reference?
In other words, the question is not how to ensure that
R-expressions ar free but how to explain that they are
sometimes coindexed with other R-expressions in grammatical
sentences. |
| We would like to mention that in the present
work our discussion of pronouns is confined to the so-called
personal pronouns and our discussion of R-expressions
is confined to entities like ram and kitab;
we have not studied variables and epithets. We agree with
Chomsky that repetition of R-expressions is a consequence
of the application of stylistic rules and may be, rules
of discourse and as such, may not be the concern of core
grammar. Beyond asserting again that we do not need a
postulation to guarantee the non-coreference between two
R-expressions in a sentence since we regard this as a
violation of the definition of the R-expression as a referentially
independent expression, we do not propose to say anything
more about R-expressions here. |
| Section |
| To conclude the binding theory arrived at
in this chapter is as follows: |
| (172)
(i) |
|
an anaphor must be bound,
where bound means be assigned mukhya as antecedent,
in P, P if contains a mukhya distinct from
the anaphor, if not, an anaphor must be bound in
P, where P is equal to or smaller than CP* |
|
(ii) |
|
a pronoun must be free, that is, not
bound, in P, must be assigned non-mukhya as antecedent
in  P'
, if P contains a non-mukhya, and may choose
its antecedent freely beyond P or if P is greater
than CP*, then, beyond CP*. |
|
|
| A few observations regarding the "where"
clause in 172 (i), namely, "where,
is equal to or smaller than CP*, are in order. As mentioned
earlier, that although we choose to use the term CP*,
it is a notional variant of tensed S. It may be worthwile
to speculate as to why tensed S is to be mentioned in
a specificaiton of the binding domain of anaphors. Whether
in the form of tensed S condition or NIC or accessible
SUBJECT, reference to tensed S becomes necessary in order
to specify the binding domain of an anaphor. While specifying
the domain of the pronoun, also -where it has a free choice
of antecedents - reference was made to tensed S. So it
is necessary for the theory to explain, why reference
to tensed S is so important. The following is a tentative
attempt in this direction. |
| Anaphors and pronouns are two distinct types
of dependent elements - one getting bound within as small
a domain as possible, the other searching for an antecedent
in a wider domain which, in sentence grammar, extends
to root S. We have treated these as directly following
from thevery basic conceptionlization of anaphor and the
pronoun. Now root S and the tensed S have one feature
in common, namely, the occurrence of tense. In the so-called
simple sentences, root S and the tensed S are, indeed,
the same syntactic constituent. In such a case, the tensed
S becomes the largest domain within which a pronoun must
receive its antecedent. If the tensed S in many instance
can be the largest domain in which a dependent element
like the pronoun can be assigned antecedent, then it stands
to reason that this domain cannot ever be smaller than
the one in which the anaphor, has to get an antecedent. |
| Imagine a situaiton in which the simplex
tensed S is .
As such the pronoun must be assigned antecedent in it
as in (173) below: |
| (173)
|
|
[ram ne mohan ko [uska
kalam] diya]
j
p j
P' = CP*
ram CM mohan CM his pen give+PAST
(Ram gave Mohan his pen.) |
|
|
| Suppose it is the case that the tensed S
is not the domain in which an anaphor has to be bound.
Then, in case of constructions like (174), the binding
domain of an anaphor would be larger than that of the
pronoun because the anaphor would then be able to choose
the mukhya outside the CP* as its antecedent. Thus, in
(174) the binding domain of the anphor would be the root
S while that of pronoun would be only the embedded S. |
| (174)
|
|
shyam ne kaha ki
ram
ne mohan ko uski
apni
shyam CM say+PAST COMP ram CM mohan CM his
self
kitab di
book give+PAST
(Shyam said that Ram gave Mohan his book.) |
|
|
| But this would go against the basic property
of the anaphor and the pronoun that we have considered
virtualy definitional. This is the reason why the possibility
of an anaphor choosing an antecedent beyond the CP* must
be ruled out. Therefore, the "where" clause
in 172(i) |
| |
