| (117) |
|
anaphors choose agent
as antecedent |
| (118) |
|
A-anahors choose perceiver as antecedent
when agent is not present |
| (119) |
|
the N.A/-anaphor apna chooses possessor
as antecedent |
| (120) |
|
the N.A.-anaphor svayam relates to
the NP it immediately precedes. |
|
|
|
| Although in
certain respects this formulation is similar than
the earlier ones, it is worthwhile to explore if
further simplification is possible. In fact, it
is quite imperative to do so, because, first, at
least two construction - specific statements have
been made and secondly, (118) is unsatisfactory
because there is no real choice: the constructions
where agent occurs are those where perceiver cannot
and vice-versa. This is a consequence of the kind
of predicates that are involved. The agentive predicate
does not take a perceiver and the perceiver predicate
does not take an agent. (118) exists to account
for the dative subject construction alone and (110)
to account for the possessive phrase alone. Therefore,
an attempt to look for simpler formulations is certainly
in order. |
| Before we
proceed to address ourselves to this task, we must
separate (117) - (110) from (120) because, whereas
(117) - (119) make use of thematic roles, (120)
does not. So, we treat (117) - (`119) as belonging
to a single set and (119) to another. We will say
nothing more about (120) for the moment and will
return to it later when we discuss the binding domain. |
| The term
perceiver is being used here to refer also to a
human being in a certain state experiencing a certain
feeling, emotion, perceiving some abstract object
or recipient of an impression. Viewed thus, possession,
can be legitimately viewed as the possessor being
in a state of possession. It is quite in order for
us to merge (118) - (119) and observe that A-anaphors
and N.A-anaphor apna choose the perceiver as antecedent
when agent is not present. The question of a choice
of this kind does not really arise in case of N.A.
apna because we know that it always takes possessor
as antecedent. Thus this part of our observation
applies vacously in case of N.A. apna. Now (118)
and (119) can be coalesced into (121): |
|
| (121) |
|
A-anaphor and N.A-anaphor apna choose the
perceiver as antecedent in the absence of an agent.
|
|
| This is nothing
strange about it. Consider muninka krodh (the Sage's
anger), which can be ambiguously interpreted as
the permanent attribute of the sage or his temporary
s tate - the sage is the passessor of a permanent
attribute or state. If under the influence of anger,
the sage, who is ordinarily a calm person, cursed
someone, one can justificably use the phrase munika
kordh and interpret it as the sage's being in a
temporary state of extreme anger. Thus interpreting
"possessor" as "being in a certain
state" is not unnatural at all. |
| Although (121) makes it
possible to eliminate one construction specific
statement from the theory of binding, namely (119),
the other, (118), remains, at least in essence,
since it refers only to the dative subject construction. |
| Turning (117) and (121),
we introduce a concept which we may be called mukhya.
It mukhya is not a syntactic notion like "subject"
and is not related to other syntactic notions like
c-command. It is the argument bearing the agentive
or the perceiver theta-role. This notion, makes
it possible to arrive at the following general statement:
|
| (122) |
|
the antecedent of
an anaphor is the mukhya,
where mukhya is the argument bearing
the
agentive or the perceiver theta-role, |
|
|
| If the arguments of a
predicate are to be arranged in a hierarchy of importance,
then intuitively speaking, the agent, if it is an
agentive predicate, or the perceiver, if it is a
stative predicate, is the most important argument.
This idea of designating an argument as the most
important one of a predicate is not really unfamiliar
in grammatical literature. Classical grammars of
Sanskrit did have a similar notion in karta ,
which they called svatantra (independent) (Kiparsky,
1982). We do not think that karta and mukhya
are notational variants as theoretical concepts.
The possessor in the possessive phrase which we
have designated as the mukhya is not karta.
Karta is a karak relationship, that
is a relationship that obtains between a kriya
(verb) and its arguments. Sanskrit grammars did
not include the possessor-possessed relationship
in the karak system because they maintained
that it is not an argument - verb relationship. |
| There seems to be at least
one syntactic issue outside the domains of binding
and control which can be satisfactorily accounted
for in terms of mukhya. In Hindi, agreement
relationship holds between an NP to which no lexical
CM is attached and INFL (eventually, verb). However,
this is not adequate to account for agreement in
the following sentences because in each of the following
there are two NPs to which no lexical CM is attached. |
|
|
|
|
