The symmetrical
predicates can take either of the following as antecedent: |
a) |
|
agent and an explicit target |
b) |
|
plural agent with or without target |
|
|
|
Consider the
following examples which validate (a): |
|
(111) |
|
iran ne iraq se yudh kiya
iran CM iraq CM war do+PAST
(Iran fought with Ira.) |
(112) |
|
gorbachev ne rajiv gandhi se vartalap kiya
gorbachev CM rajiv gandhi CM conversaton do
+PAST
(Gorbachev held talks with Rajiv Gandhi.) |
(113) |
|
(113) iran ne yudh kiya
iran CM war do+PAST |
(114) |
|
gorbachev CM conversation do+PAST
Consider (115) and (116) which provide suport for
(b): |
(115) |
|
|
(100) |
|
vartalap kiya
conversaton do+PAST
(Gorbachev and Rajiv Gandhi held talks (with each
other)). |
|
Since these
are symmetric predicates, when they have plural
agents, they need not occur with explicit targets
(themes) because the symmetrical nature of the predicate
expresses the same. However, if at all they do occur
with target, then, of necessity, the target has
to be an anaphoric element expressing reciprocity.
This accounts for the optionality of paraspar
in (115) and (116). Notice that the predicates in
(115) and (116) are agentive and paraspar,
of necessity, chooses agent as its antecedent. |
Why does paraspar
not choose perceivers as antecedents? Symmetric
predicates do not take dative subjects. Predicates
in dative subject constructions take perceivers
and themes alone as their arguments. Such predicates
do not take target as one of their arguments. But
since the symmetric predicate takes target, it never
occurs in the dative subject construction. The impossibility
of the occurrence of symmetric predicates in dative
subject construction rules out the possibility of
paraspar occurring in these constructions
and therefore of paraspar choosing a perceiver
as antecedent. |
Thus we find that paraspar,
apneap and svayam all choose agents
as antecedents. In our discussion we maintained
that svayam and apneap are in some
sense "human instrumentals" and paraspar,
a target. It amounts to saying that they have semantic
roles but, then, while motivating the distinction
betwwen A and N.A-apnaphors in Hindi we had maintained
that these are not arguments because they cannot
occur in argument positions. Recall Chomsky's remarks,
in this respect (Chomsky: 1981b: 103, 147). Discussing
sentences like "John is widely/generally belived
to be a liar", Chomsky observes that the optional
adverbial has the theta-role "agent". |
In view of the above,
(15), (52), (65), (86) and (87) can now be stated
as follows: |
|
|
|
|