Íslenskt mál og almenn málfræði - 01.01.1981, Síða 175

Ritdómar
173
(11) Mary belives [John to have eaten the shark]
The resulting argument about the surface constituency of ACI in English has been
inconclusive.3
T’s argumentation for (6)b is substantially stronger than that which has been
made for the corresponding structure in English, strong enough, in fact, to make
the case. This is partly due to the nature of Icelandic, and partly to the superior
execution of T’s arguments.
His most compelling syntactic argument (pp. 389-393) is one involving the
positioning of certain adverbial elements, such as í barnaskap X, where X is a
possessive pronoun or adjective referring to the subject of the clause which the
adverbial modifies. If X = sínum, the adverbial may be inserted after Ján or after
hafa in (6)b:
(12) a Ég tel Jón í barnaskap sínum hafa étið hákarlinn
b Ég tel Jón hafa í barnaskap sínum étið hákarlinn
If X=mínum, on the other hand, it may go after Jón, but not hafa:
(13) a Ég tel Jón í barnaskap mínum hafa étið hákarlinn
b *Ég tel Jón hafa í barnaskap mínum étið hákarlinn
The constituent structure of (6)b, where Jón is the direct object of the main verb
rather than the subject of the complement, provides a clear explanation for this
difference: in (12) the adverbial modifies the subordinate clause, and so can
appear in various positions within it, while in (13) it modifies that main clause.
In (13)a, if Jón is a member of the main clause, the adverbial can follow it and
still be in the main clause. But in (13)b, where the adverbial appears between
members of the subordinate clause, it is impossible for it to be in the main clause.
Given the generally valid principle that adverbials must be members of the
clause they modify, the ungrammaticality of (13)b follows. If we asume a con-
stituent structure analogous to (11), the facts are inexplicable.
The form of this argument is not original with T: It was originated by Kuno
(1972), and was applied to English by Postal (1974:146-154). But it hasn’t worked
very well in English, partly because examples similar to (13)a tend to be rather bad
in English. See for example the discussions in Bresnan (1967:196) and Chomsky
(1981:99-100).
But, in addition, the investigators of English have not set up the argument as
carefully as T has: we do not find, for example, any attempt to establish a con-
trast such as that between (12) and (13). This contrast is necessary to establish
that it is clause-membership rather than some other restriction on adverb distri-
bution that is responsible for the facts.
T’s other major argument arises from the results of a psycholinguistic experiment
in which he used the classic ‘click location’ test to investigate the position of the
3 See, for example, Bresnan’s (1976) critique of the monumental attempt by
Postal (1974) to establish a constituent structure like (6)b instead of (11) for
English.