Íslenskt mál og almenn málfræði - 01.01.1981, Side 173

Ritdómar
171
available to them.1 In the course of trying to constrain linguistic theory, it is easy
to overshoot the mark, or take a wrong turn, and propose conditions that rule out
kinds of analyses that are actually necessary in some language.
The theoretical interest of the book stems from the extent to which T has been
able to use Icelandic to show that this has happened with a number of current
theories. Many of the analyses which T motivates were originally proposed for
English, but later discarded by certain theorists, who proposed linguistic theories
that would exclude them as possibilities for human languages. But in Icelandic the
evidence for these analyses is often substantially better than it is in English, enough
so that the theories excluding them appear to be disconfirmed.
For example, Rosenbaum (1967) analysed complement clauses as NP in English.
But they have various non-nominal properties which have led many linguists, such
as Emonds (1970), Jackendoff (1977), Koster (1978), and others, to argue that
they are not NP, and to furthermore propose or subscribe to theories of phrase
structure which entail that complement clauses cannot be NP in any human
language. Thus Koster (1978:55) writes „in my opinion, it is one of the important
results of Emonds (1970) that what we should expect on these grounds [the prin-
ciples rendering it impossible for a clause to be an NP] appears to be correct:
Ss never behave like NPs.”
But T shows that the nominal character of complement clauses is considerably
clearer in Icelandic that it is in English. Ielandic complement clauses conjoin with
ordinary NP, for example, which English complement clauses do not:
(10)a Jón spurði þessarar spurningar og [hvort María væri farin]
b *John asked this question and wether Mary had left
It would appear that clauses can be NP after all.
T’s conclusions here have effects for linguistic theory: it must be broadened,
or left broad, so as to allow clausal NP. But the effect is not very far-reaching,
and does not have strong implications for the analysis of other languages. It might
well be that English, for example, has complement clauses that are not NP.
But there are other instances where T’s conclusions have considerable wider
impact, bearing on the nature of syntactic representation in all languages. These
appear in part II. Perhaps the most important concerns the nature of the under-
lying structures of the Equi and Raising constructions.
Consider again (5) and (6), typical pairs of sentences related by Equi and
Subject-to-Object Raising, respectively:
(5) a ?Ég skipaði henni [að hún færi]
b Ég skipaði henni [að fara]
(6) a Ég tel [að Jón hafi étið hákarlinn]
b Ég tel Jón [hafa étið hákarlinn]
In T’s analysis within the Aspects theory, the infinitives in the (b) examples have
1 See Chomsky (1980), Stich (1981) for an interesting discussion of these
limitations and their significance.