Rit (Vísindafélag Íslendinga) - 01.06.1970, Page 279

277
on the contrary to gá but not to springa. In German it applies
to both (die Uhr geht, die Uhr Iduft). But we are far from the
worst possible case, in which each word would be a law unto
itself. One of the interesting consequences of semantic exten-
sion rules is just that they reveal typological regularities of
languages in the field of semantics. It seems to happen more
frequently in English than in Swedish or German that verbs
taking actions as objects can also get propositions: cf. persuade
(where Swedish distinguishes övertala and övertyga), know
(kunna, veta), insist (on) (yill nödvándigtvis, hávda), agree (on)
(gá med pá, vara ense om), decide (besluta, komma till slutsatsen),
expect (vánta sig, anta)—similarly, figure (on)—regret (ángra, be-
klaga). Examples of the opposite case, where English but not
Swedish must draw the distinction, appear to be quite rare.
A further area in which the kind of semantic rules I am
advocating may be needed is that of idioms. Treatments of
idioms in generative grammar have usually involved listing
them in the lexicon as whole subtrees with a global meaning
but with a syntactic and morphological analysis into consti-
tuents. Kick the bucket would be listed with its full constituent
structure (VP dominating V-NP, etc.). In cases like to pull
X’s leg provision has to be made for part of a tree to be listed,
with the remainder to be freely selected from the lexicon.
Suppose that we instead say that the grammar has meaning
rules like ‘kick the bucket’ => ‘die’, and that idioms are freely
generated by the syntax, but subject to these special semantic
extension rules which superimpose a derived meaning on
them. There is one crucial difference between the subtree-
listing and the meaning-rule approach. The subtree-listing
approach, in which idioms are listed separately, imposes no
constraints whatever on the syntactic structure of idioms.
There is no reason to expect, for example, that idioms should
conform to the normal syntactic structure of the language.
In fact, there should be a liberal sprinkling of syntactic mon-
sters among the idioms of every language, e.g., idioms like
*bucket kick the. These are exactly as easy to list as subtrees as