Orð og tunga - 01.06.2006, Page 49

Matthew Whelpton: Argument Structure
47
1. Rely [___on NP]
a. John relied on him.
b. *John relied under him.
c. *John relied against him.
2. Tell [___NPacc SubClthat]
a. John told him that Mary had left.
b. *John told to him that Mary had left.
c. *John told that Mary had left.
3. Say [____(to NP) SubCthat]
a. *John said him that Mary had left.
b. John said to him that Mary had left.
c. John said that Mary had left.
This information on the number and type of phrases that must follow
the Verb is called a subcategorisation frame.
The kinds of information we see in subcategorisation frames il-
lustrates very clearly what I am calling the division of interests be-
tween theoretical linguists and lexicographers. Traditional dictionar-
ies do not require the listing of grammatical information and usually
do not list it except for part of speech and examples of use - e.g. Give
something to someone. The linguistic lexicon contains little but gram-
matical information and only those idiosyncratic elements of gram-
matical behaviour which are not predictable from general syntactic
rule.
However, two general developments have changed this signifi-
cant difference in focus between the two communities of language
researchers: first, the rise of corpora and computational modelling in
language research, esp. in dictionary development; and second, the
rise of semantics, in particular a clear and well-articulated field of for-
mal semantics alongside syntax.
2.2 Semantic Theory
The work of the philosopher Richard Montague (1974) began a tra-
dition of linguistic analysis which has established the importance of
compositionality for any serious formal theory of semantics - that is,
the property that the semantic interpretation of a sentence can be read
systematically off the syntactic structure of the sentence.