Donnerstag, 10.30 Uhr
Consider the following examples, which illustrate a particular kind of scope ambiguity involving adverbial phrases :
(1)
a. ...dat Marie Jan dikwijls een boek liet lezen (Dutch)
that Mary made John read a book often}
i. often(cause(m,read(j,b)))
ii. cause(m,often(read(j,b)))
b. Noriko-ga Masaru-ni gakkoo-de hasir-ase-ta (Japanese)
Noriko Masaru school-at run-CAUS-past}
i. at-school(cause(n,run(m)))
ii. cause(n,at-school(run(m)))
(2)
The sheriff of Nottingham jailed Robin Hood for twenty years.
i. for-20y(cause(s,in-jail(r)))
ii. cause(s,for-20y(in-jail(r)))
(3)
daß Ali Baba Sesam wieder öffnete. (German)
i. again(cause(a,open(s)))
ii. cause(a,again(open(s)))
In each of the examples, an adjunct may take either scope over a complex predicate (`wide scope') or over only a part of this predicate (`narrow scope'). While the wide scope readings are easily accounted for in practically any treatment of adjuncts, the narrow scope readings are not. In (1), an incorporation, reanalysis, or argument composition analysis of the verb cluster "liet lezen" or the complex verb "hasir-ase-ta" fails to provide the subconstituent which is modified by the adverbial under the narrow scope reading. The narrow scope readings in (2) are even more striking, as these require an adverb to scope over parts of the lexical meaning of a verb, while at the same time there is no process of incorporation, reanalysis, or composition which might be held responsible for the ambiguity.
In this paper we provide an account of the ambiguity in terms of Head-driven Phrase Structure Grammar (Pollard and Sag, 94), a lexicalist, constraint-based theory of grammar. Our account rests on the following two assumptions:
a. Adjuncts are dependents of the heads they modify,
b. Semantics underspecifies scope relations.
zum Programm der AG 8
zur alphabetischen Übersicht der Abstracts
zur zeitlichen Übersicht