This article critically examines Rizzi's (1990, 1992) syntactic accoun
t of negative islands, originally noted by Ross (1984). We demonstrate
that, contrary to Rizzi's and Ross's claims, the distinctions between
referential and nonreferential expressions and between arguments and
adjuncts do no: play any direct role in determining the acceptability
status of negative sentences involving extraction. We argue that the p
henomenon is primarily controlled by a ban against extracting the focu
s of negation (irrespective of its. referentiality and argument/adjunc
t status) out of the scope of the negative element, and by a pragmatic
factor that bans questions that solicit uninformative answers.