Verbs selecting lexically Case-marked subjects
Several studies have presented tests for syntactic subjecthood in Icelandic. These tests involve a range of phenomena including the binding of reflexives and reciprocals for all speakers, coordination, expletive insertion (það), correspondence with PRO in control structures, and a number of superficially heterogeneous word order facts. Though these properties often coincide with the nominative Case NP with which finite verb forms must agree morphologically, this is not always the case. Here we mention only three of these properties as an illustration.
For many Icelandic speakers, reflexive binding is subject-oriented; thus, nonsubjects cannot serve as antecedents to reflexives. (1a) gives the standard nominative subject case; (1b) shows that a nonnominative NP may also bind a reflexive.
(1)
It is assumed that this NP is also a structural subject, like its nominative counterpart. (1c) illustrates the fact that nonsubjects cannot bind reflexives. Note that although subject-oriented binding of reflexives is not observed in all idiolects of Icelandic, the fact that it holds in some is enough to establish the argument for lexically Case-marked subjects.
A second property of nominative subjects shared with certain lexically Case-marked NPs is that in yes/no questions, the NP immediately follows the finite verb. This parallelism follows if both Haraldur in (2a) and Haraldi in (2b) are structural subjects.
(2)
A related property is that both NPs occur in so-called raising constructions where the nominative NP is assumed to be the structural subject of the “seem”-type verb but is interpreted as the subject of the embedded infinitival. The same assumption about the status of the lexically Case-marked NP in (3b) is completely natural.
(3)
There seems to be no reason to analyze the trace of Haraldi in (3b) as occupying a nonsubject position if its lexical antecedent occurs in a bona fide subject position at S-Structure.
Given this analysis and our analysis of lexical Case as being imposed by selectional features of lexical heads (for instance, V, P, and A), Icelandic differs from both Russian and German in that lexical Case selection extends to subjects as well as complements. The fact that lexical Case selection is possible for subjects leads to further differences in the distribution of lexically Case-marked NPs in Icelandic, as we will discuss in the following section.
