Íslenskt mál og almenn málfræði - 01.01.2011, Page 166

164
Asgrímur Angantýsson
only “null subjects” can be in the subject position. Two problems with this analy-
sis are that it does not account for the SF-like movements of XPs and there is no
obvious trigger for the SF. Perhaps SF should be viewed as an optional, stylistic
operation although it is not obvious how, or even to what extent, such phenome-
na should be accounted for in the syntax. However, it is clear that SF has syntac-
tic effects and obeys syntactic principles (e.g., it depends on subject gaps, it pre-
cludes the appearance of the expletive, etc.). I suggest that SF is restricted to cases
of head movement in operator environments and that “stylistically fronted” XPs
should be accounted for as Topicalization in clauses with a subject gap. However,
this analysis is not without problems.
The hypotheses regarding direct structural relations among different con-
structions did not gain any support. The main result for Icelandic was that the
speakers who were more willing to accept ET and SF than others were also more
willing to accept the Adv-Vfin order. In my view, this supports the idea that the
Adv-Vfin order in Icelandic involves adverb fronting rather than lack of verb
movement. Recall that we discovered that the conditions for the Adv-Vfin order
depend heavily on the clause type. I also observed that in certain types of sen-
tences the same speaker can easily use both SF and Expletive Insertion in cases
where they do not like to leave the subject position empty. This can be viewed as
support for a P-feature analysis of SF and Expletive Insertion. Recall, however,
that we observed that many subject gaps can easily be left empty and that SF and
Expletive Insertion are not always interchangable, which is a problem for the P-
feature analysis. In the Faroese and Ovdalian data, no significant correlations
between the morpho-syntactic variables under investigation were found.
The Rich Morphology Hypothesis (see e.g. Bobaljik and Thráinsson 1998)
sheds light on the difference between verb placement in embedded clauses in
Icelandic and Mainland Scandinavian, while allowing for V-to-I in lan-
guages/dialects with poor verbal morphology, but it does not explain why excep-
tional V2/V3 depends on the sentence type. Such differences must be due to dif-
ferent structures or “featural content” above the IP, i.e. at the CP-level. A straight-
forward analysis of (exceptional) V2 in subject-initial embedded clauses in
Mainland Scandinavian is to assume that the verb moves to a head position in the
CP-domain since one has to assume that the verb moves there anyway in cases of
Topicalization (subject-initial V2 and Topicalization occur in the same types of
sentences). However, the situation in Northern-Norwegian, where the Vfin-Adv
order is not restricted to such environments, calls for a different explanation (see
e.g. Bentzen 2007, Thráinsson 2010 and references cited there). If subject-initial
V2 in embedded clauses in Icelandic is due to V-to-C movement it has to be
assumed that an extended CP-structure is not restricted to certain clause-types
but a general property of all types of embedded clauses in the language, including
relative clauses and various types of adverbial clauses, i.e. “non-V2 contexts”. Such
an analysis would predict that Topicalization should be possible (or widely accept-