Marit Julien
Embedded V2 in Scandinavian
In Mainland Scandinavian, the finite verb is normally in second position in main clauses, while embedded clauses have non-V2 word order. INevertheless, embedded clauses sometimes also show V2 order. In the linguistic literature, the explanations that have been offered for embedded V2 in Germanic are of two types. On the one hand, it has been claimed that the semantic properties of the embedded clause itself are crucial; more precisely, that embedded V2 is more frequent in clauses that represent an asserted proposition (see Andersson 1975, G. Green 1976, Wechsler 1991, Holmberg & Platzack 1995). On the other hand, the possibility of having V2 order in an embedded sentence has been seen as a consequence of certain properties of the matrix predicate. De Haan & Weerman (1986), Iatridou & Kroch (1992), and perhaps most famously, Vikner (1995), note that V2 is seen in clausal complements of certain verbs, called "bridge verbs". But as Vikner points out, the verbs in question are not necessarily bridge verbs in the sense that they allow extraction out of the clause they have as complement. In fact, according to Vikner it seems that his "bridge verbs" are only distinguished as a group by their ability to take a V2 complement, and moreover, there is some variation between the Germanic languages as to which verbs have this property. Hence, Vikner concludes that its semantics is not what makes a verb a "bridge verb".
What I will argue is that embedded V2 is closely tied to the semantics of the embedded clause itself: only asserted embedded clauses can have V2 order. The matrix predicate only has an indirect effect, as not all predicates can have an asserted clause as complement. But as we shall see, V2-clauses can be embedded under elements belonging to many different classes, and in many cases one cannot say that it is the embedding element that makes embedded V2 possible.
I will also suggest a formal analysis of embedded V2. In my opinion, the biggest shortcoming of many earlier analyses is that they do not establish a connection between the semantics of embedded V2 clauses and their syntax. But in my analysis, syntax and semantics are tied together -- the structure that I propose results in V2 order and in the asserted status of the clause. And as we will see, this analysis can also account for V2 in root clauses, so that the V2 phenomenon gets a uniform explanation.
Finally, I will present some examples where a constituent has moved from the embedded V2-clause into the matrix clause. It is usually assumed that movement of this kind is not possible, but it nevertheless appears that some speakers of Scandinavian accept it. This fact has consequences for the analysis of embedded V2.