Keywords: syntactic theory, morphosyntax, verb movement, Scandinavian syntax, syntactic change, variation.

This paper briefly discusses previous attempts to relate “rich verbal morphology” and verb movement (V-to-I movement) and claims that they typically fail to provide any kind of theoretical explanation for the observed correlation.

The paper then outlines the theory proposed in by Bobaljik and Höskuldur Thráinsson (1998, henceforth B&T) and points out that their theory maintains that there should only be a one way correlation between rich verbal morphology and verb movement. More precisely, B&T argue that languages differ with respect to the complexity of inflectional projections in the syntax, some languages having a Split IP (separate Tense and Agreement projections along the lines originally suggested by Pollock 1989 and Chomsky 1991) but others a simplex IP projection.

B&T describe this difference in terms of the Split IP Parameter and claim that a clear distinction between tense and agreement markers in the morphology will serve as a morphological trigger for a positive setting of the Split IP Parameter. Then B&T maintain that a split IP structure will require movement of the finite verb out of the VP to satisfy checking requirements under standard minimalist assumptions about checking domains, whereas a simplex IP will not require such movement and hence rule it out under economy conditions.

But since syntactic evidence (including V-to-I movement itself!) can also trigger a positive setting of the Split IP Parameter, a given language (or a given grammar) can have split IP without an unambiguous morphological evidence for such a structure (and thus possibly require a morphological analysis involving null morphemes or a morphologically unnatural splitting of morphemes to match the functional structure in the syntax). Under B&T’s assumptions, then, the biconditional (if and only if) holds between the split IP structure and verb movement of the kind under discussion, whereas the correlation between rich verbal morphology and verb movement is one way only.

Given this, B&T’s theory predicts that it should be possible to find languages (or dialects) without rich verbal morphology but with V-to-I movement (more precisely, at least V-to-T). Furthermore, it maintains that there is no reason to expect an immediate loss of V-to-T once the verbal morphology has been simplified.

The present paper surveys Scandinavian diachronic and dialectal evidence and shows that this prediction fits the facts: Verbal morphology is always simplified at least a couple of centuries before V-to-I disappears and in some Modlern Scandinavian dialects there are at least remnants of V-to-I although the verbal morphology appears to be quite simple. This includes Faroese, as is extensively documented in this paper, both on the basis of text research and on the basis of a survey among Faroese High School students.

The general trend seems to be that V-to-I is on the way out in Faroese, although many writers still make considerable use of it and some speakers accept it quite readily in embedded clauses (although in general less readily in relative clauses and indirect questions than in other types of embedded clauses).

But the fact that embedded V-to-I seems to be in some sense optional for many speakers of Faroese will be puzzling for many syntacticians, since syntactic rules of this kind “should not” be optional. It seems, however, that this is a general pattern in the diachronic development of this syntactic phenomenon (and many others).

As pointed out in the paper, such a state of affairs is obviously compatible with the so-called double-base hypothesis of Kroch (1989) and others, but its theoretical implications are not discussed in any detail here.