Adverbial Incompatibilities and Topicalizability
As noted in Citko and Grohmann 2001, certain adverbs may not occur in between two wh-elements either, and it can be shown that the property they all share is nontopicalizability—the impossibility of be(com)ing topicalized. Obviously, this could hardly be captured under a barrier inspired, quantifier-induced intervention approach, even leaving aside LF issues. I will present the relevant facts in this section and address an interesting twist as well.
For the sake of presentation, I confine myself to manner adverbs, the class of adverbial modifiers that arguably sit very low in the clause structure. In German declarative clauses, these must follow the direct object, unless it is focused.
(1)

In combination with multiple wh-phrases, these adverbs may not appear between them, regardless of their order. (Recall that German does not show any obvious superiority effects in monoclausal interrogative structures.)
(2)

Moreover, such adverbs may not be fronted, in contrast to sentence adverbs like gestern ‘yesterday’, adverbs implying a propositional argument such as vielleicht ‘perhaps’, or epistemic modal adverbs like wahr scheinlich ‘probably’. Since the latter cases exhibit neutral intonation and stress, I understand this kind of fronting as topicalization (taken through out to be a syntactic operation).
(3)

As it turns out, however, such prima facie nontopicalizable adverbs may be fronted if they are contrastively stressed. In these cases, we would expect them also to be legitimate in between two wh-elements. And in deed this expectation is borne out. When any of these adverbs is fronted through focalization, it may also appear in between the two wh-phrases.
(4)

The same contrasts can also be found in embedded contexts, thus ruling out that this is a matrix property possibly related to the verb-second constraint (a traditional indication of topicalization).
(5)


(6)

In this section, I have argued that the phenomenon of (non)topicaliz ability is not only more general than a QUIB approach in terms of intervention, it also follows from more general (and independently justified) principles of grammar, and—‘‘crucially’’—may involve definitely non quantificational material (such as adverbs). A Beck-inspired approach would be more than hard pressed to capture these data.
It now looks like interveners are finally a nonissue, at least as far as German multiple wh-questions are concerned. Previous, intervention driven approaches (starting with Beck 1996a) rely on their being quantificational, however that is to be defined (universally or on the basis of monotonicity). This simply does not cut the pie right. The data in this section and the last suggest that the class of interveners—which block binding of a trace created by LF movement or one’s favorite interpretation of ROLF—has one more property, which does not depend on interaction with any other material (such as those traces purportedly created at LF): they cannot be topicalized. For this, their quantificational properties do not matter at all.
Let me sum up the anti-intervention discussion and put any such notion to rest with the Topicalizability Generalization.
(7)

Admittedly, generalization (7) is only that, a generalization, with the proviso that it is possibly restricted to German monoclausal questions with two wh-elements—but it is one that works, unlike intervention. As a generalization, (7) has no explanatory power.