Question about long-distance dependencies

[Gert Webelhuth]

Hi everybody,

I am interested in analyses or empirical facts where the properties of a
filler are different from the properties of its trace. E.g. there is an
analysis of German partial VP topicalization where the topicalized element
is assumed to be [LEX -] but its trace is treated as [LEX +]. Empirically,
there are cases where an NP that is extracted must have a different case
than it would have if it is realized locally, e.g. in a construction of
Hungarian.

Are there any additional cases like that?

Thanks for your help!

Gert