Feature 9. Discontinuous NPs

found in question(s): 20a, 20b, 20c, 20d, 20e, 69, 70

Definition and illustration

Some languages allow noun phrases to be discontinous. An example from German ist given in (1):

(1)

It has been observed (Fanselow & Féry 2007) that such inverted splits (where the noun precedes elements that it follows in non-split NPs) tend to occur in languages with A-scrambling (the deeper causes for which remain to be explored). In the database, it is essentially used as an EPP-diagnostic: If a language is subject to the EPP, splitting would have to occur after movement to the subject position and should thus cause a freezing effect. In a language without an EPP, this problem does not arise, splitting can take place in a non-derived position. Question 20 merely tests whether subextraction from NPs is, in principle, possible in the language. This is taken to be a precondtion for splits (and is thus unrelated to the EPP- and VO-/vs. OV-question). Question 70 then investigates whether a language has inverted splits. Question 69 in addition tests for stacking of adjectives, which is taken to be a precondition for splits where one adjective occurs in the fronted position with the noun and one in the part of the NP left behind. Given that only VO-languages are taken to be subject to the EPP, one expects subject splits to be limited to VO-languages. One therefore expects the following correlations:

Correlations

• VO → *split subjects
• OV → split subjects
• EPP → *split subjects
• *VS → *split subjects (VS is an indication of the lack of the EPP, hence, if that is not available, one can assume that language does have the EPP and hence cannot split subjects).

References