| We report two offline studies in German testing the
interpretation of the remnant in complex sluicing like The police is
looking for a repeat offender who robbed a store. Do you know
who/which? Remnants were morphologically disambiguated as to the
correlate and focus on the correlate was manipulated using context.
Ratings showed a preference for focused correlates. While this finding
is consistent with Frazier & Clifton's (2005) results, it
contrasts with previous data that showed a somewhat lower acceptability
of correlates in constructions that may be analyzed as syntactic
islands partially repaired by sluicing, cf. Merchant (2005). Taken
together, the findings support the direct interpretation analysis of
sluicing (Culicover & Jackendoff 2005, Ginzburg & Sag
2000).
|