MIT CogNet, The Brain Sciences ConnectionFrom the MIT Press, Link to Online Catalog
SPARC Communities
Subscriber : Stanford University Libraries » LOG IN

space

Powered By Google 
Advanced Search

 

The reanalysis of case features in German: Accusative and dative make a difference

 Michael Meng, Markus Bader and Josef Bayer
  
 

Abstract:
In German, case plays a central role in determining the syntactic function of NP-arguments. If abstract case features cannot be identified immediately, e.g., on the basis of morphological information, syntactic function ambiguities arise. If such an ambiguity leads into a garden-path, not only phrase structural reanalysis may be required but also reanalysis of abstract case features: Case features on NP-arguments have to be deleted and different case features have to be assigned. In this poster, we report the results of four experiments which show that reanalysis of case features may be associated with differential costs depending on which new case feature is to be assigned during reanalysis.

The first experiment establishes the presence of a garden-path effect in filler-gap ambiguities like (1). These sentences contain a wh- phrase which is initially analysed as subject (i.e., assigned nominative) but which turns out to be a dative object with receiving the sentence final disambiguating NP. The second experiment contrasts (1) with sentences like (2). These sentences are identical except for the verb in the embedded clause: In (1), the verb (half "helped") requires a dative object whereas the verb in (2) (sah "saw") requires an accusative object. This means that in (1) and (2), different abstract case features have to be assigned to the wh- phrase during reanalysis: dative in (1) but accusative in (2). As we will show, the garden-path effect in (2) is much weaker than in (1).

Our interpretation for this difference in garden-path strength builds on the assumption that in German, dative is subject to stricter morphological licensing conditions than accusative. We hypothesize that assigning dative to the wh- phrase during reanalysis necessitates lexical reprocessing of the wh- phrase in order to check whether it is morphologically compatible with dative. This interpretation is supported by the results of Experiments 3 and 4 which show that processing asymmetries between sentences containing a dative or accusative assigning verb not only hold with respect to garden-path sentences but also with respect to the processing of ungrammatical sentences.

 
 


© 2010 The MIT Press
MIT Logo