Datensatz

Zusammenfassung

Reanalysis in language comprehension provides a window on how superficially similar processes of conflict resolution may differ depending on the context in which they are initiated. Thus, previous ERP studies have shown that reanalyses towards object-initial orders in German sentences with dative-active verbs (e.g., folgen, ‘to follow’) engender N400 effects, while reanalyses with accusative verbs (e.g., besuchen, ‘to visit’) elicit P600 effects. This difference appears surprising since these two verb classes are both associated with a subject-initial base order. The present paper reports two ERP experiments designed to shed further light on the nature of the conflict resolution processes involved in each case by examining structures in which word order disambiguation is separated from verb class disambiguation. Experiment 1 contrasted dative-active verbs with accusative verbs, while Experiment 2 compared dative-active and dative object-experiencer verbs (which are associated with an object-initial base order). Our results show that the reanalysis pattern for dative-active constructions is context-dependent: when verb class disambiguation precedes word order disambiguation, an N400–P600 pattern results. By contrast, the reanalysis patterns for the other two verb types are context independent: object-experiencer verbs invariably show an N400 and accusative verbs invariably show a P600. We argue that (a) the N400 is a general marker of reanalysis in dative sentences, reflecting an argument reindexation, while (b) the P600 in accusative sentences reflects a structural recomputation. The variable pattern for dative-active sentences reflects the (in)applicability of “good-enough” representations during conflict resolution in garden path sentences.