Cataphoric pronoun dependencies in Dutch: An ERP study

The processing of cataphoric pronouns has been shown to follow the same mechanisms as the processing of wh-dependencies in that the parser actively searches [1]for an antecedent to interpret the pronoun within the sentence, except in those cases where the pronoun must obey principle C of the binding theory(a name cannot be c-commanded by a nominative pronoun) [2]. This is the case for the Gender Mismatch (GMM) effect, a slowdown effect that shows that the parser tries to link an antecedent to a preceding pronoun only when the pronoun can be bound by it [3,4]. The present study uses the GMM paradigm used by [4] in order to test if encountering a pronoun triggers the search for an antecedent in the upcoming context. This study aims to test whether the binding principle constrains the antecedent-search process in Dutch employing Event Related Potentials (ERP). If the parser attempts to bind the pronouns zijn and haarin (1a) and (1b) to the masculine antecedent Lodewijk, we expect a GMM effect at the position of the potential antecedent in (1b). On the other hand, if the parser respects the Binding Principle C we do not expect to find any ERP difference at the position of the potential antecedent Lodewijkin (1c) and (1d), showing that the parser does not try to link the pronouns hijand zij to the name. We conducted an ERP experiment where EEG was continuously recorded while 24 native speakers of Dutch read silently 36 sentences such as (1a-d). Results show that there is a central anterior negativity in the 200-600m