Biondo, N. 1 , Vespignani, F. 1 , Rizzi, L. 2, 3 & Mancini, S. 4
1 Department of Psychology and Cognitive Science, University of Trento
2 Department of Linguistics, University of Geneva
3 Center for Cognitive Studies on Language - Department of social, political and cognitive sciences (CISCL-DISPOC), University of Siena
4 Basque Center on Cognition, Brain and Language (BCBL)
Feature sharing between constituents encompasses well studied phenomena as subject-verb agreement but also others as temporal concord. In this eye-tracking study we analyzed Number, Person and Tense mismatches on an inflected verb. At least two dimensions distinguish these relations: the type of constituent involved in the relation and the discourse relevance of the analyzed features. Number and Person mismatches (1) concern the verb and a mandatory subject while Tense mismatches (2) concern the verb and an optional adverb. Person and Tense entail deictic information while Number does not.
(1) Man?ana el viajero3SG regresara?3SG/*a?n3PL/*a?s2SG a casa con un amigo
(2) Los viajeros man?anaFUT regresara?nFUT/*regresaronPAST a casa con un amigo
Results showed that at the target region only subject-verb violations gave longer first-pass while both subject-verb and adverb-verb violations show longer reading times in late measures (go-past and total); longer reading times for Person violations compared to Number at the verb region (total) and at the post-verb region (all measures). Interestingly, Person and Tense violations differently from Number show sustained effects on the spillover region. Data suggest that the checking of congruence between features across constituents may affect different stages of parsing as a function of both structural and discourse-level relevance.