Object relatives are not always moredifficult to process, even in Spanish. Evidence from a study of relative clauses comprehensionwith psychological predicates

MARISOL MURUJOSA; CAROLINA GATTEI; DIEGO SHALOM; YAMILA SEVILLA

Amherst

Congreso; 33rd Annual CUNY Conference on Human Sentence Processing; 2020

University of Massachusetts

The asymmetry in the processing of subject (SR) and object (OR) relative clauses (RC) is welldocumented in literature and seems to be present in a wide range of languages: in languageacquisition (e.g. Friedmann et al. 2009), in sentence production (e.g. Belletti & Contemori 2010;in Spanish, Muñoz Pérez & Lago 2012); and in sentence comprehension (e.g. Gordon et al.2004; in Spanish, Sánchez et al. 2017). However, this asymmetry has only been studied insentences with transitive activity predicates, while there has been no evidence reportedregarding the processing of RCs with psychological predicates. Since Friedmann et al. (2009), ithas been argued that the advantage for SRs can be explained by the Featural RelativizedMinimality theory (fRM; Rizzi 2004) as an effect of syntactic intervention. As both the subject NPand the object NP share the [+N] syntactic feature, the subject NP functions as an intervenerwhen the object NP moves to the left periphery, hindering the establishment of the syntacticdependency. However, the syntactic structure of sentences with third class psychologicalpredicates such as gustar (?to like‟) differs from the structure of sentences with transitive activityverbs (Belletti & Rizzi 1988). In Spanish, the ?object‟ of psychological predicates is generated ina higher syntactic position than the subject (Pujalte 2015), i.e. as a high applicative phrase(Pylkkänen 2008). Therefore, the effects of syntactic intervention should occur in the oppositecondition. Hence, we expect that the processing of ORs with these kind of predicates entailslower cognitive cost than SRs. Design: we designed a sentence comprehension task in whichparticipants (n=33) were asked to listen to a sentence; then were showed an image and theyhad to decide whether the sentence they heard was true or false regarding the image they saw.The stimuli (n=20) consisted of RCs with third class psychological predicates (1) and withtransitive activity verbs (2). In Spanish third class psychological predicates project the theme ofthe event with nominative case and the experiencer with dative case, so we chose transitiveactivity predicates that also project the agent of the event with nominative case and the patientwith dative case. We manipulated the type of RC with each predicate: RCs with the nominativecase argument as the antecedent or ?subject‟ (1.a and 2.a) and RCs with dative arguments asthe antecedent or ?object‟ (1.b and 2.b). The image showed after the sentence was randomlyselected to make the sentence either true or false. Fig. 1 and Fig. 2 show sample picture stimulifor sentences in (1). Response accuracy and reaction times (RTs) were measured during thetask. Results: on average, participants answered 89% (SE= .8) of the total stimuli correctly;Table 1 shows the mean of correct answers and standard error per condition; Table 2 shows themean of RTs and standard error. Linear mixed-effect models were fitted for data analysis.Results show that RCs with psychological predicates were harder to comprehend (p=.02) andwere processed more slowly (p