Cognitive effort during the processing of relative clauses with psychological predicates in Spanish: a pupillometric study

MARISOL MURUJOSA; YAMILA SEVILLA; CAROLINA GATTEI; DIEGO E. SHALÓM

Potsdam

Conferencia; 26TH ARCHITECTURES AND MECHANISMS FOR LANGUAGE PROCESSING CONFERENCE; 2020

University of Potsdam

The asymmetry in the processing of subject (S) and object (O) relative clauses (RC) is well documented in literature and seems to be present in a wide range of languages (e.g, in Spanish; in English; in Dutch; in German; and French). However, this asymmetry has mostly been studied in sentences with transitive activity predicates (ACT), while there has been no evidence reported regarding the processing of RCs with psychological predicates (PSY).Memory based accounts of this phenomena predict a general locality preference for shorter filler-gap dependencies (Gibson, 1998; Lewis & Vasishth, 2005). In the case RCs with ACTs, ORCs instantiate a longer filler-gap dependency than SRCs and this explains why the latter could be easier to process. However, considering the structural properties of PSYs in Spanish (see note 1), it?s in the case of ORCs that a shorter filler-gap dependency is established.Therefore, the processing of ORCs with these predicates should entail less difficulty and lower cognitive effort than SRCs. Task-evoked pupillary responses (TEPR) have been used as a reliable neurophysiological index of cognitive effort in different domains (Beatty, 1982; Beatty & Lucero-Wagoner, 2000), including perception, mental arithmetic and working memory load. In the sentence comprehension domain, TEPRs have provided a measure of differential processing cost according to the type of structure and its syntactic complexity (e.g,Just and Carpenter, 1993; Scheepers & Crocker, 2004; Schluroff, 1982). Design: 33 subjects participated in an auditory sentence comprehension task. They were asked to listen to a sentence; then were showed an image and were prompted to judge whether the image they saw faithfully reflected the content of the sentence heard or not (see Fig. 1). The stimuli (n=20) consisted of RCs with PSYs (1) and with ACTs (2) (see note 2). We manipulated the type of RC with each predicate: SRCs (1.a and 2.a) and ORCs (1.b and 2.b). The images selected were counterbalanced to make the sentences either true or false. Response accuracy and reaction times (RTs), as well as TEPRs, were measured during the task. Pupil diameter was monitored using a desktop-mounted, video-based eye tracker (EyeLink 1000, SR Research Ltd., Ontario, Canada) at a sampling rate of 1000 Hz. Results: on average, participants answered 89% (SE= 0.8%) of the total stimuli correctly; Figs. 2 and 3 show mean correct answers and standard error, and mean RTs and standard error (only RTs of correct answers were considered) according to condition respectively. Linear mixed-effect models were fitted for data analysis. Results show that RCs with PSYs were harder to comprehend (p=.02) andwere processed more slowly (p