Morales, L. 1 , Paolieri, D. 1 , Bajo, T. . 1 , Valdés, J. . 2 , Gerfen , C. . 2 & Dussias, P. E. 2
1 University of Granada
2 Pennsylvania State University
Bilinguals are slower in naming pictures that are incongruent in gender between their two languages (camaFEM in Spanish, lettoMAS in Italian, bed) than gender-congruent pictures (bufandaFEM in Spanish, sciarpaFEM in Italian, scarf, Paolieri et al., 2010). Few studies have examined the presence of this effect during bilingual spoken-word recognition (Dahan et al., 2000). We tests for the gender-congruency effect in spoken language comprehension using the visual world paradigm (Allopenna et al., 1998). The eye-movements of 32 proficient Italian-Spanish bilinguals were monitored while viewing two objects on a screen. Participants listened to instructions in Spanish (Encuentra laFEM bufandaFEM; Find the scarf) and were asked to click on the picture of the named object. Grammatical gender of the object names was manipulated, so that one object name was gender-congruent between the two languages (e.g., bufanda) while the other was incongruent (cama). If bilinguals access gender information from their L1, they should orient their eyes to the L2 target object (bufanda) sooner in incongruent trials (when bufandaFEM is presented with camaFEM) than in congruent trials (when bufandaFEM is presented with ventanaFEM, in Italian finestraFEM, window), because the information provided by the Spanish feminine article la should preempt looks to the object with incongruent gender in Italian. This is precisely what we found. The results indicate that during spoken-word recognition in L2 bilinguals access L1 gender information, as predicted by models of non-selective activation (Kroll & Stewart, 1994).