ABSTRACT
In the present study we examined whether overlap in language across texts influences the integration of information into a coherent discourse representation for bilingual readers. Across two experiments highly proficient Spanish–English bilinguals read pairs of expository passages describing two fictional science facts while their eye-movements were monitored. One of the facts was revised in the second passage, requiring a discourse updating. The language of the two passages and follow-up questions was fully crossed. Accuracy was lower for questions pertaining to revised facts when the second passage was in the second language (L2). This cost was exacerbated when the first passage was in the dominant language, suggesting strong interference from the representation of the first passage which impeded updating the discourse model in the L2. This interference was eliminated in Experiment 2 when second passages were written based on a refutation-style text structure. Analyses of reading times on the pseudo-terms before and after the revised fact was stated indicated that the previous version of the fact was reactivated and interfered with processing. This interference was similar regardless of whether passages were written in the same or different languages.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. Because of the limited number of items, models did not converge with random slope adjustments across items.