Kaan, E. 1 , Ballantyne, J. 2, 3 , Choi, J. 2 , Rowlings, B. 1 & Wijnen, F. 1
1 University of Florida, USA
2 Utrecht University, The Netherlands
3 Utrecht University College, The Netherlands
This study investigated the effects of language proficiency and cognitive control on the resolution of language conflict in second-language sentence processing. Low- and high proficiency native English speakers and advanced Dutch late-learners of English were tested in a speeded grammaticality judgment study. Word order and use of pronouns were manipulated, creating potential language conflicts for the language learners. Object relatives with a subject/verb agreement mismatch, cf. (1), are ungrammatical in English, but grammatical as subject relatives when translated literally into Dutch (1b). In the pronoun conditions (2), the pronoun him is ungrammatical in English when referring to an inanimate object, but grammatical in Dutch when the antecedent is a noun with common gender (Conklin et al., 2007).
(1) a. That is the director who the dancers *has. admired …
b. Dat is the regisseur die de dansers heeft bewonderd …
“That is the director who has admired the dancers ….”
(2) Although the helicopter was available, the rescuers could not use it/*him in the mountains.
Test sentences were presented RSVP and participants gave an end-of-sentence grammaticality response. Cognitive control measures were obtained by the Attentional Network Task (Fan et al., 2002). Both native English speakers and Dutch learners of English showed higher response accuracy with higher English proficiency as measured by the C-test (Keijzer, 2007). In the Dutch-English group, however, higher English proficiency related to better performance especially in the language conflict condition involving pronouns (2), but not word order (1). This was not influenced by cognitive control. These results suggest that interference from the native language is strongest when mediated by lexical information (i.e., the gender of (pro)nouns). Furthermore, proficiency, more than cognitive control, affects first-language interference in second-language sentence processing. This supports models in which activation of language-specific information in bilinguals is not mediated by cognitive control.