Tarlowski, A. 1 , Wodniecka, Z. 2 & Marzecová, A. 2
1 Universidad del País Vasco
2 Jagiellonian University, Krakow, Poland
It is debated to what extent language control relies on inhibitory mechanism and whether the mechanism is sensitive to language membership or to specific properties of utterances. The primary source of evidence for language-general inhibitory mechanism has been the asymmetrical switching cost (i.e. the effect demonstrating greater switch costs to a dominant language than to a weaker language). However, symmetrical switching cost and a reversed dominance effect were reported in high-proficient bilinguals (Costa & Santesteban, 2004), as well as in unbalanced bilinguals self-monitoring switches (Gollan & Ferreira, 2009). These studies constrain the inhibitory model of language control. Past research focused selectively on production of nouns and numerals. To extend the scope of empirical evidence, we explored production of simple grammatical structures. In the reported study, Polish-English unbalanced bilinguals used their L1 and L2 to name either ongoing or completed actions. We assumed that there are no processing differences between different grammatical structures in L1, while for L2 such differences occur. For speakers of Polish, processing English perfective seems to be more cognitively demanding than processing English progressive (the two structures differ in morphological and conceptual complexity, as well as in AoA in L2). We observed two diverging patterns of switching costs depending on a type of grammatical structure. The results showed asymmetrical switching cost for ongoing actions and symmetrical switching cost with reversed dominance for completed actions. Participants performed like high-proficient bilinguals when L2 grammar was relatively easy (English progressive), and they performed like low-proficient bilinguals when L2 grammar was relatively difficult (English perfective). The results suggest that past findings on language switching extend to production of grammar and that mechanisms responsible for switching costs are sensitive to specific properties of utterances rather than their language membership. The results are discussed in light of current models of language control.