Abstract
The BIA model is an implemented language nonselective access model for bilingual word recognition which assumes an integrated lexicon for words of different languages. In this paper, simulations with the BIA model are presented that account for the effects of lexical competitors from the same and another language on target word recognition in two different paradigms. First, the model simulates recent lexical decision results obtained with French/English bilinguals in two masked orthographic priming experiments (Bijeljac-Babic, Biardeau, & Grainger, 1997). Differences in the result patterns for subject groups varying in second language (L2) proficiency (monolingual, beginning, and more advanced bilinguals) are simulated by assuming a reduced frequency range for L2-words. In demonstrating that the monolingual participants in this study very likely knew some L2-words, the model displays considerable heuristic value. Second, the BIA model simulates the effects of intra- and interlingual neighborhood density on target word recognition in blocked and mixed progressive demasking (PDM) with Dutch/English bilinguals (Van Heuven, Dijkstra, & Grainger, 1998). The model accounts for a shift in result patterns observed in high-proficiency subjects over parts of the blocked PDM experiment by varying the degree of asymmetric cross-language inhibition from language node to word level. Finally, the model simulates the over-all result pattern in the mixed PDM experiment by manipulation of both cross-language inhibition and the frequency range for L2-items. A comparison with some other nonselective and selective access models indicates that the BIA model produces better overall simulation results because the two mechanisms allow the model to capture the asymmetric cross language effects observed in unbalanced bilingual participants.