| Abstract: | In two experiments, Greek-English bilinguals alternated between performing a lexical decision task in Greek and in English. The cost to performance on switch trials interacted with response repetition, implying that a source of this "switch cost" is at the level of response mapping or initiation. Orthographic specificity also affected switch cost. Greek and English have partially overlapping alphabets, which enabled us to manipulate language specificity at the letter level, rather than only at the level of letter clusters. Language-nonspecific stimuli used only symbols common to both Greek and English, whereas language-specific stimuli contained letters unique to just one language. The switch cost was markedly reduced by such language-specific orthography, and this effect did not interact with the effect of response repetition, implying a separate, stimulus-sensitive source of switch costs. However, we argue that this second source is not within the word-recognition system, but at the level of task schemas, because the reduction of switch cost with language-specific stimuli was abolished when these stimuli were intermingled with language-nonspecific stimuli. |
