Language-switching and retrieval-based learning: an unfavorable combination.

Language-switching costs arise when learners encode information in one language and subsequently recall that information in a different language. The assumed cognitive mechanism behind these costs is the principle of encoding specificity that implies language-dependent representations of information. The aim of our study was to test this mechanism and to gain insights into the impact of language-switching on subsequent learning. To this end, we used retrieval-based learning as a carrier-paradigm. In a 2×3-design, 117 participants learned mathematical concepts with a practice-test or a restudy opportunity (within-subjects factor). In addition, the sample was divided into three groups regarding language-switching (between-subjects factor): one group without switching, one switched for the final tests, and one switched between initial learning and subsequent learning. Results show the expected main effects: participants performed better for the items learned via retrieval-based learning (testing-effect) and worse in conditions with language-switching (language-switching-costs). Most importantly, we were able to find an interaction between learning condition and language-switching: retrieval-based learning suffers particularly from language-switching. Additionally, our results indicate that language switching before subsequent learning seems to be particularly detrimental. These results provide both validation for encoding specificity as mechanism underlying language-switching costs and new information on the impact of the time of language-switching that can be considered in educational designs such as "Content and Language Integrated Learning."

Investigating the modulation of stimulus types on language switching costs: Do semantic and repetition priming effect matter?

Introduction: In the present study, I investigated the influence of stimulus types on bilingual control in the language switching process. The commonly employed stimuli in language switching studies - Arabic digits and objects - were compared to further investigate the way in which inhibitory control could be modulated by semantic and repetition priming effects. The digit stimuli have two unique characteristics in the language switching paradigm, for example, they are present repeatedly and are semantically related to each other, compared with pictural stimuli. Thus, these unique characteristics might influence the operation of inhibitory control in bilingual language production, modulating the size and asymmetry of switching costs. Methods: Two picture control sets were set up to match those characteristics: (1) a semantic control set, in which picture stimuli belong to the same category group, such as, animals, occupations or transportation and specific semantic categories were presented in a blocked condition; and (2) a repeated control set, in which nine different picture stimuli were repeatedly presented like the Arabic digits from 1 to 9. Results: When comparing the digit condition and the standard picture condition, analyses of naming latencies and accuracy rates revealed that switching costs were reliably smaller for digit naming than for picture naming and the L1 elicited more switching costs for picture naming than for digit naming. On the other hand, when comparing the digit condition and the two picture control sets, it was found that the magnitude of switching costs became identical and the asymmetry in switching costs became much smaller between the two languages.

Investigating the modulation of active preparation and passive dissipation on inhibitory control processes in the language switching paradigm.

Introduction: Previous language-switching studies have received scholastic attention and the observed switching cost patterns have provided empirical evidence for bilingual language control. However, results are inconsistent as the size of and (a)symmetry in switching costs differ across studies. In addition, there are various methodological differences that go beyond stimulus differences, such as the language proficiency of the participants (the participant-level factor) and the preparation time (a task-related level factor), which might be responsible for these inconsistent results. Methods: With a focus on task-related factors, the present study was designed to examine whether and how preparation time modulates the size and (a)symmetry in switching costs by using the language-switching paradigm with cue-to-stimulus and response-to-cue intervals manipulated. Results: Replicating previous literature on language switching and task switching, a clear preparation effect was observed in all trials (stay and switch trials) for both L1 and L2. The switching costs were modulated by the cue-to-stimulus intervals, and specifically, switching costs decreased when the preparation time increased. Another intriguing finding was that even when participants were offered enough time to fully prepare for selecting the target language at the cue window, the switching costs were not completely eliminated. In terms of the passive preparation at the response-to-cue interval, switching costs could be modulated by the response-to-cue interval - the time for passive dissipation of inhibitory control applied in previous trials. The size of switching costs was clearly modulated by manipulating response-to-cue intervals and switching costs decreased as the waiting time after a naming response increased. Discussion: This study provides empirical evidence for the modulation of preparation effects on switching costs and inhibitory control mechanisms in bilingual language production.