Bilingual language control : the effects of sequential predictability on language switching
Year: 2014
Type: dissertation
Abstract: Every time bilinguals produce language they run the risk to involuntarily produce in the non-target language (e.g., Poulisse & Bongaerts, 1994). Yet, selection of words in the non-target language is constrained by a process known as language control, which makes it more probable that words from the target language are selected for production (e.g., Green, 1998). An important experimental approach to investigate language control is language switching (e.g., Costa & Santesteban, 2004; Meuter & Allport, 1999), which typically requires participants to switch between two languages. A large amount of studies have shown that higher error rates and longer reaction times are obtained when switching from one language to another than when the same language is repeated (e.g., Costa & Santesteban, 2004; Meuter & Allport, 1999). This performance cost is usually referred to as “switch costs”. The current study investigated whether these switch costs are affected by preparation, which is an effect that has mostly been neglected in language switching. This is surprising, since some models (e.g., Allport et al., 1994) have postulated that switch costs mainly measure interference, whereas language control is a process of language interference resolution. Preparatory effects on switch costs, on the other hand, are assumed to be a more informative marker for endogenous language control, since these allow for advanced (i.e., preparatory) interference resolution. A novel language switching paradigm was constructed to investigate the effect of preparation on switch costs. This paradigm, named the sequence-based language switching paradigm, contains no visual input. To this end, both the language and concept sequence follow a fixed sequence, which the participants have to memorize. Since the sequence-based language switching paradigm contains a memorized language sequence and concept sequence, both the language sequence and concept sequence are predictable and can thus be prepared. In the first part of this dissertation, the effect of predictable responses in language switching was investigated. According to some models of control, no switch costs should be observed in language switching when responses can be prepared (i.e., reconfiguration model; e.g., Rogers & Monsell, 1995). On the other hand, according to the proactive interference model (Allport et al., 1994) and its bilingual variant (i.e., inhibitory control model; Green, 1998), predictable responses could instigate switch costs. The results of the experiments using the sequence-based language switching paradigm showed that switch costs can still be observed in a language switching task with predictable responses. Switch costs were even found when not just the language sequence and concept sequence were predictable, but also the response onset was predictably known. Hence, these results are more in line with the proactive interference model and the inhibitory control model than with the reconfiguration model. In the second part of this dissertation, the effects of language predictability and/or concept predictability in language switching were investigated. According to the reconfiguration model and the inhibitory control model, knowing both the upcoming language and concept should reduce switch costs to a larger extent than only knowing either the upcoming language or concept. Furthermore, only knowing the upcoming target language should also reduce switch costs according to these two models, whereas no effect on switch costs should occur due to only knowing the upcoming target concept. By manipulating the predictability of the language sequence and the concept sequence in the sequence-based language switching paradigm, smaller switch costs were found when both the languages and concepts were predictable than when only the language sequence or concept sequence were predictable. Yet, no influence on switch costs was found when contrasting language switching with either a predictable language sequence or concept sequence against language switching with no predictable language sequence and concept sequence. Hence, these results are somewhat in line with the reconfiguration model and inhibitory control model. Yet, not all the results can be explained with these two models. In sum, the results obtained in this dissertation indicate that the models of control still lack in some respects. To this end, a modified inhibitory control model is put forward that can explain the data patterns obtained in this dissertation and those of a large number of language switching studies.
Authors: Mathieu Declerck, Iring Koch
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W2292211963