Poster D58, Friday, August 17, 4:45 – 6:30 pm, Room 2000AB
Pardon My Code-Switching: Electrophysiological Effects of Mixing within the Determiner Phrase
Leah Gosselin1, Michèle Burkholder1, Laura Sabourin1;1University of Ottawa
In multiple Canadian francophone communities, switching between languages within a single sentence (intrasentential code-switching) is regarded as typical linguistic behaviour (Poplack, 1980). While code-switching was originally believed to reflect lack of linguistic competence on the speaker’s behalf, recent research suggests that code-switching is rule-governed and systematic. As such, notable regularities have been observed in corpus studies among language-mixers. For instance, in the case of mixing at the level of the determiner phrase (DP), Spanish-English bilinguals typically employ a Spanish determiner with an English noun (e.g. la house). The opposite direction (e.g. the casa) is disfavoured (Liceras, Fernández Fuertez, & Spradlin, 2005). Since the gender systems in French and Spanish are similar, French-English mixed DPs should carry the same pattern as those in Spanish-English. A handful of studies have examined code-switching more generally, with the use of Event-Related Potentials (ERP). Well-known ERP components may indicate specific reactions to linguistic events such difficulty with semantic integration (Kutas & Hillyard, 1980) or structure violation and reanalysis (Hagoort &Brown, 1999). However, all but one of these studies included bilinguals who did not frequently mix their languages. While most of the literature demonstrates that language-mixing incurs processing costs (indexed by an N400), it is likely that ERP components were impacted by the fact that participants were not habitual code-switchers. To this date, there are no known online studies examining mixed DPs. The current study tests 12 English-French simultaneous and early bilinguals from the Ottawa-Gatineau region, who are habitual code-switchers. Participants read code-switched sentences, via the rapid serial visual presentation paradigm. The conditions of the target DP itself varied according to the language of the determiner, the gender of the noun and the gender-relation between the determiner and noun (match/mismatch). While the participants read these sentences, their neural activity was measured with the electroencephalogram. Preliminary results show significant differences between mixed and unmixed DPs in the 500-700ms time window (p < .01). As such, mixed DPs indexed an increased positivity which was distributed equally across the head. Furthermore, results show no significant differences between mixed DPs with an English determiner and those with a French determiner (p > .23) at all time windows. Finally, data analysis shows no significant differences between mixed DPs with matching and mismatching gender between determiner and noun (p > .36). These results seem to indicate that French-English mixed DPs do not follow the patterns demonstrated by Spanish-English bilinguals in corpus studies: French determiners did not incur less structure reanalysis than English determiners. In parallel, it seems that “gender violations” in mixed DPs are not altogether unacceptable to language-mixers. Furthermore, this study did not find any N400 effects for mixed DPs, which support the fact that previous results in the literature are potentially a consequence of participant samples (i.e. a lack of habitual code-switchers). To conclude, this type of research could give valuable insights into the human language faculty. By demonstrating that code-switching is highly natural and constrained, this project could also help remove the stigma around language mixing.
Topic Area: Multilingualism