Leibniz-Zentrum Allgemeine Sprachwissenschaft Leibniz-Gemeinschaft

Multilingualism, Acquisition, and Development (MAD) colloquium: Code-switching as a signaling cue: Evidence from cloze and ERP studies

Vortragende(r) Onur Özsoy
Institution(en) Universität zu Köln
Datum 22.10.2026, 15:00 - 16:30 Uhr
Uhrzeit 15:00 Uhr
Ort Online via Teams

Code of Conduct for ZAS events: The ZAS is committed to fair, respectful, and professional interaction at its events. Therefore, please observe the Code of Conduct for this event.

Abstract

In bilinguals, code-switching (changing languages in the middle of a sentence) is ubiquitous. Prior psycholinguistic studies have mainly been concerned with the cognitive costs of code-switching. However, code-switching often has a communicative function. Previous studies have shown that code-switching often precedes information that is less expected (Myslín & Levy, 2015), and that bilinguals can use this as a cue that unusual information is coming up (Tomić & Valdés Kroff, 2021). To investigate the signaling function of code-switching in two different language communities, we conducted a cloze-completion experiment (Exp 1) and an ERP experiment (Exp 2) on (i) highly proficient late bilinguals with L1 German and L2 English (GerEng) in Germany; and (ii) early Spanish-English bilinguals (SpaEng) in the US.

In Study 1, participants (GerEng n=124, SpaEng n=122) read highly constraining sentences with the final word omitted and were asked to fill in the blank. The material preceding the blank was either entirely in English (“Every morning he drank a glass of __”), or contained a switch to English from German or Spanish, 3-6 words before the blank (“Jeden Morgen/ Cada mañana he drank a glass of __”). For the SpaEng group cloze probabilities were lower in code-switched than in unilingual English sentences (b=0.63, SE=0.01, p<.0001). This effect was not seen for the GerEng group (b=0.01, SE=0.01, p=.38), suggesting that the early but not the late bilingual group used code-switching as an information signal in production.

Study 2 collected ERPs while participants (GerEng n=44; SpaEng current n=18) were reading materials similar to Study 1, but now ending in a high-cloze (“milk”) or low-cloze word (“wine”). If code-switching prepares the reader for less expected information, the N400 for the low-cloze endings was predicted to be smaller after code-switches compared to unilingual items. This was borne out for the GerEng group (b= –0.48, SE=0.15, p< .01). Preliminary analyses of the Span Eng group, on the other hand, showed an increased frontal positivity for endings after a switch compared to the unilingual condition (SpaEng b=1.03; SE =0.47; p< .05). This suggests that in the GerEng group a code-switch signals upcoming unexpected information; in the SpaEng group a code-switch may make upcoming information more salient, leading to a more thorough updating of the mental representation of the discourse when that information is encountered. Altogether our results suggest that the informational mechanisms underlying code-switching may therefore differ between production and comprehension, and between bilingual groups.

Please contact garcia@leibniz-zas.de for the Teams link and password.