Séminaire de Recherche en Linguistique

Do bilingual children show effects of Relatived Minimality in the processing of wh-questions? Evidence from the visual-world paradigm

Previous research has shown that sequential bilingual children process morphosyntactic information in a similar way as monolingual children but their speed of processing is slower (e.g., Chondrogianni & Marinis, 2012; Marinis & Saddy, 2013; Chondrogianni et al., 2015). Most of these studies used grammatical violation paradigms involving errors of omission (e.g., tense, articles, pronouns) or commission (e.g., gender, case). Less is known about how bilingual children process sentences without grammatical violations that may require reanalysis, like filler-gap dependencies (but see Marinis & Saddy, 2013 for reanalysis in passives). 

In this talk I will present data on how bilingual children growing up in the UK process d-linked wh-questions using the visual world paradigm. This is work conducted by my PhD student George Pontikas at the University of Reading and is a follow up from the study by Contemori, Carlson, & Marinis (2018). 

Contemori et al. (2018) investigated subject and object d-linked wh-questions of the type “Which donkey is carrying the zebra/Which donkey is the zebra carrying?” and manipulated number in the two NPs to address whether mismatch in number facilitated interpretation. In line with previous research, object- were consistently less accurate than subject-questions. Initial subject-bias interpretation of object-questions was evidenced through an initial increase in looks to a competitor image depicting the same action with reversed argument structure (subject-question) followed by a protracted reorientation to the correct image. Crucially, mismatch in number between the two NPs aided recovery from misinterpretation (e.g. “Which donkey are the zebras carrying?” was easier to process than “Which donkey is the zebra carrying?”) in line with Relativized Minimality (Rizzi, 2004).

In the present study, 31 8-11-year-old bilingual children of mixed linguistic backgrounds but highly proficient in English and 37 age-matched monolingual children took part in a similar visual world paradigm experiment as in Contemori et al (2018). We manipulated question type (subject vs. object which-question) and number of the second NP (match or mismatch to first) to address the following research questions: 1) do monolingual and bilingual children differ in the timecourse of wh-question processing? 2) Is there incrementality and similar recovery? 3) Do bilingual children make use of number mismatch as a facilitatory cue? 

Overall looks to target did not differ quantitatively between groups. There was a significant interaction of group with time but not question type, reflecting an across-group subject-bias but reanalysis was slower for the bilinguals. The effect of number mismatch was significant in the gaze data for object-questions only for the monolinguals. In the bilinguals, it was found for time needed for picture selection after hearing a sentence, potentially a spillover effect. 

The results suggest equally accurate and qualitatively similar, incremental but time-wise more protracted processing of which-questions for bilinguals. Both groups utilised number mismatch, in line with Relativized Minimality, but the effect in the bilingual children was after the end of the sentence providing additional support from non-violation paradigms that bilingual children process morphosyntactic information in a similar way as monolingual children but at a slower speed. -