Hanulíková, A. , Davidson, D. & Carreiras, M.
Basque Center on Cognition, Brain and Language (BCBL). Donostia. Spain
An important property of speech is that it explicitly conveys features of the speaker’s identity. Although previous studies have shown that speaker’s identity affects semantic processing (Van Berkum et al. 2008), it is not clear how a speaker’s identity will influence grammatical processing. Here we investigate subject-verb agreement in Slovak, when the agreement depends on the speaker’s gender (as cued by his/her voice) compared to when it depends on the formal grammatical gender of the subject. In the Slovak past participle, for example, agreement in the 3rd person singular depends on the formally-marked grammatical gender of the subject (e.g., žena išla domov ‘(the) womanFEM wentFEM home’). However, in the 1st person singular, the agreement is based on the speaker’s gender (e.g., ja som išla domov ‘IFEM/MASC amFEM/MASC wentFEM home’). The personal pronoun ja ‘I’ is not formally marked for gender, it is the speaker’s gender that determines the pronoun-verb agreement. In this study we compared the ERP response to verbs disagreeing with the formally-marked subject’s gender (e.g., ‘(the) neighbours were upset because (the) mother-in-lawFEM *stoleMASC plums’), and to verbs disagreeing with subject’s gender as conveyed by the speaker’s voice (e.g., ‘(the) neighbours were upset because IFEM *stoleMASC plums’). Slovak native speakers (n=32) listened to 240 sentences spoken either by a woman or a man, and received 24 comprehension questions to draw their attention to the content. Gender disagreement between the formally-marked subject and the verb resulted in a P600 effect (larger for violations compared to controls) preceded by an anterior negativity (larger for violations). However, gender disagreement based on the speaker’s voice did not elicit a P600, but rather a posterior negativity (larger for violations). We will discuss how checking or repair mechanisms might be different when agreement is mediated by the speaker’s voice as compared to formally-marked syntactic features.