Poster B23, Thursday, August 16, 3:05 – 4:50 pm, Room 2000AB
Voice (mis)match in Korean right dislocation constructions: An ERP study
Bum-Sik Park1, Kiyong Choi2, Daeho Chung3, Wonil Chung1, Say Young Kim3, Myung-Kwan Park1;1Dongguk University, 2Kwangwoon University, 3Hanyang University
It has been reported that voice mismatch in elliptical contexts leads to degradedness, suggesting that (limited) identical syntactic forms are required for ellipsis licensing (Merchant 2007). The purpose of this study is to investigate the processing effect of voice mismatch in Korean right dislocation (RD) constructions, where the sentence final RDed remnant is assumed to be derived via ellipsis (cf. Park and Kim 2009, Ott & de Vries 2015). We conducted an ERP experiment, employing 240 sets of eight elliptical RD conditions in Korean with 2x2x2 factorial design (voice types of the antecedent clause, explicit/implicit correlate, (mis)match of the remnant). Representative materials are below, where implicit correlates are within parentheses: (1) Active condition (explicit/implicit, voice match/mismatich): (Max-ka) namwu-lul caluasse, Max-ka/*Max-eyuyhay. ‘Max cut the tree, Max/*by Max’. (2) Passive condition (explicit/implicit, voice match/mismatch): (Max-eyuyhay) namwu-ka caliuasse, Max-eyuyhay/*Max-ka. ‘The tree was cut by Max/*Max’. First of all, we collected the ANOVA results of the offline acceptability task with the materials, and they showed that Korean speakers consistently judged the three mismatch conditions unacceptable except the passive implicit mismatch condition in (2). In particular, the mismatch in the active implicit condition was rated the worst among them in (1). The results were generally consistent with online acceptability task except the explicit and implicit condition In general, these results were also consistent with those from the ERP recordings. Sixteen Korean speakers participated in the experiment, and ERPs were measured at the sentence final RDed remnant (Max-ka ‘Max-Nom’/Max-eyuyhay ‘Max-by’), the case/postposition marker of which cues a different voice type of the elliptical clause. At this critical element, similar/different ERP components were obtained among conditions. In pairwise comparison, the explicit/mismatch/passive condition in (2) recorded P200, P400, and a marginal P600/sustained positivity, whereas the implicit/mismatch/passive condition recorded P200 followed by a significant P600. By contrast, the explicit/mismatch/active condition in (1) recorded a significant N400 and N600, whereas the implicit/mismatch/active condition recorded a marginal N400 followed by significant N600. We take these ERP results to point to three things. First, the P200 and the P600 component elicited commonly in the passive/mismatch condition in (1) can be ascribed to the presence of structural case on the remnant (i.e., Nominative), This amounts to the structural case mismatch on the remnant, and possibly resulted in eliciting the P600(-like) component as an index of structural repair of voice mismatch. Second, by contrast, the components N400 and N600 evoked commonly in the active/mismatch condition is due to the lexical/morphological case/postposition mismatch on the remnant, which ended up yielding the component as an index of semantic integration. Finally, while the explicit conditions involving voice mismatch had ERP effects around 400 ms, the implicit conditions involving voice mismatch had prolonged ERP effects around 600 ms. This suggests that the implicit correlate yielded a intensifying effect of voice mismatch.
Topic Area: Grammar: Syntax