Poster D32, Friday, August 17, 4:45 – 6:30 pm, Room 2000AB
An ERP study of Q-particle licensing and question-answer congruence in Korean
Daeho Chung1, Kiyong Choi2, Wonil Chung3, Say Young Kim1, Bum-Sik Park3, Myung-Kwan Park3;1Hanyang University, 2Kwangwoon University, 3Dongguk University
This ERP study examines the licensing relation between a clause-final question particle (Q-particle) and the preceding wh-phrase (WP) and the congruence between a yes/no or wh-question and its answer in Korean Kyengsang dialect, which employs different Q-particles depending on the types (-na for yes/no-question; -no for wh-question). The experimental materials consisted of 240 sets of 8 items, with a 2x2x2 factorial design: (i) the clause-final Q-particle factor (-na/-no), (ii) the Q-particle-licensing factor ((im)proper licensing between an embedded WP and the Q-particle), and (iii) the question-answer (in)congruence factor (a(n) (in)correct answer to a yes/no or wh-question). The structure of the materials is as follows; A: ni-nun yenghuy-ka mwusun chayk-ul ilk-ess-tako (or -nunci) alko iss-no (or -na)? ‘Do you know whether Yenghi read some book?/Which book do you know that Yenghi read?’ B: soselchayk-ul (or ung, alko iss-e) ‘A novel (or Yes, I do)’. Twelve Korean speakers (male: 8) participated in the ERP experiment. The ANOVA results based on the descriptive data of the offline acceptability task showed that there are significant main effects of such factors as clause-final Q-particle (F(1,11)=20.90, p<0.001), Q-particle-licensing (F(1,11)=8.92, p<0.05), and question-answer congruence (F(1,11)=9.65, p<0.01). ERPs were measured at the matrix verbs with Q-particles (alko iss-na/-no) and the answers (soselchayk-ul/ung, (alko iss-e) ‘A novel/Yes, I do’). In pairwise comparison between the matrix verbs with the Q-particles, there was a significant anterior negativity effect between a Q-particle-licensing violation (the embedded and matrix verbal complex: ilk-ess-nunci alko iss-na) and its control (ilk-ess-tako alko iss-na), but no such effect between a wh-island violation (ilk-ess-nunci alko iss-no) and its control (ilk-ess-tako alko iss-no). In pairwise comparison between a Q-particle-licensing violation and its control, there was a RAN effect at 350-500 ms at the anomalous fragment answer to the yes/no-question (e.g, sosel-ul ‘A novel’) relative to the felicitous counterpart to the wh-question, and an anterior P600 effect at the anomalous yes/no answer to the wh-question (e.g., ung ‘yes’) relative to its counterpart to the wh-question. In pairwise comparison between a wh-island violation and its control, there was a left sustained negativity effect at the anomalous fragment answer to the yes/no-question (e.g., sosel-ul ‘A novel’) relative to the felicitous counterpart to the wh-question, and a sustained positivity effect at the anomalous yes/no answer to the wh-question (e.g., ung ‘yes’) relative to its counterpart to the wh-question. The results indicate that the wh-island violation is not in full force since the embedded WP is not required to be associated with the embedded Q-particle. However, the Q-particle-licensing requirement comes into effects, in that the embedded WP tends to be associated with the matrix wh-licensing Q-particle across the embedded non-Q complementizer. Regarding the question-answer (in)congruence, the incongruent fragment answers elicited a negativity, whereas the incongruent polarity answer particles evoked a positivity. One outstanding thing to note is that when the embedded verb complex contains a Q-particle, the positivity or negativity of ERP responses is sustained. This suggests that the conditions with this kind of structural configuration call for more cognitive demands in meeting the question-answer congruence.
Topic Area: Grammar: Syntax