Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Prior knowledge benefits nap-mediated consolidation of Cantonese tones through talker generalization
Poster Session D, Saturday, October 26, 10:30 am - 12:00 pm, Great Hall 3 and 4
Zhen Qin1, Ruofan Wu2, Caicai Zhang2; 1The Hong Kong University of Science and Technology, 2The Hong Kong Polytechnic University
Post-training sleep is beneficial for consolidating new sounds in second language (L2) learners. However, it is not yet clear how prior phonological knowledge affects the consolidation of new lexical tones. This study investigates whether nap-meditated memory consolidation of L2 tones depends on prior tonal knowledge, by manipulating the similarity of L2 Cantonese tonal contrasts (level-level or contour-level tonal contrasts) to L1 Mandarin tonal contrasts. While contour-level tonal contrasts are familiar to Mandarin listeners, level-level tonal contrasts are novel and perceptually challenging. Importantly, lexical tones are characterized by high variability across talkers, and previous studies have shown that sleep-dependent consolidation promotes talker generalization. Therefore, the present study adopted a nap design to investigate whether nap helps Mandarin speakers consolidate Cantonese tones through talker generalization and whether the consolidation effect depends on prior knowledge of tonal contrasts. Ninety Mandarin speakers from North China were recruited for the experiments. They were trained with two Cantonese contour-level (T5-T6) tonal contrasts and two level-level (T3-T6) tonal contrasts, respectively (240 trials each tonal pair, order counterbalanced), followed immediately with the 1st tone identification (ID 1) task using the stimuli produced by the trained (male) talker (80 trials for each tonal pair). Participants, matched in nap habits and sleep chronotypes, were then pseudo-randomly assigned to either the nap group (50 participants), who napped for 1.5 hours with brain EEG activities recorded, or the non-nap group, who rested and stayed awake by watching silent documentaries for 1.5 hours. After the nap manipulation, the participants were tested again with the trained talker in the 2nd ID task and a novel (female) talker in the 3rd ID task. Mixed-effects models were conducted on the ID accuracy across sessions. The ID 1-2 model, comparing the accuracy before and after the nap, showed the main effects of Tone (z = −18.88, p < .001) and Session (z = 2.27, p = .02) but no significant interaction effects involving Group. The ID 2-3 model, comparing the accuracy of trained and novel talkers, revealed a 3-way interaction between Group, Session, and Tone (z = 2.37, p = .02). Follow-up models showed that only the nap group had an interaction between Tone and Session (z = −2.41, p = .02), but not the non-nap group. The nap group showed a smaller ID difference between the trained and novel talkers for contour-level than for level-level tonal contrasts, whereas the non-nap group did not show the effect. The results of sleep-related EEG activities showed that sleep spindle density (z = 2.28, p = .02) and slow-wave sleep percentage (z = 2.75, p < .01) positively predicted the talker generalization effect in the contour-level tonal contrasts. The findings support the beneficial role of naps, which enhanced the nap participants’ tone consolidation by promoting talker generalization. Importantly, nap appears to prioritize the generalization of contour-level tonal contrast to a new talker, an effect attributable to the participants’ prior knowledge of contour-level tonal contrasts. The beneficial effect might reflect brain activities indexing the hippocampal-neocortical cycle underlying memory consolidation processes.
Topic Areas: Prosody, Speech Perception