Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Tracking overnight lexicalization of non-native phonological contrasts
Poster Session C, Thursday, October 1, 10:45 am - 12:45 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Kangdi Liu1, Quentin Zhen Qin1; 1The Hong Kong University of Science and Technology
Overnight sleep facilitates the lexicalization of newly learned words. For instance, Liu et al. (2024) had participants learn one list of words (non-minimal pairs, differing by multiple phonological units) on Day 1 and another list on Day 2. An identification test was administered following the Day 2 learning. Results revealed faster reaction times (RTs) for Day 1 than Day 2 words, indicating that Day 1 words (learned earlier, remote) had the sleep benefit, whereas Day 2 words (learned before testing, recent) did not. We extend this to minimal pairs in a non-native language (differing by one phonological unit). Cantonese has a more complex tonal and vowel system than Mandarin. Mandarin speakers may struggle with Cantonese phonological contrasts (e.g., Tone 5 (T5)–T6, /ɛ:/-/ɔ:/). Thus, this study investigates whether Mandarin-speaking participants can lexicalize non-native phonological contrasts after an overnight interval. Mandarin-speaking young adults (aged 18–35 years) were recruited (data collection ongoing). Auditory stimuli were 32 Cantonese disyllabic pseudo-words forming eight minimal quadruplets contrasting in tone and vowel (e.g., /pɛ:5fu:1/, /pɛ:6fu:1/, /pɔ:5fu:1/, /pɔ:6fu:1/), paired arbitrarily with 32 everyday objects. Participants learned 16 words on Day 1 and the other 16 on Day 2. Daily training (45 min) comprised one familiarization phase (free listening and viewing, 5 min), after which participants completed separate training for tone and vowel contrasts. For each contrast, they performed fixed learning (word-object matching with feedback; 192 trials; 10 min) and then free learning (clicking an object and hearing “this is [target word], not [the minimal tone/vowel pair of the target word]”, 10min). The order of tone and vowel training was counterbalanced across participants. Learning outcomes were assessed by an identification test (256 trials). Independent variables were Type (minimal vs. non-minimal, with minimal further divided into tone and vowel) and Condition (remote vs. recent). A mixed-effects logistic model was analyzed on accuracy, showing a main effect of Type. Accuracy was higher for non‑minimal than for minimal pairs (β=1.80, SE=0.21, z=8.63, p<.001). Within minimal pairs, vowel contrasts yielded higher accuracy than tone contrasts (β=0.79, SE=0.18, z=4.31, p<.001). For RTs, Type produced a significant main effect (β=-74.27, SE=25.31, z=-2.94, p=.01), with non‑minimal pairs eliciting shorter RTs than minimal pairs (vowel and tone contrasts did not differ from each other). Importantly, the main effect of Condition was significant (β=-58.86, SE=16.93, z=-3.48, p<.001), with shorter RTs in the remote than recent words, suggesting sleep benefits for both minimal and non‑minimal pairs. In the current study, accuracy revealed that minimal pairs were more difficult to learn than non‑minimal pairs, with tone contrasts being the hardest. Critically, despite this difficulty, sleep still provided overnight benefits in RTs: remote words (with overnight sleep) showed faster RTs than recent words (without sleep) for both tone and vowel contrasts. Thus, sleep supports language learning even for the most challenging phonological contrasts. These findings advance our understanding of how difficult L2 sounds can be learned at an early stage. We are currently collecting ERP data on lexicalization. A future study will examine another experimental group trained using virtual reality.
Topic Areas: Speech Perception, Language Development/Acquisition