Recent projects:
Individual predictors of response to biofeedback training for second-language production
Previous studies reported that the application of visual biofeedback facilitates speech sound learning in L2 learners and individuals with speech sound disorders. However, individual responses to biofeedback training have been highly heterogeneous and the reason is not well explained in the literature. The current study explored predictors of learning success from biofeedback training. English speakers were trained to produce Mandarin vowels and they were randomly assigned to either a visual-acoustic or ultrasound biofeedback training condition. Individual learners’ auditory and somatosensory acuity were measured and hypothesized to be correlated with progress in production accuracy. A few other potential predictors (e.g. phonological awareness) were also examined. This line of research could inspire individualized learning that speech training paradigms may be customized based on individual learners’ sensorimotor characteristics.
(collaborated with Samantha Ayala, Daphna Harel, Douglas Shiller & Tara McAllister)
Effect of Biofeedback Training on online correction in L2 vowel production
Online correction in natural speech production has been reported, showing within-syllable correction from peripheral to more prototypical realization of the target in acoustic space. In the production of vowels, online correction could be measured as a reduction in acoustic variability from vowel onset to the midpoint, in which onset variability indicates feedforward control and reduction of variability may reflect feedback correction. Production of L2 vowels might demonstrate less online correction because of the weak sensory targets and unfamiliar motor plans. This study investigates the effect of short-period biofeedback training (visual-acoustic & ultrasound biofeedback) on online correction in L2 speech sound production. Given that visual-acoustic and ultrasound biofeedback may facilitate the establishment of auditory and somatosensory targets of L2 vowels respectively, it is hypothesized that the training might not affect onset variability but will increase the magnitude of online correction. The effect of biofeedback type will also be examined.
(collaborated with Samantha Ayala, Daphna Harel, Douglas Shiller, Caroline Niziolek & Tara McAllister)
Perception and production of lexical stress by Cantonese and Mandarin learners of English
This study examines the perception and production of English lexical stress by Mandarin and Cantonese learners of English. Mandarin and Cantonese do not use stress contrast at the word level. However, Mandarin has neutral tone that shares acoustic similarities with the English unstressed syllable. Therefore, Mandarin speakers might have an advantage over Cantonese speakers in perceiving and producing native-like English lexical stress contrast. Mandarin and Cantonese speakers who are late learners of English participated in the study. A sound discrimination task was used to test the stress perception, and a word repetition task was conducted to examine the production. A facial capture system was used to track lips and jaw movement in three dimensions during the repetition task. Both kinematic and acoustic analyses were conducted. Results revealed that Mandarin speakers had a generally more native-like performance in the perception and production of lexical stress contrast. The results suggested that perception and production of non-native lexical stress could be influenced by L1 prosodic features.
(collaborated with Maria Grigos)
Previous projects:
Cross-modal association between vowels and colours: A cross-linguistic perspective
Download full paper Copyright (2019) Acoustical Society of America. This article may be downloaded for personal use only. Any other use requires prior permission of the author and the Acoustical Society of America. The article appeared in The Journal of the Acoustical Society of America 2019 145:4, 2265-2276 and may be found at https://doi.org/10.1121/1.5096632.
This project examined cross-model association between colors and linguistic components (vowels and pitch in a tone language) in non-synaesthetes. Native and non-native speakers of the tone language were tested, and the effect of language background was also examined. We found associations between red and /a/, light colors with /i/ and dark colors with /u/. A high tone is generally associated with lighter colors. Speakers with different L1s also had variable performance. This project suggested that cross-model association is a general phenomenon and linguistic factors could play a role.
(collaborated with Peggy Mok, Guo Li, Hezul Ng & Him Cheung)
Orthographic effects on the perception and production of L2 mandarin tones
Mandarin Chinese makes use of an opaque writing system (Chinese characters) and a transparent spelling system (Pinyin). This study investigated how the orthography may affect the performance in perceiving and producing Mandarin lexical tones in Cantonese learners of Mandarin. Interaction between the type of orthography and language learners’ L2 proficiency was observed. Results showed that a transparent system is not always more facilitative than an opaque system.
(collaborated with Peggy Mok, Albert Lee & Robert Xu)
The Acquisition of English Lexical Stress by Cantonese-English Bilingual Children at 2;06 and 3;0
Acoustic cues related with lexical stress in English are made use of differently in Cantonese, a tone language that has a syllable-timed rhythm. This study investigated the development of English lexical stress in simultaneous Cantonese-English bilingual children at the age of 2;06 and 3;0. The aim was to understand how the bilingual children’s experience with Cantonese prosodic features would impact the manipulation of acoustic cues in delivering English lexical stress at different developmental stages.
(collaborated with Peggy Mok)