research
大象传媒 researchers present at New Sounds conference
, a conference that was first organized in 1990 with 30 participants, now attracts 200+ attendees. Presenters share new insights on second language speech in a range of keynotes, oral presentations and poster sessions.
The aim of the conference is to bring together researchers, teachers and learners interested in multilingualism, phonetics and phonology. This year, New Sounds took place during April 23rd 鈥� April 25th and was hosted by the University of Toronto.
Professor Yue Wang, Director of the Language and Brain Lab, presented a keynote titled Harnessing a symphony of cues in nonnative speech perception. Wang鈥檚 talk explored how clearer sound and visual cues help nonnative speakers better understand a new language, using behavioral and computational methods to uncover insights for language learning and speech technology. The work was co-authored by Allard Jongman, Joan A. Sereno and Dawn M. Behne, along with collaboration from 大象传媒 colleagues Ghassan Hamarneh, Angelica Lim, Paul Tupper, Henny Yeung, and students from the Language and Brain Lab. You can view the slides here. See also the abstract below.
A new aspect added to this year鈥檚 New Sounds was the Doctoral Workshop during which PhD students presented their research. Department of Linguistics PhD student Danielle Deng presented research from her dissertation in a talk titled Acoustic and sentential cues on English learners鈥� adaptation to non-native English accents: Using artificially generated voices to test accent adaptation. Deng鈥檚 work focuses on L2 speech perception, especially in accent adaptation and word recognition.
Leon Lee, a new alumnus of the MA program at the Department of Linguistics who has recently been accepted to UBC鈥檚 doctoral program for speech-language pathology, presented research titled Beyond binary English word stress: A comparison of English, Mandarin, and Russian native speakers鈥� production, co-authored by Magdalena Ivok and Henny Yeung. Lee鈥檚 MA research focused on how orthography input influences learners鈥� L2 phonological processing, a topic explored during his second presentation titled The impact of congruent and systematic orthographic input on L2 word learning (co-authored by Henny Yeung).
Department of Linguistics PhD student Han Zhang presented Native-nonnative speech adaptations in conversation: Effects of L1 background, co-authored by Danielle Deng, Paul Tupper, Joan A. Sereno, Allard Jongman, Dawn M. Behne and Yue Wang. This research explores the dynamic adaptive strategies nonnative speakers use to mitigate miscommunication, in particular how their strategies change in spontaneous conversations and are affected by L1 backgrounds. You can read the abstract below.
Abstracts
Harnessing a symphony of cues in nonnative speech perception
By Yue Wang, Allard Jongman, Joan A. Sereno and Dawn M. Behne
View the slide presentation here
Nonnative speech perception often relies on resources beyond the speech signal, such as expectations of clearer speech and increased attention to visual cues (e.g., facial movements). This talk explores how language-general and language-specific clear-speech cues across auditory and visual modalities are orchestrated to enhance nonnative speech intelligibility.
We begin by examining how native speakers modify their speech through language-general modifications (e.g., speaking louder or slower) and language-specific changes (e.g., preserving phonemic distinctions) with the goal of improving clarity. We then investigate how nonnative perceivers respond to these cues in different speech styles and modalities. Our findings suggest that nonnative perceivers benefit more from language-general clear-speech cues than native perceivers.
However, despite their increased reliance on visual cues, they are also more vulnerable to visual modifications which blur categorical distinctions. Furthermore, native-language experience influences whether intelligibility gains (or losses) are driven by language-general or language-specific effects.
These findings highlight the importance of effectively integrating auditory and visual clear-speech cues to optimize speech intelligibility for nonnative perceivers, with implications for second-language learning and speech technology applications.
Native-nonnative speech adaptations in conversation: Effects of L1 background
By Han Zhang, Danielle Deng, Paul Tupper, Joan A. Sereno, Allard Jongman, Dawn M. Behne and Yue Wang
Phonetic adaptations occur when speakers adjust their speech productions by converging to or diverging from their interlocutor鈥檚 [2,6]. Although such adaptations are also observed in native-nonnative speaker interactions [4,5], less is known about the dynamic adaptive strategies nonnative speakers use to mitigate miscommunication, particularly how their strategies change in spontaneous conversations and are affected by L1 backgrounds.
We address these questions by examining phonetic adaptations in unscripted natural conversations between native English speakers and nonnative speakers with different L1s (Japanese, Mandarin) in an interactive computer game task involving English words contrasting tense and lax vowels (e.g., sheep-ship). Japanese and Mandarin speakers, due to their different L1 experience with spectral (quality) and temporal (duration) vowel attributes, are paired with English speakers. While native English speakers distinguish tensity primarily by vowel quality and secondarily by duration [3], Japanese speakers weigh temporal cues more than spectral cues, presumably being influenced by phonemic vowel length contrasts in Japanese [1]. By contrast, Mandarin speakers lack L1 experience with either vowel tensity or phonemic length distinctions [7].
Acoustic analyses (F1, F2, duration) are conducted to examine changes in vowel productions during the conversation task to determine the extent to which native and nonnative vowel productions adapt to overcome miscommunication and how nonnative speakers鈥� (spectral and temporal) cue-weighting strategies change over the conversation.
Preliminary results of four English-Mandarin interlocutor pairs reveal a spectral (but not temporal) learning trajectory by Mandarin speakers as the conversation progresses, with increased F1 and F2 differences that enhance tense-lax vowel contrasts. Interestingly, native English speakers are also found to accommodate nonnative spectral patterns to resolve miscommunications. Analyses of English-Japanese data and additional English-Mandarin data are in progress.
These initial findings suggest that native and nonnative speakers dynamically adapt their phonetic cue-weighting strategies in spontaneous conversations, offering insights for how adaptation strategies benefit L2 speech intelligibility and learning.
References
[1] Bohn, O. S., & Flege, J. E. (1990). Interlingual identification and the role of foreign language experience in L2 vowel perception. Applied Psycholinguistics, 11(3), 303-328.
[2] Fusaroli, R., & Tyl茅n, K. (2016). Investigating conversational dynamics: Interactive alignment, interpersonal synergy, and collective task performance. Cognitive science, 40(1), 145鈥�171.
[3] Hillenbrand, J., Getty, L. A., Clark, M. J., & Wheeler, K. (1995). Acoustic characteristics of American English vowels. The Journal of the Acoustical Society of America, 97(5 Pt 1), 3099鈥�3111.
[4] Kim, M., Horton, W. S., & Bradlow, A. R. (2011). Phonetic convergence in spontaneous conversations as a function of interlocutor language distance. Laboratory Phonology, 2(1), 125鈥�156.
[5] Olmstead, A., Viswanathan, N., Cowan, T. & Yang, K. (2021). Phonetic adaptation in interlocutors with mismatched language backgrounds: A case for a phonetic synergy account. Journal of Phonetics, 87.
[6] Pardo, J. S. (2006). On phonetic convergence during conversational interaction. The Journal of the Acoustical Society of America, 119(4), 2382鈥�2393.
[7] Yuan, J. (2013). The spectral dynamics of vowels in Mandarin Chinese. Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. 1193-1197.