Double-Walled Sound Cabin located in Phonology Lab, SW149, UTSC campus
Research Overview
My Ph.D. training primarily focused on theoretical phonology. Since then, the scope of my research has expanded to include the interface between phonology and phonetics, with sociolinguistics and psycholinguistics.
A common perspective that underlies multiple strands of my research is the view that linguistic phenomena are, at their core, learning problems. How do speakers come to exhibit attested linguistic patterns given the input available to them? What is the function that maps this input to the output speakers actually produce? What structural constructs mediate the mapping between input and output? Are there innovations in speakers’ output that are not straightforwardly derivable from the input—and if so, what do these innovations reveal about the nature of the learning mechanism? Is there evidence that the mapping function selectively attends to certain aspects of the input over others—and if so, what is the source of this bias? Do extra-grammatical factors affect the mapping between input and output—and if so, how?
I address these questions in part by examining what can be considered “non-traditional” sources of linguistic data, such as loanword adaptation, language change and dialectal variation, and heritage languages. These domains are especially fruitful for investigating the relationship between input and learned output, as they often involve innovations where the output does not directly mirror the input. Loanword adaptation and language change or variation, by definition, involve mismatches between input and output. Heritage speakers exhibit distinct linguistic traits—presumably due to different patterns and amounts of exposure to learning data compared to monolingual first-language learners—which can be viewed as a type of innovation. These contexts provide opportunities to test aspects of the mapping function that might remain obscured in traditional phonological data.
Phonetic variation in speech is also a productive area of research. For a long time, speech variability was seen as a problem to be overcome in the search for invariance. More recently, however, researchers have come to recognize that highly structured variation actually facilitates communication by providing cues to linguistic context and speaker identity. Listeners possess intricate knowledge of this variation and can adjust their perception accordingly to identify intended speech targets. My current research examines this variation through corpus-based and experimental studies and explores how and when perceptual adaptation fails—and what this reveals about mechanisms of sound change.
In all my research, I aim to combine strong theoretical grounding with rich empirical data drawn from large-scale fieldwork and corpora. A detailed understanding of primary data is a valuable and lasting contribution in its own right, and it also allows for the sharpening of hypotheses about the nature of the mapping function, which in turn informs further empirical inquiry. Methodologically, I draw on corpus, computational, statistical, and experimental techniques.
Below are brief synopses of my current research areas:
Loanword Phonology
My research investigates diverse factors conditioning loanword adaptation, including perceptual, phonological, historical, and sociolinguistic influences. I use data from experimental studies (production and perception), corpora, dictionaries, and elicited production tasks. Computational and statistical tools are employed to uncover and test empirical generalizations.
Collaborators: Sungwoo Han, Alexei Kochetov & Eunjong Kong
Funding: SSHRC Standard Research Grant (2011-12)
Sound Change
I study sound change in progress in Korean, using apparent-time corpora and fieldwork from multiple Korean dialects. These include Seoul Korean, Heritage Korean in Northern China and Canada, and Hamkyoung Korean spoken by recent North Korean refugees. I address theoretical issues such as: (i) the word frequency effect in sound change and its implications for the mental representation of speech sounds and the mechanisms of change, and (ii) the interaction between production and perception in ongoing sound change.
Collaborators: Sungwoo Han
Funding: SSHRC Insight Grant (2013-18)
Dialect Variation and Contact
We explore phonological and phonetic variation, as well as contact-induced changes, in Korean dialects. Data are drawn from the NIKL Dialect Survey Corpus and fieldwork. Examined dialects include Seoul Korean, two dialects of Chinese Korean, and Hamkyoung Korean spoken by North Korean refugees.
Funding: SSHRC Partnership Development Grant (2013-15)
Heritage Language Phonetics and Phonology
This research focuses on phonetic and phonological variation in Heritage Korean and Tagalog in Toronto. Topics include the role of phonological universals, inter-language interaction in bilinguals, and sound change in heritage languages.
Funding: SSHRC Insight Grant (2013-19)
Hyperspeech as a Diagnostic for Sound Change Mechanisms
This emerging line of research investigates how hyper-articulated speech (as opposed to hypo-articulated speech) can reveal dynamic aspects of the mental representation of speech sounds. It aims to provide crucial evidence about the mechanisms underlying ongoing sound change.
Funding: SSHRC Insight Grant (2019-26)
Speech Rate, Age, and Sound Change
We explore how speech rate differences—particularly between younger and older speakers—might drive sound change. Younger speakers tend to speak faster, and rapid speech promotes the reduction of durational and segmental contrasts. At the same time, listeners compensate perceptually for speech rate variation. I examine intra- and inter-speaker variability in rate compensation and its potential to create conditions for "mini-sound change." This work uses both production and perception experiments involving speakers and listeners across age and gender groups.
Funding: SSHRC Insight Grant (2019-26)
Phonology of Gender in First Names
We are conducting cross-linguistic studies on the relationship between sound and perceived gender in personal names. This project aims to document patterns across typologically and genetically diverse languages, providing an empirical foundation for further research into how these patterns interact with language-specific morphological and phonological constraints.
Funding: SSHRC Insight Development Grant (2022-25)
Name gender and sentence processing
In this research, we examine how the gender of personal names is processed during sentence reading, especially when the gender is ambiguous but biased toward one interpretation. We also explore how these tasks can be extended to nonce names, where the phonological features of made-up names may affect gender interpretation.
Speech foundation models, interlanguage perception, and loanwords
In this SSHRC-funded project, we investigate whether speech foundation models trained solely on speech input (without orthographic input) can reproduce human behavior in interlanguage perception (e.g., English input → Korean output) and loanword adaptation, using the same stimuli presented to human listeners in adaptation experiments. By manipulating different aspects of the models (such as the amount of bilingual exposure), we probe their inner workings and gain insights into the underlying processes. Furthermore, the models can be experimentally manipulated to assess the factors crucial to their success.
Funding: SSHRC Insight Development Grant (2025–28)
Neural encoding of secondary acoustic cues in speech perception
In this research, we investigate how English and Korean speakers encode laryngeal contrasts cued by VOT (the primary cue in English) and F0 (the primary cue in Korean).
Funding: University of Toronto Scarborough International Research Fund (2024–25)
A common perspective that underlies multiple strands of my research is the view that linguistic phenomena are, at their core, learning problems. How do speakers come to exhibit attested linguistic patterns given the input available to them? What is the function that maps this input to the output speakers actually produce? What structural constructs mediate the mapping between input and output? Are there innovations in speakers’ output that are not straightforwardly derivable from the input—and if so, what do these innovations reveal about the nature of the learning mechanism? Is there evidence that the mapping function selectively attends to certain aspects of the input over others—and if so, what is the source of this bias? Do extra-grammatical factors affect the mapping between input and output—and if so, how?
I address these questions in part by examining what can be considered “non-traditional” sources of linguistic data, such as loanword adaptation, language change and dialectal variation, and heritage languages. These domains are especially fruitful for investigating the relationship between input and learned output, as they often involve innovations where the output does not directly mirror the input. Loanword adaptation and language change or variation, by definition, involve mismatches between input and output. Heritage speakers exhibit distinct linguistic traits—presumably due to different patterns and amounts of exposure to learning data compared to monolingual first-language learners—which can be viewed as a type of innovation. These contexts provide opportunities to test aspects of the mapping function that might remain obscured in traditional phonological data.
Phonetic variation in speech is also a productive area of research. For a long time, speech variability was seen as a problem to be overcome in the search for invariance. More recently, however, researchers have come to recognize that highly structured variation actually facilitates communication by providing cues to linguistic context and speaker identity. Listeners possess intricate knowledge of this variation and can adjust their perception accordingly to identify intended speech targets. My current research examines this variation through corpus-based and experimental studies and explores how and when perceptual adaptation fails—and what this reveals about mechanisms of sound change.
In all my research, I aim to combine strong theoretical grounding with rich empirical data drawn from large-scale fieldwork and corpora. A detailed understanding of primary data is a valuable and lasting contribution in its own right, and it also allows for the sharpening of hypotheses about the nature of the mapping function, which in turn informs further empirical inquiry. Methodologically, I draw on corpus, computational, statistical, and experimental techniques.
Below are brief synopses of my current research areas:
Loanword Phonology
My research investigates diverse factors conditioning loanword adaptation, including perceptual, phonological, historical, and sociolinguistic influences. I use data from experimental studies (production and perception), corpora, dictionaries, and elicited production tasks. Computational and statistical tools are employed to uncover and test empirical generalizations.
Collaborators: Sungwoo Han, Alexei Kochetov & Eunjong Kong
Funding: SSHRC Standard Research Grant (2011-12)
Sound Change
I study sound change in progress in Korean, using apparent-time corpora and fieldwork from multiple Korean dialects. These include Seoul Korean, Heritage Korean in Northern China and Canada, and Hamkyoung Korean spoken by recent North Korean refugees. I address theoretical issues such as: (i) the word frequency effect in sound change and its implications for the mental representation of speech sounds and the mechanisms of change, and (ii) the interaction between production and perception in ongoing sound change.
Collaborators: Sungwoo Han
Funding: SSHRC Insight Grant (2013-18)
Dialect Variation and Contact
We explore phonological and phonetic variation, as well as contact-induced changes, in Korean dialects. Data are drawn from the NIKL Dialect Survey Corpus and fieldwork. Examined dialects include Seoul Korean, two dialects of Chinese Korean, and Hamkyoung Korean spoken by North Korean refugees.
Funding: SSHRC Partnership Development Grant (2013-15)
Heritage Language Phonetics and Phonology
This research focuses on phonetic and phonological variation in Heritage Korean and Tagalog in Toronto. Topics include the role of phonological universals, inter-language interaction in bilinguals, and sound change in heritage languages.
Funding: SSHRC Insight Grant (2013-19)
Hyperspeech as a Diagnostic for Sound Change Mechanisms
This emerging line of research investigates how hyper-articulated speech (as opposed to hypo-articulated speech) can reveal dynamic aspects of the mental representation of speech sounds. It aims to provide crucial evidence about the mechanisms underlying ongoing sound change.
Funding: SSHRC Insight Grant (2019-26)
Speech Rate, Age, and Sound Change
We explore how speech rate differences—particularly between younger and older speakers—might drive sound change. Younger speakers tend to speak faster, and rapid speech promotes the reduction of durational and segmental contrasts. At the same time, listeners compensate perceptually for speech rate variation. I examine intra- and inter-speaker variability in rate compensation and its potential to create conditions for "mini-sound change." This work uses both production and perception experiments involving speakers and listeners across age and gender groups.
Funding: SSHRC Insight Grant (2019-26)
Phonology of Gender in First Names
We are conducting cross-linguistic studies on the relationship between sound and perceived gender in personal names. This project aims to document patterns across typologically and genetically diverse languages, providing an empirical foundation for further research into how these patterns interact with language-specific morphological and phonological constraints.
Funding: SSHRC Insight Development Grant (2022-25)
Name gender and sentence processing
In this research, we examine how the gender of personal names is processed during sentence reading, especially when the gender is ambiguous but biased toward one interpretation. We also explore how these tasks can be extended to nonce names, where the phonological features of made-up names may affect gender interpretation.
Speech foundation models, interlanguage perception, and loanwords
In this SSHRC-funded project, we investigate whether speech foundation models trained solely on speech input (without orthographic input) can reproduce human behavior in interlanguage perception (e.g., English input → Korean output) and loanword adaptation, using the same stimuli presented to human listeners in adaptation experiments. By manipulating different aspects of the models (such as the amount of bilingual exposure), we probe their inner workings and gain insights into the underlying processes. Furthermore, the models can be experimentally manipulated to assess the factors crucial to their success.
Funding: SSHRC Insight Development Grant (2025–28)
Neural encoding of secondary acoustic cues in speech perception
In this research, we investigate how English and Korean speakers encode laryngeal contrasts cued by VOT (the primary cue in English) and F0 (the primary cue in Korean).
Funding: University of Toronto Scarborough International Research Fund (2024–25)
Collaborators
Samuel Akinbo (University of Toronto)
Adam Albright (MIT)
Hye-Young Bang (McGill University)
Ellen Broselow (Stony Brook University)
Meghan Clayards (McGill University)
Ewan Dunbar (University of Toronto)
Sungwoo Han (Inha University)
Manami Hirayama (Seikei University)
Chiyuki Ito (Tokyo University of Foreign Studies)
Arvind Iyengar (University of New England)
Hioronori Katsuda (University of Kansas)
Michael Kenstowicz (MIT)
Say-Young Kim (Hanyang University)
Alexei Kochetov (University of Toronto)
Eunjong Kong (Korea Aerospace University)
Dave Kush (University of Toronto)
Seunghun Lee (International Christian University)
Naomi Nagy (University of Toronto)
Philip J. Monahan (University of Toronto)
Andrea Hòa Phạm (University of Florida)
Na-Young Ryu (Penn State University)
Jessamyn Schertz (University of Toronto Mississauga)
Morgan Sonderegger (McGill University)
Benjamin Storme (CNRS)
Lisa Sullivan (University of Manitoba)
Connie Ting (McGill University)
Tae-Jin Yoon (Sungshin Women's University)
Suyeon Yun (Chungnam University)
Adam Albright (MIT)
Hye-Young Bang (McGill University)
Ellen Broselow (Stony Brook University)
Meghan Clayards (McGill University)
Ewan Dunbar (University of Toronto)
Sungwoo Han (Inha University)
Manami Hirayama (Seikei University)
Chiyuki Ito (Tokyo University of Foreign Studies)
Arvind Iyengar (University of New England)
Hioronori Katsuda (University of Kansas)
Michael Kenstowicz (MIT)
Say-Young Kim (Hanyang University)
Alexei Kochetov (University of Toronto)
Eunjong Kong (Korea Aerospace University)
Dave Kush (University of Toronto)
Seunghun Lee (International Christian University)
Naomi Nagy (University of Toronto)
Philip J. Monahan (University of Toronto)
Andrea Hòa Phạm (University of Florida)
Na-Young Ryu (Penn State University)
Jessamyn Schertz (University of Toronto Mississauga)
Morgan Sonderegger (McGill University)
Benjamin Storme (CNRS)
Lisa Sullivan (University of Manitoba)
Connie Ting (McGill University)
Tae-Jin Yoon (Sungshin Women's University)
Suyeon Yun (Chungnam University)
