My ongoing research currently centers around three projects: cumulative markedness effects, phonology in non-native speech perception, and the phonology of Paraguayan Guarani.
cumulative markedness effects
My research on phonotactics centers on the effects that multiple instances of infrequent sound patterns have on languages and its speakers, and if and how such effects are learned from the distributional properties of sounds in their language(s). For example, the English th sound occurs in frequent words such as think and teeth, but in English there are rarely any words with two or more th sounds and English speakers likely find words like '’theeth’’ to be weird potential words of English. Patterns such as this one, along with speakers’ intuitions about them, are interesting because infrequent sounds can have a stronger or weaker effect when they occur with another instance of a particular sound structure compared to when they occur independently, posing interesting questions about the variation and learnability of these patterns. And, these patterns challenge mainstream models of linguistic knowledge in meaningful ways.
Dissertation
Papers
-
One-shot vs. competitions phonotactics in modeling constraint cumulativity. Annual Meeting in Phonology (AMP) Proceedings paper. [abstract] [slides] [proceedings]
This work details the computational properties of two different maximum entropy model architectures, which I call ‘‘one-shot’’ and ‘‘multiple competitions’’ models of phonotactics. This work describes the findings for counting cumulativity – multiple vilolations of the same phonotactic pattern.
-
Young infants are sensitive to accumulating phonotactic violations. In prep, with Megha Sundara.
In this paper we find that 5-month-old and 8-month-old infants are able to tell apart words with zero, one, and two violations of phonotactic constraints.
Paraguayan Guarani
My research on Paraguayan Guarani centers on the language’s nasalization system. Paraguayan Guarani Tupian language spoken by 5 to 6 million people in Paraguay and neighboring areas of Argentina and Brazil. The nasalization system of Guarani is quite complex because the leftward and rightward spread of nasality are strikingly different, therefore making it particularly interesting from the perspective of phonological theory and typology.
Papers
-
Suffix independence in Paraguayan Guarani: stress, nasality and nasalization. Under review.
-
Realizations of the tritonal pitch accent in Paraguayan Guarani, with Sun-Ah Jun, Jian-Leat Siah, and Hunter Johnson. Poster and proceedings paper at International Congress of Phonetic Sciences (ICPhS). [poster] [proceedings]
-
Nasal harmony in Paraguayan Guarani: Positional effects and the representation of nasality. MA Thesis, UCLA.
Presentations
-
Paraguayan Guarani progressive nasalization as phonologically conditioned allomorphy. Presented at the Manchester Phonology Meeting (mfm) in 2025 [handout] [slides]
-
Suffix independence in Paraguayan Guarani: stress, nasality, and nasalization. Colloquium talk at California State University, Fresno in 2025 [slides] [slides (no transitions)]
-
Suffix independence in Paraguayan Guarani nasal harmony. Presented at SSILA in 2024. [abstract] [handout]
-
Wh-question formation strategies in Guarani, with Hunter Johnson. Presented at SSILA in 2024. [handout]
non-native speech perception
More recently, I study how listeners perceive, or rather, misperceive, sound patterns that are unattested in their language. In addition to speakers knowing which sound sequences occur and don’t occur in their language, they also know how their language repairs some of these underlying sequences. For example, English speakers know that words should not begin with [pt] or [ps] in English, but they also know that words such as pterodactly and psychology are pronounced by omitting the first sound as opposed to saying something like peterodactyl or perodactyl. We know that speakers misperceive sound sequences unattested in their native language, but it remains unclear if speakers’ perception of these non-native patterns is also biased by how their language repairs them.
- I’m currently running online speech perception experiments with monolingual speakers of Korean to see if their misperceptions of illegal word-medial consonant clusters are consistent with the native repair processes of these clusters in Korean. I’m also comparing the predictions of various Bayesian models of speech perception that vary in their ability to encode information about the repair processes of the language.