This work seeks to address the question of what counts as a syntactic “clause” in the mapping of intonational phrases (IPs) in Seoul Korean (henceforth Korean) by comparing the predictions of different theories of IP mapping against experimental data from a previous study on Korean (Kang & Speer 2005), that looks at the role of prosody in resolving attachment ambiguities. Through an Optimal-Theoretic implementation, I argue that there is a divergence in the predictions that the Match family of constraints (Selkirk 2011) and Align constraints (Selkirk 1986, 1996) make, despite the former being considered to supplant the latter, and that there is a need to (re)introduce ALIGN constraints back into the theoretical tools used to model syntax-prosody correspondences.
This work applies the mechanisms of Fenger's (2020) theory where marked phase heads stop head movement and induce (cyclic) spell-out to a set of Uyghur data from Major et al. (2023), featuring a range of word-formation strategies. I argue that in Uyghur, like other languages such as Turkish and Japanese according to Fenger (2020), locations of prosodic word boundaries conditioning morphophonological changes can be predicted by the presence of featurally marked phase heads in the narrow syntax. Specifically, it is argued that synthetic forms involving the past tense marker in Uyghur arise when there is no feature marking at the phase head Asp(ect), and periphrastic forms arise when there is marking at the Asp phase head, such as for the perfective participle. This work expands upon cross-linguistic work in word-formation strategies and provides evidence in favor of a strong correspondence between locality domains in syntax and phonology, specifically with respect to the prosodic word.
In this work, I pose a new “intermediate" approach to deriving FAs, where more structure is generated for FAs than the nonsentential approach of Barton & Progovac (2005) and Progovac (2013), and less structure than the sentential approach of Merchant (2004): a vP. Under this theory, the answer vP is generated with all its arguments, including the questioned constituent, and through PF-deletion, only the questioned constituent ends up being pronounced. I demonstrate that generating more than what the nonsentential approach proposes (i.e., a minimal DP/VP/PP) provides the minimal necessary structural configurations to license phenomena like (non-default) accusative case and anaphor fragment answers. By generating less structure than the sentential approach, the intermediate approach accounts for the felicitiy of fragment answers that cannot undergo fronting, such as NPIs as well as bare QPs in English and VPs in Icelandic, which are otherwise predicted to be illicit in Merchant's (2004) "move-then-sluice" approach. I then use the mechanics of Champollion's (2014) theory of event semantics to show how fragment answers at the vP level can receive a full compositional interpretation.
Converbs are a widespread linguistic phenomenon, yet there are relatively few structural accounts of them in the generative literature. To provide a preliminary syntactic account of converb constructions, I elicited data for the imperfective converb in Khalkha Mongolian (suffixed as -ж/ž) and the imperfect converb in Manchu (-me), and analyzed these constructions' argument structure and relations to the main verb clause. Looking at issues of subjecthood, valency, extraction, adjunction, and scope of negation, I argue that converbs project their own vPs which adjoin to a higher-positioned functional projection that is directly below the matrix TP. Dependent Case Theory is used to account for some of the case facts observed in these constructions.
Evidentiality has been researched in detail in a handful Formosan languages, with the exception of Squliq Atayal. To investigate the presence of evidential markers and/or strategies in Atayal, I developed a novel translation task that prompts speakers to translate sentences based on the semantic parameters of evidentiality established in Aikhenvald (2004). I find evidence to suggest that the aspectual marker wal has evidential extensions that indicate both indirect and direct sources of evidence. Interestingly, I find that wal is felicitous with each semantic parameter except for non-visual sensory. I provide an analysis of wal that shows a conceptual overlap between indirect and direct evidentiality, and conclude that its distribution contradicts the conventional (in)direct dichotomy.
Since 2021, I have been collaborating with speakers of Sibe, a dialect of Manchu to develop online documentation materials for the speaker community. These documentation materials include a Sibe-Mandarin-English dictionary with audio recordings, an introductory phrasebook, a video archive of authentic language material, and more. All of these resources can be found on Mini Buleku (lit. 'My Dictionary'), an online platform I coded and developed to make our resources available to the community of speakers in China and abroad.
From July 2022 to August 2023, I served as Documentation Manager of the HUC-JIR Jewish Language Project, where I led the organization's work in documenting endangered Jewish language varieties — primarily Judeo-Iranian and Jewish Neo-Aramaic. For both languages, I coordinated the filming, transcription/translation, and dissemination of over 10 authentic language sample videos. I also collaborated with the community of Jewish Neo-Aramaic speakers to lay the groundwork for an online recorded dictionary, as well as a speaker of Judeo-Arabic to produce an oral history in the language. In November 2022, I applied for a grant with the Wikimedia Foundation and was awarded approximately $50,000 to create and upload content in endangered Jewish languages to the various Wikimedia Projects, including Wikimedia Commons and Wikipedia. In Fall 2025, I taught a 6-week virtual course on language documentation, where we as a class collectively documented Judeo-Kashani.
In June 2024, I traveled with members of Harvard University's fieldwork (WOLF) lab to the Alabama-Coushatta tribe, with whom our lab is collaborating to document the Alabama language and develop teaching materials to be used in the community. At the tribe, we held individual elicitation sessions with native speakers of Alabama, and my elicitations particularly focused on (1) investigating the language's morphophonology of agent agreement and (2) collecting textual and audio examples of useful everyday phrases that can be turned into pedagogical materials.
As a member of Harvard's fieldwork (WOLF) lab, I developed a program that creates LaTeX-ready numbered examples from unformatted text. Some features of the program are that it can: support different LaTeX packages (linguex and gb4e), generate glosses with(out) the original language's orthography, convert linguistic annotations into small caps, create sub-examples, clean up whitespaces, and more. This program is currently accessible through a notebook on Google Colab.