Siyu Liang

I am a fifth-year PhD candidate in Computational Linguistics at the University of Washington advised by Drs. Gina-Anne Levow and Richard Wright. My research develops computational methods for multilingual and under-resourced speech grounded in linguistic theory and fairness.

I study how large speech language models represent and generalize across phonetic and structural variation, and design methods to evaluate and improve their performance across resource levels and dialectal variation. As a linguist, I am particularly interested in studying speech variation through both lab phonology and corpus-based approaches. I am also interested in how computational approaches can be applied to linguistic analysis and documentation, and I have worked on Iranian, Tibeto-Burman, Turkic, Nilotic, and areally Southeast Asian languages. I actively conduct fieldwork and collaborate with field linguists to build models and resources in real-world documentation settings.

×

When I’m not working, you can find me on red-eye flights ✈️, at concerts 🎶, in the mountains 🏞️, and occasionally up the wall 🧗 or under the sea 🤿.

You can reach me at liangsy [@] uw [dot] edu.