Youngah was invited by Peking University to discuss her recent research on how phonological knowledge emerges from phonetics. She covered four studies conducted with Ivy and Frank, demonstrating through computational models how abstract phonological patterns—such as phonotactic constraints and segmental categories—can be learned from continuous acoustic and articulatory data without relying on pre-defined categories.
Her models, including distributional learners and neural networks, showed that statistical regularities guided by formal biases can produce phonological generalizations, effectively bridging phonetics and phonology. The work suggests that similar biased learning mechanisms support both first-language acquisition and adult language learning, challenging the idea of innate primitives.
