🎉 I’m very excited to share my first peer-reviewed journal publication in Journal of Phonetics!

“Unpredictable temporal feedback perturbation dissociates mismatch detection from correction,” co-authored with Professor Sam Tilsen, is now published in Journal of Phonetics.

Like singers using in-ear monitors to hear their own voice, we continuously hear and monitor our own speech in everyday conversation.

The best way to study this feedback mechanism is to mess with it. So, what happens when we unexpectedly hear our own speech change while we’re talking?

We developed a novel paradigm that briefly and subtly stretches and compresses auditory feedback at unpredictable moments while speakers are talking. Despite these small perturbations, speakers responded with a delayed slowing of their speech, rather than immediately compensating for the change. This suggests that mismatch detection and mismatch correction may involve distinct processes in speech motor control.

Because the timing of the perturbation was unpredictable, we modeled responses both at the level of individual phones and continuously over time. We trained custom Hidden Markov Models (HMMs) in Kaldi to segment the speech data into individual phones, then used Generalized Additive Mixed Models (GAMMs) to model how phone durations changed with perturbation timing, along with local relative rate (LRR) to capture continuous changes in speech rate.

This project has been a huge part of my PhD journey since I started it in 2023, and seeing it become my first publication is a particularly meaningful milestone.

I’m incredibly grateful to Sam for his mentorship and collaboration, to all the participants who endured saying “I saw papa yesterday” 280 times for science, to the Cornell Phonetics Lab, and to the editor and three anonymous reviewers for their insightful feedback.

Access the paper here