Projects  /  2022 – present

Projects

Three labs and one thread: how speech is coordinated with the rest of the body, and how the brain follows it. Newest first.

Whole-body speech planning

Does gesturing change how far ahead a speaker plans? A deadline task makes planning visible in response time while motion capture, force plates and a respiration belt record what the body does. The recording system is built, the task is piloted, and the planning model and power analysis are done.

The full project The recording space

Rhythmic stimulation and hearing near threshold

The lab’s first brain stimulation study. It tests whether stimulating motor regions, such as the supplementary motor area, in rhythm with a sequence of sounds helps listeners detect tones near the limit of hearing.

I designed the stimulation protocol with participant safety in mind, wrote the ethics protocol, and built the timing between the stimulator and the sounds.

Fast speech in noise

People speed up online lectures and listen to them in noisy places. I designed an EEG experiment to ask how the brain tracks time-compressed speech in background noise, and what listeners remember afterward. Speeding speech up and adding noise turned out to affect memory and neural tracking independently.

  • ManuscriptDych, W. P., Ikeda, M., Lee, M. Y., & Lim, S.-J. Bottom-up noise enhancement versus top-down time compression deficits: Independent effects on neural speech tracking and memory. Under review.
  • PresentedIkeda, M., Dych, W. P., & Lim, S.-J. (2025). Acoustic degradation of speech through time compression and background noise independently affects content memory and neural tracking of speech. Society for the Neurobiology of Language, Washington, D.C.
  • PresentedIkeda, M., Dych, W. P., & Lim, S.-J. (2025). Impact of time-compression and background noise on content memory recall and neural tracking of speech. 8th International Conference on Auditory Cortex, Maastricht.

Timing precision in online experiments

An online experiment gave results that did not line up, and the cause looked like the participants’ own equipment. I designed three follow-up experiments to isolate how differences in devices, such as headphones and internet connection, limit the timing precision of auditory studies run in a web browser.

  • ManuscriptDych, W. P., Zi, L., & Lim, S.-J. Uncontrolled hardware variation introduces limits on timing precision in online auditory experiments. Under review.

Automating gesture annotation

I started in Kathryn Franich’s lab annotating speech and gesture recordings by hand, and saw that hand annotation set the pace of the research. I proposed a tool that could find and label gestures automatically with as little equipment as possible, and built it with guidance from Kathryn Franich and Karee Garvin, using rules grounded in how people can and cannot move.

The first version used markerless tracking of ordinary video. We compared the gesture apexes it computed with apexes annotated by hand, in video interviews with speakers of Medʉmba, and the two agreed closely.

  • PaperDych, W., Garvin, K., & Franich, K. (2023). Comparing manual vs. semi-automated methods for the coding of co-speech gestures. Proceedings of the 20th International Congress of Phonetic Sciences, Prague. Repository record
  • PresentedDych, W., Garvin, K., & Franich, K. (2024). The MultiGest tool: Creating multimodal phonetic corpora for co-speech gesture research. CorpusPhon, Seoul.
  • PresentedDych, W., Garvin, K., & Franich, K. (2024). A toolkit for automating co-speech gesture data annotation and analysis. Second International Multimodal Communication Symposium, Frankfurt.

The software

Gesture magnitude in Medʉmba

My first research project, funded by the University of Delaware Summer Scholars program: measuring how large co-speech gestures are in Medʉmba, a Grassfields Bantu language.

  • PresentedDych, W. (2022). Gesture magnitude mapping in Medʉmba: A Grassfields Bantu language. University of Delaware Summer Scholars Symposium, Newark, DE.