Undergraduate research assistants Taijing Chen, Jenna Krakauer, and Emma Baumgardt, are presenting their research projects at the UW-Madison Undergraduate Research Symposium on April 8 & 9th, 2021.
Modelling sensorimotor adaptation in speech through alterations to forward and inverse models (Taijing Chen, April 9th @ 9am)
Previous work has shown that sensorimotor adaptation of a particular vowel can be transferred to other, untrained, vowels. Moreover, the magnitude of this transfer decreases as the acoustic distance between the training and the transfer vowels increases. In this study, we developed a neural-network based architecture to computationally simulate behavioral results of speech motor learning and transfer and assess whether updates to internal control models could account for the observed generalization patterns. Our results replicate behavior in previous experiments: the model produced changes in speech production that counteracted the imposed perturbation, and the model showed gradient transfer of learning dependent on acoustic distance. These results suggest that updating paired forward and inverse models provides a plausible account for speech motor learning.
The Role of Attention in Compensation for Altered Auditory Feedback (Jenna Krakauer, April 8th @ 12pm)
Speech production relies on auditory feedback (hearing one’s speech) to correct for potential errors as they occur in real time. These feedback-based corrections (compensation) can be studied using altered auditory feedback, in which auditory feedback is perturbed in either pitch or vowel formants (the resonant frequencies of the vocal tract). In response, speakers oppose these perturbations. However, the magnitude of compensation varies both within and between participants, and the source of this variability is currently unknown. Previous studies have shown that divided attention modulates feedback responses for vocal pitch control, but not reaching. Here, we examine the role of attention in speech motor control through formant alterations. We hypothesize that divided attention will reduce compensation, suggesting that attention contributes to the observed variability in compensation.
The impact of sensorimotor adaptation on speech intelligibility (Emma Baumgardt, April 9th @ 11am)
In a previous study, speakers were presented with altered auditory feedback that decreased their vowel contrast, and learned to oppose the perturbation by increasing their vowel space. Here, we determine whether this sensorimotor adaptation also impacts speech intelligibility, by asking participants to identify words from recordings, masked by additional noise, made before and after speakers adapted their speech to increase their vowel space. Because adaptation resulted in more acoustically distinct vowels, we hypothesize that the words produced after adaptation will be more perceptually distinct than words produced before adaptation. Because individuals with motor speech disorders have less distinct vowels and lower speech intelligibility, studying whether sensorimotor adaptation increases speech intelligibility will allow for a greater understanding of how speech intelligibility can be intentionally increased.