Individual strategies in adaptation to altered auditory feedback

Abstract

Speakers monitor themselves while talking. When they hear a real-time altered version of their speech, they will change their articulation so that when they hear their altered speech, it matches their acoustic target [Houde and Jordan (1998, Science 20;279(5354):1213–1216)]. The experiment presented here used the novel addition of ultrasound imaging to reveal how speakers (n = 30) change their articulations in response to two different formant perturbations: raising of F1 in “head” and F2 in “hood.” Principal components analysis was used to identify speakers' individual strategies during adaptation. Some speakers use a single strategy for an entire adaptation block, while others change strategy. Speakers are also known to change production in a formant that was not altered [Katseff et al. (2010, JASA 127(3), 1955)]. The ultrasound analysis shows that at least for some speakers, change in two formants is linked to independent and uncorrelated articulatory components and possibly serves a perceptual purpose in compensation, rather than being an unintended result of the compensatory response. Preliminary results (n = 4) of adaptation to raising F3 in “heard” will also be presented. Modeling with the Maeda and Manzara synthesizers [Bakst and Johnson (2016, JASA 140(4), 3223)] correctly predicted speakers' articulatory strategies.