Computer-Assisted Challenge Point Intervention for Residual Speech Errors

Abstract

Purpose This preliminary case series investigated the effects of biofeedback intervention for residual rhotic errors delivered within a modified challenge point framework. In the challenge point framework, practice difficulty is adaptively adjusted with the goal of enhancing generalization learning. This study more specifically evaluated the feasibility of a computer-mediated implementation of challenge point treatment for rhotic errors using a custom open-source software, the Challenge Point Program. Method Participants were five native English speakers, ages 7;3–15;5 (years;months), who had established but not generalized correct rhotic production in previous treatment; overall treatment duration was flexible. Treatment incorporated either electropalatographic or visual-acoustic biofeedback and was structured by challenge point principles implemented using the Challenge Point Program software. Results Participants were highly variable in the magnitude of generalization gains attained. However, the median overall effect size was 4.24, suggesting that participants' response in treatment tended to exceed the minimum value considered clinically significant. Conclusion These findings provide preliminary evidence that computer-mediated implementation of the challenge point framework can be effective in producing generalization in some participants. Supplemental Material https://doi.org/10.23641/asha.13244501

PERCEPT-R: An Open-Access American English Child/Clinical Speech Corpus Specialized for the Audio Classification of /ɹ/

“…Segment boundaries and orthographic labels were annotated using Praat TextGrids. In the case of more recent studies these segmentation labels were placed automatically using intensity detection or by our treatment software, CPP [18], in which case the automatically generated segmentations and transcripts were manually verified by trained research assistants. The utterances were extracted using Praat [19] or the Parselmouth API [20] for Praat, depending on the originating study.…”