Vowel formant dispersion reflects severity of apraxia of speech

ABSTRACT

Background: Apraxia of Speech (AOS) has been associated with deviations in consonantal voice onset time (VOT), but studies of vowel acoustics have yielded conflicting results. However, a speech motor planning disorder that is not bound by phonological categories is expected to affect vowel as well as consonant articulations.

Aims: We measured consonant VOTs and vowel formants produced by a large sample of stroke survivors, and assessed to what extent these variables and their dispersion are predictive of AOS presence and severity, based on a scale that uses clinical observations to rate gradient presence of AOS, aphasia, and dysarthria.

Methods & Procedures: Picture-description samples were collected from 53 stroke survivors, including unimpaired speakers (12) and speakers with primarily aphasia (19), aphasia with AOS (12), primarily AOS (2), aphasia with dysarthria (2), and aphasia with AOS and dysarthria (6). The first three formants were extracted from vowel tokens bearing main stress in open-class words as well as VOTs for voiced and voiceless stops. Vowel space was estimated as reflected in the formant centralisation ratio. Stepwise Linear Discriminant Analyses were used to predict group membership and ordinal regression was used to predict AOS severity, based on the absolute values of these variables as well as the standard deviations of formants and VOTs within speakers.

Outcomes and Results: Presence and severity of AOS were most consistently predicted by the dispersion of F1, F2, and voiced-stop VOT. These phonetic-acoustic measures do not correlate with aphasia severity.

Conclusions: These results confirm that the AOS affects articulation across-the-board and does not selectively spare vowel production.

KEYWORDS:

• Apraxia of speech
• aphasia
• vowel formants
• speech acoustics
• voice onset time

Acknowledgements

This work was supported by the National Institute on Deafness and Other Communication Disorders (P50 DC014664, PI JF; T32 DC014435, AB). The authors thank Grigori Yourganov, PhD., for consultation on statistical methods.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the National Institute on Deafness and Other Communication Disorders [P50 DC014664,T32 DC014435].