Identifying segmental and prosodic errors associated with the increasing word length effect in acquired apraxia of speech

Abstract

Purpose: Individuals with stroke-related apraxia of speech (AOS) plus aphasia tend to produce more speech errors with increasing word length. The Words of Increasing Length task (WIL) uses a 3-point scale to score word accuracy but penalises for error types that can arise either from language or motor impairment, reducing the test’s sensitivity and specificity. The purpose here was to identify error types explaining variance in the WIL score, and those associated with AOS and word length.

Method: Speech errors were perceptually identified on the WIL task for 51 Australian English-speaking adults with stroke-related aphasia, 25 with concomitant AOS. Multiple regression and linear mixed effects modelling were applied.

Result: Variance in WIL scores was best explained with four error types: consonant additions, incorrect number of syllables, false starts and consonant substitutions/distortions. False starts were significantly associated with AOS diagnosis. Incorrect number of syllables, consonant omissions, false starts, and lexical stress errors increased in frequency for longer words and, while the interaction with diagnosis did not reach significance, the effect appeared driven by the AOS group.

Conclusion: Findings provide further support for using polysyllabic word production to assess apraxic speech. The WIL task has limitations that may bias patients’ performance and clinicians’ perceptual evaluation. Data provide valuable information for designing a more sensitive diagnostic protocol for AOS.

Keywords:

• apraxia of speech
• aphasia
• phonetics
• lexical stress
• prosody
• stroke

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was funded by National Health and Medical Research Council Project [Grant #632763] to Ballard; Md Nor was supported by funding from The University of Sydney, and Masso by an Australian Research Council Discovery Early Career Research Award DE200101078.