Item Restricted to University of Alberta Users

Log In with CCID to View Item
Usage
  • 254 views
  • 70 downloads

Predicting the Speech Intelligibility Scores of Children with Dysarthria and Cerebral Palsy from Phonologic and Phonetic Measures of Speech Accuracy

  • Author(s) / Creator(s)
  • Measures of speech sound accuracy (e.g., percent consonants correct or PCC) and direct measures of intelligibility (e.g., percent words identified correctly) are used to determine the severity of a child’s speech disorder. However, the relationship between these measures has not been reported for children with dysarthria. This study examined the relationships between several segmental (PCC, percent vowels correct or PVC, percent phonemes correct or PPC, PCC-R, PVC-R, PPC-R) and whole word (percent whole word accuracy; proportion whole word proximity) measures of speech sound accuracy obtained from phonetic transcription and direct measures of intelligibility based on word identification by unfamiliar listeners. Measures were based on audio recordings of TOCS+ imitated word and sentence samples obtained from 12 children with dysarthria and cerebral palsy (CP). Children ranged in age (4 - 12 years) and severity of CP (levels I to V on the GMFCS-ER). Phonetic transcription of the recordings incorporated a subset of narrow diacritics from the extIPA to capture sound error patterns of children with dysarthria and CP. PCC had the strongest correlation with intelligibility for the word samples (r = 0.764) and PCC and PCC-R had the strongest correlation (r = 0.931) with the sentence samples. While PCC rank ordered the children by severity of speech disorder in a similar way to the intelligibility scores, the magnitude of the difference between PCC and intelligibility scores varied substantially by severity of speech disorder, with the greatest difference for children with the lowest intelligibility scores.

  • Date created
    2011-06-27
  • Subjects / Keywords
  • Type of Item
    Report
  • DOI
    https://doi.org/10.7939/R3PZ51P9B
  • License
    Attribution-NonCommercial 3.0 International