Search

Purpose: To develop a system for numerically quantifying a speaker's phonetic accuracy through transcription-based measures. With a focus on normal and disordered speech in children, the authors describe a system for differentially weighting speech sound errors on the basis of various levels of phonetic accuracy using a Weighted Speech Sound Accuracy (WSSA) score. The authors then evaluate the reliability and validity of this measure. Method: Phonetic transcriptions were analyzed from several samples of child speech, including preschoolers and young adolescents with and without speech sound disorders and typically developing toddlers. The new measure of phonetic accuracy was validated against existing measures, was used to discriminate typical and disordered speech production, and was evaluated to examine sensitivity to changes in phonetic accuracy over time. Reliability between transcribers and consistency of scores among different word sets and testing points are compared. Results: Initial psychometric data indicate that WSSA scores correlate with other measures of phonetic accuracy as well as listeners' judgments of the severity of a child's speech disorder. The measure separates children with and without speech sound disorders and captures growth in phonetic accuracy in toddlers' speech over time. The measure correlates highly across transcribers, word lists, and testing points. Conclusion: Results provide preliminary support for the WSSA as a valid and reliable measure of phonetic accuracy in children's speech. [ABSTRACT FROM AUTHOR]

Purpose: Some children with speech sound disorders (SSD) have difficulty with literacy-related skills, particularly phonological awareness (PA). This study investigates the PAskills of preschoolers with SSD by using a regression model to evaluate the degree to which PA can be concurrently predicted by types of speech sound errors. Method: Preschoolers with SSD (n = 43) participated in PA and speech sound production assessment. Errors from a 125-item picture naming task were coded in 2 ways: (a) considering all consonant errors equally (percentage of consonants correct [PCC]) and (b) using a 3-category system that captures component features of sound errors (typical sound changes, atypical sound changes, and distortions). PA tasks included rhyme matching, onset matching, onset segmentation and matching, and blending. Results: Variance in a PA composite score could be predicted partly by vocabulary and age (33%). Atypical sound changes accounted for an additional 6% of variance in PA, but distortions and typical errors did not account for significant variance. When the same consonant errors were analyzed using PCC, speech errors did not predict significant variance in PA. Conclusions: Poorer PA is associated with lower receptive vocabularies and more atypical sound errors. Results are interpreted in the context of the accuracy of phonological representations. [ABSTRACT FROM AUTHOR]