Measuring speech naturalness of children who do and do not stutter
Edge, Robin Lynne
MetadataShow full item record
The two studies presented in this dissertation investigated the speech naturalness of children who do and do not stutter and the effect of rater training and stuttering severity on speech naturalness. Study One evaluated three groups of judges’ speech naturalness ratings of children who stutter. Judges’ rater agreement was also reported. The effect of training using a modified version of the naturalness portion of the Stuttering Measurement System (SMS) on rater agreement was also investigated. Results of Study One showed statistically significant improvement in the training group’s rater agreement when compared to the two control groups. Study Two was a follow-up replication study that used a larger sample size and added normally speaking children. The effect of stuttering severity (normally speaking children versus children who stutter at mild, moderate, and severe levels) and SMS speech naturalness training on speech naturalness ratings and agreement ratings was investigated. Unlike Study One, the SMS training was not found to change listeners’ agreement ratings to a statistically significant level. Stuttering severity was found to affect both speech naturalness ratings and rater agreement with normal speakers rated the most natural (lowest naturalness ratings) and children who stutter at a severe level rated the least natural (highest naturalness ratings). Agreement was higher for normal speakers and children who stutter at a severe level than children who stuttered at mild and moderate levels. These two studies provide initial data from inexperienced listeners using Martin et al.’s (1984) 9-point speech naturalness scale to rate the speech naturalness of children who do and do not stutter and their agreement levels when rating speech naturalness. The effectiveness of the SMS training program is inconclusive at this time as data supporting its effectiveness were reported in Study One, but it was not shown to be effective in Study Two. Possible explanations for this as well as future research addressing identified issues are discussed. This follow up research is needed before this measure can be recommended for use with children.