Variability of sound productions in apraxia of speech: Perceptual analysis

Variability in speech sound errors has long been considered a hallmark of apraxia of speech (AOS). Errors were considered to be extremely unpredictable with regard to both the location of the error (i.e., consistency of error location) and type of error (i.e., variability of error type across repeated productions of the same stimuli). However, subsequent research has suggested that errors may not be variable. The purpose of this investigation was to examine speech production in individuals with AOS and aphasia, specifically the consistency in the location of errors and the variability of error types across repeated sampling times.;Results revealed speakers produced a significantly greater number of errors on final phonemes in the blocked condition (p < 0.05). Speakers frequently exhibited a dominant error type across conditions and sampling times for target phonemes, but this pattern was more robust for final phonemes. Distortions tended to be the prevailing error type across phonemes. A greater number of errors occurred in the final position of words across word length regardless of conditions or sampling times. Within-subject statistical analyses revealed statistical differences among the group on particular phonemes examining condition of stimuli presentation by time and then by word length.;For this investigation, narrow phonetic transcription provided subtle articulatory detail that would otherwise be missed. Future research would benefit from combining narrow transcription with other methods of analysis (i.e., acoustic or physiologic) in order to gain additional information regarding motor control during speech tasks.;This study involved a single group, repeated measures design. Fourteen individuals with AOS and aphasia participated in the investigation. Stimuli were elicited on three sampling occasions over a 7-day period with each participant. Sampling times were separated by 2 days. Stimuli were comprised of mono-, bi-, and trisyllabic words. At each sampling occasion, stimuli were elicited under two conditions: blocked presentation (blocked by sound) and randomized presentation. Speech samples were analyzed perceptually utilizing narrow phonetic transcription. Analyses primarily focused on 14 target phonemes (i.e., 7 initial and 7 final).