Abstract
Background: Variability in speech sound errors has been regarded as a primary characteristic of apraxia of speech (AOS). Early research deemed errors extremely unpredictable, resulting in a number of different error types on repeated productions of the same stimuli. However, recent research has suggested that errors may not be variable, but there are limited data regarding variability over time (i.e., beyond a single sampling occasion). Furthermore, the influence of conditions of stimulus presentation (i.e., blocked vs random) on sound errors remains unclear.
Aims: The purpose of this investigation was to examine variability of sound errors in 11 individuals with AOS and aphasia. Of particular interest were the effects of repeated sampling and method of speech elicitation on the variability of error types as evaluated with narrow phonetic transcription.
Methods & Procedures: A total of 28 monosyllabic words served as experimental stimuli. There were four exemplars for each of the seven initial target phonemes (i.e., /h, f, m, d, s, r, n/). Stimuli were elicited on three sampling occasions over a 7-day period with each sampling occasion separated by 2 days. At each sampling time productions were elicited under two conditions: blocked presentation (blocked by sound) and randomised presentation. Speech productions were analysed perceptually utilising narrow phonetic transcription.
Outcomes & Results: Findings revealed a similar overall mean percentage of errors for the group in both conditions of stimulus presentation across the three sampling times. The target phoneme with the least number of errors was /h/. The target phoneme with the greatest number of errors was /s/. The predominant error type across target phonemes was distortions. However, the predominant error type varied across target phonemes and appeared to be influenced by number of errors.
Conclusions: Repeated sampling or method of speech elicitation did not influence errors, with a similar overall mean percentage of errors for the group in both conditions of stimulus presentation across the three sampling times. Distortions were found to be the predominant error type for the majority of target sounds. A comparison of the number of error types produced by the group in each condition across the three sampling times found no obvious pattern of responding by the group in either condition for individual phonemes. That is, condition of elicitation did not appear to influence the variability of error type for any given sound.
Acknowledgments
This research was supported in part by Rehabilitation Research and Development, Department of Veterans Affairs.