Characteristics of Speech Motor Functions in Two Low-Functioning Individuals with Autism
Objectives: To obtain a more comprehensive profile of LFA speech production, using a combination of perceptual rating, transcription, acoustic analysis and weighted error coding
Methods: Two verbal LFA (autism diagnosis confirmed by ARI-R and ADOS, in their mid-20s) with severely unintelligible speech were administered the Goldman-Fristoe Test of Articulation-2 (GFTA-2) and a battery of motor speech profile(MSP) tests via the MSP program from KayPentax CSL. Assessment recordings and acoustic analysis were carried out using the CSL4150 model at 44.1 kHz 16-bit with a Shure SM48 microphone. The GFTA-2 was given multiple times and target words (both standard and modified) were perceptually rated and transcribed. Articulation errors were coded and an overall accuracy/intelligibility score generated using the Upper LIPP system with the Weighted Speech Sound Accuracy program (courtesy of Dr. J.L.Preston).
Results: Subjects’ raw scores on the GFTA-2 placed their performance at 2-0 to 2-5 age-equivalent level, indicating severe sound acquisition delay. Common major speech errors included cluster simplification, final consonant deletion and palatal fronting. MSP analysis showed significant deviation from normal production range in tasks that examined vowel elongation (e.g. 0.34±0.03 seconds in LFA vs. 28.7±7.20 for the norm), repetitive motion rate (e.g. 1.30±1.39 syllable/second in LFA versus 5.97±0.46 for the norm); reduced loudness and pitch variations (e.g. 8.1% frequency and 28.1% amplitude variability in LFA vs. 20.3±4.8% frequency and 40.5±5.3% amplitude variability for the norm). Perceptual intelligibility rating (PR) on target words, which the subjects demonstrated adequate semantic knowledge, revealed less than 10% met screening criteria for intelligibility (≥75%). Additionally, multiple productions of the same word on different trials showed great baseline variation (e.g. up to 38.8% in standard deviation (SD)) with the greater SD values (15-38.8%) coming from words with mean baseline intelligibility (MBI) between PR of 15 to 60% and smaller SD (≤15%) from words with MBI either above 60% or below 15%.
Conclusions: 1) The significant deviation in MSP values captures and confirms clinical observations of speech features of our LFA subjects (i.e slow, soft, and monotone) in addition to articulation errors characterized by GFTA-2. This suggests a strong motor component to the inaccuracy in their speech production. 2) Tremendous variability (measured by PR) in subjects’ repeated production of the same target word, especially in the low-mid intelligibility range, implies instability in the underlying motor function that brings about production of words not yet mastered. These data should serve as one foundation for further characterizations of speech production in LFA, and as a basis for guiding therapy and improving performance both on general principles, and on an individual basis.