A comparison of tongue shape variability in children with and without speech sound disorder using ultrasound

Background: Increased tongue shape variability has been observed in children with Speech Sound Disorders (SSD), which may be indicative of subtle speech motor control issues. However, a lack of data from typically developing children (TD) hinders our understanding of how the two groups compare and what is the expected range of variability in a TD child.
Method: In this study, data from TD children (n= 10, 7-14 years) was collected using ultrasound tongue imaging (UTI) at a public engagement event. This was compared to 10 age matched children with SSD from open access corpora. Both groups of children repeated the consonants /s/, /t/, /k/, /ɹ/ in /aCa/ ten times. Using Articulate Assistant Advanced software, we automatically traced contours of the tongue’s surface at the maximum point of articulation. The mean Nearest Neighbour Distance (NND) was calculated for the ten repetitions of each consonant of each child. We predicted that children with SSD would have larger mean NNDs, indicating increased variability across repetitions. A linear mixed effects model was fitted with mean NND as the dependent variable and diagnosis as the independent variable and random intercepts per participant.
Results: The factor, diagnosis, was found to significantly affect mean NND in the linear mixed effects model, with the SSD group showing significantly larger mean NNDs for the consonant repetitions than the TD group.
Conclusions: This result supports the hypothesis that children with SSD have more variable tongue shapes compared to TD children when producing multiple consecutive repetitions of consonants. However more research is needed to establish variability norms for TD children to understand the range of expected variability and how this compares to children with SSD.
Note: Further TD data will be collected on March 18th which may impact results presented at the conference.