Research has shown that children’s articulatory knowledge and perceptual knowledge contribute to phonological awareness, an important language skill that robustly predicts early literacy skills and reading development (Rvachew and Grawburg, 2006; Johnson, 2022). However, characterizing early articulatory knowledge is difficult because children’s early word productions often contain covert contrasts that are not readily notated by native listeners when using conventional phonetic transcription. Using longitudinal data from 143 English-speaking children, we measured spectral characteristics of /t/-/k/ bursts and /s/-/ʃ/ frication. We computed the robustness of contrast (ROC) between these two phoneme pairs as the percentage of phonemes correct by comparing the target sound and the model’s prediction using spectral measures as input (Holliday et al., 2015). We examined how phonetic differentiation as indexed by these ROC measures at 3 years of age predicted later performance in a series of standardized speech and language assessments at 5 years of age. Path analysis showed that /s/-/ʃ/ differentiation at age 3 predicted speech production accuracy and speech perception scores in two standardized tests at age 5. This work underscores the need for researchers to develop automated acoustic analyses that clinicians can feasibly use to supplement existing assessment measures that solely depend on conventional transcription methods.
Wong et al. (Wed,) studied this question.