Abstract
The fricative /s/ is generally acquired later than other speech sounds by children. Spectral moments have been used to quantify comparisons of both childrens' and adults' fricatives, but it is difficult to link moment values to a particular production difference: e.g., the first moment may be relatively low if place is more posterior, the noise source is weaker, or voicing persists into the /s/. This study adapts parameters originally developed for analysis of adults' fricatives to children's fricatives. Twelve children (6 boys, 6 girls), 10-15 years old with typical speech, were recorded saying 55-60 words with initial /s/, in singletons or clusters. Multitaper spectra were computed at the beginning, middle and end of each /s/, and parameters were derived from them to capture the main mid-range resonance, relative amplitudes for mid-to-low and mid-to-high ranges, and their change during /s/. Results showed that the mid-to-low-range relative amplitudes increased mid-/s/; the mid-to-high-range relative amplitude decreased; the main peak's frequency decreased in labialized contexts, all of which occur in adult /s/. One parameter indicated low-frequency spectral variability. These parameters thus capture known acoustic effects of source and filter changes occurring during /s/ and are promising both for measuring developmental changes and assessing misarticulations.