The goal of the current study was to construct a reference database against
which misarticulations of /s/ con be compared. Acoustic data for 26 typica
lly speaking 9- to 15-year-olds were examined to resolve measurement issues
in acoustic analyses, including alternative sampling points within the /s/
frication; the informativeness of linear venus Bark transformations of eac
h of the 4 spectral moments of /s/ (Forrest, Weismer, Milenkovic, & Dougall
, 1988); and measurement effects associated with linguistic context, age, a
nd sex. Analysis of the reference data set indicates that acoustic characte
rization of /s/ is appropriately and optimally (a) obtained from the midpoi
nt of /s/, (b) represented in linear scale, (c) reflected in summary statis
tics for the 1st and 3rd spectral moments, (d) referenced to individual lin
guistic-phonetic contexts, (e) collapsed across the age range studied, and
(f) described individually by sex.