Deriving individualised /r/ targets from the acoustics of children's non-rhotic vowels.

The remediated group was only two children, so the target formula and classification results cannot be treated as general evidence. The study tested acoustic measurements, not whether children improved, stayed improved, or found therapy easier. The revised formula was fitted to typically developing children but is intended for children with speech sound disorders. It only covers stressed syllabic /ɝ/ (as in 'fur'), not all /r/ positions. The normative sample may not represent the children studied, possibly because of dialect, physical size, or task differences. Two children were excluded because their formant values were outliers, and both were bidialectal, so the result may not apply to all dialects. Gender differences were seen, but the sample was too small to say they are stable.

Do not read the 95.6% correct classification in two children as proof that biofeedback therapy works or that the target formula is ready for all children. It came from two remediated children, and the norm-derived target formula overpredicted actual /r/ F3 values.