Variability and sequential order of consonant normalization in children with speech delay

A source of discrepancy among attempts to establish a rank order in the development of English consonants has been the inability to cope concurrently with principled generalization and individual variability. This problem may be surmounted by survival analytic techniques. From the conversational speech of 24 children with speech delay (SD) recorded over 2 years, a Kaplan-Meier (KM) survival analysis provided a rank order of acquisition for all probabilities above .75. The rank ordering by chronological age uncovered an alignment by place and manner of articulation comparable to, but not identical with, the predictions of Jakobson (1941, 1968). When the derived normalization probabilities for the speech delayed children were lagged according to the strong delay hypothesis (Shriberg, Gruber, & Kwiatkowski, 1994), they were shown to converge with previous normative studies and the age of speech-sound determinations (Prather, Hedrick, & Kern, 1975; Sander, 1972). The convergence of results is interpreted both as an argument for generalizability and as support for the strong delay hypothesis.