An experimental analysis of misarticulating children's generalization

Five children who produced /theta/ for /s/ substitutions as a misarticulation were trained to produce /s/ correctly in three syllables. Untrained exemplars of syllables and words were tested throughout baseline and training. The 60 probe items contained both spontaneous and imitated words and syllables combined with high, low, front, back, vowels, and consonants. A functional analysis reversal design was used, and the gernalization patterns were analyzed. The effect of context was found to be less influential than expected while other factors such as stimulability, amount of training, and subject characteristics appeared as important variables in generalization.