This project began with a theory about the prediction of errors in childrens' attempts to pronounce utterances modeled for them by adults. Subjects were children from 15 to 54 months old in all positions in the family. The sample from each child was an unprompted utterance judged to be an attempt to say something in English. Differences between the mother's utterance and the child's are errors; the results show more errors based on place, friction, or duration than on voicing and nasality. In a study of phones attempted by the subjects, the investigator found that the co-existence of success and error is strongly correlated with the extent to which the phone is attempted. A study of phonal position reveals that the initial position confers an advantage upon phones of all 3 types at all ages. Medial position is next most advantageous for consonants, but least advantageous for vowels. Success in pronunciation depends upon more than the frequency of utterance. (RS)