A child's baseline speaking skills are determined prior to training through analysis of 100 spontaneous utterances. Analysis of this language sample considers linguistic constructions, features within these constructions, number of words/utterance and number of morphemes/ utterance. The resulting information enables clinicians to program language learning tasks for a specific child in accordance with developmental language sequences. Language programs developed by clinicians follow conventional programming and reinforcement principles and are appropriate for individual or small group training. The training procedure follows these steps: (1) language sampling; (2) analyzing the sample; and (3) training.This paper presents a system of training and testing based upon analysis of language samples. The method incorporates linguistic as well as behavioral principles and applies to children with mean utterance lengths ranging from two to six words. The method can be used for both group and individual teaching, since it provides personalized instruction in either setting.Narrowly interpreted, the procedures described in this paper apply to those children whose language delay is serious enough to warrant intervention. Broadly interpreted, it provides an instance of individualized instruction which may be helpful to a clinician or teacher interested in developing remedial programs based on children's performance.Our rationale for the method to be described follows from recent research at the Scottish Rite Institute for Childhood Aphasia and elsewhere (Morehead & Ingram 1973, Fygetakis & Ingram 1973, Bloom 1970, Brown 1973). To provide a scale of normal language acquisition with which language samples from individual children can be compared, Ingram and Eisenson (1972) developed an outline of syntactic structures based on Morehead and Ingram's study. The outline presents five levels, each of which is derived from mean sentence length. Across the five levels, mean sentence length varies from two to six words. Ingram and Eisenson's developmental sequence is not a complete listing of the forms and constructions children use. Rather, it is based on the most frequently occurring structures in the data analyzed by Morehead and Ingram. We do not presume to have complete knowledge of children's linguistic development at this time. Nonetheless, sufficient data are available on which to base language training. Since we are still collecting data on children's linguistic performance over extended periods of time, our procedures are not presented as research completed and fully docuVo/ume 7, Number 4, April, 1974 13 at