To determine whether language behavior represents an early conditioned verbal response or whether it changes with age and experience was the purpose of this study which attempted to define unique isolates of language on the basis of actual language produced by young children. Tape recorded data were collected for 12 years from 211 children in Oakland, California. Data collected during the first three grades were used to define eight "language style groups" (research groups) and statistics recorded during grades 10-12 were used to assess and predict language facility and growth. To create the research groups, three test or rating variables (e.g., intelligence test and verbal performance scores) and 15 language variables (e.g., "average length of communication unit") were utilized. The basic hypothesis--children will not change with age their relative positions to each other in language behavior--was supported with respect to speech conventionality but not supported with respect to problems of mazes (groups of words not resulting in meaningful communication). It was supported with respect to fluency, dependent clauses, and elaboration index for students who began as poor users of oral English. These results have several implications for curriculum development, especially in the teaching of reading. (LH)