Chicano and Amerindian vocabulary scale responses from the Stanford-Binet (LM) and Wechsler Intelligence Scale for Children were item-analyzed for 1,009 subjects. The response patterns differed both by ethnic group and test, as well as by age. The most common, and recurring, pattern found was "level-of-difficulty" gradient inconsistencies. The item-analysis method employed in this report was that of the covariance matrix where the sum of the matrix is the total-test's variance; vocabulary scale totals here, for both the SB-LM and the WISC. Tribes sampled were Navajo, Apache, and a group of subjects with mixed-tribal backgrounds designated "called Indian." Most of the protocols that could be analyzed on the vocabulary scales were pulled from the testing-course files at Arizona State University; other protocols were obtained from the reservations mentioned above. In the epilogue, based on the findings, it is suggested that bilingual programs similar to those being used in Texas' NIE funded program for the coming academic year be adopted for other large minority groups whose primary language is not either English or Spanish. (Author)