Computer simulation as a diagnostic tool in language evaluation

Abstract

The language of congenitally deaf children is generally less advanced and contains many more errors than that of hearing children. Because of the complex structure of language and the many types of errors that can occur, it is not always clear which errors or syntactic forms should receive special attention in teaching for the most rapid improvements in language. An attempt was made to obtain such information by means of computer simulation. Samples of the written language of several deaf children were entered into the computer together with syntactic corrections, suitably coded, whenever an error occurred. The computer then reproduced each child's written sample several times, systematically eliminating various errors and combinations of errors. The modified versions of each written sample, which covered the range from original to a completely corrected sample, were then rank ordered in terms of overall quality. From these rankings it was possible to determine on a child-by-child basis which errors, if eliminated, would produce the greatest improvements in the child's language. In addition, statistical analyses can be performed to determine the most common error types both between and within children. [Supported, in part, by New York State Education Department, CREED Project ED 73-001DC.]