Token loss in speech imitation training.

Abstract

The use of tokens as positive reinforcers for correct responses and no consequences for incorrect responses was compared with the use of tokens contingent upon correct responses and loss of tokens contingent on incorrect responses. Two magnitudes of token gain to token loss were also compared: (1) three tokens were given for a correct response and one token removed for an incorrect response; (2) one token was given for a correct response and one was removed for an incorrect response. The subjects were a six-year-old child with delayed language development and a seven-year-old with an articulation problem. Training items consisted of the verbal imitation of words. Results indicated that the no-token-loss condition was more effective in decreasing incorrect responses than either magnitude of token loss. Results of the two token-loss procedures indicated, however, that their effectiveness is partly determined by the ratio of tokens gained to tokens lost. It is concluded that token loss could be effective but that related variables should be carefully examined.