Developed a computerized version of the Silly Sentence task, and to check that the scores it produces correspond to those obtained from the existing pencil-and-paper task from the Speed and Capacity of Language Processing (SCOLP) test in 3 experiments. In Exp 1, 64 Ss were presented with sentences and asked to press the right-hand button of the box in response to sensible sentences and the left-hand button in response to silly sentences. In Exp 2, 33 Ss (aged 7-11 yrs) were presented with the same stimuli as in Exp 1, except that after the computerized task each child also completed sets A and B of Raven's Coloured matrices. In Exp 3, 21 Ss (aged 18-33 yrs) were presented with the same stimuli as Exp 1 except that the sentences were translated into Kiswahili, the task was used in Tanzanian schools, despite the absence of an electricity supply and a very different cultural background. Results show that the decision latencies had a test-retest reliability of 0.69 over 5 months, and were independent of age and baseline decision speed. The task appears appropriate for longitudinal studies, including those in developing countries. Given its simplicity and the correlations with the original SCOLP version of the task, it may also be useful in studies on literate adults. (PsycINFO Database Record (c) 2000 APA, all rights reserved)