Reading while listening: A linear model of selective attention

The first of two experiments measured the performance of 27 subjects on pairs of concurrent verbal tasks. Subjects monitored auditory sentences for stop consonants, adjectives, or time reference words (primary task) while reading randomized, syntactically correct but semantically anomalous, or normal English passages (secondary task). The results were in conflict with several models of attention but were in quantitative agreement with a linear law of attention-sharing between verbal tasks: Performance on a secondary task when combined with a primary task is proportional to performance on the secondary task in isolation. The primary tasks were unimpaired with the exception of the detection of stop consonants. The second experiment indicated that the decrement with stop consonants was not due to differences in resource/data limitations; it may result from demands upon a proposed fine-timing processor.