Measuring the contribution of printed context information to acoustical word recognition by normal subjects

The present study focuses on measuring the contribution of context in the form of visually presented printed sentences toward the acoustical identification of a word. The results obtained demonstrate the feasibility of the objective selection and calibration of sentence material that could serve to quantify an individual's efficiency in utilizing contextual information during speech perception. In the employed experimental procedure, a reference group of normal-hearing subjects are presented acoustically with test words either in isolation or together with context in the form of a written sentence. The test words are presented under various levels of superimposed noise and the median signal-to-noise ratio at which identification occurs in the reference group serves to characterize the intelligibility of a particular presentation of a word or word-sentence combination. This median value is shifted toward lower signal-to-noise ratios when contextual information in the form of a sentence is provided along with the test word. The resultant context-induced shift can be adopted as an objective and quantitative measure of the effect of contextual information provided by the written sentence on the intelligibility of the test word. The data obtained in this study demonstrate that such context-induced shifts are large enough to be easily measured with the adopted experimental approach.