The Ohio State University, Columbus, Ohio
Fifty words were recorded five times by five speakers. The five recordings represented five amounts and kinds of "context" (word in isolation, word said twice, word followed by a warning against making a common error, word in second-order, and word in third-order approximation to language). Listening panels heard the recordings and from their responses measures of word intelligibility and disparity, represented as information, were computed. The disparities among the responses decreased progressively from one condition to another in the order enumerated above; similarly, intelligibility increased, with the exception that no advantage resulted from saying a word twice.