Tests of intelligibility, or "articulation," must be assumed to be tests which are sensitive to the behavior of a listener and a speaker—and often to a listener who is also a speaker. For this reason a proper model for a valid articulation testing method may require the conscious employment of both speaker and listener in a "speech chain." A model for such a test is that suggested by Allport and Postman (1947) in their investigations of the distortions which occur in the serial transmissions of rumor. In the present study chains of five or more persons were asked to pass along, "exactly as heard," short speech samples, including nonsense words and real words, varying in length, and in segmental and supersegmental structure. Marked serial transmission distortions were found for even short speech chains, particularly for naive speakers. These distortions appeared to be similar to those found by Allport and Postman, including the distortions of "leveling" and "sharpening." Only the more highly conventionalized information survived the serial transmission. The method is suggested as a promising tool for analyzing the conventionalization of phonetic perceptions in speaker-listener chains.
![]()
CiteULike
Connotea
Delicious
Digg
Facebook
Reddit
Technorati
Twitter What's this?