Abstract :
A representative sample of stop-consonant sounds was recorded on magnetic tape. These were then cut out and spliced close together; they were played back and their oscillograms recorded with a camera of the continuously motor-driven type. On this film the required parts of the sounds, their gap, burst, transition, etc., can be located. By simple proportion the corresponding regions are marked off on the original recording tape. Each one of these parts is then cut out in turn and the part from another sound inserted in its place. A large number of these resulting sounds are collected on one long tape. The final recording is then played to an appropriate audience who judge the sounds and record what they hear. Their responses are submitted to mathematical analysis, which gives an indication of how the information is distributed in the stop-consonant sounds. In addition, evidence is obtained and ideas are advanced of the manner in which stop consonants are recognized. These have important implications for devices intended to recognize stop consonants efficiently.
