Method for Measurement of Speech Identification

Abstract

A new method for measuring speech-identification behavior is described. Thirty closed-message sets were constructed. Each set contained 10 synthetic sentences, representing 4 levels of approximation to a “real” sentence. Sentence length and informational content were controlled. Message identification under conditions of low-pass filtering and periodic interruption was studied in 30 subjects. Performance varied systematically with informational content. Articulation functions under normal and degraded conditions are compared with analogous functions for more conventional-speech audiometric materials. [This investigation was supported by National Institute of Neurological Diseases and Blindness, National Institutes of Health, Public Health Service, U. S. Department of Health, Education, and Welfare.]