A Comparison of Four Feature Systems Using Data from Three Psychophysical Methods Four feature systems (Miller-Nicely, Singh-Black, Wickelgren, and Chomsky-Halle) were compared under three different perceptual tasks involving the ability of subjects to auditorily discriminate pairs and triads of 22 prevocalic English consonants. More than 1000 listeners participated in these experiments. The features in each system were weighted using multiple regression techniques. ... Research Article
Research Article  |   December 01, 1972
A Comparison of Four Feature Systems Using Data from Three Psychophysical Methods
 
Author Affiliations & Notes
  • Sadanand Singh
    Ohio University, Athens, Ohio
  • Gordon M. Becker
    University of Nebraska at Omaha, Omaha, Nebraska
Article Information
Research Article   |   December 01, 1972
A Comparison of Four Feature Systems Using Data from Three Psychophysical Methods
Journal of Speech, Language, and Hearing Research, December 1972, Vol. 15, 821-830. doi:10.1044/jshr.1504.821
History: Received April 30, 1971
 
Journal of Speech, Language, and Hearing Research, December 1972, Vol. 15, 821-830. doi:10.1044/jshr.1504.821
History: Received April 30, 1971

Four feature systems (Miller-Nicely, Singh-Black, Wickelgren, and Chomsky-Halle) were compared under three different perceptual tasks involving the ability of subjects to auditorily discriminate pairs and triads of 22 prevocalic English consonants. More than 1000 listeners participated in these experiments. The features in each system were weighted using multiple regression techniques. The weights differed from one data collection method to another for the same feature system, and from one system to another for the same data collection method. The 11-feature Chomsky-Halle system showed the highest multiple correlation for two of the three data collection methods. However, when weights obtained from one method were used to predict the data from another method, the Chomsky-Halle system was best on only one of the three sets of data. The results raised some questions regarding the generality of the systems, the appropriateness of the features, or the appropriateness of the analysis technique.

Order a Subscription
Pay Per View
Entire Journal of Speech, Language, and Hearing Research content & archive
24-hour access
This Article
24-hour access