An attempt to predict speech discrimination from intensive analysis of peripheral data on a group of hard-of-hearing subjects



Journal Title

Journal ISSN

Volume Title



An attempt was made in the present study to predict speech discrimination from a thorough analysis of pure tone data. An assumption was made throughout the study that speech discrimination is the resultant of two abilities on the part of the subject--the ability to receive peripheral (pure tone) information by his ears, and the ability to combine that information into a meaningful experience by some more central sort of function or functions. The problem was investigated from two approaches--using individual ears as cases, in which the central function was left uncontrolled, and using differences between ears as cases, as one measure of controlling this central function. In each phase of the problem, zero order correlations were obtained between all available peripheral measures that seemed pertinent and the criterion, which was a measure of speech discrimination. An original peripheral measure, which appears particularly promising in studying speech discrimination, was derived from the pure tone thresholds. It was called the measure of masking, and gives a relative estimate of the amount of masking of speech components by other components of the same speech sounds. In the phase of this study using individual ears, the highest correlations obtained were those between speech discrimination and the following: 1. Pure tone threshold at 2000 cycles per second. 2. Pure tone threshold at 3000 cycles per second. 3. Total masking of 500, 1000, and 2000 cycles per second. 4. Cochlear reserve at 500 cycles per second. 5. Speech reception threshold. These five measures were subjected to the Wherry-Doolittle method of multiple correlation. The multiple R resulting was found to be .873. The cochlear reserve measure was eliminated from this group, because it correlated negatively with the criterion, and a new multiple R was calculated, using the remaining four tests. The new multiple R, omitting the cochlear reserve measure, was found to be .716. The entire process was repeated, using differences between ears on both the criterion measure and the peripheral measures. It was reasoned that this approach controlled somewhat the variable effects of the central function. Using this now approach, the highest correlations wore obtained between speech discrimination and the following measures: 1. Pure tone threshold at 1000 cycles per second. 2. Pure tone threshold at 2000 cycles per second. 3. The masking measure at 4000 cycles per second. 4. Cochlear reserve at 2000 cycles per second. 5. Speech reception threshold. When these tests were investigated with the Wherry-Doolittle technique, only two were selected as the peripheral battery, the pure tone threshold at 2000 cycles per second and the speech reception threshold. Addition of the second selected test increased the multiple R to .468. Selection of further tests failed to add to the multiple R. It is reasoned, pending further investigation, that using differences between ears is a method employing purer data than the method of using individual ears as cases. If it is indeed a purer measure, then the method of using individual ears adds a certain amount of spuriousness, possibly due to the uncontrolled central function, to any correlation employing it. The conclusion is drawn, again very tentatively, that the lack of correlation between the pure measures represents, very roughly, an estimate of the contribution of the central functions to speech discrimination. In other words, if we assume that speech discrimination is a resultant of only two major factors (peripheral reception and central interpretation), and if we correlate peripheral reception with the criterion, controlling the central interpretation, then the difference between the correlation obtained and perfect correlation should represent very roughly the correlation between the central function and the criterion. Much further research is needed on this problem. Recommendations were made for further research along the lines of attempting to discover what the components of the central function are, and how much this function overlaps into other areas of communication. Further recommendations were made for research toward a more thorough fitting of hearing aids to individual patterns of hearing loss.



Audiometry., Hearing disorders.