mscmMMAnON ~8Y.:Noa-MAL1.V HEARING 508151215 +011: FlLTER‘ED spew-1 UNDER commonsm HEARING A11: mum-11214110111; ”Thesis far the Degree of M. A. M1cH1GAN STATE umvmsw {Canstgance Rad» Walton W 1964 1'” IBIS LIBRARY Michigan State University ABSTRACT DISCRIMINATION BY NORMALLY HEARING SUBJECTS FOR FILTERED SPEECH UNDER CONDITIONS OF HEARING AID AMPLIFICATION; by Constance Rae Walton The purpose of this study was to analyze the results obtained from normally hearing subjects as they responded to filtered PB Word Lists under three conditions of amplifi- cation by hearing aids. The CID Auditory Test w-22 List 1A was filtered to represent four different hearing loss patterns. Twenty-four university students participated as subjects for the study. Subjects were randomly assigned to the four different filter patterns, six subjects to each pattern. Each of three conditions of amplification was assigned to two of the subjects under each filter pat— tern. A screening test for normal discrimination of Speech eliminated all subjects whose ability to discriminate was not within the defined limits of normality. Subjects assigned to Condition One listened without a hearing aid to the filtered PB Word List. Subjects assigned to Con- dition Two listened with a selected hearing aid to the filtered PB Word List. Subjects in Condition Three listened to the filtered PB Word List by means of a randomly-selected Constance Rae Walton hearing aid. Subjects utilizing a hearing aid adjusted the volume control to the most comfortable loudness level as determined by a recording of continuous Speech. The Speech signal was presented at 55 decibels (sensation level) for all conditions. Under each condition the sub- jects reSponded to the filtered word list and recorded their reSponses on forms provided for them. The results of this study showed that in comparing the discrimination scores of normally hearing subjects, a significant difference existed between the filter patterns employed. No significant differences between the dis— crimination scores were observed, however, for the three conditions of amplification. A very slight difference in favor of the selected hearing aids was observed, but this was not statistically significant. The conclusion drawn from this study was that there is a significant difference in the Speech discrimination scores of normally hearing subjects as a result of lis- tening to PB Word Lists under conditions of four differ— ent filter patterns. DISCRIMINATION BY NORMALLY HEARING SUBJECTS FOR FILTERED SPEECH UNDER CONDITIONS OF HEARING AID AMPLIFICATION By Constance Rae Walton A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Speech 1964 TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES LIST OF APPENDICES Chapter I. STATEMENT OF THE PROBLEM Introduction . . Purpose of Study. Importance of Study. Definition of Terms. II. REVIEW OF THE LITERATURE Frequency Distortion and Intelligibility. Summary. . . . . III. SUBJECTS, EQUIPMENT, MATERIALS, AND TESTING PROCEDURES . . Subjects Equipment Materials Procedure Summary. IV. RESULTS AND DISCUSSION Introduction Discussion V. SUMMARY AND CONCLUSIONS Summary. . . Conclusion. . Implications for Future Research BIBLIOGRAPHY APPENDICES ii Page iii iv DON KO ONU‘l-ITH i—' NH LIST OF TABLES Table ‘Page I. Summary for Analysis of Variance . . . . . 36 iii LIST OF FIGURES Figure Page 1. Block Diagram of Equipment Arrangement for Filtering Procedure . . . . . . . . 28 2. Mean Per Cent Scores for Filter Patterns . . 38 3. Mean Per Cent Scores for Conditions of Amplification . .. . . . . . . . . 41 iv LIST OF APPENDICES Appendix A. Filter Pattern Number Filter Pattern Number Filter Pattern Number Filter Pattern Number 41'me Selected Hearing Aids Selected Hearing Aid "A" Selected Hearing Aid "B" Selected Hearing Aid "c" Selected Hearing Aid "D" Randomly~Se1ected Hearing Randomly-Selected Hearing Randomly-Selected Hearing Randomly-Selected Hearing Randomly-Selected Hearing Subject Response Form . Assignment of Subjects According to Condition and Filter Pattern Raw Scores Per Cent Correct for Experimental Conditions . . . . Aid Aid Aid Aid "E... "F n "G" Aid "H" 72 CHAPTER I STATEMENT OF THE PROBLEM Introduction Auditonycommunication is the most important function 1 and the loss of it can produce of hearing for modern man, practical difficulties as well as adverse psychological effects. The difficulties resulting from a loss of hearing appear to be somewhat proportional to the severity of the impairment. Those individuals with hearing impair— ment generally desire to hear and communicate more effec- tively. Fortunately for some, a hearing aid adequately suits their need. Unfortunately for others, the degree and nature of the loss are Such that amplification does not prove useful. A review of the literature by this writer has indi— cated a disparity in the information available regarding the effects of amplification on the ability to discriminate. Amplification appears to improve discrimination in some individuals with hearing loss, whereas with others discrim- ination is reduced with the increased intensity provided lHallowell Davis, ”The Articulation Area and the Social Adequacy Index for Hearing," LaryngOSCOpe, 58 (1948), 761. by a hearing aid. "The first objective of a hearing aid is to make Speech intelligible."l If this objective is not attained a hearing aid will provide little benefit to the hearing impaired individual insofar as oral communi- cation is concerned. The ability to discriminate appears to be an impor- tant factor in hearing aid selection. Since this is a perceptual rather than a sensory function, it becomes extremely important to evaluate disturbances of auditory perception as a Specific 2 symptom that often accompanies loss of hearing. . . Tests of the ability to understand the Spoken word are generally recognized as the most realistic, valid, and (when prOperly performed) the most sen- sitive tests of auditory function. The instrument that provides maximum intelligibility for the suitable hearing aid candidate is generally considered to be the best instrument for him. Davis feels that the distinction between hearing loss and discrimination loss is fundamental. The Shift downward (discrimination loss) is due to a failure of sense organ, nerve, or brain. It cannot be offset by mere increase in loudness; but sometimes 1Hallowell Davis and 8. Richard Silverman, Hearing and Deafness (New York: Holt, Rinehart and Winston, Inc., 1960), p. 265. 2Otto J. Menzel, "Auditory Discrimination " e, Ear, Nose and Throat Monthly, 41 (October, 1962), fig?“ 3Hallowell Davis, et al., "The Selection of Hearing Aids," Laryngoscope, 56 (1946), 144. it can be off- -set wholly or in part by psychological factors such as auditory training. Portmann and Portmann in a discussion on discrimination ability state that, It appears theoretically difficult that a hearing aid should be able to procure better discrimination than the ear itself, especially as one interposes an amplification which can deform words. However, very often testing proves that the hearing aid is capable of improving the selective power of the different phonetic elements. Studies directed toward the selection of hearing aids (to be discussed in Chapter II) have been made in an attempt to determine those frequency response characteris— tics that are best suited for the hearing impaired individ— ual. Generally the subjects for those studies have been persons with hearing losses. It appears that additional information on ability to discriminate might be obtained by providing amplification via a hearing aid to a normal ear. By presenting phonetically balanced word lists which have been filtered to approximate the frequency distortion typical of several hearing loss patterns, an objective evaluation could be made of the effects of amplification with regard to frequency distortion. Other factors of distortion present in the pathological ear, therefore, would not be influential. lDavis, op. cit., 768. 2Michael Portmann and Claudine Portmann, Clinical Audiometry (Springfield, Illinois: Charles C. Thomas, I961), p. 294. Purpose of the Study The purpose of this study was to analyze and compare the results obtained from normally hearing subjects as they responded to the CID Auditory Test W—22 List 1A under the conditions of amplification as follows: (a) no hearing aid, (b) a selected hearing aid, (c) a randomly-selected hearing aid. The test lists were filtered to approximate the frequency distortion typical of several hearing loss patterns. From this analysis it was hoped that it might be determined whether the speech discrimination performance of normally hearing subjects for filtered word lists could be improved with amplification by a hearing aid and, if so, to what degree. The following null hypotheses were formulated: 1. There is no difference in discrimination scores among normally hearing subjects due to filter patterns (See Appendix A). 2. There is no difference in discrimination scores among normally hearing subjects due to the following different conditions of amplification: (a) no hearing aid—~1istening to filtered Phonetically Balanced (PB) Word Lists. (b) a selected hearing aid--1istening to filtered PB Word Lists with a selected hearing aid (see Appendix B). (c) a randomly—selected hearing aid—-1istening to filtered PB Word Lists with a randomly— selected hearing aid (see Appendix C). 3. There is no difference in the discrimination scores of normally hearing subjects due to inter— action of filter patterns by conditions of amplification. Importance of the Study This study is considered to be important in that it may yield information that will be helpful to those who conduct hearing aid evaluations and counsel the hearing aid wearer. Since auditory communication is the primary function of hearing in man,1 the significance of assessing the ability to discriminate in the hard of hearing individual becomes apparent. If it is determined that certain means of amplification prove more useful to the ability to dis- criminate than others, then it follows that these means should be utilized by the hard of hearing individual to improve his discrimination performance. By assessing the ability of normally hearing subjects for discrimination of filtered word lists, other distortion factors present in the pathological ear are eliminated permitting an objective evaluation of the performance of lDavis, op. cit. several conditions of amplification. Any observed differ— ences in the subjects' ability to discriminate should be the result of the effects of amplification of a selected band of frequencies by the hearing aid and not the function of distortion factors present in the hearing impaired ear. Definition of Terms For the purpose of this study the terms used are defined in the following manner. Amplification.—-To increase the intensity of sound, in this case by means of a hearing aid. Normally hearing subjects.--Persons who are able to attain a discrimination score of 94% or better monaurally. The CID Auditory Test W-22, Lists 3A and 3B were presented as the Speech signal at 55 decibels sensation level, sound field, and with 55 decibels effective masking (saw tooth noise) in the contralateral ear. CID Auditory Test W—22.——A discrimination test con— sisting of four lists of 50 phonetically balanced one- syllable words. These words are usually presented to the patient at a level above his Speech reception threshold and are available in recorded form.l Speech discrimination test.--A test which allows for evaluation of a patient‘s ability to differentiate among lIra J. Hirsh et al., "Development of Materials for Speech Audiometry," Journal of Speech and Hearing Disorders, 17 (September, 1952), 322—323. acoustically similar sounds or among words that contain acoustically similar sounds. The experimenter uses the terms "discrimination" and "intelligibility" interchange- ably for the purpose of this study. Filter.——An electrical network composed of reactive elements used for attenuating or removing a given band 1 or audio frequencies. Filter pattern,-—The characteristic response of the output Signal from the filter—attenuator system. Frequency reSponse.--"Frequency reSponse is the relative acoustic gain of the hearing aid expressed as a function of the frequency."2 Randomly-selected hearing aids.--Hearing aids with different frequency response characteristics selected randomly from the clinic stock without regard to specific response characteristics,acoustic gain, maximum output, etc. Selected hearing aids.——Hearing aids selected by the experimenter. An effort was made to compensate for the frequency distortion of the subject's filter pattern by choosing the instrument that selectively amplified those frequencies for which the intensity was most distorted by filtering. l"Filter,” Audio Cyc10pedia (Indianapolis 6, Indiana: Howard W. Sams and Co., Inc., 1959). 2"Methods for Measurement of Electroacoustical Charac- teristics of Hearing Aids," American Standards Association. Bulletin, S3.3 (1960), 7-15. /» Organization of the Study Chapter I contained the statement of the problem that led to this study. An introduction to the topic and an outline of the purpose of the study were included. Hypo— theses were stated that were to be considered in this study, the importance of the study was discussed, and terms to be used were defined. Chapter II contains a review of the literature per- tinent to this topic, and Chapter III consists of a dis- cussion of the subjects. Equipment, materials, and testing procedures utilized in this study are also within Chapter III. Chapter IV discusses the results of the study, and Chapter V contains the summary and conclusions. Implications for further research are also discussed in the final chapter. CHAPTER II REVIEW OF THE LITERATURE Clinical hearing aid evaluations are performed to guide the hard of hearing in the selection of an aid. Differences exist in the way such evaluations are conducted. Generally, however, differences among aids are sought with respect to speech discrimination, gain, tolerance, and noise thresholds.l In 1949 Carhart suggested that more time than is necessary is being Spent in the testing of several aids on each individual.seen for a hearing aid evaluation. ". the time should come when otology is exerting positive guidance by referring problem cases to hearing clinics and other cases directly to reputable dealers."2 Shore, Bilger, and Hirsh reported that: .most of the adults seen at the hearing clinic at Central Institute for the Deaf appear not to re- quire the extensive evaluation they receive, and we find that a good many of our hearing-aid recommenda- tions are made on the basis of non—auditory factors like price, size and availability of service. lIrvin Shore, Robert c. Bilger, and Ira J. Hirsh, 'Hearing Aid Evaluation: Reliability of Repeated Measure— ments," Journal of Speech and Hearing Disorders, 25 (1960), 152. 2Raymond Carhart, ”Hearing Aid Selection by University Clinics," Journal of Speech and Hearinngisorders, 15 (1950), 112. 3Shore, Bilger, and Hirsh, op. cit. 9 10 An investigation of procedures in other hearing clinics might well indicate similar practices. Several studies have been reported in the literature concerning hearing aid selection. The studies reported in this chapter include only those where monaural aids were under consideration. At the Tenth Anniversary Meeting of the Acoustical Society of America in 1939, Knudsen alluded to results of tests performed in the hearing laboratory at University of California. These, he stated, showed not only the value of high quality amplification but also the advantage of selective amplification in certain cases. In 1940 Watson and Knudsen tested listening intelli- gibility of hearing impaired individuals using both uniform and selective amplification. Conclusions of their study H were that, .the superiority of properly prescribed selective amplification over uniform is greater for ears with perceptive impairment than for those with conduc- tive impairment."2 They also suggested tentative criteria to determine whether or not to prescribe selective amplifi- cation as well as how to prescribe it. These criteria were: lVern 0. Knudsen, "An Ear to the Future," The Journal of the Acoustical Society of America, 11 (July, 1939), 30. 2N. A. Watson and V. 0. Knudsen, "Selective Amplifi- cation in Hearing Aids," The Journal of the Acoustical Society of America, 11 (April, 1940), 418. ‘U ‘- I III | 1 1! l' 1l1| ll: I'll lll ll 1. Determine the air and bone conduction threshold curves and the air conduction hearing loss for Speech at threshold. 2. Determine the most comfortable level above threshold for listening to Speech; take several articulation lists at this level, and from these calculate the corresponding percent— age syllable articulation; find the hearing loss for speech (in decibels) at this most com- fortable level by comparison with the normal curve. . . 3. If the hearing loss for Speech at the most comfortable level is approximately equal to the hearing loss for Speech at threshold, pre- scribe uniform amplification. . . 4. If the hearing loss for Speech at the most comfortable level is less than the hearing loss for speech at threshold, prescribe selec- tive amplification by means of the criterion based on the "most comfortable equal loudness curve" . . . . 5. If the hearing loss for Speech at the most comfortable level is appreciably greater than the hearing loss for speech at threshold, prescribe selective amplification by means of the criterion based on the "most comfortable equal loudness curve” Perhaps one of the most revolutionary studies in- volving the transmission of Speech by hearing aids was that of the Psychoacoustic Laboratory at Harvard Univer— H sity. Known as the "Harvard Report, this study was H O concerned with the .theoretical analysis of the general problem of 'fitting' a hearing aid, and a critique 2 of several present and proposed 'fitting' procedures." 1Ibid. 2Davis et al., 0p. cit., p. 86. 12 The relative value of several widely different pat— terns of frequency response characteristics was obtained for 25 hard of hearing ears by means of a Master Hearing Aid. The The patterns which yielded the best results were found to bear very little relation either to the subject's audiogram or to an equal-loudness contour. Even if the audiogram was used only as a general guide to determine whether or not additional am— plification should be provided for high frequency, it proved actually misleading in several instances. 0n the other hand, a particular set of frequency- response patterns proved uniformly successful for all ears tested. These results, definitely con- trary to the original expectations of the experi- menters, seem to Show that it is possible to Specify the desirable frequency characteristics of a hearing aid more successfully by a simple general rule than by an interpretation of the patient's audiogram. Experimental evidence seems to Show that the principle of 'selective amplification' to com en- sate for impairment of hearing is fallacious. authors concluded that, The appropriate frequency characteristic for a hearing aid is not correctly indicated by current principles of 'audiogram fitting' or 'selective amplification.‘ A uniform frequency characteris- tic that can be varied by a tone control between 'flat' and a moderate accentuation of high tones will provide the most satisfactory performange for all or nearly all cases of hearing loss. Davis and others studied appropriate design objectives for the construction of hearing aids. Using a Master Hearing Aid on 18 hard of hearing subjects, the relative value of the various frequency patterns for each of these selected ears was determined. 1Ibid., p. 102—103. 2Ibid., p. 87. 13 The consistent superiority of moderate high-tone emphasis in making Speech intelligible to hard-of- hearing ears disproves the pOpular theory that the best frequency pattern for a hearing aid is one which compensates for a patient's individual hear- ing loss by 'mirroring‘ his audiogram. . . . As a practical matter, the best choice for all ears lies only between a flat pattern and moderate high— tone emphasis. In 1948 Hudgins et a1. studied the problem of design objectives in a portable hearing aid. The experi- mental hearing aid incorporated most of the desirable features emphasized from previous studies. Its perform- ance on a group of six hard-of—hearing persons was compared with the performance of a Master Hearing Aid and two representative commercial hearing aids. Test materials consisted of phonetically balanced word lists which were presented with static noise in a signal to noise ratio of +15 to +20. Articulation curves for the subjects were derived for all instruments. Results of this study confirmed the findings of the previous experimental study in that all types of hard—of-hearing subjects performed better with an instru— ment that provided adequate gain and a relatively broad, flat, high—fidelity frequency response.2 lHallowell Davis et al., Hearing Aids: An Experi- Study of Design Objectives (Cambridge, Massachusetts: Harvard University Press, 1947), p. 2C. V. Hudgins et al., "The Comparative Performance of an Experimental Hearing Aid and Two Commercial Instru- ments," The Journal of the Acoustical Society of America, 20 (May,l948), 241—254. 14 A study by Shore, Bilger, and Hirsh concerned possible significant differences in using different hearing aids, different tone settings, and different testing days on 15 hard—of-hearing patients. The subjects all had mild to moderate hearing losses in three diagnostic categories: conductive, mixed, and sensory—neural. Eight specific combinations of four body-type hearing aids with two tone settings were tested on each subject on each of four testing days. The auditory measures of gain, or residual hearing level for speech, Speech dis— crimination in quiet, and speech discrimination in noise were analyzed statistically for each patient. Results indicated that, . .differences attributable to different hearing aids occur most often for gain. . .less often for discrimination in quiet. . .and not at all for discrimination in noise. . . .It is concluded that the reliability of these measures is not good enough to warrant the investment of a large amount of clini- cal time with them in selecting hearing aids. Jeffers investigated quality judgments as a criterion in hearing aid selection. Five hearing aids arranged in pairs were tested on 32 subjects with conductive type hearing losses. Subjects listened to one—minute speech recordings and were asked to list the preferred hearing aid which transmitted the best quality for him. 1 Shore, Bilger, and Hirsh, op. cit., p. 167. 15 Results of her study showed that, ". . .(1) the 'typical' acoustic differences in the hearing aids were sufficient to result in real differences in the quality of the reproduced Speech and (2) that the subjects were excellent judges of these differences."1 Jerger, Carhart, and Dirks tested the Speech intel- ligibility of 48 subjects with bilateral sensorineural hearing loss. The following conditions of amplification were employed: (a) binaural, (b) monaural—head, and (c) monaural—body. Two listening conditions were imposed. In the first condition, the subject heard 50 PB words through one loudSpeaker while, simultaneously, 50 sentences from the Bell Telephone Intelligibility Lists were heard through another loudSpeaker. The subject was asked to repeat back each of the 50 words to the experimenter. In the second listening condition, 30 questions and commands were heard from one loud speaker and continuous discourse from the other. The primary signal Shifted in an unpre- dictable fashion from one loudspeaker to another to create a relatively unstable listening condition. Subjects were tested individually and the order of the successive amplification conditions was counterbalanced from one subject to the next. In general, the results lJanet Jeffers, ”Quality Judgment in Hearing Aid Selection," Journal of Speech and Hearinngisorders, 25 (1960), 266. 16 failed to demonstrate any appreciable advantage of binaural hearing-aid amplification over monaural amplification for these Specific listening conditions.l One of the more recent studies in hearing aid selection is that carried out by Zerlin. Zerlin's study involved the recording of hearing—aid outputs onto dual— channel magnetic tape. By using the method of paired comparisons, the listener was able to compare two aids at a time. Running Speech in the presence of cafeteria noise was fed into six different hearing aids as well as half- lists of the CID W-22 recordings of PB Word Lists. Hard— of-hearing subjects listened to each set of two aids and made a paired-comparisons choice. This ultimately gener- ated a rank-ordered preference series for all six aids. "It was noted that while the intelligibility score results did not differentiate among the aids, preference scores based on the paired comparisons yielded clear—cut dis— criminations among five of the six."2 One of the most recent studies reported is that of Shore and Kramer. They devised a new testing procedure lJames Jerger, Raymond Carhart, and Donald Dirks, "Binaural Hearing Aids and Speech Intelligibility,” Journal of Speech and Hearing Research, 4 (1961), 137—148. 2Stanley Zerlin, "New Approach to Hearing Aid Selection," Journal of Speech and Hearing Research, 5 (1962) 375—376. 17 consisting mainly of audiometric testing and counseling without the recommendation of a specific aid. Questionnaires were mailed to two groups of people, those for whom a specific hearing aid had been recommended and those for whom such a recommendation had not been made. Replies were statistically analyzed to note any significant differ- ences between the two groups. The two groups were statistically different on only a few items of the questionnaire. . . .The authors conclude that evidence from this study as well as the experience with the new testing procedpre suggest the recommendation of this procedure.” Frequency Distortion and Intelligibility A great deal of research has been reported in the literature with regard to the effects of distortion upon the intelligibility of Speech. Steinberg reports that, The waves of Speech sounds are characterized by three quantities, amplitude, frequency and phase, all three of which are essential to the correct recognition of the sounds by an auditor. In gener- al, the term distortion refers to relative changes in one or more of these quantities, in the process of transmitting the sound waves from Speaker to auditor. lIrvin Shore and Joan C. Kramer, ”A Comparison of Two Procedures for Hearing Aid Selection,” Journal of Speech and Hearing Disorders,28 (May, 1963) 165. 2John C. Steinberg, ”Effects of Distortion Upon the Recognition of Speech Sounds,ll The Journal of the Acoustical Society of America, 1 (1929) 121. 18 It follows that if Speech is distorted in any one or more of the three quantities mentioned by Steinberg, par- ticularly amplitude and frequency, the discrimination of the auditor will be affected. The following summary reviews the literature with regard to frequency distortion and intelligibility of Speech. In order to determine the importance of various frequency ranges, Steinberg filtered portions of Speech using two sets of low—pass and high-pass filters. The effects of frequency distortion upon articulation was tested by changing the cut-off frequency of these filters over the speech frequency range. Conclusions showed that, ". . .for the case of syllable articulation, which involves all of the speech sounds, this range extends from about 200 cps to 8000 cps. Each of the various groups of sounds does not require this whole range."1 It was observed that in the case of the stop and frica- tive consonants, .the high pass filter tests indicate that these sounds have no frequencies of very great importance below 1000 cycles. However, when all frequencies above 1000 cycles are eliminated, (low-pass filter tests), articulationscn?an order of 50-70% are obtained, which appears inconsistent with the high-pass filter tests. 1Ibid., p. 128. 2 Ibid., p. 129. l9 Pollack attempted to determine the intelligibility of various ranges of low and high Speech frequencies against a background of white noise. Standardized articu— lation testing procedures were utilized. Low-pass and high—pass filters were employed to remove various fre- quency ranges. Two trained talkers read lists of mono- syllabic words to nine experienced listeners. Results of this study indicated that, in general, .speech intelligibility increased as (1) the inten— sity level of the Speech signal and (2) the frequency "1 It was observed that the con- range were increased. tribution of the higher Speech frequencies and very low Speech frequencies alone was very small. When only those frequencies above 2375 ops were passed, .intelligibility in terms of word articulation was a bare 5% at maximal gain. . . .practically no words were recognizedewhen the frequencies below 425 cps alone were heard. Closely related to Pollack's investigation of intelligibility of filtered speech in noise is that of Egan and Weiner' study. Egan and Weiner conducted artic— ulation tests using a large number of communication lIrwin Pollack, ”Effects of High Pass and Low Pass Filtering on the Intelligibility of Speech in Noise,” The Journal of the Acoustical Society of America; 20 (May, (1948), 262. 2Ibid., p. 263. 20 systems having band widths ranging from about one—half octave to a system covering the entire range of Speech frequencies. Two spectra of masking noise were used, and each system was tested over a wide range of Speech— to-noise ratios. The Speech was filtered before mixing with noise in one group of experiments and in the other both Speech and noise were passed through the filter. For each of the band-pass systems a relation between syllable articulation and level of received Speech was obtained. Families of equal articulation contours were derived from these gain functions. These contours showed how the gain had to be changed for a given change in the band width in order to maintain a constant articu- lation score.l Studies have been reported in the literature on frequency and amplitude distortion with respect to intel- ligibility. Pollack investigated the combined effects of these two variables on the intelligibility of speech in noise. The recorded Psychoacoustic Laboratory PB Word Lists were subjected to sharp cut—off filtering, then to infinite peak clipping. The lists were then filtered again and mixed electrically with masking noise 1James P. Egan and Francis M. Wiener, ”0n the Intelli- gibility of Bands of Speech in Noise,” The Journal of the Acoustical Society of America, 18 (October, 1946), 435. 21 and presented to earphones. Listeners reSponded to the distorted Speech signal and the percentage of correct words was used to determine the intelligibility of Speech. Results reported by Pollack were that, The intelligibility of unclipped speech, relative to that of the peak-clipped signal under corres- ponding experimental.conditions, is a function of the signal-to-noise ratio under test and is, to a rough approximation, independent of the frequency range of the Speech signal passed. At high S/N ratios, the intelligibility of the unclipped Speech signal is higher than that of the severely peak— clipped signal. Under low S/N ratios, however, the intelligibility of the latter is considerably higher than that of the unclipped signal.1 In the early 1950's Hirsh, Reynolds, and Joseph reported an investigation where Speech material was presented under two conditions: filtering and masking by white noise. Listeners were six male college students. The authors determined that most of the speech frequen- cies had to be eliminated before intelligibility of any kind of speech material is seriously impaired. When frequencies below 3200 cps were eliminated articulation scores decreased noticeably. Under low-pass filter conditions, it is only when all the frequencies above 800 cps are eliminated that the articulation decreases rapidly from its maximum.2 1Irwin Pollack, "0n the Effects of Frequency and Ampli— tude Distortion on the Intelligibility of Speech in Noise.” The Journal of the Acoustical Society of America, 24 (Septem- ber, 1952), 538-540. 21. J. Hirsh, Elizabeth G. Reynolds, and Maurice Joseph, "Intelligibility of Different Speech Materials,” The Journal of the Acoustical Society of America, 26 (1954), 530-538- 22 Owens studied the relationship between word famili- arity and intelligibility. He used seven monosyllabic word lists divided into groups of three, two and two, each group list being matched phonetically and varied systematically with respect to familiarity according to the Lorge count. Each list was recorded under seven conditions of distortion achieved by low pass filtering and presented auditorily to seven listener groups of 30 college students each. Each group listened to all seven lists under a Single condition of frequency dis— tortion. The continuous, although irregular, increase in discrimination errors on all lists as frequency dis- tortion became greater, illustrated the basic importance of the phonetic elements. As sounds were eliminated, attenuated, or distorted, all words were more difficult to discriminate. The author concluded that although word familiarity played an important part in the discrimination of Speech, phonetic content remained the dominant factor. Owens tenuously related the filter settings in his study and high frequency hearing loss. He generalized that a person with such a loss would have noticeably more difficulty discriminating Speech sounds to the extent that frequen- cies below 1560 cps were involved.1 lElmer Owens, ”Intelligibility of Words Varying in Familiarity," Journal of Speech and Hearing Research, (1961), 113-129. 23 In 1963 Giolas and Epstein compared the intelligi- bility scores for monosyllabic word lists with a sample of continuous discourse. The sample of continuous dis— course and four word lists were reproduced on magnetic tape under seven conditions of filtering. One hundred and seventy-five normally hearing subjects served as listeners for the study. W—22 word lists yielded con- sistently higher scores than did the PB—5O Word Lists. In all situations, however, errors increased as distortion increased.l Summary An overall view of the literature concerning hearing aid evaluations revealsthat prior to 1946, the concept of selective amplification was predominant in the selection of hearing aids. The ”Harvard Report" resulting from studies at the Psychoacoustic Laboratory at Harvard University proved revolutionary in refuting this concept. This study, also supported by further investigations, showed that a hearing aid with a relatively flat fre- quency response provided the most satisfactory perform- ance for nearly all cases of hearing loss. 1Thomas G. Giolas and Aubrey Epstein, "Comparative Intelligibility of Word Lists and Continuous Discourse, Journal of Speech and Hearing Research, 6 (December, 19537} 357. 24 With regard to frequency distortion and listener intelligibility a general trend is evident in the litera- ture. AS the frequency distortion increases, subjects' ability to discriminate decreases. However, most of the Speech frequencies must be eliminated before the ability to discriminate is severely impaired. The very high frequencies and very low frequencies appear to contribute very little to the intelligibility of speech. CHAPTER III SUBJECTS, EQUIPMENT, MATERIALS AND TESTING PROCEDURES Subjects Twenty-four university students, 12 females and 12 males, were subjects for this study. It was determined audiometrically that each subject had normal hearing. A criterion in selection of subjects was that all subjects attain a discrimination score of 94% or better monaurally, sound field, at 55 decibels (sensation level). Fifty-five decibels effective masking (saw tooth noise) was introduced into the contralateral ear. Subjects responded to the CID Auditory Test W-22 Lists 3A and 3B. The age of the subjects ranged from twenty-one to thirty-two years, and all had four or more years of college education. Equipment The following equipment was employed for the purpose of this investigation: Four body-type randomly—selected hearing aids Four body—type selected hearing aids Plastic ear inserts of general sizes Four band-pass filters (Model 25 Allison, Serial numbers 116, 117, 118, and 119. 25 26 One tape recorder (Ampex Model 601, Serial 2204) One Thorens turntable and tonearm Four Hewlett—Packard attenuators (Model 350A, Serial numbers 00305340, 00304345, 00305347, and 00305363) Testing suite (dimensions 7'5" by 9'6") Allison audiometer (Model 20B, serial 13) Beltone audiometer (Model 12A, SD 1102) One Bruel and Kj&r sound level meter (Model 2203, serial number 94578 1 One Bruel and KjXr volt meter (Model 2409, serial number 6329?) Materials The materials utilized in this study consisted of: Magnetic tapes (3M 202) CID Auditory Test w-22 recording (Lists 1A, 3A, and 3B Continuous Speech--Fulton Lewis, Jr. recording Forms for the recording of subject's reSponses (see Appendix D) Procedure Preliminarypprocedures.—-A 1000 cps calibration tone was recorded at the beginning of each tape in order that the Speech stimulus could be presented at the same level to all subjects. The recorded CID Auditory Test W—22 List 1A was played on the turntable. The signal was introduced to four different band—pass, filter—attenuator systems. The 27 Slope at the cut-off points was approximately 30 decibels per octave. The Signals were than recombined and repro- duced on magnetic tape. The attenuators reduced the signal so that the fil- tered band fell on the threshold of intensity represented by typical hearing loss patterns (see Appendix A). The filter permitted a band of frequencies to be filtered to approximate the level of the band of frequencies repre— sented on the audiogram form. The out-put meter recorded the intensity of therecombinedsignal and served as a means of checking the desired intensity of the filtered bands (see Figure 1. for a block diagram of arrangement of equipment). The procedure was followed for the filtering patterns of four typical hearing loss patterns. A recording of continuous discourse by Fulton Lewis, Jr. was also passed through the filter four times to correSpond to each of the W—22 word list filter patterns. Forms for the recording of the subjects' responses were devised by the examiner and are included in Appendix D. Testing procedure.--The following testing procedure was established. The 24 subjects were divided according to a table of random numbers1 into four groups of six subjects each, group lHerbert Arkin and Raymond R. Colton, Tables for Statistics (New York: Barnes and Noble, Inc., 1950), p. 142. Amomm Hope: sang eds Hossmv Lopez DSQDSO Aaom .opsoooogm wzfisouafim pom H0602 onEm mm mm om me am am mm mm 06H m o o .nnsm .oz poo as... was was... ass... UT. UT. UTL UT. 0.39 0.89 0.89 0.32 o o T: o «a .L 0 O T: 0 «u .L w 1. D. m .+ D. m m. D. w. M“ .9 “H m. mm m. mm o. .A he as as as so 3T. 8T: 3T: 3.... ID. ID. ID. ID. 9 B B B m... m m m a .oz m .oz m .oz a .oz spouumm spoupmm camppmm spouumm Access pandas unease socflfim illii Alll‘ mZOHEHono AdeMZHmmmNm mom Bommmoo BZMO mMm mmmoom 3