THE EDENTKFEATEON EIF SPEAKER CHARACTERISTICS CPN THE BASIS OF AURAL UES Thesis for the Degree of Ph. D. MECWGAN STATE? UNIVERSITY G. PATRICK NERBGNNE 19:67 “it!"“‘l'J'zllfl'lJU'l'flNflIW'fllflflflufl" ' R A R :1: “5:325:33! AB STRACT THE IDENTIFICATION OF SPEAKER CHARACTERISTICS ON THE BASIS OF AURAL CUES by G. Patrick Nerbonne The major purpose of this investigation was to evalu- ate the performance of listeners in identifying correctly the speaker characteristics of Sex, Age, Ethnic Group, Edu- cation Level, Physical Size, and Dialect Region of the Country on the basis of aural cues alone. Of additional interest were the effects of listener training, normal as op— posed to telephone speech, spontaneous as opposed to pre- pared SPeech, and speech sample duration on the performance Of listeners, Six groups of 23 persons each served as listener SUbjeCtS- Three of the groups were exposed to training pro— Cedures while the remaining three groups served as untrained liStener subjects. One trained and one untrained group made identifications of the speaker characteristics on the basis of liStefling to speech samples under each of three duration Con ' ' . dltlons: five seconds, ten seconds,- and 15 seconds. SBeech samples were provided by selected speakers. Si X Speakers provided the stimuli for identifying the Sex G. Patrick Nerbonne characteristic, three males and three females. The same number of speakers provided the stimuli for the Ethnic Group and Physical Size characteristics. These were, respectively, three Negro and three Caucasian speakers and three big (equal to or larger than six feet in height and 180 pounds in weight) and three small (equal to or smaller than five feet, six inches in height and 121 pounds in weight) speakers. Nine speakers provided the stimuli for identifying the Age characteristic, three between the ages of 20 and 30 years, three between the ages of 40 and 50 years, and three between the ages of 60 and 70 years. The same number of speakers provided the stimuli for the Education and Dialect Region characteristics. For the former, these were three speakers with less than a high school education, three speakers who had graduated from high school, and three speakers who were college graduates. For Dialect Region, three speakers were from the East, three were from the South, and three were natives of Michigan. All speakers were males with the ex— ception of the females employed for the Sex characteristic. Each listener subject made eight identifications of the Sex, Ethnic Group, and Physical Size characteristics, two under each of the following speech treatment combinations: (1) spontaneous normal speech; (2) spontaneous telephone speech; (3) prepared normal speech; and (4) prepared tele- phone speech. For the Age, Education, and Dialect Region G. Patrick Nerbonne characteristics, each listener subject made 12 identifi— cations, three under each of the above speech treatment combinations. Speech samples were presented to the listener sub— jects by means of magnetic tape. Re-test measures were made 48 hours after administration of the actual test for each group. Listener subject performance was measured in pro— portion correct. Results indicated the following conclusions: (1) Listeners can differentiate accurately between (a) male and female speakers, (b) Negro and Caucasian speakers, and (c) big and small speakers, and among (d) 20 to 30 year old, 40 to 50 year old, and 60 to 70 year old speakers, (e) speakers with less than a high school education, high school gradu— ates, and college graduates,.and (f) speakers from the Eastern, Southern, and General American dialect regions. (2) Groups of listeners seem to be highly consistent in their identifications of the speaker characteristics on separate occasions. However, the same conclusion cannot be drawn with regard to single listeners. (3) Short term training, as it was employed in the present investigation, does not signifi- cantly improve the performance of listeners in making cor- rect identifications of the speaker characteristics. (4) Differentiations among age categories of speakers can be made more effectively on the basis of spontaneous aural cues G. Patrick Nerbonne (extemporaneous speech) than on the basis of prepared aural cues (oral reading). (5) Differentiations between Negro and Caucasian speakers and between big and small speakers can be made more effectively on the basis of prepared aural cues (oral reading) than on the basis of spontaneous oral cues (extemporaneous speech). (6) Differentiations among the age categories of speakers and among speakers from the three dialect regions of the country can be made more effectively on the basis of aural cues provided by telephone speech than by the aural cues provided under conditions that simulate face-to-face conversation. (7) Differentiations between big and small speakers can be made more effectively on the basis of the aural cues provided by conditions that simulate face— to—face conversation than by aural cues provided by tele— phone speech. (8) Aural cues of ten seconds duration are a more effective means by which to make differentiating judg- ments relative to the speaker characteristics than are aural cues of five and 15 seconds duration. THE IDENTIFICATION OF SPEAKER CHARACTERISTICS ON THE BASIS OF AURAL CUES BY 5 GT Patrick Nerbonne A THESIS submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Speech 1967 TABLE OF CONTENTS Page LIST OF TABLES . . . iii LIST OF FIGURES vii LIST OF APPENDICES ix Chapter I. INTRODUCTION . . . . . . . 1 Purpose of the Study 1 Importance of the Study 4 Definitions . . . . . 10 Limitations of the Study 12 Organization of the Report 14 II. REVIEW OF THE LITERATURE . 16 The Speaker Characteristics 19 Training . . . . . . . . . 73 The Stimulus . . . 77 III. EXPERIMENTAL PROCEDURES . 84 Subjects 84 Apparatus 85 Procedures . . . 86 IV. RESULTS AND DISCUSSION . . 117 Results . . . . . . . . . . . . . . 118 Discussion . . . . . . . . . . . . . 162 V. SUMMARY, CONCLUSIONS. AND RECOMMENDATIONS 185 Summary . . . . . . . . . . . . . . 185 Conclusions . . . . . . . . . 189 Recommendations for Further Research 192 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . 196 APPENDICES . . . . . . . . . . . . . . . . . . . . . . 206 ii Table LIST OF TABLES Mean fundamental frequency used by males and females for "pure vowels" . . . . . . Optimum and habitual pitch measures made on speakers representing the Sex characteristic . . . . . . . . . . . . Obtained and expected grand mean proportion correct and results of single-sample Z tests for each characteristic . . . . Grand mean proportion correct on the retest and product-moment correlations for test- retest performance for each characteristic Mean proportion correct and standard devi- ation for each treatment and difference between the treatment means of the three main factors for the Sex characteristic Summary of analysis of variance comparing the effects of listener training, spon— taneous as opposed to prepared speech, and normal as opposed to telephone speech on listener performance for the Sex characteristic Product-moment correlation coefficients for test-retest reliability on the Sex charac- teristic for trained and untrained subjects, spontaneous and prepared speech samples, and normal and telephone speech samples Mean proportion correct and standard devi- ation for each treatment and difference between the treatment means of the three main factors for the Age characteristic iii Page 22 119 121 122 129 130 131 132 Table Page 16- Product-moment correlation coefficients for test-retest reliability on the Education characteristic for trained and untrained subjects, spontaneous and prepared speech samples, and normal and telephone speech samples . . . . . . . . . . . . . . . . . . 143 17. Mean proportion correct and standard devi- ation for each treatment and difference between the treatment means of the three main factors for the Physical Size charac- teristic . . . . . . . . . . . . . . . . . . 144 18. Summary of analysis of variance comparing the effects of listener training, spon- taneous as opposed to prepared speech, and normal as opposed to telephone speech on 'listener performance for the Physical Size characteristic . . . . . . . . . . . . . . . 145 19. Product-moment correlation coefficients for test-retest reliability on the Physical Size characteristic for trained and un— trained subjects, spontaneous and pre— pared speech samples, and normal and tele- phone speech samples . . . . . . . . . . . . 148 20. Mean prOportion correct and standard devi- ation for each treatment and differences between the treatment means of the three main factors for the Dialect Region characteristic . . . . . . . . . . . . . . . 149 21. Summary of analysts of variance comparing the effects of listener training, spon- taneous as Opposed to prepared speech, and normal as opposed to telephone speech on listener performance for the Dialect Re- gion characteristic . . . . . . . . . . . . 150 22. Product-moment correlation coefficients for test—retest reliability on the Dialect Region characteristic for trained and un- trained subjects, spontaneous and prepared speech samples, and normal and telephone speech samples . . . . . . . . . . . . . . . 152 Table Page 23. Mean proportion correct and standard devi- ation for each treatment of the duration ‘factor over all speaker characteristics . . 156 24. Summary of analysis of variance comparing the effects of listener training, spon- taneous as opposed to prepared speech, normal as opposed to telephone speech, and duration of the speech samples on listener performance over all speaker characteristics . . . . . . . . . . . . . . 157 25. Product-moment correlation coefficients for test-retest reliability over all speaker characteristics for five, ten, and 15 second speech samples 159 vi Figure 10. LIST OF FIGURES Telephone-microphone coupler Block diagram of telephone recording system Response characteristics of the telephone recording system Block diagram of instrumentation used in re- cording the master tapes . . . . . . . . Block diagram of the instrumentation used in correcting the erroneous speech sample excerpts Schematic representation of the operation of the timer and electronic switch Graphic representation of the individual treatment combination means for the : spontaneous-prepared and normal—telephone characteristic . . . . . . . . . Graphic representation of the individual treatment combination means for the spontaneous-prepared and normal- telephone interaction on the Education « . characteristic Graphic representation of the individual treatment combination means for the spontaneous-prepared and normal-telephone interaction on the Physical Size characteristic . . . . . . . Graphic representation of the individual treatment combination means for the spontaneous-prepared and normal-telephone interaction on the Dialect Region characteristic . . . . . . . . vii Page 87 88 90 104 106 106 135 141 146 150 F igure Page 11. Graphic representation of the individual treatment combination means for the du- ration and training interaction over all speaker characteristic . . . . . . . . . . 158 viii Appe ndix LIST OF APPENDICES PARAGRAPHS USED AS SPEECH STIMULI CONSTRUCTION OF THE MASTER TAPES TEXT OF THE TRAINING PROGRAM AND PRACTICE TEST . . . . . . . . . . . . . . . . . PRACTICE ANSWER FORM FOR TRAINED GROUPS AND TEST ANSWER FORMS FOR MASTER TAPES I, II, AND III . . . . . . . . . . . . . . . . PROPORTION CORRECT ON EACH SPEAKER CHARAC- TERISTIC OVER ALL TRAINING, SPEECH, AND DURATION FACTORS FOR THE TEST AND RETEST CONDITIONS . . . . . . . . . . . PROPORTION CORRECT ON EACH SPEAKER CHARAC- TERISTIC FOR THE TRAINED AND UNTRAINED GROUPS UNDER THE SPEECH TREATMENT COMBI- NATIONS FOR THE TEST AND RETEST CONDITIONS PROPORTION CORRECT OVER ALL SPEAKER CHARAC- TERISTICS FOR THE TRAINED AND UNTRAINED GROUPS UNDER THE SPEECH TREATMENT COMBI- NATIONS FOR THE TEST AND RETEST CONDITIONS ix Page 206 219 246 259 276 285 306 CHAPTER I INTRODUCTION The communication that takes place by means of the CursiJL interchange of symbols is not determined solely by the Emilrt:icular units uttered. The manner in which the units are SE¥Cikten, i.e., tone of voice, rate of speech, vocal pitch, arléi similar verbal dimensions, can also take on value and cotltribute significantly to the listener's interpretation of tl1e3 verbal message. In addition, it was also believed that S‘1<2}1 dimensions can provide the listener with information 313(Dut the speaker. This aspect, of verbal behavior, was the cc>11<2ern of the present study. Purpose of the Study It was the purpose of this study to determine the extlent to which various characteristics of speakers can be aczczlarately and reliably identified by listeners through ex- 9053lire to aural cues alone. Specifically, the speaker Chelit-acteristics in question were: (1) Sex; (2) Age: (3) Eth- ‘rit1 Group; (4) Education Level: (5) Physical Size; and (6) Dieilect Region of the country. Of additional interest in tarts of identifying these characteristics were the effects of (l) listener training; (2) differing conditions of aural c:L1e duration; (3) spontaneous (extemporaneous) as opposed to prepared (oral reading) speaking conditions; and (4) simu- J.eat:ed face-to—face speech as opposed to the distorted speech Signal listeners hear over the telephone. In considering 1:rleese problems, the following questions were posed at the outset: 1. Can listeners identify correctly each 2; the speaker characteristics in question on the basis of aural cues alone and, if so, are these identifications re— liable as well as correct? 2. Does training make a difference in the performance of listeners in the correct identification of each '2; the speaker characteristics in question on the basis of aural cues alone? 3. Does presentation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech make a difference in the performance of listeners in the correct identifi- cation of each 2; the speaker characteristics in question on the basis of aural cues alone? 4. Does spontaneous presentation of the message as op- posed to prepared presentation of the message make a difference in the performance of listeners in the correct identification of each 9; the speaker charac- teristics in question on the basis of aural cues alone? 5. Is there an interaction effect between training and presentation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech on the performance of listeners in identifying correctly each 2f the speaker character- istics in question on the basis of aural cues alone? 6. Is there an interaction effect between training and spontaneous presentation of the message as opposed to prepared presentation of the message on the per- formance of listeners in identifying correctly each .9; the speaker characteristics in question on the basis of aural cues alone? 10. ll. 12. 13. Is there an interaction effect between presentation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech and spontaneous presentation of the message as opposed to prepared presentation of the message on the performance of listeners in identifying cor- rectly each 2; the speaker characteristics in question on the basis of aural cues alone? Are there interaction effects among training, presen- tation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech, and spontaneous presentation of the message as opposed to prepared presentation of the message on the performance of listeners in identi— fying correctly each 2; the speaker characteristics in question on the basis of aural cues alone? Does duration of the speech sample make a difference in the performance of listeners in the correct identification of the speaker characteristics in question on the basis of aural cues alone? Is there an interaction effect between training and duration of the speech sample on the performance of listeners in identifying correctly the speaker charac- teristics in question on the basis of aural cues alone? Is there an interaction effect between presentation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech and duration of the speech sample on the per- formance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone? Is there an interaction effect between spontaneous presentation of the message as opposed to preapred presentation of the message and duration of the speech sample on the performance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone? ' Are there interaction effects among training, presen- tation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech, and duration of the speech sample on the performance of listeners in identifying cor- rectly the speaker characteristics in question on the basis of aural cues alone? 14. Are there interaction effects among training, spon- taneous presentation of the message as opposed to prepared presentation of the message, and duration of the speech sample on the performance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone? 3L5. Are there interaction effects among presentation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech, spontaneous presentation of the message as opposed to prepared presentation of the message, and duration of the speech sample on the performance of listeners in identifying correctly the speaker charac- teristics in question on the basis of aural cues alone? 2L6. Are there interaction effects among training,presen— tation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech, spontaneous presentation of the message as opposed to prepared presentation of the message, and duration of the speech sample on the performance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone? Importance of the Study In terms of the content-free aspects of verbal be- haVior providing the listener with information about the speaker, Barbara has indicated that his interest in speech c0“M‘nunication has been based on his belief that speech and pet"sonality are one.1 It was his contention that the symboli- zation process is a fundamental aspect of the whole person- al~icty. Barbara felt that man creates symbols not only to cc"WI-municate verbally with others and thus establish \ 1Dominick A. Barbara, Your Speech Reveals Your figlfififlfiflgfiy (Springfield, 111.: Charles C. Thomas, 1958), ' xi. interpersonal relationships, but also as an integrating pro— ce ss within himself.; He..further’ statedithat in"'man,','...s' a—t-.. tempts to find himself in his search for inner truth and psychic unity, he, of necessity, creates a personal language of his own, with verbal and non—verbal symbols peculiar and 3P6cific to his own way of life.2 Following this same train of thought, Murray related verbal behavior and the individual Characteristics of the speaker by referring to a speech Personality which he described as developing slowly accord- ing to ". . . laws of a differentiating, expanding, and uni- tary behavior pattern . . ." just as personality. itself does. This behavior, Murray felt, is composed of thought, language, vc”ice, and action.3 He indicated that speech development in the child parallels personality development and the same con- ditions that constrict, distort, or enhance the one, con- strict, distort, or enhance the other.4 As the child grows, speech and personality grow, develop, differentiate, and be- come refined together. Speech is a phase of personality and as such, speech and personality are one and the same thing. Based upon this contention, Murray completed the cycle by \ Ibid., p. 20. t. 3Elwood Murray, "A Study of Factors Contributing to he Mal-Developing'of.;the Speech Personality," Speech Mono- %. III (September, 1936). pp. 95—108. 4E1wood Murray, The Speech Personality (Chicago: J. B‘ Lippincott Co., 1944), p. vii. indicating that genuine speech improvement depends upon pe rsonality development . 5 The question of whether or not the communication that takes place by means of conventional verbal expression is limited to content alone, i.e., the meanings individuals attach to symbols and their combinations, is of evident im— Portance. Many professional persons interested in this area feel that such verbal dimensions as the tone of voice, rate of speech, grammatical usage, fluency, and other such charac- teristics carry communicative value about the speaker $9 the message carried by his words that is not obvious or available by the words alone. Barbara, for example, stated that in our culture, people judge one's character, intel- LeCztual capacity, and social standing by the way he talks.6 In addition, Fairbanks and Pronovost found that male speakers, when asked to simulate the emotions "contempt," "anger," " fear," "grief," and "indifference" in the reading of five different paragraphs, actually used different pitch levels on each selection.7 The fact that these pitch differences were found is evidence in support of the contention that com- munication does not take place by word meaning alone. \ SIbid., p. 3. 6Barbara, op. cit., p. 3. 7Grant Fairbanks and Wilbert Pronovost, "An Experi- mel'ltal Study of the Pitch Characteristics of the Voice uring the Expression of Emotion," Speech Monographs, VI December, 1939), pp. 87-104. .EPIthhermore, when the authors presented recordings to listeners made during the oral reading of a sentence common tzcat each of the five emotion paragraphs, they found remark— able interjudge agreement. The recordings simulating "in- dij.1fference" and "contempt" were judged by 88 and 84 per cent CDiE the listeners, respectively, to be conveying these emo- tljuc>ns. Interjudge agreement on the emotions "anger" and uEilrief" was 78 per cent while 66 per cent of the listeners é‘Ialuated the simulations of "fear" as really conveying this eIt't<>tion. Licklider and Miller indicated the importance of l=>C>th of the items cited above.8 They stated that speech Serves not only as a carrier for the intelligence of ma- tlealrials and meaning, as shown by Fairbanks and Pronovost, 13111: it also furnishes secondary information about the speaker, €3«-§;., his identity, background, and so forth, as suggested I33? Barbara.10 With reference to this latter point, Murray itIC'iicated that speech development is the result of the en- ti:l:‘e history of a person. One's heredity and total environ- n[Karat make him what he is; and his speech growth, just as the (Dilller factors that make him an individual, parallels his un- IE<>JLding physical, intellectual, emotional, and moral \ 8J. C. R. Licklider and George A. Miller, "The Per— ception of Speech," Handbook of Experimental Psychology, ed. ‘ S. Stevens (New York: John Wiley and Sons, 1951), pp. 1040-1074. 9Fairbanks and Pronovost, loc. cit. 10Barbara, loc. cit. E><>wers.ll Barbara wrote that each neWborn child has his own nn<>ods, temperaments, potentialities, gifts, and particular a czaapacities. Also, each individual child has his own physical, Irleantal, and emotional attributes.12 Speech is learned with- c>111z conscious effort and as we begin to relate interdynami- czangLy with the world about us, it assumes a special character aarléi identity of its own. Depending on our own individual lyiJEe—experiences, needs, education, beliefs, attitudes, feel— ziruggs, and convictions, as well as the individualities with Which we are born, our modes of speaking assume their own form, shape, and purpose.13 The individuality of verbal expression was also de- Scliribed by Lojos.l4 Voice tells its own story. There are a million different kinds of voices . . . Some voices are formed by robust muscles; they boom like trumpets, carry their messages with graceful ease. You do not see voices, but you know them as well as if you had seen them. According to Barbara, the verbal use of symbols is a:LSKD an individual thing. He felt that the symbols one (zlaxboses to use from the whole body of those available depends \ 11Murray, The Speech Personality, p. 9. 12Barbara, 0 . cit., p. 37. 13l§l§°’ p. 6. ‘1 l4Egri Lojos, Xpur Key to Successful Writing (New Ork; Henry Holt Co., 1952), p. 57. c>r1 his physical structure, environment, and psychic health, j.r1 addition to conventionality.15 As indicated above, the present study was not con- cerned with what the content-free aspects of verbal behavior <2<>nmnunicate to listeners with regard to what the speaker is saying, but what these verbal dimensions communicate to 1.5.srteners about the speaker himself. The importance of such information is evident from several standpoints. Positive results would be applicable for law enforcement and/or se— <=lerity purposes and would be consistent with current think- .ilis3'*with regard to voice printing. More practically, hows eVex, the project was of interest from the standpoint of 13C>1213 input and output. In terms of the latter, the project ill‘Irolved spontaneous and prepared speech samples, each de- livered to the listeners under undistorted (normal) and tele- pl'lone speech conditions. In addition, the speech samples in Clue stion were of extremely short duration (from five to 15 sec<>nds). In terms of input, three aspects were of interest: (‘J-) the fact that only the auditory input channel was avail— EiIQCLe to the listeners as they attempted to make the discrimi- Ilat-ions requested of them; (2) the possibility of obtaining G‘ifferences in listener performance that could be attributed to the various output conditions described above; and (3) the DQSSible effects of training. The fact that only one input \ 15Barbara, op. cit., p. 20. 10 (:11annel (auditory) was utilized may also have certain impli- czzations for the blind. In addition to these considerations, of primary con- <2earn.to the experimenter on the question of significance or importance was the possible contribution of the study toward ea. Inore complete understanding of normal speech. Definitions Listeners.--Undergraduate college students at Michigan EStZEite University who were enrolled in Speech 108, Voice and Zklrt1iculation, Spring quarter, 1967, served as listeners. flPflease listeners are referred to in the report as listener S lejects . Listener training.--Listener subjects were trained 13)? a combination of lecture and tape recorded practice E32311nples designed to acquaint them with various cues to listen for in identifying each of the speaker characteristics. One t113511ning session took place with each group of trained J‘istener subjects. These sessions were conducted in their L18lilal classroom and occurred immediately before the actual" ‘tzeist situation. A 15 item practice test was administered at ‘:118 end of the training program. .Speaker characteristics.--The listener subjects were EaSREd to identify six characteristics of speakers. These characteristics and their various dimensions were: 11 1. Sex Male Female (rm 2. Age a. 20 - 30 years b. 40 - 50 years c. 60 - 70 years 3. Ethnic Group a. Negro b. Caucasian 4” Education Level a. Less than a high school graduate b. High school graduate c. College graduate 5. Physical Size a. Big b. Small These two terms were defined relative to standard norms for 17 to 19 year old males. "Big" speakers were equal to or larger than six feet in height and 180 pounds in wieght. "Small" speakers were equal to or smaller than five feet, six inches, in height and 121 pounds in weight. 6. Dialect Region a. Michigan (represented by speakers from Michigan) b. South (represented by speakers from Georgia and South Carolina) c. East (represented by speakers from eastern Massachusetts) .§g£§l_gg§§.—-The stimuli by which listener subjects made their identifications were tape recorded samples of 8bQech delivered to them by means of a loudspeaker. These Ques varied in three ways: (1) spontaneous speech versus Ireépared speech; (2) normal speech versus telephone speech; Q “<3 ( 3 ) duration . 12 ,Spontaneous auralficues.--These stimuli consisted of tape recorded samples of extemporaneous oral descriptions of the content of a written "neutral" paragraph. W.—-These stimuli consisted of tape recorded samples of oral readings of a written "neutral" Paragraph. NOrmal aural cues.--These stimuli consisted of samples of speech (both spontaneous and prepared) recorded on magnetic tape by means of a microphone and tape recorder in the usual manner and played back to the listener subjects unchanged in any way. Telephone aural cues.-—These stimuli consisted of samples of speech (both spontaneous and prepared) recorded on magnetic tape from a telephone receiver as the speakers ta:Llied into a second telephone during a typical station to s ~tation call . .D_urati_.Lon of aural cues.--The length of the speech s at“ples varied across all spontaneous-prepared and normal- t: Q1ephone conditions. Aural cue durations were: (1) five 8 SQ<>nds: (2) ten seconds; and (3) 15 seconds. Limitations of the Study Several limitations were built into the study, most Q E which occurred because of facilitative purposes. That is, they were introduced in order to make the study feasible. 13 The first of these limitations involved the speaker charac- teristics studied. Quite obviously, the six characteristics selected for the present study did not exhaust all possibili- ties with regard to differentiating between speakers on the basis of aural cues. These six, however, were of interest to the experimenter and did possess the additional qualification of being measurable in the environment in which the study was CC>r1c3ucted . Another limitation of the study had implications in terms of the applicability of the results. This involved the manner by which the listener subjects received their As indicated above, these cues were supplied by allI‘al cues . This limitation, magnetic tape as opposed to live voice. however, is of minimal concern due to the quality of instru- me IItation used . Perhaps the most important limitation of the study inVOlved the relationships among the speaker characteristics as different speakers were employed for each characteristic. Ideally, the same set of speakers would have been employed with these speakers being matched appropriately over all six Q I‘taracteristics, i.e., sex, age, ethnic group, education and dialect region of the country. Had I. Q"e1, physical size, Eh ~ . . . . 13 been accomp11shed, the 1nterrelat1onsh1ps. of the speaker Q . 1laracteristics could have been studied, e.g., identifyIng ethnic Speaker's age g1ven a speC1f1c comb1nation of sex, SI‘Oup, physical size, etc. This limitation was introduced 14 into the study in the interest of obtaining a wide range of information. That is, the experimenter sacrificed a rela- tively complete investigation of a few speaker character— istics in order to obtain some knowledge about several Characteristics. This decision was based on two factors: (1) the relative absence of previous studies with regard to listener identification of the speaker characteristics in question and (2) because so little is known about listener identification of speaker characteristics, it was felt that, at this point, some general knowledge was necessary before Stluldy of the interrelationships between these factors could it appeared unreason- be realistically attempted. That is, able to the author to seek specific knowledge about how the pre sence or absence of varying degrees of one or several characteristics of a speaker affects the performance of listeners in identifying another characteristic of his speech if it is not known whether or not such an identifi- Further- Q . . at?-:I.on can be made in a general or gross manner. more, the important breadth of the present study would have een unfeasible if the depth referred to above had become a iufistion of immediate concern. Organization of the Report Chapter I has introduced and stated the problem in— Q3~uding the hypotheses, the importance of the study, and the d$finition of terms. 15 Chapter II consists of a comprehensive review of the previous literature with regard to listeners identifying characteristics of speakers, the significant aural cues to be considered under each speaker characteristic in question, listener training, and considerations in terms of the various Speech and duration conditions. Chapter III is concerned with the description and se— lection of the listener subjects, equipment used in record— ing the speech samples, editing and preparing the master tapas, and presentation of the master tapes to the listener S‘~-1]=>jects and the procedures employed. Chapter IV presents the results and discussion, and Chapter V consists of the summary and conclusions. CHAPTER II REVIEW OF THE LITERATURE Kramer reported that in a fantasy novel by Robert Graves, a man of the distant future asks a twentieth century Englishman, "Do I speak with correctitude?" "With great cor— "But without the modulations of rectitude," he is assured, l6 tone we English use to express, or disguise, our feelings." Krattler used this illustration to show that all of us not only use such modulations ourselves, but also make judgments about others' current feelings and attitudes as well as about more St‘-al>le personal characteristics such as those under study in the present investigation, partly on the basis of how they "sound" to us. Sullivan has stated that these "sound ac- coutpaniments suggest what is to be made of the verbal propo- sitions statedwl7 Whether or not we can interpret them cor- rectly, whether or not the speaker and listener would agree as to their significance, these "non-verbal" but nonetheless \ IE 16Ernest Kramer, "Judgment of Personal Characteristics 1;.er Nonverbal Properties of. Speech," Psychological Bulletin, 1": (July, 1963), pp. 408-420. N 17H. S. Sullivan, The Psychiatric Interview (New York: Qrton, 1954): P. 7. l6 l7 primarily vocal aspects of the exchange," play an important part in the perception of person.18 The first major experiment on impressions of persons 19 based on voice alone was performed by Pear. He analyzed over 4,000 reports from British radio listeners who had re- sPonded to questions about nine different readers they had heard on the air. Age and sex proved easiest to estimate correctly. An actor and a clergyman were most consistently identified correctly from among the nine professions repre- sel'lted. The highest leadership ratings were given to the Speakers whose voices were professionally important to them: an actor, a judge, and a clergyman. Birthplace of the Speakers was not guessed with significant accuracy. Certain errors in guessing a speaker's profession showed significant consistency, suggesting that some voices provide a stereo- type of a certain occupation even though this is not the actl-Ilal occupation of the speaker. Such "vocal stereotypes" have remained the most frequent finding in all studies of the relationship between voice and personality.20 The studies that followed this first investigation 0 an be placed under two major categories: those which \ 18Ibid., p. 5. 5i 19T. H. Pear, Voice and Personality (London: ‘16 Hall, 1931). p. 117. Chapman 20Ernest Kramer, "Personality Stereotypes in Voice: econsideration of the Data," Journal of Social Psychology, A R Lkll (April, 1963): PP- 247-251. 18 called for listener judgments from voice about relatively stable characteristics of an individual, and those which asked for these judgments about emotional or affective vari- ables which change over relatively short periods of time. For the most part, both kinds of judgments involve the sepa— ration of nonverbal aspects of the voice from the actual Words spoken as well as problems with regard to adequate in— dependent criteria for the traits being judged. Physical Characteristics of an individual may usually be easily measured; but aptitudes, interests, and personality traits have presented special problems. Such paper and pencil in- 1 and Bell22 frequently used Ventories as those by Benreuter 24 in Such studies, have according to McKelvey 3 and Tyler. many limitations of their own. Because of these limitations, KEarner felt that even if valid judgments with regard to apti— tudes, interests, etc., were made from the voice by listeners, the correlation between such judgments and scores from the inventories listed above might frequently be low just on \ 53 21R' Bernreuter, Per§9nalitY_lBV9nt0rY (Stanford: tanford University Press, 1931). 22H. M. Bell, Adjustment Inventory (Stanford: Stan— :EQrd University Press, 1934) . 23D. P. McKelvey, "Voice and Personality," Western W, XVII (1953), Pp. 91-94. 24Leona Tyler, "Test Review Number 77," The Fourth $ntal Measurements Yearbook, ed. 0. Buros (New Jersey: ryphon Press, 1953), pp. 27-33. 19 the basis of the shortcomings of these inventories themselves . 25 All six of the speaker characteristics of interest in the present study can be described as stable character- istics of an individual. The Speaker Characteristics Sex Following Pear's initial success, little or nothing has been done with regard to ascertaining listener perform- ance in judging sex of speakers. Apparently, this charac- tel‘istic has appeared to be too obvious. Evidence that differentiation between the sex of speakers can be made is 27,28 supported in a related manner by two studies by Altus. He found that readers, i.e., receptors of graphic language, can infer the sex of professional short story writers at a leVal better than what could be eXpected by chance on the baeis of only a few words at the beginning of each story. This result, in conjunction with the" evidence obtained by \ 25Kramer, "Personality Stereotypes in Voice: A Re— QQIlsideration of the Data." 26Pear, loc. cit. 27W. D. Altus, "Sexual Role, the Short Story, and the :Eiterfl' Journal Of PSLCEhOIOQL XLVII (January, 1959), pp. 37- {I 28W. D. Altus, "Inferring the Sex of an Author," ‘sépurnal of Psychology, XLVIII (October, 1959), pp. 215—218. 20 Pear,29 would seem to indicate that the language output of males and females, whether graphic or verbal in nature, may be identifiable in terms of the gender of the producers. In terms of objective studies, i.e., without listeners, the primary differentiating feature between the speech of males and females has been found to be vocal pitch. Snidecor found that median pitch levels for women's voices were ap— Proximately two-thirds of an octave above the median pitch levels of men's voices.3O Fisher, using the term "modal" Pitch, which she defined as "the note which is used most fre- c1118111213", " was more specific. She specified the modal pitch of adult males as C3 on the musical scale and the modal pitch of adult females as GB. Hanley and Thurman felt that in- fle<=tion and intonation as well as pitch level conveys infor- 32 mation about the speaker and his attitude. They used the term "habitual" pitch and also stated the male norm as C3. Ho“lever, these two authors listed the female pitch level as \ 29Pear, loc. cit. 30John C. Snidecor, "The Pitch and Duration Character— Kfraties of Superior Female Speakers during Oral Reading," wnal of Speech and Hearing Disorders, XVI (February, 1951), E:‘ED- 44-52. ( BlHilda B- Fisher, Improving Voice and Articulation BOston: Houghton Mifflin Co., 1966), p. 165. 3ZTheodore D. Hanley & wayne L. Thurman, Developing $1 Skills (New York: Holt, Rinehart and Winston, 1963), ~ 143. 21 34 G3, slightly higher than Fisher.33 Black and MOQre; stated The male speaker who might be selected to repre— sent all male voices would have an average pitch a little higher than C below middle C, or 128 waves per second. This is approximately an octave lower than the average pitch of the representative female voice, a little lower than 256 waves per second. Fletcher indicated that the pitch of the voice when Speaking the vowels varies with different individuals, corre— SFXJnding to about 90 cps for a very deep—voiced man and to abmnit 300 cps for a shrill—voiced woman.35 In agreement with Black and Moore,36 he felt that the average pitch used by a Wonuan is near middle C or 256 cps while that of a man is abcnat one octave lower. In addition, he charted the mean furuflamental frequency used by males and females for "pure Vowels." These are listed in Table 1.37 Hahn, et al. attributed these differences in vocal frequency to four factors pertaining to the human vocal folds: (l) length-~the pitch is lowered as the length in— cteases,- (2) thickness--the pitch is raised as the tension Increases; (3) tension—-the pitch is raised as the tension \ 33Fisher, loc. cit. .51 34JOhn W- Black & Wilbur E. Moore, Speech: Code, Tifgfiiningl and Communication (New YOrk: McGraw—Hill Book Co., 955). p. 49. ( 35Harvey FletCher:.§EfieCh 99d Hearingfig Communication 1‘Tew York: McGraw-Hill Book Co., 1955), p. 49. '—""""' 36Black and Moore, loc. cit. 37Fletcher, o . cit., p. 62. 22 Table 1. Mean fundamental frequency used by males and fe— males for "pure vowels." W ==-_..-== Sound Mean Fundamental Frequency Male Female u 140 cps 270 cps 1r 138 250 o 116 237 o 112 243 A 118 253 a 113 234 ar 110 231 123 232 a: 121 247 5% 131 239 e 125 235 I 137 253 i 136 252 \ Mearls'or "Normals" 125 CPS 244 CPS __‘___ increases; and (4) density--the pitch is lowered as the den- sity increases.38 They indicated that all four factors are irrvkalved in the differences between any two voices and ex— plain the lower pitch of men's voices, as compared with t1'1<>se of women and children, on the basis of longer and ‘:}licker vocal folds. Fisher felt that the differences in “nails and female pitch levels is due to vocal cord length.39 E311eestated that male vocal cords average between three— tillarters of one inch in length while those of females average \ 38E. Hahn et al., Basic Voice Training for Speech (lfiew YOrk: McGraw Hill Book Co., 1957), p. 57. 39Fisher, op. cit., p. 157. 23 from one—half to three—quarters of an inch in length. Hollien found that low pitched individuals exhibit longer vocal folds than do individuals with higher pitch levels, both between the sexes and within a sex. In a similar study, Hollien and Curtis observed that individuals with low pitch levels exhibit larger, more massive vocal folds than do individuals with higher pitch levels.4l TIfitch, as an attribute of vowel sounds, has usually been attributed to (and considered the psychological counter- Part of) the fundamental frequency in the complex periodic wave structure of speech tones. Harbold, however, 31199631365 that; the pitch attribute of speech might not be completely dependent upon the fundamental.42 He supported this con- ter“Zion by conducting an investigation of the pitch ratings of listeners for voiced and whispered vowels. His results inciicated that the listener judgments of the relative pitch of Vowels was not independent of the vowel itself. This was illustrated by his obtained high correlation between rank and order of voiced and whispered vowels which is impressive \ 4OH. Hollien, "Vocal Pitch Variation Related to Change in Vocal Fold Length," Journal of Speech and Hearing Research, J:31]: (June, 1960), pp. 150-156. 23‘: 41H. Hollien and James F. Curtis, "A Lamenographic elldy of Vocal Pitch, " Journal of Speech and Hearing Re- ‘§§\J§£g_, III (December, 1960), pp. 361-371. ‘VV}1 42George J. Harbold, "Pitch Ratings of Voiced and .ispered Vowels," Journal of the Acoustical Society of %. XXX (July, 1958), pp. 600-601. ‘ 24 in View of the fact that the latter speech-type does not con- tain the fundamental frequency ordinarily credited as a physi— cal basis for pitch interpretation. An investigation performed by Meyer-Eppler studied the relationship of vocal pitch and whispering in a different way.43 He stated that speakers can be understood without difficulty when they whisper and that this is true even in a tonal language such as Chinese where pitch is used to differ- entiate the meaning of various lexical items consisting of Otherwise identical groups of phons. Meyer-Eppler found tZhat: these necessary whispered "pitch variations" are ac- Complished in various ways. For example, in the [e] and [o] sounds, raising the "pitch" is done by increasing whispered intensity, thus filling the gaps in the higher spectral re- gions with noisy components and eventually broadening the formants above two kc to a less—sharply profiled, fricative— 1ike spectrum. That is, for these vowels, Spectral noise alterations are substituted for pitch alterations. Their formants do not change in the process even though the speaker intends "pitch" change and the listener perceives it. A study by Snidecor related this important differ- er‘ltiating characteristic of vocal pitch to an important factor with regard to speech cues in the present study; that Q35 extemporaneous (spontaneous) as opposed to prepared \ 43W. Meyer-Eppler, "Realization of Prosodic Features 1“ Whispered Speech," Journal of the Acoustical Society of erica, XXIX (January, 1957), pp. 104—106. 25 He found, using male speakers, that (reading) speech. readings were slightly higher in mean pitch (132 cps) than were extemporaneous speeches (120 cps). In addition, mean pitch variability was greater when the subjects were reading than when they were speaking as was the mean extent of in- iflections and up—ward shifts. Silverstein, et al., in their investigation of the intelligibility of male and female speakers, found another factor which may serve to help listeners differentiate be- tween the speech of males and females.45 They observed that previous authors had found that, in terms of speech, males and females could not be considered as coming from the same population--the primary differentiating variable having been vocal pitch. Their study, however, indicated that the same differentiation could be made on the basis of intelligibility Qf untrained male and female speakers in the presence of l-1<315.se. These authors found that untrained male speakers were $1gnificantly more intelligible under these conditions than L111trained females. Furthermore, the sex of the listeners §Qemed to have little or no effect upon the intelligibility Q3 these spakers. Their results also indicated that train- {:19 removed the obtained differences and that the primary 44Snidecor, loc. cit. h- 453' Silverstein, 232;” "The Relative Intelligi— llity of Male and Female Talkers," Journal of Educational W. xxx (November, 1953), pp. 418-428. 26 contributing factor in the improvement of the intelligibility of the females after training was increased syllable duration. This combination of results is consistent with those 46 of Snidecor in terms of the reading rate. This author, in comparing his findings for women readers to Pronovost's with 47 found that, while the reading rate of males male readers, was more rapid, the mean phonation length per unit, i.e., Per syllable, word, sentence, or paragraph, was shorter for females. That is, a larger proportion of the reading time for females was spent in empty pauses between phonations while the male readers consumed a larger proportion of the reading time with actual phonation. Using his own results 48 with male readers, with female readers and Pronovost's Snidecor49 computed the mean ratio of phonated time to total reading time. These figures were 0.63 and 0.75, respective- ly, for the female and male readers and give numerical sup- port to the discussion immediately above. The conclusion Grawn from this portion of Snidecor‘s study supports the re- Suits of Silverstein, et al.50 with regard to intelligibility \ 6 . . Snidecor, loc. c1t. 4 . ‘l: . . 7Wilbert L. Pronovost, "An Experimental Study for De- emlnlng Natural and Habitual Pitch," Speech Monographs, IX (1947-): pp- 111—123. 48Ibid. 49 . . Snidecor, 10c. Cit. 0 . , . , . Silverstein, et al., loc. Clt.- 27 differences. That is, females tend to use shorter speech unit durations than males. Several studies with regard to verbal content have been made which point out the differences in male and female conversations. Moore51 strolled along Broadway in the early evening noting~ fragments of conversations he overheard and later classified these to show the predominant interests of the sexes as indicated by informal verbal expression. In doing this, he assumed that spontaneous conversation would, to a considerable extent follow the lead of a person's funda- mental interests and enthusiasms. He found that men con- Versed most frequently about business, money, and amusement While women most often discussed clothes and decoration. . Landis and Burtt52 conducted a similar study in Columbus, Ohio, and obtained results similar to Moore's. They found ' that 49 per cent of male conversations centered around busi- Iless topics while female conversations were most frequently 3190‘“ men (22 per cent), clothes (19 per cent), or other Women (15 per- cent). These authors suggested that the combi- rlation of the categories "men" and "women" into one category whiCh they would label "people" would yield a total of 37 per Qent and lend support to the notion that women are more \ 51 - - Henry T. Moore, "Further Data Concerning Sex Differ- Q“€33," Journal of Abnormal and Social Psychology: XVII (JUJ-Y' eptember, 1922), pp. 210-214. 52 g . N. H. Landis and H. E. Burtt, "A Study of Conver- “Ens," Journal iComparative Psychology, IV (February, 924): pp. 81-89. 28 interested in "people" than "things." Landis 3 is another author who investigated this topic, this time in‘London. His results were virtually identical to those of Moore54 and Landis and Burtt.55 Stoke and West took exception to the re- sults of these four authors and to the method by which they were obtained.56 They repeated the measures of conversation- al interests but attempted to control for occupation and age (using only college students), time of the conversations (night), sex of the persons involved (either all male or all female), place of the conversations (residences), social factors (fraternity and sorority members as opposed to non- affiliate college students), and race (using only Caucasians). Students, unknown to their fellow conversants, acted as re— corders during "bull sessions." These authors concluded that when the above factors were controlled, the sex differ- ences between conversational topics were rather small. Their results indicated that men still showed more interest in things and in sports while women showed more interest in Personalities, cultural topics, and social items. Stoke and X 53C. Landis, "National Differences in Conversations," ‘30urnal of Abnormal and Social Psychology, XXI (January- Imarch, 1927), pp. 354—357. 5 4H- T. Moore, loc. cit. 55 . . Landis and Burtt, loc. Cit. 56 - ' Stuart M. Stoke & Elmer D. West, "Sex leferences 1n Conversational Interests," Journal of Social Psychology, II (February, 1931), pp. 120-126. 29 West felt that other differences, as indicated by the pre- vious studies, were traceable to environmental influences. Carlson, Cook, and stromberg,57 in turn, took exception to the results of Stoke and West58 and their criticism of the previous studies. These three authors felt that the Stoke and West study59 was not directly comparable to those of 61 Moore,60 Landis and Burtt, and Landis,62 because it used only college students as subjects. Carlson, Cook, and Strom- 63 with the ex- berg used the same controls as Stoke and West cePtion of age and found what they described as "clear sex differences" in the topics of money, business, and sports for men and other women and clothes for women.64 An investigation performed by Gleser, Gottschalk, and Watkins also falls within this area of interest.65 They S7J- Spencer Carlson, Stuart W. Cook, and Elroy L. Stromberg, "Sex Differences in Conversation," Journal of Ap— lied Ps cholo , xx (1936). pp. 727—735. SBStoke and West, loc. cit. sgrpid. 60 H. T. Moore, loc. cit. lLandis and Burtt, loc. cit. 2Landis, loc. cit. 6 . 3Stoke and west, loc. c1t. 4Carlson, Cook, and Stromberg, loc. cit. , 65C’voldine C. Gleser, Louis A. Gottschalk, and John Watkins, "The Relationship of Sex and Intelligence to Choice Of.w?rds= A Normative Study of Verbal Behavior,". Journal of Clinical Ps cholo , XV (April, 1959), pp. 182-191. 3O conducted a study of the relationship of sex and intelli- gence to choice of words in a free—speech situation and found that females used a significantly higher percentage of words implying feeling, emotion, or motivation. They also made significantly more references to self and used more auxiliary words and negations. On the other hand, females used a relatively smaller percentage of words implying time, sPace, or quality. They also used a smaller percentage of words referring to destructive action than did men. In a similar kind of investigation, Penny studied the age and sex differences in the motivation toward conver- 66 He found that the conver- sation for teenage children. sations of boys of this age tended to be motivated by infor- national reasons while girls were motivated to converse for narcissistic reasons. Another area of investigation with regard to differ- entiating between the aural characteristics of male and fe- Inale Speech has been in terms of speech power. Fletcher Stated that speech power is measured in microwatts and is rated in terms of the intensity level of speech produced at one meter distance from the lips and directly in front of the s"Peakerfi'7 He reported a study performed by Dunn and ; 66 . . .R- Penny, "Age and Sex Differences in Motivational Orientation to the Communicative Act," Child Development, XXIX (December, 1958), pp. 163-171. 67 Fletcher, o . cit., p. 52. 31 White who measured the total speech power used by each of six men and five women.68 The average for the men was 34 nficrowatts (equivalent to 66.4 dB) while the women averaged 18 microwatts (equivalent to 63.4 dB). The total ranges for tie male and female speakers, respectively, were ten to 90 ndcrowatts and eight to 55 microwatts. Fletcher interpreted these findings to indicate that men and women's voices are Quite alike in.terms of the frequency-power distribution of SPGGCh-69 However, he stated further that tests with a sound level analyzer indicated that the intensities or speech power Of the components carried by frequencies above 3,000 cps Were definitely greater for women's voices than for men's. He felt that this seemed to indicate that women gave greater emphasis to siblilant sounds.7O A further differentiating characteristic, variability Cnfvocal pitch, has been studied by Snidecor,71 He found that the voices of women were less variable in vocal pitch than those of men. This particular finding refers to both ‘the EXtent of pitch variation when it occurs and to the rate Of Pitch Changes‘per unit of time. In addition, Dimitrovsky, in her doctoral disser— tation, found that male and female speakers, while uttering ‘— 68 . Ibid., p. 75. N 69 . Ibid., p. 78. 70 . Ibid. 1 . . Snidecor, loc. Cit. 32 neutral phrases, differed in the degree to which they were able to communicate specific emotional feelings to children.72 Her male speakers, as indicated by listener identifications, communicated the feelings of "sadness" and "love" signifi- cantly better than did the female speakers. With regard to the emotional feeling of "anger," the situation was reversed. On the basis of the investigations reviewed above, it appears possible to specify nine characteristics by which differentiation between male and female speech might be made on the basis of both content and vocal cues. The most ob- vious of these, of course, is that of vocal pitch. Addition— al cues, however, may be provided by: (1) intelligibility: (2) syllable duration; (3) speech power: (4) variability 0f vocal pitch; (5) accuracy of communication of emotional feel- ings; (6) conversational topics: (7) word choice; and (8) motivation for conversing. The present study, however, at— 11"?!“Ptt-Z‘d to limit the cues available to the listeners to only the Vocal type. That is, cues that are "content-free" from the standpoint of providing clues as to the sex of the speaker on the basis of the last three items above (conver— sational topics, word choice, and motivation for conversing)- $3 indicated in Chapter I, and discussed more fully in Chapter III, the speech stimulus for making the spontaneous g 72 . . Lilly Sprecker Dimitrovsky, "The Ability to Identi- fy the Emotional Meaning of Vocal Expression at Successive Age Levels," figsegition Abstracts, XXIV (January, 1964), PP: 2983-2984. 33 (extemporaneous) speaker recordings was a written "neutral" paragraph. It is hoped that this method had the effect of controlling the verbal content of the speakers' extemporane- ous utterances in this situation, while at the, same time, not interferring with the strictly vocal content possibili— ties referred to above. Age Popular literature abounds in references to vocal changes associated with aging. An example is Jaque's de— scription of the sixth age of man in Shakespeare's As You Like It: "His big manly voice, turning again toward childish treble, pipes and whistles in his sound." In the profession- al literature, however, such references are sparse. Pear's radio listeners had little difficulty judging the ages Of nine readers they heard over the air.73 These results were supported by the findings of Allport and Cantril.7.4 and Herzog (as reported by Diehl)75 who also ob- tained identifications of speaker age on the basis of voice :by listeners with better than chance accuracy. However, the x 7 . 3Pear, loc. Clt. 74Gordon W. Allport and Hadley Cantrili, "Judging Personality from the Voice," JournajL of. Social PszohOJ-ogx, V (February, 1934), pp. 37-55. 75H- Herzog, "Stimme und Personlichkeit," Ztschr. f. PSY‘FhOI- cm (1933). pp. 300—369, cited by Charles F. Diehl, "VOice and Personality," Psychplogical and Psychiatric As- Mach and Hearirg, ed. Dominick A. Barbara (Spring- fleld’ Ill.: Charles C. Thomas, 1960), p. 180. 34 latter two of these three studies found a tendency for esti- mates of age to center in the thirties.76’7‘7 In a recent study, Ptacek and Sander tested the ability of ten graduate students in Speech Pathology and Audiology to differentiate the voices of young adults (under age 35) from those of older persons (over age 65).78 These differentiations were required under three successive listening conditions of de— czreasing difficulty: (1) a prolonged vowel; (2) a tape re- corded reading sample played backward; and (3) a tape re- corded reading sample played forward. In the first two of these three conditions, content or semantic cues were re— moved. In the third condition, tape recorded reading samples played forward, the listeners were offered verbal or content Ques as well as vocal cues. The results of the study indi- Qated that this latter listening condition enabled the sub- j acts to identify correctly which of the two age groups the speakers belonged to 99 percent of the time. While listen- ing to a prolonged vowel, the subjects correctly identified the age group of the speakers 78 percent of the time. Under the second condition, tape recorded reading samples played beekward, the subjects correctly identified the age group 0f \ 76Allport and Cantril, loc. cit. 77Herzog as cited by Diehl, loc. cit. 15 78.Paul H. Ptacek & Erik K. Sander, "Age Recognition rom Voice," Journal of gpeech and Hearing Research, IX (June. 1966). pp- 273-277. .9: u» b; .11 35 the speakers 87 percent of the time. The authors felt that the overwhelming success of the subjects in making these identifications precluded the need for statistics in ana-. lyzing the results.79 Following the identifications, the authors asked the listeners to indicate the cues they used in making their differentiations. The most frequent of these were: (1) rate of reading; (2) fluency; (3) voice Quality; (4) vocal pitch; and (5) vocal intensity. The listeners felt that the older speakers read slower than the Yo unger group. Objective measures made by the authors indi— Qated that the above age 65 group had a mean reading rate of :L 52 words per minute while the younger group read at a mean rate of 200 words per minute. In terms of the remaining EON? Perceptual cues, the subjects felt that the older Speakers were less fluent, had a hoarse voice quality with eVidence of frequent vocal strain and voice breaks, used a leer Pitch and less pitch variability and less vocal i ntensity , 80 These results are very interesting when they are con— sidered in conjunction with the results of various objective studies of the effects of aging upon voice and speech. Mysak, for example, investigated the pitch and duration Q 1Harm-"teristics of two groups of older males (ages 67-79 and 79 Q: Ibi 801161 Q: 36 and 80-92) and their middle aged sons (aged 30—62).81 He found that for this group the fundamental frequency of the voice tended to rise with age, pitch variability was in- creased with age, and reading and speaking rate was reduced as age increased. A significant contribution to this re- duction in reading and speaking rate was found to be an in— creased number of pauses by the older speakers. This latter Objective result supports both the subjective listener ratings and the objective measures made with regard to reading rate by Ptacek and Sander.82 It also supports the subjective obServations their listeners made with regard to fluency. However, Mysak's83 objective measures of vocal pitch and pitch variability are in direct opposition to the subjective observations of Ptacek and Sander's84 and Dinolt pointed out that degener- listeners. Bach, Lederer, ative changes in laryngeal muscles take place with advancing age and argued that these changes result in what they termed 85 On describing the senile voice, they .| . . Se nile VOice . " \ 81Edward D. Mysak, "Pitch and Duration Characteristics of Older Males," Journal of Speech and Hearing Research, IX (March, 1959). pp- 46-54. 82Ptacek and Sander, loc. cit. 83Mysak. 12<=_-_c_i£ 84Ptack and Sander, loc. cit. 85A. C. Bach, Frances C. Lederer and R. Dinolt, "Se- nile Changes in the Laryngeal Musculature," Archives <3ij Oto- W. XXXIV (July, 1941), pp. 47—56. 37 referred to "changes" in the pitch that render the voice monotonous, flat and occasionally shrill in older people." This statement, of course, is not consistent with those of Mysak who reported a trend toward greater pitch flexibility with advancing age.86 In terms of the subjective comments made by Ptacek and Sander's listeners, it agrees with one Portion, i.e., that older speakers speak with less pitch Variability, but seems opposed to these listeners' comments to the effect that older speakers use a lower vocal pitch leVel than do the younger speakers.87 In another objective study of speech as it is af- feeted by advancing age, McGlone and Hollien investigated the vocal pitch characteristics of aged women. Their re— sults provided no evidence that pitch levels vary signifi— c"ell'ltly with advancing age once pubescent voice change is com- plete. The mean pitch levels used by their speakers, two E3r<>ups of older women aged 65—78 and 80-94, did not rise with advancing age as did those for men of equivalent ages as reported by Mysak.89 McGlone and Hollien offered an ex- planation for this discrepancy.90 They felt that since the \ 86Mysak. 10.9_9_i_t 87Ptacek and Sander, loc. cit. 88Robert C. McGlone and Harry Hollien, "Vocal Pitch gheracteristics of Ages Women," Journal of §peech agd Hear- W. v1 (June, 1963). pp. 164-170. 89Mysak. _1_<.>s_-__<.=_i_t. 90McGlone and Hollien, loc. cit. 38 aruatomical changes in the female larynx are not as extensive at puberty as are those in men, degenerative changes may not have as great an effect on women's laryngeal structures in later life. Hence, the concomitant changes in pitch, if present at all, would not be as apparent as they are in men. Ptacek, g£_al., assessed the performance of younger adult (under age 40) and geriatric (over age 65) speakers on the following tasks in assessing phonatory and related changes with advanced age: (1) maximum pitch range,- (2) diadochokinesis; (3) maximum vowel intensity: (4) maximum Vowel duratiOn; (5) maximum intraoral breath pressure; and (6) vital capacity.91 Their investigation was designed pri- I“arily to explore certain changes associated with advanced Eigiei in the functioning of the respiratory, phonatory, and airticulatory apparatus subserving speech. The results indi- cated that on all of the above named tasks, the geriatric SELLlbjects showed significantly reduced scores when compared with those of the younger adults. These results are again generally consistant with the subjective statements by the juC'Ilges used by two of these authors in a previous study92 and with the objective results obtained by Mysak93 with re- gard to older males, and with the observation of Bach” '91PaulH.iPtacek,-e£ al., "Phonatory and Related Ch'E'ulges with Advanced Age," Journai of fipeech and Hew- W, Ix (September, 1966), pp- 353-360. 92Ptacek and Sander, loc. cit. 93Mysak, loc. cit. 39 Lederer and Dinolt94 with regard to anatomical and physio- logical changes in the larynx with advanced age. In addition to the studies reported above which were concerned with primarily the vocal aspects of speech as it is affected by the age of the speaker, there have been sever— al studies with regard to the content or verbal aspects. Bloomer, for example, believed that the geriatric speaker has difficulty in all forms of "verbal communication," a term which he felt implies for the aging a somewhat broader Conception of spoken and written language than is typically eu‘Ployed.95 He felt, that within this broad context, verbal behavior underlies most of our orientations in daily living-- and, it is in this sense that the communication problems of the aging reveal themselves. Bloomer felt that practically a~11 of our psychosocial activities are mediated through the “Se of one of the major linguistic modalities--speech, listening, writing, or reading, and in addition to these modalities, we must include a number of simple daily tasks, e '9. , telling time, making simple arithmetic calculations, and making change in the purchase of articles, all of which rec{mire a personal competency on the part of the individual \ 94Bach, Lederer and Dinolt, loc. cit. 95Harlan Bloomer, "Communication Problems among Aged Eighty Hospital Patients," Geriatrics, XV (April, 1960), pp ‘295. 40 to interpret the signs and symbols of life about him. In a survey of institutions for the aged, Bloomer found that the elderly person is much more prone to develop communicative disorders than is the younger adult.96 His results indicated that 45 per cent of the insti— tutionalized population he surveyed had some kind of communi— cative disorder. This incidence figure he contrasted with the usual two percent found in a random adult population and indicated, in addition, that these disorders usually differ in etiology, prognosis, and their effect upon the total psYcho-social adjustment of the person involved. The eti- ology according to Bloomer97 and Mitchell98 is usually re- lated to the degenerative process. Specifically, Bloomer found the following communicative problems in the population he surveyed:99 (1) aphasia associated with cerebrovascular acCidents; (2) verbal evidence of mental confusion; (3) dYSlalia and dysarthria; and (4) dysphonia associated with gel'leral physical debility, paresis, personality withdrawal Symptoms, or a psychosis such as schizophrenia. In addition, \ 0.. 96m Q: 971131 QBJOYCe Mitchell, "Speech and Language Impairment in 217‘: Older Patient," 9.2m XIII (July, 1958), pp. 467- 99Bloomer, loc. cit. 41 beth he and Mitchell100 reported that the high incidence of hearing loss in the geriatric patient (35 per cent as opposed to seven per. cent in a random adult population) is also a sig- nificant factor in the communicative problems of the aged. Investigations of vocabulary ability over the life SPEEII include works covering the period of later maturity. 101 102 Brown, Results range from vocabulary loss (Ackelsberg, Shakow, Dolkert, and Goldman,103 and Shakow and Goldman104) to vocabulary growth (Christian and Patterson some investigators have ' 5 and Soren- $011106) among older people: \ 1OOMitchell, 10C. cit. 101Sylvia A. Ackelsber, "Vocabulary and Mental De- ‘3ealiioration in Senile Psychosis," Journal of Abnormal and —-S1$Eial Psychology, XXXIX (October, 1944), pp. 393—406. "A Study of Performance on a 10ZMary Matrow Brown, D?terioration Test Related to Quality of Vocabulary and '1S3idity," American Psychologist, III (October, 1944), pp. 93~406. 103D. Shakow, Marjorie B. Dolkart, and Rosaline 601d" man, "The Memory Function in Psychosis of the Aged," Dis- ‘SESEQEBrs of the Neryou§_§y§§gm, II (January, 1941), pp. 3—8. 104D. Shakow & Rosaline Goldman, "The Effect Of Age on S"lanf'ord—Binet Vocabulary Scores of Adults,’' Journal of Edu- 1 Psychology, XXIX (April, 1938), pp. 241-256. M 105Alice M. Christian and Donald G. Patterson, "GrOWth of Vocabulary in Later Maturity, " Journal of Psychology, I (1936), pp. 167-169. lOGH. Sorenson, "Mental Ability over a Wide Range of Adult; Ages," Journal of Applied Psychology. XVII (1938), pp 729~741.y _ . 42 reported no change (Fox,107 Hunt, et al.,108 and Fox and Birren‘l'og)‘. These latter authors have been particularly critical of the studies which have shown vocabulary changes with increasing age.110 They found no relation between vo- cabulary size and either age or length of institutionali+ zation for a population of persons over 40 years of age. In addition, they found no significant sex differences in vo— cabulary size at any age within this population when edu- cational levels were equated. An investigation by Ricks, h'O‘Mever, of the age and vocabulary test performance of the age groups 25—29, 45-49, 70-74, and 75 and above, contra- diCts these findings. Ricks used the vocabulary subtest of the Wechsler Adult Intelligence Scale and found that O.lcier adults averaged significantly lower in the use of syn- 011171118 in defining words, averaged significantly higher in te11‘1'ns of defining words by means of "use and description, " \ . 107Charlotte Fox, "Vocabulary Ability in Later Matur- ity, " Journal of Educationarl Psychology, XXXVIII (December, 1947 )," pp. 482—492. 108w. A. Hunt, at al., "The Clinical Possibilities of an Abreviated Individual Intelligence Test," Journal of Clini- ~Q&Psychology, XII (February, 1948), pp. 171-173. 1‘ 109Charlotte Fox and James E. Birren, "Some Factors ffe<=ting Vocabulary Size in Later Maturity: Age, Education, an Length of Institutionalization," Journal of Gerontolggy, IV (January, 1949), pp. 19-26. 110Ibid. 111James H. Ricks, "Age and Vocabulary Test Perfor— A Qualitative Analysis of the Responses of Adults," Wnygtrigs, XIX (July, 1958), p. 182. 43 i.e., interior definitions, and scored significantly higher in defining words erroniously than did the younger groups. This author interpreted these results as challenging the hy- pothesis that the vocabulary score of an older person can be regarded as reflecting performance in which little or no Rather, "they change has occurred since early adulthood. 112 support the contention that impairment does occur." In a related study, Birren, Riegel, and Robbin in— vestigated the speed of word associations made by two groups of subjects, one aged l8-33 years and another aged 60-80 Years.113 These subjects were comparable in the number of Years of completed formal education and vocabulary size. The obtained data showed an age difference in mean latency Of word associations of about 0.6 seconds per word, or about 40 pe r: centflonger for the older group. In summary, it appears, from evidence cited above, that. a speaker's age can be judged with better than chance accuracy from his voice. However, the specific vocal cues that listeners have reported using is not clearly consistent Wlth objective measures that have been made attempting to d‘l'fferentiate between the vocal aspects of speech as a fu . . . . he"zion of age. Moreover, this same incon51stency of \ 112Ibid. "A 1'1'3James E. Birren, K. F. Riegel, and J. S. Robbin, b 9e Deficiencies in Continuous Word Associations Measured y Speech Recordings," Journal of Gerontology, XVIII (Janu- ary, 1962), pp. 95-96. 44 reported results is evident from the standpoint of content or verbal differences. Ethnic Group Investigations as to the ability of listeners to identify the ethnic group of speakers have been sparse. The same can be said of objective studies in which an attempt is made to specify various vocal or conversational cues by which to make such differentiations. One study of the sub- jective evaluations of listeners investigated whether or not Jewish and Gentile subjects exhibited differential reactions to the same speaker when he was using English with a Jewish 114 accent. Results indicated that persons with an accented voice were devaluated on height, good looks, and leadership by both listener groups. The Jewish subjects tended to evaluate the accented voice more favorably on sense of humor, entertainingness, and kindness. However the Gentiles did not evaluate the accented voice as being more favorable on any trait, These results offer support to a statement by All- port and Kramer.llS —\ al 114M' Ainsfield, N- 3095, and W. Lambert, "Evaluation— no 1Reactions to Accented English Speech," Journal of Ab- ‘—£lfl£aI Psychology, va (1962), pp. 223-231. 115Gordon W. Allport and Bernard M. Kramer, "Some Riggs of Prejudice," Journal of Psychology, XXII (1946): PP- 45 The question of racial identity is of small im- portance to the person free from prejudice, yet it is of considerable importance to the bigot, and for this reason, the bigot apparently learns to observe and interpret both facial features and expressive be- havior so that he can more swiftly spot his "enemy." Cohen and Starkweather, in a study investigating vfluether subjects with high scores in ethnocentric prejudices were more sensitive to the cues provided by the expressive behavior of various ethnic groups, found that the vocal por- tioaa of speech, i.e., with verbal content removed by filter— ing,. carries information which allows English speaking sub- jGCth to distinguish between English and non-English speech tha degree which exceeded that which would have been ex— Pected by chance.116 With specific reference to the ethnic group of interest in the present study, Wise has stated, "Offhand, one“(mauld say that southern Negro speech is of all the dia- leCtSB the most strikingly different from cultivated English.117 He felt, in addition, that many of these differ- encegs are not articulatory or enunciatory and are not there- fore: easily reducible to phonetic terms. With regard to ‘flnulCZiation, Wise wrote that enunciation "tends to be lax amd (careleSS-"lle He also pointed out four primary ways in —\ the 116Alan Cohen and John A. Starkweather, "Vocal Cues to <3m> JEdentification of Language," American Journal of psy_ :Lc> , LXXIV (March, 1961), pp. 90-93. s 1170' M' Wise, "Negro Dialect,"_Quarterly Journal of —B§§EE§, XIX (NoveMber, 1933), pp. 522—523, 118Ibid. 46 which the Southern Negro differs in articulation from "culti— vated English."119 (1) vocabulary ("various word substi- tutions"): (2) declension; (3) intonation ("one of the out- :standing characteristics of Negro speech“); and (4) tongue gilacement ("the characterizing element of Negro speech"). 131 terms of the last point, Wise felt that Negroes consis— terrtly use a more elevated tongue placement than is usual in cultured English. He felt that this articulatory difference accc>unted for what he described as "a characteristic Negro VOice quality" in which vowels "become somewhat open and 120 While he felt that this tongue Placement may free:-" Possibly be a heritage from the original African speech, Wise did not believe that it resulted from any peculiar phy— sical formation of Negro resonance cavities. To support this, he FK>:Lnted out that Negroes reared among a majority of Cau- casians, e.g., Negroes living in the North, "have nothing of 121 With specific reference to 122 the Negro voice quality." this iphenomenon, Wise stated: VThen in relatively complete isolation from large rlumbers of their race who are either fresh from the aiouth or who have preserved their southern voices y reason of social segregation and forced associ- aition with their own kind, these northern Negroes cIan not be distinguished from White people through -\ 47 any voice characteristic. The test of the telephone in northern cities repeatedly proves this. An additional differentiating characteristic of southern Negro speech was felt to be a limited amount of {nasal resonance in comparison with Caucasian speech. In an objective investigation of the characteristics of lflegro speech, Hollien and Malick123 duplicated a previous stuciy of the vocal pitch characteristics of northern CaucaSian maless by Curry124 using southern Negroes. Results indicated that: for 18 year olds, i.e., post adolescents, southernNegro males display a lower median pitch level than do northern Cauca sian males. These pitch levels for the perspective groups and their approximate equivalent musical tones were 124-4L (cps (B3) for Negro speakers and 137.1 cps (C3), for the Caucasian speakers . 125 This obtained difference in pitch level is consis- 126 ‘0 tent Vvith findings of Bosohoff. He reported that the laryrhges of his Negro cadavers were somewhat larger on the average than those of his Caucasian cadavers. _\ Ch 123H. Hollien and Ellen Malick, "Adolescent Voice (MSn‘Qe in Southern Negro Males, " __S£eech Monogrj‘phs, xxxx reh, 1962), pp. 53-58. leg 1243: T: Curry, "The Pitch Characteristics of the Ado- Qe at Male Voice, " ipeech Monogrgphs, VII (March, 1940), pp‘ 48-62. 125Hollien and Malick, loc. cit. 1269. H. Boshoff, "The Anatomy of the South African Hegtt> LaryngeS,"‘§9uth African Journglyof Medical_§giggg§, X (February, 1945), pp. 35-40. 48 In another objective study, this time with regard to more general language functioning, Carson and Rabin investi- gated verbal comprehension and communication in Negro and Caucasian children from 9.5-ll.5 years of age.127 Their sub- jects were matched on age, grade in school, sex, and verbal comprehension as measured by the Full Range Picture Vocabu- lary Test. The results indicated that northern Caucasian children, even though matched as described above, manifest higher levels of verbal communication than northern Negro Children, and, in turn, the latter group manifest higher levels of verbal communication than southern Negro children. The criterion measure for the verbal communication scores were the WISC- vocabulary subtest and a specially constructed verbal version of the Full Range Picture Vocabulary Test. It appears possible, on the basis of the studies re- VieWed above, to make several tentative observations with re- gard to listener differentiation between the Caucasian and Negro ethnic groups. First of all, it has been shown that listeners can differentiate between (1) English and non- Englj-sh speakers even though verbal content has been ' hen . 127Arnold S. Carson and A. L Rabin, "Verbal Compre- JOuslon and Communication in Negro and White Children," 51 thal 0f Educatifllel Psychglogv, LI (April, 1960), pp. 47- 49 removed,128 and (2) Jewish accented speech and "pure" English speech.129 Secondly, several possible characteristics by which listeners might differentiate between Negro and Caucasian speech have been suggested. Among these are vocabulary, de- clension usage, vocal intonation, voice quality, nazalization, vocal pitch, and general verbal conununicative ability. De- spite these cues, however, and the general successes pointed out above, there is no evidence that listeners can actually In make the distinction in question in the present study. fact, it has been suggested that the task is not possible where northern residents of the two ethnic groups are Conce :rned . 130 Educa tion Not one study was encountered which investigated Whether or not listeners could identify the education levels 0f Speakers on the basis of aural cues. However Fay and Midd:Leton have investigated the identificability of related Chara ctistics from the voice as transmitted over a public addre ss system. In one of the studies, these authors asked 84 undergraduate college students to judge the intelligence °f th ird and fourth year high school boys on the basis of a _\ 128 . Cohen and Starkweather, loc. Cit. 129Ainsfield, Bogs, and Lambert, loc. cit. 130Wise, loc. cit. 50 thirty second oral presentation of "interesting reading ma- terial."131 The speakers were selected on the basis of 1.0. as measured by the Terman Group Test of Mental Ability and were matched in oral reading ability. The results indicated that the listeners tended to judge the 1.0. group to which speakers belonged relatively well but not to a degree that was significantly better than chance. In the second study referred to above, Fay and Middleton tried to determine whether or not leadership ability could be accurately judged from the transmitted voice.132 Fifteen speakers, all from one fraternity, were used. Ten seniors who had known each of the speakers for a period of at least six months rated them for leadership qualities on a seven point scale (minus three to plus three). The interjudge reliability of these ratings was + .91. The mean for each speaker was used as his criterion. Twenty-eight undergraduates then rated leader- ship on the same scale on the basis of hearing the speakers read a 30 second sample of a selection from a magazine. The Obtained correlation between each speaker's leadership cri- terion and his rating by listeners was + .08. Fay and Middle— ton concluded that the performance of listeners in accurately —\ ‘ Int 131Paul 5" Fay and Warren C. Middleton, "Judgment of dreelligence from the Voice as transmitted over a Public Ad- 33 System, " §ociometgb III (April, 1940), pp. 186-191. L 132Paul F' Fay and Warren C. Middleton, "Judgment of peadership from theTransmitted Voice,“ Jourfil of §oci§_]__ -§y°\h°_129x: XVII (1943), pp. 99-102. 51 judging leadership abilities from voice, relative to a cri- terion, was about what would be expected by chance. In another study relating intelligence to oral ex- pressive behavior, Gleser,: Gottschalk, and Watkins investi- gated the relationship of this factor and sex to choice of1x words in a free-speech situation.]‘33 They found that 1.0. differences occurred primarily among gramatical categories. The proportional use of adjectives (including articles), articles alone, prepositions, auxiliary words, and words re— ferring to quality or state of being increased sharply with increased intelligence. Interestingly enough from the stand— . Point of the present study, Gleser, Gottschalk, and Watkins Su9968ted that their obtained differences in the verbal be— havior of their subjects might well not have been due to in- telligence differences alone.]'34 They performed a product- moment correlation between the intelligence and educational level of each of the subjects and obtained an 3; equal to * '72- These authors felt, on the basis of the evident Strength of this relationship, that the obtained differences in verbal behavior might well reflect differences in edu- Qational background instead of, or as well as, I-Q- level. Johnson has also done an objective study within the general area of speech characteristics of persons of differ— Qnt educational and experience backgrounds. He found that \\ 133 Gleser, Gottschalk, and Watkins, loc. cit. 13 4Ibid. .u i . neo- Irv-o Q'AI if: b. in». ' .v, ~ ‘ In. or.‘ ‘ I“ “'ua.‘ ”V..- R 1 1“ .‘-L 52 experienced business executives scored significantly higher in English vocabulary than did recent college graduates, both in business and general academic areas, and non—college graduates.135 Johnson concluded that a large vocabulary is a characteristic of a high ranking business executive and that this characteristic is independent of past educational experience. In another vocabulary study, Schulman and Havinghurst investigated this factor in relation to social status in what they described as "a representative midwestern com- 136 They found that, on the average, children of munity." higher social status made higher mean scores on the Seashore— Eckerson English Recognition Vocabulary Test than did children of lower status. In a similar study, Rosenthal in- vestigated whether or not differences existed in the language behavior of children who belong to high and low sociometric 137 groups. His subjects were second grade children who were matched on intelligence, sex and age. Language samples were 135A. P. Johnson, "A Study of the English Vocabulary Scores of 75 Executives," Technical Report of the Human Engineering Laboratory, No 2 (1935), p. 16. 136Mary Jean Schulman and Robert J. Havinghurst, "Re- lations Between Ability and Social Status in a Midwestern Community. IV: Size of Vocabulary," Journal of Educational Psychology, XXXVIII (Nevember, 1947), PP. 437-442. 138Ered Rosenthal, "Some Relationships Between Socio- Position and Language Structure of Young Children," Journal of Educational Psychology, XLVII (December, 1957), pp. 483- 497. a‘n 5“. O~-.‘ by“: ._~‘ sun . v 7:“ h‘.. h \' " V ”c... . ":Vh b.~ " o v.3 s. ‘ ~n x“... n; 11'.) ‘A ‘U 53 collected by means of conversations with an adult. They were analyzed on the basis of: (1) structure; (2) method; and (3) communicative value. The sociometric status of each subject was determined by means of sociographic methods. Re- sults indicated that children of high sociometric status tend toward language characterized by a high percentage of mean- ingful communication. Rosenthal found that their communi- cation took place in longer units and that they manifested shorter and fewer aspects of what he termed "noise“ in language.138 These subjects also used a larger number of verbs relative to adjectives and a greater variety of words. On the other hand, his children of low sociometric status tended toward language characterized by a smaller percentage of meaningful communication. Their communication was also in shorter units. In addition, the low sociometric group used a greater number of "noise“ aspects, more adjectives related to verbs and less variety in the type of words. In studies of social class and speech systems, Bern- stein has suggested three main areas of linguistic difference between middle class and lower working class boys. 1. In working class samples particular relationships were found between verbal and non—verbal I.Q. scores; the verbal scores tended to be depressed in relation lBBIbid. S4 to the scores obtained at the higher ranges of the non-verbal test.139 2. Class differences in terms of modes of verbal ex- pression were found to exist on both a lexical and grammatical basis in the sense that working class speech is selected from a more narrow range of al- ternative structures.140 3. Two different speech systems were considered to be generated through the use of different verbal planning procedures. These differences existed not only in syntatic selection but also in mean pause duration per word.141 Bernstein's theoretical basis for these studies was the postulated existances of two linguistic codes: re- stricted and elaborated, the lower working class tending to be confined to the former with the middle class group switch- 142 ing from one to the other according to the context. Ac— cording to this author, a restricted code is generated by a 139Basil BernstEin,"Language and Social Class," British Journal of §9cioloqv, XI (1960), pp. 271-276. 140Basil'Bernstein, "Social Class Linguistic Codes and Grammatical Elements," Language_gnd Speech, V (1962), pp. 221- 240. 141Basil Bernstein, "Linguistic Codes, Hesitation Phe- nomena, and Intelligence," Language and Speech, V (1962), pp. 31-45 ' 142Bernstein, "Social Class Linguistic Codes and Grammatical Elements." 55 form of social relationship based upon a range of closely shared identifications held by the members. An elaborated code is generated by a form of social relationship which does not necessarily presuppose such shared identifications with the consequence that much less is taken for granted. Each code regulates the "area of discretion" available to a speaker and so differently constrains the limits of verbal 143 The community of like interests underlying a behavior. restricted code removes the need for intent to be verbally elaborated and made explicit. The effect of this on speech is to simplify the structural alternatives used to organize meaning and restrict the range of word choice. A restricted code can arise at any point in society where its conditions may be fulfilled but the special case of its use by the work- ing class is that the speaker is limited to this code. An elaborated code is the realm of the middle class; a middle class individual simply has access to the two codes. The lower class individual has access to only one. As indicated above, Bernstein's studies have shown that the two class groups are differentially oriented in both their structural selections toward choices. Further- more, when the subjects representing the two social groups are subdivided within each group on the basis of intelli- gence, the differences between the working class and middle class still hold up. Bernstein felt that this fact indicated Ibid. 56 that the two social groups are oriented differently toward language independent of intelligence.144 Following Bernstein's theoretical structure and re- search findings, Lawton investigated the characteristics of 145 His re- working class and middle class written language. sults indicated that the characteristics of Bernstein's re- stricted code carry over into writing "to a far greater ex- tent than might be expected."146 The consistency of results showed that inter-class differences exist not only in vocabu- lary, but also in whole classes of words (adjectives, ad— verbs, pronouns), and in structure (passive verb forms and types of subordination). In each case, working class sub-_ jects, restricted code users, select words and structure from a narrower choice of alternatives. In terms of summarizing the information relative to speech characteristics available to listeners in differenti- ating betweenspeakers of differing educational backgrounds, the above reviewed studies are valuable only to the degree that intelligence, leadership qualities, vocational success, and social status are related to this factor. The : 144Bernstein,."Linguistic Codes, Hesitation, Phenomena, and Intelligence." 145Denes Lawton, "Social Class Differences in Language Development: A Study of Some Samples of written WOrk," 'Lgnguage_and Speech, XI (1963), pp. 120-143. 1461bid. 57 The relationship between intelligence and educational level has already been pointed out.147 In terms of relating leadership qualities with edu- cation, Hollingworth has pointed out that "one indispensable quality of leadership" is extraordinary intelligence.148 This statement has been supported by the results of a study by Reynolds who investigated the characteristics of leaders and non-leaders among senior high school students.149 He found that students designated as leaders among a group tended to be superior to non-leaders in achievement, intelli- gence, and personality rating. With regard to the relationship between vocational success and educational level, Glick and Miller found that both the average annual incomes and the projected life in— comes of adult males increases with increased formal edu— cational experience.150 They found, in 1949, that the annual income from a male with eight years of formal schooling was $3,112. By contrast, the yearly income for the same age male 147Gleser, Gottachalk, and Watkins, loc. cit. 148L. S. Hollingworth, "What we Know About the Early Selection and Training of Leaders," Teachers College Record, XL (April, 1939), PP. 575-592. 149F. G. Reynolds, "Factors of Leadership Among Seniors Of Central High School, Tulsa, Oklahoma," Journgl of Edu- sgtional Research, XXXVII (January, 1964), pp. 356-361. 150Paul C. Glick and Herman P. Miller, "Education level and Parental Income," Epilege Board Review, No. 32 (1957), PP- 29-32. 58 with a high school diploma was $4,519, while this figure for the college graduate was $7,907. While these figures in themselves carry some additional implications in terms of the relationship between social status and educational attain- ment, the number of studies showing amore direct positive re- lationship between these two factors are numerous. Blake, for example, investigated the relationship between socio— economic status (as measured by the Sims Score Card for Socio-economic Status when the subjects were 12 years of age) and scholastic apptitude and intelligence (as measured by the Miller Analogies Form G, the Cattell Culture-Free In- telligence Test, and the American Council on Education's Psychological Examination, 1941, when the subjects were in college).151 Significant correlations were obtained between all measures. In addition, educational and occupational levels of the parents of the subjects showed a "pronounced positive relationship with high intelligence test scores" made by their children. Blake concluded that the residual effects of childhood socio—economic status were reflected in high school scholastic apptitude and intelligence test per- formance of his young adult subjects.152 In another study, 151Robert R. Blake, "The Relationship Between Child- hood Environment and the Scholastic Aptitude and Intelligence of Adults," W XXIX (FebruarY: 1949), pp. 37— 41. 1”mm. 59 Sewell, Haller, and Straus found that the educational and oc- cupational aspirations of college students from non-farm families was a function of the social status of their fami- lies, over and above the factor of intelligence.153 This relationship between intelligence test performance and social class status has been explored by several authors. Britton found significant correlations (ranging from + .42 to + .53) between the social status of 102 boys and 130 girls and their scores on the Hennon—Nelson, the Otis Aspha Verbal and NOn-Verbal and the Kuhlmann-Anderson tests of intelligence.154 In addition, Havinghurst has pointed out that when the upper middle class is compared with the lower class, it becomes evident that there are systematic differences in the experi- ences of children of these groups that might be expected to result in different levels of performance on the ordinary in- telligence tests.155 He felt that these differences were Cultural in nature and that they fell into four areas: (1) cultural differences in home and family: (2) cultural differ— ences in neighborhood and community life: (3) cultural 153William H. Sewell, Archie O. Haller, and Murray A. Straus, "Social Status and Educational and Occupational AsPiration," Americgn SociologicalgReview, XXII (February, 1957), pp. 67-73. 154Joseph H. Britton, "Influence of Social Class upon PeTiformance on the Draw-a-Man Test," Journal of Educational Eilggglggy, xrv (January, 1954), pp- 44-51. 155Robert J. Havinghurst, "Culture and the 1.0.," Emits Unimsitv Stugies 3 Higher Education, No. 69 (1949), PP- 42-53. 6O differences in school experience: and (4) cultural differ— ences in motivation for test performance. Toward this end, Stenquist and Lorge have pointed out that both test-makers and test—users must be aware that measures of intelligence and social status are highly related.156 Finally, with regard to the social class-educational level relationship, Bernard found a positive correlation be- tween these two factors which he felt appeared to be the re— sult of the differences in the perception characteristics of 157 He felt that the members of the various social classes. these perceptual differences facilitate educational attain- ment in middle class youth but inhibit it in working class youth because schools emphasize the abilities manifestly de— veloped in the former group. Based on the illustrated relationships between each of these characteristics, i.e., intelligence, leadership, vo- cational success, and social status, and educational level or attainment, it would appear possible that discriminations be- tween members of different educational groups on the basis 0f aural cues might be made in terms of (l) grammatical cate- gories: (2) evident vocabulary size; (3) length of speech 156John L. Stenquist and Irving Lorge, "Implications of Intelligence and Cultural Differences:, As Seen byra Test- User: As Seen by a Test-Maker," Teachers College Record, LIV (January, 1953), pp. 184-193. 157B. Bernard, "Some Sociological Determinants of Per- ception: An Enquiry into Sub-Cultural Differences," British JOurnal o§_§ociology, IX (1958), pp. 159-174. 61 units: (4) frequency of reference to self; and (5) sentence structure. Physical Size From the literature it would appear that very little has been done relative to listener judgments of the physical characteristics of speakers. The same was found to be true in terms of possible differentiating aural features. In a pair of similar studies, Barta158 and Tursky 159 had listeners match the photographs of speakers to the voices they heard from behind a screen. In both studies the speakers were chosen in accord with Kretchmer body types. Each author reported results that indicated that the pyknic type was accurately evaluated. In both studies, the listeners' evaluations were least accurate in judging the athletic type. 160 had 2,700 radio listeners In another study, Herzog in Vienna judge the height and weight of radio speakers. She found that both items were judged "more accurately than 158B. Barta, Ausdruck der Personlichkeit in der Sprech— stimme und im Photogramm,".§rch. Ges. Psychol. XCIV (1935), pp. 501-570, cited by Charles F. Diehl, "Voice and Person- ality," Psychological and Psychiatric Aspectsiof Speech and Hearing, ed. Dominick Barbara (Springfield, 111.: Charles C. Thomas, 1960), p. 180. 159H. Tursky, "Zur Phenomenologie des Zuordnungsacktes zwischen Stimme und Bild des Sprechers," Wiener Dissertation, 1932, cited by Charles F. Diehl, "Voice and Personality," Psychological anngsychiatric Aspect of Speech and Hearing, ed. Dominick Barbara (Springfield, Ill: Charles Thomas, 1960), P. 180. 160Herzog, cited by Diehl, 10c. cit. 62 one might expect from the mere operation of chance." In an- other of the Fay and Middleton161 studies, this one involving listener judgements of Kretchmerion constitutional types based on the voice as transmitted over a public address sys- 162 and Tursky163 were tem, results similar to those of Barta found. All body types studies were found to be accurately identified beyond what was expected by chance. The authors felt that the pyknic and leptsomatic types were apparently the least difficult to judge. A study by McGehee,164 however, contradicts the re- sults obtained in the investigations reviewed above. He found poor results in listener estimations of height and weight on the basis of recorded oral readings. In addition, Diehll65 found that listeners were unable to identify the height and weight of 110 speakers from a college population with accuracy that exceeded chance expectations. He concluded 161Paul J. Fay and Warren C. Middleton, "Judgment of Kretschmerian Body Types from the Voice as Transmitted over a Public Address System," Journal of Socaal Psychology, XXI (January, 1940), pp. 151—162. 162Barta, cited by Diehl, 10c. cit. 163Tursky, cited by Diehl, loc. cit. 164E. McGehee, "An Experimental Study of Voice Recog- nition," Journal of General Psychology, XXXI (January, 1944), pp. 53-65. 165Charles F. Diehl, "Voice and Personality," Psycho- lpgical and Psychgatric Aspects of Speech andigearing, ed. Dominick Barbra (Springfield, 111.: Charles C. Thomas, 1960), p. 183. 63 that the "probability of judging height and weight from the recorded voice . . . appears to be extremely poor." As indicated above, the literature is sparse with regard to suggesting vocal features that might be useful in making differentiations between physical characteristics of speakers. For this reason, the writer has taken available literature and made what appeared to him to be logical specu- lations with specific reference to "big" and "small" speakers 166 has re- as defined in this study. For example, Fletcher ported that the average speech power used by men to 34 micro— watts while the speech power used by women averages 18 micro- watts. Applying this information to the characteristic of physical size, one might expect the same kind of difference, although not to the same degree, between very big and very small speakers. From a logical standpoint, a difference between big and small male speakers might also be expected in vocal pitch. Specifically, one might expect the big speakers to have lower pitch than the small speakers. This line of thinking is sup- 167 ported by the research findings of Hollien and Hollien and Curtiss168 who found that low pitched speakers exhibit longer vocal folds than do individuals with higher pitch levels. This was found to be true both between and within the sexes. 166Fletcher, op. cit., p. 76. 167Hollien, loc. cit. 168Hollien and Curtiss, loc. cit. 64 Further support is provided by Fisher who stated that pitch differences between males and females is due to the differ— ences in the length of their vocal folds.169 170 In addition, Hahn aE_al., mentioned three factors important in determin- ing vocal pitch that may be related to physical size: (1) length: (2) thickness: and (3) density. In terms of resonance, similar thinking can be fol— fowed. Big speakers would be expected to have slightly larger resonating cavities than small speakers, thus ampli- fying lower pitched tones than the resonating cavities of the latter group. Summarizing this section, there have been conflicting results with regard to listeners identifying the physical characteristics of speakers. In general, it appears that the more specific the identification requested, i.e., judging height and weight as opposed to body build, the less accurate the performance of listeners. In terms of differentiating vocal features, it has been suggested that big speakers may speak with more intensity and with a lower pitch than small speakers. Dialect Region The United States has traditionally been divided in— to three inclusive speech regions: (1) New England, (2) the 169Fisher, 0 . cit., p. 116. 17oHahn, et al., 02- cit., pp. 57—58. 65 South, and (3) the remainder of the country. The labels attached to the first two regions from the outset of interest in this area have been Eastern and Southern speech, re* spectively.171 Labeling of the third region has not been as consistent. The South is inclined to call it Northern speech, an Obvious misnomer as, geographically, New England also lies far north and, in terms of speech, portions of the region, e.g., northern Texas and California, lie far south. Other labels that were suggested are Midwestern speech and Western 172,173 speech. These terms were considered inappropriate for the same reasons. North-and-West was another label that was suggested174 and though it had merit in terms of being nearly consistent with the other geographical regions, East and South, it apparently did not catch on. The term that was finally accepted, General American speech, has the same geographical drawbacks as those labels described above in ad— dition to giving the false connotation that it applies to the whole North American Continent. 171Claude Merton Wise, Introduction to Phonetics (Englewood Cliffs, N.J.: Prentice—Hall, 1958), p. 171. l721bid. 173Albert C. Baugh, History of the English Languagg (New York: Appleton-Century, 1935), p. 446. 174Hans Kurath, "Dialectal Differences in Spoken English," Modern Philology, XXVI (May, 1928), pp. 386-393. 66 According to Wise,175 the first person to use this term was Windsor P. Doggett in the early 1920‘s. The oc- casion was an address by Doggett at the National Association of Teachers of Speech in New York. The three'all-inclusive speech regions of the United States are called, then, in order of size and population, General American, Southern, and Eastern.176 Wise177 defined these speech areas as follows. The Eastern dialect region is bounded on the west by the Green Mountains and the Connecticut River. On the north this re— gion is, of course, bounded by Canada, while on the east and south by the Atlantic Ocean and Long Island Sound. The boundary between the Southern speech region and the rest of the country begins on Delaware Bay at approximately the lati- tude of Dover and swings in a semicircle north of Baltimore and then southward to cross the Potomac a little below Harper's Ferry. Then it follows the Blue Ridge to its ex- tremity and swings westward. At this point, data breaks off. The General American region, as indicated above, in— cludes the remainder of the country. These descriptions are consistent with those contained in the Languistic Atl§§_9£ the United States.178 175Wise, op. cit., p. 172. 176Ibid. 177Wise, op. cit., p. 178. 178Hans Kurath, Linguistic Atlas of the United States (Providence, R.I.: Brown University Press, 1939—1943). 67 179 described the Eastern speech region Gray and Wise as including New England, and the city of New York, with some deviations. They also indicated that in some ways par- ticularly in word selection, the influence of the East af— fects the speech of upstate New York and the northern third of Pennsylvania. These authors stated that the Southern dia- lect is spoken in those states which formed the Confederacy, with certain additions (Kentucky, southern Delaware, and southeastern Maryland), and with certain subtractions (the southern Appalachian highlands and the highlands of Arkansas and Louisiana, and all of Texas except a relatively small southeastern portion). Many writers have discussed the differentiating characteristics of the three American English dialects with reference to pronunciation of the vowels. These sounds are discussed below under the conventional groupings. Front Vowels.—-For this group of sounds, Akin180 cited only the Eastern substitution of the sound [a] for [a:] used by the rest of the country in such words as "fad" and "hat." Hahn, et al.,181 mentioned this same substitution and in addition, indicated that the [a] is also often used l796i1es Wilkerson Gray and Claude Merton Wise, The Bases of Speech (New York: Harper and Brothers, 1959), p. 258. 180Johnnye Akin, And So we Speak: Voice and Articu- lation (Englewood Cliffs, N.J.: Prentice-Hall, 1958), p. 27. 181Hahn, et al., op. cit, p. 154. 68 in place of [a] by Eastern speakers in such words as "art" 182 and "garden.“ Wise described the use of [a] by Eastern speakers as did Akin183 but also added that Southern speakers sometimes use this sound in pronouncing the pronoun "I." 184 the central Central Vowels.--According to Akin, vowels of interest in terms of dialect are [3, 3, 3,] and [9]. She stated that General American speakers tend to pro— nounce the "r" wherever it occurs orthographically. In ' Southern and Eastern speech, however, this sound may be pro- nounced as a one—tap trill [r] as in "very": as [s] as in "sister," [snstg]; or it may be silent as in "far," [fa]. Hahn, pp_al., described the [3] as the "r" used by speakers 185 who drop their "r's." 'They also stated that this sound becomes [9] in the unstressed position. 186 Back Vowels.--For this set of sounds, Akin indi- cated that the [a] in the General American pronunciation of "war" becomes [9] when uttered by a Southern speaker, and [u] when uttered by an Eastern speaker. These observations 187 188 were not supported by either Wise or Hahn, et a1. 182Wise, 0p. cit., p. 102. 183Atkin, loc. cit. 184Akin, op. cit., pp. 63—68. 185Hahn, et al., op. cit., p. 158. 186Akin, op. cit., p. 34. 187Wise, o . cit., p. 104-105. 188Hahn, et al., op. cit., p. 155. 69 a . Several variations in the pronunciation of the General Ameri- can [afl were also noted by Akin.189 She indicated that while General American speakers use this sound in "dog," and "soft," Eastern speakers most often substitute ID]. In ad- dition, Southern speakers often expand the [o] by the addi— tion of a [w] glide plus schwa in words such as "call" and 190 however, stated that the most frequent "caught." Wise, Eastern substitution for the General American [9! in words such as those listed above is the [to], i.e. [dog] and [ssafl]. This same alteration was mentioned for Southern speakers in words such as "orange," [rargndd and "forest," [fa—rest]. In 191 and Wise192 terms of the back vowel [u], both Akin made reference to the practice of Southern speakers inserting a [j] to diphthongize this sound in words such as "tune," "doom" and "nude." Hanley and Thurman provided both a summary of the above descriptions and some additional information relative to differentiating features of the dialects. They listed the ‘following characteristics as the major differences between General American and Eastern Speech.193 189Akin, o . cit., p. 41. 190Wise, 0 . cit., p. 107. 19lAkin, op. cit., p. 51. ? 192Wise, 0 . cit., pp. 109-110. 193 Hanley and Thurman, o . cit., p. 95. 7. 70 The "broad a" [a] is used, though not as extensively as some people believe; it appears especially before the "s," "1f," "gh," and “th” endings. ‘ There is a tendency for [ad to become [0] in words with [w]; examples are "want" [wont], "watch" [wotg ], and "water," [wata ]. There is a tendency for [e] to become [$1] in "air," "are," and "arry" words; examples are "chair" ['typ], "care" [ken], and "marry" [mgrn]. There is a tendency for [o] to become fix] in "og" words: examples are "log“ [ 1nd and "fog" [fag]. Unstressed suffixes tend to contain [1]: examples are "salad" [salxd], "roses" [rozxd, and "biggest [bxgrst] - The [3] is used in "ear," "pr," and "ir" words; ex- amples are "heard" [had] and "term" [tam]. Although the dialect seems to be changing in this "“7“”feature, preconsonantal and final "r's" are often silent; examples are "cart" [ka:t] and "for" [fo:]. The secondary stress in "ery," "ary," and "cry" words may be dropped,- examples are "cemetary" [semptr I] and "stationary“ [ steSenrfl.. . Hanley and ’I-‘hurmanlg'4 indicated that the firSt-seven characteristics listed above for the Eastern dialect also 194 Hanley and Thurman, o . cit., p. 96. 71 serve to differentiate the Southern dialect from General American speech. In addition, they felt that the following three characteristics should be considered in making this latter differentiation. 1. There is a tendency to use £3] for the final "r," as in "core" [keg], and "fire" [ fall. 2. The "drawl" or prolongation, diphthongization are considered substandard variations; examples are "pop" {poop}: "baby" [beIebn]: and "bless" [bnsjesl- 3. The omission of the second half of a diphthong is also considered substandard: examples are "ice" [as], "fine" [fan], and "oil" [91]. As a general summary, it appears that Eastern and Southern speakers both drop or alter the [ 3] and [g] in the final position in words such as "far" and "poor," while General American speakers retain these sounds. In words such as "barn," "park," "form," and "third," with the vocalic "r" preceeding a final consonant, the same situation occurs. Consonant "r's," i.e., [r], appear to be handled the same across all three dialects. The [0] sound as it is used in the General American dialect in such words as "foreign" and "orange" is flattened or broadened to an [a] in both Eastern and Southern speech. This broadening is followed through by Eastern speakers in such words as "hog" and "frog" while the Southern speaker tends most often to follow the General American use of [o]- 72 Eastern speakers also occasionally broaden the vowel as it is usually pronounced by General American speakers in such words as "ask" and "dance." Southern speakers use the same pronunciation of this vowel as do General American speakers. An occasional characteristic of Southern speech is the insertion of the [j] to form a diphthong in words such as "duty" and "tune." Despite the abundancy of research in terms of the differentiating characteristics of the three dialects of American English, no studies of listener performance in making these differentiations were encountered in the literature. In summary, there is some evidence that listeners can, upon hearing speakers' voices, make differing degrees of valid judgments about them relative to their sex, age, and physical characteristics. In terms of sex, this success was attained in the 1930's by Pear.195 With regard to age and physical characteristics, such differentiations have been very gross. No studies were encountered in the literature which dealt with listener identifications of ethnic group, education level, and dialect region. There is evidence, how- ever, that differentiations within these characteristics should be able to be made from hearing speakers' voices. 195Pear, op. cit., 117. 73 Training The term "training" as used in the present study is not synonomous with the term "learning." These two terms 196 He stated that if an investi- were contrasted by WOlfle. gator's primary interest is in the process by which knowledge is acquired, then his studies are classified under the head- ing of "learning"; if his primary interest is in the teach— ing of knowledge or skill, his work is classified under the heading of "training." The latter category was of main interest in the present study. With specific regard to training, w01fe has sug- 197 (1) the distribution of practice should be suitable for the task gested the following guidelines for such programs: to be learned; (2) active participation by the learner is superior to passive receptivity: (3) practive material should be varied so that the learner can adapt to realistic vari— ation and so that motivation during drill is improved; (4) performance records should be kept in order to evaluate pro- gress and the efforts of training: and (5) immediate knowledge should be given to the trainees regarding their performance. 196Doel WOlfle, "Training," Handbook of Experimental Psychology, ed. S. S. Stevens (New York: Wiley and Sons, 1951), PP. 1267-1268. 197Doel w01fle, "Military Training and the Useful Parts of Learning Theory," Journal of Applied Psychology (1946), pp. 73-75. 74 In terms of training experiences, Kingsley198 has stated that the smaller the interference between practice sessions, the more substantial the learning. In addition, he wrote that the shorter the practice period, the greater the learning that occurs. Mednick, however, did not support 199 He indicated that spacing out training these points. sessions only wastes time when the task to be learned is familiar. Under these conditions, a concentrated single pro— gram may have some advantages over a series of spaced sessions. Postman and Egan also offer guidelines relative to training as it applies in the present study.200 They felt that in general, the more distributed the practice trials, the better the learning. However, as with Mednick,201 they indicated that this principle varied according to the task. These authors also reported that meaningful materials are better retained than nonsense items. In addition, they stated that subjects learn best by performing tasks with full awareness of the principles guiding successful performance. 198Howard L. Kingsley, The Nature and Conditions of Learning (Englewood Cliffs, N.J.: Prentice-Hall, 1957), pp. 237-257. 199Sarnoff A. Mednick, Learning (Englewood Cliffs, N. J.: Prentice-Hall, 1964), p. 87. 200Leo Postman and James P. Egan, Experimental Psy- chology (New York: Harper and Brothers, 1949), pp. 395-462. 201Hednick, loc. cit. 75 This brief review of the literature in the area of training suggests several principles that guided the formu- lation of the training procedures evaluated in the present investigation. First, care was extended to maintain moti- vation. Second, as the task involved was largely a familiar one, i.e., judging speakers by voice on familiar character- istics, spaced practice was avoided in favor of one concen— trated session. Third, knowledge of performance was pro- vided to the subjects frequently throughout the program. Fourth, variation in practice was supplied to the extent that was possible. Finally, fifth, the subjects were in- volved actively during most of the training program. No references were located in the review of the literature relative to the use of trained listeners for identifying speaker characteristics. However, the perform- ance of these kinds of judges on similar tasks may be of significance. In terms of test-retest reliability, for ex- ample, Bryan and Wilke202 have found poor agreement between first and second evaluations of public speeches by untrained judges. These authors also found poor inter—judge agreement using small groups of these raters. They found, in terms of inter-judge agreement, that five judges yielded a correlation coefficient of + .66. Using ten and 20 untrained judges, 202Alec I. Bryan and w1ater H. Wilke, "A Technique for Rating Public Speeches," Journal of Consulting Psychology, V (March-April, 1941), pp. 80—90. 76 however, they obtained coefficients of + .83 and + .91, re- spectively. Knower, in a similar study with trained judges obtained correlations ranging from + .35 to + .46 using low numbers.203 He estimated that from eight to 16 judges would be necessary to produce reliability coefficients of + .87 or better. Eckert and Keys,204 using students as untrained listeners and teachers as trained listeners found that the former tended to be strongly influenced by what the authors called the "halo" effect when evaluating different traits of speakers. The trained judges, on the other hand, were found to be more discriminating. .In terms of evaluating defective speech, Perrin found, using the method of paried comparisons, that trained and untrained judges did not differ significantly in their evaluation of the severity of articulation defect.205 In ad- dition, both groups showed a significant amount of agreement within their own groups with respect to these ratings. How- ever, the correlation between the number of sounds the 203Franklin Knower, "A Study of Rank-Orderufiethods of Evaluating Performance in Speech Contests," Journal of Ap- plied Psychology, XXIV (October, 1940), pp. 633-644. 204R. G. Eckert and N. Keys, "Public Speaking as a Clue to Personality Adjustment," Journal of Applied Psy- chology, XXIV (January, 1940), pp. 144-153. 205Elinor Horwitz Perrin, "The Rating of Defective Speech by Trained and Untrained Observers," Journal of Speech and Hearing Disorders, XIX (March, 1954, pp. 48-51. 77 speakers misarticulated and the rated severity of the par- ticular speech problems was + .98 for the trained group and + .82 for the untrained. The difference between these figures was significant, and according to Perrin, indicated that despite the gross similarities in general severity with- in a group agreement, trained judges are more valid evalu- ators of speech problems. The Stimulus As indicated in the definition section of the pre- vious chapter, there were three important considerations with respect to the stimuli given to the subjects, i.e., the aural cues. These were: (1) duration, five, ten and fifteen seconds: (2) spontaneous as opposed to prepared speech; and (3) normal as opposed to telephone speech. As is obvious by the studies reviewed above under the section devoted to previous research on the six speaker characteristics of interest in this study,these factors have not played an im- portant role. Typically, the duration of these stimuli was not specified or was described as being of some "sufficient" length, such as 30 seconds. In terms of spontaneous as op- POsed to prepared speech, all but one of the studies reviewed used oral reading as the stimulus. Finally, with the ex- ception of the Fay and Middleton studies involving the voice as transmitted over a public address system, all stimuli con- sisted of either voice delivered by a concealed speaker or 78 or by recordings under typical face-to-face acoustic conditions. The following sections include descriptions of these departures in addition to research related to the consider- ations of stimulus duration, spontaneous as opposed to pre- pared speech, and normal as opposed to telephone speech as they applied in the present investigation. Duration The duration of speech samples used as stimuli in speaker identification studies to this point have typically not been specified. Other authors have described their samples as being sufficient or as being some specific duration that they considered to be sufficient. Duration as a factor affecting the performance of listeners in this task has not been researched. Two studies, however have been concerned with the ef- fects of duration on listener performance in rating the severity of articulation defectiveness. Morrison206 found no significant differences in the performances of her raters under three conditions of stimuli duration, five, ten, and fifteen seconds. In addition, she found that the preferences Of her judges were evenly divided between the five and ten Second samples. The judges were agreed that the fifteen \ 206Shelia Morrison, "The Severity of Articulation De- fectiveness," Journal of §peech and Hearing Disorders, XX (December, 1955), pp. 347-351. 79 second samples were too long. The author also noted that while evaluating these latter durations, judges tended to make their responses before the end of the samples, thus not using the total duration as a basis for their evaluations. 207 found no In a similar study, Sherman and Morrison significant difference in the performance of listeners in judging the severity of articulation problems under two different durations, five and ten seconds. It was interesting to note that not one of the studies reviewed mentioned the possibility of duration error. This is perhaps understandable in studies in which the goal is to present just a sufficient stimuli. However, in studies such as-Morrison's?08 and Sherman and Morrison's209 in which spe- cific stimulus duration is a factor of interest, error can- not be ignored. It was expected that the performance of listeners on the task required in the present study would improve with increased duration. This expectation was based on the pre- mise that longer duration would provide them with additional cues . 207Dorothy Sherman and Sheilla Morrison, "Reliability of the Individual Ratings of Severity of Defective Articu- lation," Journal of Speech and Hearing pasordepp, XX (December, 1955): PP. 352-358. 208Morrison, loc. cit. 209Sherman and Morrison, loc. cit. 80 Spontaneous as Opposed to Prepared Speech As reported above, all of the studies reviewed used prepared speech as the stimuli with the exception of one. Fay and Middleton,210 in one of their many investi- gations in the area of speaker identification, used spon- taneous speech in testing the performance of listeners in judging whether speakers were lying or telling the truth. In this study, an announcer asked a question and indicated to the respondent if he was to tell the truth or lie. Answers to the questions, then, were not prepared. Results indicated that listeners were able to make the required judgment correctly more often than was expected by chance. This result becomes meaningless, however, from the stand- point of comparing performance under both conditions, spon- taneous and prepared. The expectations with regard to this factor of the aural cues was that listeners would perform more accurately under spontaneous than under prepared speech conditions. This prediction was based for the most part on the contention that spontaneous speech is personal and therefore more re- vealing about the speaker than is oral reading. This was ex- pected to be especially true in the present study for two 210Fay and Middleton, "The.Abi1ity to Judge Truth- Telling, or Lying, from the Voice as Transmitted over a Public Address System," Journal_of General Psychology, XXIV (January, 1941), pp. 211-215. 81 reasons. First, the speakers did not read material of their own composition, i.e., not only was their speech during the readings removed from them in the sense of not being spon- taneous, it was also removed from the standpoint of the words they uttered not being their own. Secondly, the ma— terial they read, as has been described above, was defined as being neutral. That is, its content was selected for the purpose of controlling for interaction between individual speakers and the topic. Normal as Opposed to Telephone Speech The speech signal transmitted by a telephone has not before been used as a stimulus in speaker identification studies. There has been, however, considerable research done on telephone systems. This work has been summarized by 211 Fletcher. He indicated that the telephone conversational framework is built up by a comparitively small number of words in a great variety of patterns around the principal 212 words which are nouns. Fletcher also contrasted findings with regard to telephone conversations and common words used in written English.213 He reported that the most striking difference between the two is the large number of active verbs that occuramong the most commonly used words on the 211Fletcher, loc. cit. 212Fletcher, op. cit., p. 92. 213Fletcher, op. cit., p. 93. 82 telephone, e.g., "get," "see," "know." None of these kinds of words appeared among the 50 most commonly used written English words. In addition, among the common written words five times as many conjunctions are used as compared to tele- phone conversations. Another interesting feature between these uses of English is that in telephone conversations 14 out of the first 100 commonly used words are of mOre than one syllable. In written English there are ten. Fletcher 214 In the also pointed out differences in word origin. first 100 words, 11 on the telephone list are derived'from Latin while this is true of only two words on the written list. In terms of typical speech intensity levels over the telephone, Fletcher reported a range from 54 to 75 dB, SPL.215 He indicated that the mean for:this distribution was 68 dB, SPL.216 This is compared with an intensity of 65 dB, SPL, that Hanley and Thurman reported as the typical conversation- al speech level at three feet from a speakers' lips.217 This chapter has surveyed the previous literature with regard to listener identifications of the six speaker characteristics of interest in the present investigation 214Fletcher, op. cit., p. 94. 215Fletcher, op. cit., p. 77. 216Fletcher, op. cit., p. 395. 217Hanley and Thurman, 0p. cit., p. 127. 83 from two standpoints: (1) actual performance of listeners and (2) demonstrated vocal features that might serve as a basis upon which they could make the differentiations. In addition, sections were devoted to listener training, stimu- lus duration, prepared as opposed to spontaneous speech stimuli and telephone as opposed to spontaneous speech stimu- li and telephone as opposed to normal speech stimuli as these factors applied in the present investigation. CHAPTER III ~ EXPERIMENTAL PROCEDURES w The listener subjects participating in this study were 138 undergraduate students at Michigan State University who were enrolled in Speech 108, Voice and Articulation, during the Spring quarter, 1967. This group was comprised of six recitation sections which were randomly selected from a total of 12 sections. Each section had an enrollment of 30 students. However, due to absences during the test and/ or re-test administrations, it was necessary to set the N for each recitation group at 23. Reductions were made by means of random procedures with listener subject numbers being drawn out of a box for groups that totaled more than 23 after the test and re-test conditions had been completed. Of the six recitation sections selected as listener subjects, three were selected at random from the six sections of Voice and Articulation that met for two one and one-half hour sessions per week. These three groups of 23 listener slibjects each, were designed as trained subjects. The re- maining three recitation sections used as listener groups were selected from the six Voice and Articulation sections 84 85 which met for three one hour sessions per week. 'These three groups of 23 listener subjects each were designated as un- trained subjects. Apparatus The following instruments were used for: (1) record- ingthe original speech samples, the training program, and the instructions: (2) constructing and presenting the test Stimuli to the listener subjects: and (3) various evaluation- al procedures performed to determine the adequacy and valid- ity of portions of the constructed materials. 1.. Magnetic recording tape (3M, Type 111) 2. Five tape recorders (two Ampex, Model 601; Ampex, .Model 350-G: Magnecord, Model 1022: and wollensak, Model T-1500) 3. Amplifier-mixer (Ampex, Model MXé35) 4. Three microphones (Electro-Voice, Model 654: Bruel and Kjaer, Type 4131 with cathode follower: and Bruel and Kjaer, Type 4132 with cathode follower) 5. Microphone power supply (Bruel and Kjaer, Type 2801) 6. Pure tone oscillator (Hewlett-Packard, Model 202-C) 7. Anechoic chamber (Bruel and Kjaer, Type 4212) 8. Frequency analyzer (Bruel and Kjaer, Type 2107) 9. Power level recorder (Bruel and Kjaer, Type 2305) 10. Amplifier-speaker (Ampex, Model 620) 11. Timer (Hunter, Model lOO-C, Series D) 12. Electronic switch (Grason-Stadler, Model 829-3), 86 13. Sound treated room (Industrial Acoustics Corpor- ation, Series 400) 14. Two telephones (Western Electric) .15. Pitch instrument (Pyramid Chromatic) 16. Earphones (Telephonics, Model TDH-39) 17«- Telephone-microphone coupler 18«- Sound level meter (Bruel and Kjaer, Type 2203 with Bruel and Kjaer microphone, Type 4131) Procedures The experimental procedures involved the following steps: (1) testing the telephone recording system: (2) re- cording the speakers: (3) constructing the master tapes which comprised the actual test stimuli: (4) recording and con- Structing the training program: (5) constructing the post- training practice test: (6) recording the instructions to the listener subjeCts: and (7) administering the tests. These steps are described in detail below. Telephone Recording System In recording test stimuli from the telephone, it was desired to simulate speech, as nearly as possible, in the way a listener hears it during a conventional telephone conVersation. This desire eliminated the possibility of using any commercially available_te1ephone recording instru- ment, Such instruments were found to take the signal from the telephone system at some point before the signal reached the listener's ear. In the present study, the entire 87 telephone system was utilized.' This was made possible by the construction of a coupler that enabled the speech signal to be.cecorded~frOm the telephone receiver in much the same way as the listener receives the signal. A drawing of the coupler and its dimensions are shown in Figure 1. In terms of con- strucrtion, it was composed of two parts. The base was a standard telephone receiver cap that screws on to the re- ceiver. The cap was attached to screws to a circular plastic Piece with a hole drilled down through the center to accomo- date a condenser microphone (B and K, Type 4132, with cathode follower). With its protective grid in place, the diaphram of the microphone, while using this device, rested 4 milli- meters from the most depressed portion of the concave receiver cap. 7F 1 ' I 8 ! T ( 1' - microphone insert -.l a | 5 | é—— plastic piece on ' l h l . l 4 l. a; l (——— screw '. '\ F. pl : / 8 ,L r . ix ‘-- ---’ base [11.5.3.4 l<-—— cm ——i Figure 1.--Telephone-Microphone coupler. 88 The entire system was composed of the following: (1) sound treated room (IAC, Series 400); (2) input telephone (Western Electric, station number 61, Auditorium Building, Michigan State University» (3) output telephone with coupler (Westelanlectric, station number 60, Auditorium Building, Michigan State University): (4) condenser microphone (B and K, Tyrxe 4132, with protective grid and cathode follower and in c0rxjunction with microphone power supply, B and K, Type 2801): (5) anechoic chamber (B and K, Type 4212); and (6) Tape recorder (Ampex, Model 601). A block diagram of the System is shown in Figure 2. receiver coupler sound room Station 60 \micfophone \ T [‘ Station 61 ‘L, receiver anechoic chamber / osci 1 I atOr tape micr phone Speaker- recorder power Figure 2---Block diagram of telephone recording system. 89 The response of the telephone recording system was evaluated by means of recording pure tones from 125 to 10,000 Hz through it. An oscillator (H-P, Model 202-C) served as the sound source with its output channeled to an amplifier- speaker (Ampex, Model 620). The input telephone was attached to the speaker in such a way that its microphone portion rested directly in front of the center of the speaker at a distance of 1.5 centimeters. The output of the speaker was Set at 75 dB, SPL (B and K SPL meter, Type 2203, with a B and K microphone, Type 4131, and standard grid) at a distance of 1.5 centimeters from the center of the speaker for each pure tone. The entire input portion of the system was located in 'the sound room referred to above. As indicated in Figure 2, theeoutput telephone receiver was located in an anechoic chamber (B and K, Type 4212) and coupled to a microphone (B and K, Type 4132). The response characteristics of the tele- Phone recording system as tested by this procedure were evaluated by means of a frequency analyzer (B and K, Type 2107). The results are graphically illustrated in Figure 3. Recording of Speakers The speakers, providing the stimulus material, were selected randomly for each of the six speaker characteristics. These procedures are described below for each characteristic. Six speakers for each dimension of the six characteristics Were recorded for a total of 90. Specifically, these were as follows: 9O +20fi +10- dB 0— Gain _10_ (SPL) .201 -30— -40— 50 i 1_. A o S ‘ :2 ' J 1 :2 WWW N 10 O H N V '\D an F1 .m m Hertz Figure 3.--Response characteristics of the tele- phone recording system. Sex...... ................................... .12 speakers Male.. .. ......................... . 6 speakers Female. .................................. 6 speakers Age...... . . . ........... . . ....... .18 speakers 20-30 .................................... 6 speakers 40-50 ........ . ........................... 6 speakers 60-70 ......... . .............. . ..... . ..... 6 speakers Ethnic Group.......... ...................... .12 speakers Negro ..... . .............................. 6 speakers Caucasian ................................ 6 speakers Education .................................. 18 speakers Less than high school. .............. ... 6 speakers High school graduate. ....... ...... ..... .. 6 speakers College graduate ................ . ...... .. 6 speakers 91 Physical Size ................................ 12 speakers Big ...................................... 6 speakers Small .......................... .. ........ 6 speakers Dialect Region......... ...................... 18 speakers Michigan.... ............................. 6 speakers South ................... . ................ 6 speakers East............... .............. . ...... . 6 speakers No speaker was used for more than one characteristic and, of course, within each characteristic, each speaker was eligible to be used under only one dimension. Speaker se- lection was made as follows. .§a§.--Twelve adult speakers, six males and six fe- Inales, were selected randomly from volunteers from the Michi- ‘gan.State All-University Choir. The males were all tenors i3nd were selected randomly from a population of ten volun- ‘beers. The females were all contraltos and were selected randomly from a population of 13 volunteers. All 12 speakers ‘were from 18 to 25 years of age. They were all of the Cau- casian ethnic group and spoke English as a native language. In addition, they were all natives of Michigan, i.e., they had'been‘born, and attended elementary and secondary schools in this state. léga.--Eighteen speakers, six between 20 and 30 years 0f age, six between 40 and 50 years of age, and six between 60 and 70 years of age, were selected randomly from the popu- 1iltion of graduate students majoring in Speech and Hearing science and the faculty members of the Department of Speech and other departments at Michigan State University. No one 92 was used as a speaker who was involved as an instructor with the population from which the listener subjects were drawn, i.e., the 12 recitation sections of Speech 108, Voice and Articulation. All 18 speakers were Caucasian males with English as their native tongue. The median ages for the three groups of six speakers was 26 years for the 20-30 year Old group, 42 years for the 40-50 year old group, and 62 Years for the 60-70 year old group. pfihaic Group.--Twelve speakers, six of the Caucasian ethnic group and six of the Negro ethnic group, were randomly Selected from the male population of students enrolled in Speech 101, Public Speaking, and Speech 108, Voice and Articu- ilation, at Michigan State university during the Winter quarter, 1967. Specifically, the six Caucasian speakers were selected from.a population of 407 and the six Negro speakers were se- lected from a population of 31. A11 12 speakers were between 18 and 22 years of age and had been born and attended ele- mentary and secondary schools in Michigan. Education.--Eighteen adult speakers, three groups of six each, were selected as follows. Six speakers with less than a high school education were randomly selected from a population of ten such persons, all of whom were either suc- cesSfully employed or in training for employment. .Of the six selected as speakers in the study, two were employed as serVicestation attendants, two as janitors, one as an 93 iJuiustrial maintenance man, and one as a warehouse worker. These speakers ranged from 19 to 32 years of age and had (xxnpleted an median education level of grade ten. The six high.school graduates were selected from a population of 13. These speakers were younger in age than the above group, ranging from 19 to 25 years. Two were employed as combination truck drivers-delivery men-salesmen, one as a delivery man for a dry cleaning establishment, one as a grocery store clerk, one as a service station attendant, and one worked for a florist, performing general greenhouse duties as well as Selling duties. NOne of these six speakers had attended col- ilege. The six college graduate speakers were all employed as teachers and were randomly selected from a population of 11, all of whom were similarly employed. Age ranged from 23 to 29 years. In addition, all speakers representing this category were Caucasian males who spoke English as a native language and were natives of Michigan. Physical_§ipe.--Twelve speakers, six who qualified as "big" and six who qualified as "small," were chosen from the male students enrolled in Speech 101, Public Speaking. and Speech 108, Voice and Articulation, at Michigan.8tate University during the Winter quarter, 1967, who met the Specified size qualifications. As indicated in Chapter I, the terms "big" and "small" were defined relative to standard 94 ruprms for 17 to 19 year old males.218 Thus, "big" speakers had a height equal to or larger than one standard deviation above the national mean for their age group (five feet, nine 219 inches ) and had a weight equal to or larger than one standard deviation above the national mean for males in their age group who are one standard deviation above the 220). With these national average in height (160 pounds criteria, "big" speakers had to be at least six feet tall and weigh 180 pounds. "Small" speakers were defined in the same manner using deviations below the national height and weight averages for their age group. That is, they were equal to or'smaller than one standard deviation below the national 221 mean for height and had a weight equal to or smaller than r1s. However, the relationship between the two combi- natixarls of telephone speech samples (SpTe and PrTe) is different. Under the Ethnic Group characteristic, the means for tliese two treatment combinations were not significantly different from each other. For the present interaction, 142 however, they are. In addition, under the Ethnic Group characteristic, both of the telephone speech treatment com- bination means were significantly larger than the mean for the spontaneous normal treatment combination. For the Edu- cation group, however, only the spontaneous telephone com- bination mean was larger. It can also be noted that within the normal speech treatment combinations a significantly higher mean propor- tion correct was obtained for the Education characteristic under the prepared speech condition than under the spontane- ous speech condition. In addition, within the prepared speech treatment combinations, significantly higher means were obtained for prepared normal speech than for prepared telephone speech. Within the spontaneous treatment combi- nation, this order was reversed. Significantly higher means were obtained under conditions of telephone speech than under conditions of normal speech. In reality, it can be seen that all possible meaningful treatment combinations com- posing this interaction effect were significantly different from each other. The only two means underlined by a common line are those for the spontaneous normal treatment combi- nation and the prepared telephone combination. The product-moment correlation coefficients for test- retest reliability for each treatment under the three main effects for Education are presented in Table 16. 143 Table l6.—-Product-moment correlation coefficients for test- retest reliability on the Education characteristic for trained and untrained subjects, spontaneous and prepared speech samples, and normal and tele- phone speech samples. Correlation Treatment Coefficient Untreated subjects .43 Trained subjects .41 Spontaneous speech .47 Prepared speech .37 Normal speech .37 Telephone speech .46 Inspection of the coefficients in Table 16 indicates that again, training did not appear to have an effect upon test-retest reliability. This table also indicates that test—retest reliability was slightly higher under the'spon- taneous and telephone speech conditions than under the pre- pared and normal speech conditions. NOne of the coefficients approach acceptability in terms of explained variance. Physical Size.——The fifth 2 X 2 X 2 analysis of vari- ance for repeated measures was performed to test the effects of listener training, spontaneous as opposed to prepared speech, and normal as opposed to telephone speech on listener subject identifications of the Physical Size characteristic. The mean proportion correct under each treatment of the three 144 main factors and their differences and standard deviations are presented in Table 17. Table l7.—-Mean proportion correct and standard deviation for each treatment and difference between the treat- ment of the three main factors for the Physical Size characteristic. Standard Treatment Mean Difference Deviation Untrained subjects .73370 '03260 .32850 Trained subjects .76630 ' .27750 Spontaneous speech .70109 09782 .32243 Prepared speech .79891 ' .27693 Normal speech .79529 09058 .29655 Telephone speech .70471 ' .30561 TOTALS .75000 .30424 The summary table for the analysis of variance is presented in Table 18. Inspection of this summary table indicates that for Physical Size there were three signifi- cant F statistics. Two of these were for the main effects spontaneous as opposed to prepared speech and normal as op— posed to telephone speech. As indicated in Table 17, the difference in terms of the spontaneous-prepared factor was due to a significantly higher performance by the listener subjects under the prepared condition than under the spon- taneous condition. In terms of the normal-telephone factor, 145 the significant F came from the listener subjects performing better under the condition of normal speech than under the condition of telephone speech. Table 18.—-Summary of analysis of variance comparing the ef— fects of listener training, spontaneous as op— posed to prepared speech, and normal as opposed to telephone speech on listener performance for the Physical Size characteristic. Mean Source of Variation df Square F* Between subjects 137 Training (A) 1 0.1467 1.76 Ss within groups 136 0.0835 Within subjects 414 Spontaneous-Prepared (B) 1 1.3207 17.27# A X B 1 0.0290 0.38 B X Ss within groups 136 0.0765 NOrmal-Telephone (C) 1 1.1322 12.52# A X C 1 0.0652 0.72 C X 88 within groups 136 0.0904 B x c 1 1.0435 10.87# A X B X C 1 0.0888 0.92 B X C X Ss within groups 136 0.0965 *F.2 3.84 required for significance at the .05 level, with l, 136 df. #Significant beyond the .05 level. The third significant F mentioned above was for the spontaneous-prepared and normal-telephone interaction. The 146 individual treatment combination means for this interaction are represented graphically in Figure 9. . 80 Pr: . SP. Pr Proportion .70 Correct 260 SP .50 No Te Speech Condition Figure 9.--Graphic representation of the individual treat— ment combination means for the spontaneous—pre- pared and normal-telephone interaction on the Physical Size characteristic. A Newman—Keuls difference of means test, as described above, was performed to determine the origin of the signifi- cance. The following results were obtained: igng SpNo PrTe PrNQ These results indicate that the listener subjects were able to attain the highest mean proportion correct under the prepared normal treatment combination. In addition, they indicate that significantly higher means were obtained under the spontaneous normal, prepared telephone, and prepared normal treatment combinations than were obtained under the spontaneous telephone combination, but that these higher means did not differ from each other. Also, it can be seen 147 that the means for the two prepared (PrNo and PrTe) and two normal (SpNo and PrNo) treatment combinations did not differ from each other while the two spontaneous (SpNo and SpTe) and two telephone (SpTe and PrTe) treatment combinations did. That is, within the spontaneous speech treatment combination, the listener subjects were able to identify correctly a sig- nificantly larger proportion of the physical size speech sam- ples when they heard them under the normal speech condition than when they heard them under the telephone speech cone dition. Likewise, within the telephone speech treatment com— binations, they correctly identified a significantly larger proportion of these speech samples when they heard them under the prepared condition than when they heard them under the spontaneous condition. Finally, it is interesting to note that all of the interactions encountered to this point, i.e., under the Ethnic Group, Education, and Physical Size charac- teristics, have involved the spontaneous-prepared and normal- telephone treatment combinations and, that in each case, the prepared normal treatment combination has been the highest in terms of mean proportion correct. The product-moment correlation coefficients for test— retest reliability for each treatment under the three main effects for physical size are presented in Table 19. Inspection of the coefficients in Table 19 indicates relatively consistent test-retest reliability under the 148 various treatments. However, as before, they leave much to be desired in terms of explained variance. Table 19.—-Product-moment correlation coefficients for test— retest reliability on the physical size character- istic for trained and untrained subjects, spon- taneous and prepared speech samples, and normal and telephone speech samples. Correlation Treatment w -‘ W ~5COeffiCient Untrained subjects .51 Trained subjects ..54 Spontaneous speech .53 Prepared speech .48 Normal speech .44 Telephone speech .59 Iniglggt_3§gigg.-—The sixth and final 2 X 2 X 2 analy- sis of variance for repeated measures was performed to evalu— ate the effects of listener training, spontaneous as opposed to prepared speech, and normal as opposed to telephone speech on listener subject identifications of the Dialect Region characteristic. The mean proportion correct under each treatment of the three main factors and their differences and standard deviations are presented in Table 20. The summary table for the analysis of variance is pre- sented in Table 21., InspectiOn.of this table indicates the presence of two significant F statistics. These were for the main effect normal-telephone speech samples and, as. 149 before, for the spontaneous—prepared and normal-telephone interaction. By inspection of Table 20, it can be seen that the significant F for the main effect was due to the listener subjects obtaining a significantly larger proportion correct for the Dialect Region characteristic under conditions of telephone speech than under conditions of normal speech. In terms of the significant interaction effect, the individual treatment combination means are presented graphically in Figure 10. Table 20.--Mean proportion correct and standard deviation for each treatment and differences between the treatment means of the three main factors for the Dialect Region characteristic. Standard Treatment Mean Difference Deviation Untrained subjects .66489 01583 .27657 Trained subjects .68072 ' .26189 Spontaneous speech .69293 04025 .24818 Prepared speech .65268 ' .28774 Nermal speech .61241 .12289 .28578 Telephone speech .73420 .23656 150 Table 21.--Summary of analysis of variance comparing the ef- fects of listener training, spontaneous as opposed to prepared speech, and normal as opposed to tele- phone speech on listener performance for the Dia- lect Region characteristic. Mean Source of Variation df Square F* Between subjects 137 Training (A) 1 0.3460 0.49 83 within groups 136 0.0700 Within subjects 414 Spontaneous—Prepared (B) 1 0.2236 3.60 A X B 1 0.0437 0.70 Box 88 within groups 136 0.0620 NOrmal—Telephone (c) 1 2.0807 33.37# A X C 1 0.1993 3.20 C X 88 within groups 136 0.0623 B x c 1 1.4031 20.60# A X B X C 1 0.2510 3.68 B X C X Ss within groups 136 0.0681 *F.2 3.84 required for significance at the .05 level, with l, 136 df. #Significant beyond the .05 level. .80 Pr Proportion .70 - Sp Sp Correct .60 Pr .50 No Te Speech Condition Figure 10.--Graphic representation of the individual treatment combination means for the spontaneous—prepared and normal-telephone interaction on the Dialect Region characteristic. 151 The test for differences between means yielded the following results: _g£NQ SpNo SpTe Igrgg Thus it can be seen that significantly higher mean performance was obtained under the prepared telephone treat— ment combination than under any of the other three. It will be recalled that on occasions when this interaction effect has been significant on other characteristics, i.e., Ethnic Group, Education, and Physical Size, the highest mean cor- rect performance was obtained under the prepared normal treatment combination. However, as indicated above, under the Dialect Region characteristic, the mean proportion cor— rect under this treatment combination was significantly lower than any of the other three treatment combination means. Al- so, within the two prepared speech treatment combinations (PrNo and PrTe), it can be seen that the listener subjects were able to identify correctly a significantly higher mean proportion of the telephone speech samples correctly than of the normal speech samples. This same relationship holds within the two spontaneous treatment combinations (SpNo and SpTe) but the differences are not significant. Within the two telephone treatment conditions (SpTe and PrTe), signifi- cantly higher means were obtained under the prepared con- dition than under the spontaneous condition. Under the two normal conditions (Spr and PrNo), however, this pattern was reversed. The listener subjects obtained significantly 152 higher means under conditions of spontaneous speech than they did under conditions of prepared speech. The product-moment correlation coefficients for test- retest reliability for each treatment under the three main effects for dialect region are presented in Table 22. Table 22.--Product-moment correlation coefficients for test— retest reliability on the dialect region charac- teristic for trained and untrained subjects, spon- taneous and prepared speech samples, and normal and telephone speech samples. ——‘ — _ — Correlation Treatment Coefficient Untrained subjects .49 Trained subjects .41 Spontaneous speech .52 Prepared speech .40 NOrmal speech .46 Telephone speech .40 Inspection of the coefficients in Table 22 again indi- cates relatively consistent test-retest reliability among the various treatments, especially for the trained and untrained listener subjects and for normal and telephone speech samples. As before, however, these relationships are not encouraging from the standpoint of accounting for the variation between the two tests. The following null hypotheses were tested by the six three-way analyses of variance described above: 153 There will be no significant difference in the per- formance of trained and untrained listeners in identi- fying correctly each of the speaker characteristics in question on the basis of aural cues alone. There will be no significant difference in the per- formance of listeners in identifying correctly each of the speaker characteristics in question on the basis of aural cues alone under the conditions of presentation of the message by means of normal speech as Opposed to presentation of the message by means of telephone speech. There will be no significant difference in the per- formance of listeners in identifying correctly each of the speaker characteristics in question on the basis of aural cues alone under the conditions of spontaneous presentation of the message (extemporane- ous speech) and prepared presentation of the message (oral reading). There will be no significant interaction effect be- tween training and presentation of the message by means of normal speech as opposed to presentation of the message by telephone speech on the performance of listeners in identifying correctly each of the speaker characteristics in question on the basis of aural cues alone. There will be no significant interaction effect be- tween training and spontaneous presentation of the message as opposed to prepared presentation of the message on the performance of listeners in identify- ing correctly each of the speaker characteristics in question on the basis of aural cues alone. There will be no significant interaction effect be- tween presentation of the message by means of normal speech as opposed to presentation of the message by means of tale-speech and spontaneous presentation of the message as opposed to prepared presentation of the message on the performance of listeners in identifying correctly each of the speaker characteristics in question on the basis of aural cues alone. There will be no significant interaction effects among training, presentation of the message by means of normal speech as opposed to presentation of the mes- sage by means of telephone speech, and spontaneous presentation of the message as opposed to prepared 154 presentation of the message on the performance of listeners in identifying correctly each of the Speaker characteristics in question on the basis of aural cues alone. Null hypothesis number two, dealing with the effects of training, is not rejected. Although the mean correct per- formance of the trained listener subjects was higher in magni- tude than the mean correct performance for the untrained listener subjects on all speaker characteristics, no signifi- cant F statistics were obtained for this effect. For null hypothesis number three, dealing with the effects of normal speech samples as opposed to telephone speech samples, partial rejection is possible. Specifically, it is rejected for the Age, Physical Size, and Dialect Region speaker characteristics but not for the Sex, Ethnic Group, and Education characteristics. Partial rejection is also possible for null hypothe- sis number four, dealing with the effects of spontaneous as opposed to prepared speech. Listener subject performance was significantly affected by this factor for the Age, Ethnic Group, and Dialect Region characteristics. Rejection was not possible for the Sex, Education, and Physical Size characteristics. The remaining four null hypotheses (five through eight) are concerned with interaction effects. It will be recalled that only one such effect was significant for the above analyses. This, of course, was the spontaneous- prepared and normal—telephone combination. Thus, the fifth 155 sixth, and eighth null hypotheses listed above, dealing with the training by normal as opposed to telephone speech inter- action, the training by spontaneous as opposed to prepared speech interaction, and the three factor interaction, cannot be rejected. In terms of the seventh null hypothesis, how- ever, partial rejection is possible. Significant results were obtained on the spontaneous-prepared by normal—telephone interaction for the Ethnic Group, Education, Physical Size and Dialect Region characteristics. Rejection is not possi— ble for the Sex and Age characteristics. Duration.—-The final analysis of variance was a four— way procedure (2 X 3 X 2 X 2) testing the effects of speech sample duration in conjunction with the other three factors on listener subject performance over all speaker character- istics. As such, the effects of interest were only those in- volving the duration factor. The mean proportion correct for each treatment of the duration factor over all speaker characteristics and their standard deviations are presented in Table 23. The summary table for the analysis of variance is presented in Table 24. Inspection of this table indicates that four significant F statistics were obtained, two of which are not of interest in terms of the questions and null hypotheses posed. One of these significant effects that is not of interest was obtained for the training effect. This lack of interest is based upon the tests for the training 156 effect for each individual characteristic as described in the above sections. The obtained result, however, is in- teresting from the standpoint that the training factor was not significant for any of the individual characteristics. This over-all significance is evidently due to the fact that the trained listener subjects obtained a numerically higher mean proportion correct than the untrained listener subjects on every characteristic with the exception of one (Ethnic Group, see Table 11). Table 23.--Mean proportion correct and standard deviation for each treatment of the duration factor over all speaker characteristics. Standard Treatment Mean Deviation 5 second speech samples .69176 I .11299 10 second speech samples .75740 .09251 15 second speech samples .68772 .09381 The second significant F that is not of interest was obtained for the over-all normal-telephone effect. The same reasoning in terms of non-interest stated for the over-all training effect applies here. It should also be noted that this normal-telephone factor had significant effects upon listener subject performance under the Age, Physical Size, and Dialect Region characteristics. 157 Table 24.--Summary of analysis of variance comparing the ef- fects of listener training, posed to prepared speech, spontaneous as op- normal as opposed to telephone speech and duration of the speech samples on listener performance over all speaker characteristics. Mean Source of Variation df Square F Between subjects 137 Training (A) 1 6.7697 6.39* Duration (B) 2 28.1541 26.58# A x B 2 3.8330 3.61# Ss within groups 132 1.0593 Within subjects 414 Spontaneous-Prepared (C) 1 2.2225 2.88 A X C 1 0.5510 0.71 E X C 2 1.9815 2.57 A X B X C 2 0.1904 0.24 C X Ss within groups 132 0.7710 Normal-Telephone (D) 1 5.0976 6.02* A X D 1 0.8628 1.02 B X D 2 1.8668 2.23 A X B X D 2 2.5338 2.98 D X Ss within groups 132 0.8461 2.98 C X D 1 0.7028 0.78 A X C S D 1 2.5359 2.80 B X C S D 2 2.4605 2.72 A X B X C X D 2 1.8787 2.07 C X D X Ss within groups 132 0.9046 *Significant beyond the #Significant beyond the .05 level with l, 132 df. .05 level with 2, 132 df. It can be noted in Table 24 that significant over- all results were obtained under the duration effect. A test of differences between means for this factor indicated the 158 following results. The numerals 5, 10, and 15 represent the three speech sample durations. 15 5 '19 These results indicate that the over-all mean ob- tained under the ten second duration condition was signifi— cantly larger than those obtained under the five and 15 second speech sample durations. Interestingly enough, the lowest performance by the listener subjects was obtained under the 15 second duration condition. It should be noted, however, that this over-all mean was not significantly lower than the over-all mean obtained under the five second con— dition. The remaining significant F was obtained for the interaction involving the duration factor and the factor of listener training. The individual treatment combination means for this interaction are presented graphically in Figure 11. .80 Un - Proportion . 70 Tr 7 V Tr Un Tr Un Correct .60 .50 5 10 15 Figure ll.--Graphic representation of the individual treat- ment combination means for the duration and training interaction over all speaker characteristics. 159 A difference of means test was performed on this combination with the following results. Un15 Un5 _T;15 Tr5 TrlO Un10 These results indicate that significantly higher means were obtained under the ten second duration condition for both the trained and untrained groups. In addition, it can be noted that the trained groups making identifications under the five and 15 second duration conditions performed significantly better than did the untrained groups under these two duration conditions. The fact that the order of the magnitude of the five and 15 second duration means re- mained the same for the trained and untrained groups is con- sistent with the significant main effect for duration de- scribed above. The product—moment correlation coefficients for the test-retest reliability under each treatment of the duratiOn factor over-all characteristics are presented in Table 25. Table 25.-—Product—moment correlation coefficients for test- retest reliability over-all speaker characteris- tics for five, ten, and fifteen second speech samples. Correlation Treatment Coefficient 5 second speech samples .51 10 second speech samples .50 15 second speeCh samples .39 160 Table 25 indicates that test-retest correlations under the three treatments of duration were relatively con- sistent from treatment to treatment. It is interesting to note that the lowest test-retest reliability was obtained under the longest duration condition. The following null hypotheses were tested by the four-way analysis of variance described above: 9. 10. 11. 12. 13. There will be no significant difference in the per- formance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone under three conditions of speech sample duration: (1) five sec; (2) ten sec; and (3) 15 sec. There will be no significant interaction effect be— tween training and duration of the speech sample on the performance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone. There will be no significant interaction effect be- tween presentation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech and duration of the speech sample on the performance of listeners in identify- ing correctly the speaker characteristics in question on the basis of aural cues alone. There will be no significant interaction effect be- tween spontaneous presentation of the message as op— posed to prepared presentation of the message and duration of the speech sample on the performance of listeners in identifying correctly the speaker charac- teristics in question on the basis of aural cues alone. There will be no significant interaction effects among training, presentation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech, and duration of the speech sample on the performance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone. 161 14. There will be no significant interaction effects among training, spontaneous presentation of the message as opposed to prepared presentation of the message, and duration of the speech sample on the performance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone. 15. There will be no significant interaction effects among presentation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech, spontaneous presentation of the message as opposed to prepared presentation of the message, and duration of the speech sample on the performance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone. 16. There will be no significant interaction effects among training, presentation of the message by means of normal speech as opposed to presentation of the message by means of telephone speech spontaneous presentation of the message as opposed to prepared presentation of the message, and duration of the speech sample on the performance of listeners in identifying correctly the speaker characteristics in question on the basis of aural cues alone. Table 24 indicates that only null hypotheses nine and ten can be rejected. In terms of null hypothesis number nine, the results showed that the listener subjects were able to perform significantly better over all characteristics under the ten second duration condition than under the five second durations. Rejection of the tenth null hypothesis was possible on the basis of the significant interaction effect obtained for the training and duration treatment combination. The hypotheses stated above that were not rejected (11 through 16) were all null statements regarding interactions 162 between duration and training, spontaneous-prepared speech, and normal-telephone speech treatment combinations. Discussion Several questions were posed at the outset of this study regarding listener identifications of the speaker characteristics Sex, Age, Ethnic Group, Education, Physical Size, and Dialect Region. These questions are re-stated in the following sections with an effort made to answer them in terms of the present results. The first question asked whether or not listener sub— jects could correctly identify each of the speaker character— istics, and if so, to what degree are these identifications reliable. Research was cited in Chapter II that suggested an affirmative answer to this question for some of the charac- teristics. For others, however, the only information avail- able was in terms of possible differentiating vocal characteristics. .gex.--It can be noted in Table 3 that the listener subjects were able to identify correctly the sex of speakers 99 per cent of the time across all training, speech, and du— ration conditions. This performance was significantly higher than what was expected on the basis of chance. As indicated in Table 2 of this chapter, the optimum and habitual pitch levels used by the tenors and contraltos 163 utilized as speakers underethis.characteristic.were;fl.. somewhat higher than normal in the case of the males and somewhat lower than normal in the case of females. Direct 236 (see comparisons can be made to the findings of Fletcher Table 1, Chapter II). He found that the mean fundamental frequency for male speakers producing pure vowels was 125 Hz. By converting the musical scale to the frequency scale it can be seen in Table 2 that the males used as speakers under the Sex characteristic averaged approximately 154 Hz. More specifically, speakers one, two, and three, i.e., the male speakers used on the master tapes, had habitual pitches that averaged 156 Hz. This appears to be somewhat different than the figure Fletcher237 cited as "normal." For females, Table 1 suggests a normal fundamental frequency of 244 Hz. In terms of the present speakers, conversion of the musical notes in Table 2 indicates a mean habitual pitch of 231 Hz. Again, more specifically, speakers seven, eight, and nine, i.e., the female speakers used on the master tapes, had habitual pitches that averaged 235 Hz. This fundamental is somewhat lower than the "normal“ as cited by Fletcher.238 However, the difference between the female contraltos measured in the present study and the measurements presented in Table 236Fletcher, op. ci ., p. 62. 237Ibid.~ 238 Ibid. 164 l for females in general is not as large as the difference between the two male samples. The important feature of these differences in terms of the present investigation is whether or not they influenced the identifications of the listener subjects on the Sex characteristic. By inspecting the original answer forms for the test condition, it was found that the Sex characteristic was incorrectly identified on only 12 occasions out of a possible 3,192 (138 listener subjects judging four speech samples for each of six speakers, three males and three fe- males). Thus, it appears that, for the general case, what differences in pitch that were demonstrated did not seriously affect listener judgments in terms of sex. In terms of individual items, five of the twelve oc- cations upon which the Sex characteristic was identified in— correctly involved a male speaker being mistaken for a fe- male. In addition, on all five of the instances in which a male speaker was misjudged as a female, the same male speaker provided the stimulus--Sex speaker number one. It_can be seen by Table 2 that this speaker's habitual pitch was measured as being D3 or approximately 145.Hz. ;This level is approximately one musical note or 20 Hz above the normal as 239 cited by Fletcher. However, the fact that Sex speaker number two, a male whose habitual pitch was measured as G3, 239Ibid. 165 approximately 195 Hz, a pitch higher than that of Sex speaker one, was not, on any occasion, misjudged as a female is an argument against the hypothesis that Sex speaker number one was misjudged as a female soley because of his pitch level. In terms of females being misjudged as males, five of the total of seven such misjudgments involved Sex speaker number seven. It can be noted in Table 2 that this speaker's habitual pitch was measured as A4, a frequency of approxi- mately 230 Hz. The small difference between this pitch and the normal (244 Hz) for females as cited by Fletcher240 again cannot be the sole reason for the misjudgments. This is sup- ported by the fact that the other two occasions for females being misjudged for males involved Sex speaker number eight. Inspection of Table 2 indicates that this female speaker's habitual pitch was measured as B approximately 245 Hz and 4, virtually identical to the cited normal in Table l. .§g§.--Table 3 indicates that the listener subjects were able to identify correctly the Age characteristic 67 per cent of the time across all training, speech, and duration conditions. This performance was significantly higher than what was expected on the basis of chance. It is also cone. sistent with the previous research in terms of objective studies in this area. In terms of specific proportions, how- ever, the obtained performance in the present investigation 24OIbid. 166 was somewhat lower than the 99, 87, and 78 per cent correct performances of Ptacek and Sander‘s subjects.24l These differences may well be due to the different listeners em- ployed. It will be recalled that these two authors obtained 99 per cent correct performance on oral reading speech samples using graduate students in Speech Pathology and Audi- ology as listeners. Also, the evaluation required by these listeners was of a lower order than the differentiation re- quired of the listeners in the present study. These authors asked only that their listeners differentiate between young adults (under age 35) and older speakers (over age 65). One might expect better performance under these conditions than when differentiations among three specific ranges are requested. In terms of the individual listener subject identifi- cations, it was noted that misjudgments tended to move toward the center dimension. That is, when the age group of the 20 to 30 year old speakers was misjudged, the error tended to be the result of listeners judging them as being between 40 and 50 years of age. This same error occurred for the 60 to 70 ‘year old speakers. In addition, it was noted that the ex- tremes, i.e., the 20 to 30 year old speakers and the 60 to 70 year old speakers were identified correctly more often than the 40 to 50 year old speakers. Two speakers, one in 241Ptacek and Sander, loc. cit. 167 the 20 to 30 year old group and one in the 60 to 70 year old group, were consistently misjudged. The errors in terms of the younger speaker were evenly distributed between the 40 to 50 and 60 to 70 year old categories. The errors in terms of the older speaker were limited in judging him as being 40 to 50 years of age. Ethnic Group.--Table 3 indicates that the listener subjects were able to identify correctly the Ethnic Group characteristic 64 per cent of the time across all training, speech, and duration conditions. This performance, as were those for Sex and Age, was significantly higher than what was expected on the basis of Chance.- Inspection of the individual answer forms indicated that almost all of the errors made by the listener subjects for this characteristic consisted of misjudging speech samples furnished by Negro speakers. Caucasian speech samples were evaluated as being uttered by a Negro. In terms of indi- vidual speakers, one Negro man was consistently identified as Caucasian and one was consistently identified correctly. This suggests the influence of the vocal stereotypes so fre— quently referred to in Chapter II. The results for this characteristic show need for further investigation. Of particular interest would be studies involving more precise controls in terms of the speakers. It appears evident from the literature discussed 168 in Chapter II that what is frequently referred to as "Negro dialect" is a cultural or environmental phenomenon. This factor played an important part in the selection of a label for the present speaker characteristic. As such, more con- trol could be exercised in speaker selection than was used in the present study. For example, practically all Negro families now residing in Michigan at one time migrated to this state from the South. With this in mind, it would ap- pear logical to expect some speech differences on the basis of when this migration occurred. That is, one would not ex- pect the speech of a child whose parents migrated to Michigan from the South to be representative of the typical General American dialect. The same would be true for a child whose grandparents had migrated to this state from the South. The cultural effects on speech and language behavior discussed under education in Chapter II may also be of interest in terms of Negro speakers. Education.--Table 3 indicates that the listener sub- jects were able to identify correctly the Education charac- teristic 55 per cent of the time across all training, speech, and duration conditions. This performance was also signifi- cantly higher than what was expected on the basis of chance. In terms of the literature reviewed in Chapter II, this re- sult is not consistent with the results obtained by Fay and Middleton with regard to listeners differentiating between 169 243 242 and leadership ability. However, intelligence levels it is consistent with the studies reviewed in terms of the relationships of social status, occupational success, and cultural background with speech and language behavior. In terms of individual identifications, as with the Age characteristic, errors tended to consist of evaluating the speech samples of speakers with less than a high school education and those of college graduates as coming from speakers with a high school education. One of the speakers in the less than a high school graduate category was con- sistently judged as being a high school or a college graduate. Two of the college graduate speakers were misjudged by ap- proximately half of the listener subjects. These errors con- sisted of identifying their speech samples as being repre— sentative of a high school graduate most frequently. In- terestingly enough, this stereotyping did not occur on all of their speech samples. For one speaker, there was high listener agreement in terms of his being a high school gradu- ate or lower onvonly one of his four speech samples. For the other college graduate speaker, however, this stereotyping ap- peared to occur on three of four speech samples. A. 242Fay and Middleton, "Judgment of Intelligence from the Voice as Transmitted over a Public Address System." 243Fay and Middleton, "Judgment of Leadership from the Transmitted Voice." 170 Physical Size.—-Table 3 indicates that the listener subjects were able to identify correctly the Physical Size characteristic 75 per cent of the time across all training, speech, and duration conditions. This performance was sig- nificantly higher than what was expected on the basis of chance. Inspection of the individual answer forms did not indicate the presence of stereotyping as has been suggested for some of the other characteristics. The absence of this consistent error is evident in part by the relatively large mean proportion correct obtained by the listener subjects. Dialect Region.--Table 3 indicates that the listener subjects were able to identify correctly the Dialect Region characteristic 67 per cent of the time across all training, speech, and duration conditions. This performance was sig- nificantly higher than what was expected on the basis of chance. This result demonstrates that the differentiating features of the three dialect as described in Chapter II could be accurately categorized by the listener subject. As might be expected, the primary errors in listener subject identifications for this characteristic were of two types: (1) misjudging Eastern speakers as Southern speakers and (2) misjudging speakers from both of these regions as General American speakers. No real evidence of vocal stereo- typing, i.e., listener subjects consistently erroneously identifying a speaker as belonging to a particular dialect 171 region, was noted. The errors mentioned above seemed to oc- cur on specific speech samples rather than for particular speakers as a general rule. Further study of this speech characteristic in the future may become more interesting. With increased facili- ties, interest, and time for travel by residents of all re- gions of the country, speech boundaries and differences cannot help but undergo modifications. These changes are already occurring to some extent. Physical boundaries such as the Green and Berkshire Mountain Ranges which used to separate the Eastern dialect region from the rest of the country have been eliminated by transportation improvements and a growing interdependency among the various areas of the United States. Speech region boundaries are also being broken down by occu- pational changes. The South, for example, is no longer uni- fied by its primary agricultural crop—-cotton. In addition, with increased transportation facilities, we are seeing children frequently settle in a different part of the country than their parents, perhaps long distances from where they grew up. Thus, it can be seen in response to the first question that all speaker characteristics were identified by the listener subjects to a degree that was greater than chance occurrence. However, with regard to the second portion of this question, i.e., reliability of the identifications, the results are not as favorable. Inspection of Table 4 172 indicates that the highest test-retest reliability obtained was + .71 for the Ethnic Group characteristic. These corre- lations indicate, despite the close agreement of the retest means (also shown in Table 4) with the test means (Table 3), that a given listener subject's score on the retest cannot be predicted with desirable accuracy from his score on the test. That is, even for the highest obtained correlation coefficient, only 50 per cent of the variation in listener subject per- formance on the retest can be accounted for by the variation in listener subject performance on the test. It appears, then, that despite the close agreement between test and re- test means, the magnitude of the correlation coefficients places severe limitations upon the usefulness of the demon— strated success of the listener subjects in correctly identi- fying the characteristics. In addition, it can be seen in Tables 7, 10, l3, l6, 19, 22, and 25 that, for the most part, reliability was not altered by the treatment effects. The effects of training on test-retest reliability were mixed. For four of the six characteristics, the trained listener subjects obtained smaller correlation coefficients than the untrained group. In terms of the spontaneous as opposed to prepared speech factor, the spontaneous condition produced higher test-retest reliability than did the prepared speech condition. Results under the normal-prepared factor were mixed as for the train- ing factor. The normal speech condition produced the highest 173 reliability for the Sex, Age, and Dialect Region character- istics while for the Ethnic Group, Education, and Physical Size characteristics, the highest test—retest reliability was obtained under telephone speech conditions. The highest test-retest correlation coefficients under the various treat— ments were obtained for the Ethnic Group characteristic and here the magnitude of some E's approaches what would be de- sirable. As indicated by the above discussion, however, for the most part, the coefficients obtained can only lead one to conclude that the test-retest reliability was poor for the task under investigation. For this reason, no tests of sig- nificance were performed on the over-all correlations and no tests of differences were made between the correlations Ob- tained under the various treatments for any of the character- istics. It was felt that because of the insufficient magni- tude of the E's, the presence or absence of statistical sig- nificance or significant differences was irrelevant. The test-retest reliability obtained was particularly low for the Sex characteristic. One would expect a high test- retest correlation for this characteristic, both on the basis of the logical ease of making such identifications and from the standpoint of the excellent performance demonstrated by the listener subjects. The ease of this task for the listener subjects is shown not Only by their obtaining 99 per cent correct in their identifications, but also by the relatively low standard deviation in their performance. As 174 shown in Table 4, this statistic was .07 for the Sex charac— teristic. The standard deviations obtained under the other five characteristics are shown in Tables 8, 11, l4, l7, and 20. Inspection of these tables indicates standard deviations ranging from .27 for the Age and Dialect Region character- istics to .36 for the Ethnic Group characteristic. These statistics are somewhat larger than the .07 standard devi— ation obtained under the Sex characteristic. In spite of the fact that this low standard deviation lends support to the relative ease of the task, it does not contribute to obtaining high test-retest correlations. McNemar244 has pointed out that when a sample is restricted in range, the correlation obtained will be relatively low. This point is also supported by Peters and VanVoorhis.245 They indicated that when the coefficient of reliability of a test is given, it is important to know the ranges of the distributions from which the r was computed. They stated that the size of a coefficient of correlation is very much affected by the heterogeneity of the population on which it is computed and gave examples showing that the more homogene- ous the distribution is, the less likelihood there is of ob- taining high reliability. The two distributions in question 244Quinn McNemar, Psychological Statistics (New York: John Wiley and Sons, 1962), p. 144. 245Charles C. Peters and Walter R. VanVoorhis, Sta- tistical Procedures and Their Mathematical Bases (New York: L McGraw-Hill Book Co., 1940), p. 208. 175 here both have very narrow ranges. This can be noted by the data in Appendix F. The obtained r of .27 on this set of test and retest scores indicates that given a listener sub- ject's test score and the linear prediction rule, the possi— bility of predicting the retest score for that listener sub- ject would be small. Inspection of the data, however, indi- cates that regardless of a listener subject's test score, his retest score was practically always a proportion of 1.00. The reverse was also true. Regardless of a listener sub- ject's retest score, his test score was practically always a proportion of 1.00. In this way, the narrow ranges of these distributions severely limited the extent to which a linear relationship could be established between these two sets of data and reduced the value of the linear prediction rule and the 5 in describing the relationship. The second question asked whether or not training made a difference in the performance Of listeners in identi— fying correctly each of the speaker characteristics on the basis of aural cues. As pointed out above, a negative answer to this question is indicated by the results of the analyses of variance performed on each speaker characteristic. This is true even though the mean proportion correct obtained by the trained listener subjects was numerically higher than the mean proportion correct obtained by the untrained listener subjects for all but one of the characteristics. 176 Except for Ethnic Group, the means for the trained and un- trained groups were identical. The most obvious explanation for these obtained re- sults is that the training program was inadequate. This is suggested by the fact that exposure to the program made no difference in the performance of the listener subjects. That is, they performed as if they had not been trained. If this is where the fault lies, alterations in the training pro- cedures are indicated. Possibilities for such changes would include a longer training experience and more opportunity for practice, repetition of or increased content, and spaced sessions. Another possible explanation for the similar per- formance of the two sets of listener subjects lies in the procedures actually used. It will be recalled that the training program immediately preceded the administration of the actual test. The identification task for the trained listener subjects, then, involved approximately an hour and a half of steady work. As a result of this, it may be possi- ble that the depressed performance of the trained listener subjects is a reflection of boredom or fatigue rather than inadequacy of the program itself. Another possible explanation that must be considered is that the nature of the task may not lend itself to train- ing. That is, perhaps listener identification of speaker characteristics cannot be improved by training. This 177 possibility may be especially indicated for the Sex charac— teristic. In considering this explanation, however, it should be noted that in terms of numerical results, the trained groups consistently demonstrated better performance on the task than did the untrained groups. This was true even for the Sex characteristic and casts doubt on the possi- bility that training cannot, when properly administered, im- prove listener performance in identifying the characteristics in question. It seems probable that the best solution in terms of training lies in the combination of the first two suggestions. That is, that the training program be altered to include several short periods of training and practice and that the actual test be administered during a separate session. The third question asked whether or not making their identifications from normal as opposed to telephone speech conditions had a differential effect upon the performance of the listener subjects. The results indicated an affirmative answer for three of the speaker characteristics—-Age, Physi- cal Size, and Dialect Region. However, these results were not consistent in terms of favoring one speech condition over the other. For Age and Dialect Region, listener sub- jects performed significantly better under the normal speech cOndition. In terms of the latter characteristic, Physical Size, it would appear that some differentiating cue was available to the listeners under the normal speech condition 178 that was not available to them under the telephone speech condition. As indicated in Chapter II, large male speakers are stereotyped as having a low pitched, relatively intense voice. It is possible that under the filtering conditions imposed by telephone speech, as indicated in Figure 3, Chapter III, these vocal features were altered to an extent to make differentiation between the big speakers and the small speakers, with their stereotyped higher pitched, less intense voices, more difficult. This hypothesis suggests further in- vestigation of the present data in the form of inspection of individual speech samples and the corresponding listener identifications made on them. In terms of the significantly higher mean performance under telephone speech conditions for Age and Dialect Region, it appears logical to assume that these characteristics are more often evaluated by listeners while using the telephone. That is, possibly, the difference in the obtained means was due to the listener subjects being more accustomed to making judgments such as those required for Age and Dialect Region under conditions of telephone speech than under conditions of normal speech. The fourth question asked whether or not making their identifications from spontaneous as opposed to prepared speech samples had a differential effect upon the performance of the listener subject. :The results indicated‘an affirma- tive answer for three of the“$peaker characteristics-tAge, 179 Ethnic Group, and Physical Size. Again, however, these re- sults were not consistent in terms of favoring one speech- condition Over the other. For Ethnic Group and Physical Size, significantly better performance was obtained under prepared speech conditions. The opposite was true for the Age characteristic. The results in terms of the first two characteristics, Ethnic Group and Physical Size, are diffi- cult, at best, to justify. It would be expected, particu- larly in terms of Ethnic Group, that the spontaneous speech condition would allow for the presence of considerably more differentiating vocal features. It may be possible, however, that the oral reading involved in the prepared speech con- dition had some differentiating effect. One might expect, for example, on the basis of cultural influences, that Negro speakers would be less adept in oral reading than Caucasian Speakers. In terms of Physical Size, the stereotype of the big, non-academically inclined athlete may have some appli- cation to the results. ,Both of these hypotheses indicate the need for future investigation of individual speech samples. The fact that the listener subjects performed better in terms of identifying the Age characteristic correctly under spontaneous speech conditions is consistent with ex— pected results.' Evidently, this speech condition allowed such differentiating cues as rate, speech hesitancies and vocal inflectiOn patterns that are associated with the age 180 groups used in the present study to operate while the con— dition of oral reading evidently tended to mask these differ- entiating characteristics. The fifth and sixth questions inquired about inter- action effects between listener training and (1) normal as opposed to telephone speech samples and (2) spontaneous as opposed to prepared speech samples. The results indicated that neither of these treatment combinations significantly af- fected listener performance for any characteristic. It is felt that the remarks set forth above in terms of the in- adequacies of the training procedures may have application to these insignificant results. The seventh question asked whether or not the inter— action of the normal-telephone and spontaneous—prepared speech treatment combinations had a significant effect upon listener-performance for each characteristic. The results indicate an affirmative answer to this question for Ethnic Group, Education, Physical Size, and Dialect Region. For the first two characteristics, this significance was apparent- ly due to a differential effect of the spontaneous-prepared treatment combinations. That is, within the normal speech treatment combinations, a higher mean was obtained under the prepared condition than under the spontaneous condition. This order was reversed under the telephone treatment combi- nations. Within this latter factor, the highest of the two 181 means was obtained under the spontaneous telephone rather than the prepared telephone condition. In terms of this interaction effect for the Physical Size characteristic, the prepared normal treatment combina- tion again produced the highest mean. For this character— istic, however, the mean obtained under this combination was not significantly different from the means obtained under the prepared telephone and spontaneous telephone treatments. The fact that these three means were not significantly different from each other but were significantly higher than the lowest mean obtained, that for the spontaneous telephone treatment combinations, indicates that the significant over-all effect can be attributed to the latter factor. (For the Dialect Region characteristic, the signifi- cant interaction effect Of the normaletelephone and spontaneous-prepared treatment combinations was due to a differential result in terms of the latter factor. Within the telephone treatment combinations, significantly higher means were obtained under the prepared speech condition than under the spontaneous speech condition. This trend was re- versed under the normal speech treatment combinations. With- in this factor a significantly higher mean was obtained under the spontaneous speech condition than under the prepared speech condition. The final question with respect to the performance of the listener subjects on each speaker characteristic 182 inquired about the effect of an interaction among listener training, spontaneous as opposed to telephone speech. A negative answer to this question is indicated for all six characteristics. The remaining questions were inquiries about the ef- fects of duration by itself and in conjunction with listener training, spontaneous as opposed to prepared speech, and normal as opposed to telephone speech on listener subject performance across all speaker characteristics. The first of these questions asked about the main effect of duration. Re- sults indicated that the performance of the listener subjects under the ten second speech sample condition was significant— ly higher than the performance of the listener subjects ex- posed to the five and fifteen second speech duration. The listener subjects whose identifications were made under the latter condition obtained the lowest mean proportion correct. This result, while not expected, was consistent with the 246 to the effect that judges of articu- findings of Morrison lation proficiency preferred five and ten second speech samples to fifteen second samples. She found under the latter condition that judges tended not to listen to the en- tire sample before making their answers. This tendency was also noted during administration of the test procedures in the present investigation. It was also noted that the 246Morrison, loc. cit. 183 listener subjects making their identifications on the basis of the fifteen second samples tended to become more restless and seemingly disinterested in their task than did the listener subjects who were exposed to the other two speech sample durations. It is felt that this evident disinterest may have been the primary contributing factor to the lowest mean performance being obtained under the fifteen second speech sample treatment. The remaining questions with regard to the effects of duration on listener subject performance over all character- istics were concerned with interactions. The only one of these questions for which the results indicated an affirmative answer asked whether or not the treatment combination of training and duration significantly affected the performance of the listener subjects. It was found that both the trained and the untrained listener subjects exposed to durations of ten seconds in length performed significantly better than did the other groups. In addition, under the five and fifteen second speech sample durations, the trained groups performed significantly better than did the untrained groups. This chapter has presented and discussed the results obtained by means of the procedures outlined in Chapter III. In summary, it is noted that all six of the speaker charac- teristics were identified correctly to a degree that exceeded what was expected on the basis of chance. The training factor, 184 as defined in the present investigation, had no significant effect on listener subject performance for any of the charac- teristics. Both the normal—telephone and the spontaneous- prepared factors affected the performance of the listener subjects under some characteristics but not others. Finally, listener subject performance in making correct identifi— cations over all characteristics was found to be significantly affected by the duration of the speech sample from which the judgments were made. CHAPTER V SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS The basic purpose of this research was to evaluate the performance of listeners in identifying correctly the speaker characteristics of Sex, Age, Ethnic Group, Education, Physical Size, and Dialect Region on the basis of aural cues alone. Of additional interest were the effects of listener training, normal as opposed to telephone speech, spontaneous as opposed to prepared speech, and speech sample duration on the performance of the listeners. Summary Six groups of 23 listener.subjects each were selected for this study. Three of the groups were exposed to training procedures while the remaining three served as untrained listener subjects. One trained and one un- trained group made identifications of the speaker charac— teristics listed above on the basis of listening to speech samples under each of the three duration conditions - five seconds, ten seconds, and 15 seconds. Each listener subject made eight identifications of the Sex, Ethnic Group and Physical Size characteristics, two under each of the following speech treatment combinations: 185 186 (l) spontaneous normal speech; (2) spontaneous telephone speech; (3) prepared normal speech: and (4) prepared telephone speech. For the Age, Education, and Dialect Re— gion characteristics, each listener subject made 12 identi- fications three under each of the above speech treatment combinations. Speech samples were provided by selected speakers. Six speakers provided the stimuli for identifying the Sex, characteristic, three males and three females. The same number of speakers provided the stimuli for the Ethnic Group. and Physical size characteristics. These were, repectively, three Negroes and three Caucasians and three big speakers and three small speakers. Nine speakers provided the stimuli for identifying the Age characteristic, three between the ages of 20 and 30 years, three between the ages of 40 and 50 years, and three between the ages of 60 and 70 years. The same number of speakers provided the stimuli for the Educa- tion and Dialect Region characteristics. For the former these were three speakers with less than a high school educa- tion, three speakers who had graduated from high school and three speakers who were college graduates. For Dialect Region, three speakers were from the East, three were from the South, and three were natives of Michigan. All speakers were males with the exception of the females employed under the Sex characteristic. Speech samples were presented to the listener subjects by magnetic tape. Re-test measures were made 48 187 hours after administration of the actual test for each group of listener subjects. Results indicated that all six of the speaker charac- teristics could be identified correctly by the listener subjects to a degree that exceeded what was expected by chance. Listener training, as defined by the procedures followed in this study, did not have a significant effect upon subject performance under any characteristic. Listening to spon- taneous speech samples enabled the listener subjects to make a significantly greater number of correct identifications for the Age characteristic than they could make for this charac— teristic on the basis of prepared aural cues. This pattern was reversed for the Ethnic Group and Physical Size charac- teristics. Listener subjects performed significantly better under prepared speech sample conditions than under spontaneous speech sample conditions for these two characteristics. This factor did not significantly affect listener subject perfor— mance in correctly identifying the Sex, Education, and Physical Size characteristics. Partial significance was also found under the normal- telephone speech sample condition. For the Age and Dialect Region characteristics, the listener subjects were able to make a significantly greater number of correct identifications under telephone speech sample conditions than while listening to normal speech. This pattern was reversed for the Physical Size characteristic. For this characteristic, listener sub- 188 jects made a significantly greater number of correct identi— fications under the telephone speech conditions. The effects of speech sample duration were measured over all speaker characteristics. Results indicated that the listener subjects were able to make a significantly greater number of correct identifications on the basis of listening to ten second speech samples than they were able to make on the basis of aural cues of five and 15 seconds in length. The interaction between the spontaneous-prepared and normal-telephone speech conditions was significant for four of the speaker characteristics. The use of individual comparison procedures indicated that for the Ethnic Group, Education, and Physical Size characteristics, the listener subjects made a greater number of correct identifications under the treatment combination of prepared normal speech than they did under the other treatment combinations. For the Dialect Region characteristic, individual comparisons for this significant interaction indicated that the listener subjects made a greater number of correct identifications under the treatment combination of prepared telephone speech than under any of the other combinations. In addition, under the condition of prepared normal speech samples, they made significantly fewer correct identifications than they did under the other treatment combinations. The interaction between listener training and speech sample duration also had a significant effect upon listener 189 performance. This effect appeared to be due to a differential result in terms of training with the three durations. Under the five and 15 second durations, the performance of the un- trained listener subjects was significantly lower than was the performance of the trained listener subjects under the ten second duration condition, however, this order did not hold. The untrained listener subjects achieved higher per- formance, although not significantly so than did the trained listener subjects. Test-retest correlation coefficients computed for each characteristic over all treatments and under each individual treatment were low.‘ This indicated that while the test-retest means for each characteristic did not appear different from each other, the individual listener subject reliability with- in each characteristic was poor. Conclusions Within the limits of the dimensions employed within each characteristic and present speaker, testing and training procedures in addition to the use of undergraduate college students as listener subjects, the following conclusions seem warranted: l. Listeners can differentiate accurately between male and female speakers on the basis of a combination of sponta- neous (extemporaneous), prepared (oral reading), normal (simulated face-to-face), and telephone aural cues. 2. Listeners can differentiate accurately among 20 to 30 year old speakers, 40 to 50 year old speakers, and 60 to 190 70 year old speakers on the basis of a combination of spontaneous (extemporaneous), prepared (oral reading), normal (simulated face-to-face) and telephone aural cues.. 3. Listeners can differentiate accurately between Negro and Caucasian speakers on the basis of a combination of spontaneous (extemporaneous), prepared (oral reading), normal (simulated face-to-face), and telephone aural cues. 4. Listeners can differentiate accurately among among speakers with less than a high school education, speakers who are high school graduates, and speakers who are college graduates on the basis of a combination of spontaneous (extemporaneous), prepared (oral reading), normal (simulated face-to—face), and telephone aural cues. 5. Listeners can differentiate accurately between big and small speakers on the basis of a combination of spon— taneous (extemporaneous), prepared (oral reading), normal (simulated face-to—face), and telephone aural cues. 6. Listeners can differentiate accurately among speakers from the Eastern, Southern, and General American dialect regions of the United States on the basis of a combination of spontaneous (extemporaneous), prepared (oral reading), normal (simulated face-to-face), and telephone aural cues. 7. Listeners taken as a group are highly consistent in differentiating between and among the dimensions specified above on the basis of aural cues, but when these same listeners are considered individually, the same conclusion is not 191 warranted. 8. The training of listeners in the task of making the above differentiations on the basis of aural cues does not improve significantly their performance. 9. Differentiations among age categories of speakers can be made more effectively on the basis of spontaneous aural cues (extemporaneous speech) than on the basis of prepared aural cues (oral reading). 10. Differentiations between Negro and Caucasian speakers and between big and small speakers can be made more effectively on the basis of prepared aural cues (oral reading) than on the basis of spontaneous aural cues (extemporaneous speech). 11. Differentiations among the age groups of speakers and among speakers from the three dialect regions of the United States can be made more effectively on the basis of aural cues provided by telephone speech than by the aural cues provided under conditions that simulate the acoustical signal received by listeners during face-to—face conversation. 12. Differentiations between big and small speakers can be made more effectively on the basis of the aural cues provided by simulated face-to-face conversation than by aural cues provided by telephone speech. 13. Aural cues of ten seconds in duration are a more effective means by which to make differentiating judgments relative to the sex, age, ethnic group, education level, physical size, and dialect region of speakers than are aural 192 cues of five, and 15 seconds in duration. Recommendation for Further Research As indicated in Chapter I, one of the limitations imposed upon the present investigation at the onset was that it be broad in nature. This decision was based upon the relatively small amount of information available with regard to listener identifications of the speaker characteristics. in question on a general basis, and specifically, in terms of such identifications on the basis of differing types of aural cues. Obtaining this knowledge was regarded as being a preliminary step that was necessary before the inter- relationships among the speaker characteristics and among the various training, speech and duration factors could be studied in detain. This approach is illustrated by the questions posed in Chaper I and is consistent with the manner in which each was discussed in Chapter IV relative to the results obtained. Thus, the existing data and speech sample stimuli may contain answers to questions that were not considered appropriate or practical in terms of the nature of the present study. Such questions of interest may be: 1. What are the actual similarities and differences between the speech samples collected for each dimension of the Various characteristics and how do these similarities and differences relate to consistent misjudgments of speakers? 2. What are the specific effects of the spontaneous and Prepared recording conditions upon such speech characteristics 193 as grammatical and vocabulary usage, sentence length, speaking length, speaking rate, and so forth? In addition, do trends exist relative to differentiating between the dimensions of each characteristic on the basis of these recording conditions? 3. What are the spectral characteristics of the speech samples recorded under the normal and telephone speech conditions? Do differences that are apparent between speech samples representing different dimensions of a characteristic under the normal speech condition tend to change in some way under the telephone speech condition? 4. What are the specific effects of the three speech sample durations in providing the differentiating vocal features for each speaker characteristic as discussed in Chapter II and discovered by answering the above questions? 5. Is there a consistency of performance for individual listeners in terms of their identifications across and within speaker characteristics and the speech and duration factors? The following suggestions are made in terms of continued research in the area of listener identificaion of characteristics of speakers on the basis of aural cues: l. The present results indicate the need for further research on the effects of listener training as it relates to the identification of speaker characteristics. 2. The results of the present investigation also suggest a need for clarifying the demonstrated interrelationship be- tween the normal-telephone and spontaneous-prepared speech 194 treatment combinations with regard to listener identification. The present results in terms of test—retest reliability indicate the need for further study. This factor is of vital importance with regard to the applicability of results. 3. The present results need to be substantiated by further research on a more detailed basis. Specifically, this substantiation should include the use of greater number of speakers representing the various characteristics and should investigate the possibility of listeners making more precise differentiations within each characteristic. 4. As indicated in Chapter I, as more information becomes available with respect to listener identification of individual characteristics, the need will be evident for studying the interrelationships among such characteristics and the effects of various combinations upon listener performance. 5. The relationship between listener performance in differentiating between speakers on the basis of acoustic cues and the performance of persons in making such diff- erentiations on the basis of visual cues as in voice printing is seen as essential in terms of specifying the important cues for success in each task. 6. Finally, the need for continued study of the Ethnic Group and Dialect Region characteristics is urged from the standpoint of the expected changing nature of the differenti- ating vocal features within each. The increased tendency 195 for travel and the reduction of geographical iminations in terms of family location will slowly eliminate the distinctive features of the speech regions of the United States. In addition, the recent and continued activity in the interest of providing equal rights and opportunities for all citizens is expected to be influential in modifying the speech as well as other cultural differences between Negro and Caucasian individuals. BIBLIOGRAPHY BIBLIOGRAPHY Books Akin, Johnnye. And So we Speak: Voice and Articulation. Englewood Cliffs, NTJ.: Prentice-Hall, 1958. Barbara, Dominick A. You; Speech Reveals Yggr Personality. Springfield, 111.: Charles C. Thomas, 1958. Baugh, Albert C. .gistopy of the English Language. New York: AppletoneCentury, 1935. Bell, Him. Adjustment Inventory. Stanford: Stanford University Press, 1934. Bernreuter, R. Personality Inventory. Stanford: Stanford University Press, 1931. Black, Hohn W. and Moore, Wilbur E. Speech: Code, Meaning, and Communication. New York: McGraw-Hill Book Co., 1955. Blalock, Hubert M, Social Statistics. New York: McGraw-Hill Book Co., 1960. Fairbanks, Grant. Voice and Articulation Drillbook. New York: Harper and Brothers, 1940. Fisher,'Hi1da B. Improving VOice and Articulation. Boston: Houghton Mifflin Co., 1966. Fletcher, Harvey. Speech and Heaping in Communication. New York: McGraw-Hill Book Co., 1955. Gray, Giles Wilkerson and Wise, Claude Merton. The Bases of Speech. New York: Harper and Brothers, 1959. Hahn, E. E; 31. Basic Voice Trainingjfor Speech. New York: McGraw—Hill Book Co., 1957. Hanley, Theodore D. and Thurman, Wayne.L. Developing Vocal Skills. New York: Holt, Rinehart, and Winston, 1963. Hays, William L. Statistics for Psychologists. New York: Holt, Rinehart, and Winston, 1963. Kingsley, Howard L. The Nature and ConditiOns of Learning. Englewood Cliffs, N. J. : Prentice-Hall, 1957. 197 198 Kurath, Hans. Linguistic Atlas of the United States. . V Providence, R.I.: Brown University Press, 1939—1943. Lojos, Egri. Your Key to Successful Writing. New York: Henry Holt Co., 1952. McNemar, Quinn. Psychologicgl Statistics. New York: John Wiley and Sons, 1962. Mednick, Sarnoff A. Learning. Englewood Cliffs, N.J.: PrenticfirHall, 1964. f Murray, Elwood. The Speech Personality. Chicago: J.B. Lippincott Co., 1944. Pear, T.H. Voice and Personality. London: Chapman and Hall, 1931. Peters, Charles C. and VanVoorhis, Walter R. Statistical Procedures and Their Mathematical Bases. New York: McGraw-Hill Book Co., 1960. Postman, Leo and Egan, James P. Experimental Psychology. New York: Harper and Brothers, 1949. Sullivan, H.S. The Psyghiatric Interview. New York: Norton, 1954. Winer, B.J. Statistical Principles in Experimental Design. New York: McGraw-Hill Book Co., 1962. Wise, Claude Merton, Introduction to Phonetics. Englewood Cliffs, N.J.: Prentice-Hall, 1958. Articlesjand Periodicals Ackelsber, Sylvia A. "Vocabulary and Mental Deterioration in Senile Psychosis," Journal of apnormal and Social Psychology, XXXIX (October, 1944), pp. 393-406. Ainsfield, M., Bogs, N., and Lambert, W. "Evaluational Reactions to Accented English Speech," Journal of Abnormgl Psychology, LXV (1962), pp. 223-231. Allport, Gordon W. and Cantril, Hadley. "Judging Personality from the Voice," Joggnal of Social Psychology, V (February, 1934), pp. 37—55. Allport, Gordon W. andiKramer, Bernard M. "Some Roots of Prejudice," Journal of_Psychology, XXII (1946), pp. 9-39. 199 Altus, W.D. "Sexual Role, the Short Story, and the Writer," Journal of Psychology, XLVII (January, 1959), pp. 37-40. "Inferring the Sex of an Author," Journal of Psychology, XLVIII (October, 1959), pp. 215-218. Back, A.C., Lederer, Frances C., and Dinolt, R. "Senile Changes in the Laryngeal Musculature," Archives of Otolaryngolgy, XXXIV (July, 1941), pp. 47-56. "Bees," The WOrld Book Encyclopedia, II (1964), p. 154. Bernard, B. "Some Sociological Determinents of Perception: An Enquiry into sub-Cultural Differences," British Journal of Sociology, IX (1958), pp. 159-174. Bernstein, Basil. "Language and Social Class," British Journal of Sociology, XI (1960), pp. 271-276. "Linguistic Codes, Hesitation Phenomina, and Intelli- gence," Language and Speech, V (1962), pp. 31-45. "Social Class Linguistic Codes and Grammatical 1} Elements,“ Language and Speech, V (1962), pp. 221-240. "Birds," The WOrld Book Encyclopedia, II (1964), p. 250. Birren, James E., Riegel, K.F., and Robbin, J.S. "Age Deficiencies in Continuous WOrk Associations Measured by Speech Recordings," Journal of Gerontology, XVIII (January, 1962), pp. 95-96. Blake, Robert R. "The Relationship Between Childhood ' Environment and the Scholastic Aptitude and Intelligence of Adults," Journal of Social Psychology, XXIX (February, 1949), pp. 37-41. Bloomer, Harlan. "Communication Problems among Aged County Hospital Patients," Gereatrics, XV (April, 1960), Boshoff, P.H. "The Anatomy of the South African Negro Larynges," .§outh African Journal of Medical Sciencez X (February, 1945), pp. 35-40. Britton, Joseph H. "Influence of Social Class upon Performance on the Draw—a-Man Test," Journal of Educational Psychology, XLV (January, 1954), pp. 44-51. 200 Brown, Mary Marrow. "A Study of Performance on a Deterioration Test Related to Quality of Vocabulary and Rigidity," American Psycholoqigp, III (October, 1944), pp. 393-406. Bryan, Alec I. and Wilke, Walter H. "A Technique for Rating Public Speeches." Journal of Consulting Psychology, V (March-April, 1941), PP. 80-90. Carlson, J. Spencer, Cook, Stuart W., and Stromberg, Elroy L. "Sex Differences in Conversation," Journal of Applied Psychology, XX (1939), pp. 727-735. Carson, Arnold S. and Rabin, A.I. "Verbal Comprehension and Communication in Negro and White Children," Journal of Educational Psychology, LI (April, 1960), pp. 47-51. Christian, Alice M. and Patterson, Donald G. "Growth of Vocabulary in Later Maturity," Journal of Psychglogy I (1936), Pp- 167-169. Cohen, Alan and Starkweather, John A. "Vocal Cues to the Identification of Language," American Jogrnal of Psychology, LXXIV (March, 1961), pp. 90—93. Curry, E.T. "The Pitch Characteristics of the Adolescent Male Voice," Speech Monographs, VII (March, 1940), pp. 48-62. Diehl, Charles P. "Voice and Personality," Psycholigical and Psychiatric Aspects of Speech and Hearing, ed. Dominick Barbara. Springfield, 111.: Charles C. Thomas, 1960. Dimitrovsky, Lilly Sprecker. "The Ability to Identify the Emotional Meaning of Vocal Expression at Successive Age Levels," Dissertation Abgtracts, XXIV (January, 1964), PP. 2983-2984. Eckert, R.G. and Keys, N. "Public Speaking as a Clue to Personality Adjustment," Journal of Applied Psycholggy, XXIV (January, 1940), pp. 144-153. Fairbanks, Grant and Pronovost, Wilbert. "An Experimental Study of the Pitch Characteristics of the Voice 5 During the Expression of Emotion," Speech Monographs, VI (December, 1939), pp. 87-104. 201 Fay, Paul J. and Middleton, Warren C. "Judgment of Kretschmerian Body Types from the Voice as Trans- mitted over a Public Address System," Journal of Social Psychology, XXI (January, 1940), pp. 151-162. "Judgment of Intelligence from the Voice as Trans- mitted over a Public Address System," Sociometry, III (April, 1940), pp. 186-191. "Judgment of Leadership from the Transmitted Voice." Journal of Social Psychology, XVII (1943), pp. 99-102. "Fish," The WOrld Book Encyclopedia, VII (1964), p. 139. "Flowers," The WOrld Book Encyclopedia VII (1964) p. 234. Fox, Charlotte. "Vocabulary Ability in Later Maturity," Journal of Egucational Psychology, XXXVIII (December, 1947). PP. 482-492. Fox, Charlotte and Birren, James E. “Some Factors Affecting Vocabulary Size in Later Maturity: Age, Education, and Length of Institutionalization," Journal of Gerontology IV (January, 1949), pp. 19-26. Gleser, Goldine C., Gottschalk, Louis A., and Watkins, John. "The Relationship of Sex and Intelligence to Choice of WOrds: A NOrmative Study of Verbal Behavior," Journal of Clinical Psychology, XV (April, 1959), pp. 182-191. Harbold, George J. “Fitch Ratings of Voiced and Whispered Vowels," Journal of the Acoustical §gciety of America, xxx (July, 1958), pp. 600-601. Hollien, H. "Vocal Pitch Variation Related to Change in Vocal Fold Length," Journal of Speech anngegging Research, III (June, 1960), pp. 150-156. Hollien, H. and Curtis, James F. "A Laminographic Study of Vocal Pitch," Journal of Speech and Hearing Research, III (December, 1960), pp. 361-371. Hollien H. and Malick, Ellen. "Adolescent Voice Change in Southern Negro Males," Speech Monographs, XXIX (March, 1962), pp. 53-58. Hollingworth, L.S. "What We Know About the Early Selection and Training of Leaders," Teachers College Record, XI (April, 1939), PP. 575-592. 202 Hunt, W} . g; §l° "The Clinical Possibilities of an Abbreviated Individual Intelligence Test," Journal of glinical Psychology, XII (February, 1948), PP- 171-173. Knower, Franklin. "A Study of Rank-Order Methods of Evaluating Performance in Speech Contests," Journal of Applied Psychology, XXIV (October, 1940), pp. 633-644. Kramer, Ernest. "Personality Stereotypes in Voice: A Reconsideration of the Data," Journal of Social Psychology: LXII (April, 1963), PP. 247-251. "Judgment of Personal Characteristics from Non- Verbal Properties of Speech," Psychological Bulletin, LX (July, 1963), PP. 408-420. Kurath, Hans. "Dialectical Differences in Spoken English," Modern Philology, XXVI (May, 1928), pp. 386-393. Landis, C. "National Differences in Conversation," Journal of Abnormal and Social Psyohology, XXI (January-March, 1927). pp. 354-357. Landis, H.H. and Burtt, H.E. "A Study of Conversations," Jouppal of Comparative Psychology, IV (February, 1924). pp. 81-89. Lawten, Denes. ”Social Class Differences in Language Development: A Study of Some Samples of Written WOrk," Language and Speech, XI (1963), pp. 120-143. Licklider, J.C.R. and Miller, George A. "The Perception of Speech," Handbook of Experimental Psychology, ed. S.S. Stevens. New York: John Wiley and Sons, 1951. McGehee, F. "An Experimental Study of Voice Recognition," Journal of General Psychology, XXXI (January, 1944), pp. 53-65. McGlone, RObert C. and Hollien, Harry. I'Vocal Pitch Charac- teristics of Aged WOmen," Journal of Speech and HearingoResearch, VI (June, 1963), pp. 164-170. McKelvey, D.P. "Voice and Personality," western Speech, XVII (1953), pp. 91-94. Meyer-Eppler, W. "Realization of Prosodic Features in- Whispered Speech," Journaljof the Acogstical Society of Americay XXIX (January, 1957), pp. 104-106. 203 Mitchell, Joyce. "Speech and Language Impairment in the Older Patient," Geriatrics, XIII (July, 1958), pp. 467-476. Moore, Henry T. "Further Data Concerning Sex Differences," Journal of Abnormal and Social Psychology, XVII (July-September, 1922), pp. 210-214. Morrison, Sheila. "The Severity of Articulation Defectiveness,“ Journal of Speech and Hearing Disorders, XX (December, 1955), pp. 347-351. Murray, Elwood. "A Study of Factors Contributing to the Mal-Developing of the Speech Personality," Speech Monographs, III (September, 1936), pp. 95-108. Mysak, Edward D. "Pitch and Duration Characteristics of Older Males," Journal of Speech and Hearing Research, II (March, 1959), pp. 46—54. Penny, R. "Age and Sex Differences in Motivational Orien- tation to the Communicative Act," Child Development, XXIX (December, 1958), pp. 163-171. Perrin, Elinor Horwitz, "The Rating of Defective Speech by Trained and Untrained Observers," Journal of Speech and Hearing Disorders, XIX (March, 1954), pp. 48-51. Pronovost, Wibert L. "An Experimental Study for Determining Natural and Habitual Pitch," Speech Monographs, IX (1942), pp. 111—123. Ptacek, Paul H. and Sander, Erik K. "Age Recognition from Voice," Journal of Speech and Hearing Research, IX (June, 1966), pp. 273-277. Ptacek, Paul H°.§E.§l' "Phonatory and Related Changes with Advanced Age," Journal of §peech and Hearing Research, IX (September, 1966), pp. 353-360. "Rain," The WOrld Book Encyclopedia, XVI (1964), p. 122. Reynolds, R.G. "Factors of Leadership Among Seniors of Central High School, Tulsa, Oklahoma," Journal of Educational Research, XXXVII (January, 1944), pp. 356-361. Ricks, James H. "Age and VOOabulary Test Performance: A Qualitative Analysis of the RespOnses of Adults," Dissertation Apstracts, XIX (July, 1958), p. 182. 204 Rosenthal, Fred. "Some Relationships Between Socio-Position and Language Structure of Young Children," Journal of Educational Psyehology, XLVII (December, 1957), pp. 483-497. Schulman, Mary Jean and Havinghurst, Robert J. "Relations Between Ability and Social Status in a Midwestern Community. IV: Size of Vocabulary," Journal of Educational Psychology, XXXVIII (N0vember, 1947), pp. 437-442. Sewell, William H., Haller, Archie 0., and Straus, Murray A. "Social Status and Educational and Occupational Aspiration," American Sociological Review, XXII (February, 1957), pp. 67-73. Shakow, D., Dolkart, Marjorie, B., and Godlman, Rosaline. "The Memory Function in Psychosis of the Aged," Disorders of the Nervous System, II (January, 1941), pp. 3-8. Shakow, D. and Godlman, Rosaline. "The Effect of Age on Stanford-Binet Vocabulary Scores of Adults," Journal of Educational Psychology, XXIX (April, 1938), pp. 241-256. Sherman, Dorothy and Morrison, Sheila. "Reliability of the Individual Ratings of Severity of Defective Articulation," Journal of Speech and Hearing Disorders, XX (December, 1955), pp. 352-358. Silverstein, B°.§E.§1- "The Relative Intelligibility of Male and Female Talkers," Journal of Educational Psychology, XIX (November, 1953), pp. 418-428. Snidecor, John C. ”The Pitch and Duration Characteristics of Superior Female Speakers during Oral Reading," Journal of Speech and Hearing Disorders, XVI (February, 1951), pp. 44-52. Sorenson, H. "Mental Ability over a Wide Range of Adult Ages," Journal of Applied Psychology, XVII (1938), pp. 729-741. "Stars," The WOrld Book Encyclopedia, XVII (1964), p. 660. Stenquist, John L. and Lorge, Irving. "Implications of Intelligence and Cultural Differences; As Seen by a Test-User; As Seen by a Test-Maker," Teachers College Record, LIV (January, 1953), pp. 184-193. 205 Stoke, Stuart M. and west, Elmer D. "Sex Differences in Conversational Interests," Journal of Social Psyehology, II (February, 1931), pp. 120-126. Tyler, Leona. "Test Review Number 77,” The qurth Mental Measurements Yearbook, ed. 0. Buros. New Jersey: Gryphon Press, 1953. WOlfle, Doel. "Training," Handbook of Experimental Psychology, ed. S.S. Stevens. New York: Wiley and Sons, 1951. "Military Training and the Useful Parts of Learning Theory," Journal of Applied Psychology, (1946), pp. 73-75. Wise, C.M. "Negro Dialect," Iguarterly Journal of Speech, XIX (November, 1933), pp. 522-528. Reports Glick, Paul C. and Miller, NOrman P. "Educational Level and Parental Income," College Board Review, No. 32, 1957, pp. 29-32. Havinghurst, Robert J. "Culturexand the I.Q.," Purdue University Studies in Higher Education, No. 69 (1949), pp. 42-53. Johnson, A.P. "A Study of the English Vocabulary Scores of 75 Executives," Technical Report of the Human Engineering Laboratory, No. 2 (1935), p. 16. Society of Actuaries. Build and Blood Pressure Study. Volume I, Chicago: Society of Actuaries, 1959. APPENDIX A PARAGRAPHS USED SPEECH STIMULI 207 Rain247 Rain is a necessity for all life. When city dwellers mutter objections during a shower or thunderstorm, they do not realize that millions of both men and animals are gladdened by the falling drops. When rain falls after a long dry spell, all nature undergoes a revival of life. Withering plants and trees become green again, and farmers who had been afraid they would lose their crops have hopes for a harvest. Streams baked almost dry by the summer sun, refill and make it possible for the fish and other water animals to live and multiply. Birds drink gladly from the pools of water. But there can be too much rain. Then swollen rivers overflow their banks. The waters rush over the river valleys, uprooting trees, sweeping away homes and buildings, destroying crops, and causing loss of life. But generally a renewal of life follows rain. Wherever there is plenty of rain during the growing season, life of all kinds is abundant. Where little or no rain falls, the land is barren and there are few forms of life. Great deserts form when such a lack of water continues. Rain is always falling somewhere on the earth. The record for the entire surface of the earth, according to computations at thousands of widely scattered weather 247"Rain," The WOrld Book Encyclopedia, XVI (1964), p. 122. 208 stations, is 45,000 thunderstorms a day, or 1,800 an hour. The island of Java is the world's most thunderous spot. It has no fewer than 223 storm days a year. 209 Fish248 There are more than thirty thousand kinds of fish. They live in the oceans, lakes and streams from the equator to the polar seas, and from elevations of more than fifteen thousand feet above the sea to depths at least that far below sea level. There are few waters that do not contain at least one kind of fish. Fossils show that this has been true for many millions of years. Fish have a great variety of forms and habits because they live in many different places. The appearance, body structure, and life ways fit each kind of fish to the particular place where it lives. In the open sea, for examples, most fish are hard to see because they are colored metalic blue-green above and silvery below. They are almost perfectly stream- lined. There are two reasons for this. In such open waters the kinds of fish that are eaten have no place in which to hide. They must swim swiftly to avoid being eaten. The kinds that eat other fish must also swim swiftly in order to catch their prey. Fishes that live near the bottom of bodies of water are colored like the bottom. Many of them are eel-shaped so that they can slip into burrows or cracks. Others cannot move fast because they have developed a 248"Fish," The WOrld Book gpcyclopedia, VII (1964), p. 139. 210 bony armor or strong spines. This covering is their protection. Fish that live in seaweeds are often marvelously camouflaged. They may even resemble the seaweeds in shape and movement, as well as in color. Many fish that live in the dark, deep sea carry their own lights. 211 Birds249 The bird is an animal with feathers. Birds live in all parts of the world. Some make their homes in the cold places near the North Pole. Others live in the hot, green jungles of South America and Africa. Some birds live in fields and some in the mountains. Others stay near water. The colors of some birds brighten the places where they live. The feathers of other birds are dull and help to hide them. All birds have wings, but some, such as the ostrich, cannot fly. Birds have many ways of life. Birds called swifts spend most of their waking hours flying.‘ The penguin waddles on the ice and swims in the ocean, but cannot fly. Many birds migrate, or travel long distances, at certain seasons. For this reason, many kinds of birds you see in the summer leave for the winter. They return the next summer to court their mates and build their nests. The graceful flight of birds and their beautiful colors and sweet songs have inspired artists, poets, and musicians. Inventors studied birds for hundreds of years before man, too learned to fly in gliders and airplanes. Every bird hatches from an egg, and has two legs and a bill, or beak. Birds have built—in "air-conditioning." That is, some of their bones are hollow and their bodies have air sacs, or pockets. The hollow bones also make birds 249"Birds," The world Book Encyclopedia, II (1964), p. 250. 212 lighter so they can fly easily. Many birds are valuable to man. Chickens and other poultry provide meat and eggs for food. Birds help the farmer by eating insects that attack his crops. A few birds eat the farmer's grain and fruit- But the good that birds do, more than makes up for the damage they cause . 213 Flowers250 The flowers, or blossoms, of plants and trees bloom almost everywhere on earth. Some flowers grow on high mountains at the edges of snow fields and glaciers. Others live in the shallow parts of oceans. Even hot, dry deserts have many bright blossoms during and after the rainy season. Most flowers need soil in which to grow, but some can grow on tree brances. Others float on lakes and streams. About the only places flowers do not grow are in the ice—covered parts of the Arctic and Antarctic and in the open seas. The word "flower" may mean either (1) the blossom or (2) the whole plant. Botanists, scientists who study plants, use the word flower to mean only the blossom of a plant. They call the whole plant - blossom, stem, leaves, and roots— a flowering plant. Any plant that produces some sort of flower, even a tiny, colorless one, is a flowering plant. Grasses, roses, lillies, apple trees, and oaks are all flowering plants. Flowers are the reproductive parts of flowering plants. The plants could not develop seeds and reproduce without them. Man depends completely on flowers and flowering plants for his food. Flowering plants include almost all 250"Flowers," The world Book Encyclopedia, VII (1964), p. 234. 214 of our grains, fruits, and vegetables. Even the animals that we use for food, such as cattle, hogs, and sheep live on flowering plants. 215 Stars251 The stars are suns. Some of the stars are bigger than our own sun, and some are smaller and fainter. Our sun seems so much brighter and larger than all other stars simply because it is much nearer to us than any of the rest. Our sun in only about 93 million miles away. Yet it is far enough away that a rocket from earth traveling 25 thousand miles an hour, or seven miles a second, would take 152 days and eight hours, or about five months, to reach the sun. But the nearest star except for our sun is so far away that our seven miles a second would take almost 115 thousand years to reach it. Even this star is a close neighbor, as stars go. Others are millions of times farther away. we see two kinds of star-like objects in the sky. One is the stars themselves. The other is the sun's "family of plants" that circle the sun as the earth does. Since very early times man has looked up at the stars and wondered about these diamond-like points of light that shimmer above him. Some ancient carvings show men who lived 5,000 years ago studied the heavens. But ancient men probably had no idea what the stars are really like. They did not dream that the stars are other suns, far out in space. They thought that stars made the outlines of animals or persons in the sky, and they called 251"Stars," The World Book Encyclopedia, XVII (1964), p. 660. 216 these shapes constellations. The ancient Greeks thought that one group of stars looked like a winged horse, which they named Pegasus. Other groups of stars were named after other animals or persons, and these animals and persons became part of the legends and folk tales that have come down to us through thousands of years. 217 3868252 The bee is an insect that lives in almost every part of the world except near the North and South Poles. There are ten thousand species of bees, but only honeybees make honey and wax that man can use. Bees are the only insects that produce food eaten by man. we use the wax from the nests of bees in making such products as candles and lipsticks. We use their honey in cooking and as a sweet spread on bread. When bees fly from flower to flower, they help both man and the blossoms they visit. Many fruits and vegetables would die if bees did not help fertilize flowers. Bees gather nectar and pollen from flowers. They make honey from the nectar and use the honey and pollen as food. Some people are afraid of bees because they sting. But bees do not sting unless they are fightened or hurg. Like most other insects, bees have three pairs of legs and four wings. A bee has a special stomach, called a honey stomach, in which it carried nectar to the nest. Honeybees are social insects. They live and work together in large groups. They form a colony, or group, of thousands of bees. A single honeybee can live alone only a few days, but the colony can go on living for many years. 252"Bees," The WOrld Book Encyclopedia, II (1964), p. 154. 218 One worker can do little by itself, but the many thousands of workers in a colony, working as a group, can do many things. They fly into the fields and woods to gather food and water. They build their own home in a box, a hollow tree, or a bee hive. They store honey and pollen and eat it in winter, just as squirrels eat the nuts they store. Honeybees even air-condition their hive to keep it warm or cool. APPENDIX B CONSTRUCTION OF THE MASTER TAPES Master Tape I Speaker Identification Number Char. # Cond.~ Characteristic Dimension 1 PS 3 Spon—Tel Physical Size Big 2 DR 1 Spon-Tel Dialect Region Michigan 3 DR 7 Spon-Tel Dialect Region East 4 E 3 Prep—Nor Education -High School 5 DR 2 Prep—Nor Dialect Region Michigan 6 A 14 Prep—Nor Age 60—70 7 E 7 Prep-Nor Education High School 8 DR l3 Spon-Nbr Dialect Region South 9' EG 2 Spon—Tel Ethnic Group .CaucaSian 10 PS 7 Prep—Nor Physical Size Small 11 EG l YPrep—NOr Ethnic Group Caucasian 12 A 8 Prep-Nor Age 40—50 13 DR 2 Spon—NOr Dialect Region Michigan 14 ES 3 Spon-Nor Ethnic Group Caucasian 15 PS 7 Prep-Tel Physical Size- Small 16 DR 8 Prep-Nor Dialect Region East 17 PS 8 Spon-Tel Physical Size. “Small 18 EG 8 Spon—Tel Ethnic Group; Negro 19 A 7 Spon-Tel Age A 40-50 20 DR 8 Spon-Nor Dialect Region East 21 PS 3 Spon-NOr Physical Sizel Big 22 E 8 Prep—Nor Education High School 221 :5::::r Chaidentificagizg. Characteristic Dimension 23 PS 2 Prep-Nor. Physical Size Big 24 EGO l Prep-Tel. Ethnic.Group Caucasian 25 E 15 Prep—Nor Education College 26 PS 7 Spon-Nor Physical Size Small 27 E 7 Spon—Nor Education High School 28 E l Spon—Tel Education -High School 29 DR 2 Prep—Tel Dialect Region Michigan 30 E 14 Spon-Tel Education College‘f 31 EG 9 Spon—Nor Ethnic Group Negro 32 DR 8 Prep-Tel Dialect Region East 33 E 15 Prep-Tel Education College 34 DR 14 Prep-Tel Dialect Region South 35 A 13 SpOn—Tel Age 60-70 36 S 9 Spon—Nor Sex Female 37 S 7 Prep—Nor Sex Female 38 EG 8 Prep—Tel 'Ethnic Group Negro 39 A 2 Spon—NOr Age 20—30 40 S 3 Prep-Nor Sex ,Male 4l A , 13 Spon—Nbr Age 70-70 42 E , 2 Prep—Tel Education -High School 43 DR 15 Spon—Tel Dialect Region, South 44 S' 9 Prep-Tel' Sex Female 45 S‘ 2 Prep—Tel Sex Male' 222 SEEEZEr Chaidentfificagiig. Characteristic Dimension 46 E 3 Spon-Nor Education -High School 47 A l Prep—NOr Age 20—30 ‘48 E6 8 Prep-Nor Ethnic Group Negro 49 A 2 Spon—Tel Age 20—30 50 A 14 Prep-Tel Age 60—70 51 S 8 Spon-Tel Sex ~Female 52 A 8 Spon—Nor Age 40—50 53 S l Spon—NOr Sex Male 54 A l Prep-Tel Age 20-30 55 A 8 Prep—Tel Age 40-50 56 S 2 Spon-Nor Sex Male 57 DR 15 Prep-Nbr Dialect Region South 58 PS 2 Prep-Tel Physical Size Big 59 E 7 Prep—Tel Education High School 60 E 13 Spon-Nor Education College 223 Master Tape II giggzir Chggentifiicatéggd. ; Characteristic Dimension 1 DR 9 Spon-Tel Dialect Region East 2 E 7 Spon-Te1. Education High School '3 A 3 Prep-NOr Age 20-30 4 PS 8 Prep-Tel Physical Size Small 5 DR 7 Spon—Nor Dialect Region East 6 PS 1 Spon-Tel Physical Size Big 7 S 2 Spon—Tel Sex Male 8 A 7 Prep-Nor Age 40-50 9 EG 3 Spon-Tel Ethnic Group Caucasian 10 E 2 Prep—Nor Education -High School 11 EG 8 Spon-NOr Ethnic Group Negro 12 PS 9 Prep-Nor Physical Size Small 13 DR 1 Prep-Nor Dialect Region Michigan 14 DR 7 Prep-Tel Dialect Region East 15 E 14 Prep—Tel Education College 16 A l Spon-Tel Age 20—30 17 S 9 Spon-Tel Sex Female 18 E 3 Spon—Tel Education -High School __119 DR 15 Spon-NOr Dialect Region South __¥20 PS 3 Prep—Tel Physical Size Big __¥21 PS 3 Prep-Nor Physical Size Big ___22 EG 9 Prep-Tel Ethnic Group Negro giggzir Chigentiiicatéggd. Characteristic Dimension 23 E 15 Spon—Tel Education College 24 S 7 Spon-Nor Sex Female 25 A 15 Spon-Nor Age 60-70 26 S 9 Prep-Nor Sex Female 27 S 8 Prep-Tel Sex Female 28 EG 9 Prep-Nor Ethnic Group Negro 29 PS 1 Spon—Nor 'PhySical Size Big 30 A 9 Spon—Nor Age 40-50 31 DR 14 Spon-Tel Dialect Region South 32 S l Prep-Nor Sex Male 33 E 13 Prep—Nor Education College 34 E 14 Spon-Nor Education College 35 A 2 Prep-Tel Age 20—30 36 A .l Spon-Nbr Age 20-30 37 DR ;1 Spon—Nor Dialect Region Michigan 38 EG 1 Spon-Nor Ethnic Group Caucasian 39 E 8 Prep—NOr Education High School 40 DR 13 Prep-Nor Dialect Region South 41 A 8 Spon-Tel Age 40-50 42 S l Prep—Tel Sex Male 43 E 3 Prep-Tel Education High School 44 A 7 Prep-Tel Age 40—50 45 DR 1 Prep—Tel Dialect Region Michigan 225 gsggzir Chigentiiicatéggd. Characteristic Dimension 46 A 13 Prep-Nor Age 60—70 47 S 3 Spon—Nor Sex Male 48 E l Spon—Nor Education —High School 49 DR 9 Prep-NOr Dialect Region East 50 DR 3 Spon-Tel Dialect Region Michigan 51 EG 2 Prep—Nor Ethnic Group Caucasian 52 A 15 Spon—Tel Age 60-70 53 PS 7 Spon-Tel Physical Size Small 54 EG 7 Spon-Tel Ethnic Group Negro 55 E 8 Spon-Nor Education High School 56 EG 2 Prep-Tel Ethnic Group Caucasian 57 DR 15 Prep—Tel Dialect Region South 58 A 15 Prep—Tel Age 60—70 59 E 8 Prep-Tel Education High School 60 PS 8 Spon-Nor Physical Size Small Master Tape III Identification 35:22:: Char. # Cond. Characteristic Dimension 1 PS 8 Prep—NOr Physical Size: Small 2 DR 3 Spon—Nor Dialect Region Michigan 3 EG 7 Prep—Tel Ethnic Group Negro 4 A 9 Prep-Nor Age 40-50 5 E 9 Spon-Nor Education High School 6 PS 9 Spon—Tel Physical Size Small 7 DR 2 Spon—Tel Dialect Region Michigan 8 E l Prep—Tel Education -High School 9 PS 1 Prep—Nor Physical Size .Big 10 E 13 Prep—Tel Education College 11 E 9 Prep—Tel Education High School 12 A 2 Prep-Nor Age 20-30 13 A 9 Spon-Tel Age 40-50 14 A 14 Spon-Nor Age 60-70 15 S 7 Prep—Tel Sex Female 16 EG 2 Spon-NOr Ethnic Group Caucasian 17 E6 3 Prep-Tel Ethnic Group Michigan 18 DR 3 Prep-Tel' Dialect Region Michigan 19 S 3 Prep-Tel Sex Male 20 DR 13 Spon-Tel Dialect Region South 21 A 13 Prep—Tel Age 60-70 22 S 8 Prep-Nor Sex Female 227 Efigggir . Chggentifiicatéggd.‘ Characteristic Dimension 23 DR 7 Prep—Nor Dialect Region East 24 ‘E 1 Prep—Nor Education -High School 25 A 14 Spon-Tel Age 60-70 26 A 3 Prep—Tel Age 20-30 A 27 EG 9 Spon-Tel Ethnic group Negro 28 S 3 Spon-Tel Sex Male 29 PS 2 Spon—Tel Physical Size Big 30 S 2 Prep—NOr Sex Male 31 DR 9 Prep—Tel Dialect RegiOn; East 32 DR 3 Prep-NOr Dialect Region Michigan 33 EG 3 fPrep—Nor Ethnic Group Caucasian 34 PS 9 Spon-Nor Physical Size Small 35 DR 13 Prep—Tel Dialect Region South 36 PS 9 Prep-Tel Physical Size Small 37 PS 1 Prep—Tel ~Physical Size Big 38 E 9 Spon—Tel Education High School 39 DR 9 Spon-Nor Dialect Region East 40 EG 7 Prep—NOr Ethnic Group Negro 41 A 7 Spon-NOr Age 40-50 42 E 14 Prep-Nor Education College 43 A 9 Prep—Tel Age. 40—50 44 EG 7 Spon—NOr Ethnic Group Negro 45 E 2 Spon-Tel Education -High School 228 A Speaker Identification Number Char. # Cond. Characteristic Dimension 46 EG l Spon-Tel Ethnic Group Caucasian 47 A 15 Prep-Nor Age 60-70 48 E 9 Prep-Nor Education High School 49 DR 14 Prep—Nor Dialect Region South 50 S 7 Spon-Tel Sex Female 51 E 15 Spon—Nor Education College 52 DR 14 Spon-Nor Dialect Region South 53 S l Spon—NOr Sex Male 54 DR 8 Spon-Tel Dialect Region East 55 PS 2 Spon-Nor Physical Size Big 56 A 3 Spon-Nor Age 20-30 57 E 2 Spon-Nor Education —High School 58 S 8 Spon—Nor Sex Female 59 'A 3 Spon-Tel Age 20-30 60 E 13 Spon—Tel Education College APPENDIX C TEXT OF THE TRAINING PROGRAM AND PRACTICE TEST 230 My name is Pat Nerbonne. I am a doctoral student in the Department of Speech and am presently doing my disser- tation--which is why I am here with you today. What we will be doing should be beneficial to both of us. For me, you will serve as subjects'and furnish the data for the experiment I am doing. For you, I hope to be able to furnish a good start toward your becoming critical listeners. This is necessary before one can improve his voice and articulation, and of course, this is what Speech 108 is all about. What I am trying to find out is whether or not persons, such as yourselves, can, after listening to someone talk, identify some particular characteristic about him. Specifically, I want to know if you can tell me: (1) what sex a person is; (2) his age; (3) what ethnic group he be- ongs to; (4) the extent of his formal education; (5) some- thing about his physical size, that is, whether he is "big" or "small;" and (6) what part of the country he comes from-— the South, the East, or right here in Michigan. There are some other factors involved too that might change the extent to which you will be able to make these identifications. These have to do with the kind of speech you will be hearing. Specifically, you will hear four kinds of speech. The differ- ence between two of these is due to the way they were re- corded. One was recorded in a standard way, using a micro- phone and tape recorder. The other, however, was recorded from a telephone. The speech from these two kinds of_;'“ffli 2.31.. recordings will sound different to you. The telephone speech will probably seem tinny and thin compared to the speech re- corded in the usual way. The difference between the other two kinds of speech you will hear is due to the fact that one is oral reading while the other is spontaneous, off the cuff speech. Oral reading is usually smoother and less jerky than off the cuff speech, but it is not as expressive. The fact that you will be hearing speech under these various conditions will not make your job any different. Your task will be the same for every speaker you hear--to indicate what sex he or she is, what age group he belongs to, and so on. YOur job will be made easier by the fact that you will only judge each speaker you hear on one thing. This is indicated on the practice answer forms you have. On the first page you can see under the column labeled "character- istic" that you are being asked to evaluate only what sex each speaker is. This is the way all of the answer sheets are except the last few pages. On the last few pages, the "characteristic” column is all mixed up. But, even on these pages, you will only judge each speaker on one characteristic-— the one listed for him or her. Now, let's get started with a little practice. Refer to page one of your answer sheet. Listen to the first speaker and then circle the appropriate sex, that is, "maleP or "fe- male," under the "dimension" column on the right hand side of your answer sheet. 232 Five, ten, or 15 sec sample of sex speaker 4 (male), spontaneous normal condition. That was a male speaker. How many of you circled "male?" How many circled "female?" The main difference be- tween male and female speakers, of course, is the pitch of the voice. Listen to the second speaker and circle what you think is the appropriate sex. Five, ten, or 15 sec sample of sex speaker 10 (female), spontaneous normal condition. That speaker was a female. How many circled "male?" Besides the pitch of the voice, there are other ways to differentiate between male and female speakers. It is usual- ly easier to understand males, for example. Their speech is more intelligible. Males also use a longer syllable duration. This makes it seem as though females talk faster than males. Listen to the next speaker now and circle what you think is the appropriate sex. Five, ten, or 15 sec sample of sex speaker 5 (male), prepared normal condition. That speaker was a male. How many circled "female?" Okay, now let's listen to the next four speakers. Speakers four through seven on your answer sheet. In each case, circle what you think is the appropriate sex. 233 Five, ten, or 15 sec sample of sex speaker 4 (male), prepared normal condition; sex speaker 11 (female), prepared normal condition; sex speaker 10 (female), spontaneous telephone condition; and sex speaker 5 (male), prepared telephone condition. Okay, I'll read the answers. I'll pause after each answer and I would like those of you who circled the wrong sex to raise your hands. This is just to give me an idea of how you're doing. Speaker number four was a male . speaker number five was a female . . . speaker number six was a female ... . and speaker number seven was a male. Okay, good. Now let's move on to something a little more difficult. Turn to page two of your answer sheet. This time you will be judging how old the speaker is. As you can see in the "dimension" column on the right hand side of your paper, you will have three choices for each speaker: (1) the age group 20 to 30 years; (2) the age group 40 to 50 years; and (3) the age group 60 to 70 years. Your job will be to circle the age group you think the speaker belongs to. Let's try the first one. Listen and then circle what you feel is the appropriate age group for speaker one. Five, ten, or 15 sec sample of age speaker 4 (20-30), spontaneous normal condition. That speaker was between 20 and 30 years of age. How many of you circled one of the other age groups? Listen to 234 speaker two and circle the age group you feel is appropriate for him. Five, ten, or 15 sec sample of age speaker 10 (40-50), spontaneous normal condition. That speaker was between 40 and 50. How many got it correct? New let me tell you some things that might help you differentiate between these age groups. First of all, about the older group; those between 60 and 70 years of age. These speakers have been found to talk and read slower than younger speakers and they sometimes have voice breaks. It has also been found that older speakers tend to speak with reduced intensity or loudness; with less force. These things differ- entiate them from both the 40 to 50 year old speakers and the 20 to 30 year old speakers. Differences between these last two groups are not as obvious. Some of the same differences are present, such as, slower speech and reading for the 40 to 50 year old group, but many of the others, such as, voice quality and reduced intensity do not stand out. Now let's listen to a 60 to 70 year old speaker. Don't bother to mark your answer on this one. Five, ten, or 15 sec sample of age speaker 16 (60—70), spontaneous normal condition. Okay, now let's listen to the next three speakers; speakers four, five, and six. For each one, circle what you feel is the appropriate age group. 235 Five, ten, or 15 sec sample of age speaker 11 (40-50), prepared normal condition; age speaker 17 (60-70), spontaneous normal condition; and age speaker 5 (20—30), prepared normal condition. I'll read the answers. Raise your hand if you have an answer different than what I read. Speaker number four was 40 to 50 years of age . . . speaker number five was 60 to 70 years of age . . . and speaker number six was 20 to 30 years of age . . . Okay, good. NOW let's try some more. we'll hear speakers seven through 12. For each one, circle what you feel is the appropriate age group. Five, ten, or 15 sec sample of age speaker 10 (40—50), spontaneous normal condition; age speaker 4 (20-30), prepared normal condition; age speaker 5 (20-30), pre— pared normal conditipn; age speaker 16 (60-70), spon- taneous telephone condition; age speaker 11 (40-50), spontaneous telephone condition; and age speaker 17 (60-70), prepared telephone condition. Okay. Did you hear the telephone speech? Let me read the answers. You indicate the ones you missed by rais— ing your hand. Speaker number seven was 40 to 50 years of age . . . speaker number eight was 20 to 30 years of age speaker number nine was 20 to 30 years of age . . . speaker number ten was 60 to 70 years of age . . . speaker number 11 was 40 to 50 years of age . . . and speaker number 12 was 60 to 70 years of age . . . Okay, let's try three more samples of telephone speech. Circle what you feel are the appropriate age groups for speakers l3, l4, and 15. 236 Five, ten, or 15 sec samples of age speaker 4 (20—30), prepared telephone condition; age speaker 5 (20-30), prepared telephone condition; and age speaker 10 (40-50), spontaneous telephone condition. Okay, here are the answers. Raise your hand if you missed. Speaker number 13 was 20 to 30 years of age speaker number 14 was 20 to 30 years of age . . . and speaker number 15 was 40 to 50 years of age. Very good. Now let's talk about the ethnic group characteristic. Turn to page three. Your task here will be to indicate whether you think the speaker is Negro or Cau— casian. This is often a very hard differentiation to make. As a matter of fact, research offers us very little to go on in this area. This is especially true for the speakers you will be hearing--Caucasians and Negroes who were born and at- tended elementary and secondary schools in Michigan. One possible differentiating factor that has been suggested is in voice quality. This difference is due to Negro speakers sometimes articulating some of their vowels in a more open and free manner than Caucasian speakers. It has also been suggested that Caucasians tend to have more nasality in their speech. Another factor that has been suggested is that Negro speakers use a lower pitch than Caucasians. There is also a possibility that some differences in vocal intonation or ex— pression patterns might exist between Caucasian and Negro speakers. This isn't much to go on, but let's give it a try. 237 Listen to the first three speakers and indicate under the "dimension" column on the right hand side of your answer sheet which ethnic group you feel they belong to--Negro or Caucasian. Five, ten, or 15 sec sample of ethnic group speaker 4 (Caucasian), spontaneous normal condition; ethnic group speaker 10 (Negro), spontaneous normal condition; and ethnic group speaker 5 (Caucasian), spontaneous nor— mal condition. All right. Speaker one was Caucasian. Did anyone circle "Negro?" Speaker two was Negro. Raise your hand if you circled "Caucasian.“ Speaker three was Caucasian. How many missed him? Fine. Let's listen to the next four speakers. Circle what you feel is the appropriate ethnic group for each speaker. Five, ten, or 15 sec sample of ethnic group speaker 11 (Negro), spontaneous normal condition; ethnic group speaker 10 (Negro), prepared normal condition; ethnic group speaker 11 (Negro), prepared normal condition; and ethnic group speaker 4 (Caucasian), spontaneous telephone condition. Okay, ready? I'll read the answers. You indicate if you missed by raising your hand. I'll pause after each answer so you'll have time. Speaker number four was Negro speaker number five was Negro . . . speaker number six was Negro . . . and speaker number seven was Caucasian. Fine, Let's try the last three speakers now. These will all be telephone recordings. 238 Five, ten, or 15 sec sample of ethnic group speaker 10 (Negro), prepared telephone condition; ethnic group speaker 5 (Caucasian), prepared telephone con- dition; and ethnic group speaker 4 (Caucasian), pre- pared telephone condition. All right, here are the answers. Raise your hand again after the ones you missed. Speaker eight was Negro speaker nine was Caucasian . . . and speaker ten was Caucasian. Okay, turn to page four. The next speaker charac— teristic we are going to practice is education. Your task in hearing these speakers will be to circle the level of for— mal education you feel the speaker has. For each speaker, you will have three choices: (1) less than a high school education, indicated on your answer form as "minus H.S.:" (2) a high school education, indicated on your answer form as "H.S.;" and (3) a college education. This characteristic is also sometimes hard to identify. Probably the outstanding speech difference you might notice is grammer. This, of course, is bbvious. The less formal education a person has, the more grammatical mistakes you would expect to hear in his speech. Another differentiating factor is often the use of adjectives. Better educated speakers use more adjectives, that is, modifying and qualifying words; or you could say, more complex sentences. You would also expect speakers with little formal education to be less at ease in speaking and probably less fluent in their reading. Vbcabulary usage 239 might also be a give away. Let's try a few speakers. Listen to the first three and circle the education level you feel is appropriate for each one. Five, ten, or 15 sec sample of education speaker 4 (-H.S.), spontaneous normal condition; education speaker 10 (H.S.), spontaneous normal condition; and education speaker 16 (college), spontaneous normal condition. Raise your hand if you have the wrong answer for any of the speakers. Speaker number one was less than a high school graduate . . . speaker number two was a high school graduate . . . and speaker number three was a college gradu- ate——what you all hope to be someday . . . Now let's try six more. Again circle what you feel is the appropriate edu— cation level for each speaker. Five, ten, or 15 sec sample of education speaker 11 (H.S.), prepared normal condition; education speaker 17 (college), prepared normal condition; education speaker 5 (-H.S.), prepared normal condition; education speaker 4 (-H.S.), prepared normal condition; education speaker 16 (college), prepared normal condition; and education speaker 10 (H.S.), prepared normal condition. Okay, raise your hand if you have the wrong answer. Speaker number four was a high school graduate . . . speaker number five was a college graduate . . . speaker number six ‘was less than a high school graduate . . . speaker number seven was also less than a high school graduate . . . speaker .number eight was a college graduate . . . and speaker number 240 nine was a high school graduate. Okay, good. Let's listen to four more. Remember to circle your answers. Five, ten or 15 sec sample of education speaker 16 (college), spontaneous telephone condition; education speaker 11 (H.S.), spontaneous telephone condition; education speaker 5 (-H.S.), spontaneOus telephone condition; and education speaker 17 (college), pre— pared telephone condition. Okay, let's see how you did. Speaker ten was a col— lege graduate . . . speaker 11 was a high school graduate speaker 12 was less than a high school graduate . . . and speaker 13 was a college graduate. Now, two more characteristics to practice. Our next one is physical size. Turn to page five. As you can see by your answer sheet, your job on this characteristic will be to indicate whether you think the speaker is "big" or "small." Let me define these two terms a little more for you. By "big" I mean someone about six feet, one inch, weighing approxi- mately 205 pounds. The small speakers average about five feet five and weighw about 120 pounds. So, we're talking about two opposite extremes. There is very little research in this area. However, I think we all have some expectations for these speakers. In general, we tend to expect a big, booming voice from the large fellows and perhaps a thin, weak voice from the little guys. This is, of course, somewhat exaggerated and, in some cases, not very accurate. Let's listen to a few and see how you do. Circle your answers as before. 241 Five, ten, or 15 sec sample of physical size speaker 4 (big), spontaneous normal condition; physical size speaker 10 (small), spontaneous normal condition; physical size speaker 5 (big), spontaneous normal con- dition; and physical size speaker 11 (small), spon— taneous normal condition. Okay, here are your answers. Raise your hand if you were incorrect. Speaker number one was big . . . speaker number two was small . . . speaker number three was big and speaker number four was small . . . All right, let's try five more. Be sure to circle your answers. Five, ten, or 15 sec sample of physical size speaker 4 (big), prepared normal' condition; physical size speaker 10 (small), prepared normal condition; physi- cal size speaker 5 (big), spontaneous telephone con- dition; physical size speaker 11 (small), spontaneous telephone condition; and physical size speaker 10 (small), spontaneous telephone condition. Now, let's see how you did. Raise your hand on the ones you missed. Speaker number five was big . . . speaker number six was small . . . speaker number seven was big speaker number eight was small . . . and speaker number nine was small . . . Okay, very good. This characteristic is not as easy as it SoundS' at first. Let's try three more. Five, ten, or 15 sec sample of physical size speaker 4 (big): spontaneous telephone condition; physical size speaker 11 (small), prepared telephone condition; and physical size speaker 5 (big), prepared telephone condition. 242 All right. Let's see how you did on this group. Raise your hand again on the ones you missed. Speaker ten was big . . . speaker 11 was small . . . and speaker 12 was big. All right, fine. NOW for the last characteristic we will practice. Turn to page six. Your job this time will by to indicate what part of the country you think the speaker is from. As you can see on your answer sheet, you will have three choices for each speaker: (l) the East; (2) the South; and (3) Michigan. What you will be listening for, of course, will be dialects. That's why we call this characteristic "dialect region." we expect a person from the South to speak with a Southern dialect. Likewise, we expect someone from the East to use an Eastern dialect. we expect the General American dialect from a speaker from Michigan. This type of speech is what you've been hééiifié from me for the past 35 minutes. The Southern and Eastern dialects are different from my speech. For example, let's consider the "r" sound as it occurs in the final position of words such as "far" and "poor." Eastern and Southern speakers seem to drop these "r's" and say something resembling [fag] and [p33]. The same thing happens with other words such as "barn," "form," and "third." In Eastern and Southern speech, these often come out as [bush], [foam], and [63d]. Another charac- teristic of Eastern speech is their tendency to broaden or flatten some vowel sounds. An example of this is in the 243 words "foreign" and ”orange," which I just said in my usual manner. Some Easterners would say [faren] and [arondg]. This broadening of vowels also occurs in Eastern speech in such words as "ask" and "dance." Someone from the East might say [ask] and [dons]. These examples are somewhat exagger- ated but they'll give you the general idea. In addition to their use of the "r" sound, Southern speakers can often be identified by the way they say the sound "i.“ They often seem to broaden this sound, changing "like" into [lakJ and "white" into [wut]. Another thing they sometimes do is insert the [j] sound before the [u] sound in words like "duty" and "tune." These then become [djutI] and [tjunlo In addition, we are all familiar with the Southern drawl. This makes it seem as though Southern speakers talk slower than people from the rest of the country. On the other hand, the clipped, staccato type speech that some Eastern speakers use makes it seem as though they talk very fast. Let's listen to a few speakers and see how you do. Five, ten, or 15 sec sample of dialect region speaker 4 (Michigan), spontaneous normal condition; dialect region speaker 16 (South), spontaneous normal con- dition; and dialect region speaker 10 (East), spon- taneous normal condition. Okay, here are the answers for the first three speakers. Raise your hand if you circled an answer differ- ent from what I read for any speaker. Speaker one was from Michigan . . . speaker two was from the South . . . and 244 speaker three was from the East . . . All right, let's listen to three more; speakers four, five, and six. Five, ten, or 15 sec sample of dialect region speaker 17 (South), prepared normal condition; dialect region speaker 11 (East), prepared normal condition; and dia- lect region speaker 16 (South), spontaneous telephone condition. Raise your hand again if you missed any as I read the answers. Speaker four was from the South . . . speaker five was from the East . . . and speaker six was from the South Okay, good. ILet's jlisten to three more. Five, ten, or 15 sec sample of dialect region speaker 10 (East), prepared telephone condition; dialect region speaker 11 (East), prepared telephone condition; and dialect region speaker 17 (South), prepared telephone condition. Okay, raise your hand if you missed any. Speaker seven was from the East . . . speaker eight was from the East . . . and speaker nine was from the South. Very good. NOw turn to page seven. As you can see, there are fifteen speakers on this page and the characteristic to be judged is different for each one. Your job will be the same as before though. Just circle the dimension on the right that you feel is appropriate for each speaker according to the characteristic he or she is to be evaluated on. For example, you can see that speaker one is going to be judged on ethnic group. Just as before, you circle either "Caucasian" or 245 "Negro." Speaker two will be judged on dialect region. You will circle "South," "East," or "Michigan." Okay, ready? Here we go. Sfiiggir Identification Characteristic Dimension Char. # Cond. 1 EG 12 Spon—Nor ’Ethnic Group Negro 2 DR 18 Spon-Tel Dialect Region South 3 E 6 Spon—NOr Education -High School 4 E 18 Spon-Tel Education College 5 A 6 ASpon—Nbr Age 20—30 6 A 12 Prep-Nor Age 40—50 7 PS 12 Spon-NOr Physical Size Small 8 DR 12 Prep—Tel Dialect Region East 9 A 18 Spon-Tel Age 60-70 10 EG 6 Prep—Tel Ethnic Group Caucasian 11 S 6 Prep-Tel Sex Male 12 E 12 Spon—Nor Education High School 13 S 12 Spon—NOr Sex Female 14 PS 6 Spon-Nor Physical Size Big 15 DR 6 Spon—Tel Dialect Region Michigan APPENDIX D INSTRUCTIONS TO SUBJECTS 247 MASTER TAPE I Okay, now turn to page eight. Listen carefully to the following instructions. You are being asked to make judgments about various characteristics of speakers. There are six characteristics in all and you will make a judgment about just one for each speaker you hear. Please refer to your answer forms. As you can see, there are three categories across the top of the page: at the left is "Speaker Number;" in the center, "Characteristic;" and at the right, "Dimensions." Each speaker you hear will be designated by a number in the left-hand column. The characteristic you will be evaluating or identifying for each speaker is listed in the center column. In the right-hand column are listed the various dimensions of the characteristic you are to evaluate for each speaker. You are to circle only one of these dimensions for each speaker. For example, the first speaker you will hear will be speaker number one. The characteristic you will be evaluating about him is Physical Size as indicated in the center column. In the right-hand column are the dimensions of this characteristic; in this case "big" and "small." You are to circle just one of the dimensions. As you can see from your answer form, for the second speaker (speaker number two) you will be evaluating Dialect Region and the dimensions for this characteristic are again listed in the right-hand COlumn. You will circle only one just as you will have done for speaker number one. 248 As indicated by your answer form, this procedure will continue for 60 speakers. Are there any questions at this point...? The six characteristics you will be evaluating for the speakers are: (1) Sex; (2) Age; (3) Ethnic Group; (4) Education Level; (5) Physical Size; and (6) Dialect Region of the country. As we have already seen, there are three dimensions for the last characteristic, Dialect Region. These are, the South; the East; and Michigan. There are two dimensions for the Sex characteristic; male and female. This is illustrated by speaker number 37 on the second page of your answer form. For the Age characteristic, there are three dimensions; the age group 20 to 30 years, the age group 40 to 50 years, and the age group 60 to 70 years. For each speaker.for which a judgment on the Age characteristic is requested, for example, speaker number six on your answer f‘Orm, you are to circle one of these categories or groups. For the Ethnic Group charac- teristic, there are again two dimensions; Caucasian and Negro. For each speaker for which a judgment on the Ethnic Group characteristic is requested, for example, speaker number nine on your answer form, you are to again circle just one of these two dimensions. For the Education characteristic, there are three dimensions; less than a high school graduate, a high school graduate, and a college graduate. An example for this characteristic is speaker number four on your answer form.’ 249 we have already discussed an example of evaluating a speaker on the Physical Size characteristic. As indicated for speaker number one, there will always be a choice - of two dimensions when this evaluating is indicated - big or small. Are there any questions at this point...? As I indicated before, you will be hearing 60 speakers. However, the speech sample you will hear for each of these 60 speakers will be very short - five seconds. For this reason, you will have to listen closely. Following each speech sample, you will have just ten seconds to circle the dimension you feel is appropriate. Again, this short period of time means you will have to pay close attention. The tape will run continuously during the entire session, even during the time you will be marking your answer form. The tape is arranged so that you will hear a number referring to a speaker, such as "Speaker One;" the five speech sample for that speaker; and then ten seconds of silence during ' which you will circle the dimension you feel is appropriate for the speaker on the characteristic in question. Are there any question at this point...? In summary, your answer forms list the speakers by number and the characteristic to be evaluated for each in the order that you will hear them. You are simply to listen to each speech sample and then circle the one dimension of the 250 characteristic indicated for that speaker that you feel is appropriate. Are there any questions before we begin...? 251 MASTER TAPE II Okay, now turn to page eight. Listen carefully to the following instructions. You are being asked to make judgments about various characteristics of speakers. There are six characteristics in all and you will make a judgment about just one for each speaker you hear. Please refer to your answer forms. As you can see, there are three categories across the top of the page: at the left is "Speaker Number;" in the center, “Characteristic;" and at the right, "Dimensions." Each speaker you hear will be designated by a number in the left- hand column. The characteristic you will be evaluating or identifying for each speaker is listed in the center column. In the right-hand column are listed the various dimensions of the characteristic you are to evaluate for each speaker. You are to circle only one of these dimensions for each speaker. For example, the first speaker you will hear will be speaker number one. The characteristic you will be evaluating about him is Dialect Region as indicated in the center column. In the right-hand column are the dimensions of this characteristic; in this case "East," "Michigan," and "South." You are to circle just one of these dimensions. As you can see from your answer form, for the second speaker (speaker number two) you will be evaluating Education Level and the dimensions for this characteristic are again listed in the right-hand column. You will circle only one just as 252 you will have done for speaker number one. As indicated by your answer form, this procedure will continue for 60 speakers Are there any questions at this point...? The six characteristics you will be evaluating for the speakers are: (1) Sex; (2) Age; (3) Ethnic Group; (4) Education Level; (5) Physical Size; and (6) Dialect Region of the country. As we have already seen, there are three dimensions for the last characteristic, Dialect Region. These are, the East, Michigan, and the South. There are two dimensions for the Sex characteristic; male and female. This is illustrated by speaker number seven on your answer form. For the Age characteristic, there are three dimensions; the age group 20 to 30 years, the age group 40 to 50 years, and the age group 60 to 705years. For each speaker for which a judgment on the age characteristic is requested, for example, speaker number three on your answer form, you are to circle one of these age categories or groups. For the Ethnic Group characteristic, there are again two dimensions; Caucasian and Negro. For each speaker for which a judgment on the Ethnic Group characteristic is requested, for example, speaker number nine on your answer form, you are to again circle just one of these two dimensions. we have already discussed an example of evaluating a speaker 253 on the Education Level characteristic. As indicated for speaker two, there will always be a choice of three dimensions when this evaluation is indicated - less than a high school graduate, a high school graduate, and a college graduate. For the Physical Size characteristic, there are two dimensions; big and small. An example for this charac— teristic is speaker number four on your answer form. Are there any questions at this point...? As I indicated before, you will be hearing 60 speakers. However, the speech sample you will hear for each of these 60 speakers will be very short - ten seconds. For this reason, you will have to listen closely. Following each speech sample, you will have another ten seconds to circle the dimension you feel in apprOpriate. Again, this short period of time means you will have to pay close attention. The tape will run continuously during the entire session, even during the time you will be marking your answer form. The tape is arranged so that you will hear a number referring to a speaker, such as "Speaker One;“ the ten second speech sample for that speaker; and then ten seconds of silence during which you will circle the dimension you feel is appropriate for that speaker on the characteristic question. Are there any questions at this point...? In summary, your answer forms list the speakers by number and the characteristic to be evaluated for each in 254 the order that you will hear them. You are simply to listen to each speech sample and then circle the one dimension of the characteristic indicated for that speaker that you feel is appropriate. 0 Are there any questions before we begin...? 255 MASTER TAPE III Okay, now turn to page eight. Listen carefully to the following instructions. You are being asked to make judgments about various characteristics of speakers. There are six characteristics in all and you will make a judgment about just one for each speaker you hear. Please refer to your answer forms. As you can see, there are three categories across the top of the page: at the left is "Speaker Number;" in the center, "Characteristic;" and at the right, dimensions." Each speaker you hear will be designated by a number in the left- hand column. The characteristic you will be evaluating or identifying for each speaker is listed in the center column. In the right—hand column are listed the various dimensions of the characteristic you are to evaluate for each speaker. You are to circle only one of these dimensions for each speaker. For example, the first speaker you hear will be speaker number one. The characteristic you will evaluating about him is Physical Size as indicated in the center column. In the right-hand column are the dimensions of this charac— teristic; in the case "big" and "small." You are to circle just one of the dimensions. As you can see from you answer form, for the second speaker (speaker number two) you will be evaluating Dialect Region and the dimensions for this characteristic-are again listed in the right—hand column. You will circle only one just as you will have done for 256 speaker number one. As indicated by your answer form, this procedure will continue for 60 speakers. Are there any questions at this point...? The six characteristics you will be evaluating for the speakers are: (1) Sex; (2) Age; (3) Ethnic Group; (4) Education Level; (5) Physical Size; and (6) Dialect Region of the country. As we have already seen, there are three dimensions for the last characteristic, Dialect Region. These are, the South; the East; and Michigan. There are two dimensions for the Sex characteristic; male and female; This is illustrated by Speaker Number 15 on your anSwer form. For the Age characteristic, there are three dimensions; the age group 20 to 30 years, the age group 40 to 50 years, and the age group 60 to 70 years. For each speaker for which a judgment on the Age characteristic is requested, for example, speaker number four on your answer form, you are to circle one of these age categories or groups. For the Ethnic Group characteristic, thare are again two dimensions; Caucasian and Negro. For each speaker for which a judgment on the Ethnic Group characteristic is requested, for example, speaker number three on your answer form, you are to again circle just one of these two dimensions. For the Education characteristic, there are three dimensions; less that a high school graduate, a high school graduate and a college graduate. An example for this characteristic is speaker 257 number five on your answer form. We have already discussed an example of evaluating a speaker on the Physical Size characteristic. As indicated for speaker number one, there will always be a choice of two dimensions when this evaluation is indicated — big or small. Are there any questions at this point...? As I indicated before, you will be hearing 60 speakers. However, the speech sample you will hear for each of these 60 speakers will be very short - 15 seconds. For this reason, you will have to listen closely. Following each speech sample, you will have just en seconds to circle the dimension you feel is appropriate. Again, this short period of time means you will have to pay close attention. The tape will run continuously.during the entire session, even during the time you will be marking your answer form. The tape is arranged so that you will hear a number referring to a speaker, such as "Speaker One;" the 15 second speech sample for that speaker; and then ten seconds of silence during which you will circle the dimension you feel is appropriate for that speaker on the characteristic in question. Are there any questions at this point...? In summary, your answer forms list the speakers by number and the characteristic to be evaluated for each in the order that you will hear them. You are simply to listen to 258 each speech sample and then circle the one dimension of the characteristic indicated for that speaker that you feel is appropriate. Are there any questions before we begin...? APPENDIX E PRACTICE ANSWER FORM FOR TRAINED GROUPS AND TEST ANSWER FORMS FOR MASTERTAPES I, II, AND III 260 Page 1 PRACTICE ANSWER FORM FOR TRAINED GROUPS Efiigtir Characteristic (circlengfiigs;:28(l) for each speaker) 1 Sex Male Female 2 Sex Male Female 3 Sex Male Female 4 Sex Male Female 5 Sex Male Female 6 Sex Male Female 7 Sex Male Female 261 Page 2 gfiigfiir l Characteristic l (circlenggigs;:28(l) for __;*_ each speaker) Afi__ 1 Age 20:30 ‘ 40-50 60-70—— 2 Age 20-30 40—50 60-70 3 Age 20—30 40-50 60—70 4 Age 20—30 40-50 60—70 5 Age 20—30 40-50 60-70 6 Age 20—30 40-50 60-70 7 Age 20-30 40-50 60-70 8 Age 20-30 40—50 60-70 9 Age 20—30 40-50 60-70 10 Age 20—30 40—50 60-70 11 Age 20-30 40-50 60-70 12 Age 20-30 40-50 60-70 13 Age 20-30 40—50 60—70 14 Age 20—30 40—50 60—70 15 Age 20—30 40—50 60—70 262 Page 3 :3:g::r Characteristic (circleDégigsézgs(l) for _f each speaker) 1 Ethnic Group Negro Caucasian 2 Ethnic Group Negro Caucasian 3 Ethnic Group Negro Caucasian 4 Ethnic Group Negro Caucasian 5 Ethnic Group Negro Caucasian 6 Ethnic Group Negro Caucasian 7 Ethnic Group Negro Caucasian 8 Ethnic Group Negro Caucasian 9 Ethnic Group‘ Negro Caucasian 10 Ethnic Group Negro Caucasian 263 Page 4 Dimensions gsggtir Characteristic (circle only one (1) for each speaker) ‘ 1 Education —H.S H.S. College 2 Education’ -H.S. H.S. College 3 Education -H.S. H.S. College 4 Education -H.S. H.S. College '5 Education —H.S. H.S. College 6 Education -H.S. H.S. College 7 Education -H.S. “.H.S. College‘ 8 Education -H.S. H.S. College 9 Education -H.S. H.S. College 10 Education -H.S. H.S. College 11 Education -H.S. H.S. College 12 Education —H.S. H.S. College 13 Education -H.S. H.S. College 264 Page 5 Dimensions (circle only one (1) for each speaker) ‘ " Speaker N ]er,_ Characteristic 1 Physical Size Big Small 2 Physical Size Big Small 3 Physical Size Big Small 4 Physical Size Big Small 5 Physical Size Big Small 6 Physical Size Big Small 7 Physical Size Big Small 8 Physical Size Big Small 9 Physical Size Big Small 10 Physical Size Big Small 11 Physical Size Big Small 12 Physical Size Big Small 265 Page 6 Speaker Number Characteristic Dimensions (circle only one (1) for each speaker) Dialect Region Michigan South 2 Dialect Region East Michigan South 3 Dialect Region East Michigan South 4 Dialect Region East Michigan South 5 Dialect Region East Michigan South 6 Dialect Region East Michigan South 7 Dialect Region East Michigan South 8 Dialect Region East Michigan South 9 Dialect Region East Michigan South 266 Page 7 33:22:? : Characteristic (circleDggigséigs(l) for ‘ ‘ ‘ “' each speaker)‘ 1 Ethnic Group Negro Caucasian 2 Dialect Group (East Michigan South 3 Education -H.S. H.S. College 4 Education -H.S. H.S. College 5 Age 20—30 40-50 60-70 6 Age 20-30 40—50 60-70 7 Physical Size Big Small 8 Dialect Region East Michigan South 9 Age 20—30 40-50 60-70 10 Ethnic Group Negro Caucasian 11 Sex Male Female 12 Education -H.S H.S. College 13 Sex Male Female 14 Physical Size Big Small 15 Dialect Region East Michigan South MASTER TAPE 267 I ANSWER FORM Page 8 Sfigggir Characteristic (circlenggigséfigs(l) for ‘ each speaker) ‘ ' 1 Physical Size Big Small 2 Dialect Region East Michigan South 3 Dialect Region East Michigan South 4 Education -H.S. H.S. College 5 Dialect Region East Michigan South 6 Age 20—30 40-50 60-70 7 Education —H.S. H.S. College 8 Dialect Region East Michigan South 9 Ethnic Group Negro Caucasian 10 Physical Size Big Small 11 Ethnic Group Negro Caucasian 12 Age 20—30 40—50 60-70 13 Dialect Region East Michigan South 14 Ethnic Group Negro Caucasian 15 Physical Size Big Small 16 Dialect Region East Michigan South 17 Physical Size Big Small 18 Ethnic Group Negro Caucasian 19 Age 20-30 40—50 60-70 20 Dialect Region East Michigan South 21 Physical Size Big Small PLEASE TURN TO NEXT PAGE 268 Page 9 gfifiggir Characteristic (circleD6§I386:28(l) for ,. each speaker) 22 Education -H.S. H.S. College 23 Physical Size Big Small 24 Ethnic Group Negro Caucasian 25 Education -H.S. H.S. College 26 Physical Size Big Small 27 Education -H.S. H.S. College 28 Education -H.S. H.S. College 29 Dialect Region East Michigan South 30 Education —H.S H.S. College 31 Ethnic Group Negro Caucasian 32 Dialect Region East Michigan South 33 Education -H.S. H.S. College 34 Dialect Region East Michigan South 35 Age 20-30 40—50 60-70 36 Sex Male Female 37 Sex Male Female 38 Ethnic Group Negro Caucasian 39 Age 20-30 40-50 60-70 40 Sex Male Female 41 Age 20-30 40—50 60-70 42 Education -H.S. H.S. College PLEASE TURN TO NEXT PAGE 269 Page 10 Dimensions (circle only one (1) for each speaker) Speaker Number-:1 Characteristic 43 Dialect Region East Michigan South 44 Sex Male Female: 45‘ Sex Male Female '46 Education -H.S. H.S. College 47 Age 20—30 40-50 60-70 48 Ethnic Group Negro Caucasian 49 Age 20-30 40-50 60-70 50 Age 20—30 40—50 60-70 51 Sex Male Female 52 Age 20—30 40-50 60-70 53 Sex Male Female 54 Age 20—30 40-50 60-70 55 Age 20—30 40-50 60-70 56 Sex Male Female 57 Dialect Region East Michigan South 58 Physical Size Big Small . 59 Education -H.S. H.S. College 60 Education -H.S. H.S. College MASTER TAPE 270 II ANSWER FORM Page 8 gsggzir ‘Characteristic (circleD62I386328(l) for each speaker) l Dialect Region East Michigan South 2 Education -H.S. H.S. College ‘3 Age 20-30 40-50 60-70 4 Physical Size Big Small 5 Dialect Region East Michigan South 6 Physical Size Big Small 7 Sex Male Female 8 Age 20—30 40-50 60-70 9 Ethnic Group Negro Caucasian 10 Education -H.S H.S. College 11 Ethnic.Group Negro Caucasian 12 Physical Size Big Small 13 Dialect Region East Michigan South 14 Dialect Region East Michigan South 15 Education -H.S. H.S. College 16 Age 20—30 40-50 60-70 17 Sex Male Female 18 Education -H.S. H.S. College 19 Dialect Region} East Michigan South 20 Physical Size : Big Small 21 physical Size : Big Small PLEASE TURN TO NEXT PAGE 271 II T U Page 9 33:2:27 “Characteristic ‘ (circleD62I386328(l) for each speaker) 22 Ethnic Group Negro Caucasian 23 Education -H.S. H.S. College 24 ‘ Sex , Male Female 25 Age 20-30 40-50 60-70 26 Sex Male Female —r27‘ Sex Male Female 28 Ethnic Group Negro Caucasian 29 Physical Size Big Small 30 Age 20—30 40—50 60-70 31 Dialect Region East Michigan South 32 Sex Male Female 33 Education -H.S. H.S. College 34 Education -H.S. H.S. College 35 Age 20-30 40—50 60-70 36 Age 20-30 40-50 60—70 37 Dialect Region East Michigan South 38 Ethnic Group Negro Caucasian 39 Education —H.S. H.S. College 40 Dialect Region East Michigan South 41 Age 20-30 40-50 60-70 42 Sex Male Female PLEASE TURN TO NEXT PAGE 272 II U Page 10 33:22:: (Characteristic (circleD6§I386::S(l) for ' ' ‘ each speaker) 43 Education —H.S. H.S.: Collage 44 Age 20—30 40-50 60-70 45 Dialect Region East Michigan South 46 Age 20—30 40-50 60-70 47 Sex Male Female 48 Education -H.S. H.S. College 49 Dialect Region East Michigan South 50 Dialect Region East Michigan South" 51 Ethnic Group Negro Caucasian 52 Age 20-30 40-50 60-70 53 Physical Size Big Small 54 Ethnic Group Negro Caucasian 55 Education -H.S H.S. College 56 Ethnic Group Negro Caucasian 57 Dialect Region East Michigan South 58 Age 20—30 40-50 60-70 59 Education -H.S H.S. College 60 Physical Size Big Small 273 MASTER TAPE III ANSWER FORM Page 8 gagggir Characteristic; (circleD62I386328(1) for each speaker) 1 Physical Size 7 Big Small 2 Dialect Region East Michigan Small 3 Ethnic Group Negro Caucasian 4 Age 20-30 40—50 60-70 5 Education —H.S H.S. College 6 Physical Size, Big Small 7 Dialect Region East Michigan South 8 Education -H.S H.S. College 9 Physical Size Big Small 10 Education -H.S H.S. College 11 Education —H.S. H.S. College 12 Age 20—30 40-50 60—70 13 Age 20-30 40—50 60-70 14 Age 20-30 40-50 60-70 15 Sex Male Female 16 Ethnic Group Negro Caucasian 17 Ethnic Group Negro Caucasian 18 Dialect Region East Michigan South 19 Sex Male Female 20 Dialect Regionv East Michigan South 21 Age 20-30 40—50 60-70 PLEASE TURN TO NEXT PAGE 274 III U Page 9 gaggzir Characteristic (circlenggigséggs(l) for each speaker) 22 Sex Male Female 23 Dialect Region East Michigan South 24 Education -H.S. H.S. College 25 Age 20-30 40—50 60-70 26 Age 20-30 40-50 60-70 27 Ethnic Group Negro Caucasian 28 Sex Male Female 29 Physical Size Big Small 30 Sex Male Female 31 Dialect Region East Michigan South 32 Dialect Region East Michigan South 33 Ethnic Group Negro Caucasian 34 Physical Size Big Small 35 Dialect Region East Michigan South 36 Physical Size Big Small 37 Physical Size Big Small 38 Education -H.S H.S. College 39 Dialect Region East Michigan South 40 Ethnic Group Negro Caucasian 41 Age 20-30 40-50 60-70 42 Education If -H.S. H.S. College PLEASE TURN TO NEXT PAGE 275 III U Page 10 gfiiggir Characteristic (circleD62I;s(I?Sfor each speaker) 43 Age 20—30 40-50 60-70 44 Ethnic Group Negro Caucasian 45 Education -H.S. H.S. College 46 Ethnic Group Negro Caucasian 47 Age 20-30 40-50 60—70 48 Education -H.S. H.S. College 49 Dialect Region East Michigan South 50 Sex Male Female 51 Education -H.S H.S. College 52 Dialect Region East Michigan South 53 Sex Male Female 54 Dialect Region East Michigan South 55 Physical Size Big Small 56 Age 20—30 40-50 60-70 57 Education -H.S H.S. College 58 Sex Male Female 59 Age 20—30 40-50 60-70 60 Education —H.S H.S. College APPENDIX F PROPORTION CORRECT ON EACH SPEAKER CHARACTERISTIC OVER ALL TRAINING, SPEECH, AND DURATION FACTORS FOR THE TEST AND RETEST CONDITIONS 277 mm. ac. ca. cm. ca. om. mm. mm. em. ac. oo.a oo.a m am. am. ca. om. cm. ac. cc. ca. cc. om. oo.H oo.a aam ma. ac. ac. ac. mm. ac. cc. cc. ac. om. oo.a oo.a cam ac. ac. ac. ac. om. mm. ca. om. ac. mm. oo.H oo.a cam mm. ac. om. om. ac. ac. ac. ac. ac. ac. oc.H oc.a cam ac. cc. om. ac. mm. ac. mm. mm. ac. mm. cc., oo.a mam mm. ca. ca. ca. ac. co. ca. ca. ac. om. oo.a oo.a aac aa. mm. ac. ac. ac. cm. cc. ca. om. om. oo.a oo.a Ham om. ac. oc. om. ac. mm. ca. ca. om. mm. oo.H oo.a cam am. am. om. ma. am. am. am. ac. cc. am. oo.a oo.a mm om. ac. ac. ca. mm. ca. mm. mm. om. om. oo.a oo.a cc ca. ac. ca. om. ca. ca. ca. ca. ca. mm. oo.a oo.a ac ac. ac. ac. ca. ac. ma. ca. cc. cc. mm. oo.a oo.a cc ac. ac. oc. om. mm. mm. mm. ca. om. cc. cos. oo.a cc ac. am. ac. ac. ac. ac. cc. ca. ac. ac. oo.a oo.a ac ma. ac. ac. om. mm. ac. cc. ma. ac. ca. cc. oo.H mm ac. om. ca. om. ac. om. ma. mm. ac. ac. oo.a oo.a am mm. mm. ca. mm. om. om. mm. mm. ac. ma. oo.a oc.a Hc cmcccm pace umcccm came umcccm pace cmcccm pace cameos pace cacucm pace pocansc coamwm .Hman mNflm .mhnm coaumospm msouw oacnum wm¢ xmm 278 mm ca. am. mm. mm. am. am. ca. cc. ca. ma. oo.a oo.a m ma. ac. ma. cm. aa. mm. ca. am. am. am. oo.a oo.a cam mm. om. oo.a am. am. ac. ac. ac. ma. ac. oo.a oo.a cam ma. ca. ca. ac. aa. aa. ca. ca. ca. ac. oo.a oo.a mac ac. ac. mm. mm. mm. ac. ac. cc. ac. mm. mm. aa. aac ca. ma. mm. mm. ca. ac. ca. ac. ca. ac. oo.a oo.H Ham am. am. ma. ma. ma. ac. ac. ac. ac. ca. oo.a oo.H omm ca. ma. cc. ca. ac. cc. ma. ca. ca. ma. oo.H oo.a mam om. ac. cc. oo.a ac. ac. ac. ca. ca. ac. oo.a oo.a cam cc. mm. mm. oo.a ca. ac. ca. cc. ca. ac. oo.H oo.a aam ca. ac. cc. oo.H am. am. ca. ca. ac. ma. oo.a aa.. cam ma. ma. cc. ma. am. ac. ma. ca. ca. ac. oo.a oo.a cam mm. ac. cc. mm. mm. ac. ca. ca. ca. ac. oo.a oo.a cam ca. mm. ca. ca. mm. aa. ca. ca. ac. om. oc.a oo.a mac ca. ac. om. om. aa. aa. ma. ma. mm. mm. oo.a oo.a aac ca. ca. om. ac. ac. om. ac. ma. cc. aa. oc.a oo.a Ham mm. om. ac. ac. ca. aa. ma. ma. aa. aa. oo.a oo.a cam ac. mm. ma. ma. ma. mm. mm. ca. aa. aa. cc. oo.a mac pmmuom umwe uwmumm owes umouom Dame pmmumm umma unopom umoB ummuom umoB pomhnsm coamom .Hman mNHm .mmsm coHumospm msonw oacnum wm< xmm 279 gm ca. cc. ac. mm. am. am. om. om. aa. aa. oo.a oo.a m ac. mm. ca. ac. ac. ca. ca. am. am. am. oo.H oo.a amc ac. ma. ca. ac. ac. om. aa. aa. cc. ac. oo.a oo.a amc am. am. ca. mm. mm. ca. ca. ca. am. am. oo.a oo.a acc em. ca. mm. mm. om. ac. oo.a co. ca. cc. cc. ma. ocm am. ac. cc. ac. mm. ac. ma. cc. am. am. oo.a cc. mam ca. am. ac. ca. aa. aa. ca. cm. ma. ma. oo.a oo.a mac am. am. ca. ma. ma. ac. cm. ma. om. mm. oo.a oc.H aec cc. ac. cc. ca. om. mm. ac. ac. ca. ca. oo.H oo.a cam ac. om. mm. mm. om. mm. ca. om. am. am. oo.a oo.a cam ac. om. cm. oo.a aa. aa. cc. ca. ma. ca. oo.a oo.a gem ac. aa. cc. ma. mm. ac. ac. cc. ca. cc. oo.a oo.a mac ma. ac. cc. ca. ca. ca. ac. cc. am. am. oo.a oo.a aec mm. ma. cc. cc. om. ac. ma. cm. am. am. oo.a oo.a Hcm ac. am. ma. cc. ac. om. ca. cc. ma. ac. oo.H oo.a cam ac. om. ca. cc. ca. ac. ma. am. am. am. oo.a oo.a mac ma. ac. ma. ca. ac. ac. ca. cc. ma. ca. oo.a oo.a mac ca. ac. cc. ac. ac. mm. mm. mm. ca. ac. oo.a oc.a amm umwumm umwa umouwm umwa ummpwm umme phenom umma pmmuom umoB ummuom umma pooflnsm coammm .Hman mmam .mhnm coHumosom QSOHO oasnum mmd xmm 280 Nb ac. cm. om. ca. ac. cc. ac. om. cm. ac. oo.a oo.a c ac. ca. ma. cc. ac. ac. ca. ca. mm. mm. oo.a oo.H Hac cc. om. ac. cc. cc. cc. ca. ca. ac. ma. oo.a oo.a oac ac. ac. ca. ca. ac. ac. om. cm. ac. ac. oo.a oo.a ccc ac. ac. ca. ca. om. ca. om. om. ac. ca. oo.a oo.a ccc cc. ac. ca. ac. cc. aa. cm. cm. ca. ac. oo.a oo.a acc cm. ac. ac. ac. cc. aa. om. cm. ac. ac. oo.H oo.a ccc cc. cm. cc. ca. om. am. ca. cm. ca. ma. oo.a oo.a mcc cc. ac. ca. ac. ac. ca. cm. om. mm. mm. oo.a cc. ccc mm. mm. ca. ca. cc. aa. ca. ca. ma. cc. oo.a oo.a mcc ma. om. ma. ca. cc. ma. om. om. ma. ac. oo.a oo.a acc cm. aa. ca. ma. cc. om. om. om. ac. ac. oo.a oo.a Hcc ca. ca. cc. cc. om. ac. cm. cm. ac. ma. cc. cc., occ ac. oo.a ca. cc. ac. ac. ca. ma. ca. cc. oo.H oo.a ccc ac. ac. ca. ac. om. aa. om. cm. aa. aa. oo.a oo.a cmc cm. om. ca. cc. mm. mm. cm. cm. ca. ac. oo.a oo.a acc ac. oo.a ca. cc. aa. om. cm. ca. ac. ca. oo.a cc. ccc cc. ma. ma. ca. cc. aa. ca. om. ac. cm. oo.a oo.a cmc ummumm ummB ummumm umma ammuwm ummB ummuwm uwwE “wwwmm ume ummumm umwe Homnndm coammm .Hamn mnam .mmnm coaumospm msono oacnum mod xmm 281 om ma. ac. om. cc. ac. cm. ac. ca. ac. ac. oo.a oo.H c ca. cc. ca. ma. mm. mm. ca. ca. oc. ca. oo.a oo.a ccc mm. mm. cc. om. ca. ca. ac. ac. cc. ac. oo.a oo.a ccc ac. ac. ca. cc. aa. am. oc.a cc. mm. mm. oo.a oo.a acc ca. ac. cc. cc. cc. ac. ca. oc. om. ac. oo.a oo.a ccc om. om. ac. cc. aa. aa. cc. cc. am. am. oo.a oo.a ccc ac. cc. ac. ac. mm. mm. cc. cc. oc. cm. oo.a oo.a ecm mm. mm. ac. ca. cc. ac. ca. ca. cc. ac. oo.H oo.a mcc ca. cc. ma. cc. cm. ac. cc. cc. aa. mm. oo.a oo.a acc cc. ac. om. ac. cc. ca. cc. cc. ca. ca. oo.H oo.a acm ac. ac. ca. ac. om. ac. ac. cm. ac. ac. oo.a oo.a occ ac. ac. om. ac. ac. cm. ca. ca. cc. cc. oo.a oo.a mac cc. cc. ca. ma. ac. ac. oc. om. mm. mm. oo.a oo.H cam mm. am. am. ac. ca. ca. ac. ac. om. om. oo.H oo.a aac cc. cc. cm. ac. cc. ac. ca. ma. aa. cc. oo.H oo.a cam ac. ac. cc. ac. ca. ac. ac. cc. aa. om. oo.a oo.a cam ac. ac. ac. ac. cc. ca. ma. cc. om. om. oo.a oo.a cam ac. ac. ca. ac. cc. cm. cc. cc. mm. mm. oo.a oo.a mac cmcccm acme cmcccm acme cmcccm acme umcccm pace umcucm acme umcpcm acme cucanac conmm .Hman mNHm .mxnm coaumospm msoww cannum mod xwm 282 woa ca. ac. cc. cc. ca. cc. ac. cm. ac. ac. cc.a cc.H c ac. cc. cc. ca. cc. cc. ca. ca. cc. cc. cc.a cc.a acac ac. ac. ca. cc. cc. ac. ca. ca. ca. cc. cc.a cc.a ccac ca. ca. cc.H cc.a ac. cc. ac. ca. ac. ac. cc.a cc.a ccac ca. cc. cc. cc. ac. ac. ca. ca. cc. ca. cc.H cc.a. ecac ac. ac. ca. ca. cc. ca. ca. ca. cc. ac. cc.a cc.a ccac ac. ac. cc.a cc.H ac. ac. ca. ca. ac. ac. cc.a cc.a .acam cc. cc. ca. ca. cc. ac. ca. ac. ac. ac. cc.H cc.H Hcam cc. ac. ac. cc. aa. cc. ac. ac. cc. cc. cc.a cc.a ccac ac. cc. cc. cc. cc. ac. ca. ca. ac. ac. cc.a cc.a ccc ca. cc. cc. ca. ca. cc. ca. ca. ca. ca. cc.a cc.a ccc cc. cc. cc. cc. ca. ac. cc. ca. ac. ac. cc.a cc.a acc ac. cc. ca. ca. cc. aa. ca. ac. ca. ac. cc.a cc.H ccc cc.a ca. cc.a cc.a cc. ca. cc. cc. ac. ac. cc.a cc.a ccc ca. aa. cc.a ac. cc. cc. cc. cc.H ac. cc. cc.a cc. ccc ac. cc. cc.H cc. aa. cc. ca. ca. ac. ac. cc.a cc.a mcc ac. ac. ac. ca. cc. ac. ca. ca. am. cc. cc.H cc.a acc ca. cc. cc. ca. cc. aa. ac. ac. ca. ac. cc.a cc.a Hcc cccccc acme cccccm ccca cmcccm acme cmcccc acme cmcccm acme cmcccm acme cocaccc coammm .Hman mNHm .mwnm COHDmospm moouo oasnum mod xmm 283 mma cc. ac. ca. cc. ac. cc. ac. ca. ac. ac. cc.a cc.a c ac. ac. ca. ca. aa. aa. ca. cm. ac. cc. cc.a cc.a caac ac. ac. cc.a cc. ac. ac. ac. cm. cc. ac. cc.H cc.a aaac ac. ac. cc. cc. cc. cc. ca. aa. cc. ac. cc.H cc.a mch ac. cc. cc. cc. ca. mm. ca. cc. ca. cc. cc.H cc.a aaac ac. ac. ca. cc. ac. cc. cc. ac. cc. ca. cc.a cc.a_laaac cc. ac. ca. ac. cc. mm. cc. cc. ca. ac. cc.a cc.a cch ac. ac. ca. cc. cc. ac. cc. cm. cc. cc. cc.H cc.H caac cc. ac. ac. ca. aa. aa. ac. cc. cc. ac. cc.H cc.a caac ac. ca. ca. ca. cc. cc. cc. cm. ca. ac. cc.a cc.a aaac ac. ac. ac. cc. cc. ac. cm. ac. cc. ac. cc.a cc.a caam cc. ac. cc. ca. cc. cc. cc. cc. cc. ac. cc.a cc.a caac cc. ac. cc. cc.H aa. cc. cc. ac. ac. ac. cc.H cc.a aSc ac. ca. ca. ca. cc. ac. ca. ac. ac. cc. cc.a cc.a maac cc. cc. ca. cc. cc. cc. ca. cc. ac. ac. cc.a cc.a aaam cc. cc. ca. cc. cc. cc. ca. ca. cc. cc. cc.a cc. aaam ac. ac. cc. cc. cc. ac. ca. ca. ac. ac. cc.H cc.H caac ac. ca. cc. ca. cc. ac. ca. ca. ac. ca. cc.a cc.a ccac cccccc came cmcucm acme cccccc acme cccccc acme cmcccm cmca cccccm acme cucansc coammm .Hman caac .cacm coHumospm msonw oacnum $04 xmm 284 mma ca. ac. cc. cc. cc. cc. cc. cm. ac. ca. ca. cc.a c cc. ac. ac. ca. ac. cc. cm. cm. ac. ca. ca.. cc.a amac ac. ca. cc. ac. cc. ca. cm. cm. ac. ac. cc.a cc.H cmac ac. ac. cc. cc. aa. ca. cm. cc. cc. ca. cc.a cc.H cmac ac. cc. ca. cc. cm. cc. cm. cc. cc. ac. cc.a cc.a cmac cc. cc. ca. ca. ac. cc. cc. ca. ac. ca. cc.a cc.a mmac ac. ac. ac. cc. cc. cc. cm. ac. ac. ac. cc.a cc.a amac ca. cc. cc. cc. ac. cc. cm. cc. ac. ac. cc.a cc.a Hmac ca. cc. ca. ca. cc. cc. cc. cc. cc. ac. cc.H cc.. cmac cc. cc. ca. ca. ac. cc. ca. cc. ac. cc. cc.a cc.a caac cc. ca. cc. cc. ca. cc. ca. ca. cc. cc. cc.H cc.H caac ca. ca. ca. cc. cc. cc. cm. cc. ac. ca. cc.a cc.a aaac cccccm came uccccm acme cmcccc acme cccucc acme cmcccm acme accucm pace cocanac :onmm .Hmaa maam .mmnm 20aumospm msouw oacnum mom xwm APPENDIX G PROPORTION CORRECT ON EACH SPEAKER CHARACTERISTIC FOR THE TRAINED AND UNTRAINED GROUPS UNDER THE SPEECH TREATMENT COMBINATIONS FOR THE TEST AND RETEST CONDITIONS 286 cc.a cc.a cc.a cc. mm. mm. cc.a cc.a cc.H cc.a cc.a cc.a mega ac. mm. cc. cc. cc.a cc.a cc.a cc.a mm. mm. cc.a cc.a 02am a ac. ac. cc.a cc. cc.a cc.a cc.a cc.a cc.a ac. cc.a cc.a cecc c ac. ac. cc. cc.a ac. ac. cc. cc. ac. cm. cc.a cc.a ozcc cc.H cc.a cc. cc.a cc. ac. cc.a cc.a ac. ac. cc.a cc.a mesa ac. mm. cc. cc. ac. cc.a cc.a cc.a ac. mm. cc.a cc.a oZuc c mm. mm. cc. cc. cc.a cc.a cc.a cc.a mm. ac. cc.a cc.a cecc c ac. cc.a cc.a cc.a ac. mm. cc. cc. mm. ac. cc.a cc.a ozcc cc.a cc.a cc. cc. mm. mm. cc.a cc.a cc.a cc.a cc.a cc.a mesa ac. ac. cc. cc. cc.a ac. cc.H cc.H mm. mm. cc.a cc.H ozuc c ac. ac. cc.- cc. ac. cc.a cc.a cc.a mm. mm. cc.a cc.a mean m mm. mm. cc.a. cc.a mm. mm. cc. cc. mm. mm. cc.H cc.H 02cc ac. cc.a cc.. cc. cm. cc. cc.a ccca ccaa coca cc.a cc.a mama ac. ac. cc.. cc.a cc. cc.a cc.a cc.H mm. mm. cc.a cc.a czac a cc.a cc.a cc.. cc. cc.a cc.a cc.a cc.a ac. ac. cc.a cc.a cecc c cc.a cc.a cc.a» cc.H cm. ac. cc. cc. ac. ac. cc.a cc.a ozcc cc.a ac. cc.. cc. mm. mm. cc.a cc.a ac. ac. cc.a cc.a cane ac. cc. cc.. cc. mm. mm. cc.a cc.a mm. mm. cc.a cc.a ozac m cc.a ac. cc.. cc. mm. ac. cc.a cc.a mm. cc. cc. cc.H cecc m mm. mm. cc.a. cc.a mm. mm. cc. cc. mm. cc. cc.a cc.a ozcc cc.H ac. cc.. cc. mm. mm. cc.a cc.a cc.a cc.a cc.a cc.a came mm. mm. cc.H cc. ac. mm. cc.a cc.a cc.a mm. cc.a cc.a och a ac. ac. cc.. cc. cc.a cc.a cc.H cc.a ac. ac. cc.a cc.a cecc c ac. mm. cc.a. cc.H ac. mm. cc. cc. ac. ac. cc.a cc.a ozcc ac. ac. cc.a. cc.a ac. mm. cc.a cc.a ac. cc.a cc.a cc.a mama mm. mm. cc. cc.H ac. cc.a cc.a cc.a cc.a cc.a cc.a cc.a ozHc ac. ac. cc. cc. mm. mm. cc.H cc.a ac. ac. cc.a cc.a cacc Hc ac. ac. cc.a cc.a mm. mm. cc. cc. mm. mm. cc.a cc.a 02cc cccccm acme cccccm acme cmcccm acme cmcccm acme cccccm acme cccccm acme cocanac coaumospm moouw oacnum mm< mBUHmem QMZH0. mm. oo.H oo.H mama m mm. no. oo.H oo.H mm. 00. om. om. mm. no. oo.H oo.H 02mm mm. oo.H om. om. mm. 00. om. om. oo.H am. oo.H oo.H chum am. no. om. om. no. oo.H oo.H oo.H mm. mm. oo.H oo.H OZHm OH mm. no. om. 00. no. no. oo.H oo.H 00. no. oo.H oo.H mamm m we. mm. oo.H oo.H oo.H no. om. om. am. am. oo.H oo.H 020m 00. mm. om. om. 00. mm. oo.H oo.a am. no. oo.H oo.H mesa mm. 00. oo. om. oo.H mm. oo.H oo.H no. no. oo.a oo.H osz m no. no. om. oo.H mm. aw. oo.H om. no. 00. oo.H oo.H wamm m mm. mm. oo.a oo.H oo.H oo. om. oo. oo. oo. oo.H oo.a 02mm mm. oo.H om. om. mm. no. oo.H oo.H ac. oo.a oo.H oo.H mama mm. oo.H om. om. mm. no. oo.H oo.H no. oo.H oo.H oo.H OZHA m no. no. om. oo.H no. no. oo.H oo.H mm. mm. oo.H oo.H wemm m we. mm. oo.H oo.H mm. oo.a om. om. mm. mm. oo.H oo.H ozmm wmmuwm umwa ummuwm umms umwuwm ummB ummuwm puma ummumm pmmB uwwumm umma powhndm coammm .aman mNHm .mwcm COHDmospm msouw oaccum wm< xmm UwscfiuCOUIlmBUMmew QWZHHmlleUmEumDm QMZH‘m—BZD 308 accc. cccc. aaaa. cccc. cccc. ccca. accc. accc. mac cccc. aaac. accc. accc. accc. accc. cccc. aaac. aac acaa. aacc. accc. accc. cccc. accc. cccc. cccc. Haw ccca. aacc. mace. accc. aacc. aacc. mmcc. acac. cac aacc. accc. acac. acaa. mmaa. mcaa. cccc. cccc. cam cccc. aaac. ccac. cccc. accc. aaaa. accc. acac. cam ccec. coac. acaa. ccca. mcmc. accc. acac. accc. aam umouom Home uwouom Home uwouom some amouom umma Doonnsm oconmoaoa omummonm HmEHoz ooumooum ococmoaoa msooomucomm HmEHoz msooCMDCOQm UwSCHuCOUIImmqmzmm mummmm QZOUmm m>thlm80m5mDm QMZHfiMBZD 309 ccca. cccc. accc. accc. cccc. cccc. aaaa. ccca. c aaac. aaca. ccca. acac. ccec. accc. cccc. aaaa. cam mcaa. mmmc. ccca. acaa. ccac. aacc. aaca. ccca. vac accc. ccca. accc. acaa. accc. accc. cccc. cccc. mac accc. accc. ccac. acaa. aacc. accc. accc. ccca. aac aaca. accc. aacc. cccc. acaa. ccca. ccca. ccca. Ham accc. accc. accc. cccc. acac. cccc. ccca. accc. cam cccc. cccc. acaa. ccac. mmaa. ccec. aaca. aaaa. cc aaca. cccc. cccc. ccca. ccac. cccc. aacc. aaca. cc cccc. cccc. cccc. cmcc. ccc.a accc. acaa. ccac. ac ccca. accc. cccc. mcmc. cccc. aacc. accc. accc. cc aaac. ccmc. acaa. cccc. accc. accc. ccca. acaa. cc ccca. ccca. cccc. acaa. accc. cccc. accc. mmmc. cc cccc. accc. cccc. accc. cccc.a ccec. ccca. accc. cc aaca. aaca. ccca. ccca. cccc. aaaa. aacc. ccmc. ac cccc. aaca. mcmc. cccc. cccc. ccca. accc. cccc. Hc cmcccc acme cmcccc acme cmcccm acme cccccm acme cccanac oconmoaoa oouwmoum HmEHoz oonmmonm ocosmoaoa mooocmucomm HmEHoz wsoocmucomm mmflmiflm mummmm 9200mm ZMBIImBUMmem QNZHmmBZD 310 ccca. aaca. ccca. accc. camc. accc. acaa. aaaa. mac aaca. aaca. cccc. ccac. accc. aacc. ccca. ccac. aac accc. accc. ccac. accc. acaa. accc. cccc. accc. Ham accc. accc. accc. ccmc. cccc. acaa. acac. accc. cam ccca. ccca. ccac. aaac. ccec. accc. ccac. cccc. cam cccc. cha. aaca. acaa. mmmc. accc. ccca. cccc. cam ccca. aacc. aacc. mmmc. cccc. acaa. ccca. ccca. aac cmcccm acme cmcccm acme cccccc acme cmcccc came cocancc oconooaoe ooummoum HmEHoz ooumoonm ocosmoaoa msooCNDGOQm HmEHoz mooocmucomm DOSCHDCOUIImWQQZGm mommmm 0200mm ZHBIImBUmeDm QMZHHmllmBUMmem QNZHANMB 314 mm accc. accc. mmcc. aaca. mcmc. accc. . mmcc. mcmc. c cccc. ccca. accc. cccc. accc. ccmc. aaac. cccc. aac accc. aaaa. accc. cccc. acaa. accc. accc. cccc. Hac mmaa. acaa. cccc. cccc. accc. mcmc. mmcc. accc. cac cccc. aacc. cccc. accc. cccc. accc. cccc. cccc. cac aaac. acac. accc. cccc. cccc. ccca. cccc. cccc. cac ccca. ccca. accc. aacc. ccac. cch. accc. accc. aac cccccc acme cccccm acme pccccc ccce cccccm acme cocancc oconmoama ooummoum HmEHoz ooummoum GOSCHHCOUIImmqmzflm oconmoaoe csomcmucomm HmEHoz csoocmucomm mUflmmm QZOUWm m>HmllmBUmmem QMZHfiMB 315 aaca. aaac. ccca. cccc. accc. mmcc. accc. acaa. c ccca. accc. ccmc. cccc. ccca. aamc. acaa. mcmc. cac accc. cccc. ccca. aaca. ccca. ccec. cccc. accc. eac aaca. mcmc. accc. cccc. ccec. cccc. accc. ccca. cac ccca. cccc. aacc. aacc. mmmc. mmmc. accc. aacc. aac aacc. cccc. acaa. accc. ccmc. cccc. cccc. accc. aac accc. accc. acac. acac. accc. accc. ccac. ccac. cac cccc. ccca. cccc. accc. ccaa. mmmc. accc. ccec. cc cccc. cccc. cccc. accc. acaa. accc. cccc. cccc. cc accc. ccca. accc. accc. cccc. cccc.a accc. aacc. ac acaa. aaac. accc. ccmc. cccc. accc. mcaa. accc. cc ccca. ccac. acaa. acaa. accc. accc. ccca. cccc. cc cccc. accc. cccc. accc. mmmc. ccec. aacc. ccca. cc accc. accc. ccac. acaa. accc. mcmc. cmmc. acaa. mc ccac. aacc. cccc. cccc. aaaa. acaa. accc. accc. ac ccca. accc. accc. aaca. acaa. acaa. acaa. ccca. ac cccccm acme cccucc ccca cccccc acme cccucm acme cccaaac ococmoaoe ooammoam HmanZ ooammoam ococmwaoe csoocmucomm amEHoz cooocmucomm mMAQZGm mommmm 9200mm ZmBIImBUMmew QWZH