LISTENERS’ PREFERENCES FOR ‘THE‘RATE 0F PRESENTATION OF RECORDED INFORMATION . Dissertation for the Degree of Ph. D. A MICHIGAN STATE UNIVERSITY SAMUEL JOSEPH LEVINE 1974 This is to certify that the thesis entitled LISTENERS' PREFERENCES FOR THE RATE OF PRESENTATION OF RECORDED INFORMATION presented by Samuel Joseph Levine has been accepted towards fulfillment of the requirements for PhoDo Acmein Education #519 A) , Ajfiz Major professor III II IIIIIIIIII L, 1}“ *fi 1 II KZ3U ABSTRACT LISTENERS' PREFERENCES FOR THE RATE OF PRESENTATION OF RECORDED INFORMATION BY Samuel Joseph Levine Previous studies show that a listener has the po- tential to receive recorded information at a rate far ex- ceeding the rates that are used in conversation and for the production of tape recordings. However, few studies have examined listeners' rate preferences. ,By using an experi- mental setting that allowed listeners to autonomously man- ipulate the rate of presentation of recorded information, it was proposed that listeners would manifest a preference for rate of presentation and would demonstrate rate manipu- lation behaviors that were related to the difference be- tween the initial rate of presentation of the recorded in- formation and the listener's preferred rate. Forty-eight elementary school children in the third, fourth and fifth grade listened to a series of four recorded presentations. While listening to each recorded presenta- tion, the subjects were allowed to manipulate the rate of presentation of the recording through the use of a speech -y Samuel Joseph Levine ‘Acompressor. Each of the four presentations began at a dif- ferent initial presentation rate. All subjects listened to the same four recorded passages. The four initial presen- tation rates used in the study were 100 words per minute, 150 words per minute, 200 words per minute, and 275 words per minute. The rate manipulation behaviors of each subject were recorded on a strip chart recorder for later analysis. The results of the analyses of third, fourth and fifth grade subjects' rate manipulation behaviors support the following conclusions: 1. Children will manipulate the rate of presenta- tion of recorded information in a self-paced listening situ- ation. 2. Children demonstrate a preference for rate. 3. The extent a listener alters the listening rate is positively related to the difference between the in- itial rate of presentation of recorded information and the listener's manifest preference for rate. The findings suggest a disparity between the rate at which a student is able to listen and the rate at which he prefers to listen to recorded information. This study has also suggested that an initial presentation rate for re- corded information that varies greatly from the listener's preferred rate of presentation will stimulate greater rate change than an initial presentation rate that is close to Samuel Joseph Levine the listener's preferred rate. As such, instructional ma- terials that are designed for use in a self-paced listening environment will be more likely to be altered by the sub- ject toward a preferred rate when the initial rate of pre- sentation is more different from the listener's preferred rate . LISTENERS' PREFERENCES FOR THE RATE OF PRESENTATION OF RECORDED INFORMATION BY Samuel Joseph Levine 'A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY College of Education 1974 ACKNOWLEDGMENTS The writer would like to acknowledge the encorage- ment and support provided by the following people: To Dr. Ted Ward for providing continued help both as dissertation director and professional colleague. Ted's understanding of learning and the instructional process has served as a stimulus for my own professional development which has made the task of completing this dissertation both enjoyable and challenging. To Mrs. Lou Alonso, as director of the Great Lakes Region Special Education Instructional Materials Center and member of the doctoral guidance committee, for allowing me untold opportunity to explore all levels of instructional development. To Dr. Ed Keller and Dr. George Myers, members of the doctoral guidance committee, for providing a professional atmosphere truely conducive to advanced graduate study. To Robbie and Jaclyn Shoshana, my wife and daughter, for having the patience to see this dissertation finally un- dertaken and then providing the encouragement to see it completed.‘ To the staff of the Media and Materials Development Unit of the G.L.R.S.E.I.M.C. for their assistance and es- pecially my secretary, Mrs. Lynn Kinzel, who has spent many hours seeing this document through to completion. ii TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . LIST OF CHAPTER I. II. III. IV. FIGURES . . . . . . . . . . . . RATIONALE OF THE STUDY . . . Background . . . . . . Statement of the Proble Purpose of the Study . Importance of the Study Research Questions . . Hypotheses to be Tested Definition of Terms . . REVIEW OF RELATED LITERATURE History and Background Comprehension and Rate 0 Presentation . . . Listener Preference for of Presentation . Recent Advances . . . . METHODS AND PROCEDURES . . . Population and Sample . Treatments . . . . . . Instrumentation and Data Collection . . . . Data Analysis . . . . . f Rate FINDINGS O O O I O O O O O O O O O Manifest Preference for Rate of Presentation . Mean Manifest Preference Rate iii Page vii (DQU'IWUJNH H 13 14 21 3O 40 42 42 43 50 52 61 61 72 CHAPTER Acceptance Band Deviations Effects . V. CONCLUSIONS, Conclusions . . Discussion . . Implications . BIBLIOGRAPHY . . . . . . . . . APPENDIX A . . . . . . . . . . APPENDIX B . . . . . . . . . . Position of the Longest Non- Manipulated Duration . . . . . Non-Manipulated Segments . . . . . Altering the Rate Preference Rate Manipulation Behavior . . . Analyzing with Reduced Standard Further Analysis for Random iv DISCUSSION AND IMPLICATIONS Page 75 78 79 83 89 93 102 102 103 111 116 119 121 TABLE 3.1 LIST OF TABLES LENGTH AND ORIGINAL WORD RATE OF STORY SEGMENTS O O I O O O O O O O O O O CONFIGURATIONS OF STIMULUS TAPES . . . . DURATION AND LENGTH OF PRE-ALTERED SEGMENTS . O O O O O O O C O O O C O O O CONVERGENCE BEHAVIOR BY NUMBER OF SEGMENTS C O O O O O O O O O I O O O O O MOVEMENT TOWARD CONVERGENCE BY NUMBER OF SUBJECTS O O C O O O O O C O O O O O EXTENT OF MOVEMENT TOWARD POINT OF CONVERGENCE FOR ALL LISTENING SEGMENTS . CORRELATIONS BETWEEN THE EXTENT OF MOVEMENT TOWARD POINT OF CONVERGENCE AND THE DISTANCE BETWEEN INITIAL PRESENTATION RATE AND POINT OF CONVERGENCE . . . . . . . . . . . . . . DEMONSTRATED MANIFEST PREFERENCE FOR RATE 0 O O C O O C O C O C C O O C . ACCEPTANCE OF MANIFEST PREFERENCE FOR RATE ACCORDING TO CRITERIA #1 AND #2 MEAN MANIFEST PREFERENCE RATE BY SUBJECT . I C C C O O O O O C O O O O O MEAN MANIFEST PREFERENCE RATE FOR ALL GRADES COMBINED AND BY INDIVIDUAL GRADES POSITION OF LONGEST NON-MANIPULATED DURATION IN RELATION TO PRESENTATION ORDER OF SEGMENTS . . . . . . . . . . . Page 47 49 50 62 63 65 69 71 72 73 75 77 TABLE 4.10 DIFFERENCE FROM PREFERENCE FOR NON-MANIPULATED SEGMENTS . . . . . . BAND WIDTH OF FINAL SELECTED RATES BY SUBJECT NUMBER 0 O O C O O C O O DEMONSTRATED MANIFEST PREFERENCE FOR RATE AT DIFFERENT ACCEPTANCE LEVELS COMPARING MANIPULATION ONSET TO THE DIFFERENCE FROM PREFERENCE . . . . . COMPARING MANIPULATION TERMINATION TO THE DIFFERENCE FROM PREFERENCE . . . COMPARING MANIPULATION DURATION TO THE DIFFERENCE FROM PREFERENCE . . . COMPARING MANIPULATION ONSET TO THE DIFFERENCE FROM PREFERENCE USING THE AIJTEED DATA SET 0 O I C O O C O O O COMPARING MANIPULATION ONSET TO THE DIFFERENCE FROM PREFERENCE ON THE BASIS OF CRITERION #1 FOR SUCCESSIVELY MORE STRINGENT GROUPINGS . . . . . . . . COMPARING MANIPULATION TERMINATION TO THE DIFFERENCE FROM PREFERENCE ON THE BASIS OF CRITERION #1 FOR SUCCESSIVELY MORE STRINGENT GROUPINGS . . . . . . COMPARING MANIPULATION DURATION TO THE DIFFERENCE FROM PREFERENCE ON THE BASIS OF CRITERION #1 FOR SUCCESSIVELY MORE STRINGENT GROUPINGS . . . . . . . . vi Page 79 81 82 87 88 90 92 96 99 100 LIST OF FIGURES FIGURE Page 1.1 RELATIONSHIP OF DEFINED TERMS . . . . . . . . 12 3.1 ORDER OF EVENTS PRESENTED ON RECORDING . . . . 44 3.2 ACCEPTABLE AND NON—ACCEPTABLE DEMONSTRATION OF PREFERENCE . . . . . . . . . 55 3.3 ANALYZING DEMONSTRATED RATE PREFERENCE ON THE BASIS OF CRITERION #2 WHEN CRITERION #1 HAS NOT BEEN MET . . . . . . . . 56 3.4 ELIMINATION OF A SINGLE SKEWED TERMINAL RATE IN CALCULATING THE MEAN MANIFEST PREFERENCE RATE 0 O O O O I O I O O O O O O O 5 8 3.5 INSTANCES WHEN MEAN MANIFEST PREFERENCE RATE MUST BE CALCULATED FROM ALL TERMINAL RATES O O O O O O C O O O O O O O O O 5 9 vii CHAPTER I RATIONALE OF THE STUDY Background The differences between communicating through speech and communicating through the printed word have been exam- ined through many studies. A primary difference between the two forms of communication lies in the ability of the re- ceiver to alter the rate at which the communicated informa— tion is received. The reader is able to adjust his reading rate to suit his particular reading ability, to suit his mood, to match the difficulty of the material, and to cor- respond to the necessity of retaining the information that is presented. The listener, however, is dependent upon the rate of presentation of the person delivering the informa- tion. This dependency is found in both the face-to-face verbal interchange and in pre-recorded information delivery. In the face-to-face situation the receiver may, however, request the sender to increase or decrease the rate of pre- sentation. This is not possible when listening to record- ings. In most cases the reader operates with a self-paced system whereas the listener must utilize an externally paced system. This study has been designed to examine listening behaviors when the listener is given the opportunity to utilize a self-paced system for adjusting the rate of pre- sentation of recorded information. Statement of the Problem The use of compressed speech as a procedure for altering the rate of presentation of recorded material has been shown to be an effective tool for increasing the efficiency of learning through listening. Educators are just beginning to realize the implications of this proce— dure and schools are starting to provide pre-compressed materials for students. It has been shown that listeners can more than double the rate of presentation of recorded material without a significant decrease in their compre- hension. However, it has been assumed that a listener who is provided with a pre-compressed recording will want the recorded material presented at the fastest rate at which he can attend and still comprehend. No study has yet been con- ducted on the situation in which the listener has autonomous control over the rate of aural presentation. A break- through in technology makes this an important question. Speech compressors first became available at a cost of five thousand dollars per unit with added expense necessary for peripheral equipment. This cost drOpped drastically in 1973 with the availability of two different units that were priced at fifteen hundred dollars and five hundred dollars. Finally, in 1974, information was released regarding the availability, within a year, of a speech compressor that would sell for approximately one hundred dollars. With the advent of inexpensive speech compression equipment that provides for instantaneous playback of rate- altered recordings, many students will be given the oppor- tunity to self-pace the playback of recorded material. The impending availability of this equipment reinforces the need for research that establishes a more complete under- standing of self-paced listening behavior. The dream of widespread availability of speech compressors is soon to be realized and there is little information regarding how this equipment will be used. Purpose of the Study The primary purpose of this study was to ascertain if listeners, given the opportunity to alter the rate of presentation of recorded material, in fact will alter the rate. Further, if listeners do alter the rate, what rate- altering behaviors will they display? And, is it possible to gather meaningful data regarding rate-altering behavior for elementary school students? Finally, is it possible to determine an individual listener's manifest listening rate preference for recorded information? Importance of the Study A review of studies examining the use of compressed speech indicates a primary concern with ascertaining the upper limits at which a listener can comprehend compressed recorded material without a significant decrease in com- prehension. To fully understand the use of compressed speech it is necessary to examine the rate a listener will select for listening to the recorded speech. Such an exam— ination will allow the establishment of groundwork for the definition of the operational limits for training proce- dures that can be used to develop an individual's listening behavior from the point of preference to the point of max: imum efficiency. The point of maximum efficiency is defined as the maximum rate that does not show a significant de- crease in comprehension. It can be expected that inexpensive speech compres- sion equipment will be available for general use within the next year and used on a widespread basis within the next five years. This study attempts to establish a basis for the development of guidelines for effective use of this equipment so that listener selection of rate can become a systematic process based on the individual behaviors of listeners in relation to the potential behaviors of lis- teners. In 1967 Friedman, Graae, and Orr reported: Given the state of current technology, self-pacing is of limited practical value since it is not feasible to make available machines for extensive individual use. However, it may be practical to provide a machine for a school library for indi— vidual use as an auditory review mechanism for material with which the student is already famil- iar. (Friedman 32 al, 1967, p. 27) In the seven years since the Friedman, Graae, and Orr study was completed, technological advancements have opened up new frontiers. For the first time, self-pacing will be available and it can be expected to have great practical value. This will bring to fruition the need that Richard Kinney spoke of at the 1966 Louisville Conference on Time Compressed Speech. Ultimately, what we need in this field is a device that will give the same options to the auditory reader. A device that will lift him to the plain of the visual reader with all the visual reader's freedom of choice. This could be done, I believe, only through an individual compressor—expansor unit for playing back recorded speech. In other words, a talking book or a tape recorder that contains its own personal compressor-expansor so that the listener could speed up or slow down the speech to which he is listening according to his need just as the visual reader unconsciously does. Research Questions The study addresses itself to the basic question of whether or not a listener will express his listening-rate preference by manipulating the rate of presentation of recorded information when given the opportunity to do so. To further specify this question and to provide a set of testable hypotheses, two self—paced listening behaviors will be examined. An experimental procedure was designed 1 Kinney, Richard, "Report on Studies in Speech Compression Conducted in the Spring of 1966 by the Hadley School for the BIind," from Proceedings of the Louisville Conference on Time Compressed Speech (Louisville: University of Louis- ville, 1967), p. 41. to permit the examination of these behaviors and the col- lection of data to analyze relationships between these behaviors and a self-paced listening experience. Behavior #1 - The listener's manifest listeningérate preference for recorded information. By providing a series of rate altered listening segments, each presented at a different rate, the listener will have an Opportunity to alter any or all of the segments to better accommodate his own preference. The study will examine whether or not a manifest preference for rate of presentation does exist, whether or not a listener will alter the rate of presenta- tion to better suit his individual preference, and if it is possible to ascertain what this manifest preference is for individual listeners. Behavior #2 - The listener's manipulation of rate of pre- sentation. Three variables will be examined to better understand instructional options that may become available when self-paced listening opportunities are provided within the school setting. These three variables are the onset of manipulation of rate (manipulation onset) as compared to the initial rate deviation from manifest preference (dif- ference from preference), the termination of manipulation of rate (manipulation termination) as compared to the dif- ference from preference, and the duration of manipulation of rate (manipulation duration) as compared to the difference from preference. The listener will be presented four recorded seg- ments each presented at a different pre-altered rate. For segments that deviate greatly from the manifest preference, it is expected that the onset of the listener's manipu- lation of the rate would occur early in the listening ex- perience. It is not known whether this deviation from man- ifest preference would relate predictably to the termina— tion or the duration of the manipulation. This study examines all three of these variables to determine if relationships exist between the initial rate of presentation and manipulation behaviors. Hypotheses to be Tested Based on these questions regarding behaviors that are demonstrated in a self-paced listening situation, the following testable hypotheses were formulated: Hypothesis #1 A manifest preference for rate of presentation of recorded information will be demonstrated when the listener is given autonomous control over the rate of presentation in a self-paced listening situation. Hypothesis #2 A negative correlation exists be- tween the elapsed time before manipulation onset and the listener's difference from preference. As the difference from preference increases the manipulation onset will occur sooner in the listening experience. 8 Hypothesis #3 A correlation exists between the elapsed time before manipulation termination and the lis- tener's difference from preference. The direction of this correlation is not hypothesized. Hypothesis #4 A correlation exists between manipu- lation duration and the listener's difference from prefer- ence. The direction of this correlation is not hypothe- sized. Definition of Terms The following terms and phrases are used in the description of this study. Definitions for each term and phrase are provided to form a common basis for under- standing. Listener A subject in the study, having no known hearing deficits. Rate The speed of presentation of recorded material. For this study, rate will be presented in words per minute (wpm). Experiments utilizing compressed speech have variously used "percentage of compression," "per- centage of time saved," and "words per minute." Words per minute has been selected as the term to be used due to the expressiveness of the phrase, the immediate clarity of meaning, and the precision of measurement. In describing compressed speech, specification in terms of word rate appears to be necessary, and it is probably sufficient. Word rate is probably the most meaningful dimension in terms of cogni- tive and perceptual processes of the listener. Compressed Speech Speech that has been accelerated in rate of presentation without a resultant change in pitch from that of the original recording. Though peri- odic samples of the original recorded material have been deleted, comprehensibility is not affected by the deletions. Usually Speech compression is accomplished by processing the recorded material through a "speech compressor." Expanded Speech Speech that has been slowed down in rate of presentation without a resultant change in pitch from that of the original recording. Though small pause segments are inserted at periodic intervals, comprehensibility is not affected due to the insertions. Usually accomplished by processing the recorded material through a "Speech compressor" that is operating in the "expand" mode. Speech Compressor A specialized device, electronic or electro-mechanical in nature, for altering the rate of tape recordings. The speech compressor periodically deletes small samples of the original recording or adds 2 Foulke, E. and Sticht, T. "A Review of Research on Time Compressed Speech," from Proceedings of the Louisville Conference on Time Compressed Speech (Louisville: University of Louisville, 1967). P. 6. 10 pause segments to the original recording to affect an altered rate without subsequent changes in the pitch of the original material. Alterations of rate are accomplished through the manipulation of a rate con- trol knob. Self-pacipg» A listening situation wherein the listener has autonomous control of the rate of presentation of re- corded material. This term is used to differentiate from "external pacing" where the rate of presentation of recorded material is controlled by persons or equipment not in the direct control or influence by the listener. Chart Recorder An electro—mechanical transducer that in- scribes a line on a continuous roll of paper. The paper moves at a known rate and the scriber deflects in response to the subject's function. For this study the chart will provide a documentary record of the rate selections made by the listener. Manipulation onset The point of time, measured from the be- ginning of a listening segment, at which the listener begins to alter the rate of presentation (see Figure 1.1 at indicator 3). Manipulation termination The point of time, measured from the beginning of a listening segment, after which no further alterations of rate are made by the listener (see Figure 1.1 at indicator p). 11 Manipulation duration The elapsed time between the onset of manipulation and the termination of manipulation (see Figure 1.1 at indicator 3). Manifest preference for rate A grouping of final selected rates of a series of listening segments that does not exceed 40 wpm in width. This term is further defined in Chapter III (see Figure 1.1 at indicator g). Mean manifest preference rate A single rate computed to denote the mean of the manifest preference for rate. This term is further defined in Chapter III (see Figure 1.1 at indicator 3). Difference from preference The rate differential between the initial presentation rate of a segment and the mean manifest preference rate (see Figure 1.1 at indicator 9. Point of convergence The arithmetic mean of a subject's final selected rates that has not been altered or ad- justed through the use of any criterion defined in Chapter III. This term is used in Chapter IV to examine rate manipulation behavior for all subjects without regard for exclusions used to compute mean man- ifest preference rate (see Figure 1.1 at indicator g). Extent of Movement The rate differential between the initial presentation rate of a segment and the final selected rate (see Figure 1.1 at indicator h). 12 HYPOTHETICAL RATE-CHANGING BEHAVIORS FOR A LISTENER RATE (expressed in words per minute) TIME (expressed in seconds) Segment 1 Segment 2 Segment 3 Segment 4 100 150 200 275 0 __ (a) ‘f" (a) 15 I I I (a) 30 I I I I I 45 (C) I (b: I I 6° I (b) I I I 75 I I I I I I I I """"""""" (b) I 90 I + ———————— 0-—(h) —————— +: I 105 III . I I I IdII—I—U , '(g) I I I I 120 (elI I I +<~ —————————— (f) —————————— +I (a) (b) (C) (d) (e) (f) (g) (h) manipulation onset manipulation termination manipulation duration (for segment 1) manifest preference for rate mean manifest preference rate (for three segments) difference from preference (for segment 4) point of convergence (for all segments) extent of movement (for segment 4) W FIGURE 1.1 RELATIONSHIP OF DEFINED TERMS CHAPTER II REVIEW OF RELATED LITERATURE The review of related literature provides an over- view of the entire area of compressed Speech research and highlights those studies that have examined listeners' preferences for rate of presentation. Though very few studies have dealt with rate—preference in a self-paced listening situation, the studies that are reviewed form a logical theoretical basis for the study. The first of the four sections of this review, History and Background, deals with the key experiments in speeded listening (prior to the deve10pment of compressed speech techniques) and the first experiments that were con— cerned with the deve10pment of speech compression techniques. The viability of the use of compressed speech as a vehicle for altering the rate of presentation of recorded material without adversely altering comprehension or intelligibility is established. The second section, Comprehension and Rate of Presentation, reviews those studies that have examined rate thresholds for listening; the parameters of rate of compression, especially the probable tolerances for changed 13 14 rate, that should be used in a study of listener rate pref- erence are deduced. The third section, Listener Preference for Rate of Presentation, reviews all known research studies that have examined some aspect of listeners' preferences for rate of presentation. The review includes a range of studies wherein listeners were merely questioned as to their rate preference, were provided a finite set of rates to select from, or were provided infinite selection of rates for total self-pacing. The basis of research procedures for the proposed study is deduced from these studies. The fourth section presents Recent Advances in the technology of speech compression. History and Backgpound Goldstein (1940) provided one of the earliest inves— tigations into the effect on comprehension of increased rates of presentation of recorded materials. The study of 280 adult subjects with a mean age of 34.3 used both auditory stimulus and visual stimulus. A total of 28 passages were selected, sub-divided equally according to level of diffi- culty (14 were at a 3.5 grade level and 14 were at a 7.5 grade level). The subjects listened to a record that pre- sented each passage at rates varying from 100 words per min— ute to 322 words per minute, in increments of 37 words per minute. Though the study was primarily interested in exam- ining differences between reading and listening, the results were the first indications that listeners can process 15 auditory information at a rate faster than that which is encountered in normal conversation or that which is used for presentation. Goldstein found that comprehension de- clined as rate was increased; however, the differences between reading comprehension and listening comprehension grew smaller as the rates of both increased. The findings suggested that at faster rates both processes (reading and listening) were equally effective, or more precisely, equally ineffective. A potential source of invalidity in the measurement of listening comprehension in the Goldstein study was due to the pitch increase that accompanied the in- creased rate of presentation: the recordings were altered in rate by merely speeding up the replay to gain faster words per minute rates, simultaneously increasing the pitch. Later studies show that comprehension does not decline for increased rate listening tasks when original pitch is main- tained until a certain rate is achieved (Fairbanks 2E.El 1957b, Bixler 23.21 1961, Foulke g; Q; 1962, Sticht 1968, Gropper 1969). The pitch change in the Goldstein study is one potential reason why the comprehension of the listening passage declined at faster word rates. Garvey (1953) further points out that the Goldstein study required readers to speak at rates higher than normal when making the recordings for the study. This was done in an attempt to somewhat lessen the effects of pitch change on comprehension. Garvey states that the "method is limited by the rate at which the reader 16 is able to speak, and also by the changes in enunciation and timing involved in attempting to Speak rapidly." (Garvey, 1953, p. 102) Goldstein's findings were later supported by Fergen (1954) who found that when 438 subjects in grades four, five and six were presented with proqressively faster rates (80 me, 130 wpm, 180 wpm, and 230 me) created by having the reader read faster, that comprehension increased from 80 wpm to 130 wpm, but fell off after 130 me. Grumpelt and Rubin (1972), however, found that increased rate with resul- tant pitch change could Show higher comprehension if training at the increased rates were provided for the subjects. In their study, 66 blind high school students were divided into control and experimental groups. Both groups were equated according to age, IQ, and pre-test comprehension of recorded material. The experimental group received listening train- ing at speeds from 275 wpm to 300 wpm while the control group received similar training at only 175 wpm. On a post-test presented at 300 me, the experimental group scored signifi— cantly higher than the control group, though both groups showed a decline in comprehension from the pre-test. While the training effect was supported as a procedure for improv- ing comprehension, the study indicated that the improvement in the experimental group was still small (9.4%). In an attempt to control for the effect of pitch change as a contaminant to comprehension of listening to 17 recorded materials at fast rates, Garvey (1953) utilized a "chop-Splice" technique to alter rate without altering pitch. Garvey based his procedure on a study by Miller and Licklider (1950) who found that undistorted Speech patterns contain an excess of cues which, though utilized by the listener, are not essential to intelligibility. In their work, Miller and Licklider electronically interrupted the presentation of speech by systematically turning the Speech on and off at a desired rate. They reported that at a rate of ten times per second the intelligibility of monosyllabic words did not drop below 90% until over 50% of the original speech pattern had been removed. In this study, however, no attempt was made to close the gaps produced by turning the speech off. Garvey's study attempted to gather data on the effect of closing this gap and thereby increasing the rate while presenting Speech with fewer cues for the listener. Garvey recorded Spondaic words (baseball, sunset, etc.) on a specially modified tape recorder that moved the tape at 40 cm. per second instead of the standard 19.5 cm. per second. This procedure yielded a tape record of a word on a longer section of tape than is typical and thereby pro- vided more room for chopping and splicing the tape. Care- fully marking the beginning and end of the recorded word, he systematically removed short sections of the tape record at Specified intervals in between. The remaining 18 sections of the tape were spliced together and then trans- ferred to another tape for use in presentation to the sub- jects. Garvey studied 96 male college students, randomly assigned in groups of Six to 16 different treatment vari- ables. He was concerned with the relationships between intelligibility and a) the general effect of acceleration, b) the Size of the chop removed when acceleration was held constant at twice the original speed, and c) the percentage of the Speech pattern removed. Garvey found that a mean intelligibility score of 78% was still obtained when words were presented at 3.0 times the original speed. It was not until the speed was 3.5 times original that intelligibility approached the 50% level. He further found that discrete Spondaic words could be presented at 2.5 times original speed with no Significant decrease in intelligibility. In terms of the Size of the discarded segment, he found that intelligibility is not ad- versely affected until the discard segment is at least .0625 seconds (2.5 cm.) in length. Finally, comparing speeded rates with and without resultant frequency (pitch) shifts, Garvey found: At an acceleration of 1.75, using the same test words and identical experimental procedure, the mean of the intelligibility scores, with the concomitant fre- quency shift, was 90%; for the same degree of accel- eration without frequency shift the mean intelligi- bility score was 95%. For an acceleration of 2.0, mean intelligibility with the frequency shift was 65%, but was 95% with the "chop-splice" technique. 19 When the acceleration was increased to 2.5, a mean intelligibility score of less than 10% was obtained with the frequency Shift and a mean in- telligibility score of 93% was obtained with the "chop-splice" technique. (Garvey, 1953, p. 106) This study by Garvey substantiates the usefulness of recorded material that is presented at rates faster than "normal" if altered to compensate for pitch change. The Garvey study, however, is limited in direct application to the general field of rate of listening since spondaic words, rather than connected discourse, formed the basis of the listening experience. Fairbanks, Everitt, and Jaeger (1954) reported the development of an electro-mechanical device that would allow the automation of Garvey's chop-Splice procedure. The Vari- Vox utilized a set of four heads mounted on a cylinder that rotated as the tape passed. One and only one of the four "pickup" heads was in contact with the tape at any time. However, at the point when one head left contact with the tape and the next head began contact with the tape a short portion of the passing tape was not reproduced. In this manner, it was possible to remove short segments of recorded information. The output of the Vari-Vox was fed into another tape recorder for generating a permanent taped record of the "compressed" version of the original tape recording. The development of this electro-mechanical device for Speech compression presented the first practical procedure for com— pressing large segments of recorded information and allowing 20 future studies to examine larger units of verbal informa- tion (connected discourse) rather than the small units examined in the earlier studies. Further, it permitted a viable procedure for Speeding up recorded information without resultant changes in pitch which have been Shown to adversely affect comprehension and intelligibility. Fairbanks, Guttman, and Miron (1957a, 1957b) con- ducted studies that examined comprehension as a function of word rate and the use of "saved" time due to speech compres- sion to re-present the recorded material. In their studies, it was found that there was little difference in comprehen— sion of recorded material presented at 141 wpm, 201 wpm, and 282 me. Also, when the rate of presentation was doubled (from 141 me to 282 wpm) and the material was presented twice, comprehension for subjects who listened twice was better than for subjects who were only allowed a single pre- sentation at 141 wpm. Both groups of subjects required the same amount of time for the listening task. By the late 1950's, the examination of Speech com- pression as a technique for presentation of recorded infor— mation at increased rates was established as a viable topic for researchers. Research had shown that it was possible to comprehend material presented at rates faster than normal, the procedure of increasing rate without affecting pitch was superior for learning as compared to procedures that dis- torted pitch, and equipment was finally available that 21 allowed the researcher a convenient procedure for compressing recorded material. Comprehension and Rate of Presentation The studies reviewed examine the listening rate in comparison with comprehension. The purpose of this section is to provide an understanding of the studies that have in- vestigated the optimization of the Speech compression pro- cess in terms of most effective and/or most efficient lis- tening rates. In all of the studies reported, an externally paced listening experience has been used. Some of the studies first presented, however, attempted to answer the question of rate preference without using self-pacing proce- dures by asking the subjects to state or choose their pre- ferred rate. Through an examination of the studies, it is possible to ascertain the appropriate range of presentation rates for utilization in this study. Prior to the use of Speech compression as a tech- nique for increasing rate of presentation, a few studies examined the effect on comprehension due to increased rate that was effected through non-machine procedures. Nelson (1948) chose to study radio broadcasts and the effect of rate of presentation on comprehension as meas- ured by recall. Five different newscasters each recorded five different newscasts at five different rates that varied from 125 wpm to 225 wpm in increments of 25 me. Though 22 Nelson found a decrease in comprehension from 125 wpm to 225 wpm, this decrease was not found to be significant. No preference was shown for any Single announcer that was used in the study and an analysis of reactions indicated a preference for a rate between 175 me and 200 wpm. Nelson used this last finding to hypothesize that: The Slight increase in mean scores between 175 and 200 words-per-minute (13.70 to 13.94) might indicate that the most "efficient" rate as meas— ured by student reaction might be somewhere be- tween 175 and 200 words-per-minute. (Nelson, 1948, p. 179) Harwood (1955) supported the earlier findings of Goldstein (1940) and Nelson (1948) when he found that lis- tenability (listening comprehension score) decreased as rate of presentation increased. For this study, Harwood used a Single male voice to create separate recordings at each rate. The speaker altered his rate of speaking to match the word rates needed for the study. No electro-mechanical pro- cedure for increasing word rate was used. Harwood also found that mean listenability, the mean of listening compre- hension scores, at each of four rates of presentation (125 wpm, 150 wpm, 175 wpm, and 200 me) did not Show sig- nificant differences that favored any specific word rate. This study did Show significant findings that supported "readability, as defined by Flesch, as an indicator of "listenability." As such, Harwood suggested the use of readability as a gross predictor of listenability. 23 Diehl, White, and Burk (1959) inquired into the relationship of rate of Speech and listener comprehension by altering only pause time. A master tape was produced with spoken discourse that was presented at a rate of 145 wpm. Four modified versions of this tape were then created. Tape A had 75% of each individual pause removed thereby creating a tape with a word rate of 172 wpm. Tape B had 50% of pause time removed and resulted in a new speed of 160 wpm. Tape C had the pause time increased by 75% which Slowed the rate to 135 me. Tape D had the pause time increased by 50% which slowed the rate to 126 wpm.1 Tape E was not changed and was presented at the original rate of 145 wpm. A total of 371 college students listened to both a comprehension test and a reaction rating scale. An analysis of the data on com- prehension showed no Significant differences in comprehension for any of the five different word rates. The responses to the subjective rating scale also failed to yield Significant differences with all five tapes rated as "good" to "very good" in terms of delivery to the Speaker. The findings in- dicated conflict with Nelson (1948) in that no rate, or rate range, was generally preferred. It was pointed out, however, that such a conflict was dependent upon the comparability of this Study's use of the word "good" and Nelson's use of the l The percentages and word per minute rates are Shown exactly as reported in the Diehl g; a; article. It is assumed that there is an error in the percentages at the 135 me and 126 me rates. The procedure would be more consistent if "50%" and "75%" were reversed. 24 word "interest." Nelson asked the subjects to respond in terms of their interest in the presentation at the different rates while Diehl £3 a; asked the subjects to rate the de- livery on a scale from very good to very poor. Fairbanks, Guttman, and Miron (1957b) analyzed the use of speech compression techniques and their resultant effects on comprehension of connected discourse. Technical information was recorded and compressed at a number of dif- ferent rates up to 70% compression (30% of original record- ing time). At 50% compression (50% of original recording time, 282 wpm) response to a factual test was almost 90% of maximum. However, at 60% compression (40% of original re- cording time, 353 wpm) response fell to 50% of maximum. This pioneering study in the examination of externally paced com- pressed presentations indicates that rates of at least 282 wpm are possible without significant loss of comprehension. It should be noted that the Fairbanks, Guttman, and Miron study was one of the first to examine connected discourse rather than Single word presentations. AS such, it was pos- sible to examine comprehension effects rather than merely examinations of intelligibility as were previously reported in discrete word studies. In a study by Bixler, Foulke, Amster and Nolan (1961), 299 blind children from Sixth, seventh, and eighth grades were provided compressed listening materials at 175 wpm, 225 wpm, 275 wpm, 325 wpm, and 375 wpm. Materials of 25 both a scientific nature and a literary nature were used and subject's comprehension was measured for both types of ma- terial at the five different rates. No significant loss in comprehension was shown for the literary material at 225 wpm and for the scientific material, comprehension held up at 275 wpm with no Significant loss. Foulke pg 3; (1962) also studied the effects of word rate of compressed material on comprehension. Word rates of 175 wpm, 225 wpm, 27S wpm, and 325 wpm (similar to the Bixler g; a; study, except omitting the 375 wpm rate) were used with blind subjects in grades Six, seven, and eight (same as Bixler). Results of the study indicated that for both literary and scientific passages, a rate of 275 wpm yielded better than 90% comprehension. Orr, Friedman, and Williams (1965) conducted a study of the effects of training in listening to compressed speech as it effects comprehension. The subjects showed Signifi- cant differences, in favor of a training effect, when a rate of 425 wpm was reached. However, no significant difference was found at rates of 175 wpm, 325 wpm, and 37S wpm. Those subjects that received no training showed only a 20% decline in performance at more than double the normal speaking rate. This finding supports the previous work of Fairbanks 3; a; (1957b) and Bixler g; a; (1961) and lead the authors to state that "even without practice, it appears that normal rates of Spoken material could be essentially doubled with little or 26 no loss in comprehension." (Orr §E_3I, 1965, p. 152) Wood (1965) compressed a group of fifty imperative sentences at rates from 175 wpm to 400 wpm in increments of 25 me and presented them in a controlled manner to 90 sub- jects from the first, third, and fifth grade levels. His findings supported other studies; generally comprehension decreased as rate of presentation increased. For the first- grade group, comprehension at rates above 250 wpm was con— siderably lower than that for the third and fifth grade groups. Interestingly, the subjects exceeded a 90% compre- hension level at rates as high as 350 wpm, and at no rate was the level of comprehension less than 75%. Intelligence was not found to be a statistically Significant factor. A possible explanation for the high levels of com- prehension at the high presentation rates (350 wpm) lies in the fact that discrete sentences were used in this experi- ment as compared to extended narratives comprised of contin- uous discourse as have been used in most other experiments. This seems to indicate a need for more mental processing time for comprehending continuous discourse than for dis- crete sentences. An interesting study by Spicker (1968) compared lis- tening comprehension and retention of "normal" and retarded children. Forty-four subjects of each classification were selected for the study by matching them according to mental age (10.0 to 11.0). Three separate passages differing in 27 readability levels were presented at 125 wpm (expanded), 175 me (normal), 225 wpm (compressed), and 275 wpm (com- pressed). Data Showed no Significant differences between groups for listening comprehension and comprehension losses. For the retarded group, comprehension losses were negligible at 125 wpm and greatest at 175 wpm. Conversely, the "nor- mal" group showed the greatest comprehension losses at 125 wpm and least at 175 wpm. This study tends to support nor- mal or below normal rates of presentation for subjects with a mental age of 10.0 to 11.0. Working with men of high, average, and low mental abilities, Sticht (1968) found that a message can be time compressed by as much as 36% (275 wpm) without greatly dis- turbing comprehension. For all three groups, comprehension declined only slightly until a rate of 275 wpm was reached and then comprehension declined more rapidly. Sticht states that these results are consistent with previous observations and goes on to say: There appears to be some Special Significance attached to a speech rate of 275-300 wpm such that exceeding this rate accelerates a decline in comprehension. This rate appears to be that at which channel capacity begins to be exceeded. (Sticht, Thomas G., 1968, p. 250) In an examination of comprehension as a function of listening rate, Gropper (1969) presented narrative passages at 126 wpm, 190 wpm, 252 wpm, 312 wpm, and 380 wpm to 72 fourth-grade subjects. AS was Shown in early studies, Gropper confirmed that performance on criterion tests 28 decreased as Speed (rate) increased. Significance for this decrease, however, was found only after the speed exceeded 252 wpm. When examining the data in terms of efficiency (learning per unit of time) the 252 me rate was the most efficient. Of particular interest to the present study was Gropper's finding that: Large individual differences were obtained indicating that there is not one most efficient Speed for every- one. In most cases, however, a speed much Slower than normal will not add much to comprehension, while Speeds about twice as fast as normal will take too much away from comprehension to warrant their use. (Gropper, Robert L., 1971, p. 252) Two other studies, though not directly related to comprehension as a function of rate, further establish guidelines for the present study. Durrell (1969) compared listening comprehension and reading comprehension. Durrell examined vocabulary and found that, in terms of comprehen- sion of vocabulary words, listening vocabulary is much superior to reading vocabulary in all primary grades. For the third grade, the reading-listening ratio was found to be 76% (listening score was 46 as compared to a reading score of 35). At the fourth grade level the ratio was 83% and at the fifth grade, listening was still superior with a ratio of 90%. When a comparison of paragraphs was made (longer units of language than separate words as reported above), the differences decreased slightly but listening was still superior for third, fourth, and fifth graders (82%, 83%, 29 and 95% respectively). From grade six on, the scores on reading paragraphs are higher than those in listening. Several fac- tors may account for this superiority in reading to listening: Silent reading speed at these levels is greater than the Speed of Speech, hence more time is available for "looking back" to check com- prehension; many words are introduced in silent reading that may not be in listening vocabularies. (Durrell, 1969, p. 458) When the raw Scores of listening tests were equated to a reading grade, Durrell found that the third grade stu- dents were operating one year and one month higher in lis- tening than reading (4.6 for listening, 3.5 for reading), nine months higher for the fourth grade (5.4 for listening, 4.5 for reading), and seven months higher for the fifth grade (6.2 for listening, 5.5 for reading). These findings indicate that listening, when compared to reading, is a superior form of communication for third, fourth, and fifth grade students. In a study that examined the effects of compressing recorded material and then returning it to normal, Sticht (1970) found that: EXpanding previously compressed materials to restore the word rate to normal may restore the comprehension of the material to very near normal -- when the com- prehension/expansion is limited to 40%. When the materials are compressed/expanded by 47%, there is apparently enough noise and/or Signal distortion added to reduce comprehensibility of the material significantly below normal, although the restoration of a normal word rate appears to improve the compre- hensibility of the material to a limited degree. (Sticht, 1970, p. 107) Sticht's findings are important since the present 30 study utilized the speech compressor twice. The recorded material was first pre—altered to specified rates. Then the subject again altered the word rate of the recording during the self-paced listening experience. Sticht's find- ings establish the fact that such double-alterings of word rate do not adversely affect comprehension as long as the amount of variance does not exceed a difference of approx- imately 40% compression of the original material. Listener Preference for Rate of Presentation Few studies have been conducted that examine lis- tener preference for rate of presentation of recorded ma- terial. A predominant contaminant to the larger percentage of those few studies that have been conducted is the depen- dence on the subject's statement of preference based on a series of presentations at pre-fixed (externally paced) rates. This section reviews all reported studies of listener rate preference that utilized external pacing. The one re- ported study that utilized self—pacing is also reviewed and discussed in Special reference to the design of the present study. Foulke (1965) sent out invitations to two hundred blind persons registered with Recording for the Blind en- couraging them to participate in a research study. Of the two hundred invited, one hundred accepted the invitation. A vinyl record was sent to each person accepting. The record included samples recorded at different rates of 31 presentation through speech compression procedures. One sample began at a normal rate and gradually increased to 350 wpm. Accompanying the record was a questionnaire de- signed to assess the listener's reaction to the samples. Fifty-one percent of the listeners returned the completed questionnaire. Of those responding, 91.7% said that they would listen to material prepared in this manner if it were available. When asked which rate they found "most satisfactory," 45% indicated 275 me, 25% indicated 225 wpm, 22.5% indicated 300 wpm, and 7.5% indicated 350 wpm. In terms of those responding, 275 wpm seems to be clearly indicated as a rate preference in this study. The generalizability of these findings, however, must consider that all respondents were blind (indicating prior experi- ence with learning by listening), all respondents utilized the services of Recording for the Blind which specializes in textbooks rather than light reading (suggesting persons who may typically listen to a larger number of technical recordings than the average blind listener), and the largest percentage of respondents were students, with college stu- dents being most numerous (suggesting a respondent group that is highly dependent on listening as a vehicle for in- formation input). Only 10.4% of the respondents indicated that they "rarely" did their reading by means of recordings. In a study conducted by Foulke and Sticht (1966), 32 one hundred sighted college students were presented with five listening segments where the speed gradually ascended and five segments where the speed gradually descended. The subjects were instructed to tell the examiner to speed up (Slow down) the rate of presentation until a preferred rate was achieved. The mean preferred word rate for the total group was 207 wpm. Males indicated a mean rate preference of 212 me and the mean preference rate for females was 204 wpm. The descending trials produced a higher mean preferred rate (217 wpm) than the ascending trials (197 wpm). No attempt was made to measure comprehension. A second part of the study failed to Show any rela- tionship between preferred listening rate and anxiety as measured by the Taylor Manifest Anxietnycale. AS compared with Foulke's earlier study (1965) the Sighted subjects seemed to prefer a rate of presentation lower than preferred by the blind subjects. This preferred rate differential might have been further exaggerated if, in- stead of using a recording with an eighth grade level in the 1966 study, a recording of college level had been selected. A college level recording would have better matched the level of the subjects in the experiment and may have altered the results. This study is the first to indicate a difference may exist between preferred listening rate and potential listen- ing rate. Though not conducted in a self—paced manner (the 33 experimenters controlled the rate according to the prefer- ences verbalized by the subjects), the subjects in this study indicated a mean preferred rate (207 wpm) that is greatly different from the potential rates substantiated in earlier studies (250 wpm - 300 me) at which compre- hension can still be maintained. This gap between preferred rate and potential rate is not seen in the earlier study by Foulke (1965) which examined blind listeners. The subjects in this earlier study indicated a preferred rate that is very similar to their potential rate (275 me was indicated as preferred by the blind listeners). The assumption is that the blind listeners, reported in the study, prefer to listen at a rate closer to their potential rate and consequently are capable of operating at a more efficient level in a lis- tening situation. In a recent study by Challis (1973), 96 college junior and senior students were randomly assigned to one of four groups: normal rate (120 wpm), 30% compression (174 wpm), 40% compression (200 wpm), and a choice of 20%, 25%, 30%, 40%, 50%, or 55% compression. All subjects received the recorded information in conjunction with filmstrips. No Significant difference was found between any of the groups on measures of achievement. Further, there was no interac- tion between achievement and amount of time Spent listening. A questionnaire completed by all subjects at the end of the experiment indicated that 97% felt that learner control over 34 the compression rate was necessary or desirable for a most satisfactory learning experience. This finding is rather unique in this experiment since only 25% of the subjects were allowed control over the rate of presentation and this control amounted to a forced selection from a small set of choices (Six). Such a finding may indicate an ability on the part of the subjects, especially the 75% who were not allowed to select the rate of presentation, to sense a poor individual match between presentation rate and preferred rate. For the 25% who were allowed to select their own rate of presentation, such a finding would seem appropriate. This study implies that a listener does have a rate preference, and that it is possible for a listener to be aware of his own rate match (whether the rate of presentation is the same as that which he prefers) in situations where he is provided a rate choice and in situations where no rate choice is allowed. Friedman, Graae, and Orr (1967) conducted a series of studies that includes the only documentation to date of a study that examines learner preferences for rate of pre- sentation of recorded information through a self-paced lis- tening environment. The series of studies aimed chiefly at the determination of the ability of college students to com- prehend and to be trained to comprehend compressed Speech. The two questions asked in the self-pacing study were: 1) At what rate will a listener choose to hear material 35 which is compressed? and 2) In what way will this affect his learning to comprehend compressed Speech? Based on evidence that suggests that for young children listening is a preferred modality and for older children and adults reading is preferred, the experimenters felt that a potential explanation related to the reader's ability to peruse material at his own rate. This would in- clude Speeding up the rate, slowing down the rate, or re- reading difficult portions a second or third time. In other words, when a person is old enough to read, reading is pre- ferred Since it accommodates the individual. It was further felt, however, that listening could be considered more in- formative than reading since it contained intonational nu- ances. If the provision of a self-paced listening environ- ment could accommodate for the comparative advantages of reading (except for the advantage of re-reading) it was hy- pothesized that self-pacing would provide a more efficient means of communicating to a subject as measured by the sub- ject's ability to answer questions accurately per unit time taken to receive the information. Additional questions that were examined in the study included: What behavior would result from listener controlled speech? How frequently would the rate be changed? What mean rate would be chosen? Twelve male college students with a mean age of 19.6 were paid $1.50 per hour for their participation in the study. A $10 bonus for the best subject was promised. The 36 listening materials consisted of seven historical passages taken from a college level textbook. They were recorded and compressed to 1.5 times their original rate of 175 wpm. The compressed rate was 262 wpm. Comprehension was meas- ured by a group of five option multiple choice questions for each passage. A baseline passage at normal speed (175 wpm) was administered to all subjects at the beginning of the experi- ment. The presentation of the remaining passages began with an externally paced compressed passage (1.5 times original), then moved to three self-paced passages, and ended with two more externally paced passages (1.5 times original). Each subject served as his own control and partici- pated individually during the experiment. For the self- paced passages subjects were provided an unmarked remote control knob that was fixed to the Speech compressor and provided control of the rate of presentation. The experi- menter recorded selected rates by noting positions of the dial indicator on the speech compressor. Timings were made with stop watches. Subjects were told that not only compre- hension would be examined, but they also Should try to lis- ten at the greatest speeds. The lowest mean rate used by any subject on the self-paced task was 1.16 times normal rate and the highest rate was 2.05 times normal. The overall mean speed used by 37 all 12 subjects on the self-paced task was 1.45 times normal rate. No consistent trend was seen from the first to the third self-paced passage in terms of rate at which they were played. (Means were 1.43, 1.45, and 1.48.) There was a tendency for the number of downward changes in rate to in- crease and upward changes to decrease as the passage went on. There was no indication of superiority in terms of compre- hension of one type of pacing over the other. In their discussion, the experimenters make note of the fact that the mean rate of each of the self-paced pas- sages was very close to the rate at which they first heard compressed Speech (1.5 times normal rate). It was felt that this was due to a modeling effect of the first passage and created a Situation whereby the subjects did not want to de- viate very far from the model. The subjects tended to in- crease the rate higher and higher during the first quarter of the self-paced passages and showed more downward rate changes in each successive quarter. The experimenters pointed out three aspects of the study that may have impeded an increase in comprehension on the self-paced passages as had been hypothesized. l) The training effect of the series of passages might have contaminated comprehension scores. If the final two passages were self-paced (rather than externally paced) higher comprehension scores might have resulted. The sub- jects' comprehension might have been negatively affected 38 due to the pre-set rate that was used for presentation for the final two passages. 2) Since no feedback was provided the subjects at the conclusion of each of the listening passages regarding their comprehension performance, subjects may have been attempting to manipulate the self-pacing to their advantage (to improve their comprehension) but in fact were doing it to their disadvantage (interfering with their comprehension). 3) The mechanics of manipulating the rate may have interfered with passage comprehension. The task of compre- hending might have become secondary to the task of manipu- lating the rate. Suggestions for further experimentation included ex- periments a) to provide more practice in self-pacing, b) to control the order of self-paced and external-paced passages, c) to test the effect of feedback on subject performance, and d) to compare active and passive subjects in a self- pacing Situation. This study by Friedman, Graae, and Orr sets the stage for further studies of self-pacing listening behavior. It is certainly easy to criticize experimenters for confound- ing the experimental conditions by mixing the types of pacing within the experimental procedure if viewed entirely from a position of self-pacing. However, the design seems reasonable if viewed in light of the eXpressed goal of in- vestigating training effects. As a study of self-pacing, 39 it falls Short; but in terms of training, it yields valuable information on the effect of modeling rates of presentation and confounding comprehension by focusing attention on man- ipulation of rate. The instrumentation used in the eXperiment is a cause of concern since all data were collected through human observation of displayed rate and elapsed times during the actual experiment. A procedure for systematically recording experimental events through a mechanical system would have provided more reliable data. Though two experimenters were used to record data, each was responsible for different sets of data and no reliability checks of the recorded data were reported. The findings indicate that it is possible for a sub- ject to respond to a set of self-paced situations with pre- ferred listening rates that are similar for all of the situ- ations. Two potential reasons for this behavior are the effect of the initial rate model and the fact that subjects were instructed to return their rate selector knob to the "normal" rate (1.5. times normal) at the beginning of each passage. A procedure to control for the potential biases created by these two constraints would be to present no model prior to the self-pacing activity and to begin each self- pacing passage at a different rate. The subject would be instructed to return the rate selector knob to a zero point prior to the beginning of each passage. This zero point 40 would be arbitrary and would serve to guarantee that each presentation to the listener would begin at a known rate that is different for each passage. It Should be assumed that a dual focus for the sub- ject (manipulation and comprehension) interferes in unknown ways with the listening task. If a subject is given the dual task of manipulating the rate of presentation and also lis- tening for comprehension, he may either subvert the compre- hension task in favor of the manipulation task, or subvert the manipulation task in favor of the comprehension task. By removing the comprehension task from the listening exper- ience, it may be possible to better focus on rate manipula- tions that accompany self-paced listening experiences. In- terference of these tasks demands that they be examined separately before proceeding with studies that look at the interaction. Recent Advances The most recent advance in the area of compressed speech was announced in January 1974 when Cambridge Research and DevelOpment Group released details of licensing agree- ments with the world's first and second largest manufacturers of tape recorders for production of tape recorders that in- cluded a Variable Speech Control (VSC). Variable Speech Control uses miniaturized circuitry smaller than a pocket-Sized cigarette lighter to electronically speed up or slow down recorded Speech without distortion. The solid-state system 41 can be adapted my manufacturers to any standard cassette recorder. A simple knob on VSC-equipped players enables the user to increase the play- back speed to more than twice its original speed without any distortion. Any pre-recorded material can be used. (Cambridge Research and Development Group, Press Release, January 1974, p. 2) Estimated costs for recorders equipped with VSC are expected to add less than $50 per unit retail to the price of standard audio equipment. Arthur Fisher, writing in the "Science Newsfront" column of Popular Science says that eventually "the price differential may come down to as little as $10 or $12." An indication of potential demand for such a unit and con- sequently the expected widespread use was provided by Fisher when he wrote: The most provocative Science Newsfront item to appear in the last few years -- judging by reader mail -- did not concern any breathtaking advance in energy generation, environmental control, in- sight into the nature of the cosmos, or anything else I would have suggested. Instead, it dealt with an electronic method of Speeding up or Slowing down Speech on ordinary cassette players, using tiny integrated—circuit chips of relatively low cost. (Fisher, Arthur, "Science Newsfront," Popular Science, Vol. 204, No. 4, April 1974, p. 30) CHAPTER III METHODS AND PROCEDURES This chapter is divided into four sections. The first section will define the population and sample that was used in the study, the second section outlines the treatment that was implemented to elicit listening behaviors of a self-paced nature, instrumentation and data collection makes up the third section and describes the unique equip- ment that was utilized for the listening environment and documentation of behaviors, and the final section describes the procedures that were used for data analysis. Population and Sample The population for the study comprised elementary school children in grades three, four, and five who displayed no hearing deficits. Children were limited to those attend- ing regular public school classes (not enrolled in special education programs). The delimitation to regular school classes was provided to reduce the likelihood of atypical listening behavior that might have been attributed to a handicapping condition. A sample of 16 subjects from each grade level 42 43 (total sample of 48 subjects) was selected for participation in the study. Selection was made on a random basis by the experimenter from enrollment lists for each grade level of a local elementary school. The elementary school was lo- cated in an established area of East Lansing, Michigan. The children came from a middle to high socio—economic back- ground with a high percentage of the parents employed in professional capacities. The selection of the sample allowed comparisons across grade levels and also permitted randomi- zation of presentation sequence across subjects. Each of the 16 subjects for each grade level was randomly assigned to one of 12 different treatments. The conditions of the assignment were the following: 1) no more than two subjects were assigned to any one treatment, and 2) all 12 treatments were assigned. Treatments One at a time, each subject in the study listened through headphones to a series of pre-recorded and rate- altered listening selections. Each subject listened to the material in the same room, located adjacent to the school library. The room, approximately eight feet by ten feet, was well lighted and contained a ventilating fan to circu- late air within the room. During the listening experience the subject was given the opportunity to alter the rate of presentation of 44 the recorded material by manipulating a Single rate control knob on a metal box. The metal box had a plain appearance and was constructed so as to provide a minimal distraction to the subject. All instructions regarding the listening experience and actions to be taken by the subject during the listening experience were delivered on the tape and thereby standardized for all subjects. A tapescript of the instruc- tions is provided in Appendix A. The experimenter, the speech compressor, and the chart recorder were situated in the same room as the subject but out of the subject's direct line of Sight. Figure 3.1 presents a schematic diagram that out- lines the order of events that were presented on the tape recording. Introductory Instructions First Listening Segment Instructions for Second Segment Second Listening Segment Instructions for Third Segment Third Listening Segment Instructions for Fourth Segment Fourth Listening Segment FIGURE 3.1 ORDER OF EVENTS PRESENTED ON RECORDING 45 The four listening segments consisted of the first, second, third, and fourth quarters of a single tape record- ing. This insured a similarity of content and style of pre- sentation for all segments. The instructions, presented prior to each segment, were recorded using a "second voice" which was different than the voice used for the story nar- ration. This served as a form of audio highlighting to cue the listener that, in contrast to the story narration, in- structions were now being presented. Or, after listening to instructions, the story narration was being presented or continued. The story used in the study was selected from the third grade volume of the reading series produced by the Houghton Mifflin Company. The particular story selected was entitled "The Train That Never Came Back" and was written by Freeman Hubbard. This selection was made based on the cri- teria that a) the story must be from a recognized text that is widely used in the school setting, yet is not familiar to the sample of subjects that participated in the study, and b) the story must use primarily a narrative format rather than a format that contains extensive conversational dia- logue. Only those stories were considered that did not ex— ceed the reading ability of the youngest subjects (third grade) as equated to listening ability according to Durrell (1969). A set of four pictures were created to illustrate 46 each of the four segments of the story. These pictures, water colors mounted in a flip book, were presented to each subject at the conclusion of the recorded introduction and directly prior to presentation of the first listening seg- ment. The flip book was constructed as an easel and was positioned directly in front and Slightly to the left of the remote control unit. Instructions on the tape told the subject when to change pictures. All subjects were observed during the introductory instructions to assess which hand would be used to manipulate the rate control knob. In one case the subject used his left hand and the flip book was then positioned in front of the control unit and to the right of it. It was felt that a totally "pure" listening experi— ence, one that is devoid of any planned visual stimuli, would be an unnatural listening environment for elementary grade students. Consequently, it could be expected that students, without a provided visual focal point, would visually search the environment to seek for visual input. The providing of the four pictures was designed to control for the effects of extraneous visual stimulation and to control the type and amount of visual stimulation for all subjects. By limiting the number of pictures to one per segment, the motor activity associated with changing pictures was restricted to the in- terval of time between listening segments and did not dis- rupt the listening environment. The single pictures were 47 also felt to provide a level of visual stimulation which would not impede the auditory task. A tape recording of the story was made by a pro- fessional announcer. This recording was then divided into four segments of approximately equal length that matched appropriate break points in the story line. Table 3.1 shows the length and original word rate of each story segment. TABLE 3.1 LENGTH AND ORIGINAL WORD RATE OF STORY SEGMENTS Segment Length (words) Original Word Rate (wpm) l 444 191.7 2 492 189.2 3 503 184.0 4 450 179.2 All four segments of the tape were rate-altered to prOV1de a consistent stimulus to each subject. To control for presentation order of the segments, a set of twelve dif- ferent stimulus tapes were recorded from the original tapes. The segments appeared sequentially in the correct order, but the sequence of rate alterations were varied. The four rate alterations that were used in the study were an expanded rate (100 wpm), a "normal" rate (150 wpm), a moderately 48 compressed rate (200 wpm), and a highly compressed rate (275 me). In selecting the configurations of sequence for the twelve different stimulus tapes, it was decided to eliminate: a) any configuration that began with the "normal" rate. This exception is based on the Friedman pp 21 finding that a modeling effect occurred due to the initial presen- tation rate of 150 wpm. To eliminate the possibility of modeling the normal rate for the subjects, none of the twelve stimulus tapes began at the normal (150 wpm) rate. b) any configuration that began with the highly com- pressed rate. Since none of the subjects had prior experi— ence with compressed Speech it was assumed that a configura- tion that began with the highly compressed rate could "over- power" the subject and affect the subject's rate alterations in successive segments. All configurations of sequence, therefore, began with either the expanded rate (100 wpm) or the moderately com- pressed rate (200 wpm). The twelve configurations, stimulus tapes, comprised all combinations of the four different rates with the exception of those that began with the normal rate (150 wpm) or the highly compressed rate (275 wpm). Table 3.2 Shows the sequence of presentation rates for each of the stimulus tapes. 49 TABLE 3.2 CONFIGURATIONS OF STIMULUS TAPES Stimulus Segment Segment Segment Segment Tape # One Two Three Four 1 M E N H 2 M N E H 3 M H N E 4 M N H E 5 M E H N 6 M H E N 7 E N H M 8 E N M H 9 E H N M 10 E H M N 11 E M N H 12 E M H N E = Expanded Rate M = Moderately Compressed (100 wpm) Rate (200 wpm) N = Normal Rate H = Highly Compressed (150 wpm) Rate (275 wpm) 50 Each segment of each stimulus tape was pre-altered to the four selected initial presentation rates. Duration and rate for each segment, in the pre-altered form, are shown in Table 3.3. TABLE 3.3 DURATION AND LENGTH OF PRE-ALTERED SEGMENTS Segment Length Duration (seconds) # (words) 100 wpm 150 wpm 200 me 27S wpm 1 444 266 178 133 97 2 492 295 197 148 107 3 503 302 201 151 110 4 450 270 180 135 98 The duration of the actual listening experience was different for each subject since duration was dependent upon the rate manipulations initiated by each individual subject. The total listening experience, including the pre-recorded instructions, lasted no longer than twenty minutes for any subject. Instrumentation and Data Collection The instrumentation for the study consisted of three separate pieces of equipment. The subject was only aware of one of the pieces of equipment (the remote control 51 unit). The other two pieces of equipment (the Speech com- pressor and the chart recorder) were placed out of view of the subject. The only other instrumentation used was a single set of high quality headphones through which the subject listened to the stimulus tape and the picture flip book to accompany the tape. The remote control unit was a metal box that pro- vided the following functions at a location away from the speech compressor and chart recorder: a) rate control knob, b) volume control to adjust volume to headphones, c) phone jack for connecting headphones, and d) power switch for ac- tivating the system and begin presentation of the stimulus tape. The rate control knob included a "zero point" indi- cated by an arrow on the knob and a corresponding arrow on the body of the case. This "zero point" established the beginning rate that the subject used to start each segment of the story. Instructions to return the knob to this "zero point" were provided in the instructions directly prior to each segment of the stimulus tape. The remote control unit was placed on a desk of the appropriate height for the subjects. No other materials, other than the picture flip book, were located on the table, thereby minimizing extraneous stimuli. The speech compressor used in the study was the Varispeech I produced by Lexicon, Inc. This compressor unit is capable of both expansion and compression of recordings 52 through the manipulation of a single control. With the con- trol set to normal, rotation in a counter-clockwise direc- tion expands the recording (Slows down the rate) and ro- tation in a clockwise direction compresses the recording (increases the rate). A closed circuit jack assembly in- stalled on the Speech compressor allowed direct connection with the remote control unit and permitted the remote con- trol unit to take over the rate-altering control functions of the speech compressor. The chart recorder used in the study was a Mingraph produced by Esterline Angus. The unit was calibrated to accept dc milliamperes. A regulated power supply, control- led by an extra wafer on the rate control knob of the remote control unit, supplied power to the chart recorder of dif- ferent voltages based on the setting of the rate control knob. The chart paper, moving at a constant speed of 30 inches per hour, was used to document all settings of the rate control knob per unit of time. Directly following each subject, the chart paper that was expended for the subject was marked to identify the particular subject whose re- sponses were displayed. Data Analysis Each subject in the study was provided the listening eXperience individually in a Single session, and served as his own control. Data were examined and analyzed in a number 53 of different manners to yield information regarding manifest preference for rate and rate manipulation behaviors that occur in a self-paced listening situation. Criteria were established for this study to define manifest preference for rate and mean manifest rate prefer— ence. Two different bases were used to establish the cri— teria. First, they were based on the limited amount of available information on rate preference and manipulation behaviors. Secondly, the criteria established a structure that would allow non-manipulation of rate for no more than one of the segments for acceptance as demonstrated manifest preference for rate. Data were also analyzed by altering the criteria to further examine acceptable criteria for analysis of self-paced listening behaviors. The following criteria were established to define manifest preference for rate. Criterion #1 The final rate for all four segments of a subject's listening experience will fall within a band of 40 wpm. This criterion was established to guarantee rate alteration of at least three of the four listening segments. Since the rate differences between the four segments (50 wpm, 50 wpm, 75 wpm) were all greater than 40 wpm, a subject would have to alter at least three to yield a final band width of 40 wpm or less. This criterion also accounted for the possibility that the rate of one of the segments might 54 coincide with the subject's manifest preference for rate. In such an instance, the criterion would accept a single non-manipulated segment, but would demand that the other three segments be manipulated. However, the criterion would not accept a subject as showing a manifest prefer- ence for rate who did not manipulate two or more of the segments. The use of a 40 wpm band was further supported by the Friedman g: a; study that reported mean rates for three self-paced listening passages of 250 wpm, 254 wpm, and 259 wpm with corresponding standard deviations of 24.5, 31.5, and 42. Based on these findings, 68% of their popu- lation (college students) showed preference fluctuations of 49 wpm, 63 wpm, and 84 wpm. The greatest difference between these preference fluctuations (84 wpm - 49 me) was 35 wpm. It could be expected that 68% of their population showed manifest preference for rate within a band width of 35 wpm. When 95% of the population is considered, the band width increases to 70 me. Figure 3.2 Shows examples of terminal rates that are acceptable and non-acceptable as demonstrating prefer- ence . 55 Acceptable Non-Acceptable IIII I I I L J \___._.___.__J 40 wpm 40 wpm FIGURE 3.2 ACCEPTABLE AND NON-ACCEPTABLE DEMONSTRATION OF PREFERENCE Criterion #2 When the terminal rate for one segment is more than 40 wpm away from the band of rates delimited by the other three segments, the rates of these three segments not to exceed a band width of 40 wpm, the be excluded and the remaining three rates the manifest preference for rate. This criterion was established to occurrence of a Single terminal rate that from the band of the other three terminal fourth rate will will be considered provide for the deviated greatly rates. Such an occurrence was considered atypical of manifest preference for rate and the deviant terminal rate was discarded as being incongruent. Figure 3.3 Shows a hypothetical example of a terminal rate that is excluded according to Criterion #2 and two terminal rates that can not be to Criterion #2. excluded according 56 Three rates accepted as demon- A B C D strating manifest preference for rate ("D" is more than 40 wpm I I I I away from the band of the other L——-:’ J three rates, and the band of the 40 75 other three rates is less than 40 wpm) A B C D No demonstration of rate prefer- ence ("D" is less than 40 wpm away from the band of the other I I I I three rates, though the band of L-jfiT-TJ , the other three rates is less 35 than 40 me) A B No demonstration of rate prefer- ence ("A" is more than 40 wpm I I I I away from the band of the other 50 , , three rates, but the band of the other three rates exceeds 40 wpm) FIGURE 3.3 ANALYZING DEMONSTRATED RATE PREFERENCE ON THE BASIS OF CRITERION #2 WHEN CRITERION #1 HAS NOT BEEN MET The following criteria were established to define mean manifest preference rate. Criterion #3 A mean will be calculated for only those subjects who demonstrate a manifest preference for rate. Criterion #4 Where the standard deviation of the four terminal rates is greater than 15 and the elimination of the terminal rate for a Single segment reduces the stan- dard deviation to below 9, this single rate will not be used to calculate the mean and the mean will be calculated by 57 using the other three terminal rates. This criterion was established to provide for the Situation where three of the terminal rates are extremely similar and a single rate is divergent. In such an in- stance, it was felt that the mean manifest preference rate was more typical of the mean of the three similar rates than of all four rates. The defining of the actual limits for Criterion #4 was established through the development of a series of hy— pothetical situations where terminal rates were examined in relation to their standard deviation. In particular, this criterion allowed for the alteration of extremely skewed distributions, but did not allow for alteration of bimodel distributions, or distributions of limited skewness. Figure 3.4 illustrates the use of Criterion #4 to discard a single skewed terminal rate from use in figuring the mean manifest preference rate. Figure 3.5 illustrates two instances (limited skew- ness and bimodal distribution) where the criterion does not allow elimination of a single rate. 58 Segment Terminal Rate A 100 wpm B 100 wpm C 110 wpm D 140 wpm R = 112.5 me s = 16.39 Segment Terminal Rate A 100 me Rates Used For Calculating Mean B 100 wpm Manifest Preference Rate C 110 me i = 103.3 wpm s = 4.7 (s<9) FIGURE 3.4 ELIMINATION OF A SINGLE SKEWED TERMINAL RATE IN CALCULATING THE MEAN MANIFEST PREFERENCE RATE 59 LIMITED SKEWNESS BIMODAL DISTRIBUTION Segment Terminal Rate A 100 me B 100 me C 120 wpm D 140 wpm R = 115 wpm S = 16.58 Segment Terminal Rate A 100 wpm B 100 wpm C 120 wpm — 106.7 wpm >< II II 9.43 (S‘>9) Segment Terminal Rate A 100 wpm B 100 wpm C 135 wpm D 140 wpm R = 118.75 me s = 18.83 Segment Terminal Rate A 100 wpm B 100 wpm C 135 wpm ‘ 111.7 wpm >4 II II 16.5 (s > 9) Rates Used For Calculating Mean Manifest Preference Rate Rates Used For Calculating Mean Manifest Preference Rate FIGURE 3.5 INSTANCES WHEN MEAN MANIFEST PREFERENCE RATE MUST BE CALCULATED FROM ALL TERMINAL RATES 60 Criterion #5 A mean will be calculated by using no less than three terminal rates. This final criterion was established to guarantee at least a 75% sample of a subject's self-paced listening behavior for analysis. Data analysis was also conducted using all four rates without consideration for any of the above criteria. In these instances, "point of convergence" is used to iden- tify the arithmetic mean of all four final selected rates. Primary data analysis was carried out through a comparison of means and standard deviations for different sets of data. Pearson product moment correlations were com- puted on the data to yield information regarding the exis- tence of relationships for the different rate manipulation behaviors that were investigated. A Significance level of .05 was used to test the significance of relationships and the difference between means. Additional analyses were carried out to substan- tiate the use of the defined experimental procedures for the examination of listener rate preference behaviors. CHAPTER IV FINDINGS The data collected in the study are presented in this chapter and analyzed according to the procedures de- scribed in Chapter III. The data and analyses are organized in eight different sections with each section dealing with a separate aspect of the study. Each of the hypotheses pre- sented for study in Chapter I is examined along with other findings that were uncovered in the course of analyzing the data though not formally suggested for examination. These additional examinations were conducted in an attempt to thoroughly and systematically examine all viable proce- dures for the study of self-paced listening behaviors. Complete data for all subjects are displayed in Appendix B. Manifest Preference for Rate of Presentation The primary focus of this study was on the examina- tion of rate preference and whether a subject, when given the opportunity to self-pace the listening task, would demon- strate such a preference. The first hypothesis states that a manifest preference for rate of presentation of recorded 61 62 information will be demonstrated when the listener is given autonomous control over the rate of presentation in a self- paced listening situation. This hypothesis was examined through a series of analyses that were designed to test the hypothesis and also provide detailed information regarding the behaviors of the subjects regarding convergence toward a Single point (point of convergence). Convergence, a band width of final selected rates that was narrower than the band width of initial pre- sentation rates, was shown by all subjects in the study. The convergence behavior and non-convergence behavior demonstrated by the subjects is Shown in Table 4.1. The point of convergence is the arithmetic mean of the subject's final selected rates. TABLE 4.1 CONVERGENCE BEHAVIOR BY NUMBER OF SEGMENTS Movement Movement No Toward Away From Change Convergence Convergence Number of 158 22 12 Segments (82.3%) (11.5%) (6.25%) Total number of segments = 192 (48 subjects x 4 segments) As indicated in Table 4.1, 158 (82.3%) of the 192 listening segments experienced by subjects showed movement 63 toward a point of convergence. Only 22 (11.5%) of the seg- ments Showed subject demonstration of movement away from a point of convergence. On twelve occasions (6.25%), sub- jects showed no change from initial presentation rate to final selected rate. These non-manipulated segments will be discussed further at a later point in this chapter. Table 4.2 presents convergence data by subject according to the number of subjects who showed convergence on all four segments (100% of segments Show convergence) and the number of subjects who showed convergence on only three segments (75% of segments Show convergence). TABLE 4.2 MOVEMENT TOWARD CONVERGENCE BY NUMBER OF SUBJECTS Convergence Convergence Convergence Shown By Shown By Shown By Four Only Three Three or Four Segments Segments Segments Number of Subjects 19 26 45 As indicated by Table 4.2, 45 subjects in the study (94%) Showed convergence on either three or four of the four listening segments that each subject experienced. Of this total, 26 subjects Showed convergence on all four of the listening segments. The extent of movement toward the point of 64 convergence for each subject by listening segment was also examined. For those subjects that demonstrated convergence on at least three of the four listening segments, the ex- tent of movement was calculated and ordered in relation to the extent of the difference between the initial presenta- tion rate and the rate at the point of convergence. The data were examined in relation to the extent of movement of the segment that differed the most from the convergence point and the average extent of movement of the other three segments. Further examination was made between the extent of movement of the segment that differed the most from the point of convergence and the extent of movement of the seg- ment that differed the second most from the point of con- vergence. Both of these analyses were made to establish the strength of the extent of movement of the segment that differed the most from the point of convergence. The second of these analyses was computed to establish the potential presence of a linear relationship between the extent of move- ment and the difference between the initial presentation rate and point of convergence. Table 4.3 displays these data. As is indicated by Table 4.3, nine of the 26 subjects who had only three movements toward the point of convergence had one non-manipulated segment and three segments that showed movement toward the point of convergence. The re- maining 17 subjects of those who had only three convergent 65 TABLE 4.3 EXTENT OF MOVEMENT TOWARD POINT OF CONVERGENCE FOR ALL LISTENING SEGMENTS Subject MV MV Mv Mv Mv +Mv 1 2+Mv 3 3 Subjects with three movements toward point of convergence and one non-manipulated segment 8 0 36 80 101 38.67 *+ 10 O 64 97 92 53.67 * 15 0 2 27 98 9.67 *+ 22 0 60 67 109 42.33 *+ 38 0 89 82 86 57.00 *+ 39 0 54 120 122 33.00 *+ 45 0 61 81 115 47.33 *+ 50 5 0 84 100 29.67 *+ 51 4 36 0 91 13.33 *+ Subjects with three movements toward point of convergence and one movement away from point of convergence 2 -l 74 67 99 46.67 *+ 4 —21 39 30 80 16.00 *+ 9 -11 29 83 90 33.67 *+ 11 —21 28 63 105 23.33 *+ 17 -12 39 60 81 29.00 *+ 18 SO -3 4 164 17.00 *+ 20 -45 68 116 140 46.33 *+ 24 —44 70 128 155 51.33 *+ 27 -26 68 66 72 36.00 *+ 31 -13 62 46 71 31.67 *+ 32 -12 53 71 104 37.33 *+ 33 -ll 58 102 76 49.67 * 35 -31 43 57 79 23.00 *+ 42 -8 90 125 154 69.00 *+ 44 -12 30 81 144 41.00 *+ 49 -102 53 69 75 6.67 *+ 52 -20 42 65 72 29.00 *+ 66 TABLE 4.3 (cont'd.) Subject Mv Mv Mv Mv Mv +Mv +Mv # 1 2 3 4 l 2 3 3 Subjects with all four movements toward point of convergence l 12 13 79 130 34.67 *+ 3 30 47 62 95.5 46.33 *+ 5 15 24 69 99 36.00 *+ 7 2 70 81 108 51.00 *+ 12 24 62 94 88 60.00 * 14 38 45 102 114 61.67 *+ 16 30 27 82 181 46.33 *+ 19 42 34 87 104 54.33 *+ 21 3 37 86 101 42.00 *+ 23 6 63 94 93 54.33 * 25 29 18 99 91 48.67 * 28 20 42 81 103 47.67 *+ 29 18 65 74 117 52.33 *+ 30 18 43 62 124 41.00 *+ 34 22 52 98 105 57.33 *+ 37 23 26 70 99 39.67 *+ 43 28 46 71 90 48.33 *+ 47 21 36 59 166 38.67 *+ 53 16 39 7O 91 41.67 *+ Subjects with two movements toward point of convergence 6 110 116 129 112 46 67 0 0 149 48 97 0 62 99 le = Extent of movement for segment with least difference between initial rate and rate at point of convergence. Mv2 = Extent of movement for segment with second least difference between initial rate and rate at point of convergence. Mv3 = Extent of movement for segment with third least difference between initial rate and rate at point of convergence. Mv4 = Extent of movement for segment with greatest difference between initial rate and rate at point of convergence. + + += Mv4>’lebw3 Mw3 Mv4:>Mv3 3 67 movements had a single segment where movement toward conver- gence was not evident. Of these 17 subjects, in all but one instance the non-convergent movement was the smallest of the four movements. Further, in all but one case the non-con— vergent movement occurred in the listening segment that had an initial presentation rate that was the closest of the four segments to the point of convergence. This finding suggests that the non-convergent movements are relatively small, as compared with a subject's convergent movements, and non- convergent movements occur for those segments that have an initial presentation rate that is very close to the point of convergence. When the extent of movement for the three segments that have the smallest difference between initial presenta- tion rate and the point of convergence are averaged and com- pared to the extent of movement of the segment that has the greatest difference between initial presentation rate and the point of convergence, 3;; of the 45 subjects examined, those that showed convergence on three or four segments, showed greater extent of movement for the segment that varies the most between initial presentation rate and point of conver- gence. This finding suggests that the greatest manipulation of rate by a subject in a self-paced listening situation will occur for those listening experiences that differ the most from a subject's point of convergence. In the study, 93.75% of the subjects demonstrated this behavior. 68 When the extent of movement of the segment that has the greatest difference between initial presentation rate and point of convergence is compared to the segment with the next largest difference between initial presentation rate and the point of convergence, the data Show that the segment with the greatest difference between initial presen- tation rate and the point of convergence has the greater ex- tent of movement in 40 of 45 cases. When these 40 cases are compared to the total group of 48 subjects, it is seen that 83.3% of the subjects demonstrate this behavior. This find- ing suggests that a linear relationship may exist for a sub- stantial group of the subjects when the difference between initial rate of presentation and the point of convergence is compared to the extent of movement. An examination of this relationship is shown in Table 4.4. As is indicated by Table 4.4, 28 of the 48 subjects (58.3%) Show a correlation that is greater than chance when the difference between the initial presentation rate and the point of convergence for each of the subject's four listening segments is correlated with the subject's extent of movement. This finding suggests that 58 persons in 100 would be more inclined to demonstrate manifest preference for rate if the initial presentation rate for a listening experience is rela- tively far from the listener's preferred rate of presentation of recorded information. The first hypothesis regarding manifest preference TABLE 4.4 69 CORRELATIONS BETWEEN THE EXTENT OF MOVEMENT TOWARD POINT OF CONVERGENCE AND THE DISTANCE BETWEEN INITIAL PRESENTATION RATE AND POINT OF CONVERGENCE Subject Correlation # Coefficient 20 .9111 19 .9411 53 .9836* 48 .3819 11 .9501* 31 .9094 39 .8751 44 .9950* 45 .9981* 43 .9560* 42 .9801* 37 .9999* 34 .9967* 33 .9131 32 .9952* 30 .9935* 29 .9955* 28 .9826* 25 .9462 23 .9717* 22 .9981* 21 .9968* 17 .9981* 16 .9974* *significant at the .05 level (R >.95, P <.05) Subject Correlation # Coefficient 14 .9447 12 .9738* 10 .9599* 9 .9978* 8 .9608* 7 .9845* 5 .9955* 3 .9813* 2 .8922 l .9936* 4 .9096 6 .1622 15 .8136 24 .9459 27 .8495 35 .9179 38 .9022 46 .8046 47 .9630* 49 .7020 50 .9899* 51 .8009 52 .9762* 18 .7684 70 for rate of presentation was further examined through the use of Criterion #1. This criterion defined manifest preference for rate for this study as the total band width for the final selected rates of a subject's four listening segments to be no more than 40 wpm in width. Through this criterion, only those subjects with less than a 40 me spread between the highest final selected rate and lowest final selected rate were accepted for further analysis as having a defined manifest preference for rate. Further, Criterion #2 allowed for the elimination of a single final selected rate of one of the listening segments when that final rate exceeded a 40 me distance from the grouping of the other three final selected rates; this grouping of the remaining three rates not to ex- ceed a spread of 40 me between the highest and the lowest rates. Table 4.5 presents the findings of the study when Criteria #1 and #2 are applied to the data. The data Show that the percentage of listeners demon- strating a manifest preference for rate (68.75%) is greater than the percentage of listeners not demonstrating a mani- fest preference for rate (31.25%). The data further supports the hypothesis that a listener, when given autonomous control over the rate of presentation of recorded information in a self-paced listening situation will demonstrate a manifest preference for rate of presentation. 71 'PABIJB 4.5 DEMONSTRATED MANIFEST PREFERENCE FOR RATE # of Subjects % of Subjects Rate No Rate Rate No Rate Preference Preference Preference Preference Demonstrated Demonstrated Demonstrated Demonstrated All 33 15 68.75 31.25 Grades Thlrd 12 4 75.0 25.0 Grade F°urth 10 6 62.5 37.5 Grade Fifth 11 5 68.75 31.25 Grade Table 4.6 shows the quantification of subjects accor- ding to the acceptance criteria that were used. Of the 33 sub- jects that demonstrated manifest preference for rate, 27 sub- jects (82%) were accepted solely on the basis of Criterion #1. An additional Six subjects (18%) were added to this group on the basis of Criterion #2. subjects that comprised the study, In terms of the total group of 48 56.25% were accepted as Showing manifest preference for rate on the basis of Criterion #1 and an additional 12.5% were added on the basis of Criterion #2. 72 TABLE 4.6 ACCEPTANCE OF MANIFEST PREFERENCE FOR RATE ACCORDING TO CRITERIA #1 AND #2 Subjects Additional Subjects Total Demonstrating Rate Demonstrating Rate Subjects Preference According Preference According to Criterion #1 to Criterion #2 48 27 6 Mean Manifest Preference Rate The question of manifest preference for rate was further examined in terms of the mean manifest preference rate of the subjects. Only those subjects who demonstrated a mani- fest preference for rate according to Criteria #1 and Criteria #2 were used in calculating mean manifest preference rates. Of the 33 subjects who demonstrated a manifest preference for rate, a mean manifest preference rate was calculated on the final selected rate of all four segments for 25 subjects. Six sub- jects had mean manifest preference rates calculated using three final selected rates based on the exclusion of a single rate that was eliminated by the use of Criterion #2. An additional two subjects had mean manifest preference rates calculated using three final selected rates based on the exclusion of a Single rate due to Criterion #4. Criterion #4 provided for cases where a single final selected rate was found to inordin- ately influence the standard deviation of the mean. The mean manifest preference rate by subject with indications of the implementation of Criteria #2 and #4 are shown in Table 4.7. 73 TABLE 4.7 MEAN MANIFEST PREFERENCE RATE BY SUBJECT Mean Manifest # Of Segments Used Subject # Grade Preference Rate To Calculate Mean X 42 3 275.5 4 18 4 264 3 (a) 20 4 261 3 (b) 19 4 240 3 (b) 16 4 235.25 4 44 3 235 3 (a) 14 4 233.5 4 30 5 221.5 4 31 5 218 3 (a) 29 5 217.25 4 43 3 217 4 3 4 216.125 4 39 3 214.3 3 (a) 22 3 206.75 4 7 4 206 4 32 5 205.75 4 45 3 205 4 48 5 203.7 3 (a) 17 4 199.25 4 10 3 196 4 23 3 195.25 4 21 4 193.5 4 9 3 193 4 33 5 193 4 28 5 192.25 4 12 3 189.25 4 34 5 187 4 74 TABLE 4.7 (cont'd.) Mean Manifest # Of Segments Used Subject # Grade Preference Rate To Calculate Mean X 25 3 180.5 4 5 4 176 4 37 5 173.24 4 11 5 170.3 3 (a) 53 5 170.25 4 8 3 167 4 (a) (b) concept Final selected rate deleted on the basis of Criterion #2 Final selected rate deleted on the basis of Criterion #4 207.62, S.D. = 26.87 As indicated by Table 4.7, the data support the that preferred listening rate is an individual skill with considerable variance between subjects. For third, fourth and fifth grade students as a group, the individual mean manifest preference rate of students varies from a low of 167 me to a high of 275.5 wpm. The mean rate preference for each subject was grouped according to grade level and also examined as a total group. The total group mean and the means by grade level along with corresponding standard deviations are presented in Table 4. 8. 75 TABLE 4.8 MEAN MANIFEST PREFERENCE RATE FOR ALL GRADES COMBINED AND BY INDIVIDUAL GRADE A11 Third Fourth Fifth Grades Grade Grade Grade N 33 12 10 11 i (wpm) 207.62 206.2125 222.463* 195.658* S.D. 26.87 26.92 27.66 18.30 *Significantly different at the .05 level (p< .05) As Table 4.8 indicates, the mean manifest preference rate of the total group was 207.62 wpm. The mean manifest preference rate for the fourth grade subjects is Signifi- cantly higher than the mean manifest preference rate for the fifth grade group. The standard deviation of the total group was 26.87 and the groupings by grade level showed standard deviations of 26.92 for the third grade, 27.66 for the fourth grade, and 18.30 for the fifth grade. The high standard deviations that were computed provide further support that preferred listening rate is a highly variable attribute and that there is considerable variance among third, fourth, and fifth grade students regarding mean mani- fest preference rate. Position of the Longest Non-Manipulated Duration An analysis was made of periods within each sub- ject's listening experience where the rate was maintained 76 without manipulation for a period of time. This analysis was made to examine whether the listening segments were of ample duration to allow the demonstration of manifest pref- erance for rate. For each listening segment, the longest period of non-manipulation was identified in terms of the position of its occurrence. These non-manipulated durations were coded as occurring at either the beginning of the seg- ment, the middle of the segment, or the end of the segment. These data were grouped according to presentation order of the segments. The positions of non-manipulated durations for the first segments were grouped together, positions for the second segments were grouped together, positions for the third segments were grouped together, and positions for the fourth segments were grouped together. A total of eight segments were coded as having the longest non-manipulation duration at the end of the listening segment. Table 4.9 displays these groupings according to presentation order of the segments. AS Table 4.9 indicates, over 60% of the subjects had the longest non-manipulated duration at the end of the listening segment. This is true for all four listening segments. Table 4.9 also indicates that the smallest per- centage of subjects had the longest non-manipulated duration at the beginning of the segment and a moderate percentage of the subjects had the longest non-manipulated duration during the middle of the segment. This finding, though supporting 77 mm.em 5H.m~ m~.m Hm I m «a u m m n m v.om m.mm m~.m mm ma m bassoon nunsom mm.am m.m~ m.mH am as m ucmsmmm pawns hm.mo mm.am mo.m mm ma H ucwfimmm pcooom nm.mm ha.mm na.v mm ea N ucofimom umufim 6am macaw: mcflccflmmm 6cm maneflz maflaefimmm cowumnao coumaomwcochoz pmmmcoq cowumusa poumasmwoozlcoz umomooq “mono mo coauflmom on mewpuooom mo cofluwnom ou mcwpuooom cowumucomoum muomnnsm to a muomAESm no * m.¢ manda mfizmzwmm m0 mmnmo ZOHBdBmemmm OB ZOHBdAmm ZH ZOHBdMDQ QHBflADQHZflSIZOZ BmflUZOA mo ZOHBHmom 78 the idea that most of the subjects were able to come to rest at the end of the segment, therefore suggesting an ample lis- tening opportunity, is not considered substantial in terms of a true indication of appropriate segment length. If the segment were of appropriate length, it could be expected that a high percentage of subjects would have had the longest non-manipulated duration at the end of the listening segment. The data show that 64.58% of the subjects had the longest non-manipulated duration at the end of the segment. It could also be expected that the percentage of subjects with the longest non-manipulated duration at the end of the seg- ment would increase from segment to segment with the highest percentage of subjects showing the non-manipulated duration at the end of the segment for the fourth listening segment. The findings suggest that an opposite movement occurred in the study. Non-Manipulated Segments Of the 33 subjects who demonstrated a manifest pref- erence for rate, six different subjects did not manipulate the rate of presentation at any time during one of the lis- tening segments. This information is shown in Table 4.10. Of the six non—manipulated segments, five (83.3%) were at the moderately compressed rate (200 wpm), and one was at the normal rate (150 wpm). The largest difference between a subject's mean manifest preference rate and the rate of presentation (difference from preference) of the 79 TABLE 4.10 DIFFERENCE FROM PREFERENCE FOR NON-MANIPULATED SEGMENTS Non-Manipulated Difference Subject # Segment from Preference 8 N (150 me) 17 wpm 48 M (200 wpm) 3.7 wpm 10 M (200 wpm) 4 wpm 45 M (200 wpm) 5 wpm 22 M (200 wpm) 6.75 wpm 39 M (200 wpm) 14.3 wpm 2 (difference from preference) = 8.46 me S.D. = 5.24 non-manipulated segments was only 17 wpm. The average dif- ference from preference for all six subjects was 8.46 wpm. This finding indicates that those segments that were non- manipulated were extremely close to the listener's manifest preference for rate. Altering the Rate Preference Acceptance Band The concept of manifest preference for rate was further examined in this study by altering the acceptance band width that was used for indicating manifest preference for rate. This ex post facto analysis was conducted to assess the viability of using 40 me as a defined band width of final selected rates as an indicator of manifest 80 preference for rate. Two other acceptance band widths, a stringent band width of 20 wpm and a lenient band width of 60 me were imposed on the data. Table 4.11 displays the band width of final selected rates for all subjects in the study with brackets indicating those subjects who demon- strated manifest preference for rate at the 40 me band width, the 20 wpm band width (stringent), and the 60 wpm band width (lenient). As indicated by Table 4.11, when the acceptance level is reduced to a stringent band width of 20 wpm, a total of 14 subjects demonstrate manifest preference for rate. When the acceptance level is increased to a lenient band width of 60 me, the number of subjects with demon- strated manifest preference for rate increases to 43. Table 4.11 also graphically displays the divergence of band widths of final selected rates for individual sub- jects. There is greater variance across cases than there is within cases. Rather than any central tendency, Table 4.11 Shows a skewness indicating manifest preference for rate that moves above the band of word rates used for the orig- inal recording. The data are further broken down and pre- sented in Table 4.12 according to the total group and by grade levels. Table 4.12 indicates that 29.17% of the total group of subjects demonstrate manifest preference for rate when the acceptance band is reduced to a stringent band width of 831 TABLE 4.11 BANDWIDTH OF FINAL SELECTED RATES BY SUBJECT NUMBER ________ ‘— -I9 1 37-- I -_-32 I —22 I _48 I _44 I 5— I _3I I —]6 I _45 I —'—'7 I 8— I 23 I 9 l 34 I I7 I I I? I 42 I IO I 33 I I4 I --- 43 ''''''' J l f 50 t I : 27 I I 24 l I 52 I I 418‘“ I 35 l I x16 I I I ; 2 I l ~e~ 33 I | 47 __ _JI I 5] _. ' 4 l 1 <9 ' ‘5 “XQ I 46 I0 l 19 g» l 6 L 1 : \--- o C) o o o o o 0 <0 <3 0 o 0 <3 0 o C) o o M V L") \D (\ CD 0‘ O I-w N M V tn \0 I\ oo 0‘ C) H H o-( H r-t F! H r—i N N N N N N N N N N M ('3 RATE (words per minute) m = stringent acceptance of ___-___ ___ . _ preference (20 wpm band) defined non acceptance of preference nauumu~n~n~-= defined acceptance of Preference ‘40 ”pm band) = band width of original ———————— = lenient acceptance of word rates used for preference (60 wpm band) stimulus tapes 82 mn.mm m~.Hm m.hm mm.mm ma ma «a me Aucmwcqu an: on me.wm m.~m o.me mh.mm Ha OH NH mm an; ow m>.mv mN.Hm m.ma ha.m~ h m N «a Auamucwuumv ems om momma momma momma mmwmno mpmuo momuw momma mwcmuo numwm suuaom pawns Haé :HMflm nuuaom chase Had nuowz camm mocoummmum mcfl3onm mocmummmum mcwzonm mocmummood muomflnsm mo w muoonnsm no * mnm>mn NUZ¢Emm004 BZHMHthD Bfi madm mom mozmmmmmmm EmmmHZ¢z DmB4MBmZQ£mD NH.v mqmde 83 20 wpm. With a lenient band width of 60 wpm the percentage of subjects demonstrating manifest preference for rate in- creases to 89.58%. The data further indicate that the fifth grade group is affected the least by the alteration of accep- tance band width with a higher percentage of fifth grade sub- jects remaining in the acceptance group at both the 20 wpm band width and the 60 wpm band width. This finding sug- gests that the fifth grade group demonstrated less variance among subjects for mean manifest preference rate, as indi- cated earlier in Table 4.8 where the fifth grade group showed the lowest variance among groups, and that generally the fifth grade subjects demonstrated manifest preference for rate for the total listening experience, all four segments, that were more closely configured. Though the fourth grade subjects showed a higher mean manifest preference rate, the fifth grade subjects showed a more defined mean manifest preference rate. Rate Manipulation Behavior The second, third and fourth hypotheses for this study were established to examine possible relationships between a subject's rate manipulation behavior and the dif- ference from preference. The three rate manipulation be- haviors examined were a) the time that elapsed before a sub- ject first altered the rate of presentation (previously de- fined as "manipulation onset"), b) the time that elapsed 84 prior to when a subject last altered the rate of presenta- tion (previously defined as "manipulation termination"), and c) the time that elapsed between the subject's first alter- ation of the rate of presentation and the subject's last alteration of the rate of presentation (previously defined as "manipulation duration"). Since the examination of rate manipulation behavior was dependent upon a subject's mean manifest preference rate as one quantity in the establishment of the difference from preference, only those subjects that demonstrated a manifest preference for rate were utilized in the examination of rate manipulation behavior. All of the rate manipulation relationships were ex- amined through the use of Pearson product moment correla- tions. It was expected that negative correlations would sub- stantiate that the greater the difference from preference the sooner the manipulation onset would occur. The direc- tion of the correlations for manipulation termination and manipulation duration were not projected in the design of the study. The relationships were tested for significance at the .05 level of probability. Though a series of listening segments were needed to examine manifest preference for rate, the use of more than a single listening segment for each subject compounded the analysis of rate manipulation behavior. Since each subject in the study yielded four separate sets of scores, one for each listening segment, it was mandatory that the data be 85 blocked in four separate groupings to compensate for any statistical effect that may be caused by pooling all scores of all subjects and thereby counting a subject's four scores as four different subjects. It could be assumed that a re— lationship would exist between the rate manipulation be- haviors of the four separate segments that a single subject experienced. As such, the blocking procedure effectively turned the analysis of rate manipulation relationships into a series of studies with correlation coefficients derived for each blocking group. To compensate for any effect the blocking may have had on the analysis of the data, two different blocking pro- cedures were used. First, the scores of the listening seg- ments were blocked according to the initial word rate of the segment. All scores for segments beginning with the same initial word rate, regardless of their presentation order within the total listening experience, were analyzed as a group. Next, the scores of the listening segments were blocked according to the presentation order of the segments. All scores for segments in the same presentation position, regardless of initial word rate, were analyzed as a group. In each of the blocking procedures, a single subject was represented no more than one time in the computation of the correlation coefficient. A total of eight different cor- relation coefficients were computed for each rate manipula- tion behavior due to the use of the two different blocking 86 procedures. Table 4.13 presents the correlation coeffic- ients for the eight blocked groups when the relationship between difference from preference and manipulation onset is examined. An examination of Table 4.13 indicates that seven of the eight relationships were negative in direction. This is the hypothesized direction. Only two of the relationships were significant at less than the .05 level of significance. Of the two significant relationships, one was positive in direction and the other was negative in direction. When the subjects were blocked by presentation order of the segments the strongest relationship occurred for the first segments with the relationships getting progressively smaller for each successive blocking of segments. Little consistency was shown between correlation coefficients for the different blocked groups which suggests that there is little relation- ship between manipulation onset and difference from pref- erence as manifest by the subjects. Based on this finding, hypothesis #2 is rejected. Table 4.14 presents the correlation coefficients for the eight blocked groups when the relationship between dif- ference from preference and manipulation termination is examined. An examination of Table 4.14 indicates that seven of the eight relationships were negative in direction. None of the relationships tested as significant at less than the 87 TABLE 4.13 COMPARING MANIPULATION ONSET TO THE DIFFERENCE FROM PREFERENCE Manipulation Blocking Number Onset Correlation Group of Cases (seconds) Coefficient 100 wpm 31 i = 2.145 r = -.049 Segments S.D. = .93 150 me 30 i = 14.2 r = -.2959 Segments S.D. = 28.16 200 wpm 27 i = 7.019 r = -.2419 Segments S.D. = 10.38 275 wpm 30 i = 2.45 r = -.4594* Segments S.D. = 1.23 First 28 i = 5.839 r = -.382* Segments S.D. = 10.33 Second 30 i = 6.783 r = -.237 Segments S.D. = 15.69 Third 29 i = 5.62 r = -.217 Segments S.D. = 13.88 Fourth 31 i = 7.274 r = -.191 Segments S.D. = 21.02 *significant at the .05 level (p <.OS) TABLE 4.14 88 COMPARING MANIPULATION TERMINATION TO THE DIFFERENCE FROM PREFERENCE Manipulation Blocking Number Termination Correlation Group of Cases (seconds) Coefficient 100 wpm 31 i = 101.935 r = -.2537 Segments S.D. = 48.058 150 wpm 30 2 = 91.8 r = -.145 Segments S.D. = 47.41 200 wpm 27 i = 91.46 r = -.l66 Segments S.D. = 48.967 275 wpm 30 i = 74.23 r = -.0929 Segments S.D. = 50.67 First 28 i = 88.46 r = -.205 Segments S.D. = 34.48 Second 30 i = 103.27 r = -.151 Segments S.D. = 48.06 Third 29 i = 96.069 r = -.299 Segments S.D. = 53.84 Fourth 31 i = 72.56 r = -.182 Segments S.D. = 53.93 89 .05 level of significance. Neither blocking procedure yielded any consistent trend or movement between groupings. Little consistency was shown between correlation coeffic- ients for the different blocked groups suggesting that there" is little relationship between manipulation termination and difference from preference. Based on this finding, hypoth- esis #3 is rejected. Table 4.15 presents the correlation coefficients for the eight blocked groups when the relationship between difference from preference and manipulation duration is ex- amined. An examination of Table 4.15 indicates that six of the eight relationships were negative in direction. None of the relationships tested as significant at less than the .05 level of significance. Neither blocking procedure yielded any consistent trend or movement between groupings. Little consistency was shown between correlation coefficients for the relationship between manipulation duration and difference from preference. Based on this finding, hypothesis #4 is rejected. Analyzing With Reduced Standard Deviations An examination of the manipulation onsets of the 150 wpm segments, 200 wpm segments, first segments, second seg- ments, third segments, and fourth segments indicated stan- dard deviations that were greater than their respective means. Though statistically acceptable, such a finding 90 TABLE 4.15 COMPARING MANIPULATION DURATION TO THE DIFFERENCE FROM PREFERENCE Manipulation Blocking Number Duration Correlation Group of Cases (seconds) Coefficient 100 wpm 31 i = 99.79 r = -.253 Segments S.D. = 47.96 150 wpm 30 i = 77.60 r = -.027 Segments S.D. = 53.84 200 me 27 i = 84.47 r = -.111 Segments S.D. = 50.85 275 me 30 i = 71.78 r = -.1037 Segments S.D. = 50.84 First 28 i = 82.625 r = —.085 Segments S.D. = 36.645 Second 30 i = 96.48 r = -.068 Segments S.D. = 51.696 Third 29 i = 90.448 r = -.3388 Segments S.D. = 56.46 Fourth 31 i = 65.29 r = -.1065 Segments S.D. = 54.47 91 suggests that a movement of one standard deviation from the mean would yield a negative manipulation onset time. It is impossible to have a negative manipulation onset time since the subject would have to begin manipulating the rate of presentation before the presentation began. To accommodate for this, the individual manipulation onset times of all subjects were examined to ascertain which times had the greatest effect on the standard deviation. Starting with the manipulation onset time that varied the greatest from the mean, single times were discarded until the standard devia- tion for each of the blocking groups in question was brought numerically below the mean. By so doing, it was then pos- sible to accommodate at least 68% of the population, one standard deviation above and below the mean, as having man- ipulation onset times that were not negative. An analysis of relationships was made using this set of data to ascer- tain whether the previous correlation coefficients (Table 4.13, Table 4.14, and Table 4.15) were inordinately affected by these manipulation onset times with high variance from the mean. A total of eight manipulation onset times were discarded through this procedure. Table 4.16 displays the correlation coefficients for manipulation onset and dif- ference from preference for the altered data set. An examination of Table 4.16 indicates that six of the eight relationships were correlated in a negative dir- ection. This is a decrease from Table 4.13 where the 92 TABLE 4.16 COMPARING MANIPULATION ONSET TO THE DIFFERENCE FROM PREFERENCE USING THE ALTERED DATA SET Manipulation Blocking Number Onset Correlation Group of Cases (seconds) Coefficient 100 wpm 31 i = 2.145 r = -.049 Segments S.D. = .93 150 wpm 25+ 2 = 3.96 r = -.170 Segments S.D. = 2.95 200 wpm 24+ 2 — 3.604 r = -.184 Segments S.D. = 2.87 275 wpm 30 X — 2.45 r = .4594* Segments S.D. = 1.23 First 25+ 2 = 2.42 r = -.345 Segments S.D. = 1.618 Second 28+ i = 3.45 r = -.296 Segments S.D. = 2.86 Third 28+ 2 = 3.07 r = .1365 Segments S.D. = 2.12 Fourth 29+ 2 = 2.845 r = —.4386* Segments S.D. = 2.005 *significant at the .05 level (p <.05) +affected by the altered data set 93 unaltered data showed seven of the eight relationships were significant at less than the .05 level of significance. Of the two significant relationships, one was positive in di- rection and the other was negative in direction. The positive relationship that was significant was the same as that pre— sented in Table 4.13 since no data from the 275 wpm seg- ments were deleted as having high variance. The negative relationship that was significant occurred for the grouping by fourth segments. Of the six groupings that were altered through the elimination of manipulation onset times that varied greatly from the mean, four yielded lower correlation coefficients than were computed with the unaltered data. Neither blocking procedure yielded any consistent trend or movement between groupings. Little consistency was shown between correlation coefficients for the different blocked groups suggesting that there is little relationship between manipulation onset and difference from preference for the altered data. Further Analysis for Random Effects Further analysis of the data was conducted to ex- amine the experimental procedure in an attempt to ascertain whether the listening experience provided appropriate oppor- tunity for all subjects to demonstrate manifest preference for rate. This further analysis was based on the lack of significant relationships between difference from preference and manipulation onset, manipulation termination, and 94 manipulation duration. Since it was hypothesized that a relationship would exist, it was felt that the nonsignifi-~ cant finding could not be fully accepted until it could be shown that the data collected truly represented self-paced listening behavior and was not a function of some aspect of the study. The need for further analysis was also sug- gested when it was found that the longest non—manipulated duration, though occurring at the end of the listening seg- ments for more than half of the subjects, did not occur at the end of the listening segments for all subjects. Further, a decrease in percentage of subjects with the longest non— manipulated duration at the end of the segments occurred from segment to segment with the smallest percentage of sub- jects showing non-manipulation at the end for the fourth segments. For this analysis only Criterion #1 was employed. As such, only those subjects with all four final selected rates within a band width of 40 wpm were considered. No sub- jects were accepted for this further analysis on the basis of Criterion #2 or Criterion #4. Criterion #3 and Criterion #5 were also discarded since their implementation was de- pendent upon the use of Criterion #2 and Criterion #4. Since Criterion #2 and Criterion #4 provided for exceptions to the other criteria, their elimination in this further analysis provided a sample of subjects that was more strin- gent than that used in the preceding analyses. The mean 95 manifest preference rate for this further analysis was cal- culated on all four final selected rates for each subject. Correlation coefficients were computed for each of the hypothesized relationships for this group of subjects. Blocking procedures were again employed on the basis of the initial presentation rate of the segment and also on the basis of the order of presentation of the segments. A series of progressively more stringent acceptance levels were employed for the demonstration of manifest pref- erence for rate. This was accomplished by successively de- leting small groups of subjects with the greatest band width of final selected rates. Correlation coefficients were com— puted for each successively smaller group of subjects. A total of four different groups of subjects were analyzed in this manner. The groups consisted of 27 subjects, 21 sub- jects, 15 subjects, and 9 subjects. Each smaller group of subjects was a constituent part of all of the larger subject groupings. The data for the comparison of difference from preference and manipulation onset for each of the four group- ings is displayed in Table 4.17. Table 4.17 substantiates that a random effect was present in the collected data. This is shown by the wide variation in correlations for each level of stringency. For the group of 27 subjects, the correlations vary from .414 to -.367. The group of 21 subjects vary from .376 to -.416. The group of 15 subjects vary from .383 to -.456. The group 96 3o. v3 :53 mo. 05 on confluence“... 84.. u .o.m m.m~ u .o.m «46:6. n .o.m m.- u .o.m 3:858 hmv.u I u mm.m I m hvm.n I u b.0H I m mm~.1 n u mv.m I m mH~.| I u No.m I m . nuusom om.H I .n.m vv.H I .o.m mm.~ u .o.m mw.H I .m.m mucaauom mmo. n u Nh.~ I m mmo. u u Hm.m I m moo. u u m.m I m vmo. I u mm.~ I m cuwna mm.m I .m.m nH.m I .m.m o.mH n .o.m m.mH I .m.m mucmauom mmm.n u u Hm.v u m mmv.n I u ~m.m u m mwm.+ u u mv.n I m ovm.u I u mm.w u m ccoomm mm.m I .o.m mm.~H n .m.m o.~a I .o.m m.oa I .o.m mu8080om Hmv.u n u am.~ u m mvm.u u u mm.> u m mav.| u u ma.h u m hmm.n I u m.@ u m umumm 34 u on mmé u .o.m :4 u on 24 u on 3:053 «00. u u vv.~ I m Nmm. u u mv.~ u m mom. I u mm.m n m cvav. u u Hm.m I m 8&3 mum ov.v u .o.m no.ma I .o.w H.~a n .m.m m.oa I .o.m mucofimmm Nov.u u u an.v u m vo~.- u u mm.m u m nm~.a n u mv.m u m mH~.1 I u 0.5 I m 8&3 oom vH.m I .Q.m hm.0m u .o.m mm.0m n .m.m v.h~ I .o.m mucosmmm mmv.u I u Ha.v u m Oma.u u u mH.~H I m Hmm.n n u m.vH I m vom.n u u mv.~a n m 8&3 oma NH.H u .o.w mom. I .o.m mm. H .o.m o.H I .o.m mucwsomm 00m.n I u FH.N I m moa.u I u no.m u m mmo.| u u o.m u m m¢0.1 n u ma.m I m 8&3 ooH Amccoommv Amccoowmv Raccoommv Amccooowv ucmmowummov uwmco ucmmommmwoo ummco ucmmomumooo uwmco ucomomwuwoo ummco :oquHouuou cowumHs&«cmz cOwumHmuuou coHuoHo&«:mz cemumHmuuoo co«um~:&wcmz SOmumHouuoo noduoas&wcmz &:ouo 8&3 Hm I comm wocmu&moom 8&3 ma n comm oucmu&moum 8&3 on u comm mocmu&moom 8&3 ov n ccmm mocmu&ooo< onwxoon m I 2 ma I 2 am I 2 pm I z mOZHmbomw BZWOZHmBm mmoz Mdm>HmmmUUDm mom Ht ZOHmmBHmU m0 mefim WEB ZO ha.v mqmda muzmmNMHM& 20mm muzmmmhhHo N39 09 BmmZO ZOHB I m HH~.I u u mH.co I m ouuoom H.4m u .o.m H.mm n .o.m m.nm I .o.m H.mm I .a.m nucoauom mom. I u o.mm I m mod. I u om.nm I m own. I u mm.mo I m own. I u mm.mm I m cumoa c.8n u .o.m ~.mm u .o.m H.He I .o.m o.ov I .o.m mucusomm mma.n u u mm.haa I m mmH.u u u ab.o~H I m mmo.n I u mm.moa I m ovo.| u u mv.noa I m cooumm H.mm I .o.m m.om I .o.m m.mm u .o.m ~.vm u .o.m oucuouom mam.n I u ma.moa I m moo.u u u vm.mm I m ovm.u u u no.0m u m hmu.u I u m~.Hm I m umufim e.mv u .o.m m.om n .o.m «.mv n .o.m ~.Hm u .o.m mucusuom mm~.n u u m.Hm u m mma.u u u >.Hm u m Ho~.u u u ~m.no I m ova.n u u hm.oh u m 8&3 mow h.ov I .o.m m.mv I .o.m v.mm I .m.m >.om I .m.m muomfimmm mac. n u mm.mHH u m Hma. u u ov.~oa I m vHH.1 I u mo.Hm I m nvm.n u u vm.mm I m 8&3 oom m.~m I .m.m H.mv I .o.m o.v¢ I .o.m v.hv I .m.m muomscmm mac. n u vm.vm I m com. I u vH.Nm I m mmo.u u u mw.mm I m hmm.u u u mb.hm u m 8&3 oma ~.om n .o.m o.~v u .o.m m.vv u .o.m m.mv u .o.m mucosoom ovm.u I o om.hHH I m ~HN.I I u h~.noa I m omm.u n u vm.mm I m Hhm.u I u mm.moa u m 8&3 ooa Amcooommv Amcooommv Amcooommv Amcooommv uomonummoo camumoflauma uomemummoo oOmumoHBume uomwowummou oOHumoHEums uomwowmumou oowumofiauma ooHumamuuoo oowumao&aomz camumaouuoo oOMumHo&Homz ooHumHmuuou oOAumHo&aomz oOHumHmuuoo oowumao&womz &oouo 8&3 Hm I comm moomu&muom 8&3 ma I comm moomu&moo¢ 8&3 on I comm moomu&moo< 8&3 cm I comm moomu&moum mowxuon G" 2 ma I z Hm I 2 hN I z mUZHmbomU BZMUZHmBm mmoz NAN>HmmmUUDm mom Hw ZOHmmBHMU m0 mHmdm HEB 20. muzmmmmmM& 20mm muzmmmthQ mmfi OB ZOHBflZHzmmB ZOHB¢AD&HZ¢S UZHMdmzoo mH.v manta 100 m.mm I .o.m m.mm I .o.m v.~m I .o.m n.~m I .o.m nucqawom New. I I wh.om I m vmo. I I m.~m I m ~m~.u I I mm.om I m IHH.I I I nH.mm I m gIInom N.vm I .o.m «.mm I .o.m m.nm I .n.m ~.mm I .o.m uuomauum mmm. I I m~.om I m mmH. I I mm.mm I m mmm. I I ma.mm I m IMN. I I In.mm I m mena «.mm I .a.m o.mm I .a.m m.mv I .o.m m.am I .o.m uuouamcm mMH.I I I mo.maa I m ~mo.u I I mm.mHH I m boo. I I mv.~oa I m mmo. I I o.mm I m vacuum m.¢m I .o.m «.mm I .o.m «.mm I .o.m m.mm I .o.m muoqamom mo~.- I I Hm.~oI I m «mo. I I «H.Hm I m woo.u I I mm.mh I m mao.n I I mm.vm I m IIIII m.mv I .o.m ¢.om I .o.m m.mv I .o.m m.Hm I .o.m mucosuum nIm.- I I mo.mn I m III.I I I I~.mh I m oI~.I I I mI.mn I m mmI.I I I mo.vs I m an: mum «.mv I .o.m m.om I .o.m v.mm I .o.m «.mm I .o.m Iuouamom moo. I I «H.~IH I m How. I I vm.mm I m mmo.u I I no.~m I m va.I I I pm.om I m an: cow «.mm I .o.m m.~m I .o.m ~.Im I .o.m m.~m I .a.m mucosmom Inc. I I mm.om I m 0mm. I I mm.mn I m Ibo. I I mp.mh I m ~mo.u I I m.mh I m an: omI ~.pm I .o.m m.~v I .o.m m.vv I .o.m m.ov I .o.m muouavom mn~.n I I mm.mII I m oI~.I I I «.moI I m mwm.- I I em.hm I m II~.I I I ov.moH I m an: ooI Amcooommv Amcooommv AmcoOuwmv Amcooommv uomaoflmumoo ooflumuom uomfiuwmmwou ooHumIom uowwoflwuwou oOIumuom uomaoammooo ooHumuom oOIImHmIIOU oowumaomwomz oowumHmIIoo oOMumHomHomz oowumawuuou ooHumHomwomz oOwumHmIIoo oOaumHomfiomz macho 6&3 Hm I comm moomumwooo an: mm I comm muomummooo an: on I comm moomummouo an: em I comm ooomumouom mowxoon m I z ma I z Hm I Z pm I z mUZHmbomU BZNUZHMBm mmOZ wam>HmmmUUDm mom Ht ZOHmmBHMU m0 mHmdm mmfi ZO muzmmmmmmm Zomh muzmmmhmHO mm? 08 ZOHHNH HMH m.m NHN m.NvH m.va N HOH 5 HM vHH mHH H VNN ONH NNH N OHN H Om NO OHH MH OHN OO HO H hHN NH ON m.OO m.OO N vOH Ov m.Om m.N OOH O ON I I I vnH I I I NbH HH 5N m.Oh m.vn v vOH m.On m.HO N OOH O ON I I I mmN I I I ObN m VN HmH me m HOH vs OHH Nv vOH m MN O O O OON m.mm m.hm N OON HH NN OEHB mEHB mEHB wumm OEHB «EH9 wEHB mpmm mama # coHumuso coHumcHEumB ummco Hmch :oHumuso coHumcfleumB uwmco Hmcwm msHseHuO uomnndm ucmEOwO Ocoumm ucmfiuwm umuwm A.o.ucoov aomnmom wm «BOO 123 0.00 HOH O.N NON mO Hp O OON OH vH O.NH O.vH N HOH v n m OhH O NH NO O.vO O.N OOH N On hr OOH m HH m h v OOH m.mOH 0.00H O NOH N OH 0.00 0.00 O HHN OHH mHH m NOH n O O OH N vhH O O O OOH H O NO OO H OON OH OH N VOH v n I I I OHm I I I OhN O O OOH OOH H OOH Hv Nv H vnH O O I I I OOH I I I HNN O O ONH HNH H ONN O.mvH 0.0vH N 0.00H N m I I I ONH I I I OhH n N I I I NON I I I ONN H H wEHB mEHe mEHB mumm mEHe weHe QEHB mumm mama x coHumuao coHumcHEuwa ummco HMCHO COHumusO :oHuMCHEume ummco HMCHO msHseHum uomnnsm unmEOmO nausom acmsOmO OHHna 0.0vH an O.H VOH 0.0HH OHH O.H OOH OH mm I I I OHN I I I NNH O NO I I I OmN I I I OOH O HO OEHB OEHB wEHB mumm OEHB OEHB OEHB mumm mama * COHumusa COHOMCHEumB ummco HMGHO :oHumuso coHuOCHEhmB ummco Hmch msHSEHuO uomnnsm ucmEOmm Ocoomm unmEOmm umuHm A.@.ucoov BUMmem Mm