EEEE EEEEcEs EEE EELEEEEAEEE PROSE CGMPRESSED‘: DzssertaEm EEE EEe {Eegree OE ?h D: ..... .. . 1 EEE SHEEN STATE UEEEVEWEY 1 m1 Emmnmm : V V ,fi‘éi-fi:;;;:;. 1- ‘ 19231 s .5 m... _...L L I B R A R Y Michigan State University .- .- <‘ _ \ .' ' ., o " . ‘ J This is to certify that the . , thesis entitled V I ,1; ' _. The Effects of Telegraphic Prose, Compressed 4- Speech, and Modality Upon Comprehension presented by Polly Wietzke Peterson A \v’ has been accepted towards fulfillment 1;, of the requirements for $9? Ph.D. ,1. ‘ :, jg.- tut ' fig October 24, 1973 51? .. 0-7639 1;, ABSTRACT THE EFFECTS OF TELEGRAPHIC PROSE, COMPRESSED SPEECH, AND MODALITY UPON COMPREHENSION By Polly Nietzke Peterson Information processing and learning efficiency have become the recent focus of vehement concern and conflict in behavioral science research. As the knowledge explosion is geometrically progressing at an ever increasing rate, investigations involving new concepts and techniques designed to attempt to increase the efficiency of informa- tion input are critical from both a theoretical and practical stand- point for all levels of education. The purpose of this study was to investigate the effects of combining telegraphic style prose, time-compressed speech, and modality upon comprehension. These three independent main effects were analysed in two 3 x 4 factorial designs and in one l x 4 design with g_= 20 §s in each of 28 treatment cells. A 3 x 4 x 2 ANOVA was employed to compare the listening modality with that of combined listening and reading. The subject population was 560 University of Miami students. A 2692 word fictional passage of appropriate interest and dif- ficulty level for university §s was generated in traditional prose form specifically designed to maintain the essential information in Polly Nietzke Peterson the telegraphic versions in which 20, 40, and 60% of the words were deleted. The four versions of the prose story were each then aurally tape-recorded and time-compressed into three presentation rates: l75 wpm. 275 wpm. and 400 wpm. Subjects were assigned randomly to 24 experimental treatments in which one-half (240) of the §s were assigned to listen to one of the 12 telegraphic prose compressed speech combinations and the other half (240) were assigned to simultaneously listen and read. Eighty additional §s were assigned to read one of the four story versions and these data were used as the control measure. The dependent variables were an objective comprehension test, analysed in four parts; a measure of learning efficiency; and, for the reading treatments, reading rate and reading time. The §s completed a pre-task self-rating form on preferred mode of learning, reading rate, and conversational speech rate. Subjects assigned to the com- bined listening/reading treatments also completed a post-test modality form. Since no previous research has investigated either the combined inputs of telegraphic style prose and time-compressed speech or the effects of aural, visual, and aural/visual modalities upon this com- bination of learning techniques, hypotheses were largely exploratory in nature. In the listening treatments, a drastic decrement in com- prehension was predicted at the highest deletion levels and pre- sentation rates. The addition of the reading modality to the listening one was expected to recapture some of this decrement in comprehension. Main concerns of the study were to attempt to Polly Wietzke Peterson quantify where the highest learning efficiency occurred and where di- minishing returns occurred on the comprehension curves for these treat- ments. The issues of how informationally compact telegraphic messages are and the neurological limitations on central processing abilities were also studied. The results were highly significant for each of the three main effects: deletion level, presentation rate, and modality. Significant interaction between these main effects was found only for the learning efficiency variable (ENC). For presentation rate, comprehension was maintained at the 275 wpm speed at almost the same level as at 175 wpm for both the listening and listening/reading treatment conditions. For deletion level, comprehension for the 20% reduced version was almost as high as that for the traditional version for both listening and listening/reading at the 175 and 275 wpm rates. Comprehension was maintained significantly better than expected at the 400 wpm rate for the traditional and 20% deleted story versions for both the listening and listening/reading modalities. THE EFFECTS OF TELEGRAPHIC PROSE, COMPRESSED SPEECH, AND MODALITY UPON COMPREHENSION By Polly Nietzke Peterson A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Elementary and Special Education 1973 5? ACKNOWLEDGMENTS The writer is indebted particularly to Dr. Byron H. Van Roekel for his wise counsel, professional friendship, and support throughout her years in graduate school. Dr. Van Roekel's high standards of academic excellence, his dedication to learning, and his remarkable ability to instill these qualities in his students made the writer's entire graduate program most exciting, valuable, and pleasurable. Special appreciation is expressed to Dr. Clessen J. Martin, Chairman of the Department of Psychology and Director of the Human Learning Research Laboratory at Texas A&M University. As advisor to this research project, Dr. Martin's unfailing interest over time and distance, and his creative talent and expertise in research were not only invaluable in the writing of this dissertation but also created a lasting interest in the writer to undertake continued research studies. This project will be part of Dr. Martin's final report to the National Institute of Education, Department of Health, Education, and Welfare for his research grant entitled Development of an Analjrtical Approach to Telegraphic Communication. Sincere gratitude is also expressed to Doctoral Committee members Dr. Jean LePere whose classes were a highlight of the writer's graduate coursework, and to Dr. Robert J. Geist, Professor of English, for his continued participation and interest. ii Without the cooperation and generous assistance of the University of Miami's faculty and students, this research could not have been completed. Many thanks are thus in order to the faculty in the School of Education and in the Departments of Psychology and Architecture, and to the 560 students whose participation in and enthusiasm for this study made the data gathering a truly delightful experience. Grateful appreciation is expressed to Mrs. Ruth Ann Powell, Psychology Departmental Secretary at Texas A&M University, for the entire typing of this dissertation and for her ever pleasant and vital assistance during many months of long distance communication. TABLE OF CONTENTS LIST OF TABLES .......................... LIST OF FIGURES ......................... LIST OF SYMBOLS ......................... INTRODUCTION AND REVIEW OF THE LITERATURE ........ Overview of the Problem ................ Review of the Literature ................ Language Redundancy and Telegraphic Prose ...... Time-Compressed Speech ................ Listening and Reading Comprehension ......... Hypotheses ....................... Summary ........................ METHODS AND PROCEDURES .................. Subjects ....................... Materials ....................... Basic Structure of the Four Story Versions ...... Traditional Story ................. Dawes' Set Relations Model ............. Telegraphic Versions ................ Grammar, Punctuation, and Style Changes in the Deletion Process ............... Calculation of Readability Level . . ._ ....... Reading Material Format .............. Time-Compressed Tape Recordings ........... Evaluation Material ................. Multiple Choice Items ............... Set Relations Items ................ Pilot Test of the Questions ............ Recording and Playback Equipment ........... Additional Materials ................. Procedure and Design of the Study ........... Listening and Listening/Reading Treatment Conditions . Listening Treatment GrOUps ............. Listening/Reading Treatment Groups ......... Reading Treatment Conditions ............ "iv Page vii xi Chapter Additional Procedures ................ Comprehension Test ................. Modality Form and Posttest Comments ........ Reporting Results to the Subjects ......... Posttest Optional Participation .......... Design and Analysis ................. III. RESULTS ......................... Reading Treatments ................... Listening Treatments .................. Listening/Reading Treatments .............. Listening Treatments Compared to Listening/ Reading Treatments .................. IV. DISCUSSION ........................ Presentation Rate ................... Deletion Level ..................... Modality ........................ Reading Rate ...................... Specific Comparison of Treatments ........... Effective Word Per Minute Comprehension (ENC) ..... Prestudy Self-Rating Form ............... Posttest Modality Form for Listening/Reading . Treatments ...................... Telegraphic Prose ................... Comparison of Findings with Previous Telegraphic Prose Research ................... Telegraphic Prose and Dawes' Set Relations Model . . . Time-Compressed Speech ................. Summary of Hypotheses ................. V. CONCLUSIONS ....................... Reactions and Conclusions of Ss ............ Implications for Further Research . . . . ....... General Conclusions .................. APPENDICES ............................ A. San Francisco--Traditional Version ............ B San Francisco--Twenty Percent Deletion .......... C. San Francisco--Forty Percent Deletion .......... D San Francisco--Sixty Percent Deletion .......... 112 112 115 116 118 119 120 124 124 126 127 134 134 135 144 159 Chapter Page E. Comprehension Test for "San Francisco" .......... l64 F. Prestudy Self-Rating Form ................ 178 G. Posttest Modality Form .................. l80 H. ANOVAs for Reading Treatments .............. l8l REFERENCES ............................ T83 vi Table 2.1 2.2 2.3 2.4 2.5 2.6 2.7 3.1 3.2 3.3 3.4 3.5 LIST OF TABLES Number of Male, Female, Graduate, Cuban,Black, Physical Education and Architecture §s Randomly Assigned to the Twenty-Four Treatment Conditions .............. Number of Words in Each Grammatical Category in the Four Versions of San Francisco ....... Flesch's Reading Ease Categories ......... Flesch Readability Scores for "San Francisco" Traditional Version of 2692 Words ....... Information on the Original Recordings of "San Francisco" ................... Data on the Time-Compressed Speech Tapes for "San Francisco" ................ Summary Statistics of the Two-Part Test Developed for "San Francisco" ......... Item Analysis of the Total Multiple Choice Items (60) and Set Relations Items (20) for the Reading Treatments ............... Means and Standard Deviations for Reading Time, Reading Rate, and Effective Word Per Minute Comprehension Rate (ENC) in the Reading Treatments ................... Results of l x 4 ANOVA for Reading Rate in the Reading Treatments ............. Results of l x 4 ANOVA for Reading Time in the Reading Treatments ............... Means and Standard Deviations for Four Dependent Variables in the Reading Treatments ...... vii 35 41 44 44 46 47 49 59 61 61 62 64 Table 3.6 3.7 3.8 3.9 Results of l x 4 ANOVA for Multiple Choice in the Reading Treatments .............. Means and Standard Deviations for Multiple Choice and Set Relations for Presentation Rate and Deletion Level Main Effects in the Listening Treatments .................... Results of 3 x 4 ANOVA for Multiple Choice in the Listening Treatments ............. Results of 3 x 4 ANOVA for Set Relations in the Listening Treatments ............... Means and Standard Deviations for Nested and Disjunctive for Presentation Rate and Deletion Level Main Effects in the Listening Treatments .................... Results of 3 x 4 ANOVA for Nested in the Listening Treatments ............... Results of 3 x 4 ANOVA for Disjunctive in the Listening Treatments ............... Means and Standard Deviations for Each of the Five Dependent Variables in the l2 Listening Treatments ............... Means and Standard Deviations for ENC for Presentation Rate and Deletion Level Main Effects for Multiple Choice in the Listening Treatments ............... Results of 3 x 4 ANOVA for ENC for Multiple Choice in the Listening Treatments ........ Means and Standard Deviations for Multiple Choice and Set Relations for Presentation Rate and Deletion Level Main Effects in the Listening/Reading Treatments ........... Results of 3 x 4 ANOVA for Multiple Choice in the Listening/Reading Treatments ....... Results of 3 x 4 ANOVA for Set Relations in the Listening/Reading Treatments ...... Means and Standard Deviations for Nested and Disjunctive for Presentation Rate and Deletion Level Main Effects for the Listening/Reading Treatments .......... viii Page 65 67 69 69 7O 71 71 72 75 75 77 77 79 80 Table 3.20 Results of 3 x 4 ANOVA for Nested for the Listening/Reading Treatments .......... Results of 3 x 4 ANOVA for Disjunctive for the Listening/Reading Treatments .......... Means and Standard Deviations for ENC for Presentation Rate and Deletion Level Main Effects for Multiple Choice in the Listening/Reading Treatments .......... Results of 3 x 4 ANOVA for ENC for Multiple Choice in the Listening/Reading Treatments . . . Means and Standard Deviations for Each of the Five Dependent Variables in the l2 Listening/Reading Treatments .......... Means and Standard Deviations for Multiple Choice and Set Relations for Presentation Rate and Deletion Level Main Effects in the Listening Treatments Compared to the Listening/Reading Treatments .......... Results of 3 x 4 x 2 ANOVA for Multiple Choice in the Listening Treatments Compared to the Listening/Reading Treatments ........ Results of 3 x 4 x 2 ANOVA for Set Relations in the Listening Treatments Compared to the Listening/Reading Treatments ........ Means and Standard Deviations for Nested and Disjunctive for Presentation Rate and Deletion Level in the Listening Treatments Compared to the Listening/ Reading Treatments .............. Results of 3 x 4 x 2 ANOVA for Nested in the Listening Treatments Compared to the Listening/Reading Treatments .......... Results of 3 x 4 x 2 ANOVA for Disjunctive in the Listening Treatments Compared to the Listening/Reading Treatments ........ Means and Standard Deviations for ENC for Presentation Rate and Deletion Level Main Effects in the Listening Treatments Compared to the Listening/Reading Treatments .................. ix Page 80 81 81 82 85 86 86 88 89 90 91 92 Table 3.32 3.33 4.1 4.2 4.3 4.4 4.5 4.6 A-Hl A-H2 A—H3 A-H4 Results of 3 x 4 x 2 ANOVA for ENC in the Listening Treatments Compared to the Listening/Reading Treatments .......... Means and Standard Deviations for Each of the Five Dependent Variables in the Listening Treatments Compared to the 12 Listening/Reading Treatments ........ Percent Correct for Multiple Choice Test Showing the Effect of Deletion Level, Presentation Rate, and Modality for All 28 Treatment Conditions ............ Reading Treatment Conditions Compared to Listening and Listening/Reading for Multiple Choice Test ............. Reading Comprehension for the Traditional Version for the Multiple Choice Test ...... Comparison of All of the Experimental Treatment Conditions to the Control Reading Condition (Traditional) for the Multiple Choice Test . . . Results of Subjects' Self-Rating on Mode of Learning Efficiency, Reading Rate, and Conversational Speech Rate ........... Modality Relied Upon Most in Listening/Reading Treatment Conditions as Checked by Each Subject ................... Results of l x 4 ANOVA for Set Relations in the Reading Treatment ............ Results of l x 4 ANOVA for Nested in the Reading Treatments ............... Results of l x 4 ANOVA for Disjunctive in . the Reading Treatments ............. Results of l x 4 ANOVA for ENC for Multiple Choice in the Reading Treatments ....... Page 93 97 lOl 107 108 109 113 114 181 181 182 182 Figure 2.1 2.2 2.3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 LIST OF FIGURES Set relations existing before the earthquake as described in all four treatment versions of "San Francisco" ............... Set relations existing after the earthquake as described in all four treatment versions of "San Francisco" ............... Research design showing treatment conditions analysed by the analysis of variance model . . . Effect of deletion level on the two measures of reading rate for the reading treatments . . . Effect of deletion level on reading time for the reading treatments ............. Mean comprehension levels for multiple choice in the l2 listening treatments ......... Mean comprehension levels for multiple choice in the l2 listening/reading treatments ..... ENC rates comparing the Listening and Listening/Reading treatments and the control reading treatment ............... Interaction between presentation rate and deletion level main effects for ENC in the 3 x 4 x 2 ANOVA ................ Interaction between presentation rate and modality for ENC in the 3 x 4 x 2 ANOVA xi Page 38 39 56 63 63 73 78 84 95 96 Telegraphic Prose Deletion Level LIST OF SYMBOLS T 20 40 60 Modality Other ENC traditional story version 20% deletion of words 40% deletion of words 60% deletion of words listening (aural) reading (visual ) Compressed Speech Presentation Rate 1 l75 words per minute 2 275 words per minute 4 400 words per minute listening plus reading (aural/visual) effective wpm comprehension rate Treatment Conditions Tl 20 4o 60 1 l 1 20 402 602 traditional at l75 wpm 20% deletion at 175 wpm 40% deletion at T75 wpm 60% deletion at l75 wpm traditional at 275 wpm 20% deletion at 275 wpm 40% deletion at 275 wpm 60% deletion at 275 wpm xii T4 = traditional at 400 wpm 204 = 20% deletion at 400 wpm 404 = 40% deletion at 400 wpm 604 = 60% deletion at 400 wpm CHAPTER I INTRODUCTION AND REVIEW OF THE LITERATURE Overview of the Problem At no previous time in history have educational methodology and the process of learning been the focus of such vehement concern and conflict. A basic challenge confronting education and related pro- fessions today is definitive research on the ways individuals pro- cess information and on new means and techniques for increasing the efficiency of learning. The rationale for the crucial need to investigate the possi- bilities and potentialities of increasing rates of information input and of training high speed comprehension is twofold. First, a priori evidence exists to suggest it can be done; secondly, it can without question be shown that the need for it has never been so great. The knowledge explosion is prodigious. The volume of communication is enlarging each year as is the amount of information that is being taught at every level of education. The sum of material which an individual must master just in order to attempt to keep up academi- cally or professionally is rising astronomically and yet the methods of presenting information , with few exceptions, remain the same as 'those which have been used throughout the long history of pedagogy. In this age of technological brilliance, a number of new in- structional concepts and methods are now feasible which could make it possible to bridge the gap between the available knowledge needed and the time required to process it. The didactic potentialities of es- tablishing more efficient and effective methods of information input and storage appear limitless for application to both normal and handi- capped learners. For example, while the invention of Braille in l829 was a phenomenal breakthrough for the teaching of the blind, Braille reading rates for blind high school students average 90 wpm (Ethington, l956) as compared to 25l wpm for sighted high school level readers (Harris, l947). Thus the blind are tremendously handicapped in information processing when compared to normal learners who are themselves inundated with too much to learn and too little time to do so with the present traditional methods for the comprehension of material. One exciting technique now available which is particularly applicable to the blind as well as to the normal learner is that of time-compressed speech which permits the rate of speech to be in- creased without distortion of pitch. Research investigations in- volving time-compressed speech which relies on the aural channel and other new techniques such as telegraphic prose which, using mainly the visual channel, seeks to increase learning efficiency through elim- ination of excess verbiage are barely at the threshold despite a recent burgeoning of interest and studies in these and other innovative concepts designed to increase information input. As illustrated here and in the review of the literature which follows, learning techniques such as time-compressed speech and telegraphic prose appear to have significant potential for learners at all levels, but further scientific investigation of the contingent variables, relationship of modalities, and limitations of information input and storage is needed to specifically define and delimit the parameters involved. A relatively small amount of research work has been reported regarding the telegraphic concept of written prose which employs techniques to reduce verbiage in order to speed up information input. Significantly more research investigations have studied the concept of time-compressed speech. The compressed speech literature has shown that S; can process information via the aural modality at faster than normal presentation rates of connected speech but many questions remain unanswered or in conflict. Both of these learning techniques obviously depend on the basic communication channels of listening or reading. Despite an abundance of research efforts span- ning many years, listening comprehension and reading comprehension continue to plague investigators with unresolved complexities regard- ing teaching, rate, central nervous system limitation, individual predilection for one or the other, and the effects of simultaneous exposure to both modalities. The existing research knowledge in telegraphic prose, time-compressed speech, and listening and reading comprehension is thus necessary to this study and is reviewed in the next section of this chapter. One of the major problems of concern of the research that has been done on telegraphic prose is that of whether the message is more highly informationally compact; and, if so, whether it is the compaction which renders the message more difficult and causes the consistent decrement of approximately 50% in reading rate on the highly reduced telegraphic materials. Since it is known that there is a drop off in comprehension as compression rate increases above a certain point, listening to high wpm compression rates can be used as a vehicle to test whether the telegraphic messages are more informationally compact and therefore more difficult to comprehend. Since the rate of information input is controlled by the compressed speech tape, there should be a differential drop off in comprehension and significant interaction between the wpm Presentation Rate and the telegraphic Deletion Level if the telegraphic materials are more compact and if it is the com- paction which causes the material to be significantly more difficult and thus more time consuming to comprehend. This question is a central issue to be answered in the study of telegraphic prose. The relation- ships among reading rate, listening rate, speech compression, and telegraphic compaction require study which the design of this research will permit. No known research has investigated directly the combined learning techniques of telegraphic style prose and compressed speech. Nork analysing these with reading (visual), listening (aural), and combined listening/reading (aural/visual) modalities is also nonexistent. Thus this study is seeking definitive data in relatively uncharted territory. Although it is believed that future research and technological advances may establish more efficient ways to process information, a basic assumption upon which this research study is based is that irudividuals have a limited capacity for input of information. The intent of this research investigation is to study the effects of com- bining compressed speech and telegraphic style prose upon comprehension and to attempt to define where diminishing returns occur in tasks involving four levels of telegraphic prose combined with three rates of time-compressed speech and three modalities of material presentation. An additional concern is to pursue the relationship between the compres- sion of speech and the compaction of prose materials. The purpose of this study is to provide some critically needed additional data on the interrelationships among the quantity of infor- mation, the mode of presentation, and the rate at which it is transmitted. A major goal of this investigation is to obtain some normative data on the complexities of these interrelationships through the use of two specific learning techniques, telegraphic prose and compressed speech, combined with the effects that modality (reading, listening, and listening/ reading) has upon curves of comprehension. Review of the Literature Language Redundancy and Telegraphic Prose The concept of developing telegraphic materials was based upon the rules of formal English grammar and on the presence of redundancy in English prose. The grammatical structure of the English language results in the inclusion of words in sentences and phrases which may be insignificant in transferring the intended meaning of the sen- tence. The efficiency of grammatically correct English prose forms may thus be less than perfect in relation to the rate of informa- tion processing. Both written and spoken language contain many words and word sequences which are unnecessary for comprehension of the meaning. These sequential word dependencies and the presence of redundancy form the theory from which the telegraphic prose concept was formulated. The fact that redundancy is present in language and that sub- jects are aware of this redundancy has been demonstrated in research investigations originating primarily from the area of information theory. Although this research does not relate specifically to the telegraphic concept, it provides some support for the feasibility of increasing learning efficiency by the development of telegraphic materials. Garner (1962) has written a book on redundancy in English in which he presents a review of the literature concerned with problems in obtaining a direct measurement of redundancy in written English. The initial attempts to estimate letter redundancy were made in relation to letters which were followed by studies of word associated redundancy. Shannon (1949) found that a printed message has 50% letter redundancy; that is only 50% of the available letters were necessary for the interpretation of the message. Using a new method based on the predictability of written English, Shannon (1951) con- cluded that a letter is at least 75% determined by what has preceded it.’ Shannon's (1951) model was used by Burton and Licklider (1955) and confirmed Shannon's estimate of letters being 75% re- dundant. Newman and Gerstman (1952), however, used a different method and estimated redundancy at only 52% using a 10,000 word ex- tract from the King James version of the Bible. Chapanis (1954) studied letter redundancy by subjects' ability to replace missing letters in words. Chapanis (1954) deleted letters in six amounts ranging from 10 to 67% in both random and regular patterns from 13 prose passages which differed widely in style and content. Subjects were able to replace approximately 90% of the missing letters on the passage which had deleted 20% of the letters in a regular pattern. At 25% deletion on a regular basis, subjects were able to replace about 70%. For deletion levels above 33%, how- ever, the ability to restore the missing letters decreased drastically. Nhen 50% of the letters in theiext were deleted, only 10% of the text could be correctly restored. One of the most important find- ings was that if the redundancy in the passage was high, recon? struction was significantly easier than when the passage had a low level of redundancy. Low redundancy made the deleted letters much more difficult for the subjects to replace. This finding may be of great importance to this present study and to the other research directly investigating telegraphic prose. In 1957, Miller and Friedman replicated Chapanis' results in a similar study. In an experiment designed to explore the recall of redundant strings of letters, Miller (1961) found that in a subject population of uni- versity students, the amount of material learned increased when the letter strings were redundant, but the amount of information, measured in bits, decreased. By this information measure, Miller concluded that redundancy in the materials to be learned does not increase the efficiency of learning. The 26 letter English alphabet makes estimates and analyses of letter redundancy much easier than those for word redundancy where practially no limiting factor exists. Because of this the redundancy of words in written English has been studied in relation to general frequency of occurrence. The Thorndike and Large (1944) word count on the occurrence frequency of 30,000 words provided the initial data of word associated redundancy.’ Nord recognition studies have provided further knowledge regarding word associated redundancy. Using tachistoscopic exposures to words, Solomon and Howes (1951) found that the recognition thres- hold of the word decreases as the Thorndike-Large (1944) frequency of the word increases. In a similar study, Solomon and Postman (1952) controlled the frequency of nonsense words tachistoscopically and found that the more frequent the occurrence of the nonsense word, the more easily it was recalled. Pierce and Karlin (1957) and Staats (1968) provide evidence that reading rates increase as word familiarity in- creases. Although the research has established that a relatively large percentage of letters can be deleted without a significant loss in intelligibility, a fairly low percentage of word deletion results in serious impairment of a subject's ability to specifically replace the missing word. Morrison and Black (1957) and Aborn, Rubenstein, and Sterling (1969) found that accuracy in replacing just one deleted word from an otherwise complete sentence was only 40 to 50%. In a related study, however, Aborn and Rubenstein (l958)found that subjects were able to replace the deleted words with a similar, contextually meaningful word even though they had great difficulty in predicting the exact deleted word. The preceding research substantiates redundancy as an important feature of language. Nhile subjects are able to replace missing letters and words with appreciable accuracy because of this re- dundancy, more of these studies investigated the effect of deletion upon comprehension. The research suggests, however, that some re- dundant features may be deleted from written English without loss of comprehension which lends credence to the telegraphic concept which is based on the elimination of nonessential information. Studies on the early development of language in young children also support the concept of learning economy through use of a tele- graphic paradigm. The work of Brown and Bellugi (1964), Carroll (1961), and Leopold (1961) establishes that in early acquisition of syntax, children speak in telegraphic style using mainly essential high information nouns and verbs. Although most modifying words are deleted, such early stages of language development are comprehendable. In the initial generation of language, children use single words to express whole sentences or phrases (Carroll, 1961). In the second half of the second year, two word structures are heard with syntact- ical constructions involving three or more words occurring toward the end of the second year of life. During the third year a child induces latent structure of language from the speech to which he is ex- posed (Brown and Bellugi, 1964). Thus while one of the main charac- teristics of the speech of very young children is its telegraphic nature, children are at the same time learning the structure of language through its distinctive features and redundancy. Brown and Bellugi's work (1964) not only provided significant breakthrough knowledge regarding the acquisition of language but related as well to the feasibility of the telegraphic concept as illustrated in the following quotation: 10 Ne adults sometimes operate under a constraint on length and the curious fact is that the English we produce in these circumstances bears a formal resemblance to the English pro- duced by two-year-old children. Nhen words cost money there is a premium on brevity or to put it otherwise, a con- straint on length. The result is "telegraphic" English and telegraphic English is an English of nouns, verbs, and adjectives. . . .Ne make the same kind of telegraphic re- duction when time or fatigue constrain us to be brief, as witness any set of notes taken at a fast moving lecture. A telegraphic transformation of English generally communicates very well. It does so because it retains the high-information words and drops the low-information words. The concept of telegraphic prose which, as noted previously, is based on the deletion of excess verbiage to increase learning efficien- cy has only recently begun to be investigated. Initial research on the feasibility of this concept was undertaken by Martin and Alonso (1967) in a project utilizing telegraphic prose in Braille as a method to increase information processing in visually handicapped students. A total of 210 Braille readers were assigned to read one of three versions of material constructed in traditional form, medium tele- graphic form,and high telegraphic form. Many problems of measure- ment exist in designing deletion schemes for existing meaningful prose material. A model developed by Dawes (1964), however, makes it feasible to hold the essential information constant in varying versions of passages specifically designed to use set relations for measure- ment. Martin and Alonso (1967) employed Dawes' method to construct the prose materials and test measures in this investigation. The constructed story concerned two warring African nations. Martin and Alonso's (1967) traditional story version consisted of 1,620 words and was written in a typical English prose style. 11 The medium telegraphic version, containing 947 words, used traditional sentence form and reduced the original material 42%. The high telegraphic version contained only 455 words which was a 72% reduction of the original and was written in outline form. All re- ductions were based on Dawes' (1964) model so that the set relations or essential information were maintained in all three versions. As determined by the Dale-Chall.readability formula, this original version was found to be appropriate for average seventh and eight grade readers. The study imposed three recall conditions and analyses were done on comprehension, reading rate, and reading time. Learning and re- tention of the essential information on the telegraphic versions compared positively with the traditional version on almost all measures. Reading rate, however, decreased by almost 50% for the braille readers assigned to the highly telegraphic version of the story. It had been assumed that the same reading speed would be maintained for the telegraphic as for the traditional materials. To account for this unexpected finding, it was postulated that: l. The unfamiliar format and style may have caused readers to reduce their normal reading rate and, 2. The compactness of information may have capsed students to reduce speed in order to process the highly in-. formational stimulus. The authors felt that the first postulate was the more plausible of the two explana- tions. It had also been assumed that the descriptive narrative and extraneous detail of the traditional passage would not be recalled as it was not essential for complete understanding of the story. This assumption also proved to be wrong. 0n the delayed recall objective 12 test the results suggested that this extraneous information was con- textually helpful in processing and storing the essential information of the story. A major problem of telegraphic style prose was noted by the authors to be the absence of any systematic rules or methods by which to delete nonessential words from traditional materials. The re- sults of the study in general, however, were interpreted as supporting the feasibility of the telegraphic concept. Additional research on the telegraphic concept has been completed recently by Martin and Herndon (1971), Martin and Hope (1972), Pantalion (1972), and Sheffield (1972). The purpose of the Martin and Herndon (1971) study was to partially replicate Martin and Alonso's (1967) work to try to establish on a non-visually handicapped population the cause of the significant decrease in reading rate for highly reduced telegraphic materials. Sixty university undergraduate and graduate subjects were assigned to either read or to listen to the same seventh grade difficulty level materials about two warring African nations that were used in Martin and Alonso's (1967) study. Aurally presented messages force the student to process the infor- mation at a constant input rate rather than allowing voluntary change in input rate as reading does. It was reasoned therefore that if telegraphic materials are more difficult due to unique style or compactness, comprehension should drop for the listening condition. The results of this study found that reading rates dropped signifi- cantly on the high telegraphic materials, but comprehension was maintained at equivalent levels for both the reading and listening treatments. Since the aural tape presentation of the materials was 13 at a slow 120 wpm rate, this was stated as the reason compre- hension was as high as on the visual telegraphic messages. No in- teraction was found between the modality and material format main effects which led the authors to suggest that comprehension of tele- graphic materials is independent of the presentation mode. In another study, Martin and Herndon (1971) investigated the base levels of acceptable deletion of words in telegraphic materials. From the results, the authors concluded that 10 percent of the words in prose reduction can be randomly deleted without significantly affecting reading rate or comprehension. No solutions regarding an objective method for deleting words from existing materials was considered, however. Martin and Hope (1972) designed a study to test the two hy- potheses formulated to account for the significant decrement in reading rate on highly telegraphic material in the Martin and Alonso (1967) study with Braille readers. These authors tested the two hypotheses with undergraduate university students. Tape record- ings to provide a constant input rate were used to present the material aurally to one group of students while another student group read the same material at their own individual speeds. Three forms (Traditional, Medium Telegraphic, and High Telegraphic) of each story were constructed. The aural and the written or visual versions were identical for each form. Results of this were interpreted to support their previous hypothesis (1967) which speculated that in order to process extremely compact information effectively, students reduce their wpm reading speed. No support was found for the hypothesis that the reduction in reading rate was caused by the unique style of 14 telegraphic prose and that practice would increase reading speed. The major finding of significance in Martin and Hope's study, however, was that both actual total reading and listening time for the High- Telegraphic versions was almost 70% less than total time spent on the Traditional versions. The added finding that comprehension on the High-Telegraphic versions dropped at an average of less than 10% when compared to comprehension on the Traditional versions gives substan- tial support to the efficiency of the telegraphic prose concept. One other interesting result of the experiment was that the aural mode of presentation produced significantly lower comprehension scores for all three traditional and telegraphic versions despite a slower wpm presentation time (average recording rate equaled 128 wpm) than average reading or speaking rates. In another study using undergraduate studentssPantalion (1972) also found a significant saving of reading time in medium to high (40 to 50%) telegraphic passages in comparison to the traditional version. He also established the efficacy of a method termed SHORT (Subjective Hierarchy of Relevant Terms) which provides a basis for the generation of different deletion levels for sentences. This method is based upon subject ranking of words within each sentence on the basis of their functional communication value. Sheffield (1972) examined the effects of subjectively reduced telegraphic passages upon comprehension, reading rate, and reading time in a study which used both blind and sighted sixth grade level subjects. Results of this study indicated that telegraphic reduction of 30% was possible without loss of comprehension or increase of reading time. Blind subjects were able to process the 50% reduced 15 version better than the sighted students which implies that there is a difference in the way blind and sighted readers process materials. All of the studies on telegraphic prose have resulted in findings contradictory to those on which Dawes' set relations model is based. Dawes concluded that the extraneous material found in traditional prose would result in more overgeneralization or disjunctive errors due to information overload and that these errors would be reduced on the telegraphic passages. The opposite result was found in all of the previously cited telegraphic studies. It appears that the extran- eous detail in traditional style prose may benefit the reader's recall and storage of the essential facts. In summary, the concept of telegraphic prose appears to offer a means to facilitate the speed of information processing. Further research is needed to clarify its ideal efficiency in format or per- cent reduction and in modalities of presentation. Efficiency is, of course, defined as a measurement of the interrelationship between comprehension and the rate of information input. In order to attempt to increase information input, the telegraphic concept involves the manipulation of the basic structure of the stimulus. In addition to the pioneering work being done on the telegraphic concept, other research designed toward facilitating information input rates has been focused primarily on time-compressed speech and speed reading. Both of these techniques attempt to achieve greater efficiency by increasing the word per minute processing rate for the presented material. Time- compressed speech achieves the increased efficiency by manipulating the speed (aural) of the stimulus. Speed reading does so by altering 16 the individual's response rate (visual) to the stimulus. Reviews of pertinent literature regarding these two methods are presented in the following two sections. Time-Compressed Speech The advent of modern communication media has brought growing awareness of the educational importance of aural communication and listening skills. Only recently, however, has it become possible to gain effective control over rate of communication between speakers and listeners. Methods initially developed at the University of Illinois by Fairbanks, Everitt, and Jaeger (1954) have made it possi- ble to increase the word rate of recorded speech without affecting intelligibility through distortion of pitch or voice quality. Sig- nificant research interest in speech rate has been generated by the availability of the Fairbanks method. Studies by a number of researchers (Orr, 1966; Sticht, 1968; Foulke, 1970) have shown experi- mentally that speech rate can be increased, without significant loss in comprehension, to about 275 wpm as compared to the average speaking rate of 125-175 wpm. Great interest by educators and researchers in rate-controlled and time-compressed speech has developed. As a result, the Center for Rate-Controlled Recordings was established at the University of Louisville following the first Louisville Conference on Time-Compressed SPeech in 1966. Two volumes produced from the 1966 and 1969 Louisville Conferences respectively, are essential sources of infor- mation to the total research literature on time-compressed speech (1967, 1971)- A three volume anthology covering the entire history, 17 methods, research, and applications of compressed speech, edited by Duker is to be published in early 1974 (in press). In a review of the research on time-compressed speech, Foulke and Sticht (1969) explain the technical procedure of producing accel- erated speech tapes: The electromechanical process for accelerating speech is analogous to cutting out and discarding small, periodic samples of tape and splicing the remainder together to form a continuous tape. This process depends upon the fact that the duration of most speech elements is greater than actually needed for perception of the speech sounds. Due to this temporal redundancy, a considerable portion of a word may be deleted without totally impairing its intelligibility. Because the acceleration process reduces the amount of time required to present a message, the message is said to be time-compressed. It really should not be too surprising that faster than normal speech rates are possible to comprehend since it has long been known that the human mind is capable of processing information at a more rapid rate than the human speech organs are capable of producing. According to Nichols (1972), one reason for poor listening is that, paradoxically, individuals think too fast. His work notes that although the average person talks at a speed of 125 wpm, he thinks at up to four times that rate. This fact can be used to explain the “mind-wandering" phenomenon, described by Cohen (1956), which - constitutes one of the significant problems in effective and efficient listening comprehension. Although it has been established that time-compressed speech is an exciting and workable learning technique, many controversies exist regarding the effects of training and practice, the limits to which speech can be compressed for effective learning, and its use with other modes of learning. 18 Many researchers have shown that compression is effective for information processing until the normal speech rate is doubled. Be- yond 50% compression (275-300 wpm), however, comprehension suffers severely (Heise, 1971). In terms of learning efficiency, however, even though higher comprehension scores may be made at slower wpm rates, more efficient learning may take place at faster wpm rates (Noodcock, 1971). The two factors which determine the efficiency of compressed speech are intelligibility and comprehension. Foulke and Sticht (1967) studied the effect of compression on listening com- prehension and word intelligibility. It was found that both com- prehension and intelligibility decreased as the compression was increased but that comprehension declined significantly more rapidly than intelligibility. Sticht (1970) used 280 army inductees in a study designed to evaluate the relationship of mental ability and the capacity to comprehend compressed speech tapes of varying wpm speeds. Sticht concluded that "the comprehension of fast rates of speech appears to be a function of the information capacities of the listener." This investigation, as well as a number of other studies, has established that the processing capacity of the in- dividual learner and not the fidelity of the time compressed stimulus determines the ceiling or comprehension of fast rates of speech. The problem of increasing information input beyond this ceil- ing is thus due primarily to human, rather than equipment, short- comings. As Carver (1973) noted in a very recent article, compressed speech would be most beneficial and efficient when and if it becomes possible and practical for the individual to be able to control the input rate. Carver, who has also completed considerable research 19 on reading, even suggests that "further research on time-compressed speech may be paving the way toward the day when technical advances in the auditory presentation of verbal information make reading, and the teaching of reading obsolete." The nature of the relationship between the rate of compressed speech and comprehension is critical from both the theoretical and practical standpoint. Some amount of processing time, which must vary according to individual ability, is necessary for encoding transduction of the material. In normal speech (125-175 wpm) there is more than enough time to perfect all of the necessary operations on all of this incoming material to make it fully understandable (Overmann, 1971). Often there is extra, unnecessary time which causes the individual's attention to wander and lose information as a result. As the wpm rate is increased, however, a rate is ulti- mately reached at which there is no surplus time in which the in- dividual can perform the needed storage operations and comprehension then declines rapidly. Miller (1953, 1956) used a concept of a com- munication channel with finite capacity to explain the input threshold. If more information is presented than an individual can absorb, some of the information will be lost. Miller (1953, 1956) emphasized that compressed speech complicates the transduction process all the more. In the language of the computer storage model, fewer cues are present in the compressed word to assist the recognition process which de- mands that more items in the individual's store of vocabulary must be rejected before the correct match is discovered. The compression process essentially reduces the redundancy of the individual words in the message that is to be comprehended. Thus, as available 20 processing time decreases with increasing word rate, the growth in word uncertainty increases the demand for processing time. This ex- planation that decreased redundancy creates confusion and results in additional required processing time should be equally as applicable to telegraphic prose as to compressed speech. In telegraphic prose the visual rather than the aural stimulus is altered which reduces the redundancy and therefore, the repertoire of cues present to aid in both the word identification and the comprehension process. After the capacity of a channel is reached, comprehension should begin to decline, and the slope of this line should become gradually steeper as the rate of compression (or the level of deletion in telegraphic prose if a constant input rate were maintained) is increased. Foulke (1968) reviewed the compressed speech literature and found that the evidence offers tentative support for this hypothesis. In a study using 140 undergraduate students, Overmann (1971) attempted to further investigate the question of comprehension de- crement occurring from lack of enough processing time. An experiment was performed using both time-compressedtapes and tapes in which the compressed sentences were followed by pause time (which restored the tape time to the original normal speech production time). The author found that this additional processing time resulted in significant improvement in comprehension for the 250 and 325 wpm levels, but at the fastest 400 rate comprehension still remained below that of the contrbl group which listened to the same material at a normal, un- compressed rate. Most of the research in compressed speech has been completed with subject populations who have had no previous training in 21 listening to compressed speech. Since any exposure to speech rates of 200 wpm or over is rare in normal speaking environments, it would seem logical that practice or training sessions in listening to varying compression rates should result in improving a listener's ability to comprehend compressed materials. Studies investigating the effects of practice, however, have resulted in conflicting findings. Blind students, of course, depend almost solely on the aural channel for the processing of information so it would be expected that their capacity for aural input rates would be higher than that of most sighted normal learners. Blind students, for example, are known to have trained themselves to listen with good comprehension to normally recorded tapes or records played at a faster than recorded rate which to the untrained listener would sound like Donald Duck. Foulke (1964), however, evaluated the efficiency of four different training methods with blind subjects and found that none of the four yielded any significant improvement in the comprehension of compressed speech. Using sighted subjects Voor and Miller (1965) found that practice improved comprehension in the initial warm-up stage of adjust- ment to listening to compressed speech but not thereafter. In a study using 700 students including mentally retarded and culturally deprived in grade levels from three to six, Noodcock (1971) found that after the initial two or three exposures to compressed speech recordings, continued practice produced little improvement in performance. Orr and Friedman (1967), by contrast, found a significant practice effect on the compre- hension of compressed speech in work done with normal learners at the American Institutes for Research. According to Orr (1971), much of 22 the conflict in research findings on the effects of practice in com- pressed speech may center around the problem of measurement. Some of the shortcomings of measurement which Orr cited and cautioned other researchers to be aware of were the difficulties in constructing valid and reliable multiple choice tests, accounting for prior know- ledge of the listener, and the actual definition of comprehension. Sticht (1971) reviewed all of the research on the effects of listening to compressed materials twice in immediate succession. Although the compressed materials could be presented twice in the same amount of time required to listen to the uncompressed message once, a number of studies have found that, although comprehension improves when the compressed message is repeated, it does not improve over that obtained in a single listening session to the uncompressed messages. Sticht (1971) concluded the following on the basis of his data and that ofothers: It appears as though the technique of trading time for information has not resulted in more information being processed by the listener for short—term retention. Most significantly, this has been true for materials compressed to speech rates of 275-300 wpm for which listening "efficiency", that is, the amount learned per unit of listening times has actually been higher than obtained with "normal" materials. Thus, the implication that, because of improved listening efficiency more information can be learned in a unit of time with moderate compression, has yet to be substantiated. Sticht (1971) cautioned, however, that all of the studies on which his conclusion was based used the time saved only to immed- iately repeat, in a single setting, the compressed message. Thus, a short break or even substantial delay in repeating the material might be more beneficial. Also, the savings of time could be used by the listener to learn something entirely different or by the 23 instructor to teach another group of listeners. A number of studies in the compressed speech literature have investigated the effects of adding a visual stimulus to the compressed aural message. Noodcock (1971), in one of his interrelated com- pressed speech studies, found that listening plus viewing of slides was a more effective and efficient medium for learning than listening alone, and in this elementary school level population which included mental retardates, listening alone was better than reading as many of the subjects had not yet become good readers. Travers (1964) presented passages through compressed speech listening alone, reading alone, and combined listening and reading. At the lower com- pression rates no advantage was found for the audio-visual presenta- tion, but at higher speeds the two combined modalities proved to be superior. Parker (1971) sought to find an efficient method of learning for junior college students of varying aptitudes. 0n tasks involving both listening alone to compressed speech of three dif- ferent speeds and simultaneously listening and reading, Parker found that combined listening plus reading resulted in significantly better comprehension for all aptitude levels hearing compressed speech. For the groups hearing normal rate recordings, the combined audio- visual presentation was superior for those of average aptitude, but not for high or low aptitude students. Except for the groups with low aptitude, students' comprehension was maintained without sig- nificant loss up to the one-half or 50 percent compression rate for the simultaneous listening plus reading presentations. Comprehension decreased significantly at 50% compression (275-300 wpm) for the listening only treatment which led Parker (1971) to conclude that the 24 combined modalities should be used if the compression rate is 50% or above. Walker (1971) designed a study using compressed speech as one of two techniques tested to attempt to increase reading rates in 77 gifted sixth-grade students. All of the students had obtained a score of 130 or more on a Neschler Intelligence Scale for Children (NISC) or on a Stanford-Binet in order to qualify for the special class. Nalker noted that "these students will most likely continue their formal education after high school and will probably have the greatest need to increase their rate of reading because of the de- mands placed upon them." On the basis of sex and a pretest of reading rate the students were assigned randomly to a compressed speech, self-improvement, or control group. The experimental treat- ments were taught and practiced daily for six weeks.. The compressed speech group began practice at a compression level assigned according to the pretest reading rate. Compression levels were increased during the six week period in which the students simultaneously read and listened daily. Results indicated that compressed speech was valuable in helping children increase their reading rate and this was the first known study which individualized the tapes in accordance with the initial reading rates. The major caution was the extreme expense of providing compressed speech materials in a public school setting. Significantly more research has been completed on compressed speech than on telegraphic prose. Although much interest has been generated in compressed speech and it has been established as a 25 feasible learning technique, the preceding review has illustrated that a myriad of research questions regarding it remain unanswered or in conflict. Listening and Reading Comprehension As the foundation of the compressed speech and telegraphic prose techniques for the processing of information, listening and reading comprehension have been basic to the literature reviews presented in the two preceding sections of this chapter. An additional related review and summary of the literature on the relationship of listening and reading will complete the necessary background of knowledge pertinent to this study. A number of excellent evaluations of research are available on the relationship between reading and listening (Berg, 1955; Nitty and Sizemore, 1958, 1959). It is agreed that reading and listening are closely related receptive communication skills which are related to such factors as intelligence and age. Because of measurement and definition problems, however, many contradictions continue in the research despite a plethora of investigations. An annotated bibliography on listening by Duker (1964) notes 880 articles on listening and many hundreds more have appeared since that time. Although listening is far less emphasized at all levels of education than reading, writing, and speaking, it is the skill used to greater extent than any of the other communication skills. Markgraf (1957) and Nitty (1966) have demonstrated that well over 50% of the time that a child is awake is devoted to listening activity of $0116 kind: 26 The answer concerning the relative effectiveness of reading and listening for learning purposes has been sought in many studies for almost a century. Seymour (1965) found that college freshmen enrolled in a one semester course in listening showed a significant increase in reading ability. Other studies have shown no effects on one modality by training in the other. Much of the literature stresses the advantages of reading or the visual modality over listening or the aural modality. Foulke (1969) summarized these advantages by noting-that the visUal reader can control his own rate, can preview or retrace with no difficulty, can scan and skim; and, uses the whole organizational structure of sentences, paragraphs, and headings to assist comprehension. The listener, or a blind student who is accomplishing the process of reading by listening can do none of the above through the auditory channel. A blind indivi- dual who must read by listening is dependent on the oral reader's rate, pitch, pause durations and emphasis, for cues for understanding the message. It is widely accepted that efficient readers listen to or hear at either a conscious or subconscious level the material being read visually. This appears to be an important part of processing and storing the information read. The listening modality lacks this distinctive advantage as the listener does not receive comparable visual cues. Staats' (1968) work in reading indicates that information storage is increased if, after reading, the material is repeated by the reader to himself either vocally or subvocally. Vernon (1937) stressed that 27 in reading "some type of auditory or vocal process is always reported." Although reading courses generally attempt to break the subvocalization barrier, there is good evidence that even fluent readers use silent speech (or in a sense listening to themselves) especially when dealing with difficult text. The view that silent reading is a motor-based speech activity appears to be one of credence (Edfeldt, 1960). No means to allow the listener to control and change at will the auditory input rate is yet available. That good readers should be flexible in adjusting reading rate according to the difficulty of the material and the reading purpose, however, has long been accepted and emphasized in the field of reading. Yet remarkably little evidence exists to support that readers actually do alter speed in relation to reading purposes and materials. A number of studies have shown that even fluent readers alter reading speeds only to a small degree and tend to be rather inflexible (Levin, 1968; Rankin, 1970; Rankin and Hess, 1970). A recent study by McConkie and others (1973) increased the flexibility of university level readers through monetary rewards based on reading time and comprehension on texts of varying difficulty. Redundancy was discussed as an important factor in both listening and reading in the preceding reviews of telegraphic prose and com- pressed speech. The psycholinguistic approach to reading provides additional support for the essential role that redundancy has in aiding the fluent reader to identify essential features of letters and words and in facilitating comprehension (Smith, 1971). Another controversy exists regarding the benefits of simultan- eous listening and reading. Nhile many different kinds of research 28 studies have found that comprehension is increased by the joint use of the two modalities, some researchers believe that this is caused only because the individual can then select and rely upon the one modality to which he is most adapted. Broadbent (1958) and Jester (1966) contend that only a single modality channel, and not two, can be used efficiently at any given time. The final relationship between listening and reading that re- quires consideration is that of rate or capacity of input speed. As noted in the review of compressed speech, comprehension falls rapidly above rates of 250-300 wpm. Despite claims to the contrary, the average university student does not read over 300 wpm (Kolers, 1972) although many readers can, with good comprehension, read at 400 to 500 wpm. The added cues and advantages of the visual mode evidently allow greater capacity processing levels. Miller (1956) explains this by the "chunking" concept which is based on the reader dealing with words in units or chunks rather than singly. Since a listener is much less able to do this, the upper limits on listening rate may be significantly less flexible than those for reading. Nhile reading and listening rates can both be increased according to the individual capacities of the learner, the claims of some of the high speed reading advocates appear outrageous in the face of know- ledge about human information processing capabilities. Speed reading is very likely predicated on the same principle as the telegraphic concept of cueing in on key words of high informational value. Despite claims by Evelyn Nood (Gallup, 1964) and other advocates of high speed reading that word per minute rates of 1,000 and above without 29 significant loss of comprehension are entirely reasonable, recent research presents substantial findings to refute these. Most reading researchers agree that rates above 800-l,000 are physiologically im- possible. Taylor (1962) found that the average individual reading rate increased from 80 wpm in first grade to 280 wpm in college. Depending on ability and the difficulty of the material much reading research supports that 150-300 wpm rates are the average for readers. After studying reading research and conducting speed reading experi- ments Carver (1972) has concluded that speed reading is "5% sense and 95% nonsense." His evidence suggests that increasing reading speeds adversely affects comprehension. Nhen the limits of speed are approached he has documented a decrement in quality as well as quantity of information input. One very recent experiment completed by Carver (1972) showed a test of comprehension used for a speed reading course was insensitive to major losses in comprehension. In the research, one group of students took the comprehension test over the material without previously reading the selection. Average scores were 57% comprehension for this group while comprehension scores for the group of speed readers who had read the selection averaged at 68%. Such findings suggest significant fallacy in the means commonly used by speed reading advocates to measure comprehension or effective wpm reading rate and the need for careful analysis of design and measurement in experiments dealing with information input. A current advertisement on national television for the Evelyn Nood Reading Dynamics program features a twelve year old boy posed with a book saying, "I read about 4,000 words per minute. I read this book in less than an hour and I read every word and every comma." 30 By contrast, a recent report on reading from the National Assessment of Educational Progress project evaluated reading rate and compre- hension of 7,850 readers at four age levels by region, sex, race, parental education, and community size and type. Only 17 of these readers read faster than 750 wpm while 149 read less than 50 wpm. Average rates were 100-299 wpm. Both high speed reading and compressed speech of too high a degree fail to consider the significance of the internal sampling and reconstruction process a good reader/listener goes through to in- put and preserve coherently the message he is receiving. Kolers (1972) has done recent studies involving unusual presentations on printed letters, words, and sentences. His findings show that a good reader proceeds by generating internal grammatical messages based on clues in the text and on a kind of reconstruction process that is similar to what one does in speaking. Kolers' (1972) work emphasizes the continuing investigative challenge of unanswered questions in the way humans process information. Hypotheses The primary purpose of the present Study was to determine the effects of telegraphic prose, compressed speech, and modality upon learning efficiency and comprehension. The efficiency measure was designed for this particular study and was calculated as a function of total number of words in the traditional story version divided by the time required for task completion multiplied by the percent of comprehension (gégggngogds x % correct = ENC). ThlS measure was tape time 31 termed effective word per minute comprehension rate to be designated as ENC. The rationale for the objective of this ENC measure is explained as follows: The experimental alteration of the stimulus material (traditional written prose version) explores the possibility of in- creasing the student's learning efficiency relative to time. It is postulated that some combination of compression (Presentation Rate) and word reduction (Deletion Level) will enable students to compre- hend the information contained in the traditional version in a shorter unit of time. Thus, a means to measure this must not only consider time to task completion and comprehension but amount of information processed as extrapolated back to the traditional version. Effective word per minute comprehension (ENC) will be used to denote the word per minute rate of comprehension under experimental conditions and is to be differentiated from word per minute reading rate which refers only to the actual wpm rate based on time required for task com- pletion without taking comprehension into account. Example: A 1000 word traditional prose story is reduced to 750 highly informational words. It is pre- sented aurally at 300 wpm taking a time unit of 2.5 minutes (750wpm e 300 wpm). The subject (S) scores 80% comprehension on the test over the essential information. His effective word per minute comprehension would be: mat; x = Since no previous research has investigated either the combined inputs of telegraphic style prose and time-compressed speech or the effects of visual, aural, and aural/visual modalities upon this 32 combination of learning techniques, the present study was largely exploratory in nature. The lack of comparative data limited the predictions that reasonably could be made. On the basis of other research work completed in the separate techniques of telegraphic prose and compressed speech and on studies of listening and reading comprehension as well, the following list notes the exploratory hypotheses which were advanced and the variables that were investi- gated. l. A differential drop off in comprehension will occur as a function of increasing the Presentation Rate (com- pressed speech tapes) and the Level of Deletion or compaction (telegraphic prose). The effective word per minute comprehension rate (ENC) will be highest at the 275 wpm presentation rate for the Traditional and 20% deleted material. ENC will fall on the 40 and 60% reduced material at 275 wpm and for all story versions at the fastest 400 wpm rates. Presentation of the material at the highest reduction level and fastest presentation rate will result in significant decrements in comprehension. It is assumed that at these levels the information input rate will exceed the processing capacity of the learner. The combined Listening/Reading Modalities will enhance comprehension compared with the single listening modality or reading modality. If so, whether this is a result of the efficient use of both modalities or the selective use of one will be studied. Reliance by the majority of §s on the Reading Mddality alone is predicted for the 40 and 60% telegraphic versions at the 275 and 400 me rates for the combined Listening/Reading Modalities. The aural tapes will be extremely difficult to comprehend for these treatments. 33 An additional exploratory aspect of the study was the comparison Of all experimental treatments to determine the most ideal combina- tions of telegraphic prose, compressed speech, and modality for learning efficiency. Summary The need for more knowledge regarding learning efficiency and the processing of information is great. Learning techniques such as time-compressed speech and telegraphic prose offer innovative means to increase learning efficiency but the contingent variables, relationship of communication channels or modalities, and limita- tions of information input and storage present unresolved complexi- ties and controversies which require investigation. The specific research problem and the pertinent areas of related research have been the subject of this chapter. The methods and procedures of investigation used in the research design are reported in Chapter II. The statistical results of the experimental treatment conditions are analysed in Chapter III while the discussion of the results and general conclusions are presented in Chapters IV and V, respectively. CHAPTER II METHODS AND PROCEDURES Subjects The Ss for this study were 560 undergraduate and graduate students in the Schools of Education, Psychology, and Architecture at the Uni- versity of Miami. A total of 480 §s were assigned randomly to each of the 24 Listening and Listening/Reading treatments with 20 §s in each group. Table 2.1 presents the number of males, females, graduates, major fields of study, and minority populations in each of these 20 treatment conditions as a result of random assignment. All males, graduates, and §s with Spanish as a first language were randomized as separate populations. Participation was a class requirement for approx- imately one-half of these students and was on a volunteer basis for the other 50%. All of these S; were given appointment hours outside of class time for this project. The additional 80 §s for the four control Reading treatments were enrolled in four sections of a developmental reading course in the Department of Educational Psychology. On the basis of their individual scores on the Nelson-Denny Reading Test, the students en- rolled in this course were judged to be a representative sample of the undergraduate population. The developmental reading classes included students in class rank from freshman to senior whose purpose in 34 35 TABLE 2.1 Number of Male, Female, Graduate, Cuban, Black, Physical Education and Architecture §s Randomly Assigned to the Twenty—Four Treatment Conditions* Treatment Conditions Total Grad Cuban (Span.) Black Phys Ed Arch Listening F M F M F M F M‘ F M F M 175 T 14 6 l 4 1 l 3 20 14 6 l 2 l 1 4O 15 5 2 l 1 2 3 60 16 4 l 2 2 1 275 T 15 5 2 l 1 20 16 4 4 l l 1 4O 14 6 l 2 1 6O 13 7 2 1 1 l 3 400 T 14 6 l 2 l O O l 20 13 7 3 l l l O O 2 4O 14 6 2 l l l 60 12 8 l l l 3 Listening/Reading 175 T 15 5 1 l 1 2 20 14 6 l 4 1 3 4O 15 5 2 O 5 2 2 6O 15 5 O 1 l l 275 T 16 4 l O l 1 2 20 15 5 l 1 O O l 1 4O 15 5 l O l O 1 6O 14 6 1 O 3 1 l 2 400 T 16 4 2 l l l 20 14 6 2 l l 1 2 4O 14 6 2 3 l l l 2 60 ll 9 l l 2 l l 5 Total 344 136 35 10 29 10 16 4 4 21 2 27 *The remaining §s in each group were undergraduate Psychology or Education majors. 36 taking the course ranged from specific remediation in reading vocabu- lary or study skills to development of increased rate and flexibility in preparation for graduate school admission. Of these 80 control Ss, 59% were male as compared to 28% males in the 480 S; for the experimental treatment condflions. Participation in this study was a class requirement and was accomplished during the regular class period in each of the four sections of the course. Materials Basic Structure of the Four Story Versions Traditional Story The basic material for all 28 treatments was a fictional prose composition authored in 1972 by a technical writer specifically for use by the Texas A&M Research Foundation. This story was created as part of a federally funded (Bureau for the Education of the Handi- capped) project directed by Dr. Clessen J. Martin and entitled De- velopment of an Analytical Approach to Telegraphic Communication. Criteria followed in creating the 2,692 word Traditional (T) ver- sion of "San Francisco" and the test to accompany it were: (1) the story should be of sufficient interest to college undergraduate students to promote positive motivation among the §s in the testing situation; (2) the reading level of the prose material should be of appropriate difficulty for college students; (3) the test questions assessing students' comprehension should have high reliability; (4) the length of the story and its comprehension test should be designed to allow all §s to complete them comfortably in one 50 37 aminute class period. The story "San Francisco" focuses on the impact of a devastating earthquake in the major urban area of San Francisco Bay. Work of the city and state officials interacting in the after- math of the destruction plus the evacuation and rescue work provided by a commune group which had previously borne the resentment of the local police are the central events depicted. The story was written to be analyzable by Dawes' (1964) set relations model. The full text of the T version appears in Appendix A. Dawes' Set Relations Model As noted in Chapter I, the major feature of this model is the structure of artificially constructed meaningful material into set relationships to make it possible to keep the amount of essential in- formation constant among different versions of the specially con- structed passages. To meet the criteria of the Dawes' model, the changing relationships among the groups throughout the course of events were systematically maintained in all versions of "San Francisco." As shown in Figures 2.1 and 2.2, the essential structure of "San Francisco" is illustrated by the use of Venn diagrams. Telegraphic Versions Telegraphic forms of the story were generated from the 1 version of "San Francisco" using the SHORT (Subjective Hierarchy of Relevant Terms) method of creating telegraphic materials developed by Pantalion (1972). The 151 sentences in the T version of the story were divided into 24 sentence sets of six or seven sentences each. Two hundred and fifty-six freshman Urban Problems Committee sueostoueug ues Area surrounded by Dashes = leadership of Mayor St. John Shaded Area = Action Group Nested Relations All members of the Urban Problems Committee were senators All city councilmen were San Franciscans All San Franciscans were under the leadership of Mayor St. John All city councilmen were under the leadership of Mayor St. John All of the Action Group were under the leadership of Mayor St. John Disjunctive Relations Some Senators were under the leadership of Mayor St. John Some members of the Urban Problems Committee were under the leadership of Mayor St. John Some Senators were in the Action group Some members of the Urban Problems Committee were in the Action Group Some San Franciscans were in the Action Group Some city councilmen were in the Action Group Figure 2.1. Set relations existing before the earthquake as described in all four treatment versions of "San Francisco." . 39 Disaster Committee SUPOSIDUEJJ ue “Area surrounded by Dashes = followers of Atkins Shaded Area = Volunteers for Hope Nested Relations All of the Disaster Committee members were senators All officials of the Interim City Government were San Franciscans All San Franciscans were followers of Atkins All officials of the Interim City Government were followers of Atkins All Volunteers for Hope were followers of Atkins Disjunctive Relations Some of the senators were followers of Atkins Some of the Disaster Committee members were followers of Atkins Some of the senators belonged to the Volunteers of Hope Some of the Disaster Committee members belonged to the Volunteers for Hope Some of the San Franciscans belonged to the Volunteers for Hope Some of the officials of the Interim City Government belonged to the Volunteers for Hope Figure 2.2. Set relations existing after the earthquake as described in all four treatment versions of "San Francisco." 4O psychology students at Texas A&M University served as judges. Each judge was asked to read the entire story and then to rank order the words in each of the sentences in his assigned sentence set. Judges rank ordered words from least to most important on the basis of each word's importance in conveying the essential idea communicated in the original sentence. The rank orderings were keypunched on IBM cards for analysis by the SHORT computer program. The computer printout gave the mean rank of each word in each sentence based on the rank orderings of all judges. A Kendall Coefficient of Concordance was computed for the rank orderings of each sentence. Using the results of the Kendall coefficient, a Chi-square and a pfvalue were also obtained. For each sentence, the obtained prvalue was 25.001 which revealed a significant degree of agreement among judges in their rank orderings. The SHORT method generated telegraphic versions by deleting 10, 20, 30, 40, 50, and 60% of the words judged to be least important. The passages were then typed maintaining original sentence and para- graph form. Nhen the first word of a sentence was deleted, the next remaining word was capitalized. All versions contained the same essential information. The same sequence of events was maintained and the same persons and groups of persons were identified in all versions. The 20, 40, and 60% deleted versions were selected for use with the traditional version in this study. The full texts of the three deleted story versions appear in Appendices B, C, and D. 41 Grammar, Punctuation, and Style Changes in the Deletion Process The four versions of the story contain words by grammatical cate- gory as shown in Table 2.2. TABLE 2.2 Number of Words in Each Grammatical Category in the Four Versions of San Francisco Grammatical Number of Nords Category T 20 4O 60 Nouns 644 633 582 483 Pronouns 200 176 121 56 Verbs 533 498 434 324 Adjectives 324 274 173 82 Adverbs 232 157 89 50 Prepositions 349 265 167 66 Articles 280 7O 14 4 Conjunctions 130 85 37 10 The flowing narrative style of the traditional passage is rich in descriptive words and phraseology. The low telegraphic (20% deletion) maintains the same basic style and sentence structure. The medium telegraphic (40% deletion) is less fluid because of the omission of phrases and connecting words. The high telegraphic (60% deletion) is extremely compact and eliminates all extraneous content. Nouns, pronouns, and verbs were eliminated less frequently than words of other grammatical categories in the deleted versions. are: 42 Traditional In June that year, the first earthquake came. More than halfway up the Richter scale, it was strong enough to cause a pretty solid loss of life in the old buildings, those relics of rebuilding in 1906 that had somehow weathered urban renewal and freeway right-of—way acquisition. The Mayor hopped into his helicopter to the flight to Sacramento to seek state help. (62 words) Low Telegraphic - 20% Deletion June year, the first earthquake came. Halfway up Richter scale, it was strong enough to cause loss of life in old buildings, those relics of rebuilding of 1906 that had somehow weathered urban renewal, freeway right-of-way acquisition. The Mayor hopped into helicopter flight to Sacramento to seek state help. (49 words) Medium Telegraphic - 40% Deletion June year, first earthquake came. Halfway up Richter scale, it was strong enough to cause loss of life in build- ings, relics of 1906 weathered urban renewal, freeway acquisition. Mayor hopped into helicopter to Sacramento to seek help. (37 words) High Telegraphic - 60% Deletion First earthquake came. Halfway up Richter scale strong enough cause loss of life in buildings, relics weathered renewal. Mayor hopped helicopter Sacramento to seek help. (25 words) The total word counts for the four versions of "San Francisco 1. Traditional, uncondensed form 2692 words 2. Twenty percent deletion (Low-Tel) 2158 words 3. Forty percent deletion (Med-Tel) 1617 words 4. Sixty percent deletion (High-Tel) 1075 words 43 Calculation of Readability Level The readability level of the T version of "San Francisco" was calculated by Flesch's (1949) formula which was develOped for the evaluation of upper-grade and adult materials. The Flesch formula has been used in determining the reading level of difficulty of materials for use by the United Nations and in the development and use of technical service manuals for the Armed Forces. It also has been used widely in the evaluation of textbooks for secondary school use. The steps involved in using the formula require taking 100- word samples from each 500 words of text. The number of sentences in each sample is noted and the total number is divided into 100 to obtain the average sentence length. A sentence portion at the end of the sample that contains less than half of its total words within the sample is not counted. A syllable count is then made for the sample and the total number of syllables in the sample is located on the Flesch Formula Chart. The same procedure is followed for the sentence length scale and an intersecting line is plotted. The Reading Ease Score for each sample is that level on the center column of the chart where the line crosses. Both a numerical score and a general categorical rating of reading difficulty can be ob- tained. The Flesch Formula scores, descriptive categories, and the grade range equivalents are presented in Table 2.3 while Table 2.4 gives the Flesch Readability scores for "San Francisco." Although Flesch Readability scores were not calculated for the Flesch's Reading Ease Categories 44 TABLE 2.3 Score Descriptive Category Grade Range 90 to 100 Very Easy 5th Grade 80 to 90 Easy 6th Grade 70 to 80 Fairly Easy 7th Grade 60 to 70 Standard 8th & 9th Grades 50 to 60 Fairly Difficult 10th through 12th Grades 30 to 50 Difficult College Undergraduates O to 30 Very Difficult College Graduates TABLE 2.4 Flesch Readability Scores for "San Francisco" Traditional Version of 2692 Nords Sample Nord Count Sentences Nords Per Syllables Formula Number In Passage Per Sample Sentence Per Sample Score 1 112-211 5 20 151 59 2 553-652 8 12.5 144 72 3 1277-1376 6 16.6 154 60 4 1924-2023 5 20 168 45 5 2205—2304 6 16.6 159 56 Average 58 (Fairly Difficult) 45 telegraphic versions, the level of difficulty definitely increases with each higher percentage of deletion. Even though sentence length decreased in telegraphic forms, the word reduction process omitted significantly more one syllable connecting and modifying words than the generally more meaningful multi-syllabic words in the telegraphic versions as compared to the traditional. ReadingiMaterial Format The four versions of "San Francisco" were typed in double space on white 8 1/2" x 11" paper (as shown in Appendices, A, B, C, and D) for use by the 320 §s who were assigned to a treatment condition which required reading. Time-Compressed Tape Recordings The four versions of "San Francisco“ were sent in written form to the Center for Rate-Controlled Recordings at the University of Louisville. Each of the story versions was recorded by a male reader at his natural speaking rate. Information on these original record- ings is shown in Table 2.5. The highly compact format of the 40 and 60 versions appears to have caused the reader to automatically decrease his rate of speaking. As noted in Table 2.5, a second re- cording of the 60 (High—Tel) version had to be made because the original one was too slow (105 wpm) to be time-compressed smoothly to the fastest (400 wpm) rate which this project required. Each of these four original recordings was then time-compressed to 175, 275, and 400 word per minute (wpm) rates. A discard 46 interval of 20 milliseconds which is the optimum level of discard for intelligibility was used for all of the compressions. The recordings were open-reel recorded on separate one-half inch width tapes at seven and one-half inches per second which resulted in 12 individual tapes (four story versions x three presentation rates). Exact wpm rate accuracy plus or minus 5% is guaranteed by the University of Louisville's Center for Rate-Controlled Recordings. Table 2.6 presents the specific data on the time- compressed tapes. TABLE 2.5 Information on the Original Recordings of "San Francisco" Version No. of Nords Tape Time NPM Rate T 2692 16.25 165.7 20 2158 14.75 146.3 40 1617 13.60 118.9 60 1075 3.15* 131.9 * Second recording Evaluation Material Multiple Choice Items Sixty multiple choice questions with four alternatives for each item were constructed from the factual material available in the 60% deleted version of "San Francisco." .After the questions were constructed, they were all checked against the T, 20, and 47 TABLE 2.6 Data on the Time-Compressed Speech Tapes for "San Francisco" T/2692 wds 20/2158 wds 40/1617 wds 60/1075 wds 175 NPM *Perfect Tape Time 15.38 12.33 9.34 6.14 Actual Tape Time 15.58 12.75 9.75 6.55 Actual NPM 173 169 166 164 *Ewc 173 211 276 411 275 NPM *Perfect Tape Time 9.78 7.85 5.88 3.90 Actual Tape Time 9.93 8.23 6.23 4.23 Actual NPM 271 262 260 265 *Ewc 271 327 432 636 400 NPM *Perfect Tape Time 6.73 5.40 4.04 2.69 Actual Tape Time 6.70 5.58 4.24 2.87 Actual NPM 402 387 381 375 *Ewc 402 482 635 1 938 *Perfect Tape Time = Errorless time-compressed recording such T 2692 wds _ as W -15.38 2692 wds of T version * . = Effective NPM Comprehen510n Tape Time 48 4O deleted versions to ascertain that no conflicting detail was present which might cause confusion due to the greater amount of de- scriptive material in these versions. These items assessed mainly the recall of factual information and it was assumed that this method of measuring comprehension approximates typical methods used in the classroom. These 60 items followed the sequence of the story in their order of presentation in the text. Set Relations Items Twenty questions were constructed to test comprehension of the two types of set relations, nested and disjunctive. The first 10 items tested recall of nested sets before the earthquake occurred and the second 10 items tested comprehension of disjunctive sets of relationships following the earthquake. These questions had two alternatives for each item. As presented in Chapter I, Dawes (1964) found more overgeneral- ization errors as tested in the disjunctive items than pseudo- discrimination errors as tested by the nested items. The explanation which he presented for this was that overgeneralization errors occur as a result of information overload. Since the deleted versions con- tain less extraneous information, Dawes'model predicts proportion? ately fewer overgeneralization errors in the telegraphic forms of the story. It is the extraneous material which increases the amount of total information that may interfere with the recall of the basic set relations found in all versions of the material. Analysis of the set relations recall errors may permit the determination of whether or not telegraphic materials make less imposition upon 49 memory than T prose materials. The 20 set relations items were added in sequence to the 60 multiple choice questions to create an 80-item total test. The total 80-item test is presented in Appendix E. Pilot Test of the Questions The questions were pilot tested on male and female §s enrolled in undergraduate psychology classes at Texas A&M University. All §s read the T passage and then answered all test questions on IBM 503 answer sheets which were machine scored. Kuder-Richardson #20 correlations, point-biserial correlations, and item discrimination values were com- puted. Table 2.7 shows the summary statistics for both parts of the test. TABLE 2.7 Summary Statistics of the Two-Part Test Developed for "San Francisco" Multiple Choice Questions Set Relations Questions K-R#20 0.88 0.45 Mean Point Biserial 0.34 0.27 Mean Item Discrimination 0.27 0.28 Mean Item Difficulty 0.83 0.70 N=49 "=53 Recording and Playback Equipment Three Nollensack tape recorders equipped with listening stations were used for the presentation of the tape recorded material. 50 Each listening station had six jacks for headsets which permitted up to six §s to listen to the same tape simultaneously. The headsets which were used for each §_insured the best intelligibility from the tapes, allowed individual adjustment of the volume, and reduced dis- tractions from the presence of other §s or outside noise. Additional Materials Additional items used were a combined scheduling and self-rating form, IBM answer sheets, a modality preference sheet, a stopwatch, and a cardfile. Prior to the study, §s completed a time preference sheet in order to permit scheduling participation time at their convenience. On this same sheet, §s were asked to rate their own preferred mode of learning, reading rate, and conversational speech rate. This form is shown in Appendix F. The 240 S5 who were assigned to the combined Listening/Reading treatment groups were given a form immediately following the presen- tation task to indicate the modality they believed they relied upon most for comprehending the material (Appendix G). All §s completed the test using IBM 503 answer sheets. For the 30.§$ in the Reading treatment group, a stopwatch was used and the elapsed time was written on the blackboard in five second intervals to enable each §_to record his individual reading time. Except for the 80 §s in the Reading treatment groups, the name, sex, class, first language if other than English, and the randomly assigned treatment group for each §_were written on a 4" x 6" index card. These 480 cards were alphabetically indexed in a cardfile 51 and then refiled by treatment group number (1-24) after the parti- cipation of each S, This facilitated setting up the specific treat- ment condition for each S, kept an exact count of the status of each treatment group, and allowed the recording of Ss' specific comments following the task. Procedure and Design of the Study Listening and Listening/Reading Treatment Conditions The 480 S5 were assigned randomly to one of the 24 treatment conditions. Twelve groups of 20 Ss each were assigned to the Listen- ing treatments and 12 additional groups of 20 Ss each were assigned to the Listening/Reading treatment conditions. Equivalent situations were maintained for all Ss. The basic study was originally explained in each of 30 education and educational psychology classes from which volunteers for the project were sought. Nhen Ss came to participate (by either individual appointment time or a weekly schedule sheet given to each S), specific instructions were given individually or in very small groups. Because most of the Ss were totally unfamiliar with even the concept of compressed speech or telegraphic prose, instructions were modified according to the rand0mly assigned treatment condition. Subjects were given a detailed explana- tion of the level of speech rate and percent of word reduction to which they would be listening or simultaneously listening and reading. In a pre-study presentation of the compressed speech tapes to a number of graduate students who were not in the S population, it 52 became evident that the unique style and speed commanded Ss' atten- tion at least initially and therefore interfered with concentration on the content of the material. This occurred especially on the 400 wpm tapes and the highest deletion levels at any of the three presentation speeds. Thus, the rationale for explaining the assigned word per minute speed and deletion level to each S_was not only an appropriate but necessary procedure. The basic instructional sets for the Listening and Listening/ Reading treatments were as follows: Listenipngreatment Groupg This project is concerned with finding out how well university students can comprehend material presented at faster than average speech rates and with varying per- centages of words deleted. Listen as carefully as possible to the story on the tape. You will be listening at 275 wpm which is twice as fast as normal speech rate for tape recordings. Forty percent of the words have been deleted from the original version so it will sound something like a telegram being read at a very fast rate. (The two previous sentences were modified according to the exact treatment condition.) The volume should be set correctly, but you may adjust it if you wish. The tape will be started as soon as you have the earphones on and are ready. Some students find it helpful to close their eyes or to put their heads down while listening; you may want to try this to maintain maximum concentration. Remove the earphones as soon as the tape ends and you will then be given an 80-item objective test which covers the important facts, events, and names in the story. ListeningLReading,Treatment Groups Instructions were identical with those for the Listening groups except for the following essential changes: . . . You will be listening to the story and reading it at the same time. Therefore read or follow the written text while you listen to the identical selection on the tape. 53 The same words have been omitted from the text as from the tape recording. . . . As soon as you have the earphones on and are ready, the tape will be started and you may turn the story booklet over and begin reading at the same time. You must stop reading as soon as the aural tape is finished. Reading Treatment Conditions The 80 S5 were enrolled in four sections of a developmental reading course. Each S read the story and completed the test dur- ing one 50 minute period in one of the four sections of the class. The four story versions were distributed randomly in the classes. Specific instructions were given to each of the four groups as follows: The purpose of this study is to determine the effects of deleting varying percentages of words from a story on comprehension and reading rate. Twenty students will read the story in its original 2692-word form. Others will read the same typed story in one of three versions from which 20, 40, or 60% of the words have been deleted. You each have on your desk a story booklet and an IBM answer sheet. The exact procedure for the experiment is as follows: Read the story as rapidly as you can without sacrificing comprehension. Those of you who are reading the highly de- leted versions may find the telegraphic style disconcert- ing but read it as rapidly as possible, again without un- due sacrifice of comprehension. As soon as you finish reading raise your hand and look up at the blackboard to note your exact reading time. Your instructor will write the elapsed times on the blackboard at five second intervals as soon as the first student in the class finishes reading. [This is the procedure these classes normally followed in recording reading time.] Record your reading time on the IBM answer sheet which you have. As soon as you have done this you will be given a test booklet. The test has 80 items and takes approximately 25 minutes to complete. You will have ample time to finish it comfortably during this class period. Note that the answer sheets are numbered horizontally across the page. Please put both your name and the version of the story which you will be reading, designated by T, 20, 40, or 60 at the top of your reading booklet, on the IBM answer sheet at 54 this time. You will be given your wpm reading rate, test score, and ranking for the group of 20 students who will be reading each version of the material next week in this class. Are there any questions? Please turn your reading booklet over and begin reading. Additional Procedures Comprehension Test No time limit was imposed for the test which took 25-30 minutes for most students to complete. All tests were hand-scored with a sep- arate items correct score given for multiple choice, nested, disjunc- tive, total set relations, and for the total test. The tests were then filed according to treatment groups. Modality Form and Posttest Comments The 240 S5 assigned to the 12 Listening/Reading treatment groups completed a post-task check form asking which modality they had relied on most to comprehend the material: Aural, Visual, or Aural/Visual equally. This form was attached to the IBM answer sheet (Appendix G). All participating Ss were invited to write comments about the particular treatment conditions on this form or on the back of the answer sheet. Reportipg_Resu1ts to the Subjects In order to enhance motivation and interest in performing well, all S5 in the Listening and Listening/Reading groups were given a postal card along with the test. These 480 S§ were told to self-address the card if they wished to know their individual test score and rank in the treatment group of 20. All Ss except one wanted to receive this infor- mation and self-addressed the card. These results were mailed as soon as the study was completed. 55 For the 80 S5 in the Reading groups, individual cards listing total test score, wpm reading rate, effective wpm comprehension (ENC), and rank within the treatment group of 20 were given to every S, Posttest Opti0nal Participation After completing the test, Ss in the 24 Listening and Listening/ Reading treatment conditions were invited to listen to tape segments from other treatment conditions in order to understand more clearly the combined concepts of time-compressed speech and telegraphic prose. The great majority of the S5 were interested in doing so and many remained to listen to several complete tapes. Design and Analysis FigureZLB presents the research design and legend for the study. The experimental design was composed of two separate 3 x 4 factorial analysis designs and a l x 4 design. The factors in the 3 x 4 designs were three Presentation Rates (175, 275, and 400 wpm) and four Deletion Levels (T, 20, 40, and 60%). The modality was Listening (Aural) for one of these designs and Listening/Reading (Aural/Visual) for the other one. Factors for the 1 x 4 design were Nord Per Minute Reading Rate and the same four Deletion Levels as in the 3 x 4 designs. The study had 28 treatment conditions with N = 20 Ss for each treatment cell. The major analyses of this study involved three sets of data: (1) data from S5 who listened to the prose materials, (2) data from ‘ Ss who listened and read simultaneously, (3) data from Ss who read 56 m E // RT I R201 R40 1 RSO‘I/ 9: / RT2 R202 R402 F3662 g ’ RT4 R204 R404 , ’R604 g? — / / I, ’ 09$ 2 / / l // I- I, z ,/ 3") I75 T1 20. 40. 601 ,x 101:1 275 T2 202 402 602 J O- 400 T4 204 404 604 946“ '1' 3.9.22 m ,6_0__°/g V 5: =20 PER TREATMENT CELL x x x x 28 CELLS 1:20 So DELETION LEVEL 560 roman. e, IIRZO IR60 [L 7 20% f 40% 1 60% 1 x x x x DELETION LEVEL LEGEND 1 EXAMPLE: T=TRADITIONAL 1 = 175 WPM ‘°"‘°°’°.3§",§.I,',°" ‘1 20 =20°/. DELETION 2 = 275 WPM ___&13LE§_"‘E___ 4o =40°/, DELETION 4 = 400 WPM R4°'=”°/°g§l-§g;‘°" AT 60 =60°/. DELETION R = READING usrenms a READING FIGURE 2.3 RESEARCH DESIGN SHOWING TREATMENT CONDITIONS ANALYSED BY THE ANALYSIS OF VARIANCE MODEL. 57 the prose materials. In addition to the separate analyses performed on these three sets of data, a fourth set of analyses compared the performance of Ss who listened with that of Ss who concomitantly read while listening to the prose materials. Analyses of variance were performed on the means of the five dependent comprehension variables. The total scores from the multiple choice, set relations, nested, and disjunctive test items were ana- lyzed. A measure of comprehension efficiency called effective word per minute comprehension (ENC) was computed for each S_by dividing the total number of words (2692) in the Traditional story by the tape or reading time (TT or RT) and multiplying it by the percent correct of the multiple choice test items (C). Thus, X %C. For the Reading treatment groups, word per minute reading rate and reading time were additionally analysed. The Scheffe'post hoc comparison was used to determine the precise nature of the differ- ences among treatment means yielding significant {_values. A Item analyses were completed on both parts of the test for indices of difficulty and discrimination (for S5 in the Reading treatment conditions). The Kuder-Richardson #20 coefficient for test reliability was computed. The data were analysed through the services of the Human Learning Research Laboratory and the Data Processing Center at Texas A&M University. An IBM 360-65 computer was used and the computer program was written in NATFIV Fortran language. CHAPTER III RESULTS The results are organized into four major sections based upon the treatment modalities being analysed. Analyses of variance are re- ported for the three levels of the Presentation Rate main effect, for the four Deletion Levels main effect, and for the Modality main effect. The four major sections are: 1. Reading (Visual) Treatments - l x 4 ANOVA 2. Listening (Aural) Treatments - 3 x 4 ANOVA 3. Listening/Reading (Aural/Visual) Treatments - 3 x 4 ANOVA 4. Listening compared to Listening/Reading Treatments - 3 x 4 x 2 ANOVA Factorial ANOVAs were performed acroSs all treatment conditions for each of the following dependent variables: Multiple Choice Test Score, Set Relations Test Score, Nested Set Relations Score, Dis- junctive Set Relations Score, and Effective Nord Per Minute Comprehen- sion Rate (ENC). Two additional dependent variables, Reading Rate and Reading Time, were analysed for the Reading treatment conditions. As reading is the prime mode of learning in formal education, the four Reading groups were used as control conditions for comparison with the 24 experimental groups in which the mode of learning was Listening or combined Listening/Reading. 58 59 Table 3.1 presents the item analysis of the 60-item multiple choice test and the 20-item set relations test. The 80 tests from the four Reading treatments were selected for this analysis because reading is the modality on which most comprehension tests are based. Previous item analyses of this test have been completed on comparable S popu- lations at Texas A&M University (Table 2.7). Twenty different Ss read each of the four versions of the prose material. Mean compre- hension scores for the total 80 S; were 64 and 62%, respectively, for the multiple choice and set relations tests. TABLE 3.1 Item Analysis of the Total Multiple Choice Items (60)and Set Relations Items (20) for the Reading Treatments Item Statistics Multiple Choice Set Relations Mean Score for test 38.41 12.29 Standard Deviation 9.06 2.55 Mean item discrimination 0.37 0.29 Mean point biserial correlation 90.34 0.28 ‘ Kuder-Richardson #20 0.87 0.42 Standard error of measurement 3.26 1 1.95 Ebel (1965) states that the reliability of an item increases as the discrimination index reaches .40 and above. Thus, he recommends that items have a minimal discrimination value of .30. The .37 mean discrimination value for the multiple choice test was above this 60 while the .29 discrimination value of the set relations test approached the recommended value within acceptable limits. The mean point biserial correlation measures the relationship be- tween performance on the items and total test score. The Kuder- Richardson No. 20 (K-R#20) coefficient indicated a high degree of re- liability for the multiple choice items. Although the corresponding reliability coefficient for the set relations items was considerably lower, it is suggested that the relatively small number of items in the set relations test and the apparent difficulty of these 20 items may have been the two significant factors depressing the correlation. As shown in Table 3.1, the mean number of total correct responses for the 20-item set relations test was 12.29 with the expected chance score on this test being 10.00. Analysis of the 12.29 mean score indicated that it was not significantly above the chance level. Reading_Treatments The means and standard deviations for reading times, reading rates and ENC are presented in Table 3.2. Tables 3.3 and 3.4 show the significant ANOVAs for reading rate and reading time. Reading times were found to be significantly different [f(3,76) = 5.96, p5.005] among the four groups in the l x 4 ANOVA design. The Scheffe test revealed that reading time was reduced by 2.22 minutes on the 40 as contrasted with the T version. For the 60 material, reading time was not significantly different from the T. Subjects assigned to the 60, containing 1075 words, read it at an average of 8.39 minutes while the T Ss read 2692 words at an average of 9.33 61 TABLE 3.2 Means and Standard Deviations for Reading Time, Reading Rate, and Effective Nord Per Minute Comprehension Rate (ENC) in the Reading Treatments Treatment Reading Time Actual NPM Rate ENC Conditions N Mean SD Mean SD Mean SD Deletion T 20 9.33 1.35 294.81 45.74 235.22 54.52 20 20 7.96 1.96 285.15 62.30 228.92 71.80 40 20 7.10 1.30 236.58 54.45 221.30 68.93 60 20 8.39 2.02 135.86 35.08 191.05 52.33 TABLE 3.3 Results of l x 4 ANOVA for Reading Rate in the Reading Treatments Source df MS E_ p_ Groups 3 105904.60 41.68 <.001 Error 76 2541.05 - - Total 79 6466.25 - 62 TABLE 3.4 Results of 1 x 4 ANOVA for Reading Time in the Reading Treatments Source df MS F_ p_ Groups 3 17.07 5.96 <.005 Error 76 2.86 - - Total 79 3.40 - - minutes. This represents a significant increase in effective reading time on a word per minute basis for the 60 version. Reading rates were significantly different [E(3,76) = 41.68, .E‘°001] among all four story versions except T and 20. The Scheffe post hoc comparison revealed that reading rates were reduced 20% in the 40 version (ps.01) and 55% in the 60 version (ps.001) as compared to the T passage. Reading rate decreased 18% on the 40 (p§.05) and 53% on the 60 (p5.001) when compared to the 20 version. Reading rate decreased by 100 words per minute (43%) on the 60 material as com- pared to Ss reading the 40 version. Figures 3.1 and 3.2 illustrate the differential effects of read- ing rates and reading times for the four reading groups. Four 1 x 4 ANOVAs were employed to analyse the four additional dependent variables for the four Deletion Levels. The means and standard deviations for these variables are presented in Table 3.5. As shown in Table 3.6, the ANOVA for the multiple choice test revealed highly significant differences for the Deletion Level main . effect. 63 300 29L 285 l“ 50 235 e _____ o: 2 200 \ 191 g ACTUAL wpm READING RATE 50 ---- EFFECTIVE WPM COMPREHENSION RATE (am) 100 T 20 40 60 DELETION LEVEL FIGURE 3.1 EFFECT OF DELETION LEVEL ON THE TWO MEASURES OF READING RATE FOR THE READING TREATMENTS. mgao g El i-é 0380 _2_ as Q a) 1.1.1 :5; 7.0 05 s 9’ O T 20 4O 60 DELETION LEVEL FIGURE 3.2 EFFECT OF DELETION LEVEL ON READING TIME FOR THE READING TREATMENTS. 64 am.. mo.m mm.F oe.e mm.~ mm.N_ we.m op.em om om 40., m..e Ne._ m_.o oe._ oe.N_ oe.e m~.mm ON oe me.p o~.m we.F mm.e me.N mo.~_ mp.m oe.mm om om mm._ om.m _m.p oe.e mm.m oe.Np me.m om.ee ON A cowumpmo om eeez am new: om gee: om eeez z meowpeeeou e>eppe=hmpo eeemaz meeeeepem new eeeoeu a_aeppaz peeeeeeee muemeummch mcweemm one cw manewee> pemvcmamo eaou Low meoweew>mo neeucmum use memo: m.m m4mOeOOOOmOO Oeemez meeeeepem pew OONOOO O_Oee_Oz eeeeeeeee mmFOOOeO> mucmspemeh memcmumwo N_ we» cw mmpamwge> acmucmamo m>wd one Go comm Low meowpmm>wo nemucwum use memo: m~.m m4mm4 ZOEZMIMEQSOO 24m.)— m.m mmeE .mhzmzhhwmh 02.2m._.m_1_ 4m>m1_ zoFmJMQ ov 0N m._.m1_ zo_mzwImmn=200 24w: in meGE 4m>m.._ zo_._.m.._mo 0m 0? ON ._. m._.m4 ZO_._.m.._wo 0m 0.» ON ._. vo. 5%sz mnm 1111111 lwmm \.\\\\\ mom $‘HUI‘dfl‘lll-Ffl'l..l Ell”.- .mmN. RN \\H.\owm11 IIIIIII IMMN lmNN EON. \\\ \V.\ NmN 1:1. \ \\ ........ 111.1 I I I mmN mmN \\ ..\ \. ........... whN oW\\ NON oNn \ q \ mmn ..\ IIIIIIIIII 1... 00¢ .\ ..\ l lllll 2N \\ 825353.35... a m... \x 02.55... 0 NEE zoEfizmmmma 00_ on OON 00 00m 00 00¢ 31118 NOISNEHEHdWOO WdM BAILOEAAB 85 N0.00 OF._~O NO._ Om.O NO.. O0.0 NN._ ON.OF O0.0 Om.ON OOO O0.00 O_.NON OO.P O0.0 NN._ O0.0 OO.N OO.__ ON.O O0.0N OOO ON.OO OO.ONN ON._ O0.0 m_.N O0.0 NO.N O_.NP NN.N OO.ON OON O0.0N N0.00N OO.. ON.O OO.P OO.N OO.N OO.NF Om.OP OP.OO O» N0.00 NO.OPO OO.. O0.0 ON._ O0.0 mm.N OO.F_ O0.0 O_.Om NOO NO.OO NO.NmN OO._ O0.0 O.._ O0.0 ON.P OO.NP OO.N O0.0N NOO O_.ON O0.0mN NO.P OO.P O..N O0.0 OO.N mm.N. N..O OO.NO NON _0.0N NO.NON NF.P O0.0 OO.. mO.N ON._ OO.NF O0.0 OO.OO Ne mO.mO PO.NNN NO._ O0.0 PO._ O0.0 OO._ ON.N_ _N.O O0.00 _OO NO._N ON.OO~ Om.. O0.0 mm._ ON.O ON.N ON.N_ NO.O O0.0N POO OF.ON O0.00_ mm._ OO.O OO.F O0.0 OO.N ON.NP NO.N mm.me _ON OO._N ON.ONP ON._ OO.O NO.P O0.0 .m.N O..NF OO.N OO.NO _O OO Oeez Om Oeez Om Oeez Om eeez Om Oeez mOO_OeOOOO OON O>OOOOOOO_O Oeemez meoeeepem New OOOOOO OPOOONOZ OOOeOeeee mwpnmwee> magm2pmmeh m:_cmmm\mcwcmpmw4 up we» cw mmpnmeee> “coucmqmo m>pd we» do comm Low meowumw>mo uemucmum use meme: ON.m m4m024 Nx¢xm MI... 2_ ogw mom. whomumw 2.4—2 4w>m.._ 20_._.w1_wo 024 w._.4m 20_._.4.—.2wmwm_n_ zwwziwm ZO_._.o4mw.r2_ QM mmDGE ...m>m._ 20_._.m1_wo Om _ ov 0N ._. m... om. \\\\®m_ ..N l\.\\\ l IIIIII ANN \\ NeN.1melilli|1...IllI 11.11.1111 SON \\ .\..NON \ \ \ \ \ mom \. 1111111111 1 00¢ \. I IIIII mNN \ mt wh4m ZO_._.4._.2wmmmn_ 00. on CON om 00m On 00? BLVEI NOISNBHEIHdINOO WdM 3AI.LOB:I:IE 96 .4>024 Nxvxm NIH 2_ 03m mou— >._._.._4n_0_2 024 w._.4m 20_._.4._.2wmwmn_.2wm3._.mm 20_._.04mw._.2_ Em wmnwi m._.4m 20_._.4._.zmmmmn_ 00¢ mbN mt 02.04mm \ 02.295... 1111111111 I 02.2w._.m.._ 00. 00 00N om 00m 00 00¢ 3.13/El NOISNHHBBdI/IIOO WdM 3N103.:I:IB 97 ON.OO. O0.00N NN.F N0.0 OO._ N_.O OO.N OO.__ _0.0 OO.NN OO P0.00 O..NON ON.. ON.O ON.P O0.0 NO.N NO.P_ N0.0 ON.ON OOO _O._O NN.NON OO._ OO.N PN.N N0.0 OO.N NN.N_ OO.N Om..m OON OO.NO OO.NON ON.. ON.O NO.F NN.O _O.N NO.N. ON.O NO.NO Op ON.NO NN.OON PO.P N0.0 OO.F NF.O OO.N NO.N. OO.N m..ON NOO O0.00 OO.NON ON._ ON.O NN.~ N_.O _O.P OO.__ O0.0 O0.00 NOO ON.OO _0.0NN _N.F ON.O OO.N ON.O _O.N NO.NF OO.N OO._O NON Nm.Om O0.00. O_.. mm.m OO.. O0.0 ON._ ON.N_ OO.N OO.NO N» NN.Nm O0.0PN NO.. OO.O NN._ ON.O ON._ OO.N. NO.N NN.ON _OO OO.NN ON.OO_ ON.P ON.O OO.P NN.O OO.N OO.N_ m..m NN.ON _OO m0.0N NF.OOF OO.N N0.0 PN.. NN.O .O.N OO.N_ O0.0 OO.NO _ON OO.NN OO.NN_ _O.F N0.0 .O._ ON.O OO.N N_.N_ NO.N ON.OO _e Om Omez Om Oeez Om Omez Om Oeez Om eeez OOOONNOOOO Ozm e>_NOOOOmOO momma: OOOFOOFON New OOPOOO O_ONN_O2 Neeeoeeee mNchwcm> mocmEpmmch mcwmmomxmcwcmpmwc NF ecu Op umcmmeoo mNcNENmmcp mcwcmpmwc mco cw mopcmwcm> ocomcmaoo m>wm mco mo comm com meowpmw>oo mcmmcmom ccm memo: mm.m mmm< gum mo. gmm mm gag mg zogm mummlmcgummm ggg omg gm. mg gag egg puma gog oem ggg cg gag meg ammem>< gmo mg go. o_ ggo o zogm mama gummnw pmcoguomgm>cou mcwummm gmo m_m gmo no gem mmg gmcgcagmgg nacgnsoo ggg gm ggo g ggg mm mcgcamg gmg og_ gmm mm gog gg megcagmgg .mcgcgmmg mo mvoz acogugggu gmoz acougmm gmassz pcmugmm gmassz acougma Logan: wpmewuumpaz wpm: m—msmm cogummsm mgumgazm gagog mama cummam Focogummgm>=ou new .mumx mcgummm .aucmgugmgm mcwcgmmg mo mvoz co mcmpmmumpwm .mpumnnzm mo mupammm m.¢ m4m