THEsls This is to certify that the dissertation entitled DIFFERENTIAL COMPREHENSION PATTERNS OF MATURE SLOW AND FAST READERS IN EXTENDED DISCOURSE presented by Susan Maudie Gibson Hice has been accepted towards fulfillment of the requirements for 0 Ph. D. degree in Administration and Curriculum ; [hue September 30, 1983 MSU is an Affirmative Action/Equal Opportunity Institution 012771 MSU RETURNING MATERIALS: Place in book drop to LIBRARIES remove this checkout from w. your record. FINES will be charged if book is returned after the date stamped below. .5 A us} Ain‘»."“ .~!"‘.‘:"‘."5 3’" Or”? "‘l- "5 3‘" .. “7. . J 1 l ; r T' 54" ‘: .3 3 l1" _' it}? 3 he livkwti A“ ”#15 34.w they. in; .. 1 . .1 a 3% an ‘35?" '—' -- 'r "-5 ‘ii, 1’ vi © Copyright by SUSAN MAUDIE GIBSON HICE 1983 DIFFERENTIAL COMPREHENSION PATTERNS OF MATURE SLOW AND FAST READERS IN EXTENDED DISCOURSE By Susan Maudie Gibson Hice A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Administration and Curriculum T983 ABSTRACT DIFFERENTIAL COMPREHENSION PATTERNS OF MATURE SLOW AND FAST READERS IN EXTENDED DISCOURSE By Susan Maudie Gibson Hice The purpose of this study was to discover whether mature read- ers with slow or fast reading rates differed in the types of informa- tion recalled after reading. The study population contained l33 high school seniors who were good readers and high achievers. They were termed "mature" based on definitions of psychological maturity in perception and cognition. All had good near-point acuity. Four categories of comprehension were described in ascending order of structural complexity: detail, microstructure, main idea, and macrostructure. Multiple-choice questions testing recall in these four categories were constructed from a structural map of the study discourse. (Other study variables were Nelson-Denny Reading Test scores, SAT scores, and rank in class.) Mature readers were asked to read once through a general- interest Scientific American article of 2,698 words (4,385 syllables), using their normal internal purpose and rate for such assignments. Subjects were timed; recall in the four comprehension categories was assessed. Susan Maudie Gibson Hice Rate of discourse reading was correlated with each of the four comprehension variables; only microstructure (detail relationship) was statistically significant: r = .1609. (Nonsignificant trends were that main idea was consistently positive, whereas detail and macro- structure were positive or negative, depending on calculation method- ology.) It was concluded that (1) either speed facilitates connecting neighboring details, or (2) mature slow readers integrate microstruc- tures into existing constructs, employing considerable idea transfor- mation, so that the microstructures become less recognizable in their original form. Post hoc study groups of the 30 slowest (X'= 148 wpm) and 30 fastest (X'= 324 wpm) discourse readers were compared based on mean comprehension patterns. An apparent interaction was noted between rate groups and micro- and macrostructure comprehension: slow readers appeared superior in macrostructure and fast readers in microstructure. Middle-rate readers (X'= 212 wpm) excelled in detail comprehension. Theories maintaining that slow reading is a cause of poor com- prehension received little support within this population. It appeared more likely that rate is a reflection of the type of comprehension occurring. A "structure" hypothesis was presented maintaining that mature slow readers process and structure essentially all ideas encountered, using inductive reasoning for questions; mature fast readers focus on a smaller subset of ideas, using deductive reasoning for questions. To my parents, Andrew and Frances Gibson, whose unflagging support made completion of this dissertation possible. ii ACKNOWLEDGMENTS It is doubtful that work of this type is completed without the aid and encouragement of many other persons. It is fitting that this thesis should begin by acknowledging them and their contributions. Dr. Lois Bader, my advisor and teacher, had the great wisdom to bal- ance guidance with freedom. She allowed me to begin maturing as a scholar by not providing ready solutions to every problem. Rather, she placed in me the confidence that I could find many of the answers myself. The transition from student (sitting in class and learning from a master) to scholar (directing one's own learning) is thus made. I thank her for her confidence in me, for her guidance, and for the many things she has taught me. A debt of gratitude is also owed to other members of my doc- toral committee for their guidance and support: Drs. Laurence W. Lezotte, Eugene Pernell, and William H. Schmidt. Laurence Bates assisted in programming the data, and Sue Cooley typed the final manuscript. Several persons helped in the laborious process of data col- lection and compilation: Andrew H. Gibson, Frances Gibson, Neale E. Hice, Andrea Disbro, Bernie Olander, and Mary Ann Yaich. Although students and personnel of the participating schools cannot be identi- fied by name, a debt of gratitude is owed them for their participation and facilitation of data collection. iii My husband, Neale, and children, Jeffrey, Jennifer, and Rebecca, suffered with me through the dissertation process and were willing to take on additional duties and put up with my absences (psychological and physical) that the writing might be completed. I love them more than I can say. My parents, Andrew and Frances Gibson, and my brother John and sister-in-law Toni provided haven and psychological comfort on my out-of—state jaunts to work at MSU's library. What a lot of love and good times we shared. Last of all, I wish to thank the friends who prayed for me, that the ideas would come, that I might finish, that my family would be sustained--especially Donna and Ruth. And I thank God for answer- ing their prayers. iv TABLE OF CONTENTS Page LIST OF TABLES ......................... vii LIST OF FIGURES ......................... viii LIST OF APPENDICES ....................... ix Chapter I. THE PROBLEM ....................... 1 Introduction ..................... l Statement of the Problem ............... l Theoretical Framework ................. 1 Psychology of Reading ................ 1 Psychological-Process Indicators .......... 4 Rate of Processing ................. 7 Postulation of "Structure" Hypotheses ........ 14 Research Questions .................. l6 Limitations ...................... 19 Definition of Terms .................. 20 Summary and Overview ................. 22 II. RELATED RESEARCH AND LITERATURE ............. 23 Introduction ..................... 23 Reading Rate as a Correlate of Comprehension ..... 23 Using Standardized Tests .............. 25 Using Extended Discourse .............. 25 Summary ....................... 27 Factors Upon Which Reading Rate Depends ........ 27 Nature of the Material ............... 28 Purpose for Reading ................. 29 Rate Training .................... 31 Reader Characteristics ............... 33 Summary ....................... 35 Reading Rate as a Factor in Determining Comprehension . 35 Natural Rates .................... 35 Assigned Rates ................... 37 Rate and College Achievement ............ 40 Summary ....................... 4l Summary ........................ 42 Page III. METHODOLOGY ....................... 43 Introduction ..................... 43 Population ...................... 43 Procedure ....................... 44 Sample-Selection Procedure ............. 45 Visual Testing and Task Procedure .......... 47 Materials ....................... 50 Developing Task Questions .............. 52 Validating the Questions .............. 54 Design ........................ 55 Data Analysis ..................... 56 Hypotheses ...................... 57 Null Hypotheses ................... 57 Directional Hypotheses ............... 60 Summary ........................ 60 IV. PRESENTATION AND ANALYSIS OF DATA ............ 62 Introduction ..................... 62 Hypotheses and Statistical Tests ........... 63 Null Hypotheses ................... 63 Directional Hypotheses ............... 77 Summary ........................ 79 V. SUMMARY, DISCUSSION, CONCLUSIONS, AND RECOMMENDATIONS . . 81 Summary ........................ 81 Discussion ...................... 85 Research Questions ................. 85 General Comparisons With Background Studies ..... 91 Informal Post Hoc Analysis ............. 97 Failure of Directional Hypotheses .......... 108 Revised "Structure" Hypotheses ........... 109 Implications for Education ............. 113 Conclusions ...................... 114 Recommendations for Further Research ......... ll7 APPENDICES ........................... 120 REFERENCES ........................... 146 vi Table LIST OF TABLES Intercorrelations Between Standardized Test Results, Achievement Variables, Four Types of Discourse Compre- hension Variables, and Discourse Reading Rate ..... Frequency Data for Population Variables ......... Correlations Between Four Types of Discourse Compre- hension and Discourse Reading Rate, Assessed With Different Statistical Procedures, and Including Variance for Statistically Significant Factors ..... Summary of Results of Statistical Tests of Null Hypotheses l and 2 ................... Summary of Results of Variance Comparisons of Null Hypotheses Controlled for Certain Factors ....... Summary of Results of Directional Hypotheses ....... A Comparison of Mean Reading-Rate Changes Between the Slow-Slow and Fast-Fast Subjects in Nelson-Denny and Extended-Discourse Task Material, in Words Per Minute and Syllables Per Second ............ Subpopulation Means for Post Hoc Comparison of the 30 Slowest and the 30 Fastest Readers of Extended Discourse, for All Study Variables ........... vii Page 64 65 72 82 83 84 95 101 Figure LIST OF FIGURES A Diagram Showing the Seven Textual Ideas ("Effects") Around Which the Major Portion of Langer's "The Black Death" Article Is Organized ........... Nelson-Denny Reading Rate for Each Student as a Function of His/Her Reading Rate on the Study Material ........................ Number of Questions Per Comprehension Category Answered Correctly by Each Student as a Function of His/Her Rate of Reading the Extended-Discourse Material ........................ Mean Group Accuracy for Slowest and Fastest Readers on Four Types of Comprehension .............. Mean Group Accuracy for Slow-Discourse, Middle Group, and Fast-Discourse Readers on Four Types of Comprehension Questions ....................... viii Page 51 94 99 LIST OF APPENDICES Appendix A. VERBAL DIRECTIONS GIVEN TO EACH STUDENT BEFORE THEY READ THE TASK ARTICLE ................. B. QUESTIONS FOR BLACK DEATH ARTICLE ............ C. STRUCTURAL DIAGRAM 0F LANGER'S (1968) "THE BLACK DEATH" . D. DETAIL QUESTIONS, SHOWING CORRECT ANSWER, AND TEXT DERIVATION 0F ANSWER ................. E. MICROSTRUCTURE QUESTIONS, CONSTRUCTED FROM NEIGHBORING DETAILS, AND REPRESENTING SUPPORTING IDEAS ...... F. MAIN-IDEA QUESTIONS, SHOWING CORRECT ANSWERS AND DIVERSE ELEMENTS OF TEXT THAT THE ANSWER AND/OR THE DISTRACTORS ARE DRAWN FROM ............ G. MACROSTRUCTURE QUESTIONS, RELATING MAJOR TEXTUAL IDEAS AND SOMETIMES ALSO GLOBAL APPLICATIONS ........ H. RAW DATA ........................ ix Page 121 123 128 CHAPTER I THE PROBLEM Introduction This first chapter includes a statement of the problem, the theoretical framework of the problem, the research questions, limita- tions of the study, and a definition of terms. Statement of the Problem The purpose of this study is to discover whether mature read- ers with slow or fast reading rates differ in the types of informa- tion recalled after reading. Theoretical Framework Psychology of Reading Weaver (1977) defined the psychology of reading as "the scientific study of graphical decoding operations in the human organism.” Gibson and Levin (1975) added that reading is a self- directed, active process, and not simply the act of matching apprOp- riate sounds to written symbols. Fluent reading, according to Cunningham (1979) , is achieved when: l. a large number of high- and medium-frequency words are recognized immediately; 2. unfamiliar words can quickly and accurately be pronounced "without isolating phonemes"; 3. semantic and syntactic cues can be used to predict the meaning and usage of upcoming words; and 4. the upcoming words can be recognized without completely processing them. Good reading, however, involves more than using good "auto— matic decoding" perceptions (a la Cunningham); it also involves cer- tain thought processes necessary to good comprehension. Failure to evaluate sentence elements, in light of other paragraph elements, and purpose for reading, will result in incorrect comprehension. Incor- rect meanings can be obtained from a text: 1. "because of wrong connections with the words, singly"; 2. because certain elements are over-weighted or under- weighted; or 3. because ideas obtained from the reading are not treated as "provisional" and to be accepted gr_rejected as they are examined. So proper paragraph understanding involves weighing elements correctly in relationship to each other, and analyzing and organizing the elements properly. Much poor comprehension is not from failure to remember what is read, or failure to organize the items read. It is because the text was never properly understood to begin with (Thorn- dike, 1917). Gibson and Levin (1975) divided the psychology of reading into perception and cognition, where perception is the process of extracting the "relevant information from the manifold available stimulation," and cognition is the self-directed use of these per- ceptions. Perception involves extracting distinctive features and invariant events from multiple stimuli and finding structural rela— tionships among and between them. Cognition involves remembering, problem solving, and organizing conceptual knowledge in order that meaning may be better extracted from perceptions. Cognition in the broadest sense, of course, includes perception. Both perceptual learning and the learning of cognitive strate- gies are developmental. Eye movements, reflections of what a person must do to read (Buswell, 1937), and eye-voice span, the distance the eyes move ahead of the voice, become adult-like at the fourth grade. After grade 4, improvement in reading efficiency appears to be asso- ciated less with improved oculo-motor activity than with "superiority at a deeper level of processing" (Gibson & Levin, 1975). Piaget found that the average child reaches the level of mature adult think- ing (formal operational stage) by the age of 14 or 15 (Hafner, 1977; Wilson, 1979). Thus, the subjects of the present study, being above- average students, being between the ages of 16 and 18, and showing a vocabulary and comprehension ability above their grade level, should all have reached the basic level of perceptual and cognitive maturity necessary to adult reading. The term "poor readers," in most existing studies of the read- ing speed variable, is used to delineate readers with decoding or comprehension difficulties. In neither of these senses could subjects of the present study be considered poor readers: their decoding and comprehension abilities were excellent. Persons with decoding and comprehension problems generally read at slow rates (the definition of "slow" varying, frequently not being described). What about readers whose decoding and comprehension are excellent, but whose silent- reading speed is still slow (i.e., close to the speed of adult oral read- ing)? Buswell (1937) was referring to this population when he wrote: Many readers at the college level who understand thoroughly what they read and whose critical reactions to the material are of a high order use cumbersone [reading] procedures which so restrict their choice and amount of reading that their entire fund of ideas remains limited. What are the known psychological indicators of reading process, and may these indicators be considered as symptoms or as causes of pro- cess difficulties? Psychological-Process Indicators Buswell (1937) conducted a comparative study of psychological- process indicators. First he recorded the eye movements during oral and silent reading of 1,000 adults with varying reading abilities. Comparing the worst with the best 100 readers, Buswell found the best had: wider recognition span, fewer regressions per line, and a slightly shorter fixation duration. Mature silent-reading skills were psychologically different from oral-reading skills. The silent reading of the poorest readers (i.e., those with process difficulties) more nearly resembled oral reading than mature silent reading. On each oral-reading film the subject's voice was also recorded, so that precise word-eye relationships could be determined. The eye-voice study showed that the best readers had a wider eye- voice span (EVS). Gibson and Levin (1975) compared readers, aged 7 through adult, and found that fast readers had a longer EVS than slow readers. Fast readers were more likely to end their EVS at phrase boundaries, showing a greater use of grammatical structure. However, even the youngest and poorest readers made some use of sentence structure in reading. Buswell stressed that eye movements are symptoms of good or poor reading, not causes. They reflect what the person must do to read. Last, Buswell (1937) examined vocalizations and lip movements made by his subjects when reading tongue twisters silently. He found that there was generally less vocalization and lip movement in better readers. Edfeldt (1960) monitored the mylohyoid muscle, which responds only to speech impulses. He found less silent speech in good readers than poor readers. Easy, clearly written texts showed less silent speech than hard or unclearly written ones. Even fast readers did not circumvent this motor-base of speech activity, although they dis- played less silent speech than slow readers. Watkins (1979) effected a reduction in subvocalization levels of seventh graders, experimen- tally, but no transfer to rate of comprehension appeared to occur. In an earlier study, Buswell (1945) had shown that lip movement could be reduced by conscious teaching but that any such "improvements" would disappear over time. Subvocalization, said Edfeldt (1960), is not a cause of poor reading, but rather, a symptom; “silent speech does not occur until there is a need for extra aids in order that the text may be understood." Or, as Buswell observed (1937), one reverts to the more immature habits which accompany oral reading with particu- lar textual difficulty or under emotional tension. Huey (1908, 1968) found that auditory imagery (saying the words in one's head) characterized the silent reading of many gradu- ate students. Since this inner speech of silent reading was faster than the vocalized speech of oral reading, he deduced its nature to be simpler. Oral reading was an average of 56% to 66% slower than silent reading in one experiment. Huey calculated that at speeds above 300 to 400 words per minute it would be impossible to form auditory images of all words; persons reading at higher speeds would probably use auditory imagery in a more abbreviated manner. Persons reading at oral-speech levels are probably forming auditory images of the words in their head as they read. Inner speech may be classed with vocalization, lip movement, and mylohyoid-muscle movement as different aspects and degrees of auditory imagery processing. While perceptual processing as indicated by eye movements seems at mature levels in fourth grade (Gibson & Levin, 1975), Samuels' study (1975-76) showed that rate of perception of words continues to improve into adulthood. In comparing good with poor fourth-grade readers, and in comparing college undergraduates with fourth-grade readers, using words known to all subjects, Samuels found the more mature groups to be better at generating words from word fragments, more aware of incorrect recognitions, faster at word recognition, and processing at a faster rate. Average rates of oral and silent read- ing also improve beyond the fourth grade. Edfeldt (1960) included a chart in his study of silent speech and silent reading which shows how oral- and silent-reading rates diverge in the primary grades. In grade 1, pupils read, orally and silently, at an average rate of 45 wpm. By fourth grade the respective rates were 135 and 156 wpm. In sixth grade, the rates were 170 and 210 wpm. (It appears that the oral- and silent-reading rates of some good readers do not increase much beyond their sixth-grade levels, although cognitive strategies improve [Hafner, 1977]). Edfeldt's chart ends here, but other sources allow further approximations. The maximum adult oral-reading rate probably does not exceed 250 wpm, as it is practically impossible to be understood at rates above this level (Buswell, 1937). Since the average rate of adult speech is between 170 and 200 wpm (Gibson & Levin, 1975), actual average rates of adult oral reading probably would also be in this range. In fact, Buswell (1937) sets the average adult oral-reading rate at "below“ 200 wpm. The average college stu- dent has a silent-reading rate of 280 wpm (Harris, 1968). Rapid silent reading, however, can commonly reach 600 wpm. The faster one reads silently, the less it is like oral reading, psychologically (Buswell, 1937). Rate of Processing While aspects of eye movements and silent speech have been shown to be symptoms of poor reading, rather than causes, some theo- rists are not ready to make the same statement for reading speed. The speed of input is thought by some to be not a symptom but a cause of comprehension difficulty. Gough (1972) referred to the poor compre- hension associated with word-by-word reading to show that delays in input can affect comprehension. Gough felt this rate-comprehension effect can be explained by examining two characteristics of the primary memory (PM): 1. Seven entries must be held together in the PM (for pro- cessing) for sentence comprehension to occur, and 2. Items held in the PM decay rapidly if not processed. Gibson and Levin (1975) said, "There is a minimal speed of reading below which the syntactical and meaningful relations within a sen- tence or a larger unit of discourse do not come through." This mini- mal speed would be that of word-by-word reading. Therefore, at low enough speeds, reading comprehension is adversely affected by rate (Cunningham, 1979; Rankin, 1963). Fridal (1979) declared that for any given text there is a threshold reading speed above which the mind is occupied enough with information processing not to wander. Below this threshold, not enough stimulation of the brain occurs to keep it actively engaged in the reading task only. Thus, concentration is not as good at slow reading speeds. From his clinical experience with college stu- dents, Fridal stated that 200 to 350 wpm can be too slow to preserve concentration. (It sounds like the "concentration" thresholds he was advocating were those where auditory imagery has a lessening influ- ence on processing.) The "slow" subjects of the present study were not word-by-word readers (a natural rate of less than 200 wpm does not indicate word-by-word reading). Their perceptual and cognitive maturity were at adult levels. Can such theory apply to them? Could the reading speed of slow-reading good-comprehenders be frustrating propositional memory and causing difficulty in grasping relationships? Rothkopf (1979) paid high school, then college, volunteers to read for different specific learning goals (specific details). The control group was not given specific goals but was instructed to learn as much as they could about the text. When reading for specific goals, (1) goal-related material was inspected longer than incidental material, and (2) more goal-related than incidental material was recalled. Eye movements were analyzed for 18 paid high school volun- teers while they either read for different specific learning goals or to learn as much as they could. (Forward saccades, regressions, fixation durations, and return sweep time were recorded.) Experi- mental subjects seemed to make a judgment that a sentence was goal related after reading about two-thirds of it. They then displayed diverse styles of inspecting the material. Some gave added attention to goal-related material, employing some combination of rescanning lines, using more fixations, and/or using longer fixations; they spent longer on goal-related material than incidental material. Some experi- mental subjects continued inspecting goal-related material in the same manner as incidental material (p < .05); they spent the same amount of time on all material, regardless of goal-relatedness. Recall accuracy was essentially equivalent for these various styles displayed by the experimental group. (The control group spent as much time on each line as though it were goal-relevant. Recall was significantly better for the experimental group, in goal-related material, but sig- nificantly better for the control group in incidental material. As the number of specific goals was increased, in a separate trial, goal-related recall decreased.) As the experimental results are examined, at least three conclusions seem possible: 1. Extra processing time spent on goal-related material (details) is related to increased recall of these details. 10 2. Mere recognition that a text is goal related is suffi- cient to induce increased recall. Any extra time spent is relatively nonproductive in terms of recall of these details. 3. Extra processing time is required on goal-relevant mate- rial by some, but not all, subjects, to produce increased recall. Since Rothkopf did not correlate inspection style with rate, we do not know whether a relationship existed between them. Fast readers are known to be more flexible readers, adjusting their rate to type of reading material (Robinson, 1941). If incidental knowl- edge were assessed, faster readers would probably get lower scores than slow readers (McConkie, 1973). Fast readers read for the central contents (Dee-Lucas, 1979). Perhaps slow readers do not vary their inspection style (eye-movement patterns) when encountering goal- relevant material, but fast readers do, displaying a range of eye- movement modifications. The slow reader's style may be expressed in #1, above, and the fast reader's style in #2, above. Dee-Lucas (1979) studied what randomly assigned undergradu- ates do as they read faster. (Subjects were either rewarded for more thorough comprehension or for a higher speed of reading, producing two distinct rate groups.) She found that as students read faster (120 vs. 251 wpm), they selected material they judged to be of importance and focused on that. She thought that the group induced to read faster processed a smaller subset of material, but did so with thoroughness equal to the group induced to read slower. The slower group recalled 1.6 times as many propositions as the faster group (i.e., 37% vs. 23%); recall sentences of these two groups were 11 equally well structured. Thus the very slow speed of 120 wpm did not appear to frustrate the comprehension of these assigned slow readers, and relations within the text did come through. The faster readers picked up only half as many implications from the text as did slower readers. This may indicate that seeing implicit causal relations is facilitated in good readers by a slower reading speed. There was not as much memory return (immediate recall of propositions) for the time spent in reading at these lower rates, however. Students reading at about 120 wpm recalled about 10.7% of selection proposi- tions for each minute of reading time. But students reading at about 251 wpm recalled about 13.9% of selection propositions for each minute of reading time. Perhaps faster subjects were recalling more propositions for each minute of reading time because their con- centration was less impaired than slower subjects'. Perhaps slower subjects were recalling fewer propositions for each minute of reading time because they were dwelling more on implications. Perhaps in order to remember more details, more transformation of material (into implications) must occur. Another aspect of seeing relationships is ambiguity resolu- tion. If subjects are attempting to resolve text ambiguities, they will spend longer on the text. Resolving ambiguities takes time (Tanenhaus et a1., 1979). In reading ambiguous words, there seem to be two processing stages: all meanings of the word are accessed, then meanings which do not fit the context are suppressed. Students reading more slowly make more inferences. Making correct inferences would require properly resolving ambiguities. No support has been 12 found for Fridal's contention that students reading slowly have greater concentration problems. Perhaps the types of relationships seen at slower speeds are of a different type than at faster speeds--detail relationship rather than main-idea relationship--and this difference has led to Fridal's assertion (i.e., perhaps he identified concentra- tion with seeing main—idea relationships). Another interesting thought occurs at this juncture. Perhaps auditory-imagery usage in processing does more than thwart memory decay at slow speeds. Per- haps it facilitates the grasp of details and their relationship for the slowly reading good reader. Kerr (1973) studied how processing demand is affected by various mental operations. She supported a hypothesis wherein a central processing mechanism of limited capacity would be necessary to some, but not all, mental operations. The degree of central pro- cessing required by different mental operations was studied by "com- paring their degree of interference with a common secondary task.“ Using sentences and word lists, Kerr found that greater processing capacity was required: 1. when item difficulty was increased; 2. when list length was increased; 3. when subjects knew recall would be expected of them later; 4. when length of the retention period was increased; 5. when multiple input material (word lists, sentences) was presented more rapidly; and 6. when stimulus material was transformed into another form (as in computing answers or inferring). 13 Processing demands decreased over rehearsal time, perhaps because some items were committed to long-term memory, thus lightening the rehearsal load. Let us assume that such results apply not only to processing word lists and sentences, but to reading extended discourse, as well. Combine the Dee-Lucas and Kerr results. Mature slower readers have to retain meanings of phrases, sentences, and paragraphs longer before synthesizing them for comprehension. They also appear to focus on a larger amount of material, make more inferences, and later recall a greater volume of material than faster readers. This increases the processing load on the limited central mechanism. Are mature slower readers trying to commit material to long-term memory as they go by incorporating more rehearsal time? This would lighten their process- ing load (as would their slower speed). Perhaps making inferences, discovering unstated causal relations, is part of the slower reader's rehearsal strategy. Some good readers read at habitually slow rates; do they adopt their slower rate due to bad habit and/or a preference for understanding detail and causal relations? Mature fast readers would presumably be using more of their central processing capacity by increasing their speed of input. They appear to narrow their focus in reading, however, selecting material they consider important. They do not retain and recall as large a volume of material or make as many inferences (which cuts down on processing demands), but their time is more efficiently spent in terms of recall volume per unit of time on task. Perhaps, if mature fast readers could be evaluated, we would see that, while missing 14 some details and causal relations, they have a better view of overall topic organization and main-idea relationship. Since Dee-Lucas' subjects all were reading at induced rates below that of the average college student (i.e., below 280 wpm), we lack conclusive evidence on these points. Mature slow and fast readers do appear to Spend their processing-capacity capital differently while engaged in the various mental Operations of learning from reading, however. Processing style changes with the speed of reading (for assigned rates and natural rates) and with the goal relevance of a particular text. Two seemingly counter theories to explain how mature readers process text are Gough's and Goodman's. Gough (1972) felt that all visual data from the text are processed and that meaning is con- structed from this base. Goodman (1976) felt that "the least amount of information possible to make the best guess possible" is processed by the good reader. The intriguing thing about these theories is that Gough's sounds like possibly the mature slow reader's strategy, also resembling perhaps the inductive method of reasoning (forming conclusions from a body of one's own data). On the contrary, Goodman's sounds like possibly the mature fast reader's strategy, also resembling the deductive method of reasoning (synthesizing from clues). Gough's strategy seems best for gleaning details (something slow good readers excel at), and Goodman's strategy seems best for gleaning main ideas (something at which fast good readers appear to excel). Postulation of "Structure" Hypotheses To summarize the information-processing strategies hypothe- sized to be in use by natural slow and natural fast readers, the 15 building of details, inferences (microstructure), main ideas, and their relationship (macrostructure) will be likened to the building of a house. Slow readers build their home brick by brick. They work together all details of the first part of the substructure encountered, so that individual pieces are in proper relationship to each other. The various substructures are assembled in the order encountered until the whole text is assembled into substructures. Depending on the familiarity (Nicholson, 1978-79) of substructures encountered, sub- structure relationships are seen with varying degrees of clarity as he builds. Some substructures may be fitted together after initial construction (or the builder may be so exhausted from his effort that this final "fitting" will not occur). The builder could be left with a well-constructed house, or well—constructed components strewn hap- hazardly across the lawn, or anything in between. In other words, important relationships of main ideas may be overlooked by slow read- ers; this is the disadvantage of this strategy. The fast reader stands a better chance of having at least his superstructure intact. He puts that up first. He only fills in those details considered to be of major importance to the house, as he puts up the superstructure. One can see what the finished house would look like. With this building technique, the house will take its form fairly rapidly. The builder might be very uncomfortable moving into this house, however. Unfinished details could make the house cold and uncomfortable. He has guessed about which parts of the house were necessary and which could be ignored, and has perhaps been wrong at times. He may miss details from which unique implications can be 16 built. Especially in the field of science, many new discoveries come from building up a compendium of details into unique configurations. Of course, each builder should be able to build a house according to the demands of the situation. Building a detailed house to sleep in one night would be wasteful of time. As we shall see (Review of Literature), mature slow readers prefer to build all their structures the same way, whereas mature fast readers show more ability to adapt to situation needs. No doubt, each is most comfortable with his own building style, however. This experiment has been set up to test out aspects of this ”building" hypothesis. 00 mature slow readers excel at detail and inference comprehension, as research has indicated, and do mature fast readers excel at main idea and major relationship comprehension when allowed to read a text through once and answer questions from memory? Research Questions In the previous section, information processing was explored in an attempt to form a profile of the reader who reads slowly in spite of having good perceptual and cognitive development in reading- related tasks. A theory was formulated to predict differential types of comprehension to be expected from mature slow and fast readers on an initial read-through of extended discourse material. In this section, research questions are formulated to outline the correla- tional study that addresses this theory. 1.0 Will there be a relationship between standardized and academic performance data, and the rate at which an extended discourse is read, for thh-grade students who are mature readers? 2.0 17 Will there be a relationship between Nelson-Denny Reading Test Vocabulary scores, and the rate at which an extended discourse is read, for thh-grade students who are mature readers? Will there be a relationship between Nelson-Denny Reading Test Comprehension scores, and the rate at which an extended discourse is read, for thh-grade students who are mature readers? Will there be a relationship between Nelson-Denny Reading Test Total scores (Comprehension and Vocabulary combined), and the rate at which an extended discourse is read, for 12th-grade students who are mature readers? Will there be a relationship between Nelson-Denny Reading Test Rate scores, and the rate at which an extended dis- course is read, for thh-grade students who are mature readers? Will there be a relationship between Scholastic Aptitude Verbal scores, and the rate at which an extended discourse is read, for thh—grade students who are mature readers? Will there be a relationship between Scholastic Aptitude Math scores, and the rate at which an extended discourse is read, for thh-grade students who are mature readers? Will there be a relationship between class rank, expressed as a percentile, and the rate at which an extended discourse is read, for thh-grade students who are mature readers? Will there be a relationship between the number and type of questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade stu- dents who are mature readers? 2.1 2.2 Will there be a relationship between the number of para- phrased detail questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for thh-grade students who are mature readers? Will there be a relationship between the number of micro- structure questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for thh-grade students who are mature readers? 3.0 4.0 2.3 2.4 18 Will there be a relationship between the number of main- idea questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? Will there be a relationship between the number of macro- structure questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? Will Scholastic Aptitude Test Verbal scores have an effect on the relationship between number and type of comprehension ques- tions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? 3.1 3.2 3.3 3.4 Will Scholastic Aptitude Test Verbal scores have an effect on the relationship between number of paraphrased detail questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for thh-grade students who are mature readers? Will Scholastic Aptitude Test Verbal scores have an effect on the relationship between number of microstructure ques- tions correctly answered after reading an extended dis- course, and the rate at which that discourse is read, for thh-grade students who are mature readers? Will Scholastic Aptitude Test Verbal scores have an effect on the relationship between number of main-idea questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? Will Scholastic Aptitude Test Verbal scores have an effect on the relationship between number of macrostructure ques- tions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? Will percentage rank in class have an effect on the relationship between number and type of comprehension questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for thh-grade students who are mature readers? 19 4.1 Will percentage rank in class have an effect on the rela- tionship between number of detail questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? 4.2 Will percentage rank in class have an effect on the rela- tionship between number of microstructure questions cor- rectly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? 4.3 Will percentage rank in class have an effect on the rela- tionship between number of main-idea questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for thh-grade students who are mature readers? 4.4 Will percentage rank in class have an effect on the rela- tionship between number of macrostructure questions cor- rectly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? Limitations It is possible that the format of this study may have dis- rupted the method whereby some subjects obtain meaning from print. At least some fast readers first survey an article to determine main ideas, then reread to fill in important related information. Such a preview was not permitted in this study. Students were allowed to read through the article only once. Subjects who develop ideas as they read by looking back to check on an idea may likewise experience disruption of their method if the desired reference is located on a previous page. Comprehension responses only reflect information gleaned from an initial read-through, and they must be interpreted in this light. 20 Results of this study do not apply to 12th graders who may be excellent students, but whose Nelson-Denny Total Reading score falls below grade level. Definition of Terms Certain terms used in the research questions need further delineation. Mature readers are defined as 12th-grade students who are in Regents or accelerated Regents classes and who score at or above the 12th-grade level on the Nelson-Denny Comprehension subtest, and who also score at or above their grade level on the Nelson-Denny Reading Test Total (a combination of Comprehension and Vocabulary subscores). Piaget found that maturity in thinking, the formal opera- tional stage, is generally reached by age 14 to 15 (Hafner, 1977). Sixteen- to 18-year-old 12th-grade students reading at or above grade level, and being mostly in the top quartile of their class, are there- fore assumed to be at a psychologically mature level in perceptual and cognitive reading abilities. Details are defined as particular actions, facts, or descrip- tive information that is located in one place in the text, generally a single sentence. Microstructure is defined as the organization of details, based on stated and implied relationships. Such comprehension requires understanding the relationship of sentence details in one paragraph, or in neighboring paragraphs. Microstructures represent supporting ideas, rather than main ideas. 21 Main idea is defined as one of the major points of an extended discourse. To form the main idea, elements (microstructures, supporting ideas) must typically be gathered from several (noncontig- uous) sectors of the text, weighed to determine their relative impor- tance, and then synthesized. Main ideas in this study were not explicitly stated in the text. Macrostructure is defined as the organization of main ideas and related concepts of an extended discourse. This structure of ideas must be inferred by the reader by assigning relative importance to various categories of information (main ideas), then inferring the proper relationship of these categories. Macrostructure may also be thought of as the relationship of major textual ideas to a unifying concept (Bader, 1980). A correct perception of macrostruc- ture requires an understanding of the important textual ideas and their relationship; it also allows for application of those perceptions to new situations. Discourse is defined as any stretch of structurally related language (Clark & Clark, 1977). Extended discourse is defined as an extended stretch of struc- turally related language. For the purposes of this study, a passage 2,698 words (4,385 syllables) and 28 paragraphs in length is under- stood to represent extended discourse. Percentage class rank is defined as that value which is obtained when class rank is converted into a percentage. (class rank) x 100 percentage class rank = (number of students in class) 22 A percentage class rank of 15 means that a subject was in the top 15% of his class: 85% of the students performed at a lower academic level and 15% performed at the same level or higher. (This score was introduced as an approximate method of equating rank scores from eight different schools that had senior-class populations ranging from 39 to 250.) Summary and Overview The problem addressed in this study was introduced in this first chapter. Related theory was discussed in terms of the follow- ing topics: the psychology of reading, indicators of psychological process, and rate of processing discourse. Different information- processing strategies were hypothesized to be associated with naturally slow or fast reading rates. The chapter ended with a listing of research questions, study limitations, and definitions of terms. A review of the literature related to the study problem is presented in Chapter II. The methodology used in conducting this study is described in Chapter III, together with a list of null and directional hypotheses. A statistical analysis of the data is pre- sented in Chapter IV for each hypothesis. In Chapter V, results of statistical tests are interpreted in light of related literature and theory. The chapter concludes with a post hoc analysis of data, discussion of implications, and recommendations for further research. CHAPTER II RELATED RESEARCH AND LITERATURE Introduction The purpose of this study was to attempt to discover whether mature readers with slow or fast reading rates differ in the types of information recalled after reading. Are different flexibility pat- terns associated with these differences? The review of literature presented here is organized under three major headings involving reading rate. These headings are: (1)readingrate as a correlate of comprehension, (2) factors upon which reading rate depends, and (3) reading rate as a factor in determining comprehension. Reading Rate as a Correlate of Comprehension Several early researchers operated from the premise that as one becomes a more proficient reader, one's rate increases. The faster reader was assumed to have a better understanding of what he read than the slower reader (King, 1916). However, when researchers have attempted to establish th§_correlation of reading rate and com- prehension, widely divergent results have been obtained. King (1917) found a rate-comprehension correlation of -.47 in one study, and Eurich (1930) reported a relationship of +.67 in another. Most studies reported a low positive correlation, indicating very little comprehension advantage associated with a higher reading rate (Harris, 1968). 23 24 Researchers have attempted to identify the variables respon- sible for the wide correlational results. Results have been found to depend on testing procedures, nature of the material, and reader char- acteristics. Standardized tests of rate and comprehension are generally composed of short unrelated paragraphs. One or more comprehension questions follow each reading selection. Testing procedures have varied widely, yielding different correlational results. Sometimes rate has been determined on different textual material than has com- prehension; the presence of uncontrolled variables such as text dif- ficulty and content render such correlations (which are generally low) invalid (Letson, 1958; Tinker, 1932). Robinson (1941) reported that rate and comprehension-accuracy scores vary between different subject areas, and between fiction and nonfiction. This effect was particularly pronounced for the rate scores of fast skilled readers. Tinker (1939) showed that when rate/comprehension correlations are calculated on separate materials they decline, even when the mate- rials are judged comparable. Rate and comprehension should be measured on the same textual material. Sometimes the rate determination included the time spent reading and answering questions. When such a procedure was combined with giving subjects a strict time limit to work in, inordinately high correlations resulted (Tinker, 1939). More recent group stan- dardized tests use a longer beginning selection so that rate may be determined based on one minute of reading text only (Nelson, 1960). 25 Using Standardized Tests When standardized tests are used with rate/comprehension studies, the correlations reported will be higher than if extended- discourse material had been used (Tinker, 1932). In studies employ- ing standardized tests, correlations were also higher under the following conditions: 1. when strict time limits were imposed (Preston & Botel, 1951; Rankin, 1962), 2. when easily comprehended material was used (especially in combination with strict time limits) (Tinker, 1939), 3. when the rate score included time to read and answer com- prehension questions (Rankin, 1962; Tinker, 1932), 4. when the rate of reading incorrectly answered paragraphs was excluded from the rate calculations (Blommers, 1944), 5. when the purpose for reading was general (Carlson, 1949), 6. when intelligence of sample subjects was higher (Carlson, 1949). Using Extended Discourse Some rate/comprehension correlational studies have used extended discourse rather than standardized test material in study- ing these relationships. One long reading selection is followed by several comprehension questions. Sometimes subjects read only for a specified time period, and other times they read until they com- plete a selection. Some studies have specified the nature of reading purposes and the types of comprehension questions, and others have not. 26 Occasionally, slow and fast readers have been compared (King, 1916, 1917; Robinson, 1941). Related studies have dealt with the ideal length of extended-discourse selections to yield reliable results (ideal length having been found to vary with school grade and read- ing skills of the sample population) (Chang, 1977; Robinson, 1941). As has already been mentioned, correlations in extended discourse are generally much lower than those obtained in unrelated-paragraph tests. In studies employing extended discourse, rate/comprehension correla- tions have been found to be higher_under the following conditions: 1. when time limits were imposed and the total number of details remembered was compared (King, 1916, 1917); 2. when rate and comprehension were measured on easier mate- rial, under strict time limits (Letson, 1958), 3. when comprehension was measured as the number of correct responses, as opposed to the ratio of correct responses to items covered or tried (accuracy); this effect was especially pronounced when strict time limits were used (Letson, 1958; Robinson, 1941); 4. when social studies or literature reading material is used, as Opposed to science and math material (Shores, 1951; Thurstone, 1944); 5. when fiction material is used, as opposed to nonfiction material (art, history, science) (Robinson, 1941); and 6. when critical thinking is pot_involved (Carlson, 1949). Similarly, the rate/comprehension correlations in extended discourse have been found to be lower under the following conditions: 27 1. when science or math reading material is used (Shores, 1950), 2. when critical thinking is involved (i.e., questions requiring inferences) (Carlson, 1949), 3. when the volume of details recalled from a specific seg- ment of text is considered (King, 1916, 1917), 4. when comprehension is measured as the ratio of correct answers to the number of questions attempted (Letson, 1958; Robinson, 1941), and 5. when comprehension is measured on difficult material with a strict time limit (Letson, 1958). Summary It may be seen from these studies that there is no general rate-comprehension correlation. Likewise, speed of reading does not appear to be a unitary concept, but it varies with intelligence, pur- pose, and difficulty of material (as well as artifacts of measurement and scoring procedures) (Carlson, 1949; Tinker, 1932). Factors Upon Which Reading Rate Depends As it became more and more obvious that reading speed was not a unitary concept, researchers began changing the focus of their studies. Instead of trying to demonstrate that fast readers are the best readers (Shores, 1950), researchers began searching more syste- matically for factors upon which reading rate depended. Reading rate was found to depend on the nature of the material, the purpose for reading, rate training, and reader characteristics. 28 Nature of the Material Blommers (1944) found that good comprehenders slowed down as material became more difficult, whereas poor comprehenders continued reading at the same relative rate, from a nonstatistical comparison of data. Speed depended on an interaction of reading ability and material difficulty. Rothkopf and Coatney (1974) reported that rate decreased in more difficult material, but the more difficult material used in their study appeared to be above the subjects' reading level. (It was rated as being on a college-graduate level, but was being read by college undergraduates.) Hoffman (1978) theorized that in material above a subject's reading level, a strategy of skimming, scanning, or study must be adopted. It appears that the average reader slows down as material becomes more difficult; however, if the material becomes so difficult that it exceeds his reading ability, the average reader will speed up. Rothkopf and Coatney (1974), using average and very difficult material, varied the order of presentation of text. They found that the readability of a text interferes with the rate at which subjects begin reading subsequent material. This rate-distortion effect was especially pronounced when a more difficult passage preceded an easier passage. Blommers (1944), noting similar rate-interference effects, discounted the rate of reading the first paragraph in calculating reading rate for his study passages. Robinson (1941), in a nonstatistical comparison of data, found that college students differentially adjusted their reading 29 rate to the type of subject matter being read. Fiction was read more rapidly than nonfiction (art, history, and geology). Among nonfiction selections, art was read fastest, followed by history, then geology. Robinson concluded that the rate divergence between fiction and nonfiction material was due to difference in the speed of assimilating the ideas expressed. Material with a higher concept density (e.g., science) was read more slowly. Two different history selections, dealing with Canadian and Russian history, were read at the same average rate and with the same general accuracy. Robinson felt that each subject might require a different set of reading skills, but that the skills within any one area (e.g., history) were probably constant. A related theoretical statement was made by Buswell (1937) when he stated that the ability to read in a familiar field is dif- ferent from that of reading in an unfamiliar field. Purpose for Reading Yoakam (1928) maintained that every individual has more than one characteristic reading rate. He has at least four general rates, which vary with reader purpose: skimming, rapid reading, normal reading, and slow careful reading. Skimming is the most rapid rate and is used to obtain a quick survey or to locate a specific item. Rapid reading is used in reading for main ideas or in rereading; it is used with familiar or easy material. Normal reading occurs when reading to obtain both main ideas and details; it is that rate habitually used by a reader for most recreational and easy to mod- erately difficult factual reading. Careful reading, the slowest 30 rate, is used in difficult text, for high accuracy, or for critical evaluation; very little material is worthy of such detailed reflec- tion (Adler & VanDoren, 1972). The skilled reader, then, is the efficient reader; he can vary his speed to suit his purpose. Buswell (1938) added that attention to detail is not always desirable; it slows down rate and accomplishes a purpose that may be much broader than necessary. Some readers need to be taught to vary speed to fit the purpose for reading. McConkie (1973) demonstrated that speed of reading is influ- enced by the purpose for which subjects read. Subjects reading for multiple purposes read most slowly. Jensen (1978) verified that reading speed increases when one reads for the "central contents" only. Subjects can change the type of information they receive from a text, depending on the purpose for which they are asked to read. McConkie (1973) asked college students in repeated trials to practice reading for a specific purpose. He then tested them with a variety of question types to see what type of additional information they picked up. There was no difference in test score totals between groups, but scores on different question types varied significantly. The groups picked up different patterns of information as they read, depending on their assigned purpose. Students reading for facts and students reading for inference adopted a similar strategy: they appeared to be reading for details and making inferences, based on the facts, at a later time. Students reading with the purpose of 31 answering any type of question could recognize if a phrase came from the reading material but did poorly answering factual questions. Rate Training Hill (1964) and Rankin (1970) stated that most readers, even good ones, display inflexible reading rates (they read different mate- rials at the same general rate). McDonald (1960) tested 6,000 sub- jects of all ages in material of varying difficulty, style, content, and purpose. Most readers, regardless of age or text variables, maintained inflexible reading rates. Inflexible reading may be learned through emphasis on oral reading. It may also be related to the type of professional reading one does; it can also reflect a lack of flexibility training (McDonald, 1965). Readers should vary their rate within selections and paragraphs (not only between material categories, and for different purposes) (McDonald, 1963). Flexibility in adapting rate efficiently to content and pur- pose can be learned. Braam (1963) helped high school seniors differ- entially increase their rate in five types of content. Dee-Lucas (1979) showed that college students could adjust their rate of reading to different payoff conditions. Reading rate seems to depend in part upon rate training. A common concern is that rate improvement, however, might negatively affect comprehension. Harris (1968) stated that normal reading speed can be increased 20% to 40% without adversely affecting comprehension. 32 McConkie (1973) found that reading speed could be increased without affecting the information for which one had been instructed to read; the amount of incidental information acquired, however, was reduced. Fridal (1979), in his work with Danish college students, found that concentration improves as a by-product of rate-improvement work (concentration is defined as being attained when the mind is so occu- pied with information processing that it does not wander). Two hundred to 350 wpm can be too slow for good concentration, he stated. Speed changes in his course were from 265 (t 70) wpm to 625 (i 235) wpm. Spache (1962) used the eye-movement camera to study the reading of college students who were graduates of commercial rate-improvement programs. They read study selections at 400-600 wpm with 70% com- prehension, and books at 1,800-2,400 wpm with 50% comprehension. Harris felt these rates were poor for their purposes. So rate can, in general, be increased without disturbing comprehension of goal- related material. Excessive rate "increases" will probably impair comprehension of main ideas and details; these high rates probably reflect skimming strategies. Pauk (1965) compared a short study-skills-only class with a longer class that covered study skills, rate training, and comprehen- sion work. The study-skills-only class showed triple the grade point average improvement of the other group. Pauk was opposed to reading- pacer training, finding that it may interfere with the formation of more useful study skills (a statistically nonsignificant conclusion). Rankin (1963) gave poorly comprehending college students sequential instruction in comprehension-vocabulary and rate. He 33 found that the order of instruction did not affect comprehension- vocabulary scores, but it did affect rate scores, with bigger increases when rate instruction came first. Students studying rate improvement first reported greater grade improvement and greater enjoyment for reading. Thus Rankin advised working on rate, before comprehension, if rate improvement is one of the goals for a student. Harris (1968) advised that rate improvement should be included in a study-skills course only for extremely slow readers or for those who must read volumes of material daily. There is a difference between rate train- ing and flexibility training. Rate training works on improving read- ing speed within a given type of reading (generally within Yoakam's "normal reading" category); flexibility training teaches one to recog- nize the different reading purposes associated with various reading material and then to adopt a speed appropriate to those varying situa- tions. Reader Characteristics The characteristics of the individual reader differ. Poor reading habits, which are reflections of psychological process, include: slow eye movements, many fixations per line, and lip move- ments (Jensen, 1978). These behaviors are not causes of poor reading speed but symptoms of a particular level of reading maturity (Buswell, 1937). Thus eye movements may be used to diagnose reading difficulties or to quantify improvement. Buswell found that moving the lips in silent reading retarded the ability to rapidly fuse "words into ideas"; it coincided with 34 reading in a more word-by-word fashion. Lip movements were present in all educational levels of subjects, but to a lesser extent in the best readers. Subvocalization levels were altered through training, by Watkins (1979), but there was no apparent transfer to rate or com- prehension. Buswell (1951) found that the rate of reading may be related to the rate of thinking. Slow readers tended to be slow thinkers, whereas fast readers showed no tendency to be fast or slow thinkers, as a group. Fast thinkers tended to be fast readers, however, whereas slow thinkers tended to be low-average-rate readers. "To some extent, rate of reading is related to rate of thinking. It does no good to try to read faster than one can assimilate ideas" (Harris, 1970). Other possible causes of slow reading, besides rate of thinking, are learned inflexibility and habit carryover (Buswell, 1951). Gibson and Levin (1975) quoted a study finding that what a person gains from an Evelyn Wood's speed reading course may be related to personality characteristics (using the Rokeach Dogmatism Scales and Shostrom Personal Orientation Survey). Holmes (1954) tested college psychology students and found that slow readers were slightly more impulsive. Maxwell (1978) found a somewhat opposite result; impulsivity was correlated with initial scanning time, r = .65. Thus faster scanners tended to be more impulsive. Slow scanners applied improvement strategies in a more rigid manner, more frequently persisting in their original strategy, and tending to look at each line (even when supposedly applying the "Z" strategy). No conclusions can be drawn about reading speed as it relates to personality variables. 35 Summary From the studies reviewed in this section, reading rate is seen to depend on material difficulty; type of material being read (fiction vs. nonfiction and type of subject matter); readability of previously read text; purpose (however, with many readers being inflexible, instruction on reading for purpose may be needed); training; level of reading maturity, as reflected by psychological indicators of process; and rate of thinking, to some extent. There is some possibility that rate may also be related to personality factors, although this has not been demonstrated with any consis- tency. Reading Rate as a Factor in Determining Comprehension Studies in this section were divided into those assessing comprehension when subjects were reading at their natural rates and when subjects were reading at assigned rates. Reading rate and college achievement are then discussed. Natural Rates Natural rate has been defined as the rate at which one habitually reads textual information for understanding (King, 1916). A few studies have examined the differences between fast and slow readers reading at their natural rates. Slow readers tend to be inflexible, whereas fast readers tend to be flexible in adjusting rate to varying reading material. Robinson (1941) compared the slowest 20% and the fastest 20% of a sample of college undergraduates with respect to their average 36 reading speeds in various material. The slowest 20% varied their rates very little between different types of material: geology-- 159 wpm, history--l67 wpm, art--l66 wpm, and easy fiction--175 wpm. Only the difference between science and fiction reading rates was statistically significant. The fastest 20% of readers, however, showed great variation in their rates: geology--294 wpm, history-- 312 wpm, art--33S wpm, and easy fiction--385 wpm. All differences were statistically significant. Fast readers seemed to be varying their reading strategy with textual factors; slow readers seemed not to be, in general. Natural slow and fast readers also demonstrate different strategies in reading fiction and nonfiction materials. In easy fic- tion, slow readers increased their average rate by only two words per minute between the first and last third of a nine-minute reading period. Fast readers, however, increased their average rate by 28 words per minute. Having received a mind set for the story line early, fast readers perhaps read the rest of the story more selec- tively, attending more exclusively to central ideas. (Robinson tested "main meanings" rather than "unimportant detail.") In non- fiction material (geology, history, art), fast and slow readers both maintained their initial rates, with the fasts increasing their rate slightly and the slows decreasing their rate slightly. The major difference in rate patterns in nonfiction material was that the fastest readers adjusted their rate immediately to varying textual content, whereas the slowest readers used the same general rate 37 regardless of subject matter. Natural rate seems to affect reading- flexibility patterns. Assigned Rates As was suspected from Robinson's (1941) study, slow and fast reading speeds seem to interact with the type of information gleaned from the text. McConkie (1973) had students read for different pur- poses and rewarded them for doing this more rapidly or more slowly (for greater comprehension). No time limits were set. He found that increasing one's speed does not affect recall of purpose-related material. Increasing speed does lower the amount of incidental (non- purpose related) material recalled, however. Thus, when reading with no time limit, the average good reader would be expected to recall more from a slower reading than a faster reading of given material. Randomly assigned college undergraduates read an article covering the same topic (Rothkopf, 1974). The second half of their articles was identical, but the first half differed in difficulty. In testing comprehension of the second half of the article, subjects who had read difficult material first read this second material more slowly (292 vs. 359 syllables per minute) but with greater comprehen- sion (+4.1 vs. +3.5) than subjects reading easier material first. The higher comprehension may have been due, at least in part, to the slower reading rate. (Instructions were to read rapidly to learn "the essential elements of the content.") 38 Dee-Lucas (1979) wanted to know more specifically how a reader's comprehension would be affected if he were induced to read more rapidly or more slowly. She had randomly assigned college stu- dents to read history material under two reward conditions: (1) com- prehend at greater speed or (2) comprehend well regardless of speed. Comprehension was measured by analyzing the propositional content of written recall. The readers rewarded more for speed read at the average rate of 251 wpm, recalling 23% of all propositions. The readers rewarded less for speed read at the average rate of 120 wpm, recalling 37% of all propositions. Slower readers recalled a greater volume of propositions. Specific propositions recalled at 120 wpm were most likely to be recalled at 251 wpm, whereas propositions poorly recalled at 120 wpm were even less likely to be recalled at 251 wpm. Slower readers recalled more incidental information than faster readers. Faster readers recalled a greater proportion of case propositions (i.e., those stating action), though not a greater amount of information central to the meaning of the passage. Both groups had the same proportion of accuracy in recall. Thus, Dee-Lucas surmised that the information each group chose to examine was pro- cessed with equal thoroughness. However, the fast group, having less time, had to select a "smaller subset of information" to process. The fast group most likely attempted to identify that information "most central" to the passage and most likely to be needed for the later test (i.e., the story line). In this same study, subjects reading at the faster rate reported only half the implications of subjects reading at the slower 39 rate. However, the structure of recall (relatedness of recall sen- tences) was equally good for both groups. Slower readers, no doubt, had more time to reflect on the relationship of incidental ideas. The earliest studies of rate, as it affects comprehension, were done by King (1916, 1917). He did not accept the then—current assumption that faster readers got more from their reading than slower readers. His research questions were ahead of their time and seemingly discounted (Eurich, 1930). King (1916) randomly assigned college students to slow and fast reading conditions. The fast group read about twice as fast as the slow group. In material read by both groups, the "slows" recalled 53.3% of requested details (facts, “specific points"), whereas the "fasts" recalled 45.7% of details. "Slow" and "fast" groups were next divided according to accuracy of recall, within each group. The segment of the "slow" group compre- hending most poorly was still superior to the respective segment of the "fast" group (43.9% vs. 35%). The best-comprehending "slow" readers slightly outperformed the best "fast" readers (65% vs. 61.5%). King analyzed the data of a second similar set of experiments (1917) according to actual speeds achieved by his assigned-rate readers. Comprehension results were divided into quartiles according to rate. The lowest quartile of rate subjects recalled more details accu- rately than the highest quartile (67% vs. 60%). For the average reader, slower reading appeared to lead to greater recall of details. King (1916) identified those readers in the "fast" and "slow" groups who were reading at their natural rates and those who were forced to read at unnatural rates, in their assigned groups. When a naturally 40 slow reader was made to read fast, he was equally likely to recall more or fewer details. However, when a naturally fast reader was made to read slowly, he most generally remembered fewer details of the text. Interfering with the natural reading rate of college sub- jects was a greater handicap to fast readers than to slow readers. The reading strategy of fast and slow readers must be different. It would appear that the reading strategy of fast readers is more spe- cifically oriented, King concluded. Rate and College Achievement In a study involving 68 pairs of Harvard freshman men, intel— ligence was controlled, and reading ability was compared with scholas- tic achievement in English, history, government, or economics courses. (Subjects were chosen in pairs so that IQ was equivalent, but one of the pair was scoring 2 grade points below the other in the given subject.) Overall, faster readers had slightly better achievement. Slower readers achieved slightly better in English and economics. (The economics professor in question stressed high accuracy in fine details; since detail comprehension has been said to be aided by slower reading, the slight advantage for slower readers is not sur- prising.) Faster readers achieved slightly better in history and significantly better in government. In general, rate of reading was not significantly associated with success in these high-density reading courses (Anderson, 1941). 41 Summar Slow readers seem to have inflexible rates, whereas fast readers seem to have generally flexible rates. In easy fiction, fast readers have seemed to benefit from early knowledge of the story line, enabling them to increase their rate more as they read. Slow readers continue. through the whole story at about the same rate. Rate of reading seems to affect the quantity and quality of information obtained. Slower readers recall more of the text, both main information, details, and implications. This effect is espe- cially apparent when detail and implication recall are considered. Slow readers have consistently shown an advantage when recall of details and knowledge of implications are evaluated. Fast readers (or those required to read rapidly) seem to zero in on items of central importance to the tapic. This raises a question: do habitual rates of slow readers reflect a preference for knowing details and the inferences than can be built from them? All readers seem to structure their recall with equal effectiveness, regardless of rate. We do not know if there are differences in micro- and macrostructural recall, however. If time spent on the task were compared with volume of recall attained (compare data in King, 1916, and Dee-Lucas, 1979), the fast reader is more efficient. He remembers a greater amount of informa- tion for each unit of time spent reading. If time limits are removed from consideration, then the slow reader has the comprehension advan- tage. Slow reading does not seem to have prevented college students 42 from making equally good grades as fast readers, though the effort to attain their goals may be greater (in terms of time on task). Summar If psychologically mature readers continue to read slowly in text written within their reading ability, it would seem the explana- tion must be elsewhere than with difficulty in processing. Slowness in silent reading, for a given individual, could be due to one or more of the following factors: 1. type of material generally read (large amounts of techni- cal reading tend to induce a slower general rate); 2. bad habits stemming from inflexible training (excessive oral-reading training, lack of reading-flexibility training); 3. habitual use of auditory cues to aid comprehension and recall; 4. a slower rate of thinking (this is not well established, however); and 5. preferred style of thinking is applied to most reading situations (whether it is efficient of time or not). This could be explained by: a. a preference for knowing details and formulating inferences from them (building a structure detail by detail, inference by inference); b. a dissatisfaction with cursory knowledge of main ideas and their relationship (building a structure by first putting up the main-idea framework, then filling in major details). CHAPTER III METHODOLOGY Introduction This section contains a description of the methodology used in conducting the study. Chapter III is arranged in six main sections. The general population is first described. Next, study procedures are specified, including sample-selection procedures and vision-screening and task procedures. Third, the materials used are described, includ- ing development and validation of task questions. The fourth section outlines the study design. The fifth section presents data-analysis methods employed. The last section states the hypotheses formulated for this study. A summary of methods concludes the chapter. Population All subjects in this study were in their senior year in one of eight high schools or central schools. Six of these schools were located in a county in southwestern New York, the seventh was in a neighboring county in southwestern New York, and the eighth was in a neighboring county in Pennsylvania. All schools were within 80 miles of each other. Included in the sample were rural consolidated schools, small village schools, a suburban school, and a city school. Senior-class enrollments at these schools, respectively, were 39, 73, 84, 86, 122, 43 44 134, 162, and 250. For government-funding purposes, all these areas sampled are classified as part of the Appalachian poverty region, though subjects did not necessarily come from impoverished back- grounds. Although it is somewhat isolated, there is much vitality and opportunity in this region of the country. There are several cultural and educational advantages. The Chautauqua Institution, a summer music-arts-religion community of national renown, is located within the general area. An extensive two-county library system distributes learning materials by several methods. There is a Nature Center, patronized by a world-renowned ornithologist, which has an extensive environmental-education outreach in area schools; the center also offers free summer nature day-camp experiences to all area youngsters. A community college, a business college, practical and registered nurse programs, and two universities are also within this region. Procedure To obtain a testing sample for this study, superintendents of 12 area schools were telephoned, and the study was explained to them. Eight indicated willingness to have their schools participate, upon approval of the principals involved. Contact was then made by phone with the principals. All agreed to participate, and meetings were set up with relevant English and reading personnel from each school. Agreements reached by phone or in meetings were confirmed by follow-up letters to the persons who had been assigned to liaison duties. 45 To make participation more attractive, schools were offered the results of group screening data together with recommendations for their reading programs suggested by the data. Since most of the schools had no reading consultant and had not tested these students' reading since their ninth-grade competency test, this offer was well received. After testing had been completed at a school, the group data were explained in formal meetings or informal gatherings, depending on the wishes of cooperating personnel. All administrators were assured that no comparisons would be made between specific schools or between individual students and that the names of participating schools, school personnel, and students would remain anonymous. The testing program at each school involved two separate days per school, except for the smallest school, where all testing was completed in one day. One day of screening potential subjects was followed, within a week, by a day of individually testing all students selected as subjects. All data were collected in the first semester of the 1982-83 school year. Sample-Selection Procedure Participating schools were asked to make available for screen- ing all seniors in the top quarter of their class, according to grade point average. If, due to scheduling considerations, this was not possible, they were requested to make available the seniors enrolled in 12th-grade English Regents classes. Potential subjects were then either brought together for screening in one large classroom or 46 screened during the time they were scheduled for 12th-grade English. The object of the screening procedure was to obtain subjects who were good readers for the research task. Asking for students in the highest quartile of their class was meant to give a higher return of acceptable subjects per individual screened and to have added indi- cation of their maturity as readers by their level of scholastic achievement. The Nelson-Denny Reading Test (NDRT), Form B, was administered as a screening device (Nelson, 1960). Vocabulary, Comprehension, and Rate subtests were given to all groups of potential subjects. The Rate score was based on the first minute of reading the first paragraph of the Comprehension subtest. Each screening required 45 minutes of administration time, including explanations to the group about the purposes of the study. To enhance motivation for this testing, students were informed that they would each receive a complete set of their Nelson-Denny results, together with interpretation, at a later date. Since most subjects were planning to attend college, such feedback proved to be of substantial interest to most of them. Students had been informed by school personnel beforehand that participation was voluntary, but the positive benefits to be obtained by participation were emphasized. Any student wishing to be excluded from this project, however, was scheduled into a study area during screening. After the NDRT had been administered and scored, students were selected to be potential subjects in the study, based on their 47 results. To be selected, students had to achieve a minimum grade- 1evel score of 12.0 on the Comprehension subtest, as well as a Total Reading score that was greater than or equal to their grade place- ment (i.e., 12.1 to 12.4, depending on month of testing). School personnel scheduled these students for individual visual testing and task participation on an ensuing day. Students were scheduled so as to appear one every 10 to 20 minutes. Visual Testing and Task Procedure Liaison personnel at each school had been asked to arrange a room for visual testing and task performance for this second and final phase of the study. They were asked to provide a quiet room (size unimportant), with a table and two chairs, and another table and four chairs. It was requested that the room have an electrical outlet for the vision machine. The table with two chairs was used for visually screening students, and the table with four chairs was used by stu- dents performing their reading and question-answering tasks. Students who had passed the Nelson-Denny group screening and who had been scheduled for individual testing were responsible for keeping their appointments. Some schools reminded students who were late, whereas others did not. Only a few eligible students failed to be included in this second phase due to nonappearance. Upon arriving in the testing room, students were asked whether they required corrective lenses for reading. If reading lenses were required, students were not permitted to continue without them. Some students who wore contact lenses mainly for distance vision, but kept 48 them in for near vision too, were permitted to wear them for this reading task as well, if they successfully passed the near-point screening. Visual screening consisted of the following Keystone Tele- binocular Near Point Tests: Fusion, Usable Vision--Right Eye, Usable Vision--Left Eye, and Usable Vision--Both Eyes. Students with acuity errors below the 80% level in Usable Vision--Both Eyes were dismissed from the population sample. (Only two students were thus excluded.) Students with underconvergence abounded. They were admitted into the study as long as they saw only one image at nearpoint in Usable Vision--Both Eyes. After successfully concluding the visual screening, students were ushered to the other table in the room. There they were presented with a six-and-one-half-page typed article. They were instructed to read the article in the manner they normally read such material, but only to read through the article once. They were also asked not to look back at any page once they had passed it. They were told that their reading would be timed and that they would be asked a few ques- tions afterward. (Appendix A lists the precise directions.) Each student's reading was timed with a stop watch. After completing the article, having the time recorded, and returning the article, the subject was given a booklet containing 16 multiple- choice questions. (Appendix 8 contains a copy of the question booklet.) There was no time limit for answering the questions, just as there had been no time limit for reading the text. Both tasks were performed at a rate chosen by the student. 49 When the questions were answered, the student was given a pass to return to class. Frequently, there was time to discuss an individual's Nelson-Denny results at this point and to give some advice concerning preparing for the reading demands of college. How- ever, whether or not this was possible, all students participating in any phase of the project were supplied through the school contact person with a printed report of Nelson-Denny results, together with a few specific suggestions for improving their vocabulary, comprehen- sion, and/or rate. The whole procedure was carefully prepared so that as one student began reading, another could be tested visually. The rooms were not always completely insulated from outside noise, but approxi- mated the situation most students face when taking tests in a class- room setting. Each subject who passed the NDRT screening and the visual screening was assigned a number. All data relating to each individual were then recorded by that number. School counselors provided rank- in-class data and available Scholastic Aptitude Test data by subject number. Counselors kept the list linking student names with their assigned numbers and destroyed the list when all data were collected. This procedure assured individual anonymity in the data-analysis phase of this study, as well as compliance with federal human-subjects requirements. 50 Materials For the reading task, a well-written and -organized article of at least 1,500 words was sought. Several criteria were considered in selecting the article. It should be high-interest, general-study- type material that integrated concepts from several content areas. It should be written at a 12th-grade reading level and be unfamiliar, in terms of exact information, to the study population. A Scientific American article entitled "The Black Death" was found in an auxiliary college text (Langer, 1968) that contained sev- eral Scientific American articles included for their high readability, good structure, and general interest. The article was too long at 3,498 words but with the deletion of some digressions was pared down to 2,698 words (4,385 syllables). "The Black Death" met all the desired criteria; it was selected as the textual material to be read by subjects in this study. The article discussed the great plague years that began in Europe in 1348. After first reviewing medical/historical information, the discussion was divided into demographic, economic, psychological, moral, and religious effects of the plague, both from the standpoint of short-term effects and long-term effects. The authors hinted at an application to our present times when we are faced with the poten- tial threat of mass death from atomic war. Readers are asked to consider the great plague as an example of human stress behavior in an atmosphere of all-encompassing disaster. (See Figure l for a dia- gram of the relationship of major textual organizing elements.) mmamv mco_umpmgcou :omgewa .umwk wv:u_uq< u_umm_o;um mmumuwccw h=ou sounds go» .»_:o m:o_um_mccou =u_; .kuu_q5ou m_ acegu men do mumm acmgtumw_ .mcopuopog aceu_uwcm_m:oc so» :m>wm x—co m? mzpm> uumxm mgu “mm:~m> cowumpmccou cone: vwamw_ use mmwuw—_neacce .cmm: mew: .mucwvzum ocw>o_zue-cmgmwg uwumuwvcp meansac cop—mam mean; .mmo—u :_ xcmc mamucougmn cc co_u -amuxm mgu sup: .mcmummc Loewe» co acmew>mpsua segue; umumu_ucw mcmnE=c Lacmw; .mcowam—occou amen» co» tom: came 2mg on» :_ ”ouoz Rowe.v Auexv mopo. in cmnp. mmnp. mNNp. PFNN.1 mmop. momw. NPNO. —-—. wpm—. wsauuagumogumz AemF.v . A_zv came. own—. in FNOF. ammo. mNp—.i m—mp. weep. NmNp. oo~—. ##NF. ovum cwcz . . A-_.V guess weep. Nmmp. pmop. in weep. ammo.n mwem. meN. O¢vo. ooom. omwp. msauuagumOLUwz Amme.v “.mo.v Aomm.v . on pspo.n mN~—. ammo. movp. in —mwo.n comp. nmwm. upmp. mm—N. Noam. pwmuwo A_m_.v .e. Amao.v Am_~.v ANN_.V Axeee xv mafia. FPNN.1 mN—p.i ammo.u pmwo.u in Nvom.u enmv.u emop.u mmwN.n eo¢N.i xcmm mmmpu N a # Ammo.v rot k that Azihih huczv momm. mmo—. pmo—. mmop. mmv—. ommm.» mwcm. —~No. omom. coca. once. chgu—anmuo> 02 mean mcwuomm um: Hz ow: a xcmm & zihuh cowmcmnmcasou mmcaoum_v we mquu Lao» .mmpne_cm> acm5m>mwgue .mupzmms ammo cm~wcsmccaum comrade mco_um_wcgousmuc_tt.~ upon» 65 Table 2.--Frequency data for population variables. n. ”ma. 3:323:32. NDRT Vocabulary 132 73.36 16.64 NDRT Comprehension 132 77.97 15.26 NDRT Total Reading 132 76.95 13.79 NDRT reading rate 131 272.17 86.19 SAT Verbal 110 491.91 77.58 SAT Math 110 557.55 92.94 Percentage Class Rank 132 15.63 11.67 Detail Questions 132 2.37 1.03 Microstructure Questions 132 1.98 0.95 Main Idea Questions 132 2.65 0.89 Macrostructure Questions 132 1.64 0.87 Task reading rate: words per minute 132 218.40 63.46 syllables per second 132 5.92 1.72 aMaximum number of subjects was 133. (See raw data note--Appen- dix H. Hypotheses and Statistical Tests Null Hypotheses Ho 1.1: There will be no relationship between Nelson-Denny Reading Test Vocabulary percentile scores, and the rate at which an extended discourse is read, for 12th- grade students who are mature readers. The Pearson correlation relating NDRT Vocabulary scores and the rate of reading an extended discourse was r = .2803, p = .001, based on 132 subjects. 66 The null hypothesis was rejected. There was a statistically significant relationship between NDRT Vocabulary percentile scores, and the rate at which an extended discourse was read, for 12th-grade students who were mature readers. The relationship was low and posi- tive. The higher the NDRT Vocabulary percentile score of 12th-grade students who were mature readers, the faster those students tended to read extended discourse. Ho 1.2: There will be no relationship between Nelson-Denny Reading Test Comprehension percentile scores, and the rate at which an extended discourse is read, for 12th-grade students who are mature readers. The Pearson correlation relating NDRT Comprehension scores and the rate of reading an extended discourse was r = .3832, p = .001, based on 132 subjects. The null hypothesis was rejected. There was a statistically significant relationship between NDRT Comprehension percentile scores, and the rate at which an extended discourse was read, for 12th-grade students who were mature readers. The relationship was moderate and positive. The higher the NDRT Comprehension percentile scores of 12th-grade students who were mature readers, the faster those students tended to read extended discourse. Ho 1.3: There will be no relationship between Nelson-Denny Read- ing Test Total percentile scores (Comprehension and Vocabulary combined), and the rate at which an extended discourse is read, for 12th-grade students who are mature readers. The Pearson correlation relating NDRT Total Reading scores and the rate of reading an extended discourse was r = .3906, p = .001, based on 132 subjects. 67 The null hypothesis was rejected. There was a statistically significant relationship between NDRT Total Reading percentile scores, and the rate at which an extended discourse was read, for 12th-grade students who were mature readers. The relationship was moderate and positive. The higher the NDRT Total Reading percentile scores of 12th-grade students who were mature readers, the faster those students tended to read extended discourse. Ho 1.4: There will be no relationship between Nelson-Denny Reading Test reading rate scores, and the rate at which an extended discourse is read, for 12th-grade students who are mature readers. The Pearson correlation data relating NDRT reading rate scores and the rate of reading an extended discourse were r = .4018, p = .001 , based on 131 subjects. The null hypothesis was rejected. There was a statistically significant relationship between NDRT reading rate scores, and the rate at which an extended discourse was read, for 12th-grade students who were mature readers. The relationship was moderate and positive. The higher the NDRT reading rate of 12th-grade students who were mature readers, the faster those students tended to read extended discourse. Ho 1.5: There will be no relationship between Scholastic Aptitude Test Verbal scores, and the rate at which an extended discourse is read, for 12th-grade students who are mature readers. The Pearson correlation data relating SAT Verbal scores and the rate of reading an extended discourse were r = .3247, p = .001, based on 110 subjects. The null hypothesis was rejected. There was a statistically significant relationship between SAT Verbal scores, and the rate at 68 which an extended discourse was read, for 12th-grade students who were mature readers. The relationship was moderate and positive. The higher the SAT Verbal scores of 12th-grade students who were mature readers, the faster those students tended to read extended discourse. Ho 1.6: There will be no relationship between Scholastic Aptitude Test Math scores, and the rate at which an extended dis- course is read, for 12th-grade students who are mature readers. The Pearson correlation data relating SAT Math scores and the rate of reading an extended discourse were r = .1902, p = .023, based on 110 subjects. The null hypothesis was rejected. There was a statistically significant relationship between SAT Math scores, and the rate at which an extended discourse was read, for 12th-grade students who were mature readers. The relationship was low and positive. The higher the SAT Math scores of 12th-grade students who were mature readers, the faster those students tended to read extended discourse. Ho 1.7: There will be no relationship between percentage rank in class, and the rate at which an extended discourse is read, for 12th-grade students who are mature readers. The Pearson correlation data relating percentage rank in class and the rate of reading an extended discourse were r = .0799, p = .181, based on 132 students. The null hypothesis was not rejected. There was no statis- tically significant relationship between percentage rank in class, and the rate at which an extended discourse was read, forthh-grade students who were mature readers. There was a statistically nonsignificant 69 trend for students who ranked higher in the class to be somewhat slower readers. (See note on Table 1.) Ho 2.1: There will be no relationship between the number of para- phrased detail questions correctly answered after reading an extended discourse, and the rate at which that dis- course is read, for 12th-grade students who are mature readers. The Pearson correlation data relating number of detail ques- tions correctly answered and the rate of reading an extended discourse were r = -.0171, p = .423, based on 132 subjects. The null hypothesis was not rejected. There was no statis- tically significant relationship between number of paraphrased detail questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. Ho 2.2: There will be no relationship between the number of micro- structure questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. The Pearson correlation data relating the number of micro- structure questions correctly answered and the rate of reading an extended discourse were r = .1609, p = .033, based on 132 subjects. The null hypothesis was rejected. There was a statistically significant relationship between number of microstructure questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. The relationship was low and positive. The faster extended discourse was read by 12th-grade students who were mature readers, the more microstructure questions those students tended to answer correctly. 70 Ho 2.3: There will be no relationship between the number of main- idea questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. The Pearson correlation data relating the number of main-idea questions correctly answered and the rate of reading an extended dis- course were r = .0896, p = .154, based on 132 subjects. The null hypothesis was not rejected. There was no statis- tically significant relationship between number of main-idea questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. Ho 2.4: There will be no relationship between the number of macro- structure questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. The Pearson correlation relating the number of macrostructure questions correctly answered and the rate of reading an extended discourse was r = .0165, p = .426, based on 132 subjects. The null hypothesis was not rejected. There was no statis- tically significant relationship between number of macrostructure questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. Ho 3.1: Scholastic Aptitude Test Verbal scores will have no effect on the relationship between the number of paraphrased detail questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. Table 3 reports the Pearson and partial correlations between paraphrased detail comprehension and reading rate. The zero-order 71 partial correlation between detail and rate was .0591. The partial correlation controlled for SAT Verbal scores was -.0160. Neither of these correlations was statistically significant. The null hypothesis was not rejected. SAT Verbal scores had no significant effect upon the statistically nonsignificant relation- ship between number of paraphrased detail questions correctly answered after reading an extended discourse, and the rate at which that dis- course was read, for 12th-grade students who were mature readers. (Nie et a1., 1975, p. 304, explained the rationale for this test.) Although not significant, partialing out SAT Verbal scores appeared to have the effect of nullifying any positive relationship of detail-rate correlations in this sample. (Note Pearson correlation in Table 3.) Ho 3.2: Scholastic Aptitude Test Verbal scores will have no effect on the relationship between the number of microstructure questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. Table 3 reports the Pearson and partial correlations between microstructure and reading rate. The zero-order partial correlation between microstructure and rate was .1811, p < .05. The partial correlation controlled for SAT Verbal scores was .1278, p = .114. Only the first of these two correlations was statistically signifi- cant. The null hypothesis was rejected. SAT Verbal scores had an effect upon the statistically significant relationship between number of microstructure questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. When the 72 Table 3.--Corre1ati0ns between four types of discourse comprehension and discourse reading rate, assessed with different statis- tical procedures, and including variance for statistically significant factors. Types of Comprehension Variance , . Micro- Main Macro- Micro- COFFEIatIO" M9th°d Details structure Idea structure struchre r r r r r pearsona -.0171 .1609* .0896 .0165 .0259 Zero-order partialsb .0591 .1811* .0590 -.0180 .0328 Partials controlling for SAT Verbal -.Ol60 (.12784) .0045 (-;0933) .0163 scoresC p ' p ' Part'als C°"'F°“l”9 .0629 .1843* .0659 -.0112 .0340 for percentage rank in classc Note. Pearson correlations are presented only for comparison with zero-order (simple) partial correlations. aBased on 132 subjects. b Based on 121 subjects. cBased on 89 subjects. *p < .05. Partial correlations were based on zero-partials, not Pearson correlations. 73 microstructure/rate relationship was controlled for variance in SAT scores, the formerly significant relationship became nonsignifi- cant. Partialing out SAT scores reduced the percentage of variance explained by microstructure/rate relationship by one-half (see micro- structure r2 column in Table 3). H0 3.3: Scholastic Aptitude Test Verbal scores will have no effect on the relationship between the number of paraphrased main- idea questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. Table 3 reports the Pearson and partial correlations between main-idea comprehension and reading rate. The zero-order partial correlation between main idea and rate was .0590. The partial corre- lation controlled for SAT Verbal scores was .0045. Neither of these correlations was statistically significant. The null hypothesis was not rejected. SAT Verbal scores had no significant effect upon the statistically nonsignificant relation— ship between number of main-idea questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. Although not significant, partialing out SAT Verbal scores appeared to have the effect of decreasing the main-idea/rate correlation in this sample. H0 3.4: Scholastic Aptitude Test Verbal scores will have no effect on the relationship between the number of macrostructure questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. Table 3 reports the Pearson and partial correlations between macrostructure and reading rate, for 12th-grade students who were mature readers. The zero-order partial correlation between 74 macrostructure and rate was -.0180. The partial correlation con- trolled for SAT Verbal scores was -.0900, p = .198. Neither of these correlations was statistically significant. The null hypothesis was not rejected. SAT Verbal scores had no significant effect on the statistically nonsignificant relation- ship between number of macrostructure questions correctly answered after reading an extended discourse, and the rate at which that dis- course was read, for 12th-grade students who were mature readers. Although not significant, partialing out SAT Verbal scores appeared to have the effect of decreasing the macrostructure-rate correlation in this sample; the partialed negative relation began to approach significance. Ho 4.1: Percentage rank in class will have no effect on the rela- tionship between the number of paraphrased detail questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th- grade students who are mature readers. Table 3 reports the partial correlations between detail and reading rate. The zero-order partial correlation between detail and rate was .0591. The partial correlation controlled for percentage rank in class was .0629. These correlations were of approximately the same magnitude. Neither of these correlations was statistically significant. The null hypothesis was not rejected. Percentage rank in class had no significant effect on the statistically nonsignificant relationship between number of paraphrased detail questions correctly answered after reading an extended discourSe, and the rate at which 75 that discourse was read, for 12th-grade students who were mature readers. Ho 4.2: Percentage rank in class will have no effect on the rela- tionship between the number of microstructure questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th- grade students who are mature readers. Table 3 reports the partial correlations between microstruc- ture and reading rate. The zero-order partial correlation between microstructure and rate was .1811, p < .05. The partial correlation controlled for percentage rank in class was .1843, p < .05. These correlations were of approximately the same magnitude. Both of these correlations were statistically significant. The null hypothesis was not rejected. Percentage rank in class had no significant effect on the statistically significant relationship between number of microstructure questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. Ho 4.3: Percentage rank in class will have no effect on the relationship between the number of main-idea questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th- grade students who are mature readers. Table 3 reports the partial correlations between main idea and reading rate. The zero-order partial correlation between main idea and rate was .0590. The partial correlation controlled for percentage rank in class was .0659. These correlations were of approximately the same magnitude. Neither of these correlations was statistically significant. 76 The null hypothesis was not rejected. Percentage rank in class had no effect on the statistically nonsignificant relationship between number of main-idea questions correctly answered after read- ing an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. Ho 4.4: Percentage rank in class will have no effect on the relationship between the number of macrostructure ques- tions correctly answered after reading an extended dis— course, and the rate at which that discourse is read, for 12th-grade students who are mature readers. Table 3 reports the partial correlations between macrostruc- ture and reading rate. The zero-order partial correlation between macrostructure and rate was -.0180. The partial correlation controlled for percentage rank in class was -.0112. These correlations were of approximately the same magnitude. Neither of these correlations was statistically significant. The null hypothesis was not rejected. Percentage rank in class had no effect on the statistically nonsignificant relationship between number of macrostructure questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. It was interesting, however, that while partialing out percentage rank slightly increased the rate relationship with detail, microstructure, and main idea, it had the effect of slightly decreasing the rate/ macrostructure relationship. (Referring to Table 1, it can be seen that slower readers were nonsignificantly higher in their class rank, and persons higher in their class rank got nonsignificantly more macrostructure questions correct.) 77 Directional Hypotheses Hd 1.1: There will be a negative relationship between the number of detail questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. The Pearson correlation between detail recall and reading rate in extended discourse was r = -.0171, p = .423, n = 132. The zero- order partial correlation between detail recall and reading rate was r = .0591, p = .258, n = 121. The directional hypothesis inconsistently predicted the direction of the correlation, and the relationships were not statis- tically significant. There was not a statistically significant nega- tive relationship between the number of detail questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. Hd 1.2: There will be a negative relationship between the number of microstructure questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. The Pearson correlation between microstructure recall and reading rate in extended discourse was r = .1609, p = .033, n = 132. The zero-order partial correlation between microstructure recall and reading rate was r = .1811, p = .022, n = .121. The directional hypothesis incorrectly predicted the direc- tion of the correlation; the correlations were consistently positive and statistically significant. There is a statistically significant 10w positive relationship between number of microstructure questions 78 correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. The faster extended discourse was read by 12th-grade students who were mature readers, the more microstructure questions those students tended to answer correctly. Hd 1.3: There will be a positive relationship between the number of main-idea questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. The Pearson correlation between main-idea recall and reading rate in extended discourse was r = .0896, p = .154, n = 132. The zero-order partial correlation between main idea and rate was r = .0590, p = .258, n = 121. The directional hypothesis correctly and consistently pre- dicted the direction of the correlation, but the relationships were not statistically significant. There was not a statistically signifi- cant positive relationship between the number of main-idea questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. Hd 1.4: There will be a positive relationship between the number of macrostructure questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers. The Pearson correlation between macrostructure recall and reading rate in extended discourse was r = .0165, p = .426, n = 132. The zero-order partial correlation between macrostructure and rate was r = -.0180, p = .422, n = 121. 79 The directional hypothesis inconsistently predicted the direc- tion of the correlation, and the relationships were not statistically significant. There was not a statistically significant positive rela- tionship between the number of macrostructure questions correctly answered after reading an extended discourse, and the rate at which that discourse was read, for 12th-grade students who were mature readers. Summary The null and directional hypotheses described in Chapter III were tested in Chapter IV. The following null hypotheses were not rejected: 1.7; 2.1, 2.3, 2.4; 3.1, 3.3, 3.4; 4.1, 4.2, 4.3, and 4.4. The following null hypotheses were rejected: 1.1, 1.2, 1.3, 1.4, 1.5, 1.6; 2.2; and 3.2. One directional hypothesis correctly and consistently predicted the direction of the relationship--that of main idea (1.3). However, the positive relationship was not statis- tically significant. Two directional hypotheses inconsistently pre- dicted the direction of the relationship, with one calculation method producing a negative correlation and the other method producing a positive correlation. These two directional hypotheses pertained to detail (1.1) and macrostructure (1.4); neither method of calculation produced statistical significance. One directional hypothesis failed to predict correctly the direction of relationship; however, that relationship was statistically significant--that of microstructure (1.2). 80 In Chapter V, the results of this study are discussed and interpreted in the context of related literature and theory. A post hoc data analysis is included, informally comparing the sample means of the 30 slowest and 30 fastest discourse readers for impli- cations. CHAPTER V SUMMARY, DISCUSSION, CONCLUSIONS, AND RECOMMENDATIONS Summary The purpose of this study was to determine whether mature readers vary in the type of information they recall after reading extended discourse at slow or fast rates. The study population was composed of 133 seniors from a variety of high schools. All subjects had at least thh-grade-level reading skills, and most were in the top quarter of their class, based on grade point average. The study population was described based on mean performances on various standardized and achievement measures (Table 2), and then the relationship of these measures to the discourse reading rate was explored (Table 1). Accuracy in recall of each of four types of information was correlated with discourse reading rate to see whether any relationships existed. Then, zero-partial correlations between rate and type of comprehension were tested, controlling for Scholas— tic Aptitude Test (SAT) Verbal scores, or percentage class rank scores, to see whether these achievement variables had any effect on existing relationships (Table 3). Finally, hypotheses predicting the direction of comprehension-rate relationships were examined. Table 4 summarizes the research findings for Null Hypotheses l and 2, Table 5 summarizes the research findings for Null Hypotheses 3 and 4, and Table 6 summarizes the research findings for the directional hypotheses. 81 82 .ucmuwwwcmwm appmuwumwumpm poc mew mommcucmsma cw meowumumsacmucu .muoz -1- cmuownoc uoz msauochmosumz e.m o: Ap_euee teemee...v eeeeenee eez ewe? ewe: m.~ 0: __mumc segues... mmo. u a .nmuomnmm mczpuzcumosuwz N.~ o: -u- emuomnmc uoz Pwmpmo _.N o: Axcms smzop...v empowwmc uoz xcwm mmm—u & N.F o: meeeem Lessee... mmo. u e .eeeuehee eeez e P pmoz ~.p o: =...m>m; mcmumms cmummde ovum mcwcemm smnsaz mmcvvcwd mmcaoumwo saw: mwmmggoax: ”cowumpmsacmpcm umpmpmssou mpnmwcm> .N new _ mommcpoazz __:z mo mummy quwumvumpm mo mu—smms eo acmeszmuu.v «Peep .mmcwucwu mo mucmowwwcmwm Pmuwumwpmum mmcmsu no: ewe was» mcompsmaeou Amucewsm>v copum—msLou cu games mommgpcmcma cw muumwcm .muoz 83 Acowuozums ugmwpmv umpumwos uoz xcmm & mczuuacamocumz e.¢ o: Ammmwcocm psmwpmv cmuumwms poz xcmm x emu? awe: m.¢ o: Ammmmsucm ugmwpmv umpumnoc uoz xcmm N msauuzspmosuwz N.¢ o: Aemeeeeee Seme_mv eeeaehee eoz seem e .Peeeo ..e 0: A:o_uu:umsv voguwwms poz pensm> p H h h .msopue$ :mmusmu so; cmFFocpcou mommgpoaxc Ppsc mo mcompsmasou mucmwse> mo mupammc mo xgmesamiu.m m_nmh 84 .uceuwwwcmwm zFPmuPum_pmum we: mew mew -mgucwsen cw mppzmom .APNP n :V meowumpmsgou Fevucma gmucotosmN new Ammp u cv meowpepmscou comsmme no women .m>wuemo: we: cmgao one new m>muwmoa we: meowumpmssou umumpzupmu oz“ mo mco gasp mmueu -wccw mesa =.:owuomewu pcmumwmcoucwe mo coma m>ms op umpcogms we ecopgumswc Pmsuume on» can: .muoz Acowpumcwu ucmumwmcouch Aeoeeeeeee eseewmoev cowuumgwu m>wuwmoe Acowuumgmu ucwumwmcouch nousogasm uoz umucoaazm uoz empcoaazm 9oz emusoagsm poz w>wuwmoauumc=uoscumocuwz m>wuwmoauummuw ewe: m>wuemmctims3puaspmocowz m>wummmciupwmumo e._ e: m._ e: U N._ I _._ e: wpm”— mmssoom 5 eat: seemeeeeepem we cowaumcwo szuu< llll IIIlIiIIlII'll mme_eeea ..1|‘111.,|clll.|ll]11.1.1:1111 11! tl ii. mama mcwummm mmssoumwo ee_3 seemeo_pe_em to co_uumswo emuumuwsa consaz mwmmzuoazz _e:owpumcwo .mmmmcpoaxc Pecowpumswu mo mupzmms mo xgwsszmut.o mpneh 85 Discussion Research Questions 1.0: Will there be a relationship between standardized and academic performance data, and the rate at which an extended discourse is read, for 12th-grade students who are mature readers? The relationship between speed and comprehension in timed material has been found to be moderate to high when comprehension is measured as the number of correct responses; however, the relationship between speed and comprehension has been found to be negative and low when comprehension is measured as the number of right responses com- pared to the number of items attempted (Letson, 1958; Rankin, 1962). Thus, timed tests contain an inherent bias against the slower reader. For this reason it was expected that Nelson-Denny Reading Test (NDRT) and SAT correlations with discourse rate would be of some positive magnitude. This proved to be the case. NDRT Vocabulary scores and SAT Math scores showed a low sta- tistically significant relationship to discourse reading rate. NDRT Comprehension, NDRT Total Reading, NDRT reading rate, and SAT Verbal scores all showed moderate statistically significant relationships to discourse reading rate. (See Table 1.) Anderson (1941) found slightly better achievement, overall, from faster-reading than from slower-reading Harvard freshman men. In general, however, rate of reading was not associated with success in high-density reading courses. For this reason, it was expected that percentage rank in class would be nonsignificantly related to discourse reading rate, but that faster readers might tend to be 86 ranked higher in their class. This projection was partly borne out; the rank/rate relationship was nonsignificant. However, there was a slight tendency for slower readers to rank higher in their class (r = .0799, p = .181), as reported in Table l. The tendency for this relationship to be negative or positive might well be a function of subject selection. In this study, only good readers and higher achievers were selected as subjects. One would assume that Harvard freshman men also tended to be good readers and high achiev- ers in their high school days, but no such controls were introduced into that experiment. In any case, rate of reading appeared to be unrelated to achievement for either of these populations. The study population was characterized using frequency data from NDRT, SAT, and rank scores (Table 2). The average subject scored between the 63rd and 9lst percentile on NDRT Total Reading (or between grade-level equivalents of 13.0 and > 14.0). The average subject also scored between 414 and 569 on SAT Verbal and between 465 and 650 on SAT Math. The average subject's rank in class was in the top 4 to 27%. (These figures were obtained by subtracting, then adding, 1 $0 to the mean population figures in Table 2.) 2.0: Will there be a relationship between the number and type of questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? Scattered research in this century has begun to produce a pattern of comprehension-type/rate results that appear to allow for prediction of subject recall based on relative speed of reading. King (1916) found that any person reading more slowly (reading at 87 assigned rates) can expect to recall a greater number of details. In timed material, the faster (assigned rate) reader got a greater total volume of details through reading more material, but slower readers got more details per unit of text read. When college upper- classmen and upperclasswomen were requested to read at their natural rates, slower readers recalled a significantly greater number of details than faster readers (showing a moderate negative relation— ship). Dee-Lucas (1979) found greater recall of action-related information (case propositions) by Cornell undergraduate psychology students when they read at fast assigned rates. Subjects assigned to read slowly, however, recalled a greater total number of proposi- tions and twice as many inferences (implied causal relations) as fast readers. Dee-Lucas ascribed this result to the greater time available to slow readers for reflection. She hypothesized that slow readers probably "processed all of the information that they consid- ered relevant very thoroughly prior to the recall task," whereas faster readers were more selective in processing and chose "a smaller subset of information" to process. Dee-Lucas felt that faster read- ers (X'= 250 wpm) probably processed information of central importance to the passage equally as carefully as did the slower readers (X'= 120 wpm). Shores and Husbands (1950) found that fast readers were not the best comprehenders when reading-related problems had to be solved using critical thinking. Carlson (1949) noted that slow- reader performance was better on questions requiring inference. 88 McConkie (1973) found that reading rate depended on the pur- pose for which one read, using college undergraduates reading nggp; tific American articles. Having multiple purposes reduced reading rate the most, whereas reading only for the central contents increased the reading rate. Jensen (1978) verified that reading speed increases when one reads for only central contents. McConkie (1973) also found that students gathered different patterns of information from their reading, depending on their assigned purpose. Both students reading for facts and those reading for inference appeared to read for details initially and to make inferences based on facts at a later time. Students reading to answer any future type of question appeared able to recognize phrases from their reading, but they were poor at answering factual questions. The pattern that appeared to emerge from these studies was one wherein mature slow readers would be expected to read information in greater depth and to recall more details and more information requir- ing inference and critical thinking. Mature fast readers, on the other hand, would be expected to process a smaller subset of material, involving main ideas and "central contents." From this perceived pattern, hypotheses were constructed pre- dicting differential rate relationships with various types of compre- hension. Detail and detail-relationship (microstructure) were predicted to be negatively associated with rate for mature good readers, and main idea and main-idea relationship were predicted to be positively associated with rate. The slower readers were hypothe- sized to be using inductive reasoning to a greater extent, wherein 89 textual ideas were integrated into one's own idea structure, and recall conclusions were formed from a body of one's own data. The faster readers were hypothesized to be using deductive reasoning to a greater extent, wherein main ideas were recognized and processed and served as clues from which to synthesize recall conclusions. In comparing total discourse comprehension with discourse reading rate, no significant difference was found for the whole p0pu- lation (r = .1044, p = .117). No significant relationship was found for detail, main idea, or macrostructure and rate (although the main- idea relationship was consistently positive). However, a statistically significant low positive relationship existed between microstructure (detail-relationship) and discourse rate (Table 1). None of the pro- jected directional hypotheses was supported at a statistically sig- nificant level. Main idea behaved more as predicted than did the other three comprehension variables. Obviously, much remains to be discovered about rate/ comprehension-type relationships. The exact nature of comprehension requirements must be stated in future studies, to allow for proper inference and study reproduction. In addition, populations must be defined with relation to reading ability and possibly achievement level, as well. An example of the difficulties involved is seen with detail questions. This study employed detail questions which required that subjects had understood what they read; this was accomplished by using paraphrase in recall questions. Other studies reporting "detail recall" may have used questions of a much simpler nature, allowing for mere recognition of text phrases. 90 3.0: Will SAT Verbal scores have an effect on the relationship between number and type of comprehension questions cor- rectly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? Since SAT Verbal scores were derived from timed tests, it was assumed that a score bias would exist in favor of faster readers. This result has already been substantiated and reported. It was assumed, then, that partialing out the effect of SAT Verbal scores would have the effect of lowering rate/comprehension-type correlations. As reported in Tables 3 and 6, this assumption was substantiated. When partial correlations were calculated for these mature readers, con- trolling for SAT Verbal scores, all correlations between rate and detail, microstructure, main idea, or macrostructure decreased by .05 to .07. The only modification that affected statistical signifi— cance, however, occurred with the microstructure/rate relationship, which declined to a nonsignificant level from a previous statistically significant low positive relationship. (It is interesting that macro- structure/rate was affected so that it approached significance in a negative direction.) 4.0: Will percentage rank in class have an effect on the rela- tionship between number and type of comprehension questions correctly answered after reading an extended discourse, and the rate at which that discourse is read, for 12th-grade students who are mature readers? In the research examined, grade point average was not signifi- cantly associated with reading rate (Anderson, 1941). This result was replicated in the present study, as has already been reported. It was assumed, then, that partialing out the effect of percentage rank in class would have little if any effect on rate/comprehension-type 91 relationships. In fact, detail, microstructure, and main-idea cor- relations with rate increased slightly when partial correlations were calculated controlling for percentage rank in class, whereas macrostructure correlations with rate decreased slightly. Signifi- cance of simple correlations was not affected by controlling for grade point average. (See Tables 3 and 6.) There were obviously high-achieving students among all segments of this population with respect to rate. ' General Comparisons With Background Studies Much has been written on the topic of rate and comprehension. Early researchers attempted to identify fast readers as the best readers through rate/comprehension-correlation studies. When divergent results were consistently obtained, the focus of the studies gradually changed; researchers began looking for the factors affecting data variability. They found that reading rate depended on many factors, such as the nature of the reading material and the level of reading maturity of subjects. Some studies have investigated differences in comprehension when reading purpose or rate was varied, comparing read- ers who were assigned to read at various rates or comparing slow and fast readers reading at natural rates. Conclusions of such research were presented in Chapter II. Some of these conclusions are briefly reviewed in the following pages so that they may be examined with reference to data from the present study. In the present research, rate and type of comprehension were compared using one extended-discourse selection. The material was 92 lengthy (2,698 words or 4,385 syllables), general-study-type material, of appropriate difficulty for the study population. Rate was calcu- lated based only on time spent reading the discourse passage. It has long been established that such procedures produce the most reli- able rate/comprehension relationships (Tinker, 1939). Several conditions that tend to lower rate/comprehension cor— relations were intentionally incorporated into the extended-discourse- material selection so that any relationships found would more accurately reflect true p0pulation differences: nonfiction material was used (Robinson, 1941), and critical thinking was involved in answering microstructure, main-idea, and macrostructure questions (Carlson, 1949). Tinker (1932) and Rankin (1962) showed that rate/comprehension correlations produced with standardized test material were higher than when such correlations were calculated from extended-discourse mate- rial. The data from this study confirmed this long-standing conclu- sion. The Pearson correlation of NDRT reading rate and Total Reading was r = .2723, p = .001 (131 subjects). The Pearson correlation of extended-discourse reading rate and total comprehension was r = .1044, p = .117 (132 subjects). The NDRT correlation was statistically sig- nificant, displaying a low positive relationship, but the extended- discourse correlation was not. The only type of comprehension found to be significantly related to rate in the extended-discourse task was microstructure: r = .1609, p = .033 (132 subjects). The magnitude of that relation- ship was still below that of NDRT rate/comprehension. In contrast 93 to extended-discourse material, NDRT paragraphs employed strict time limits and more easily comprehended material; both these factors have been shown to increase the rate/comprehension relationship (Preston & Botel, 1951; Rankin, 1962; Tinker, 1939). Rothkopf and Coatney (1974) found that rate decreased in more difficult material. Robinson (1941) reported that material with a higher concept density is read more slowly due to the difference in speed with which the ideas can be assimilated. The average rate of reading the easier NDRT passage (based on the first minute of reading) was 272 wpm, with a standard deviation of 86 wpm (Table 2). The average rate of reading the more difficult extended-discourse mate- rial (based on total discourse reading time) was 218 wpm, with a standard deviation of 63 wpm (Table 2). The relationship between these two rate measures for the study population was r - .4018, p = .001 (131 subjects). This relationship is illustrated in Figure 2, using a scattergram. This study, then, supported the findings that rate decreases in material that is more difficult or that has a higher concept density. Hill (1964) and Rankin (1970) stated that even most good readers showed inflexible rates, reading all material at the same rate. When measured in words per minute, inflexible rates were not displayed by the present population as a whole; the standardized NDRT material was read more rapidly than the extended-discourse material. Table 7 compares data from the present study on the 31 subjects who most consistently read slowly on both NDRT and extended discourse; this group, called the slow-slow group, read below 226 wpm on NDRT NDRT reading rate-words per minute 450‘I 400. 350. 300- 250- 200. 150- 94 .. . O . o . o 0 o ' o o .. o O o . o o o . co ‘0‘. . O o 2 0.0.00 o o a» o O O O 00 O O .0 ° 0.0.30 o O 0 .0 '33:}; .00 O 001 0|! 0' o . .0 O. .0. Q C O O O Q C 0 O . o O U T I f I I I 150 200 250 300 350 400 450 Extended discourse reading rate--words per minute Figure 2.--Nelson-Denny (NDRT) reading rate for each student as a function of his/her reading rate on the study material (N = 133). (Note: Subjects with rates at the extremes were readin at faster rates than could be shown in this graph.1 95 Table 7.--A comparison of mean reaging-rate changes between the slow—slowa and fast-fast subjects in Nelson—Denny and extended-discourse task material, in words per minute and syllables per second. Reading Rates GVOUP Measured in Measurgd in wpm sps Slow-slow Nelson-Denny mean rate 194 i 26 4.77 i 0.62 Discourse task mean rate 159 i 24 4.27 i 0.64 (Percentage rate drop)e (18.6%) (10.5%) Fast-fast Nelson-Denny mean rate 331 i 75 8.30 i 1.70 Discourse task mean rate 295 i 59 7.98 i 1.59 (Percentage rate drop) (10.9%) (3.9%) aSlow—slow subjects read NDRT at less than 226 wpm, and the discourse at a rate less than or equal to 190 wpm (n = 31). bFast-fast subjects read NDRT at more than 226 wpm, and the discourse at a rate greater than or equal to 250 wpm (n = 32). prm = words per minute. dsps = syllables per second. ePercentage rate drop was calculated by comparing the rate dif- ference in each category with the NDRT rate, to show how much group rate decreased when task material was read. (me and sps rate drops are not proportional between groups due to a peculiarity of the NDRT rate task. In the NDRT reading, multisyllabic words became more fre- quent the further one read. Thus, a person reading to NDRT word 331, in the one-minute time limit, read disproportionately more syllables than persons reading to word 194. This difficulty was not encountered in the discourse task as all subjects were timed reading all words.) 96 and S 190 wpm on the discourse task. The 31 slow-slow subjects were compared with 32 fast-fast subjects, who most consistently read rapidly on both NDRT and extended discourse; the fast-fast group read above 226 wpm on NDRT and 2 250 wpm on the discourse task. Both words-per-minute and syllables-per-second rates were calculated and appear in Table 7 to informally test the idea that an individual's reading rate remains the same between different read- ing tasks when measured in syllables per second (Hoffman, 1978; Ritty, 1979). In this study, both those who tended to read consis- tently slowly and those who tended to read consistently rapidly reduced their reading rates for the discourse material, with the slow-slow group showing the greatest average drop in rate (Table 7). This difference has evident whether measured in words per minute or syllables per second. When the percentage rate drop was calculated, greater variability in rate adjustment was produced by measuring in words per minute than in syllables per second: slow-slow displayed an 18.6% vs. 10.5% rate drop; fast-fast displayed a 10.9% vs. 3.9% rate drop. In spite of this observation of measuring-system differ- ences, rate of reading was still varied by the individual, using either method of calculation (Table 7). Rothk0pf and Coatney (1974) reached this same general conclusion. Subjects in this study were instructed to read at a rate natural for them in such extended-discourse material. King (1916) found slow readers reading at natural rates to be less flexible in adjusting their rate to varying materials than fast readers. This conclusion was not supported by data from the present study. Table 7 97 shows that the slow-slow group displayed a greater percentage of rate change between NDRT and task material than did the fast-fast group (18.6% vs. 10.9%), both groups being good readers and tending to be high achievers. Thus the slow-slow group displayed more flexi- bility in rate between NDRT and discourse tasks than did the fast-fast group. As has been mentioned, the rates would be expected to vary between NDRT and the discourse task because of the differences associated with them. NDRT had a strict time limit, consisted of short unrelated paragraphs, and was somewhat less difficult; questions were on the same page as the text. In contrast, the discourse mate- rial had no time limit, consisted of 28 conceptually related para- graphs, and was more difficult; questions were not presented until all discourse had been read. King's conclusions on rate/flexibility relationships were based on various types of study materials with subjects reading at natural rates, without time limits. Although he used college students as subjects, one cannot be sure that they were all good readers (Marshall & Glock, 1978). Within the confines of King's study methods, his conclusions may still be valid. Informal Post Hoc Analysis In this study, subjects were asked to read "The Black Death," a somewhat condensed Scientific American general-study-type article. They were instructed to "read at the rate they normally read such material," in order to answer "a few questions." It is assumed that subjects then chose their own purposes and read at a variety of rates 98 in order to satisfy the objectives they generally set for themselves in such study-type material. The investigator hypothesized that different general styles of reading (patterns of selective attention to information) would be displayed by the population extremes, with respect to rate. This hypothesized phenomenon was tested using Pearson correlations; statis- tical significance was found only for microstructure questions. It was found that faster readers tended to answer significantly more of those questions that involved detail relationship. There was, how- ever, no statistically significant relationship between rate and the following types of comprehension: detail (paraphrased), main idea (synthesized from diverse sectors of the text), or macrostructure (which required inferring major textual ideas and their relationship and interpreting these ideas in the context of unifying concepts. From the results of Dee-Lucas' (1979) and King's (1916, 1917) research, it was anticipated that slow readers would be better at all types of detail comprehension. The present study showed no statis- tically significant rate relationship for details, however, and a rate relationship that favored faster readers for the microstructure material. The scattergrams shown in Figure 3 depict each individual's accuracy in answering each type of question, as a function of his/her extended-discourse reading rate. To formulate theoretical implications and suggestions for further research, the study population was subdivided for analysis. A slow-discourse group was established, containing the 30 slowest readers of extended discourse in the research population; all 30 of Number of questions right per student (N = 132) 99 DETAIL QUESTIONS 4- o 0. on 0‘ co no on O u 0 00 3— 00000 00“...“ 0000’... no 0 2— no 00 0 on.“ a” h “a on O“ o 0 out o o o 1- o 0 w no on 0000 o o 0'. o O O 0 JW 1 l L l l 1 l l MICRO-STRUCTURE QUESTIONS 4”- o 1.. 0 3" o .1 o or o quifllfipqlipoo 1UP Ililtqflllflhlo o 4.0 o o (I 2" our 0 oqllbo d‘llllf."'15’ O'Domp il)flbnloo .1 qu o o l- a goof. oowooogo on 0‘... o O 0-— g .0 .0 O 1. . _1' I I J l I 1 J I s___ MAIN IDEA QUESTIONS 1 4- o oo o not. 000 0000 00.0. 000 o 3" o o 00 “com 0500.. on o co 0. 0 2h- 09 o o w ‘0‘ «a «num- 000 O C! 1- o o to on on o 0 . , . I J J .J l 1 l l MACRO.STRUCTURE QUESTIONS 0 o. a» o p o o o o o. o w M“. '0 *0 ”to. a on o o 00 o “9*. $90.“ at”. filo... o o o o o 00 O o u o 9 I l 1 l l 1 l L 100 150 200 250 300 350 Extended-discourse reading rate in words per minute Figure 3.--Number of questions per comprehension category answered correctly by each student as a function of his/her rate of reading the extended-discourse material. 100 these subjects read the task at or below 170 words per minute. (In Figure 3, this level is indicated with a small mark on the rate scale just to the right of the 150 wpm mark.) A fast-discourse group was also established, containing the 30 fastest readers of extended dis- course in the research population; all 30 of these individuals read the task at or above 260 words per minute. (In Figure 3, this level is indicated with a small mark on the rate scale just to the right of the 250 wpm mark.) Means for these two groups, and the 73 subjects who fell in between them, based on task rate, are reported in Table 8 for all study variables. (NDRT reading rates were disregarded in defining these groups.) Edfeldt (1960) found the average rate of adult speech to be between 170 and 200 wpm and the maximum rate of adult oral reading to be about 250 wpm. Harris (1968) found the average rate of silent reading by college students to be about 280 wpm. Comparing the slow- discourse and fast-discourse groups to these levels produced some clear distinctions. The slow-discourse group was not only reading below the average silent-reading rate of college students, it was also reading at a rate below the average level of adult oral speech-- 170 words per minute, or less, with an average reading rate of 148.3 wpm. The fast-discourse group, in contrast, was reading above the maximum rate for understandable oral speech, and its average silent-reading rate was above that of college students--260 words per minute, or more, with an average reading rate of 323.5 wpm. There should therefore be a psychological difference between these groups, in terms of the manner in which they were processing the information 101 Table 8.--Subpopu1ation means for post hoc comparison of the 30 slowest and the 30 fastest readers of extended discourse, for all study variables. Rate Groups Based on Reading Rate Study Variables of Discourse Task Slow-Discoursea Middleb Fast-Discoursec NDRT Vocabularyd 66.97 73.51 79.77 NDRT Comprehensiond 71.30 76.86 86.70 NDRT Total Readingd 70.30 76.42 84.97 NDRT wpm rate 213.9 277.1 314.6 SAT Verbal 473.6e 484.1f 526.9g SAT Math 526.4e 663.1f 572.3g % Class Rank 16.22 15.43 15.68 Detail 2_.l7_ 2 . 51 fl Microstructure 1:23: 2.03 22131 Main idea 2469_ 2.62 2_§Q_ Macrostructure 1483_ 1.63 1:47 Task wpm rate 148.3 212.1 323.5 Note. Underlining emphasizes post hoc group comprehension means. an = 30; these subjects all read the discourse S 170 wpm. bn = 73; these subjects all read the discourse between 171 and 259 wpm. cn = 30; these subjects all read the discourse Z 260 wpm. dThese scores were all expressed as percentiles. n = 25. n = 59. n = 26. “150 (Q 102 they read (Chapter I; Buswell, 1937; Edfeldt, 1960; Gibson & Levin, 1972; Huey, 1968 [1908]). Findings of post hoc analysis. The smallest unit of discourse structure measured in this study was recall of paraphrased details. As shown in Table 8, the slow-discourse group remembered an average of 0.13 fewer detail questions than the fast group. Memory for detail relationship, called microstructure, showed an even wider gap between groups; the fast-discourse group remembered an average of 0.36 more microstructure questions than did the slow-discourse group. Microstructure, it may be remembered, was the only comprehension variable among the discourse questions that showed a significant rela— tionship to rate: r = .160, p = .033. Thus the slow-discourse readers appeared to be at a disadvantage in recalling detail and relationship (microstructure), in comparison with fast-discourse readers. The fast-discourse group also appeared to be superior when recalling main ideas; the average fast-discourse reader had .20 more main ideas correct than did the average slow-discourse reader. An interesting change occurred with macrostructure, however; the average fast-discourse reader no longer appeared to be superior. The average slow-discourse reader answered 0.36 more macrostructure questions correctly than did the average fast-discourse reader. This differ- ence was of the same magnitude as the microstructure group mean dif- ference, only the difference was in the opposite direction. There appeared to be an interaction (not statistically demonstrated) between slow- and fast-discourse groups and microstructure and 103 macrostructure question accuracy. Figure 4 graphically depicts this informal interaction of means. Macrostructure accuracy for all subjects was depicted in Figure 3. The distribution of macrostructure/rate results looked as if these results should have been significant; the upper margin formed a skewed normal curve, appearing to favor slower readers. That no statistically significant relationship was found between rate and macrostructure accuracy was probably a function of popula- tion variability in the middle rate range. Macrostructure questions represented the most complex type of comprehension tested in this study; accuracy required synthesis of all levels of information, as well as inference, critical thinking, and determining the relationship of major textual ideas to a unifying concept. It would appear that mature slow readers were occupied with a more complex type of information processing during reading than were mature fast readers. The slow-discourse group was not reading at slower rates due to difficulty with reading (Table 8), thinking (note macrostructure group means), or achievement abilities (note percentile group means). The slow-discourse group must have been using processing time differently from the fast-discourse group. Figure 5 duplicates Figure 4 and also depicts mean performance of the highly variable middle-rate readers. As shown in Figure 5, the middle-rate readers appeared to be superior to both other rate groups in mean detail accuracy: 0.21 greater than fast-discourse group means and 0.34 greater than slow-discourse group means. In 104 3'0 Fast-discourse 0"..Ogroup (n = 30) ' \ H Slow-discourse 1’ group (n = 30) 2.5 2.0 1.5 Mean number of questions correctly answered 1.0 Details iain Idea Micro- Macro- structure structure Type of comprehension Figure 4.--Mean group accuracy for slowest and fastest readers on four types of comprehension. (Slow-discourse readers were reading extended discourse at rates 5 170 wpm; fast-discourse readers were reading extended discourse at rates 2 260 wpm.) 105 3.0 «-- a 0—0 Fast-discourse '8 4 A gem-x Middle S. g a ll \ H Slow-discourse : I1 " f5 >, 2.5 "I'- S .1 U 0 d S. S. o d U m d .8 2.0 '-- t: ‘1 Q) d 3 U C q. 0 d 3 1.5 .3. .0 E at 3 C II C 1 f6 -‘='-’ . 1.0 ‘- ‘L_+ * j i Details Main Idea Micro- Macro- structure structure Type of comprehension Figure 5.--Mean group accuracy for slow-discourse, middle group, and fast-discourse readers on four types of comprehension questions. (Slow-discourse readers (n = 30) were reading S 170 wpm; middle readers (n = 73) were reading between 171 and 259 wpm; fast-discourse readers (n = 30) were reading at rates 2 260 wpm.) 106 all other question categories, their mean performance fell between slow and fast mean scores. Comparison of study and post hoc findings with comprehension literature. Auditory imagery (saying the words in one's head) prob- ably occurs with greater frequency among slower readers than among faster readers (Huey, 1908). It appears to be partially a mechanism to overcome the forgetting of idea traces in a sentence, which slow reading can produce (Gough, 1972). Perhaps auditory imaging serves an important function in macrostructure comprehension. This researcher hypothesized that the mature slow readers actually were reading with the intention of evaluating and incorporat- ing new information into idea chunks already categorized and stored in their memory. Thus, not only must the new information be under- stood and related to itself, it must also be related to information already "on file." liller's (1956) 7 i 2 hypothesis stated that an individual seems capable of holding only five to nine pieces of information in his immediate memory at one time. To hold more information in mem- ory, relationships must be formed among the items so that they are grouped into fewer categories. Thus, selective integration of information, either into existing mental categories or into new groupings, must occur for new information to be retained. Reading at a slower rate puts a greater strain on the central processing mechanism. Kerr (1973) found that transforming information into another form, such as inference, requires a greater processing 107 capacity, as does a more rapid input. Slow readers could be rehears- ing, committing some items to long-term memory, to lighten the short- term memory load (Chapter I, p. 11). Slow readers may also be resolving more ambiguities that are presented by the interaction of text and previously held assumptions. Ambiguity resolution takes time (Tanenhaus, 1979). Shores (1950) found that fast readers perform better on short- paragraph standardized tests but do not comprehend better when tested using science material. The time-limit element of short-paragraph tests is one biasing factor in favor of higher comprehension by faster readers. Also, however, short paragraphs are essentially microstructures; detail relationship is the highest level of thought required in such short text (considering the types of comprehension defined in this study). Thus, Shores' findings are not inconsistent with findings of the present study. Microstructure (detail relation- ship) comprehension may actually be facilitated at higher rates. Dee-Lucas (1979) maintained that fast readers read for central contents. This conclusion received some support from the present study; a comparison of main-idea means in Figures 4 and 5 revealed that more questions were correctly answered in the main-idea category than in any other by the fast-discourse readers. Dee-Lucas also found that the slower readers remembered twice as many inferences as the fast group. If this statement "is judged on the least complicated structural level, microstructure, the fast readers were best. (Little inference was required for microstructure accuracy.) If this statement is judged on the middle level of inference difficulty, main idea, fast 108 readers still had an advantage. But if one considers the most compli- cated understanding of test structure in this study, macrostructure, the slow readers appeared to have an advantage. Thus, mature slow readers may be tentatively judged to have been making more inferences than mature fast readers. Failure of Directional Hypotheses In Chapter I, it was stated that slower readers would pro- duce the most successfully completed information structures if their strategies worked; evidence for successfully completed structures was best judged in this study by superiority in macrostructure comprehen- sion. Failure to link different substructures successfully into macrostructures would have resulted, nevertheless, in superior micro- structure performance by the slower readers, according to the original theory. That slower readers were predicted to excel in micro- rather than macrostructure comprehension was assumed to be a failure for the slower readers' supposed reading strategies. It would have been con- sistent with the original theory to predict higher accuracy for mature slow readers in macrostructure recall. The fast readers were projected to be building an information superstructure, only ascertaining the identity of main ideas and their relationship. Three factors were overlooked, two regarding the nature of the questions and one regarding the nature of the subjects. Micro- structure questions, in this lengthy study discourse at least, repre- sented important supporting ideas and thus could have been considered important by these superior students. Macrostructure comprehension 109 required careful in-depth reading, which was less likely to occur at higher reading rates. Last, the subjects were not only good read- ers, they were also high achievers and thus more intelligent than an average high school student. With these factors operating, it is not impossible to conceive that fast readers were building more complete information superstructures. Superiority in macrostructure comprehension by mature slow readers who were excellent students should have been predicted. This conclusion fits the original slow readers' "structure" hypothesis for cases in which the text was conceptually accessible (i.e., not too difficult). Post hoc findings (not statistically tested) that slow readers exceed fast readers in macrostructure comprehension can thus be explained by the original "structure" hypotheses presented in Chapter I. Even if fast readers who were superior students attended to the supporting ideas presented by microstructure, it was thought that the more meticulous processing by slower readers would result in greater recall. Either speed really does facilitate making connec- tions between neighboring details, or else, perhaps, slower readers integrate microstructures into existing conceptual frameworks so that the microstructures are not as recognizable in their original form. Future researchers may find both of these projections to have a measure of accuracy. Revised "Structure" Hypotheses The slow-discourse group's slower average discourse rate of 148 wpm was not an indication of decoding or comprehension difficulty 110 or of word-by-word reading. These subjects were shown to be highly competent in the reading process and were also among the best stu- dents in their schools. The average slow-discourse reader scored in the 70th percentile on NDRT Total Reading; this equates with a grade- level reading score of 13.4. The average NDRT Vocabulary subscore was slightly lower, and the average NDRT comprehension score was slightly higher. The extended-discourse material read for this study was of appropriate difficulty, with a readability score of the 12th- grade level. Why, then, were these good students, who were mature readers, reading so slowly? The theoretical explanation given in Chapter I will be revised in light of study and informal post hoc study findings. Miller (1956) demonstrated that the span of immediate memory is severely limited, as is the span of perceptual judgments that can be made at any given time. People are restricted in the amount of information they can "receive, process, and remember" to about 7 i 2 bits of data. An article like "The Black Death" (Langer, 1968) obviously contains a multitude of information, much of which is new information. (Appendix C presents a structural diagram of text ideas.) Miller hypothesized kinds of linguistic recoding that allow such a multitude of thoughts to be dealt with. Most commonly, people trans- late new ideas they wish to remember into their own words, and in recall remember verbalizations and elaborate on them. People may attempt to combine the information into "chunks" of related ideas, thereby reducing the memory load. They may also attempt to memorize 111 the ideas by integrating them into already existing chunks of related ideas in long-term memory. According to the Given-New strategy explained by Haviland and Clark (1974), "the listener takes in a sentence, breaks it into its syntactically defined Given and New information, and then attempts to add the New information to memory." The Given information func- tions as a memory address, showing the reader where similar informa- tion is stored. When this "address" is located, the new information is incorporated into the already existing "file." In cases in which no file (Antecedent information) yet exists, the reader has the choices of (l) modifying a related existing category, (2) establish- ing a new category and building a relational bridge to some existing category, or (3) deciding it is not worth the processing effort and continuing to scan for ideas considered important. It will take more time to integrate the new information than simply to ignore it and to read for only more important ideas. The more inferences an indi- vidual attempts to draw, the more time, relatively speaking, the task will assume (Dee-Lucas, 1979; Kerr, 1973). This researcher theorized that the slow reader's tendency is to attempt to integrate all the ideas encountered into existing memory chunks and networks, building bridges from ideas that are read to existing schemata by extensive use of inference. This chunking or integration process follows the Given-New strategy. The mature slow reader tends not to make the decision that an idea is unworthy of this integration effort. He processes all details, examining rela- tionships, making judgments about importance and agreement and fit 112 with respect to existing memory networks, and continues reading and incorporating, adjusting perceptions. This occurs in a fashion simi- lar to that hypothesized by Gough (1972) in that all text is processed; Gough's theory must be viewed in light of Thorndike's (1917) and Huey's (1908) statements about information processing. Decisions on the lowest levels of relationship (words, syntax) are likely made without conscious awareness (Graesser, 1980; Kerr, 1973). Graesser showed that when readers attempt to form "a cohesive, well-organized, and recallable representation" of the discourse, mere resources are allocated to macrostructure analysis, or text relation- ships, than if they merely wished to comprehend the contents of the discourse. The slow-discourse reading style is comparable to the careful-reading rate for critical evaluation described by Yoakam (1928). Mature slow readers thus spend their extra time in a deeper and more extensive analysis of the text, attempting more recoding and integration of article ideas into conceptual chunks and carefully examining relationships as they occur. These slow readers use a form of inductive reasoning in comprehension/question-answering activities, making the data their own, then answering later discourse questions by drawing conclusions based on the chunked, related, integrated data. The slow readers are better thought of, not as building an isolated house, as postulated in Chapter I, but as building a structure that must be tailored to fit the contours of existing related idea chunks, fitting the structure being erected but also fitting the structures that already exist. 113 Whereas mature slow readers process essentially all material they encounter, constructing their comprehension from the entire body of data, mature fast readers are more selective. They select infor- mation they feel is worth in-depth processing, attending more to main ideas and supporting ideas. They construct a framework of this infor- mation, which allows a good understanding of general information structure to emerge. They use a more deductive-reasoning style in comprehension/question—answering activities; the ideas they have selectively attended to become, for them, clues from which the correct answer may be synthesized. Goodman's information-processing theory most nearly reflects their strategy (l976)--processing the least infor- mation possible to make the best guess possible. Implications for Education Few implications for education can be drawn at this stage of theory development. However, it appears likely that mature readers who read slowly in all printed material may be wasting precious time resources. It is unlikely that the heavier reading load imposed by a college schedule would allow the time for such a consistent in-depth approach. One is also mindful of Adler and Van Doren's (1972) opinion that very little writing is worthy of such an in-depth treatment. It would therefore seem useful to teach mature slow readers how to vary their reading strategy to fit course requirements, teacher expec- tations, and their own purposes. They need to learn that every word is not necessarily golden. 114 Nature faster readers would seem better prepared than slower readers to cope with the vast reading expectations of typical college courses. Their method of reading does not appear as "scientific," however; they are apparently not as thoroughly testing ideas before incorporating them into memory structures. It would seem that they need to learn to use a more rigorous reading style in scientific topics, especially if they become more involved with use of the scien- tific method; one must know as certainly as possible what is "given" and what is "new" and untested, to operate successfully within a scientific framework. For nonscientific purposes, however, their reading strategy seems superior. No doubt, all mature readers, what- ever their reading strategies, could benefit from study-skills instruc- tion that teaches them to recognize more efficiently the idea frame- work of any extended discourse. Conclusions In Chapter I of this dissertation, the writer discussed theory related to information processing and formulated conclusions from the literature concerning how mature fast and slow readers might differentially process extended discourse. Reading was defined as a self-directed, active process that goes beyond symbol-sound matching into complex thought processes requiring evaluation and organization of ideas encountered. Maturity in perception and cognition was defined, and by applying the definition to selection of the study p0pu1ation, it was possible to say that all study subjects were "mature" readers, having reached perceptual and cognitive maturity. 115 Reading-rate theories were divided into two groups, one hypothesizing that slow reading speed may be a cause of poor compre- hension and the other hypothesizing that the type of comprehension that occurs varies with the rate at which a text is read. A struc- tural hypothesis was formulated to account for the reading strategies of mature slow and fast readers, to predict the types of information that would best be recalled at slow and fast rates, and to test the two general theory divisions. Four types of structurally related comprehension were described: detail, microstructure, main idea, and macrostructure. In Chapter IV, the results of hypotheses testing were dis- cussed. No statistically significant relationship was found between the total number of extended-discourse questions correctly answered and the rate of discourse reading for the population. Also, no sta- tistically significant relationship existed between detail, main-idea, or macrostructure comprehension and extended-discourse reading rate, for the population. The only statistically significant comprehension/ rate relationship was between microstructure and discourse rate; the relationship was low and positive. In Chapter V, the p0pu1ation was subdivided for further trend analysis. The 31 subjects who most consistently read Nelson-Denny material and extended discourse slowly (slow-slow group) did not differ significantly from the 32 subjects who most consistently read both these measures rapidly (fast-fast group): F = .5280, p = .591. These two groups also did not differ significantly on any of the four types of comprehension. 116 It appears that the first group of reading-rate theories-- those stating that slow reading might cause poor c0mprehensi0n--were not generally applicable to this study population. With the exception of microstructure (detail relationship), rate did not appear to be related to comprehension. No data supported a causative rate- comprehension statement. The microstructure data supported a limited rate-relationship statement. In a post hoc study, the slowest 3O discourse readers (X'= 148 wpm) were informally compared with the fastest 30 discourse readers (X'= 324 wpm). Although no significance tests were performed, patterns of comprehension were suggested by the mean sample comprehen- sion data. Fast-discourse readers achieved slightly higher mean scores than slow discourse readers in detail and main-idea comprehen- sion. Whereas fast-discourse readers appeared superior in micro- structure comprehension, slow-discourse readers appeared to be superior in macrostructure comprehension. The middle group of read- ers (X'= 212 wpm) displayed great score variability; their group means showed them to be superior to both slow and fast readers in detail comprehension. This middle group attained mean accuracy scores on the other three types of comprehension that fell in between slow- and fast-discourse readers' mean scores. The seeming superiority of slow readers in macrostructure comprehension was explained by their detailed method of processing essentially all ideas encountered in the data. Incorporating these ideas into existing memory structure required some idea transformation, which may have hindered detail, microstructure and main-idea recall. 117 Fast readers were pictured as processing a smaller subset of data-- the central contents. Macrostructure was not as well understood by fast readers because less inference and information integration had occurred, but microstructure and main-idea comprehension were facili- tated. Microstructure recall seemed to increase significantly with increased speed of reading. A middle rate of reading extended dis- course appeared best for detail comprehension. Further investiga- tion concerning comprehension/rate patterns is recommended. Recommendations for Further Research Research regarding how the brain processes, chunks, and remem- bers ideas is in its infancy. Resolution of these ideas will undoubt- edly involve more than psychological research; biochemical research will eventually assume the greater significance. This study offered a general overview of how comprehension and rate may differentially interact in mature readers. Speculations are made about some of the general factors that may underlie the surface observations. A further study is suggested to test for the hypothesized presence of a rate interaction with micro- and macrostructure compre- hension. It is suggested that subjects be selected who have at least grade-level scores in NDRT Total Reading and that the same material and questions be used as in the present study. Two randomly assigned groups would be constructed and would be assigned different purposes for reading. One group would be asked to read for an under- standing of detail relationships (microstructures), and the other group would be asked to read in order to understand the interrelationship 118 of ideas of the entire text (macrostructure). What patterns of infor- mation recall are exhibited in these circumstances? Are fast readers still best at microstructure and slow readers at macrostructure? If it seemed likely to yield additional data of worth, a third group could be added to read for just the main ideas or central contents. Many of the subjects from the present study are accessible for further research. (Most indicated in writing their willingness to be subjects on a continuing basis.) It would be interesting to have all available original subjects engage in free written recall, writing everything they remember about the article one year after initially reading it. Any conclusions they came to, or any changes in the way they viewed things, formed from the reading, would be included. Is there a difference in the types of recall based on previous discourse rate? In examining the profiles of subjects in this population, it was noted that they seemed to have higher SAT Math than SAT Verbal scores. It would be interesting to perform this study with mature readers whose SAT scores varied, so that some were substantially better in Math and some in Verbal. SAT Math scores were more highly related to achievement than were SAT Verbal scores; thus it would be predicted that such groups would have group mean achievement differ- ences. Would they also differ in patterns of information recalled at various rates? As shown in Table 8, the highly variable middle group had SAT Verbal mean scores that resembled slow-discourse group means. and SAT Math mean scores that resembled fast-discourse group means. 119 This middle group had the highest group means in percentage class rank, detail comprehension, and total discourse comprehension. Further investigation into the relationship of these variables to each other is recommended. At this point, it appears likely that rate differences in mature readers reflect different information-processing strategies and that different rates are more appropriate for recall of differ- ent types of information. It does not appear that rates below oral- speech levels are hindering comprehension of these mature slow readers, with the possible exception of microstructure comprehension. There is much room for further research on these questions, however. APPENDICES 120 APPENDIX A VERBAL DIRECTIONS GIVEN TO EACH STUDENT BEFORE THEY READ THE TASK ARTICLE 121 122 Verbal Directions Given to Each Student Before They Read the Task Article I have here a copy of a Scientific American article called "The Black Death.” I want you to read this article in the manner that you would normally read such material upon seeing it for the first time. (Pause) Read through the article just once. After you have passed a page, you are not to return to it. Tell me when you have finished because I will be timing you. (Show the st0pwatch for emphasis on the rate factor.) You will be asked a few questions afterward. Do you have any questions? (If there are questions about rate, say: "If you generally read fast the first time through an article, do so now; if you generally read at a slower pace the first time through an article, do so now.") Begin. APPENDIX B QUESTIONS FOR BLACK DEATH ARTICLE 123 124 Questions for Black Death Article For each question, print your answer in the blank provided. (Please make sure your printing is legible. Thanks.) 1. The Black Death was so named because A. it was carried mainly by black rats. 8. it was transmitted from ports in the Black Sea. C. Black Magic was thought to be responsible for the plague. 0. black spots appeared on plague-stricken persons. 2. People had different theories for what caused the plague; a common belief of the time was that A. fleas, carried by rats, were transmitting the disease. 8. Satan had sent the plague thru witches and/or Jews. C. physicians were spreading the plague by poisoning wells. 0. Death would overtake them unless they fled. 3. During the plague years of the 14th century, the loss of population A. exceeded thirty percent, but rose above sixty percent in some areas. 8. approached twenty percent, with England especially hard hit. C. was greatly increased because many countries were at war. 0 was so great that Europe did not recover until the 18th century. 4. The fear and despair gripping Europe during the plague did not directly influence A. the moral outlook. B. the art and literature. C. the economic changes. 0. the demographic changes. 5. The 1348-1350 plagues claimed the most victims each year during A. January-March. B. April-June C. July-September. 0. October-December. 6. The attitude of most people during the plagues may be sum- marized as follows: A. Take whatever you want; we won't live to pay the price for our deeds. Seek God; we are being punished for our sins. Eat, drink, and be merry for tomorrow we may die. Suicide is preferable to enduring the agony of the plague. COG) 10. 11. 125 What was the financial situation of the pe0p1e immediately after the 1348-1350 plague years? A. Rural areas were in a desperate financial situation but towns and cities were not as there were fewer people to share wealth. B. Towns and cities were in a desperate financial situation but rural areas were not because of their food reserves. C. Towns, cities and rural areas were in a desperate finan- cial situation because the social-economic structures collapsed. D. Towns, cities and rural areas were not immediately in desperate financial situations because there were fewer people to share food and wealth. Pretend you are on a task force to plan for civilian needs in the event of a disaster in this country. You must consider how people would react to a nuclear bombardment which kills a huge segment of the population. From your study of human behavior during the European plague years, which response would you consider to be least likely after a nuclear disaster? A. There would be a significant increase in criminal behavior. B. People would attempt to atone for their sins. C. There would be a resurgence in witchcraft and superstition. 0. There would be a great concern for helping survivors. Flagellation cults of the time were concerned with people A. purifying themselves of evil. 8. making extravagant vows to God about the future. C. praying that the end of the world might not come. 0. whipping themselves to stop the advance of the plague. These several hundred years of plagues in Europe may have produced the Reformation because A. constant mental pain and the nearness of death led people to reconsider their relationship to God. 8. people discarded their belief in God when prayer had failed to protect them from the plague. C. people were upset with clergy who had turned to supersti- tion and hedonistic behavior. 0. Calvin, Wycliffe and other church leaders had fled each plague outbreak, abandoning the common people. The term pandemic refers to all the outbreaks of plague occurring between 1348 and the late 17th century in Europe. the final large outbreak of plague in 1665 in Europe. all the outbreaks of plague occurring between 1348 and 1350 in Europe. all the outbreaks of plague occurring after 1350 all over Europe. DOW) 12. 13. 14. 15. 126 During the European plague years, major shifts in population occurred as A. people fled to the towns and cities during the few years of continuous plague. people fled to the towns and cities during the years of intermittent plague. C. people fled to the country during the few years of continuous plague. 0. people fled to the country during the years of inter- mittent plague. E. A and B H. B and C F. A and C I. B and D G. A and D J. C and 0 Which segment of society made a lot of money immediately after the 1348-1350 plagues? A. the clergy, because they were profiting from the labor of the poor. 8. the landowners, because they sold farmland for a great profit. C. the survivors, because they stole what they wanted from deserted houses. 0. the craftsmen, because there was a demand for their work. Why did people behave as they did as the plague approached? A. Since they had no way of preventing the plague from reaching them, hastening death by entering quarantined areas seemed reasonable. 8. They were terrified because there was no known cure for the plague. C. The psychological shock caused by so many persons dying led to feelings of omnipotence. D. The Bacillusgpestis organism affected not only the lungs, lymph glands, or blood, but also the emotions and mental attitude of people. Why were sick and dying people frequently dumped in burial pits with persons already dead? This was done A. because loved ones had abandoned them and there was no one to care for them. 8. in a desperate attempt to keep the plague from spreading. C. because sick people were considered to be as good as dead; people who contracted plague invariably died in a few days. 0. because burial crews were mostly composed of hardened men who were more interested in body fees than human suffering. 127 16. This article has presented a variety of responses which pe0ple had to the plague, discussing certain individuals for illustration. What generality concerning human behavior during major catastr0phe can most logically be made? A. Pe0ple can generally choose how they will respond to major catastrophe. A person can never be sure how he/she will respond to major catastrophe. C. A person's environmental background will determine how he/she will respond. 0. A person's economic status will determine how he/she will respond. Discussing any part of this experiment might influence the results obtained. People tend to respond in relationship to their expecta- tions. Therefore, I earnestly requestgyou not to discuss any phase of this experiment with anyone until all persons at your school are finished with it. Thank you very much for your participation. I woqu be happy to make results and conclusions known to you, upon your request, at a later date. Results of the Nelson-Denny Reading Test will be returned to you in the next few days. Thank you. Note. These questions may be used only with the written permission of S. M. Hice, and with the condition that the source is properly acknowledged. C) Susan Hice, 1982 APPENDIX C STRUCTURAL DIAGRAM 0F LANGER'S (1968) "THE BLACK DEATH" Note: Diagram refers to the condensed version of this article prepared by S. M. Hice. 128 {I’m-III!“ by ' black nt- un-Ied by CQYKI n rod-n:- rmnfl : uu‘KIll mn- I! understood nod-y m a mum In 801 610 L1: 1-3 a: I each [011 until] flu] in 5<6 til I outbrnk Ln ch. noun a- in Nedical - Historical Aspects lprnd mum“ in an: 2 year- (s sin W 6 men were armed “ruck harden: manna tn vim-r been our. . m Ln Sp: 621 to I to I vuk Ln no 01-“ no" Amman!) it shows human behavior n ma 5 t t—o situation u. there is now a threat of km" exterminat on by more Why 15 1t studxed today? rapid means forces that make for c 01152 were streng ened and given impetus by it amp-apex: (mun. aym of! or total.) THE PLAGUE Blac Dear. (Great Dying) Short~Tem Effects (1348-1350) w m. .m. Sim an.“ Hitchol t A! Cllvln "w “W p - Jon c it rm sum l GIN. an a loan-Hon liciln- I'P ‘ to hll [la XIII uten- pnenai 16.1 ml/ : u - um rm 0! u:- . plagu- Socxal Aspects u or ell-I In“ It:- (I no “1 I n "a ouur mm... ablation mu not not We v notch for uuu 1mm . u m M Lon ~Term Effect (13th-17th Centuries) ©1982 SUSAN HICE oul - I: u. u: run-an: 1 It." or P Rnforllcion 130 Structural Diagram of Langer's (1968) "The Black Death" THE PLAGUE (Black Booth) (Great Dying) 1 .0 Medical Aspects why is it studtad today? it man hman bahavior nan teat-o g “mation than 1a now a threat of utauflnation by ma rapid moans (man that. me for change war. atrmgthanad and givm impetus by it 2.0 Short-ram Effects Social Aapecta 3.0 Lon Jam Effects (DEUEK CamuS MEDICAL/ HISTORICAL ASPECTS 1.0 as understood fix plague tines caused by 13] cause unknown prevmtive nusures unknown cure unknown transaitted_by 71“. black rats carried by as understood today Bacillus certain rodents pestis squirrels there pnanonic “952) known in 3 (all fatal) Lepticnic (blood) high fever death toll synptaas dark skin h- as Hatches Black Death is on see onizin in 801 die in 2-3 days 1368-50 ‘ n c history of occurrence in mat-ope usually fatal in 5-6 days outbreaks in the aaaan aspire A————°‘1-l6m °" after this outbreaks were rare until cmtury brought by nerchant ships habl carried (ran a as ports spread northward in next 2 ears in Watchman. 5:790. 53%!“V3; which were crowded struck hardest F 13 towns 1 and unsanitary an in Italian ports ' subsided in winter ‘ 1351-1665 rose to a eak in i ate at - broke 03;: in in nost abundant) further ream-rat outbreaks outbreaks occurred av 10 ears 0' r Iess Nbonicjlggle outbreak 1;“ 1 ' ‘ killed of London in 1665 9° “ °° ‘ thereafter. pl e vanished 5 western and cuts-a1 Europe 132 ‘ aany persons died in the streets I]; of mropeans 1m .1. MIMI. Agrevedhieri id ed in m.“ problaa of Wu .. three years WI. ‘1'. ‘ unscrupulous V <17 ... O. m in p¥t outside of town d%g§ic (Massive ‘ . mu 50‘ ‘4‘ n‘m'm A ° ° P'°P' c ace 641nm varied over 1!: of mro- W 9"" “d filial! qgation fell by ad Of 50 ".u v 1 . . to . I m ”iguana Roch o . loss inc. “attained LF—{Eg'flthe year war Sllm'r- for 2w ”Us . SPEC He sent it thru £32559 E T8 1 or coasts ._____2'° aost felt God out it as ain‘ thru vuaaous molecules lanai-hut ‘ via 3 :1;- PO sonar arrows rch f Satan m.“ witches t it can; see or P asu- Jews t it fro satan a cause siciane m“ t to ha :1“ “111 “O“. .W (60‘) (la ellation "“1 “ch “he! ‘ cults m p 1. to ' Fitted or Day reopen... 3,,“ ‘ throw of Atoll-mt P'leosioal/aoral/ ‘0 hypothetical 3:16?“ 00 ‘ cases a aercy r sous-Dee gum gm of ch. indul use in witchcraft inc. plague (Satan o are kill Jan-witches-dootors upper class left other responses no 0 v search for cause . suicide ; hedoni- EFFECTS 3. 0 ‘ sychol ical- apocalyptic [despair 133 period of rapid population growth territorial expansion eneral eri urvivors inherit much . pro-Great by .m i towns 8. post-Great dd” Dying short flush arti t t of work of mm W (than) cmrches share in new wealth. n ma:— _a_g_r_icultural dggession overtakes cities surplus of crops n ices a favor people to buy than rural areas (Mediate fern Powlt- move to towns for work collapse) tion fell landlords t5: to keg forcibly thousands of fans 8 villages eserted - fans revert to wilderness rats a land values ninor land-owning gong disappear povrished ‘ last a l tine thoughts aolded of disease f“: 3 of death pro-occupation with graves! disea seL decay . inorality & loose livin reli ious - la e stage or Protestant Refouaation? © Susan Hice, 1982 cl sett 20L)». hood reflected in Christ's suffering painting Hell's torment death shown as in a-“ dance of death skeleton fl 4' old m1 ed God pl 33 673%— y or sat a e escr y to redress sin ilt J31?— boccaccio mercenary—L— ev (experimced by | g: reuorse ‘ rich truendous will-known Er: ilt poor guilt people. too) . Martin Luther I ‘ ”“1 (or) an ered nuc fra- fear of death Calvin fled in terror from each plane pmitmtial processions il es 1 commal ilt ‘ “yarn-Far? nass preaching_ assertion of life in esence of death . aarked tudency toward wild defiance dqneted - dead or fled parishes ready for no: leaders-church not stopped I: Ref. leaders affected E 2152“” relationehiE reli ious outlooks changed thr_ou_§Lchronic grief acyo a APPENDIX D DETAIL QUESTIONS, SHOWING CORRECT ANSWER, AND TEXT DERIVATION OF ANSWER 134 135 Detail Questions, Showing Correct Answer, and Text Derivation of’Answer l. The Black Death was so named because black spots appeared on plague-stricken persons. From paragraph 2, last sentence: "The Black Death got its name from)dark blotches produced by hemorrhages in the skin." (Langer, 1968 2. The l348-l350 plagues claimed the most victims each year during July-September. From paragraph 4, last sentence: "Each year the epidemic rose to a peak in the late summer, when the fleas were most abundant, and subsided during the winter, only to break out anew in the spring." (Langer, 1968) 3. Flagellation cults of the time were concerned with people purify- ing themselves of evil. From paragraph 10, last sentence: "In the streets half-naked flagellants, members of the century-old cult of flagellantism, marched in processions whipping each other and warning the people to purge themselves of their sins before the coming day of atone- ment." (Langer, l968. 4. Which segment of society made a lot of money immediately after the 1348-1350 plagues? --the craftsmen, because there was a demand for their work. From paragraph 21, middle sentence: "They built elegant houses and went on a buying spree that made work (and highprices) for the manufacturing artisans." (Langer, 1968) © Susan Hice, 1982 APPENDIX E MICROSTRUCTURE QUESTIONS, CONSTRUCTED FROM NEIGHBORING DETAILS, AND REPRESENTING SUPPORTING IDEAS l36 I37 Microstructure Questions, Constructed From Neighboring Details, and Representing Sgpporting Ideas l. During the plague years of the l4th century, the loss of population exceeded thirty percent, but rose above sixty percent in some areas. From paragraph 7 (8). Requires recognizing that "more than l/3" is compatible with over thirty percent, and "more than 2/3" is compat- ible with above sixty percent. (Text from Langer, l968) Is essen- tially)memory of Details. (All distractors based on paragraphs 7 and 8. 2. What was the financial situation of the people immediately after i the 1348-1350 plague years? Rural areas were in a desperate finan- ‘ cial situation but towns and cities were not as there were fewer people to share wealth. From paragraphs (20), 2l, 22. Requires understanding and remember- ing a clear comparison made in the text. No inference is required, just syntactic and semantic understanding. (All distractors found in paragraphs 20, 2l, 22.) 3. The term "pandemic" refers to all the outbreaks of plague occurring between l348 and l350 in Europe, From paragraphs 4 and 5. This is a vocabulary question and requires noting the usage of "pandemic" in the sentences in which it occurs. Some inference is required. (All distractors are composed of other date possibilities.) 4. Why were sick and dying people frequently dumped in burial pits with persons already dead? This was done because burial crews were mostly composed of hardened men who were more interested in body fees than human suffering. From paragraphs (12) and l3. Requires understanding the relation- ship of sentence details in one complete paragraph. No special inference is required, just implications from word usage. (All distractors came from paragraph l2, except one that was invented.) © Susan Hice, l982 APPENDIX F MAIN-IDEA QUESTIONS, SHOWING CORRECT ANSWERS AND DIVERSE ELEMENTS OF TEXT THAT THE ANSWER AND/OR THE DISTRACTORS ARE DRAWN FROM I38 T39 Main-Idea Questions, Showing Correct Answers and Diverse Elements of Text That the Answer and/or the Distractors Are Drawn From 1. People had different theories for what caused the plague; a common belief of the time was that Satan had sent the plague thru witches and/or Jews. Requires putting together elements of text from paragraphs l0 and 17. Distractors came from paragraphs 2 and 9, among others. Thus the main idea is pieced from several microstructures throughout the text. The question must also be closely read, to remember that "belief" refers to belief about the cause of the plague. Summarizing, this answer required synthesizing of related information in different paragraphs and referred to one of the main ideas in the text. 2. The attitude of most people during the plagues may be summarized as follows: Seek God; we are being punished for our sins. The answer occurs in paragraph l0; half the answer appears again in paragraph 17, and half in paragraph 22. Distractors came from paragraphs 12, l4, and 15. This main-idea question was very like microstructure, except for its repetition in diverse areas of text, its being a main idea, and the distractors coming from diverse areas of text. Some inference was required to produce the correct answer. 3. These several hundred years of plagues in Europe may have produced the Reformation because constant mental pain and the nearness of death led people to reconsider their relationship to God. Requires synthesis of information from paragraphs 27 and 28. Para- graphs ll and 14 give clues that the distractors are wrong. This main-idea question differs from microstructure in the importance of the idea, the synthesis required, and the location of distractor information in diverse sections of the text. 4. Why did people behave as they did as the plague approached? They were terrified because there was no known cure for the plague. Required synthesizing information presented in one paragraph of text, paragraph 9. Behavior was gleaned from paragraphs 10, ll, 14, 15, and 16. Distractors were located in paragraph 12, or fabricated based on paragraph 2. This main-idea question differs from micro- structure in the importance of the idea, the s nthesis required, and the necessity of knowing that the diversely located distractor material was incorrect. © Susan Hice, l982 APPENDIX G MACROSTRUCTURE QUESTIONS, RELATING MAJOR TEXTUAL IDEAS AND SOMETIMES ALSO GLOBAL APPLICATIONS I40 l4l Macrostructure Questions, Relating Major Textual Ideas and Sometimes Also Global Applications (See Figure l) l. The fear and despair gripping Europe during the plague did pp§_ directly influence the economic changes. Requires an understanding of the whole article. The four possible pairings must be weighed, with an understanding of each necessary to answer correctly. This question concerns the interaction of psychological effects with other societal effects. 2. Pretend you are on a task force to plan for civilian needs in the event of a disaster in this country. You must consider how people would react to a nuclear bombardment which kills a huge segment of the population. From your study of human behavior during the European plague years, which response would you consider to be least likely after a nuclear disaster: there would be a great concern for helping survivors. Requires an understanding of human behaviors presented in the article. These must be weighed for which was least present during the plague, then applied to the situation proposed. 3. During the European plague years, major shifts in population occurred as pegple fled to the towns and cities during the years of intermit- tent plague and peOple fled to the country during the few years of continuous plague. Requires an understanding of short-term demographic effects and long-term demographic effects. The answer may be gleaned from para- graphs ll, l9, and 22, which makes it a bit like main idea. How- ever, the idea interaction is on a more complex basis. 4. This article has presented a variety of responses which people had to the plague, discussing certain individuals for illustration. What generality concerning human behavior during major catastrophe can most logically be made? A person can never be sure how he/she will respond to major catastrophe. Requires inference from statements made of behavior patterns during the article to know that this is the most appropriate answer. Also requires a global generalized understanding of the article. C) Susan Hice, l982 APPENDIX H RAW DATA I42 T43 Raw Data Data for 133 subjects. Percentile rank has been converted from rank. task wpm has been converted from task time. and D. Mic. HI. and Hac have been tabulated from scoring each individuals test by these categories. Subject ND Zage Task Number N'D-V ND-C ND-T m SAT-V SAT-H Rank D Mic HI has Rm 001 77 99 94 333 650 560 2.8 2 3 2 1 330 002 98 98 98 344 630 570 4.8 2 2 2 1 360 004 75 84 81 269 540 510 17.6 3 2 3 3 239 005 71 90 83 333 460 400 22.0 2 2 1 1 322 006 89 84 88 195 640 650 8.0 4 1 3 3 218 007 66 84 77 226 430 620 6.8 2 3 2 2 200 008 81 81 82 356 550 710 1.6 2 4 2 1 347 011 99 93 99 561 700 690 0.8 4 3 4 1 204 012 95 93 94 586 560 530 12.8 2 3 2 2 231 013 83 74 79 226 460 550 10.0 3 3 2 2 186 014 56 59 58 425 530 630 6.0 2 3 2 1 292 016 32 74 55 203 470 660 6.4 2 2 3 2 179 015 69 74 71 245 490 530 20.4 2 3 2 O 182 017 64 59 62 235 380 460 4.8 2 0 3 0 200 018 73 69 71 235 410 640 14.4 2 3 2 2 200 019 93 9O 92 299 550 560 4.4 3 3 3 3 264 020 44 69 58 235 390 550 25.2 1 2 2 2 161 021 73 93 85 245 460 340 22.4 2 4 3 4 203 023 69 78 73 226 410 650 11.6 2 1 2 1 194 028 87 84 87 226 580 460 14.8 4 3 2 3 188 029 95 69 88 368 450 540 16.0 3 3 2 1 199 031 50 59 55 195 460 630 18.0 2 2 2 2 158 032 89 97 97 203 550 690 7.2 3 4 3 1 277 033 50 74 63 195 430 520 18.8 2 2 3 1 130 034 61 54 58 235 340 540 23.6 2 3 2 1 162 035 81 97 91 279 530 740 2.4 3 3 1 2 216 036 53 87 73 309 460 550 5.6 3 3 1 2 256 037 64 54 59 299 390 590 13.6 2 3 3 2 232 039 71 64 68 475 450 680 16.0 3 3 0 1 205 040 69 64 67 309 470 530 10.8 2 1 1 0 178 041 59 49 54 309 410 450 14.0 2 4 1 1 204 043 95 90 94 391 570 640 7.2 2 2 2 2 232 044 53 81 69 203 440 580 13.2 3 4 3 1 144 046 66 49 58 159 450 480 12.0 1 4 2 2 159 047 75 74 75 226 510 620 16.0 3 3 0 2 158 048 66 98 88 195 640 670 0.4 4 2 2 2 165 050 73 97 88 257 580 640 5.2 1 2 1 1 215 051 38 81 63 226 370 550 20.0 3 2 2 1 180 052 64 74 69 165 550 730 3.6 3 4 1 4 152 053 38 69 55 269 410 580 8.4 3 2 1 2 231 T44 055 66 90 81 290 520 540 9.6 2 3 2 1 269 056 81 78 80 188 500 690 9.2 3 3 3 1 175 057 69 78 73 344 420 560 2.0 3 2 1 3 192 058 35 69 54 195 430 480 22.8 1 3 1 1 158 059 53 64 59 188 390 370 24.4 1 2 2 1 118 060 95 97 96 450 660 660 2.8 4 4 4 2 250 062 90 95 93 177 500 580 8.8 2 4 1 1 183 064 69 9O 82 195 510 670 12.4 4 2 3 2 167 066 38 74 58 344 410 350 23.2 1 2 O 1 208 071 69 49 59 214 390 530 26.0 1 3 3 2 215 072 61 49 55 203 440 390 27.4 4 2 O 1 159 073 95 98 97 195 550 630 1.4 2 3 1 2 199 074 87 87 88 153 - - 9.6 2 3 1 2 145 075 90 84 88 327 560 620 19.2 3 1 2 1 234 076 50 78 65 177 410 560 20.5 1 2 2 1 156 077 61 64 63 257 380 520 24.7 2 2 0 1 249 078 61 84 75 299 440 550 6.9 3 2 1 1 250 079 94 98 96 356 550 660 5.5 2 3 2 2 269 080 89 49 72 195 510 510 12.3 3 3 1 2 161 081 77 81 80 488 410 560 32.9 2 4 2 2 272 082 77 84 82 226 550 590 45.2 3 3 2 2 215 083 92 95 94 235 480 640 2.7 4 4 2 3 208 084 96 69 89 356 530 660 8.2 1 3 3 1 298 085 97 74 91 290 410 390 34.3 2 3 2 2 239 091 38 64 53 188 380 550 12.8 2 3 2 2 127 092 79 84 83 195 490 520 15.4 0 2 0 1 347 093 86 69 79 214 - - 18.0 3 3 3 2 193 094 59 84 73 279 400 480 25.6 2 2 2 2 311 095 91 99 96 203 530 690 2.6 2 4 3 0 300 096 S9 78 69 188 480 540 23.1 3 3 3 1 249 097 71 69 70 203 480 570 5.1 1 2 2 2 154 098 71 87 81 235 480 550 7.7 O 2 1 2 267 099 75 81 79 245 520 540 10.3 3 1 2 1 191 100 73 93 85 269 - - 20.5 3 2 3 1 255 101 50 74 63 257 420 500 28.2 2 3 3 1 184 111 92 98 95 413 600 600 4.8 4 4 1 2 283 112 86 99 94 561 560 690 1.2 4 2 4 4 378 113 47 93 76 188 440 530 16.7 3 3 1 0 194 114 71 49 60 - - - 11.0 4 1 2 1 149 115 77 87 84 299 - - 14.3 4 2 2 0 192 116 90 99 96 279 - - 6.0 2 3 3 3 253 117 79 93 88 269 540 620 7.1 3 3 3 2 207 118 73 74 73 195 - - 16.7 0 1 2 1 122 119 S6 90 77 391 440 320 22.6 1 4 1 2 128 120 79 97 90 319 530 620 15.5 2 2 1 1 269 121 83 99 94 368 610 630 3.6 4 3 3 2 327 122 87 54 73 226 - - 8.3 2 3 1 4 186 123 50 64 58 299 - - 20.2 4 1 1 2 222 1115 125 71 54 63 413 - - 23.8 3 3 1 2 189 126 91 84 89 203 630 720 9.5 0 3 3 4 170 127 56 54 55 269 - - 2.4 3 3 1 3 165 128 64 74 69 245 - - 19.1 2 3 2 1 217 129 64 97 85 279 - - 13.1 2 4 2 2 308 130 71 49 60 279 - - 29.8 1 3 2 1 184 131 77 69 73 299 - - 36.9 1 3 1 1 192 141 59 59 59 165 450 400 20.5 2 2 3 1 314 142 95 9s 95 356 600 600 21.3 1 3 3 3 559 143 66 74 70 290 540 520 7.4 3 4 2 2 306 144 91 74 85 299 570 530 4.1 2 4 3 3 169 14s 66 69 68 129 - - 4.9 2 2 2 2 180 146 69 93 84 235 - - 34.4 2 2 3 1 265 147 73 74 73 257 460 540 18.0 2 3 3 1 231 148 71 59 65 438 490 650 1.6 2 1 3 1 176 149 64 49 57 195 440 420 17.2 1 2 3 1 153 151 83 74 79 195 470 390 40.7 2 2 0 1 106 152 84 59 73 226 540 480 16.3 2 4 2 0 176 153 91 97 94 319 - - 12.8 1 3 3 2 184 154 56 44 50 379 - - 30.2 1 4 1 1 245 155 93 93 93 356 470 500 9.3 3 3 2 0 282 156 93 9o 92 299 450 550 14.0 2 4 3 1 240 157 96 99 98 290 620 630 8.1 2 3 3 2 254 158 95 98 96 290 440 420 38.4 4 3 1 1 146 159 59 69 64 269 - - 62.8 2 2 2 1 274 160 83 59 72 214 - - 26.7 4 2 3 1 809‘ 161 95 81 91 391 470 530 32.6 4 3 2 1 284 171 84 87 87 24s - - 8.6 2 2 1 2 235 172 97 99 98 279 670 480 11.1 4 3 2 3 136 181 81 81 82 269 500 700 23.1 4 3 4 2 247 182 84 54 71 _ 214 450 460 17.2 3 3 2 3 188 183 53 54 54 327 390 420 29.1 3 2 1 1 209 184 73 59 67 117 420 460 31.3 2 2 3 2 109 185 86 69 79 290 420 530 18.7 2 o o 1 215 186 97 95 97 257 510 450 40.3 4 2 2 2 260 187 90 84 88 195 510 470 44.8 1 2 3 2 258 188 44 69 58 195 360 440 17.9 3 3 3 3 215 189 87 87 88 319 550 630 2.2 4 4 2 2 192 191 75 87 83 269 530 560 20.9 1 2 1 1 264 192 32 69 53 279 - - 61.9 4 3 3 1 182 193 83 95 90 299 590 610 14.9 4 2 3 2 248 194 83 84 85 106 570 510 12.7 3 2 2 1 244 195 77 84 82 235 - - 2.4 2 2 1 2 163 196 84 69 78 235 510 590 7.5 1 4 1 1 250 197 64 69 67 235 400 630 3.0 2 3 2 3 217 124 so 39 44 195 - - 10.7 1 4 1 2 146b Note. This table is a data summarization. Comprehension accuracy was actually entered as individual question responses, then corrected and tabulated into categories by computer. Task wpm rate was entered as time units and converted by computer into me and Sps by apprOp- riate formulas. Percentage rank was also calculated by computer from class rank and school senior population. aThis subject's speed was known to have probable error; speed could have been any value between 324 and 809 wpm on task. Deleted for Pearson r. bThis subject was not included in study due to low NDRT Total score. REFERENCES 146 REFERENCES Adler, M. J., 8 Van Doren, C. (1972). How to read a book. New York: Simon & Schuster. Anderson, I. H., & Dearborn, W. F. (l94l). Reading ability as related to college achievement. Journal of Psychology, 11, 387-396. Bader, L. A. (l980). Reading diagnosis and remediation in classroom and clinic: A guide to becoming an effective diggnostic-remedial teacher of reading and language skilTE. New York: Macmillan. Blommers, P., 8 Lindquist, E. F. (1944). Rate of comprehension of reading: Its measurement and its relation to comprehension. Journal of Educational Psychology, §§(8), 449-472. Braam, L. (l963). Developing and measuring flexibility in reading. The Reading Teacher, lg, 247-254. Buswell, G. T. (l937). How adults read. Chicago: University of Chicago. Buswell, G. T. (l945). Non-oral reading: A study of its use in the Chicago Public Schools. Chicago: University of Chicago. Buswell, G. T. (l95l). Relationship between rate of thinking and rate of reading. School Review, 49, 339-346. Carlson, T. R. (1949). The relationship between speed and accuracy of comprehension. Journal of Educational Research, 42, 500-5l2. Chang, S. S. (l977). Reliability of reading rate measures as a func- tion of warm-up passage and of time limit. Dissertation Abstracts International, 38, 5347A. (University Microfilms No. 78-00, 805) Clark, H. H., & Clark, E. V. (1977). Psychology and language. New York: Harcourt, Brace & Jovanovich. Cunningham, J. W. (l979). An automatic pilot for decoding. ng_ Reading Teacher, 32(4), 420-424. Dee-Lucas, D. (l979). Reading speed and memory for prose. Journal of Reading Behavior, 11(3), 22l-245. I47 I48 Edfeldt, A. W. (1960). Silent speech and silent reading. Chicago: University of Chicago. Eurich, A. C. (1930). The relation of speed of reading to comprehension. School and Society, 32(821), 404-406. Fridal, G. (1979). Improving concentration in reading. Journal of Reading, 23(3), 198-199. Gibson, E. J., & Levin, H. (1975). The psychology of reading. Cambridge, MA: MIT Press. Goodman, K. S. (1976). Behind the eye: What happens in reading. In H. Singer & R. Ruddell (Eds.), Theoretical models and processes in reading (pp. 490-491, 2nd ed.). Newark: International Read- ing Association. Gough, P. B. (1972). One second of reading. In J. F. Kavanagh & I. Mattingly (Eds.), Language by ear and gy eye: The relation- ships between speech and reading (pp. 35l-358). Cambridge, MA: MIT Press. Graesser, A. (1980). Structural components of reading time. Journal of Verbal Learning and Verbal Behavior, 13, 135-151. Hafner, L. E. (1977). Developmental reading in middle and secondary schools. New York: Macmillan. Harris, A. J. (1970). How to increase reading ability: A guide to developmental and remedial methods (5th ed.). New York: David McKay. Harris, A. J. (1968). Research on some aspects of comprehension: Rate, flexibility, and study skills. Journal of Reading, 12, 205-210, 258-260. Hill, W. P. (1964). Influence of direction upon the reading flexibility of advanced college readers. Yearbook of the National Reading Conference, 13, 119-125. Hoffman, J. V. (1978). The relationship between rate and reading flexibility. Reading World, 11, 325-328. Holmes, J. A. (1954). Factors underlying major reading disabilities at the college level. Genetic Psychology Monographs, 13(1), 3-95. Huey, E. B. (1968). The s cholo and eda 0 of reading (First publ. by Macmillan, 1 . am r1 ge, : MI Press. Jensen, P. E. (1978). Theories of reading speed and comprehension. Journal of Reading, 21, 593-600. I49 Kerr, B. (1973). Processing demands during mental operations. Memory and Cognition, 1(4), 401-412. King, I. (1916). A comparison of slow and rapid readers. School and Society, 1000), 830-834. King, I. A. (1917). A comparison of the efficiency of slow and rapid readers. School and Society, 3, 203-204. Langer, W. L. (1968). The black death. In G. Hardin (Ed.), 39 steps to biology: Readings from Scientific American (256-261). San Francisco: W. H. Freeman. Letson, C. T. (1958). Speed and comprehension in reading. Journal of Educational Research, 32, 49-53. Marshall, N., & Glock, M. D. (1978). Comprehension of connected discourse: A study into the relationships between structure of text and information recalled. Reading Research Quarterly, 11(1), IO-56. Maxwell, M. (1978). Learning style and other correlates of performance on a scanning experiment. Journal of Reading Behavior, 19, 49-55. McConkie, G. W., Rayner, K., & Wilson, S. J. (1973). Experimental manipulation of reading strategies. Journal of Educational Psychology, §§,(1), 1-8. McDonald, A. S. (1960). Factors affecting reading test performance. Yearbook of the National Reading Conference, 3, 29-35. McDonald, A. S. (1963). Flexibility in reading. In J. A. Figurel (Ed.), Reading as an intellectual activity (International Read- ing Association Conference Proceedings, 3, pp. 81-84). New York: Scholastic Magazines. McDonald, A. S. (1965). Research for the classroom: Rate and flexi- bility. Journal of Reading, 3, 187-191. Miller, G. A. (1956). The magical number seven plus or minus two: Some limits on our capacity for processing information. Psycho- logical Review, 33, 81-97. Nelson, M. J., & Denny, E. C. (1960). Examiner's manual: The Nelson- Denny Reading Test. Boston: Houghton Mifflin. Nicholson, T. (1978). The relative effects of different error types on children's understanding of connected discourse (abstracted). Reading Research Quarterly, 11(2), 259-264. I50 Nie, N., Hull, C. H., Jenkins, J. G., Steinbrenner, K., & Bent, D. H. (1975). Statistical package for the social sciences: SPSS (2nd ed.). New York: McGraw-Hilll Pauk, W. (1965). Scholarly skills as gadgets. Journal of Reading, 3, 234-239. Preston, R. C., & Botel, M. (1951). Reading comprehension tested under timed and untimed conditions. School and Society, Z1, 71. Rankin, E. F., Jr. (1963). Sequential emphasis upon speed and com- prehension in a college reading improvement program. Journal of Developmental Readigg, 1, 46-54. Rankin, E. F. (1962). The relationship between reading rate and comprehension. Yearbook of the National Reading Conference, 1], 1’5. Rankin, E. F., & Hess, A. K. (1970). The measurement of internal (intra-article) reading flexibility. In G. B. Schick & M. M. May (Eds.), Reading: Process and pedagogy. Nineteenth Yearbook of the National Reading Conference. Milwaukee: National Reading Conference. Ritty, J. M. (1979). The relationship of selected units of measure- ment in passages of varying readability to reading rate and number of visual fixations. Dissertation Abstracts International, 13, 3918A-3919A. (University Microfilms No. 80-01, 495) Robinson, F. P., & Hall, P. (1941). Studies of higher level reading abilities. Journal of Educational Psychology, 32(4), 24l-252. Rothkopf, E. Z., & Billington, M. J. (1979). Goal-guided learning from text: Inferring a descriptive processing model from inspection times and eye movements. Journal of Educational Psychology, 11(3), 310-327. Rothkopf, E. Z., & Coatney, R. P. (1974). Effects of readability of context passages on subsequent inspection rates. Journal of Applied Psychology, 33(6), 679-682. Samuels, S. J., Begy. G., & Chen, C. C. (1975). Comparison of word recognition speed and strategies of less skilled and more highly skilled readers. Reading Research Quarterly, 11(1), 72-86. Shores, J. H., & Husbands, K. L. (1950). Are fast readers the best readers. Elementary English, 21, 52-57. Spache, G. D. (1962). Is this a breakthrough in reading. Reading Teacher, 13, 258-263. ISI Tanenhaus, M. K., Leiman, J. M., & Seidenberg, M. (1979). Evidence for multiple stages in the processing of ambiguous words in syntactic contexts. Journal of Verbal Learning and Verbal Behavior, 13(4), 427-440. Thorndike, E. L. (1917). Reading as reasoning: A study of mistakes i3 paragraph reading. Journal of Educational Psychology, 3(6), 3 3-332. Thurstone, L. L. (1944). A factorial study of perception. Chicago: University of Chicago Press. Tinker, M. A. (1939). Speed versus comprehension in reading as affected by level of difficulty. Journal of Educational Psy- chology, 33, 81-94. Tinker, M. A. (1932). The relation of speed to comprehension in reading. School and Societ , 33(918), 158-160. Watkins, M. W. (1979). Modification of subvocalization levels in seventh-grade students with electromyograph biofeedback training and its effect on reading rate and comprehension. Dissertation Abstracts International, 13, 1372A. (University Microfilms No. 79-219 0047 Weaver, W. W. (1977). Toward a psychology of reading and language (A. J. Kingston, Ed.). Athens: University of Georgia Press. Wilson, M. M. (1979). The processing strategies of average and below average readers answering factual and inferential questions on three equivalent passages. Journal of Reading Behavior, 11(3), 235-245. Yoakam, G. A. (1928). Reading and study. New York: Macmillan.