={III-IIII_Ii IIIIEEE— '. ‘ n. 1-. , .- k g{”—¥"Va‘fll‘.\q .mm . gaff-Ivy: .iii’zfll'i-‘r . . oi .. R“ It! .v x .l. vest... turqliv o 1!"! lirbvvkr: r! (villa-4 ’ltrt. fr! u o b'o. it! .v {Fifi [I i. 9 11 . I! tivka v (CI lvtclfollrecV §.r.t.|ic fill-.90.: IE... rib 09%.... g!:.l 117': v?! .v-vfl...fveur..;\t.llfl . ‘.§§$rvvo%’q$.iig iii ‘ f gt’IIlrttnv it}!!! b. Evi‘ggigr , .is.%tt.t.t¢x-§ 555. 258%... {it . ........ gliitittttnflptbmvdlctzhhigii: rt Eliot... -EE‘R‘! . :zi‘t K ... I v L v {I . ts . z} .tflmflmfifi‘ti {1.1. SE , Shit .4 F . ‘ V . . u . .O . . ‘ .53.... .Pt‘itdgwtltstfgz it... . . §J . gait 7'73““: i‘flmuhnt‘: ”(ff-«UV: . . I “"0: D'§:§t 19‘1“ k. tu“;i . ind... . ivn§_¥.§i 2.5:: txi. «.21.»-.. . $3§gik§¥i¥ft §.u$..t . t . gitettkunnnfikzénylihnfinfiws 11...?»3 nip-TA? .1700. E?! ti.» .7 . ....\ £3 var; i. . igfii‘dl‘tz‘v“ O Et“k.§§fv II . - .r\ I.§i3£§§vflflwl vlgzj , , V . . st... thrrgtéflfitiagfiéd. Id; . 5%}. $343.. w {3%} 3... ‘ ix titikgltranflxhwmmufi. {knotssx 2:22 . . ‘ . s L.....§...It§fib..tvci Inca}. It): ‘ . ~ , A . {52v 5. €10.Mr.£nvlsfii= . .3 tiaranurii 1», [ll [.3 i iié , £9112. attitooufl 4 drug ‘gihfl! xv . . . , . . . ‘ . S . ~\ slit-1 ~. 3%.: 4 '1... [:3 2:31... . . ‘ , . ~15!!! g I; .. vi . . . {Oi leii . . A . 2:31.. 9.5%.94; gt. #0... 3... ll . .lo .. 4‘... 3.33. igly‘ 4.. .1. E! Héiébl it"s cunt: u «a «Ell... 65st} at“ iii-$5.53“. . WM t It“ 51 ‘ f. 5).. . .. .‘oh ilxlrfiixdikt ”1' JV“... "mum"... “N . \htfn. .nvgggii rat. nun}. :. 1}: 0 o It . .1 . . .e flan»... u: 5......”iimmmnxs itifia; ‘ 53?: .iWL‘I’: .iu.§§fi?§§§af , . .. 41L; {‘11.}? big e131 , A . 3 V . é a :1 . — I . . Flirts/(riff? Egg WI. 3. $16. E. k , . “trig’Aéli‘i ' A ‘ urn-)1”); ‘ it... {Artur Eff” EEC!” vhl . $331.?! ,Ett.t£t!i¥£,ktt$tltgpwr . riauwlutlt,.iillrugt+§£ri t, . . ‘ iii-lift}. ; . fig-using . . {.11 . 2E§.mfitu.hstsi}uzi tmwflkammtfi EA%¢§»2¥. Li. , king”. E28, .2999... L1 .‘tE 1‘! I u n. .‘ r'ig . {$11.1 \ , .. L ‘ haul} . k.”- éxa .rtilt . xx .. .4.) if‘; xxxxx .1 iigisoxysl . lit 1.2.}! ‘6. Halli! . -. A . . fNrRtto‘v’Pu igtzrlxr} . ‘ Lt LLB} ‘ , .. L§{§.i§§fi¥ .. I. gag? \htjixfittbnh“: hiring“; , ‘ a t... .2. . F....vr1t.§+th¥t‘ . ,gtnfiiigiler , . . . #11,...) {it . iii...xliihluufl..}sxislufiivts.§ .1. éfluigtétlifia \ Exit. ..AbLA5§ItLranva\Ll},-, “NH . t .1 {éi‘t‘i-Et‘tx 111E ti§§1 ‘Itv‘uhgtflt & It... a 5 via 2. i .% . . :31. ?t1 It ii . 1‘ , Eitihtflhg "at , it .0). {[72 .. L \i I... ptuw.‘§.itrru&x t. 1: L x; . .Jz S IIIIIIIIIIIIIIIIIIIIII )r- "_~\ I II 3 12931 University This is to certify that the thesis entitled LISTENING TO CHILDREN'S LITERATURE READ ALOUD: ITS EFFECT ON SELECTED ASPECTS OF THE NARRATIVE WRITING FIFTH GRADE STUDENTS presented by Janice Hart—w i ck Dressel has been accepted towards fulfillment ’ of the requirements for Ph.D. degree in Teacher Education Major profess MHW 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution MSU RETURNING MATERIALS: Place in book drop to remove this checkout from UBRARms _ your record. FINES will be charged if book is returned after the date stamped below. 17 K?9'L MW ”2’ pm o’29a :5 “a APQW “Uiéfi9-fifiwr LISTENING TO CHILDREN'S LITERATURE READ ALOUD: ITS EFFECT ON SELECTED ASPECTS OF THE NARRATIVE WRITING OF FIFTH-GRADE STUDENTS By Janice Hartwick Dressel A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Teacher Education 1986 COpyright by JANICE HARTWICK DRESSEL I986 ABSTRACT LISTENING TO CHILDREN'S LITERATURE READ ALOUD: ITS EFFECT ON SELECTED ASPECTS OF THE NARRATIVE WRITING OF FIFTH-GRADE STUDENTS By Janice Hartwick Dressel This research examined the effect of listening to literature read aloud on selected aspects of the writing of fifth-grade children with different reading abilities as determined by a standardized test. Literary quality and genre development were evaluated holistically using a primary trait instrument. The number and length of T-units and vocabulary maturity were also measured. Forty-eight children were randomly assigned within reading levels to two treatments. After the children had written a pretest detective story, literature was read aloud daily for eight weeks to both treatment groups. Children's series detective stories, rated high on all selected measures, were read aloud in Treatment l. Treatment 2 heard a different, less highly rated series. During the second half of each hour-long session, the children participated in a structured writing program, first discussing the professional literature from the author's perspective, then developing and writing their own detective stories. —4 WWW ' {mg-L Janice Hartwick Dressel A multivariate analysis of covariance showed an interaction effect significant at p < .11 for the holistic measures. Literary quality improved the most in the writing of the lowest readers hearing the sophisticated literature. Development of genre traits improved the most in the writing of the best readers hearing the sophisticated literature. A second MANCOVA found no difference by treatment or group for any of the atomistic measures. Results of the MANCOVAs were sup- ported by both the Mann-Whitney U and the Median Tests. Conclusions are that high-quality literature positively affects writing and should be read aloud in upper—elementary classrooms, although less experienced readers may benefit from hearing less sophisticated literature as they initially attempt to discern the narrative structure of a genre. If listening is to improve atomistic aspects of writing, they may need to be specifically called to children's attention. This research found no correlation between standardized reading comprehension scores and writing, when that writing was measured holistically for content. Since children identified as low readers appear able to control lan- guage when they are producing it, educators need to seriously consider using productive in addition to receptive language modes when evaluat- ing language competence. ACKNOWLEDGMENTS I wish to express my deep appreciation to the many people who helped make this study possible. To Dr. Sheila Fitzgerald, my major advisor, who has been a continual source of encouragement, a mentor, and a friend. To Dr; Patricia Cianciolo, whose willingness to share her love for the beautiful has sharpened my thinking and enriched my world. To Dr; William Schmidt, who taught me the value of statistics through his job in scholarship. To Dr. Larry Sarbaugh, who stimulated, encouraged, and assisted in this work, frequently over long distances. To my husband, Dave Dressel, who saw the value both in the doing and in the completion of this task, who willingly gave of his time, energy, and expertise, who worked beside me in everything that made this possible. To my daughter, Sherri, and to my son, Mark, who have under- stood my preoccupation, accepted my absences, and contributed in more ways than words can express. To Helen and Ernie Hartwick, who encouraged the pursuit of excellence. ii To the Reverend Oscar and Elsie Dressel, who provided a continual source of support. To Bill Main for his unending patience in helping me understand statistics; to Joshua Bagakas for his help in analyzing the data; to Ann Tracy and Mary Ellen VanCamp for the hours they spent rating the children's writing; to Gloria Boodt and Bette Bosma for their reading and rating of the children's literature used in this research; to Nancy Kopka for her faith, encouragement, and understanding; to the principal, teachers, and children who helped make this project a reality. iii TABLE OF CONTENTS LIST OF TABLES I I I I I I I I I I I I I I LIST OF FIGURES Chapter I. INTRODUCTION . . . . . . . . The Need for This Study . Background for the Study . Statement of Purpose . . . Assumptions . . . . Statement of Research Questions Quality of Writing in General Quality of Writing Within a Genre Fluency, Complexity, and Vocabulary Writing Improvement and Reading Ability Statement of Research Hypotheses . . . . . Quality of Writing in General Quality of Genre Writing . . . . . . Fluency, Complexity, and Vocabulary Writing Improvement and Reading Ability Limitations Definition of Terms . . . . . . . . . . . Overview . . . . . . . . II. REVIEW OF THE LITERATURE . . . . . . . The Relationship Between Children's Writing and Listening/Reading Summary . . . . I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I Writing and Children's Literature in the Classroom Summary Structure, Conventions, .and Children's Writing . . Summary Detective/Mystery Stories--Structure, Convention, and ChiIdren's wr1t1ng O I O C C 0 O O O I C I 0 Summary I I I I I I Page vii viii III. DESIGN AND IMPLEMENTATION OF THE STUDY, SELECTION OF THE LITERATURE, PROCEDURE The Pilot Study . . . . . . tation of the Study The Design and Implemen The Population Sample Random Assignment . Scheduling and Organi Selection and Analysis The Primary Trait Sco Rating of the Literat T-Units, Words Per T- T-Units and Words Per Weighted Vocabulary Procedure and Methodolo Introduction and Pret Treatment and Posttes Compilation of the Da Holistic Rating . . Atomistic Rating . . Overview . . . . . . . IV. ANALYSIS OF THE EXPERIMEN Overview . . . . . . . Correlation of Reading Analysis of the Holisti Literary Quality . . Development of Genre Literary Quality and Analysis of Ratings of Literary Traits . . Genre Traits . . . . Additional Findings . Summary . . . . . . . V. SUMMARY OF THE STUDY . . Overview of the Study Summary of the Findings Research Question l Research Question 2 Research Question 3 Research Question 4 Additional Findings Implications . . . . . Discussion and Observato 8, AND METHODOLOGY zation . . . . . of Literature . . ring Instrument . ure . . . . . . . Unit, and Vocabulary S T-Unit . CO scoreIIIIIIIII gy . est t D ta I I I I I I I I I I I I I I TAL RESULTS . . . and Writing Scores c Ratings . . . . Traits . . . . . Development of Genre T a t Individual Traits I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 60.0 I I I I I ore 152 158 158 163 163 167 174 176 178 181 187 Page Suggestions for Further Research . . . . . . . . . . . 196 sumary O O C O O O I I C O O l I O 0 C I O C C O I O 199 APPENDICES I I I I I I I I I I I I I I I I I I I I I I I I I I I 20] AI LITERATURE RATING FORM I I I I I I I I I I I I I I I I I 202 B. QUESTIONS DESIGNED TO FOCUS THE CHILDREN'S ATTENTION ON THE TRAITS OF THE CLASSICAL DETECTIVE GENRE . . . . . 206 C. CHILDREN'S WRITING BOOKLET . . . . . . . . . . . . . . . 216 D. CHILDREN'S EDITING WORKSHEET . . . . . . . . . . . . . . 231 E. SAMPLES 0F CHILDREN'S WRITING . . . . . . . . . . . . . 233 F. PRETEST/POSTTEST DATA . . . . . . . . . . . . . . . . . 253 BIBLIOGRAPHY I I I I I I I I I I I I I I I I I I I I I I I I I I 256 vi Table 3.1 4.1 4.2 4.3 4.4 4.5 LIST OF TABLES Gates-MacGinitie Reading Comprehension Scores (Normal Curve Equivalent) . . . . . . . . . . . . . . . . . . Professional Literature: Correlation of Rater Ratings of Literary Quality . . . . . . . . . . . . . . . . . Professional Literature: Correlation of Rater Ratings of Genre Deve1opment I I I I I I I I I I I I I I I I Professional Literature: Ratings From Literature Rating FOT‘m, Part I--L1t9ral"y Oua11ty I I I I I I I I I I I Professional Literature: Ratings From Literature Rating Form, Part II--Genre Development . . . . . . . . . . Professional Literature: T-Unit Counts . . . . . . . . Professional Literature: Vocabulary Score . . . . . . . Low-Frequency Words in Children's Writing Representing Immature Word Choice . . . . . . . . . . . . . . . . Comparison of Standardized Reading Test Scores (Gates- MacGinitie) With Pretest Writing Scores (Literary Qua11ty) I I I I I I I I I I I I I I I I I I I I I I Median Test: Literary Quality . . . . . . . . . . . . . Mann-WhI‘tney U Test ReSU1tS I I I I I I I I I I I I I Parameters for Individual Traits, Literature Rating Form, Part I--Literary Quality . . . . . . . . . . . Parameters for Individual Traits, Literature Rating Form, Part II-—Genre Development . . . . . . . . . . Children's Writing: T-Test--Literary Quality . . . . . Children's Writing: T—Test——Number of T—Units . . . . . vii Page 87 95 96 97 98 101 106 127 133 137 139 145 148 150 151 Figure 3.1 3.2 4.2 LIST OF FIGURES stUdy 09519" I I I I I O I I I I I I I I I I I I I I I Four Visuals: Literature as Developed in the Detective Sto ry I I I I I I I I I I I I I I I I I I I I I I I MANCOVA Results: Literary Quality of Children's writing I I I I I I I I I I I I I I I I I I I I I I MANCOVA Results: Development of Genre Traits . . . . . viii Page 82 111 136 140 CHAPTER I INTRODUCTION The purpose of this study is to determine how chderen's writing is affected by Titerature read aioud to them in the eiementary ciassroom. ATthough considerabie attention has focused on the importance of reading aioud to chderen before they enter schooT (Hoidaway, 1979; McConneTi, 1982), reading aioud to children during the schoo] day, particuTariy in the upper e1ementary grades, is frequentiy sacrificed to find additionai time to teach those subjects considered by many to be the basics. Lundstein (1979) pointed out that Tistening has been graveiy sTighted in instructionai programs of the past even though a1] of the Tanguage arts are both directiy and indirectTy dependent upon it (pp. 14-16). She cautioned that chi1dren may not have had much chance to Tisten to the Titerary use of Tanguage and encouraged the use of chderen's Titerature in the c1assroom (pp. 111- 12). Resuits of a weTT-known study by Chomsky (1972) indicated that exposure to compiex Tanguage goes hand in hand with increased knowiedge of the Tanguage. She suggested that chi1dren be permitted access to written Tanguage "we11 'above [their] Tevei' to take from it whatever they can" (p. 33). However, Chomsky a1$o noted that: Listening to books read aloud decreases sharply after first grade as the children's own reading begins to replace their listening. Even in the first grade independent reading is beginning to predominate for the more able readers. (p. 27) Although reading aloud to children is one way of providing access to written language "above their level," teachers read aloud even less as children advance in grade level. In a 1969 survey of teachers'read— aloud practices, Tom (1969) found that by fourth grade, fewer than 50% of the teachers read aloud regularly, and by sixth grade, only 26% do so (p. 174). Although he gave no evidence concerning the effect of reading aloud on children's writing, Trelease (1985), in his Bead-Aloud Hgndboo , extrapolated from this study by Tom to note that the first signs of a national decline in grade—level reading performance are shown by the same children, those in grades three and four, to whom teachers are reading aloud less (p.7). During the last 15 years there has also been increased interest in children's writing and in the relationship between the written discourse to which children are exposed and the writing they produce. Writings by Graves (1983a) and Holdaway (1979) as well as the work of Stewig (l978, 1983) and others attest to this trend. While much of the empirical research investigating this relationship has assumed a corre- lation between reading and writing, regrettably little has recognized the interaction between listening and writing. The distinction between listening and reading is frequently overlooked or ignored. In fact, research studies investigating the relationships between the receptive and the expressive aspects of language are often difficult to interpret since researchers and research reviews, particularly if their primary focus is on reading, often fail to differentiate between listening and reading as the mode of stimulus. Stotsky (1983), for example, in a synthesis of "findings from all the correlational and experimental studies that can be found on reading/writing relationships," found that there is very little research examining "the influence of .. . reading experience on the development of writing ability"(p. 627). Yet as reading experience, she cited both studies in which the subjects listened and studies in which they read. Chomsky (1972), in her clas— sic study concerning the stages of linguistic development and reading exposure, went even further and explicitly stated that it makes little difference whether children listen to or read a book: From the point of view of exposure to the written language. it may matter little whether the child has the book read to him, as would be the case with the younger children in our study, or reads it himself, as do the older children. It is possible, perhaps even likely that in both situations the contents, style, and language usage of the book are made available to the child with little difference in effectiveness. (p. 23) Much of the seminal work by cognitive researchers attempting to explain the active rather than passive role of language users has been applied primarily in the area of reading comprehension. The actual research, however, was often conducted using information transmitted to subjects through listening rather than reading. Pearson and Fielding (1982) argued convincingly that since much of the research in cognitive psychology, language arts, and literature on the active role of learn- ers has been conducted using listening rather than reading as the input mode, all the recent talk about active readers who construct a model of meaning for a text, all the work on schema theory and its applica— tion to reading practice should be regarded, if anything, as even more applicable to listening than it is to reading comprehension. (p. 625) Although it appears that "both reading and listening are controlled by the same set of cognitive processes" (p. 623), and training in specific tasks is often transferable between the two, Hildyard and Olson (1982) found that children use different strategies to comprehend narrative discourse when they listen than they do when they read. When they listen, children attend to information necessary for the coherence of the story theme and for the introduction of new events; as readers, they pay more attention to details. Research studies have provided empirical evidence that chil- dren's writing, as well as their reading achievement and vocabulary, is affected by the written discourse they hear and discuss in classroom— like settings. Their writing reflects the literary elements, such as plot, character type, motifs, and genre elements found in the stimulus literature (Duncan, 1981; Mikkelsen, 1983), and they demonstrate improvement in writing skills (Cohen, 1968; Mills, 1974L McConnell (1982) found that second-grade children with higher levels of prior literature exposure and writing practice received significantly higher holistic ratings on their original narrative writing than those with low literature exposure and infrequent writing practice. Mills (1984) conducted a study designed to teach composition using literary models. She found improvement in the writing of primary students from lower socioeconomic backgrounds. Working with fifth-grade children, Pinkham (1968) taught lessons during which she used literature to exemplify the characteristics of good writing. Although she obtained conflicting results, she concluded that the series of lessons "had a beneficial effect upon the written expression of urban children" (p. 122). She could not make such claims about her suburban sample. Duncan (1981) found indication, in preliminary analyses, that listening to a piece of literature followed by discussion of its narrative structure resulted in greater improvement by less able seventh-grade writers from middle— class backgrounds than by more able writers. Findings such as these are vitally important because they indicate that listening to stories may do more to improve the writing of elementary students than reading stories. The Need for This Studv A majority of studies involving the effect of listening to literature on reading or writing have been conducted either with children in the primary grades (e&L, Lyons, 1972; McConnell, 1982; Michener, 1985; Mills, 1974) or with children from lower socioeconomic backgrounds (Cohen, 1968; Mills, 1974; Pinkham, 1968; Porter, 1969). Many of the existing studies have been conducted using portions of literature selections rather than a piece of literature in its entirety (Duncan, 1981; Pinkham, 1968) and/or have attempted to use literature as a model to teach specific skills (Duncan, 1981; Mills, 1969; Pinkham, 1968). Of the studies investigating the effects of listening to literature on reading comprehension and/or narrative writing which have failed to find significant results, some have severely restricted the time to which the children were exposed to literature (Lyons, 1972; Nielsen, 1980) and others have attempted to measure results with standardized tests (Cohen, 1968; Lyons, 1972; Mills, 1969; Pinkham, 1968L Few studies have investigated the effect of the extended reading aloud of entire pieces of literature and discussing them upon the narrative writing of middle-class children in the upper elementary grades. There is a need to examine the effect of listening to litera- ture read aloud over an extended period of time on the narrative writ- ing of children in the middle grades. There is a need to know whether hearing a piece of literature read aloud in its entirety and discussed as a unified whole provides a context within which children can develop the necessary components of a given genre as well as the literary quality of their writing. There is a need to know whether the quality of the literature heard is reflected in the subsequent writing of the listener, either in the literary quality of that writing or in the development of the genre traits. There is also a need to know whether atomistic measures of the literature read (i.eu number of T-units, average length of T-units, maturity of vocabulary) will be reflected in the children's writing. Background for the §tudy Even very young children recognize that "story" language is different from everyday language (Applebee, 1978L Different uses of language are expected to be different and are approached with different expectations. In his analysis of language use, Britton (1970) called everyday language expressive language, language that relies on the shared representation of experience. It draws heavily on what people know in common. Story language differs from expressive language. With story language, authors attempt to create a new world, to lead an audience beyond the present into a unique experience generated by the author and structured by the "relationships among the various parts of a work" (Applebee, 1978, p. 13). When writing stories, authors guide the expectation and response of the listener or reader by carefully structuring the presentation of experience. The structure of the work itself must guide and provide a boundary for the response of the audience, for if the reaction of the audience deviates too far from that of the author, the pattern of interrelationships upon which the work is built begins to break down. According to Brown (1977), children, in their own writing, are able to produce cohesive stories—-those in which events are connected to a theme-—by the age of nine. Atwell (1981) found, however, that such writing is by no means automatic, and in actual practice, writing which reflects the absence of a guiding sense of structure is common even into college years. In 1979, King and Rentel attempted to set forth a theory of early writing development. They stated, What is needed is a framework for understanding how children's intentions in learning interact with varying learning contexts as they make the transition from speech to writing and, in particular, a framework that focuses on how children deyelop control over the written medium.(p.243) Reviewing studies concerned with children's telling and retelling of stories, King and Rentel concluded that children use an internalized story schema--an abstract representation of structural features—-to guide their telling or retelling of stories. Updating their theory in 1981, they further explained that children have already learned the underlying structure of stories by the time they go to school, that that structure is canonically organized, and that four and five year olds appear to rely on that schemata to tell about common events in their lives. In light of the evidence, they concluded that "story schemata may constitute one of the fundamental cognitive bases for the rhetorical scaffolds employed by beginning writers" (p. 147). They identified several reasons supporting their view that beginning writers derive this rhetorical framework from folk and fairy tales: children's texts "mirror" many of the elements identified as elements of folk and fairy tales; children reconstitute stories according to a stereotypical canonical pattern; children respond positively to and become deeply engaged with folk tales; and such stories have a highly convention— alized structure (p. 149). In order to guide production, King and Rentel hypothesized that any story schemata must contain those abstract elements that are necessary and sufficient to the story and that estab— lish the relations between characters and events while also providing the basic structure of action. Within this abstract framework, the teller has a range of options from which to select in order to give variety and dimension to his/her story. While King and Rentel acknowl- edge that "the extent to which such schemata guide production is not really known—-however appealing or likely such a notion might befl'they concluded that research has indicated that "children . . . acquire a general frame for fiction and then differentiate it into specific story genres" (p. 147). In light of their theory, King and Rentel believed that folk and fairy tales provide a rhetorical framework for early writers. Billman (1984), writing in Children's literature in education, argued that series mysteries pick up where fairy tales and fairy-tale- 1ike novels leave off. Highly popular with the preadolescent child, these series stories reflect the conventional nature of much folk literature and thus are predictable in style, content, and narrative pattern. As in fairy tales, mysteries are part of the comic tradition in their conventional triumph of the hero/heroine and their final predictable resolution. Beyond the psychological comfort provided by this predictable overcoming of evil and successful resolution, series mysteries assist the immature reader in the acquisition of what Billman called "literary literacyd' Such stories reinforce and extend the knowledge of story schemata for these readers and "help solidify for the inexperienced reader the elements and patterns of fictional narrative" (p. 34). By providing preteens with "comfortable fiction that they can come quickly to perceive as a recognizable story with recognizable ingredients, [and] by offering engaging cognitive games that show child readers their own responsibility in making literature meaningful" (p. 40), they provide the foundation for the reading of more sophisticated literature. They draw the child into interaction 10 with the text, into the predicting and confirming of predictions which is an integral part of the reading process. In spite of the fact that the reading of series mysteries has never been held in particularly high regard by educators, they are widely read by preadolescent children. The Stratemeyer Syndicate alone sells six million copies of series stories per year, many of them mysteries (Donelson, 1978L Although many of these mysteries rely heavily on coincidence to the detriment of logical deduction, all incorporate the traits or conventions of the classical detective story to a greater or lesser degree. Holbrook (1967) argued that conven- tional stories such as these provide preadolescent children with a socially acceptable and yet depersonalized means of expressing indi- vidual emotion while at the same time relating to a broader social structure. In stories such as folk tales and mystery stories which have highly conventionalized structures, the pattern of interrelationships is provided for both the author and the audience. To write such stories, however, children must become consciously aware of this structure, ixz, of the elements of the particular genre. To help children achieve the feeling of competence and the sense of liberation that accompanies it, Wilde and Newkirk (1981) believed that children must be helped to capitalize on what they already know. They maintained that: Students possess an intuitive awareness, a grammar, of the rules that govern detective story writing. This awareness comes from listening to and reading stories as well as from watching 11 television. But in order to use these rules in their writing, students must gain a conscious critical awareness of them. (p. 287) This study was designed to use series detective stories as a stimulus for children's writing in the same genre. There is strong evidence to support the appropriateness of these stories for the developmental level of fifth-grade preadolescents, thus providing them with an acceptable vehicle for involvement and self—expression. There is also strong evidence that children of this age level are at least unconsciously aware of the underlying structure and conventions of these stories. Upon becoming consciously aware of what they already know, children should be able to draw upon this knowledge and make it explicit in their writing. It is expected that by reading a different series aloud to each group, one series more highly developed than the other in terms of literary quality and development of traits of the classical detective genre, it will be possible to determine whether the quality of literature which children hear read aloud in the classroom affects their writing. Statement of Purpose The purpose of this study is to determine if the literary and linguistic quality of literature which children hear and discuss in the classroom is reflected in the quality of the stories they write concur- rently. This study has two objectives. The primary objective is to determine whether the quality of the literature which children hear and discuss is reflected in their writing. That is to say, if the children 12 hear stories which exhibit a high degree of literary craftsmanship, will the children themselves write better stories than children who hear stories less well crafted? Will their stories contain more ideas, more involved syntax, or more extensive vocabulary? Will they be more cohesive as reflected in the development of genre elements? The secondary objective is to determine whether the writing of middle-grade children, identified by a standardized reading test as being relatively low in reading ability, is affected by hearing and discussing litera- ture more than is the writing of middle—grade children with above— average or high reading ability. If children through the sixth grade are able to comprehend better through listening than through reading, as research has suggested (Hildyard & Olson, 1982; Stricht, 1974), it would seem that children for whom it is difficult to re-create a writ- ten text would benefit the most from hearing literature read aloud and that that benefit would be reflected in their writing. Assumptions 1. Children of school age have internalized the elements of the mystery/detective genre (Billman, 1984; Hubert, 1976; Wilde & Newkirk, 1981). 2. Children nine years and older have the developmental ability to write cohesive stories in which events are connected both to each other and to a central theme (Applebee, 1978; Brown, 1977; Emig, 1982). 3. Since writing ability is closely tied to socioeconomic class (Loban, 1963), children from middle to upper socioeconomic 13 backgrounds will be most likely, in a short-term study, to demonstrate improvement in literary quality and development of the traits of a genre. Statement of Research Questions The following research questions are examined in this study: anlity of Writing in general Does the quality of literature which is read to and discussed with children affect the quality of writing by those children? More specifically. do fifth-grade children who listen to stories of high literary quality write more literary stories, when those stories are evaluated holistically, than fifth-grade children who listen to stories of less sophistication? u lit 0 ritin it in enre Does the quality of detective literature which children hear affect their writing of that genre? More specifically, do fifth-grade children who hear stories in which the traits of the classical detec- tive genre are especially well developed receive higher scores on a primary trait analysis of the mystery stories they write than do children who hear less sophisticated detective stories? Fluenc om 1e it nd oc bu Are atomistic characteristics of the literature which children hear reflected in their writing? More specifically, does the writing of fifth-grade children reflect the degree of language fluency, 14 syntactic complexity, and quality of vocabulary characteristic of the stories which are read to them? Writing Improvement and Beading Ability Does improvement in writing correlate with type of treatment for children of all reading abilities? More specifically, does listen- ing to more sophisticated literature result in improvement in the writing of readers having differing reading abilities (as measured by standardized reading tests)? Do they respond in different ways to the two treatments? St teme t of ese rc ot eses The following research hypotheses were derived from the preced- ing research questions. ualit of ritin i ener l The holistic writing scores of the class which hears the literature rated as higher in literary quality will exceed the scores of the class which hears the literature rated as lower in literary quality. Quality of Genre Writing The scores of the class which hears the more highly rated literature will exceed the scores of the class which hears the less highly rated literature when evaluated by a pcimary trait analysis of detective stories they write. 15 F enc om exit nd oc bu r The class which hears the literature with greater language fluenCy, syntactic complexity, and quality of vocabulary will receive higher scores in corresponding measures of those qualities, 1.9" number of T-units, length.of T-ugits, and weighted vocabulary score than the class which hears literature less replete with these qualities. ritin m rovemen nd e in bilit Readers with lower ability will show greater improvement in both groups than readers with average or above-average ability on all measures of writing: holistic scores, primary trait analysis, fluency of language, syntactic complexity, and quality of vocabulary. Lower- ability readers in the group hearing literature rated high in literary quality will show greater improvement on all measures than the group hearing the literature rated lower in quality. Limitations 1. For purposes of the study, it was important that all of the children write under similar conditions. Therefore, the children were permitted to write only during the listening/writing sessions or arranged make—up sessions; the researcher collected the stories and kept them until the next session. Children were not permitted to write these stories at other times during the school day or at home. This meant that the children had to stop writing when the class period was over and could not write again until the next writing session. For 16 children, as for any author, this frequently resulted in the interrup- tion of thought patterns and ideas which the children complained they "couldn't remember" when they returned to their stories and attempted to continue. Children frequently pleaded for more time to finish. When told they had to go to the next class, they would ask if they could plaaaa take it home and write. Whenever such a request was forthcoming, the feeling of immediacy--of having to get it down now-— was apparent. Tomorrow was too far away. It is likely that the arti- ficial constraints imposed by this limitation diminished the quality of writing of some if not all of the children. 2. The study was designed to span eight weeks, although it actually extended beyond the ninth week to enable each of the children to complete a finished product. Even though this encompassed the entire language arts time for one-quarter of the school year, it is a relatively short period of time in which to hope for an effect of literature on children's writing. 3. Since children in both treatment groups were drawn from two classrooms, it is possible that the children discussed the different treatment literature upon returning to those classrooms. Since the treatment was delivered orally, however, it is unlikely that this caused a contamination in the study; the children could not replicate the stories or the exposure to the quality of literature. Comments from the children indicated that many, if not all, of them were unaware, even at the end of the study, that the treatment groups were hearing different stories. 17 4. The process of revision was modeled briefly for the children in this study, and they were provided with the opportunity to participate in a guided activity involving revision. They appeared to have little prior experience in revising their own work, however. This was demonstrated by the fact that a majority of the children made no content changes upon the completion of their first draft, even though this process had been modeled in class and they were encouraged to do so both in their writing booklet and orally by the researcher. As a result, most of the stories were edited first drafts, not revised pieces of writing. W Mysteryldetective story. For the purposes of this study, mystery/detective story refers to the classical detective story. A classical detective story is defined as one in which the problem is created and solved by the author and presented in such a way that an astute reader has the opportunity to reason out the solution. The problem is solved by the assembling of evidence and the use of deductive reasoning, not by a trick of the author (Wells, 1929). Quality of literature. Literature of high quality has met the criteria of literary excellence. This literature exhibits outstanding quality in the literary elements of plot, characterization, style, setting, and theme, and it appropriately reflects the characteristics of the genre within which it is written (Ross, 1982). Not definable in statistical terms alone, the quality of literature is often measured by holistic rating by competent, trained raters (Cooper, 1977). 18 Holistic measures of yriting. In evaluating writing holis- tically, samples of writing are considered only as whole entities. Such evaluations may view the sample of writing as representative of all discourse, as in the method employed by the Educational Testing Service, or may identify a particular subcategory of discourse and determine the appropriateness of the sample within the established boundaries of that subcategory, as in the Primary Trait scoring method developed by the National Assessment of Educational Progress (Lloyd- Jones, 1977). Atomistic measures of writing. Although the boundary between holistic and atomistic measures of writing is sometimes disputed, "atomistic tests rely on the assessment of particular features asso- ciated with skill in discoursing" (Lloyd-Jones, 1977, p. 33). These features are isolated from the context and scored separately, often through counting (Cooper, 1977; Lloyd—Jones, 1977). Atomistic features selected for evaluation in this research are the number of T—units used in the writing, the length of those T-units, and the maturity of the vocabulary used. Language fluency. This is an index of the number of ideas expressed and measured by the total number of Communication Units (T-units) in a piece of writing (Evanechko, Ollila, & Armstrong, 1975). Syntactic complexity (also labeled syntactic maturity by Hunt, 1979, and syntactic fluency by Odell, 1981). This term refers to the elaboration of language or sentence maturity, ten the level of syn- tactic development. The level of maturity is reliably indicated by the 19 average length of a writer's T-units, and was one of 11 language characteristics displayed by the subjects in Loban's (1963) study who teachers felt had impressive language power. "Growth in syntactic fluency is reliably indicated by increases in the average length of a writer's T-units" (Odell, 1981, p. 121). Jfiiuflla Also called a minimal terminable unit or a Communica- tion Unit, this term is attributed to Hunt (1970). It is defined as an independent clause with its modifiers, i.e., a group of words that cannot be further divided without a loss of their essential meaning. It is an index of the number of ideas presented by a writer, and thus a measure of fluency in language (Evanechko et a1., 1975, p. 324% Loban (1976) used a similar unit in his research but called it a Communication Unit. Evanechko et al. found the number of Communication Units in a piece of assigned writing to be a significant predictor of children's reading achievement as measured by a standardized test (Bond—Balow-Hoyt New Developmental Reading Test Intermediate Level, 1965, with reliability and validity recognized by Buros, 1972). Length of T-unit. This is a reliable indicator of growth in syntactic maturity. The length of T-units in a piece of writing is directly related to linguistic complexity; i.e., the longer the T-unit, the more complex the language is likely to be in transformational terms (Applebee, 1978, p. 182). Maturity of vocabulary. The maturity of the vocabulary used by an author depends on the frequency with which the individual words chosen by that author are used in the English language within a 20 specified context. The more rarely a word is used, the more mature its use, and the higher its rating (Cohen, 1968). Maturity of vocabulary was determined by comparing each different word used by a writer with the American Heritage Word Frequency List (Carroll, 1971) to determine a point total. genres A genre is a distinctive category or class of litera- ture which can be described by its primary and secondary traits. Primary trait scoring. Primary trait scoring is a sophisti- cated, holistic method of measuring writing which focuses the rater's attention on specific features of a particular kind of discourse (Mullis, 1980), in this case the classical detective genre. Primary traits. Primary traits are those traits that make one genre of writing different from another. These are also the conven— tions of that mode of writing. For a summary of the traits of the classical detective story see Appendix A, Literature Rating Form: Part II. Qverview Chapter I contained the introduction to the research study, including the statement of the problem, the need for the study, the background for the study, a statement of purpose, the assumptions of the study, and a statement of the research questions and hypotheses. The limitations of the study and definitions of terms used in this research project concluded Chapter I. 21 Chapter II contains a review of the literature pertinent to this study. The review is concentrated in four areas: (a) the relationship between children's writing and listening/reading; (b) writing and childrenksliterature in the elementary classroom; (c) structure, conventions, and children's writing; and (d) detective/ mystery stories--structure, convention, and children's writing. Chapter III describes the pilot study, the design of the study and the population sample, the selection and analysis of the literature selections, the procedure and methodology, and the compilation of the data. Chapter IV contains the analysis of the data which was submit- ted to a multivariate analysis of covariance. Chapter V contains a summary of the study, reflections and observations, and suggestions for further research. CHAPTER II REVIEW OF THE LITERATURE The review of the literature is divided into four major areas which are pertinent to this research study: (a) The Relationship Between Children's Writing and Listening/Reading; (b) Writing and Children's Literature in the Classroom; (c) Structure, Conventions, and Children's Writing; and (d) Detective/Mystery Stories--Structure, Convention, and Children's Writing. A summary concludes each major section. Ine Belationsnip Between Children's Writing and Listening/Reading A now-classic study by Chomsky (1972) was designed to investi- gate children's acquisition of syntactic structures and to explore the relationship between the children's exposure to written materials and their rate of linguistic development. Thirty-six children between the ages of six and ten participated in the experiment. Chomsky's findings indicated: (a) the sequence of acquisition of syntactic structures is developmental, and (b) there is a strong relationship between exposure to the more complex language available in books and the child's linguistic development (pp. 29-33). Of particular interest to this study is Chomsky's finding that for the older children, who were all from relatively high socioeconomic backgrounds, linguistic 22 23 progress was related to the child's own activity in relation to reading (p. 28). The effect of exposure to literature and early writing practice on children's narrative writing was the focus of a study conducted by McConnell (1982). Using scores from a literature inventory and from teacher and home questionnaires, McConnell categorized 144 second-grade children based on amount of prior literature exposure and writing practice. Samples of the children's original narrative writing were 1 then rated holistically and analyzed for vocabulary and story structure elements. Using an analysis of variance, McConnell found that those children with high literature exposure and a variety of early writing experiences received significantly higher holistic ratings than chil- dren with low literature exposure and infrequent writing practice. She did not find significant differences for vocabulary or story structure elements. The most extensive investigation to date of the relationship among the language arts is a 13-year longitudinal study, K-12, con- ducted by Loban (1963, 1976) and begun in 1952. The original sample for this study consisted of 338 children who were chosen using the stratification variables of sex, racial background, and intellectual ability together with proportional allocation of socioeconomic background. The statistical analysis for the study was conducted using the data for 105 of the 211 subjects remaining at the conclusion of the study. Thirty-five children were selected on the basis of their high general language ability, 35 on the basis of their low general language 24 ability, and 35 were selected for a random group by using a table of random numbers. General language ability was determined by a vocabu— lary test administered in kindergarten together with a rating from each elementary teacher on the child's activity in speaking, writing, and reading (1976, p.2). Loban (1976) found that writing was related to socioeconomic status. Those children who had parents or guardians with occupations identified as professional, semi-professional, managerial, clerical, skilled trade, or retail business were above average in writing. Those whose parents were in lower occupational categories were rated below average in writing (p. 85). Loban (1976) also found a high correlation for the upper- elementary grades between reading achievement and writing scores. Reading achievement was measured by a standardized test and was found to be related to general language ability. Writing scores were obtained from compositions written by the children in response to a picture prompt. The writing was measured by two judges, both teachers of writing, using a five-category classification: superior, good, inferior, illiterate, and primitive (pp. 25-26). These classifications involved the evaluation of content, mechanics, style, and vocabulary. Loban found that the correlation between reading and writing became stronger as children went up in grade level. Of fourth-grade children in the two upper quartiles in reading, determined by averaging the paragraph and word meaning scores on the Stanford Achievement Test, 26% were below average in writing (illiterate according to the 25 classification system). Thirty percent of the students in the lowest two reading quartiles were above average (superior) writers. By sixth grade, however, the relationship became stronger. Only 20% of students above average in reading were judged to be below average in writing, and only 17% of those below average in writing were above average in reading (1976, pp. 72-75% In his conclusions, Loban argued that proficiency with language for elementary students comes through opportunities to grapple with their own thoughts as they attempt to successfully communicate with a real audience. He believed teachers can support this by providing models of effective communication and by encouraging students, both individually and in small groups, to reason and search for meaning (p. 88). In their survey of literature relating to listening comprehen— sion, Pearson and Fielding (1982) highlighted what they perceived to be a discrepancy between the ability to comprehend the spoken word and the inability to comprehend the written word for children who do poorly on reading comprehension tests. To substantiate this, they summarized two lines of research pertaining to the development of listening and reading. The first pertained to the advantages of listening and read- ing at different age levels. The second compared the linguistic tasks in which readers and listeners must engage to make sense of the graphic or auditory data, respectively (pp. 621-23). In summarizing the research pertaining to the advantages of listening and reading at different age levels, Pearson and Fielding 26 drew heavily on a review by Stricht, Beck, Hanke, Kleiman, and James (1974). Stricht et al. reviewed 31 studies that compared reading with listening comprehension at different grade levels (p. 60). They found that, for grades one to six, the results of these studies showed that children comprehend better through listening than through reading (p. 62). The second line of research presented by Pearson and Fielding drew on work by Schreiber (1980) and focused on the differing linguis- tic tasks facing readers and listeners. Schreiber compared reading and listening comprehension by focusing on particular aspects of auditory messages not well transmitted through the written medium. Schreiber contended that the acquisition of reading comprehension, beyond the ability to recode, involves the reader‘s ability to compensate for, Le., to supply, the oral signals or prosodic cues not present in the graphic representation of language (p. 178). Drawing on his own research as well as the research of others such as Samuels and homsky, Schreiber presented strong evidence that hearing the oral eading of competent adults is a crucial component for many children n the movement from decoding to reading with comprehension and fluency p. 186). Pearson and Fielding noted that much of the research reflecting he cognitive perspective of language, which argues for activeness ather than passiveness on the part of the language processor, has been onducted with listening rather than reading as the mode of input. his view of the language user as an active participant is reflected in 27 the work of Rosenblatt (1939), Smith (1971, 1978), Goodman (1965), Goodman and Goodman (1979), Graves (1978, 1982), and others. Pearson and Fielding argued that the results of this research are even more applicable to listening than to reading comprehension even though little has been written about listening from the cognitive perspective (p. 624). Cohen (1968) investigated the effect of reading literature aloud on the vocabulary and reading achievement of second-grade chil- dren. Beginning with 580 original subjects, Cohen's final analysis involved 285 subjects. Her data were collected in seven New York City schools designated as Special Service Schools because of academic retardation, low socioeconomic population, and a high percentage of ethnic and racial minorities. The study spanned the length of the school year during which time the experimental and the control groups both continued to receive reading instruction from a basal series. Teachers of the experimental groups were asked to read a story aloud every day of the year and to engage their classes in follow-up activi- ties suggested in a prepared manual. Books to be read were selected according to specific criteria and provided by the researcher. Teach- ers of the control groups were asked to follow their usual routine, reading stories as an occasional treat if at all. These stories were not chosen according to any specific criteria. Using an analysis of covariance, Cohen found the following results: (a) the experimental group demonstrated an increase in vocabulary development over the control group, significant at the .005 28 level; (b) the experimental group demonstrated an increase in word knowledge, as measured by the Metropolitan Reading Achievement Test, over the control group, significant at the .005 level; (c) the experi- mental group showed an increase in reading comprehension, as measured by the Metropolitan Reading Achievement Test, over the control group, ignificant at the .01 level; (d) the experimental group was superior to the control group in quality of vocabulary at the p < .05 level; e) the three experimental classes composed of the lowest reading froups, as determined by reading and reading-readiness levels at the and of first grade, showed an increase over the three control classes 'n Word Knowledge and quality of vocabulary at the.05 level, and in leading Comprehension at the .005 level. Cohen concluded that comprehension of meaning through oral anguage is basic to growth in the language arts and that regular istening to story books chosen for their ease of conceptualization and notional appeal aids, among other things, narrative sense, recall of tretches of verbalization, and the recognition of newly learned words 1 other contexts. While the relationship Cohen found between reading :ories orally to the children and the children's improvement on read- 9 measures may have been influenced by the use of follow-up activi- es "to strengthen comprehension of the story and individual words" 211) with the experimental but not the control groups, this study particularly significant because of its length and the number of '1dren involved. 29 Attempting to determine whether reading aloud to inner—city, iddle-grade children would affect their reading achievement and iterest in reading, Porter (1969) conducted a study involving a sample :pulation of 1,202 fourth-, fifth-, and sixth—grade students. High .hool juniors read literature aloud to the experimental group for 20 eks; the control group did not participate in the oral reading ssions. To determine reading achievement, Porter used a series of tests to analyze the pre- and post-reading achievement scores of 954 udents on the reading section of the California Comprehension Test of sic Skills. She found that the reading achievement scores of the aerimental group were significantly higher than the scores of the itrol group after listening to the literature. Lyons (1972) conducted a study to investigate the influence of ) programs designed to enhance language development on the reading lievement of middle—class children in grades one and two. The first natment involved reading aloud to children from children's literature ected on the basis of its syntactic complexity and richness of :abulary and discussing that literature. The second treatment pro- .ed the children with increased opportunities for active language use Lugh discussion and oral language activities but did not involve ‘tening to children's literature. The treatments were administered f69 children in eight classrooms, one of each grade in four schools. ladditional classroom from each grade provided a no-treatment con- 1. The lessons were conducted for 20 minutes three times a week for I %eeks. While analysis of the data indicated that neither treatment 1 4 1 30 :as significant, Lyons did find some evidence to indicate that reading :0 children more often and reading from literature recommended for rades higher than the one hearing the literature can affect reading chievement. Lyons reasoned that, since her subjects were from middle rather han lower socioeconomic backgrounds, the language of the literature sad might not have provided as great a contrast to the child's daily inguistic environment as had been true in previous studies which used is reading aloud of literature as the experimental treatment. She Iggested that this may have been one of the factors precluding the ndings of significant results. It also seems probable that using andardized tests to evaluate the reading achievement of first- and cond-grade children may have failed to measure improvement which tually did occur (Lyons used the California Achievement Test-- ading, 1970, Level 1, the Peabody Picture Vocabulary Test, and a nguistic Structures Repetition Test). Increasing the length of eatment or analyzing the data by reading ability level or sex rather an by total class might also have increased the chances of finding 3nificant results (Chomsky, 1972; Cohen, 1968; Porter, 1969). In a quasi-experimental study, Michener (1985) investigated the fects of reading aloud on the written composition of third-grade %ldren. Students from four third-grade classes were stratified to iresent low and middle socioeconomic levels. Forty-seven students e then chosen at random to participate in the 12-week study. chers of the two experimental groups were trained and asked to read 31 aloud for 15 minutes a day from a set of SO literature selections; teachers of the two control groups did not read aloud during this time. Samples of writing were elicited using parallel sets of writing prompts. Pretest writing samples were collected and used as covariates in a posttest and a delayed posttest multivariate analysis of covariance. The writing was measured for writing style using holistic impression scoring, for syntactic maturity using a T-unit analysis, and for seman- tic maturity using an analysis of specifically selected words. No differences were found due to treatment or group in either the posttest or delayed posttest analysis. Michener recommended the need for more valid and reliable measuring instruments for writing as well as con— trolled field settings. Strickland (1971) investigated the effect of listening to children's literature and participating in oral language activities on the acquisition of standard English. Strickland collected data from 94 hildren selected from eight kindergarten classes located in lower ocioeconomic areas of two metropolitan communities. Using 50 selected hildren's books placed in the classrooms, teachers read aloud daily, rom November to May, to children in the experimental and control roups. Following each reading session, the experimental group articipated in activities designed to involve the children in oral anguage participation; the control group participated in an activity hich did not encourage oral language participation. Strickland found hat those children who participated in oral language activities after aring stories read made significant gains in the acquisition of 32 standard English without negating their native dialect. The control group, which was exposed to the oral reading but did not participate in a follow—up oral language activity, made no measurable gain. For the population in this study, hearing literature read aloud did not, by itself, bring about change in language use. Only the group which participated in oral language activity following the oral reading made significant gains in language expansion. Evanechko et al. (1975) asserted that "the concept of a strong relationship among language skills has become axiomatic" (p. 315), but that the nature of these interrelationships has not been made clear through research. They concluded that: If it could be determined that certain language competencies are common to both reading and writing, then development in these areas could be planned to occur in these subjects concurrently to permit reinforcement as the child uses skills learned in the receptive reading process to apply in the expressive writing process. (p. 316) Their study was designed to determine which aspects of writing perform- ance would best predict reading achievement. The subjects of the study were 118 sixth-grade children in four classrooms of a school located in a middle-class neighborhood in Victoria, British Columbia. Writing samples were obtained by asking all classes to write a paper on a :ommon theme. Instructions were common to all classes, and there was set period of time for writing. Syntactic complexity of the chil- ren's writing was determined using a modified formula developed by otel and Granowsky (1972). The criterion measure was the Bond-Balow— oyt New Developmental Reading Test Intermediate Level (1965)-—subtests n Basic Reading Vocabulary, Reading for Information, Reading for 33 Relationships, Reading for Interpretation, Reading for Appreciation, Literal Comprehension, Creative Comprehension, and General Comprehen- sion. Both measures were administered by the classroom teachers. An analysis of the data revealed that "the single best predictor of read— ing achievement was the number of Communication Units [same as Hunt's T-unit]. This measure was the most powerful predictor on all subtests of the reading achievement test" (p. 322k 0f the 13 syntactic forms measured in the children's writing, only four were significant predictors of reading achievement, and two-- the Communication Unit (also called a T-unit) and Two Count Structures (i.e., passives, paired conjunctions, dependent clauses, comparatives, participles, infinitives as subjects, appositives, and conjunctive adverbs)-—were consistently first in the regression equation. Evan- echko et a1. concluded that if fluency (as measured by the Communica- tion Unit) and syntactic complexity (as measured by Two Count Structures) are key language competencies underlying reading achieve- ment, then building on these two competencies in oral and written discourse could improve reading achievement. While other studies have sed T-units to measure children's free writing, their rewriting (Hunt, 1965, 1970), or have averaged a T-unit score with other measures of hildren's writing (Loban, 1963), Evanechko et a1. here introduced the se of the T—unit to measure children's assigned writing. Exposing children to written discourse through listening ppears to have different results than exposing them to written iscourse through reading. In a study involving 72 children, 36 third 34 graders and 36 fifth graders, Hildyard and Olson (1982) found that readers and listeners adopt somewhat different strategies in compre- hending narrative discourse. In understanding a narrative story, chil- dren appear to listen more to what is meant, but they read more for what is said. Listeners pay attention to information which is neces- sary for the coherence of the main story theme and for the introduction of new events. They do not recall, as well as readers do, elaborating information unnecessary to comprehension of the story as a whole. Readers, on the other hand, pay attention to all specified details, including incidental ones, and they are more able than listeners to identify whether statements are actually presented in the story or implied by it. A 1966 study by Bergdorf provided data compatible with the {findings of Hildyard and Olson. Her study involved 432 fourth, fifth, :and sixth graders from various socioeconomic levels. Bergdorf Robtained results indicating that children are better able to interpret lliterary materials read to them than material they read themselves. EHer findings also indicated that children attend to different aspects bf language when reading than when listening. 1 Stotsky (1983), in a synthesis of correlational and experi- mental studies on reading/writing relationships, found that better Writers tend to be better readers, to read more, and to produce more syntactically mature writing than poor readers. In her summary, 1 however, Stotsky reviewed studies involving listening and studies involving reading without distinguishing between them. Her findings, If— """ ‘ /' v. .7;_7_ a; y ,, ,_ , y/ 35 therefore, applied to receivers of both oral and written language, not necessarily to readers as such. Based on findings drawn from these studies, she concluded that "reading experience" is a consistent cor— relate of, and may be a critical factor in, developing writing ability. Although she found that studies using literary models noted significant gains in writing, a majority of these studies were conducted at higher educational levels, used expository or informational reading material, and were concerned with ways to improve the acquisition of academic discourse. She reviewed only one study relating to narrative writing (Mills, 1974L That study was conducted with elementary students and examined the effect of listening to literature read aloud on selected aspects of the children's writing. Stotsky recommended further research to determine the nature of the influence of "reading expe- riencefl'i.e” to determine whether certain types of "reading experi- ences" are particularly beneficial to developing writers and whether the quantity and quality of the "reading" that good and poor writers do is important. Research, as well as common sense, has indicated a strong elationship among the language arts. The amount of exposure to the omplexity of language as captured in the written word appears to be trongly correlated with linguistic development. Exposure to complex aterials through listening and/or reading is characteristic of the ighest stage of development (Chomsky, 1972; McConnell, 1982L 36 Linguistic development continues well into the elementary years and is reflected in the relative consistency of performance in all of the language arts (Loban, 1976). The general relationship between reading achievement, as measured by a standardized test, and writing ability, as measured by standardized tests, T—units, and teacher judgment, has been well substantiated (Cohen, 1967; Evanechko et a1., 1975; Loban, 1976). Research studies in the area of reading achievement, vocabu— lary, and language development of children from lower socioeconomic backgrounds have shown that exposure to children's literature, espe- cially when extended through discussion and follow-up activities, has a significant effect on performance in the other language arts. Listen- ing to literature read aloud resulted in improvement in vocabulary, reading achievement, narrative sense, and use of standard English for the children in these studies (Cohen, 1967; Porter, 1969; Strickland, 1971). Less is known, however, concerning the effect of listening to children's literature on children's writing when their writing is evaluated holistically, or when the children are from other than lower socioeconomic backgrounds. Such research has been inconclusive at best (Lyons, 1972; Michener, 1985). Relatively recent research has indi- ated that children attend more to information necessary for the coher— nce of the story theme and the introduction of new events when they isten than they do when they read (Hildyard & Olson, 1982L Research as also indicated that children up through the sixth grade comprehend etter through listening than through reading (Stricht et a1., 1974L his may be due to the fact that reading comprehension beyond the level 37 of simple recoding involves the ability to assign prosody to a written text. For many children, especially those with reading difficulties, hearing the oral reading of a competent adult aids in the movement from recoding to fluency and comprehension (Schreiber, 1980% Writing and Childnen's litaratune in the Classroom Mitchell (1977), a classroom teacher, clearly stated the dilemma of teacher and student when she said, I [also] knew that students needed models for their writing . . . to ask them to write a story leaves most of them staring blankly ahead. When they can write about anything, they are generally overwhelmed and retreat. There are lots of ideas floating around in their heads but these ideas need some sort of magnet to pull them together and give them focus. I saw that using children's literature could be another way of providing limits and focus for their writing topics. (p. 62) The dilemma of how to interrelate literature and writing has been a topic of much concern in educational literature. Haley-James (1981) summarized the literature pertaining to the use of children's literature in the classroom as follows: Professional references generally support the positive effect that listening to and reading good literature has on composition, but as for teachers actually using such literature as a model children can emulate, opinion is divided. Prior to 1950, it was not uncommon to see such a recommendation. However, while Hughes Mearns, who published primarily in the 1920B» and the majority of authorities publishing since 1950 have advocated exposing children to good literature, they objected to the use of models of adult writing. . . . James Moffett (1968) and John Stewig (1980) have taken a middle of the road position by recommending examples from published literature be used just to initiate writing sessions. (p. 4) In more recent years, the work of Graves (1983) and his ssociates has spurred a resurgence of interest in the incorporation of 38 children's literature into the classroom. Graves, however, advocated that no distinction be made between the children% writing and the writing of professionals. Both are treated as important, and both are examined critically (pp. 65-76). How teachers can best help children see the relationships between professional literature and their own writing remains a controversial issue. The literature reflects a continuum of studies and opinions that range from learning by imitation to learning through exposure. McCampbell (1966) described the use of literature models as one technique for improving the teaching of composition. He advocated analyzing models for patterns of written language, i.eu how to express, not what to express. To do this, students inductively analyze models and then imitate phrases, sentences, paragraphs, or organiza- tional structure first in whole-class writing, then in small—group writing, and finally in individual writing. They also consider the connection between conventional literary forms and particular kinds of meanings, observing that form is often a clue to meaning. McCampbell maintained that the use of a model is not restrictive, but is an aid in improving expression. A 26-week study to determine the effect of models imitation on the writing of seventh-grade students was conducted by Martin (1980L Sixty-one students in the experimental group participated in 20 lessons involving models imitation. The control group of 62 students studied 0 composition lessons taken from a standard seventh-grade textbook. ive pretreatment and five posttreatment writing samples provided the 39 data for analysis. Writing was evaluated for overall quality by a panel of nine teachers using subjective ratings and for syntactic maturity by scores on six indices. Martin did not find significantly more improvement in the overall quality of writing or in syntactic maturity for the group using models imitation. He did find pre—post treatment growth in words per T—unit, words per clause, and adjective clauses per 100 T-units. This study indicated that while the imitation of another's writing may affect atomistic features, it does not seem to stimulate an increase in overall quality of writing. Olson (1983), a cognitive psychologist at the Ontario Institute for Studies in Education in Toronto, discussed the use of modeling in an interview for language Arts. He dismissed as trivial the behavior- istic view of modeling, which he saw as imitation since it does not contain the component of comprehension. According to Olson, comprehen- sion involves interpretation by the child of what is seen to determine what was done and why. He maintained it is this interpretation, not imitation, that guides performance. It appears, however, that reading materials in the classroom may become models for children's writing by default as well as by design. In an exploratory study, Eckhoff (1983), a PhlL student with Carol Chomsky at Harvard, obtained writing from two groups of second- grade children in response to writing stimuli adapted from the 1969-70 NAEP. She found that the writing of each group reflected the linguis- tic structures as well as the format and stylistic features of the particular basal text in which they were reading. Using Huck's 40 Inventory of Children's Literature, she determined that children in both groups had equal knowledge of children's literature, indicating that their outside reading was unlikely to have influenced the observed differences in their writing. Although Eckhoff did not indicate whether oral reading and listening were involved, it is highly unlikely in light of current teaching practices that children assigned to basal readers 1.2 and 3.1 read only silently. It is thus possible, there- fore, that these results may be attributable, at least partially, to listening rather than reading. In a frequently quoted article, OWJea (1965) criticized five approaches to the teaching of writing. Two of the approaches identi— fied by CfiDea are particularly relevant to this discussion: (a) that students learn to write by reading great literature and (b) that stu— dents learn to write essays by analyzing professionally written essays. CfiDea opposed the first approach because of its demand for close rather than wide reading. He contended that many teachers spend a great deal of time reading and discussing one piece of literature on the pretense that it will improve writing, while, in essence, such teaching is the result of the teacher's desire to teach literature rather than writing. (NDea opposed the second approach because of its assumption that students must be given something to write about, which results in forcing the student "to parrot or to assume an artificial rhetorical stance because of what he has read" (p. 329). In an overview of research on using models to teach writing, Smelster (1978) defined modeling as an approach that uses imitation to 41 develop the skill of writing through a study of ideas and structures. Smelster indicated that a number of authorities in the field find value in the method, even though each adds his/her own particular cautions. She noted that Miles Myers, Administrative Director of the Bay Area Writing Project, has found that this approach may be effective if used together with an understanding of the writing process, sentence combining, and students' views of the world. Sara Lundsteen feels that intermediate steps are needed between the presentation of the model and the child's own writing to give the child a sense of ownership. Even (rDea, who identified the idea that students learn to write by reading great literature as one of his myths, conceded that the assumption has some truth but argued that the models provided are usually not clearly related to the writing assignment. Smelster cautioned that the teacher must be sure that the product does not become the goal. Instead, models should be used as part of a sharing process in which children are encouraged to build on ideas gained, and to give those ideas their own interpretation and treatment. In her PhIL study, Mills (1967) examined the use of literary models in teaching composition to students in four fifth-grade class- rooms from two schools in Clarke County School District of Athens, eorgia. For 24 weeks, written composition lessons were taught to an xperimental group of 45 children twice a week for an hour. A control roup of 77 did not receive instruction. Lessons used were based on he literary models found in specifically selected children's litera- ure. Mills obtained conflicting results from her two measures--pre 42 and post Sequential Tests of Educational Progress (STEP) Writing Tests and pre and post Writing Samples. While a significant difference after treatment was not found on the STEP Writing Test, a significant differ- ence after treatment was found on the Writing Sample. Mills felt that the Writing Sample was more relevant to the emphasis of the study. In the October 1974 issue of Elementary English, Mills summarized a similar three-year longitudinal study conducted in two schools in Madison County, Georgia. Beginning in the first grade, the experimental group participated in weekly half—hour sessions, taught by the researcher, for 24 weeks each year. The socioeconomic level of the students was predominantly low. The children listened to literature read aloud and then examined that literature as a model for their own writing. Mills stated that "in all cases, first in importance was the aure enjoyment of the literature for literature's sake. Secondly, in arefully measured small doses, what that literature could do as a odel to help improve one's own writing was utilized" (p. 91). The essons were based on an expansion of the University of Georgia's roject English materials and focused on skills such as compound words, sage and sequential order, and on concepts such as differences between antasy and realism and the reflection of time and place in setting. esults were calculated during the fourth year for 40 students emaining out of the original 70 in the experimental group. A fourth- rade class receiving no special treatment was used as a control group. he experimental group scored significantly higher than the control "cup on Writing Samples (proficiency was rated by Veal's Rating 43 Scale), Iowa Tests of Basic Skills—-Capitalization, Punctuation, and Total Language Tests. In a particularly well—planned study, Pinkham (1968) theorized that a systematic series of lessons, based on what she found to be the recognized approaches to the improvement of composition, would improve the writing of fifth—grade children. Her primary approach was to use excellent literature as a model for writing. She designed lessons which focused, through discussion, on the characteristics of good writing as exemplified in selected works of children's literature. Into these lessons she then incorporated the basic tenets of the three remaining approaches: (a) much practice in writing, (b) revision of written expression by the child after evaluation by the teacher or class group, and (c) expanding the pupil's actual or vicarious experi— ence background. Pinkham designed l4 bi-weekly lessons to use with the experimental group-~a 45-minute period for the literature and writing period and a 30-minute period later in the week, after the writing was corrected by the teacher, for discussion, revision, and rewriting of student material. Both periods followed by the same pattern and were highly structured by the teacher. Each lesson focused on a specific haracteristic of good writing; portions from literature selections ere read aloud and then analyzed through direct discussion to deter- ine the methods and techniques used by the authors to attain that haracteristic. After discussion to stimulate imagination and think- ng, students were encouraged to try these methods in their own writ- ng, incorporating their own experience. The students' writing was 44 discussed during the following lesson to demonstrate excellence or the need for improvement. The control group spent equal time listening to literature and writing compositions stimulated by that literature, but did not participate in the systematic series of lessons. Two class- rooms in each of four schools participated in the study; one room was used as part of the experimental group, and the other became part of the control group. Two of the schools were urban, two suburban. Using an analysis of covariance and t—ratio techniques, Pinkham controlled for differences in chronological age, intelligence, and writing ability between classes within her experimental and control groups. At the conclusion of her study, she compared her groups using gain scores submitted to analysis of variance and t—tests. As did Mills (1967), Pinkham found inconsistent results on her final measures. She found significant differences in favor of the total experimental group and the experimental subgroups, urban and suburban, on the Sequential Tests of Educational Progress (multiple choice) Writing Test in the areas of organization, conventions, critical thinking, effectiveness, and appropriateness. The experimental urban subgroup was the only group, however, to show significant gains in written expression as measured by the Quality of Thought and Style areas of the STEP Essay Test (p < .05). Pinkham concluded that the series of lessons "had a beneficial effect upon the written expression of urban children" (p. 122). Pinkham's study raises the important question of whether reading the piece of literature in its entirety, with its structure 45 intact, rather than focusing on selected sections within that piece of literature, would have resulted in different findings. A descriptive study by Keech (1984) provided an interesting perspective on the failure to find expected results in children's writing. Keech's study suggested that unexpected drops in scores on sequences of essay tests written by high school students may be false rather than true regression of performance. She supported her conclusions with an analysis of a longitudinal sample of high school students' writing in addition to three case studies. Her analysis suggested that as students attempt more in writing, they may lose control of the rhetorical features rewarded by readers or they may mix aspects of text construction appropriate to different kinds of texts they have not fully differentiated. As a result, their performance scores drop. While continued composing within a type of discourse may result in improved writing scores, switching to a new type may result in a drop in scores since the student must master composing skills different from those previously used. Duncan (1981) summarized a year-long study of the writing growth of eight middle-class seventh-grade volunteers. The students varied in academic ability and had little experience in creative writing. The study was based on an underlying assumption that "active comprehension of a distinctive model of literature provides awareness of the structure of narrative prose which transfers to the student's own writing" (p. 345), and attempted to determine whether knowledge of literary models could be transferred to personal writing. Guided 46 discussion was used to focus the students' attention on aspects of story structure and authorksstyle in order to develop a fuller under— standing of prose construction. Duncan used a program of sequential activities, which she called "Listen, Discuss, Write," involving target discussion of a literature model followed by writing. Weekly sessions involved 15- to 20-minute discussion periods followed by 15- to 20- minute writing sessions. As the program developed, oral discussion became brief and informal, with writing sustained for longer periods. All sessions were limited to 50 minutes. Although difficult to deter- mine from Duncan's description of the project, it appears that a film strip or film presentation was used to present an overview of the literature model. Selected passages of the literature were then read in order to focus on specific literary components and story relation- ships, and discussion guided by teacher questioning followed. Children were subsequently assigned writing tasks using the discussed tech— niques; these writing tasks eventually developed into story writing. Preliminary results of holistic scoring and feature analysis (a) appeared to validate the "Listen, Discuss, Write" procedure for fostering growth in composition skills; and (b) indicated that although all students appeared to profit from the targeted discussion of narra- tive elements, greater improvement was shown for less able writers. Duncan recommended that future replications include early attention to a sense of audience. Duncan's study is of interest because of its focus on listening 0 selected sections from literature followed by targeted discussion of 47 ie authoris development of structure through the use of literary amponents. The study also provided preliminary evidence that after lscussion of narrative structure, the writing of less able writers may low greater improvement than the writing of more able writers. Like inkham, Duncan did not read the selections in their entirety. Her ascription of the study indicated, however, that she focused on the :ructure of narrative to a greater extent than Pinkham by selecting we portions of literature to be read aloud expressly for this purpose. Stewig (1980) is a major proponent of the value of literature >r children as writers. He believes in a planned sequential writing/ iterature program in which teachers read to children for at least 20 lnutes a day in both the primary and intermediate grades. Apprecia- ion, discussion, and using literature as a basis for specific writing :periences are separate yet interrelated strands of his program, which ‘cludes four components: intensive literature input, writing assign- nts based on observations, writing assignments based on literature, d editing as a means of improving compositions. At the Eighth Annual Conference of the Children's Literature sociation, held in March 1981, Anita Moss was a member of a panel terested in bridging the gap between literary critic and child ader. In her presentation, she described an attempt to involve a aup of "regular" fourth-grade students in responding to literature tically as well as emotionally and cognitively. Her teaching plan , developed from a theoretical base which suggests that "immersion in basic structures of literature should enable children to read and 48 to make sense of texts more effectively and to write more effectively as well"(p. 201). Moss conducted 45-minute classes during which the children read and discussed a story or completed a writing assignment. She began by reading folktales to the children and introducing them to literary concepts such as type of journey, character types, or patterns of experience. Moss did not indicate that any formal measurements were conducted, nor did she compare the performance of subgroups of chil- dren. She believed, however, that the children learned and applied the concepts taught to the stories they read, to their popular culture, and to their own personal experience. In their writing, by the end of the term, she found that the children were using chronological order as an >rganizational structure for narrative, classification, cause and affect, comparison and contrast, and increased unique vocabulary. Nielsen (1980) conducted a study to investigate the effective- iess of using quality children's literature as a stimulus for effecting narrative writing of good quality by fourth—grade children. Her study vas based on the assumption that emphasis on the mechanics of composi- ,ion is not as fundamental to writing of good literary quality as is The stimulation of ideas. She divided students in nine fourth-grade lassrooms into four treatment groups. Two of the groups listened to he teacher read selections from children's literature; the other two roups chose books from a designated collection of quality books and sad them silently. Following the reading, one group from each reading reatment was given questions, to answer individually and silently, alating to literary elements of stories. The other two groups did not 49 receive questions. Children from all four groups were then given 20 minutes to write a story on any subject of their choice. The writing was done individually and silently. Although students perceived them- selves as writing longer and better stories at the end of the study, Nielsen found no difference by treatment in the literary quality of the children's writing when evaluated using the Glazer Narrative Composi- tion Scale. Since the children participated in reading and writing only ten times over a period of five weeks and wrote only twice a week for 20 minutes a time, with no possibility for revision, the length of the study and the frequency of treatment may have precluded the finding of significant results in this study. The restriction on the children's interaction with each other during the writing process may also have influenced the results. In an interesting article concerning the relationship between the child writer, the professional, and the teacher, Moran (1980) argued that the transfer of technique from reading to writing is possible but not automatic or inevitable; it will occur to the extent that the reader, in effect, becomes the writer. He saw the teacher's role, therefore, as one of presenting the author as a writer, of helping the student to identify him/herself as a writer like the author, and through this identification of writer with writer, of facilitating the transfer of technique. The attempt to facilitate the child's viewing of him/herself as an author is also a primary focus of the process writing approach. Conferencing and using an “author's 50 chair" are frequently used to help reach this goal (Graves & Hansen, 1983). The concept of communication between author and reader was further examined in a study by Roskelly (1984L Although Roskelly's study pertained to the integration of literature and writing in the teaching of freshman English, there is no evidence to suggest that her argument is applicable only to the college level. She argued that the act of interpretation remains the same in literary reading and in writing and suggested that students who become aware of the operations they perform as they read literature can learn to control those opera- tions and manipulate them in their writing. Perhaps it is Wright (1974), in her article "Wishes, Lies and Dreams: Pedagogical Prescriptions," who presented one of the most cogent arguments concerning the use of models with children. She ‘argued that through the process of writing, writers formulate experi— jence through language and develop constructs to order the world. She lmaintained that training children in the use of techniques develops éconformity through language. By using models, she argued, the teacher -does this ordering and the child is relegated to filling in the ‘Eblanks (p. 551). 1 l i um r ‘1 How best to interrelate children's literature and writing in lthe elementary classroom is a much-debated issue in educational lcircles. The professional literature reflects a continuum of studies Wand articles advocating methods from imitation to simple exposure. 1 l 51 While the use of modeling through imitation may seem inviting as an aid to expression, such imitation seems to have more effect on atomistic measures of writing than on overall quality (Martin, 1980; McCampbell, 1966). A preliminary study by Eckhoff (1983), however, indicated that children may unconsciously participate in the imitation of reading materials provided in the classroom. The extent and effects of this type of modeling are thus far unknown. There is some evidence, although conflicting, that helping children to focus on techniques used by authors of children's litera- ture in their creation of literature and then encouraging them to use those techniques in their own writing results in improvement for some children. Discussion seems to be an important component of these attempts. Inadequate or inappropriate measures of writing as well as attempting to write in unfamiliar discourse modes may be a contributing factor in the inability to find consistently significant results in these studies (Keech, 1984; Mills, 1967; Pinkham, 1968L Writers, especially those less able, seem to profit from dis— cussion of narrative elements designed to highlight the structure of narrative (Duncan, 1981; Moss, 1981). It appears, however, that this focusing needs to be part of a planned, structured literature and writing program in which good literature models are provided and class- room emphasis is on meaning, not superficial aspects of form (Mills, 1967, 1974; Olson, 1983; Pinkham, 1968L 52 Structure Conventio 5 nd C ildren's riti Applebee (1979) found that children enjoy hearing and telling stories because stories relate to but are separate from the real world. In explaining the appeal of stories, he said: Children are engaged in a search for meaning, a search for struc- tures and patterns that will suggest order and consistency in the world around them. The patterns of meaning they find are primarily socially determined--rules of language, thought, and behavior that are transmitted by a range of social structures, stories among many others. One source of pleasure in such encounters is the pleasure of mastery--knowing the rules and being able to manipulate them so that things come out right in the end. This is a particularly important factor in the fascination of certain highly stereotyped formula stories--cowboys and Indians, later comic books, and later still the detective story or James Bond. .. . Another satisfaction derives from working out the implications of the rules and conven- tions that govern our lives .. . and part of the usefulness of stories is that through them we can explore those limits without losing the game.(p. 645) He maintained, however, that any such experience with literature "depends on the mastery of the underlying conventions which govern the exchange between author and audience; without the conventions no exchange can take place" (p. 645). l i l Applebee (1978) determined through his research that the con- Wentions of stories, including complexity which is handled by the , l i l imposition of structure, are both developmental and learned. They emerge gradually and over time, proceeding through a process similar to the stages of concept development described by Vygotsky (1962L Apple- )ee found that for young children, a story is primarily a patterning of events. Early adolescents, however, recognize that these patterns are fonsciously ordered for a purpose and older students recognize that l , articular works imply a broader theme or message (p. 134% 53 Calder (1984) designed a study to examine the effect of story models of differing complexities on children's concepts of story (as measured by Applebee"s levels of story structure), their response to literature (as measured by the classification systems of Applebee and Purves), and the quality of their narrative writing (as measured by Tway's Literary Rating Scale). Calderis sample was composed of 20 pairs of suburban third graders matched according to teacher judgment of reading and writing abilities; the matched pairs were divided into two groups. For one group, Calder used, as models, stories written according to Applebee's (1978) Unfocused Chain Model, a structure slightly above the children's level of story production. He followed these stories with activities explicitly based on concrete incidents in those models. For the second group, he used stories written according to Applebee's criteria for a True Narrative. The activities for this ‘ group were based on abstract foci or themes derived from the story ‘ heard. All stories were written by the researcher. All children i participated in parallel activities, including drama, visual arts and ‘ music, and in training pertaining to their respective story grammar model. The children interacting with stories above but more closely ; matched to their own level of production showed significant gains over 1"the matched group on all three measures. Calder recommended further tresearch using story models closer to the child's stage of development. Brown (1977), in his article "Development of Story in Chil— ldren's Reading and Writing," reinforced the importance of children's tparticipation in the spectator role (Britton, 1970) which permits the 54 child to safely examine and modify his/her own relationship with the world while remaining separate from it. Through listening/reading and telling/writing stories, the child practices using his/her understand- ing of the world to predict events and actions, an ability she/he needs as a participant in real—life situations. The ability to operate with stories and to understand their rules or conventions is both develop— mental and learned. It is related to age, experience with stories, and facility with language. Although the ability to create a story in which the events and climax are tied to the theme is developmentally possible by about the age of nine, a child's sense of story may remain immature much beyond that. Brown noted that the ability to use story language is closely tied to reading achievement and that reading diffi- culties are due more to lack of knowledge concerning the variety and sophistication in syntactic patterns of written language than to diffi— culties in sounding out. He believed that the ability to predict syntax and gain meaning may depend on the agtant of a child's sense of story. Brown recommended telling or reading stories aloud throughout the elementary school and cited Cohen (1968) and Porter (1969) in his recommendation for discussion and follow-up activities to develop mem- ory of content, meaning, relationships, vocabulary, and story language. He also recommended the dictating or writing of stories by children to focus their attention on form and sense of story, including the logic and syntax generally associated with them. In the second edition of her book The thld as Critic, Sloan (1984) reaffirmed that literature cannot be taught. It can only be 55 experienced and, as experience, it cannot be transmitted from one person to another. She believed, however, that experience alone does not create an awareness of the coherence of literature as an entity representing humanity's quest for identity. Literature must be treated as an art; children need to learn to respond to the total structure of a story. They need to see that a work is a unified entity, shaped and structured by the elements within it. Knowledge of story form and structure becomes personal when the child attempts to write a story. This is particularly true if the story the child is writing incorpo- rates details from his/her own personal world into the overarching structure of an archtypical story. Both traditional and popular lit- erature, with their easily observable literary conventions, exemplify this archtype. Conventions, according to Smith (1981), are the basis of society's shared understanding and communication. Conventions are unpredictable at first, thus reflecting the arbitrariness of language, but once learned, "they are used as the basis for prediction, and prediction is the basis for comprehension" (p. 104). In an article entitled "Learning to Write by Listening to Literature," Moss (1977) described a writing-literature unit designed to help children discover specific elements of narrative structure through exposure to many stories with a common and personally meaning- ful theme. Drawing on the theories of psycholinguists like Frank Smith, Moss hypothesized that, if immersed in literature, the children would develop a 56 "theory of narrative" from which to make sense out of the literary world. This theory of literature would emerge as the child actively engaged in the process of comprehending and producing literature and it would serve as a base for interpreting each new literature experience and for generating new prose and poetry. (p. 541) Working with six- and seven—year—old children, Moss read aloud many pieces of children's literature which focused in a variety of ways on the theme of animals. She encouraged the children's spontaneous response to each piece and then continued the discussion of the story focusing on some element of narrative. Following the reading and the discussions, pictures were taken of each child with his/her favorite stuffed animal. The children then wrote descriptive paragraphs to accompany the photographs, and each decided on a specific point of view from which to tell his/her story about an imaginary animal. The final stories were illustrated, combined with the photo—stories, and compiled into a class book. Moss believed that the essential ingredients of this or any successful writing-literature unit are using ideas that are meaningful to the children and having appropriate tools, i.en narra- tive elements, for expressing those ideas on paper. Yatvin (1981) described a program that used models of the lwriting task to be completed as supportive devices for the writing of ‘fourth- and fifth-grade students. The group of teachers involved in the project found this method superior to explanations by the teachers since the model "contains everything writers need to know and it remains with them while they write" (p. 54). The teachers helped the children deduce the significant characteristics from the text, supplied multiple 57 texts, and permitted the writers to use as much or as little of the model as they needed. A study by Atwell (1981) was based on a sociopsycholinguistic perception of language and involved the analysis of narrative writing by college undergraduates of above-average and below-average writing ability. Atwell hypothesized that the coherence of the text and the character of the process would change, depending on whether or not the writer could read his/her emergent text. She found that texts of the above-average writers involved greater local and overall coherence than the texts of the Basic Skills writers and that poorer writers were severely affected when they could not read their emerging text. Atwell explained that the writers who had clear mental schemes to guide their writing were less affected by the inability to reread their text. Atwell's findings indicated that poorer writers, even at the college level, lacked an overarching sense of structure and thus found it difficult to develop coherence in their writing. Summarizing research in response to teachers' questions, Holdzkom, Reed, Porter, and Rubin Otd.) reported that children's writing becomes more cohesive with age and experience. Their texts y become more cohesive "when they are provided good models, well-written stories for example, which they can experience in a variety of ways. Also important is a classroom emphasis on meaning rather than on 7 learning form" (p. 37). Squire (1983) felt that, in addition to ‘ exposure, writing in a particular mode requires instruction and ‘ directed practice in writing that form. He stressed, however, the 58 importance of helping students to grasp the piece of discourse as a unified whole so that they have a context within which to develop the necessary components. In response to Bettleheinfls encouragement in The Uses of Enchantment that fairy tales be reexamined for new pedagogical approaches by those in the social sciences and humanities, Harp (1978) concluded "if a student is to write about his own experience and that of his society, he must first have acquired a framework within which he can see and focus these things" (p. 158). He suggested that imagina- tive literature is the best way to begin to teach writing and that the first literature taught should be the most basic and elemental: the fable, fairy tale, parable, proverb, and myth. Building on Bettle- heim's contention that nearly every incident and character in a fairy tale is drawn from common adolescent experience, Harp suggested that students can retell many of these elemental forms using their own personal experience and, in so doing, relate that experience to social groups and structures not immediately their own. Mikkelsen (1983) analyzed 52 stories written by 15 children, from seven to ten years of age, immediately after they listened to folk tales. The children met regularly over a three-month period, listened to folk tales told or read, discussed the stories, and then told or wrote their own stories. Mikkelsen divided the children's writing into five categories: (a) retellings--the original was retold with only small additions and/or deletions; (b) borrowings-—details of the plot were changed, episodes from parallel plots were included, or a 59 new version was created; (c) re-creations—-a single literary element, usually a character, was placed into the child's own personal experi- ence; (d) blendings—-the child borrowed and integrated material from multiple literary sources as well as popular culture; and (e) trans- formations-—using themes as scaffolding, the child integrated existing elements with new features in such a way that an entirely new product was created. She concluded that maturation plays a part in children's reshaping of literature. Children move gradually from retellings, where there is heavy reliance on an author's ideas, through the inter— mediate stages of borrowing, re-creation and blending to the final stage of transformation, where there is greater independence in the creative process. She also found that children do use literary motifs in their own stories and find ways of intricately fusing those elements with their own personal experience. Mikkelsen recommended further investigation into the relationships between literature genre and chil- dren's writing. The idea that removing oneself from the role of participant helps one deal with experience is the basis of Britton's (1970) concept of spectator and participant roles. A study by Edmund (1960) indicated that preadolescents engaged in working out their relationships within society may make extensive use of the spectator role. Edmund found that middle- and upper—grade children do not automatically write, in school, about their concerns. He randomly selected 64 fifth and sixth graders from a large suburban Pennsylvania school system and required them to list five or more fears, personal problems, or things they 60 worried about. The following week, they were given an assignment to write a composition on a topic of their choice. Five of the 64 wrote about their problems, 59 did not; 5 wrote about their worries, 59 did not; 3 wrote about their fears, 61 did not. Stewig (1980) saw an implication of this study being the need of older elementary children i for vicarious experiences through literature. Holbrook (1967), a British educator concerned with children's response to literature, found considerable success in freeing even reluctant writers to write by using traditional and popular story structures. According to Holbrook, these conventionalized structures provide children with socially acceptable, depersonalized forms through which to seek self-identification and express individual emotion. He found that permitting children to express themselves through a conven- tional and thus depersonalized form frees the child from "having to consider explicitly something he could only grasp implicitly"(p. 115% um r The structuring of personal and vicarious experience into narrative often involves a search for meaning and identification (Applebee, 1979; Harp, 1978; Holbrook, 1967L To communicate with others, however, a writer must master the conventions of language and story which are the basis of prediction and thus comprehension and which reflect and transmit socially determined rules of thought, language, and behavior (Applebee, 1978; Smith, 1981). Knowing the rules and being able to manipulate them means being able to control them, both in stories and in onefls own life. Highly stereotyped, 61 formula stories provide this opportunity (Applebee, 1979; Holbrook, 1967). Those students who have a clear mental scheme to guide their writing create more coherent text (Atwell, 1981). The ability to write a cohesive story with a thematic center is developmentally possible for children by about the age of nine, although it may not be developed until much later (Brown, 1977). If children's writing is to become more cohesive with age and experience, however, it is important that they be provided with well—written stories as examples in classrooms which emphasize meaning rather than superficial aspects of form (Holdzkom et a1., 1982; Moss, 1977; Squire, 1983; Yatvin, 1981). In contrast to much current practice, there is evidence that pre- and early adolescent children may need depersonalized, socially acceptable story structures through which they can express their individual emo- tion while at the same time relating that experience to the large group ‘and social structure (Edmund, 1960; Harp, 1978; Holbrook, 1967L DetectiveZMystery Storjes-—Structure, ConventionI ‘ and Children's Writing M The mystery story is a highly conventional form of literature, which originated in the 18005. In one of the classic writings pertain- wing to this genre, Wells (1913) suggested a three—way classification of l mestery stories: ghost stories, riddle stories, and detective stories. T4She defined the detective story as one in which the solution is Jdetected, not guessed; the problem is invented and solved by the author 1 lin such a way that it is possible for an astute reader to reason out 62 the solution. According to Wells, since all detective stories have the same plot, the only difference between a detective novel and a detec- tive short story is one of length. In contrast to later writers, early authorities including Wells and Freeman (1941) maintained that the primary interest of the detective story is intellectual, involving a mental duel with the author as well as ratiocination. Thus, they argued that the plot must be free from fallacies, with the conclusion being the only one possible given the clues presented. Particularly in its early history, authorities attempted to elucidate aspects of the detective story which were immutable. By the 19405, however, things began to change. Steeves (1941) was arguing that critics could not tell writers the rules of the game, and Chandler (1944) maintained that detective fiction depended not on the actual logic of the story (since even well-known mystery writers, including Sir Arthur Conan Doyle, made significant blunders) but on whether or not the reader accepted the assumptions of the author. By 1950, Aydelotte was able to write that the world of the detective story was a simplified one where troubles were caused by external circumstances, cause and effect were simplified to the extent that one detail could provide the solution, and solution of the crime returned everything to normal. Aydelotte was one of the first to recognize or acknowledge that the reader plays a spectator role, not attempting to solve the mystery but deriving a sense of satisfaction from watching the detective do so. He explained that detective stories were not a test of the reader's intelligence, were often not free of 63 fallacies of reason, and were not factually accurate. Instead, they create a delusion of intelligence by convincing the reader that she/he has displayed intellectual proficiency by following the steps in the analysis to arrive at the correct solution. In actuality, the reader is totally dependent on the detective. Russell Nye (1970) of Michigan State University maintained that the conventions of the classical detective story were fairly well defined by 1900. Edgar Allan Poe and Sir Arthur Conan Doyle had devel- oped the amazingly logical, amateur detective figure, the correspond- ingly brilliant criminal adversary, the obscure support figure, and the ineffective, often blundering police. They had also set the basic narrative pattern which subsequent writers would elaborate and vary, but would not significantly change. The hard-boiled detective story appeared after World War I. A variation of the classical detective story, it diminished the emphasis on the answer to the puzzle and increased the emphasis on the character of the detective and the dangerous and violent world in which she/he operated. According to Nye, the affinity for unraveling riddles and solving puzzles accounts for the genre's popularity. Unlike other current authorities (e.gu Cawelti, 1976; Reilly, 1979), Nye maintained that the mystery "tells a story, nothing more; it doesn't involve symbols, multileveled interpre- tations, underlying myth-ritual patterns, nor elaborate psychologizing" (p. 268). Today, the detective mystery with its variants is the litera- ture most read by the reading public. It accounts for one out of every 64 four books published or reprinted in the United States each year, for at least half of all books in rental libraries, and for a third of all library borrowings. In spite of the fact that most professional writ- ers claim they read widely and indiscriminately as children, the read- ing of series books has never been held in particularly high esteem by educators. The Stratemeyer Syndicate alone, however, sells six million copies of series stories a year, many of which belong to the mystery genre (Donelson, 1978L The reading of detective fiction has never held much status in the literary world, and, according to Reilly (1979, “until recently much of the best writing about detective fiction [was] intended to legitimate the form"(p. 64% In contrast to Nyek;exp1anation, Reilly believed that the primary purpose for reading detective fiction is not to figure out the puzzle, but for enjoyment and escape. He explained that all detective stories are basically the same, with the conventions used to organize the narrative repeated in each one and the anticipa— tion of the reader fulfilled in the closure of the story. Reilly identified three particular conventions employed in the detective story: (a) the illusion of reality; (b) the puzzle, which involves a disruption followed by a restoration of the existing order; and (c) the demand for closure. The illusion of reality is created by using details and mundane facts to simulate the real world but remain detached from it. Thus, the story is not judged against reality but takes place in the created world designed to assure enough distance from the real world so that 65 the disruption of order and its resolution become pleasurable, not disturbing or real. The disruption and restoration of order, i.e., the crime itself and its solution, is also a convention of the detective story, closely connected with the convention of demand for closure. The distance effected by the creation of a secondary world identical to the real world separates the crime from the reader's immediate experience and permits enjoyment of the plot for plot's sake. The demand for closure ensures that the detective will make things turn out all right. According to Reilly, The criminal problem of the story will not intrude upon our lives by making reference to occurrences we know directly, but it pre- serves the outline of feelings we know we have. The repetition of conventions from one story to another promises certain resolution to the criminal problem, and, doing so, promises as well a resolu— tion to the feeling of anxiety so that we can experience the entire narrative as art. The detective story, thus, touches deep-seated feelings, just as all art does, but it does not require us to translate those feelings into problems to be confronted in our consciousness. (p. 65) Cawelti, whose book Adventure, Mystery, and Romance Formula Stories as Art and Popglar Culture (1976) is recognized by many as the most important contribution to date in the understanding of popular American culture, identified the detective/mystery as formula litera- ture. He defined a formula as the embodiment of themes specific to a particular culture within universal or archtypical story forms. Formu- las use "standardized" conventions that are familiar to both readers and writers. Past experience with the formula provides the reader with an anticipatory set for each new individual example and thus provides 66 satisfaction and basic emotional security. Cawelti quoted Warshow in explaining that the reader approaches any individual formula work "with very definite expectation, and originality is to be welcomed only in the degree that it intensifies the expected experience without fundamentally altering it"(p. 9). Warshow went on to explain that the formula appeals only secondarily to onefls experience of reality. "More immediately it appeals to previous experience of the type itself; it creates its own field of reference" (p. 10). Thus, the quality of any formula work depends on the writer's ability to imbue characters, plot, and setting with originality and vitality while remaining within the prescribed boundaries of the stereotypes required by the formula itself. Cawelti explained that a major characteristic of formula lit- erature is its goal of providing escape or entertainment. Describing what he saw as a pejorative distinction between fine art and formula literature, he explained that mimetic literature, traditionally seen as serious literature or fine art, "confronts us with the world as we know it, while the formulaic element reflects the construction of an ideal world without the disorder, the ambiguity, the uncertainty, and the limitations of the world of our experience" (p. 13). The art of formu- laic literature relies on the author's ability to develop excitement and danger based on the reader's knowledge of reality without confront— ing the reader with the insecurity and anxiety which accompany that excitement and danger in the real world. This is accomplished pri- marily through three literary devices-~suspense, identification, and 67 the creation of a slightly removed, imaginary world. Suspense is created by developing uncertainty regarding the fate of a character the reader cares about. It is intensified by tension developed by the reader's acceptance of the formula convention that everything must work out, and his/her suspicion that the writer might deceive him/her. While identification in mimetic fiction is designed to make the reader confront his/her participation in the real world, identification in formula fiction permits the reader to see him/herself as the idealized self, the super protagonist. This is accomplished by focusing on the story action and by using stereotyped characters who reflect a conven- tional view of life and society. The writer creates a world with which the reader is intimately familiar but removes it to the extent that the reader is not tempted to measure it by the usual standards of plausi- bility and probability. To explain the escapist dimension of formulaic art, Cawelti quoted Piaget's description of play, thus unintentionally indicating its inherent appropriateness for children: Conflicts are foreign to play, or, if they do occur it is so that the ego may be freed from them by compensation or liquidation whereas serious activity has to grapple with conflicts which are inescapable. The conflict between obedience and individual liberty is, for example, the affliction of childhood and in real life the only solutions to this conflict are submission, revolt, or coopera- tion which involves some measure of compromise. In play, however, the conflicts are transposed in such a way that the ego is revenged, either by suppression of the problem or by giving it an effective solution. .. . It is because the ego dominates the whole universe in play that it is freed from conflict. (p. 20) In describing the pattern of the classical detective story, Cawelti identified four major components: the situation, the pattern 68 of action, the characters and their relationship to each other, and the setting. 1. The situation involves some crime, important in itself but not personally important to the detective, which needs to be solved; 2. The action focuses on the detective's investigation and solution of the crime and involves six steps which, while present, may vary in sequence and may be collapsed into each other; a. the introduction of the detective which establishes his/her competence and detachment from the world of the crime; b. the crime and the clues—~although the crime appears insolvable, the clues indicate that it can be solved; c. the investigation conducted by the detective who mentally or physically examines or follows up on the clues; d. the announcement of the solution, often dramatic to increase interest in the detective's explanation; e. the explanation, usually derived, not from the clues, but by seeing the problem from a unique perspective; f. the denouement, including the apprehension of the criminal and his/her confession. 3. The characters consist of a victim--necessary but of least importance, the criminal--often brilliant, but always defined as bad and not permitted to detract from the focus on the detective, the detective—-detached, brilliant, eccentric, and possessing an intuitive insight and astounding ability for inductive reasoning, and finally, others involved with the crime—-friends and acquaintances of the detective who need his help, the bungling police, false suspects who need an advocate to clear them; 4. The setting is comprised of two delimited spaces—~the isolated headquarters of the detective, symbolizing order and ration- ality, and the separated, often isolated location in which the crime occurs and in which the clues can be silhouetted. (pp. 80—98) According to Cawelti, the hard-boiled detective story resembles the classical detective story but is different in two significant 69 respects: the detective%;active involvement in the pursuit and accomplishment of justice overshadows the process of intellectual detection, and the detective becomes judge as well as investigator after being tempted and intimidated. Billman (1984) in her article "The Child Reader as Sleuth" discussed the mystery genre in children's literature. While she drew heavily from Cawelti's book, her application to children's literature is of particular interest to this research. Billman maintained that the principle identified by Cawelti as fundamental to the mystery——"the investigation and discovery of hidden secrets—~15 psychologically appropriate for early adolescents and underlies time fantasies, contemporary realism, and historical fiction" (p. 33). She contended that series mysteries pick up where fairy tales leave off for the pre- and early adolescent whose interest in fantastic stories is becoming quiescent. These series stories provide psychological comfort, including the escape and consolation Tolkien and Bettelheim believed children find in fairy tales. According to Billman, series books in which children encounter the mystery/detective genre ask little of readers in terms of literary sophistication or mental or emotional involvement. Instead, children are provided with an opportunity to repeatedly work out the basic narrative pattern using the particulars of an individual storyline. Billman quoted Todorov in explaining that formulaic stories encourage '"metareading,'a process by which'we note the methods of. .. narrative instead of falling under its spell"'(p. 33). She contended that broa for oug str "cc um Th1 70 that the child reader reads to discover "how the genre--and more broadly, literary narrative as a whole-—operates" (p. 30), as well as for the answer to the puzzle. She explained that series books help solidify for the inexperienced reader the elements and pat- terns of fictional narrative. . . . Series mysteries that are as demonstratively formulaic as the fairy tale will be useful in providing for the preteen further clearcut data about literary construction and narrative categories. .. . [They] pick up where fairy tales and fairy-tale—like novels typically leave off;they are equally conventional, and hence predictable, in style, content, and pattern of the narrative. (p. 35) She concluded that while many series mysteries for children ought more accurately to be called adventure stories since they lack strict adherence to the principles of ratiocination, they do provide "comfortable fiction" containing cognitive games that lead children to understand their own responsibility for making literature meaningful. They thus provide the foundation for the reading of more sophisticated literature (p. 35). Vardell (1983) described a study designed to compare "the developing capacity to compose and the reader‘s response to story as it develops across age levels" (p. 47% In this study, sixth, ninth, and twelfth graders were encouraged to read and reread a professionally written mystery/detective story and to prewrite, write, and revise their own mystery story. They also responded to the stories they read and wrote by means of a questionnaire. Vardell made no mention of any additional teaching or any intentional focus on the conventions or structure of the genre. In analyzing the results, she found, as Apple- bee did, that students responded to literature in ways consistent with thei Van tio: men' tha per cor tyl fr 5’: 71 their level of intellectual development. To analyze the writing, Vardell used content analysis to focus on portrayal of characteriza- tion, use of foreshadowing, creation of ending, total story develop- ment, and overall quality. In relation to characterization, she found that sixth graders developed their characters from people they knew personally'or vicariously, ninth graders designed prototype characters consistent with the genre, and twelfth graders expanded those proto- types with individual characteristics. ‘Vardell saw this as a movement from the develOpment of flat characters to the development.of'more stereotypical ones, and finally to fuller characterization. She felt that it illustrated a growing ability to integrate personal experiences with story knowledge. In this study, only twelfth graders were able to incorporate foreshadowing and to develop natural rather than unrealis- tic or "epilogue-type" story endings. In terms of story development and overall quality, Vardell found that although ninth and twelfth graders understood that characters had an effect on the events in a story, they had difficulty realizing this relationship in their own stories. She also found that only the writing of the older students reflected the ability to integrate the real world with story conven- tions, thus producing a story that represented reality but was not limited to it. Vardell found that while students borrowed or recreated some aspects of the professionaly written stories, their stories were generally unique and varied. She recommended (a) using literature as a model or stimulus, thus enabling students to manipulate the conventions studl gene accc 198' inS' the on 72 of literature, and (b) using open-ended questions designed to help students integrate their own experience with literary conventions. The disdain of many critics and writers for genre writing in general and for the detective story in particular (i&%, for writing according to identified conventions or a formula) was discussed in a 198l article by Valin. Valin, an author of detective novels and an instructor of creative writing at Washington University, explained that the argument, which he believed to be more apparent than real, centers on the supposed antithesis between rules and imagination. Noting E. M. Forster's contention that authors are stuck with the convention of a storyline, Valin maintained that to say a detective story writer is "stuck with a mystery is similar to saying that a sonnet writer is stuck with a particular rhyme scheme. He argued that conventions provide both limits and structure and maintained that what a writer does within the conventions determines whether she/he is writing by formula or imagination. Drawing on l2 years of writing "serious (not detective) fiction," Valin contended that a writer must decide what conventions mean to hisgher work, 1.6" how does the form provide a vehicle for the ideas which the author brings to the genre? He main- tained that "pure" fiction touches the real world and yet seems apart from it; he argued that creative detective writers likewise "don't speak of the world as it is, but about the world as we would have it be——as we would like to imagine it" (p. 2l). In a book discussing how to teach writing within a genre frame— work, Hubert (1976) related her own experience with writing, including her 11 t6 have whel own beg mec‘ son we re pc 73 her feelings of inadequacy when she compared her writing with classic literature. She felt children must learn hgw to tell the stories they have inside of them. Hubert explained that a writing assignment works when it captures the writer's imagination and incorporates the writer's own personal stories or fantasies. She argued that writing should begin where the children are in their reading, television viewing, and media since a child's voice is strongest when she/he is writing about something she/he knows and likes. Hubert believed that students are well aware of story forms. Although their writing often does not reflect this awareness, their ability to anticipate and predict within popular genres gives evidence of their internalization of the elements of those genres. Hubert maintained, however, that the forms with which students are so familiar will not work for them in their own writing until they learn the conventions of genre. Discussing the place of the mystery/detective story in the classroom, Hunter (1982) suggested that students' natural interest in these stories can be used to improve their problem-solving and reason— ing abilities. Hunter believed that incorporating puzzle mysteries into the curriculum can help students master particular problem-solving and reasoning processes such as logical analysis, attention to details, and the formation of accurate and appropriate assumptions which are as applicable to everyday decision making as they are to math. Wilde and Newkirk (l981) found the writing of detective stories an effective means of helping children develop from writing sequence stories to writing those involving causality. They felt that the writ toge by E iic‘ ire per the au4 th 74 writing of detective stories provides a transition from fiction held together merely by chronology to the autobiographical writing promoted by Graves. Wilde and Newkirk suggested that children are drawn to fiction because of the power and mastery they can develop in its freedom from real-world constraints. They felt that children should be permitted to develop this power and mastery in their writing through the conventions of fiction writing. To promote this mastery, the authors developed a detective story unit, which was taught during the last eight weeks of third grade. The unit was designed on the premise that students possess an intuitive awareness, a grammar, of the rules that govern detective story writing. This awareness comes from listening to and reading stories as well as from watching tele- vision. But in order to use these rules in their writing, students must gain a conscious critical awareness of them. (p. 287) Wilde and Newkirk maintained that the issue is not whether the teacher should introduce writing constraints, but instead, what kinds of con- straints these should be. They believed that if the constraints are reasonable and if they clarify the way appealing fiction works, they can be liberating rather than unpleasant or restrictive. Referring to Howard Gardner's work in the field of art education, these authors suggested that, at least in this culture at this time, there may be a critical period in writing as in art during which a child needs to obtain a sense of competence in the craft if she/he is to see him/ herself as a writer. If, by the time of puberty, the disparity between the child's writing and the child's perception of competence is too great, the authors proposed that the race may be lost. Newki rk and Spe Pa! st de 75 The fol lowing is a summary of the sequential steps Wilde and Newkirk followed in their unit: 1. The teacher read mystery stories to the class. Each dealt with a different crime and different kinds of detectives. The teacher pointed out aspects of the mystery genre and of auth- cn"s style which he felt the students would need to deal with in their own stories. During the first two weeks the class discussed the stories; additional reading of mysteries from a class library was encouraged throughout the unit. During the initial period of reading, the teacher emphasized three con- straints of the detective form-~(a) the criminal is as smart as the detective, (b) there may be false clues, and (c) the solu- tion must depend logically on the clues, not on coincidence; 2. The class generated lists of descriptions of possible character traits fol lowing which the students used those descriptive words to write a character description of a classmate, or a photograph. These were read aloud for peers to comment on particularly effective language choice. Then, students generated lists for their own characters, wrote, and again shared their descriptions; 3. Settings, clues and descriptions of the crime were similarly developed; 4. The teacher put a possible plot structure on the board. Chil- dren first met with the teacher to outline a tentative plan and were then paired with a peer editor; the entire class met once a week. Emphasis was on the constraints mentioned earlier, and on the plausibility of sequence; 5. Finally, after revising and editing, the stories were compiled into books. (pp. 287-89) Palmer and Wiermaa (1982) described a project in which fifth- and sixth-grade students, ranging from gifted children to those in a special learning resource center, wrote a novel of 5 to 25 typewritten pages. Although not specifically'related to the writing of detective stories, this was a sequential writing program designed to help stu- dents cope with problems inherent in an extended piece of writing and thus seemed particularly applicablerto this study. According to the authc ment coul The recc hi 5, the tea add WI" fo 76 authors, the children, despite a wide range of skills, shared a commit- ment to write a long story called a novel, had the conviction that they could be successful in the venture, and became involved in the process. The project lasted from February through May and was structured with recorded checkpoints along the way so that each child could work at his/her own pace and also quickly compare his/her rate of progress with the progress of others. These same checkpoints indicated to the teacher which children were ready for conferencing or instruction. In addition to individual conferencing, periodic class discussions pro- vided the opportunity forichildren to:share writing samples as well as writing problems. The writing tasks were assigned according to the fol lowing outline: l. The children and teacher discuss novels, including how authors develop storylines, settings, and characters. After several days, the children are asked to give a definition for novels, identify different types of novels, and to list their common characteristics; 2. ‘The children write a preliminary storyline, a description of each main character, and a description of the setting-~each in 25 words or less; 3. The children work in small groups, write, and draw a picture of each main character to engrain the character in the mind of the student; 4. The students expand their storylines to develop a detailed plot outline; 5. The students illustrate three important parts of the story and write a one-page description of each scene; 6. Children write a rough draft, revise, edit, submit the final copy to an editing committee, and finally send the final copies to a high school typing class after which the manuscripts are bound for sharing. ia H dr Ik IIUIMI, iii 77 §ummg PM Mystery/detective stories can be seen as an extension of the fairy tale for early adolescents, providing psychological, social, and literary frameworks (Billman, T984; Cawalti, l976). Although chil- chtwfls series mysteries frequently lack strict adherence to the princi- ples of ratiocination, they are as formulaic as fairy tales, using standardized conventions to embody themes specific to a culture within archtypical story forns. While the major goal of formula literature for adults is escape or entertainment, such literature provides chil- dren with the ability and opportunity to control the universe as they do through play activities (Cawelti, l976). In detective stories, conventions are employed to create a world, similar to the real one and yet apart from it, in which disruption is always resolved and events are not judged by the laws of plausibility and probability operating in the actual world (Cawelti, l976; Reilly, 1979L. Although children are well aware of the detective story form from their reading, television viewing, and media participation, the transfer to writing is not auto- matic. To tell the stories they have inside of them, children must make their knowledge of form functional by learning the conventions of the genre. 'This can be accomplished by using their natural interests together with a structured literature-writing program (Hubert, l976; Valin, l98l; Wilde & Newkirk, l98l). the p the s comm clas tic‘ was chi att gra mai de: Se f1 CHAPTER III DESIGN AND IMPLEMENTATION OF THE STUDY, SELECTION OF THE LITERATURE, PROCEDURES, AND METHODOLOGY Chapter III describes the pilot study, the design of the study, the population sample, selection and analysis of the literature read to the subjects, and the procedures and methodology. Ihe Pilot Study A five-week pilot study was conducted immediately before the commencement of the study. The pilot was conducted in a fifth-grade classroom of 30 pupils in a community separate from the one that par- ticipated in the actual study itself. The purpose of the pilot study was (a) to provide an opportunity for the researcher to experience childrenhs responses to the questions she had designed to focus their attention on the classical detective genre; (b) to determine how fifth- grade children would respond to the structured writing program and the materials designed for their use during that program; and (c) to deter~ mine how those materials might be improved. This early study was designed to approximate the actual study itself as closely as possible. Several adjustments were made, however. 'The pilot extended for five weeks rather than for the nine-week period of the actual study; the researcher met with the children for a shorter time each 78 day-- prete be c< read read enti chi' the fir PO‘ As pr 79 day-—approximately 30 minutes; the children wrote only one story (no pretest); and because of the time restriction and since no data were to be collected, one short story from the classical detective genre was read to the children at the beginning of the five weeks rather than reading daily. The children in the pilot study participated in the entire writing program, although the classroom teacher helped the children finish revising and editing their stories. The pilot study indicated that the children responded well to the structured writing program which asked them to focus on literature first from the point of view of a listener/reader, then from the view— point of the professional author, and finally as authors themselves. As a result of this original study, no changes were made in this procedure. A number of changes were made, however, in the material which the children received during the writing program. Individual worksheets used during the pilot study were combined into a booklet which was given to each child at the beginning of the study. This change from individual worksheets to a self-contained booklet resulted in a number of improvements. First, it eliminated the necessity of keeping track of many sheets for each child, and it helped ensure that the children's work would not get lost. Second, binding the individual sheets into a unified booklet presented the writing task as a whole and encouraged the children to see the cohesiveness of the project as well as to examine the details. Third, whereas the children in the pilot study called the individual sheets "more worksheets" to be done, the children in the actual study identified the booklets as "my story." to t Firs num‘ res ac th of 80 i The pilot study identified two additional concerns that needed {to be addressed in selection of the literature for the main study. iFirst, it became apparent that fifth graders were familiar with a great number of mystery/detective stories and authors. Unknown to the (researcher, the particular story read aloud during the pilot study v(lhe Cgckling ghost by Sid Fleishman) had previously been shown on ltelevision and seen by a number of children in the class. This high— \ lighted the necessity of ensuring that the literature read during the actual study not be familiar to any of the subjects participating in the research. Second, selection of literature below the maturity level 1of the listeners appeared to be both inappropriate and self—defeating for the study. At the beginning of the project, the children in the pilot study were troubled that the literature read was "too easy" for fifth graders. Regardless of the fact that the genre traits were easily identifiable, and that the story was to be used only as an example of writing in a particular genre, the stigma of "too easy" seemed to affect their motivation to participate. The lack of sophis— tication in the plot also resulted in a second problem: a number of students imitated it quite closely in the writing of their own stories. This experience demonstrated that the literature read aloud needed to be at or above the expected maturity level of the students in the study to stimulate interest and motivation as well as to ensure that the plot could not be reproduced by the students in their own writing. ’- des COD SCC rai 81 e Desi n d e e t tion of he ud The basic design of the study was a 3 x 2 randomized block design with a pretest and posttest. (See Figure 3.1.) The pretest consisted of an initial story written by each child. Five different scores (Appendix J) were obtained from this story: (a) a holistic rating score representing the literary quality of the writing, (b) a holistic rating score representing the development of the genre traits in the story, (c) the number of T—units, (d) the number of words per ‘ T-unit, and (e) a vocabulary score representing the maturity of the vocabulary used. After completing the pretest, the children were randomly assigned to one of two treatments. The treatments consisted of reading aloud to the children each day. Group 1 heard novels from a children's literature detective series (hereafter called the Literary Series) determined by qualified raters (a) to be of high literary quality and (b) to contain the traits of the classical detective genre, each trait being well developed. Group 2 heard novels from a chil— dren's literature detective series (hereafter called the Popular Series) determined by the same raters (a) to be lower in literary quality than the series read to Group l and (b) to contain the traits of the classical detective genre, but these traits were less well developed than they were in the Literary Series. Both groups partici- pated in discussion of the literature according to the same predeter- mined format; both groups followed the same procedure in developing their second stories. The second story written by each child provided the anal Re 82 the five posttest scores (Appendix.J). The results of the study were analyzed using a multivariate analysis of covariance (MANCOVA). Reading Ability Treatment Literary Series Popular Series Total Treatment 1 Treatment 2 Group l--High reading scores 8* 8* l6 g Group 2--Middle reading scores 8 8 l6 Group 3--Low reading scores 8 8 l6 24 24 48 *These two groups originally consisted of nine subjects each. According to University regulations, archild is allowed to withdraw from any research study at will. During the first week of this study, a child in Treatment l, Group l, obtained a letter from his parents stating that he had their permission to withdraw from the study if he so desired. Thereafter, whenever he did not wish to participate in the activities of the study, he would indicate, both to the classroom teacher, the other children, and the researcher his intention to with- draw. The classroom teacher and the researcher concurred that the long-range needs of this child would not be met if this situation continued. It was decided that he should discontinue participation in the research and remain with the classroom teacher for the duration of the study. Since an equal number of subjects was needed in each cell for analysis, one subject was selected at random and dropped from Treatment 2, Group l. Figure 3.l.--Study design. not sid stu use the 83 According to Cornfield and Tukey (l9560,'this sample, although not randomly selected from a larger population, can plausibly be con— sidered a random sample from such a population, and results of the study can be generalized to subjects "like these" if the population used is clearly described, the correct model is used for analysis, and the purpose of the analysis is clear. T 9 Po u io S le The subjects participating in this study reside in a rapidly growing suburban community adjacent to a metropolitan center in mid- Michigan. The population includes employees of the State of Michigan, white-col lar employees of General Motors, and professors and students of a major Michigan university. Many of these are professional people with an income over $50,000. Four elementary schools, one middle school, and one high school are attempting to meet the needs of a community in which the school enrollnemt has sky-rocketed. With 264 new students, ennollment during the year this research was conducted increased at three times the normal rate. Enrollment in the elementary schools had grown by l80 students during the past 18 months. ‘While the school district spent $3,072 a year per child compared to the state average of $2,767, it faced the necessity of severe program reduction unless an upcoming millage proposal was passed. An emphasis on writing was one of the priorities set by the school board for the l985-86 school year. the str if} of CO 5&3 84 The elementary school in which this study was conducted served a more diverse economic population than the other three in the commu- nity. (Although a majority of its population of 439 children were from the upper middle class, it enrol led children from all socioeconomic strata—-low5 middle, and high. .Approximately 8% of the students qual- ify for free or reduced-price meals. 'The principal describes his staff of 30 as having high expectations for themselves and as being strongly committed to a high level of student academic growth. 'The teachers speak of the principal as a supportive facilitator. 'The school prides itself on the high level of community involvement at all levels, including yearly participation of approximately'ZOO parent volunteers in programs such as Project Charlie (drug education), Junior Great Books, a publishing center, and Picture Person (presentation of famous works of art). Although the entire teaching staff is Caucasian, approximately l5% of the students are from other ethnic groups--3% Black, 10% Asian, and 2% Hispanic. Two fifth grades were initially recommended for this study by the building principal. One of these classes was taught by two teachers, one in the morning and one in the afternoon. Because of this shared teaching situation, scheduling proved impossible and a third fifth-grade teacher was asked if she would be willing to participate. Her acceptance provided two fifth-grade classes, one with 26 students, one with 27. Forms for permission to participate in a research study were immediately distributed to the parents of all children in the two classes. Permission was received for El children to participate, 26 frm sub exp sco eqr 85 from the first room, 25 from the second. One child from the first room subsequently moved, and a child from the second room left the study as explained earlier. Following the completion of the treatment, the scores of an additional child were selected at random and discarded to equalize the groups for analysis. ndo ssi ent Once the sample was determined, a reading comprehension score from the standardized test routinely administered by the school district was obtained for each child having permission to participate. The Gates MacGinitie Reading Test (l978), Level D, Form 1 had been administered to all children during May 1983—84, the school year before this study. An equivalent form was administered to new students. ‘The NCE (normal curve equivalent) score was used. For this sample, the stanine scores were as follows: Stanine l--0% Stanine S--23% Stanine 2--2% Stanine 6--l7% Stanine 3--2% Stanine 7--27% Stanine 4--6% Stanine 8--lO% Stanine 9--l2.5% Whereas a normal distribution of scores would have 40% below, 20% at, and 40% above the fifth stanine, this sample was skewed to the top half of the normal curve as expected, with l2% below, 23% at the fifth stanine, and 67% above that. After the scores were obtained, an assistant to the researcher assigned a number to each name and corresponding score, then detached the names from the scores and filed them separately. 'The researcher then divided the scores into three groups: Group l--higher com] (NCI ins “wt J . 86 comprehension scores (NCE 73-99), Group 2--middle comprehension scores (NCE 55-73), and Group 3--lower comprehension scores (NCE 20-55). In instances where it was necessary to assign subjects with the same score to different groups, the Gates NCE vocabulary score was used to deter- mine which subject was assigned to which group (Table 3.l). Within the three groups, each subject was then randomly assigned to one of the two treatments. 'To assure that no child could be identified by reading level during the treatment, the three groups within each treatment were randomly mixed using a random numbers tablet At no time did the researcher see the names of the children and their reading or vocabu— lary scores together. cheduli d r niz tion By rearranging their daily schedules, the teachers were able to devise a schedule for researcher/student contact which provided one and one-half hours on Monday morning and Thursday afternoon, and two hours on Tuesday afternoon, Wednesday afternoon, and Friday morning for each of nine weeks. The time per day was divided equally between the two treatments. Random assignment of the two classes complicated scheduling because it meant that only approximately half of each class would be meeting with the researcher at any one time. Rather than combine the rest of the children into one group, the two teachers decided each would conduct class with the children remaining in their respective classrooms. This arrangement provided continuity for the children and assuaged the teachers' concerns that parents wanted the major class writ‘ Tabl Grom . .__;__.I_4__n__n_l 87 major portion of instructional time to be under the direction of the <3lassroom teacher. In light of the school districtfls emphasis on writing, the researcher became known as a special writing teacher. Table 3.l.--Gates-MacGinitie reading comprehension scores (normal curve equivalent). No. of NCE Compre- NCE Vocabu- Group Students hension Score Stanine lary Score l l 99 9 l 3 93 l l 88 l l 84 8 l l 8l 1 3 78 l 3 76 7 l 3 73 l03, 84, 82 2 3 73 73, 73, 57 2 2 7l 2 3 66 2 l 64 6 2 2 62 2 2 60 2 3 58 3 3 55 S 3 2 53 3 2 51 3 2 49 3 2 47 3 2 44 4 3 l 41 3 l 40 3 l 29 3 3 l 20 2 and res semi mai see stu ass nai ro ch ti The two treatments met back to back. To minimize disruption and encourage a smooth transition between groups, the teachers and the researcher concurred that it was better to have the groups meet in the same order each day rather than to alternate. This arrangement was maintained throughout the treatment period since the researcher could see no benefit in reversing the meeting order halfway through the study. To avoid the connotation of ability difference frequently associated with reading instruction groups, the treatments were desig— nated as the "early group" and the "late group" according to the order in which they met. To provide optimum facilities for the study, one of the class~ room teachers volunteered his room. PE» in turn, took his remaining children to the art room, the library, or a small room off the back of the classroom, but separate from it. 'The claSsroom used by the researcher was large, roomy, and carpeted except for the back portion. In addition to a desk for each child, it contained two round tables, three carrels, and several extra desks for independent or small-group work. The desks were usually arranged in small groups, a large square, or rows in which desks abutted each other. The room was windowless but had a door which opened onto a main corridor. Because of constant distraction in the hall, the door was generally kept closed, often causing the room to become warm and stuffy. Although the room was equipped with a circulation system, the noise generated by the fan interfered with the children's ability to hear the literature being read and thus was seldom used. ‘Having her own room meant that the chi att anc ric 0P! F0- bc 89 children, not the researcher, moved. ‘This, together with the positive attitude of the teachers and the principal, contributed to the accept- ance of the program as the language arts portion of the regular cur- riculunt During each treatment session, half of the children had the opportunity to sit in their own desks while the children from the other room used a desk belonging to someone else. While occasional minor irritations resulted when visiting children disturbed the contents of a borrowed desk, this caused no major problems. e ectio d sis of iter ture Specific literary selections to be read aloud to the students were chosen from the classical detective genre. Literature for the study was selected by this researcher based on the following criteria: l. The novels to be read aloud were part of a series, all were written by a single author, and both series contained at least four titles. 2. Each title used met the criteria of the classical detective genre. 3. The publication date was recent enough that the language and style used by the author were not dated. 4. The literature was reviewed in professional journals as being appropriate for children in the middle to upper elementary grades. 5. While series books and/or detective stories for children are seldom considered worthy of review by authorities in the field of i . l1“ "I. :I. 90 childrenhs literature, all titles used in this study were reviewed by major professional journals. Each title in the Literary Series was positively reviewed in two out of three major reviewing journals, namely Hohhbook, Bookljst, and eti o e Cen e o C i d enks Eggks, Each title in the Popular Series was reviewed in at least one of these same journals as acceptable or better. 6. The characters, plot, and setting of the two series were comparablen the main characters in both series were children, with a member of the police force as a "significant other"; the plot in each novel of both series met the criteria for a classical detective story; and both series were set in England, albeit one series pertained to a contemporary time period while the other occurred around the turn of the century. Having met these criteria, the following titles were selected and used in the study: Literary Series Newman, Robert. The Case of the yghishing Corpse. Atheneum, 1980. __ . e se 0 e re tened in . Atheneum, l982. It . Ihe Qgse of the Fcightehed Etiend. Atheneum, l984. Popular Series Dicks, Terrance. T ke tree rre l r i e s e Missing Mgsterpiece. Elsevier/Nelson, 1979. ___L . e ker treet e u r i e se 0 the l c i o s. Elsevier/Nelson, l981. ___ . e ker tree rre rs i e C s o t e Ghost firgbbers. Elsevier/Nelson, l98l. 91 The Primgry Irgit Scoring Instrument After their selection, these titles were rated by two profes- sionals in the area of children's literature, who validated the deci— sion of this researcher. Both raters had obtained their Ph.D.'s from najor midwestern Land Grant universities with majors in the area of :hildren's and adolescent literature. One is currently a professor of :hildren's literature at a college in a major Michigan city. The other is the reading consultant for a large city school system in a mid-size nidwestern city. 'The series were rated on their quality as literature and on the extent to which they developed the characteristics of the :lassical detective genre. 'The rating was performed on a primary trait scoring instrument developed by the researcher (Appendix A0. Primary trait scoring is a holistic scoring method which requires evaluators to assess a piece of writing as a whole. For scoring purposes, how- ever, the particular aspects of writing to be attended to in the holis- tic evaluation are identified. According to White (1985), "the primary trait score is a single number representing the quality of the paper according to the scoring guide for that trait" (p. 23). The instrument used in this study was designed to determine the extent to which the writing in these two series reflected the qualities of good literature and to determine how successful the series were in developing the characteristics of the classical detective genre. From a careful and thorough search of the literature pertaining to the qualities of literature, five elements of literature were identified: plot, setting, characters, style, and mood. The primary trait separe each s provh also puzzl retur disc: SGCOI astu desi to sub ree or) 92 'ait instrument asked the raters to consider and evaluate each element sparately. Due to the interrelationship of these elements within rch story, however, the ratings on the five elements were combined to 'ovide a total score representing the literary quality. Ten traits pertaining to the classical detective genre were so identified. The primary trait of this genre is its creation as a Izzle story in which a normal but currently disrupted situation is aturned to normal by a major character who uses clues present or ‘scovered to solve the problem through logic and deduction. Nine :condary traits were also identified: l. The detective is adequately established as being mentally stute. 2. The actions of the characters are consistent with the asignated role the character plays. 3. The plot centers on the detecti ve's investigation, solution ’ the crime, explanation of the solution, and final resolution of the ituation. - 4. The setting imitates the real world but is removed from it > the extent that the events and actions in the story world are not ijected to the standards of plausibility and probability used in the aal world. 5. The crime is not important as such, but provides a back- ‘ound for the story and a reason for the deductive activity of the atective(s); it is not the focus of the story. integ once out a abow pres pers ing ten gen App Edu f0! USi 93 6. (Blues presented are unobtrusive on a first reading, yet are integral to the story and can be seen as contributing to the solution once that solution is known. 7. In spite of the formula restriction that things will turn out all right, the author creates suspense by developing uncertainty about the fate of a character. 8. ‘The solution of the crime is a logical outcome of the clues presented in the story when those clues are seen from the appropriate perspective. 9. 'The story ends with a resolution of the disruption, includ— ing an explanation of the solution. These 15 traits, five reflecting general literary quality and ten reflecting the extent of development within the classical detective genre, compose the two sections of the Literature Rating Scale in Appendix A. Each trait was measured on a six-point scale. ‘While the Educational Testing Service generally defines four levels;of competency for a trait (Mul lis, l98l), 5-, 6-, 9-, and lS-point scales are also in use. This researcher chose to use the six-point scale, which "is becoming more and more standard for such tests" (White, 1985, p. l25). The six-point scale permits greater precision by the raters, particu- larly since it does not offer a midpoint, thus forcing a choice on mid— range papers (Myers, 1980; White, l985), and permits greater discrimi- nation in statistical analysis than do scales using fewer points. Liter comp‘ ti on high The cie Pop thi 94 Bgting of the Litehgture Copies of each novel were sent to both raters along with a Literature Rating Form: Part I and Part II, which they were asked to complete for each novel and for each series as a whole. The correl a- tion between the ratings of the two readers and the researcher was very high: 0n literary quality: researcher and rater l -- .96 researcher and rater 2 -- .94 rater l and rater 2 -- .84 On genre quality: researcher and rater l -- .83 researcher and rater 2 -- .85 rater l and rater 2 -- .99 The data used to compute the Pearson product-moment correlation coeffi- cient between the ratings are given in Tables 3.2 and 3.3. All three evaluators rated the Literary Series higher than the P0pular Series on Part I of the Literature Rating Form, indicating that the literary quality of the Literary Series was higher than the liter- ary quality of the Popular Series, and on Part II of the Literature Rating Form indicating that the traits of the classical detective genre were better developed in the Literary Series than in the Popular Series. ‘The differences between these ratings were all significant with p < .001. (See Tables 3.4 and 3.5.) Both raters felt strongly that, while the Literary Series deserved the higher rating it received, it was less likely to be read by children because of its sophistication; therefore, it was less appropriate for its reading audience. ‘This professional evaluation lent credence to the purpose of this study: to determine whether and Tabl 95 Table 3.2.--Professional literature: correlation of rater ratings of literary quality. x v xv x2 Y2 Rater 1 and Researcher Literary Series Novel 1 2A 27 688 576 729 Novel 2 30 30 900 900 900 Novel 3 28 28 78A 78“ 78“ Popular Series Novel 1 15 17 255 225 289 Novel 2 15 20 300 225 #00 Novel 3 15 21 315 225 “Al Total 127 183 3,202 2,935 3,5h3 ny = .96 Rater 2 and Researcher Literary Series Novel 1 28 27 756 781i 729 Novel 2 27 30 810 729 900 Novel 3 25 28 700 625 78“ Popylar Series Novel 1 l7 17 289 289 289 Novel 2 21 20 #20 hhl #00 Novel 3 21 21 #41 hhl #81 Total 139 Th3 3,916 3,309 3,5h3 ny = .SA Rater l and Rater 2 Literary Series Novel 1 24 28 672 576 784 Novel 2 30 27 810 900 729 Novel 3 28 25 700 78k 625 Popular Series Novel 1 15 17 255 225 289 Novel 2 15 21 315 225 Ahl Novel 3 15 21 315 225 Ah] Total 127 139 3.067 2.935 3.309 R - .BA xv Table 96 Table 3.3.--Professional literature: correlation of rater ratings of genre development. x Y xv x2 Y2 Rater 1 and Researcher Literary Series Novel 1 56 54 3,024 3,136 2,916 Novel 2 6O 59 3,540 3,600 3,481 Novel 3 59 58 3,422 3,481 3,364 Popular Series Novel 1 30 50 1,500 900 2,500 Novel 2 34 55 1,870 1,156 3,025 Novel 3 30 47 1,410 900 2,209 Total 269 323 14.766 13.173 17.495 ny e .83 Rater 2 and Researcher Literary Series Novel 1 51 54 2,754 2,601 2,916 Novel 2 52 59 3,068 2,704 3,481 Novel 3 52 58 3,016 2,704 3,364 POpular Series Novel 1 46 50 2,300 2,116 2,500 Novel 2 48 55 2,640 2,304 3,025 Novel 3 47 47 2,209 2,209 2,209 Total 296 323 15.987 14.638 17,495 ny = .90 Rater 1 and Rater 2 Literary Series Novel 1 56 51 2,856 3,136 2,601 Novel 2 60 52 3,120 3,600 2,704 Novel 3 59 52 3,068 3,481 2,704 Popular Series Novel 1 30 46 1,380 900 2,116 Novel 2 34 48 1,632 1,156 2,304 Novel 3 30 47 1,410 900 2,209 Total 269 296 13,466 13,173 14,638 ' .99 R XY iiiiiiiii Table 58111 Rtrl S’ol¢a~mmm able 3.4.--Professiona1 literature: ratings from Literature Rating Form, Part |-- literary quality. eries as a Whole Novel 1 Novel 2 Novel 3 Rtrl Rtr2 Rchr Rtrl Rtr2 Rchr Rtrl Rtr2 Rchr Rtrl Rtr2 Rchr Literary Series 5 6 5 4 6 5 6 6 6 S 6 6 5 5 6 3 5 5 6 5 6 5 5 5 6 5 5 6 5 5 6 5 6 6 4 5 6 5 6 6 6 6 6 5 6 6 S 6 6 5 6 5 6 6 6 6 6 6 5 6 28 26 28 24 28 27 30 27 30 28 25 28 Mean = 5.47 Mean = 5.27 Mean = 5.80 Mean = 5.40 o = .32 o = .88 o = .41 o = .63 P0pular Series 3 4 3 3 3 3 3 5 5 3 4 4 3 5 4 3 4 3 3 5 5 3 5 5 3 4 3 3 3 3 3 4 4 3 5 4 3 3 3 3 3 3 3 4 4 3 3 4 _4' 5 5 3 4 5 3 5 5 3 4 4 l6 1 18 15 17 17 15 23 23 15 21 21 Mean = 3.67 Mean = 3.27 Mean = 4.07 Mean = 3.80 o = .82 o = .59 o = .88 o = .77 T = 7.22:2 T = 7.236 T .. 6.88% T = 6.20* *Significant at the p < .05 level. Table 3.5.--Professional literature: genre deve10pment. 98 ratings from Literature Rating Form, Part 11-- Series as a Whole Rtrl Rtr2 Rchr Novel 1 Rtrl Rtr2 Rchr Novel 2 Rtrl Rtr2 Rchr Novel 3 Rtrl Rtr2 Rchr Literary Series S 5 6 6 5 6 6 . 5 6 6 5 6 6 6 6 6 6 6 6 6 6 6 6 6 6 5 6 6 4 6 6 5 6 6 4 6 6 5 6 5 S 5 6 5 6 6 5 6 5 6 6 5 6 5 6 6 6 6 5 6 6 5 5 6 5 5 6 4 5 5 5 5 6 5 6 6 5 6 6 5 6 6 5 6 5 5 5 5 S 4 6 S 6 6 5 5 6 5 6 5 4 5 6 5 6 6 6 6 _6 .2 .2 .2 .2 .2 .6. 2 .2 .2 .2 .2 , 57 53 58 S6 51 54 60 52 59 59 52 58 Mean = 5.60 Mean = 5.37 Mean = 5.70 Mean = 5.63 o - .49 o = .67 o = .54 g = .56 POpular Series 4 5 5 3 5 5 3 5 5 3 5 5 4 5 5 3 5 5 3 5 6 3 5 4 4 5 5 3 4 6 3 5 6 3 4 4 5 6 5 3 6 5 3 6 6 3 6 5 4 4 5 3 4 5 3 4 5 3 5 4 3 3 4 3 3 4 4 3 4 3 3 4 4 4 4 3 4 4 4 4 4 3 4 5 3 4 4 3 4 5 3 4 4 3 3 4 3 5 6 3 5 S 4 S 6 3 5 6 .2 .2 .2 .2 6 6 .2 .2 .2 .3. .2 .2 38 47 49 3o 46 50 34 48 53 30 47 47 Mean = 4.47 Mean = 4.20 Mean - 4.43 Mean = 4.10 o e .90 o = 1.10 o - 1.10 o c 1.09 T = 6.044 T = 4.98:2 T - 5.654 T - 6.84* *Significant at the p < .05 level. 111111 1‘6'C ure: frm wor obt all nal 99 in what ways the quality of literature read aloud to children, and thus re-created or reconstituted for them, affects their writing. T-units, flords Pen l-unit. and yocgbulary §core In addition to the two primary trait scores, three other meas- ures. later to be compared with the children's writing, were obtained from the professional literature-~the number of T-units, the number of words per Teunit. and a weighted vocabulary score. 'These measures were obtained by using a random sample from each title in each series read aloud. ~Since each series contained substantial amounts of dialogue and narrative, and since all of the above measures differ depending on whether dialogue or narrative is being used, both were sampld and the results averaged to obtain a final measure. An equivalent of approximately two pages of each, dialogue and narrative. was analyzed. Samples were collected by identifying a random page, then taking from that page all extended pieces of discourse. dialogue and narrative. An extended piece of discourse was defined as running text more than one sentence in length. Dialogue of one sentence or less occurring between sections of narrative was not counted. nor was narra- tive of one sentence or less occurring between sections of dialogue. No paragraph of narrative was counted if any portion of it was dia- logue. The equivalent of approximately two full pages of text was compiled from each of the six books, three from each series. cl 2 ati bet 100 I-units and words Per I-unit lhuflfis (1970) structural definition of a T-unit as "one main clause plus any subordinate clause Or nonclausal structure that is attached to or embedded in itfl (p. 4)‘was used for this research because it is objective and relatively easy to apply. T-units, T-units per hundred words, and words per T—unit for the professional literature read aloud are shown in Table 3J1 In all instances, the means of the random samples favored the Literary Series; ldh, the Literary Series had fewer T—units per hundred words, and those T-units were longer: T—units[]00 words HordsZT-gnit Literary Series Mean = 36.24 Mean = 11.8 Popular Series Mean = 44.54 Mean = 10.0 weighted yocgbulary §core The vocabulary score was obtained by assigning points to words identified by the American Heritage Intermediate Corpus (AHI) and reported in the American Heritage flord Fceguency Book (Carroll. 1971) as those appearing infrequently in material used in schools with children from grades three through eight. The less frequently the word was used, the higher the score. The American Heritage word list was chosen for several reasons: 1. It used as its source "textual samples from published materials to which students are exposed in grades 3 through 9" (Richman. 1971, p. v). ‘Texts, library fiction. library nonfiction, library general reference, and magazines, as well as other school materials, were samples. .Since the American Heritage sample was drawn Table 101 Table 3.6.--Professiona1 literature: T-unit counts. No. of T-Units/ Words/ Words T-Units 100 Words T-Unit Literary Series Dialogue: Novel 1 387 43.00 11.11 9.00 Novel 2 387 41.69 10.77 9.28 Novel 3 387 39.75 10.27 9.74 Mean = 41.48 10.72 9.34 Narrative: Novel 1 468 33.57 7.17 13.94 Novel 2 468 32.36 6.91 14.46 Novel 3 468 27.06 5.78 17.29 Mean = 31.00 6.62 15.23 Totals: Mean = 36.24 8.48 11.80 Popular Series Dialogue: Novel 1 387 49.67 12.83 7.79 Novel 2 387 48.24 12.47 8.02 Novel 3 387 46.70 12.07 8.29 Mean = 48.20 12.46 8.03 Narrative: Novel 1 468 41.27 8.82 11.34 Novel 2 468 30.21 6.46 15.49 Novel 3 468 40.33 8.62 11.60 Mean = 37.27 7.97 12.81 Totals: Mean = 42.74 10.01 10.00 102 from materials written by adults for a child audience, it seemed that this word list would reflect both the breadth and the depth of the language used in the literature read aloud and could be used to measure the extent to which the children's writing reflected this language. In contrast. the Lorge-Thorndike and the Kucera lists are both based on materials composed primarily for adults (Kucera 8 Francis, 1967; Thorn- dike & Lorge. 1963). Other frequency lists considered as possible sources of comparison for this study were not considered adequate because the frequency count was either based on too small a sample or was obtained from materials written or intended for a younger or older population. For example, although Horn examined over five million running words, these words were drawn from adult written materials with a heavy emphasis on letters. Durr's (1970) list was composed of words found in library books for primary grades; he only sampled approxi- mately 100.000 words. Harris and Jacobson (1972) examined adult- authored elementary textbooks. involving a sample of overALS million words. but the textbooks were intended for grades one through six and thus did not contain a broad enough sample of vocabulary for this study. A study by Hillerich (1966) which tabulated the words used by children in their creative writing done in grades one through six was inviting. but again the total corpus was small. totaling only 380.340 running words. 2. It is the most recent large-scale word list to be published (the material was collected in l969L Another word list, called the Kucera list and based on a corpus collected out of Brown University. 1., 103 was published about the same time, but that material was collected in 1963-64 and sampled adult printed material (Kucera 8 Francis, 1967). Since vocabulary use changes constantly. it:was important for this researcher to use as recent a frequency list as possible. The word list compiled by Thorndike and Lorge (1963) contains vocabulary col- lected from 1921 through 1944. The Rinsland study was compiled from vocabulary collected between 1936 and 1945; the Horn list was compiled in the late 19305 (Hillerich. 1966L 3. The rank order of this corpus. unlike other lists such as the one by Kucera. includes derivatives. not just base words. Thus. it was possible to give every word a specific point value. 4. The actual frequency values of words in the AHI have been adjusted to estimate their true frequency. thus assuring, am}. that ‘values of words used frequently in detective stories but not elsewhere will be adjusted to reflect the relatively infrequent use over the theoretically'infinite sample represented by the corpus. This lower rating will more accurately reflect their true usage. yocgbulary Scoring Procedure The estimated true probability of each word's occurrence is reported as a Standard Frequency Index (SFI). An SFI of 90 indicates that a word may be expected to occur once in every 10 words, and SFI of 80 represents an expected occurrence once in 100 times. and an SFI of 50 indicates an expected occurrence of once in 100.000. 104 The prototype used in this portion of the study is that of Finn (1977). who, in his model for computer—aided description of mature word choices, defined rare words as those with an SFI of 50 or less (p. 77% Words having a low SFI are not bizarre or little—known words; they are, rather. relatively uncommon and used with relatively low frequency. To obtain a list of low-frequency words used in the literature. the researcher developed a program for use with the Apple IIe home computer. Using a commercial program entitled "Sensible Speller IV" distributed by Sensible Software. Inc.. the researcher was able to create an original dictionary for use in the study. Based on Finn's definition of rare words as those with an SFI of 50 or less. all words in the AHI Corpus with an SFI of 50.99 or less were typed onto a computer disk and used as a dictionary against which to compare the random samples from the novels. The program ignored all words with an SFI over 50.9 but identified those with an SFI of 50.9 and lower. All proper names and other proper nouns not found in the corpus were deleted because of their highly specific referents. All numbers were also eliminated since the computer program recognized numbers written as text but eliminated those written as numerals. Words remaining were considered to be relatively uncommon and were assigned point values. Point values were determined by looking up the word in the alphabetical list of the corpus and then finding the assigned value in the rank order list. The number indicating the rank of the word in the rank order list became the weighted vocabulary score ass ass 105 assigned to that word. Individual words not found in the listing were assigned points in the following manner: 1. 10.000--any word derived from a root word with an SFI over EKL9 or listed in the footnotes as appearing once in samples designed below seventh—grade level; 2. 25.000--any word derived from a root word with an SFI less than 50:9 or listed in the footnotes as appearing once in samples designed for seventh grade or above; 3. 40.000--any foreign or antiquated word such as bloke, organ grinder. or porte cochere. Finally. since the computer program eliminated hyphenated words, treating each as two individual words. the literature samples were reread for such words. These were then compared with the corpus and assigned their appropriate value. . The number of words receiving points as well as the number of points received for the professional literature read aloud are shown in Table 35L. The Literary Series had an average of 26 words per sample which were identified as mature word choices. These 26 words each received an average of approximately 16.000 points. ‘The Popular Series had an average of 39 words per sample which were identified as mature word choices. 'These words. however, received an average of only 14.000 points apiece. This means that the words identified in the Literary Series as mature vocabulary appeared less often in materials to which children are exposed in school. Since this series was set in 1900 England. it contained many words from the turn of the century, Table 3.7.--Professional literature: 106 vocabulary score. No. of Words Total Avg. Points/ Receiving Points Points Word Literary Series Dialogue: Novel 1 34 598.624 17.606.59 Novel 2 21 269.681 12.841.95 Novel 3 34 561.811 16.523.85 MeanD = 29.67 476.705.33 16.068.72 Narrative: Novel 1 31 487.283 15.718.80 Novel 2 14 251.768 17.983.43 Novel 3 24 342.864 14.286.00 MeanN = 23 360.638.33 15.679.93 Totals: MeanD+N = 26.33 418.671.83 15.898.93 Popular Series Dialogue: Novel 1 31 429.016 13.839.22 Novel 2 40 473.473 11.836.83 Novel 3 34 544.432 16.012.71 MeanD = 35 482.307 13.780.20 Narrative: Novel 1 36 515.306 14.314.06 Novel 2 37 482.887 13.051.00 Novel 3 54 760.808 14.089.04 MeanN = 42.33 586.333.67 13.850.40 Totals: Meanp+w = 38.67 534.320.33 13.818.63 fie the 107 amp. porte cochere. The Popular Series contained more words identi- fied as being mature choices. but each was more likely to appear than the words used in the Literary Series and thus received a lower number of points. Procedure and Methodology The researcher met with the two groups of children on 49 of 52 consecutive school days. The researcher did not meet with the chil- dren on Halloween and on two days when she attended the National Coun- cil of Teachers Annual Convention. With the exception of the literature read aloud. the procedure for both groups was the same. After a child had finished writing. revising. and editing his/ her story. that child no longer attended the sessions but remained with the regular classroom teacher. The researcher met with the children and fol lowed the procedure of "reading aloud. then write" until the last child was finished. Every child heard literature read aloud until she/he was finished writing. revising. and editing. oduc ion d Pretest The researcher. drawing on experience gained during the pilot study. ascertained that none of the children was familiar with either author or series to be used in the research. The week before the onset of the study. with the approval of the principal. she had the school librarian remove from the shelf for the duration of the study the only book the library possessed by either author of the two series used in this research. This book, The Case of the Someryille Secret by Robert L Newme also chil cl as clas Mi“ "Mi ini dur eii $9! 108 Newman. was not used in the study. During this week. the researcher also spent three afternoons observing in the two classrooms whose children were to be part of the study. Friday of that week. the classes were divided into groups for the first session of what the classroom teachers termed the "writing class." The first six days during which the groups met were devoted to writing an initial detective story. ‘This story became the pretest which provided the covariates for the analysis of covariance. 'The initial part of the first session was used as a get-acquainted session. during which time it was established that no child had read a book by either author to be used during this study. The last part of the session was devoted to reviewing the five components of literature: characters. plot. setting. mood. and style. At this time. the researcher also introduced the ten traits of’a classical detective story: 1. The author creates a puzzle story in which a normal but currently disrupted situation is returned to normal by a major char- acter who uses clues present or discovered to solverthe problem through logic and deduction. 2. The detective is adequately established as being mentally astute. 3. The actions of the characters are consistent with the designated role the character plays. of th situa to ti subji real grou dete int 00C out abc 109 4. The plot centers on the detective's investigation. solution of the crime. explanation of the solution. and final resolution of the situation. 5. The setting imitates the real world but is removed from it to the extent that the events and actions in the story world are not subjected to the standards of plausibility and probability used in the real world. 6. The crime is not important as such. but provides a back— ground for the story and a reason for the deductive activity of the detective(s); it is not the focus of the story. 7. Clues presented are unobtrusive on a first reading, yet are integral to the story and can be seen as contributing to the solution once that solution is known. 8. In spite of the formula restriction that things will turn out all right. the author creates suspense by developing uncertainty about the fate of a character. 9. The solution of the crime is a logical outcome of the clues presented in the story when those clues are seen from the appropriate perspective. 10. The story ends with a resolution of the disruption. includ- ing an explanation of the solution. In a large group. the children identified books and television programs they had read or seen which they felt met one or more of the criteria presented and discussed how those examples met those criteria. ___J ent Th1 dei res Cal 110 The second day. the researcher read aloud a short story entitled "Footprint in the Sky." part of a series by John Dickson Carr. This story was chosen because it contained all of the characteristics of the classical detective genre. was written by an award-winning detective story writer. was part of a series. and was felt by the researcher to be appropriate for fifth-grade children. In addition to Carr's short story series. 50 of his 70 mystery novels belong to series featuring one of the three detective characters he created. Another novel. 111e,,Life of Sir Arthur Conan Dovle (1949) was highly successful. Carr was president of the Mystery Writers of America in 1949. received the Mystery Writers of America Edgar Allan Poe Award in 1949 and 1969. the ler ueen's ster ine award twice. and the Grand Master Award in 1962 (Reilly. 1980). The third day. the researcher again presented the characteris— tics of literature as developed in the classical detective story. this time with a visual (Figure 3.2). This visual was composed of four large posterboard posters. one for each of the components of litera- ture. These general headings were flollowed by specific characteristics unique to the classical detective story. These four posters were displayed each day throughout the study and were frequently referred to by the researcher. Using the short story as an example. the children were asked to help compose a brief storyline containing the essence of the primary traits for that story. ixx. how the detective in the story used clues and logic to solve a problem which had developed. The discussion guidelines in Appendix D. "From the Listener's/Reader's wer am 111 Point of View." were then used by the researcher in guiding a discus- sion with the children to help them focus on the development in the story of the other nine characteristics. Extensive brainstorming of topics for detective stories yielded three blackboard sections full of ideas such as a dog-napping. stolen jewels. a lost bike. a murder. a disappearance. a kidnapping. stolen money. or a missing pencil. The children and the researcher then composed several storylines for stories which an author such as the children might write. The children were told that the next day they would begin writing their own stories. and they were asked to begin thinking about a storyline for that story. Characters P ot - detective - problem or crime - victim - clues - suspects - investigation - culprit — solution by logic - others - explanation - resolution e ti Mood - real but unreal - certainty - suspense Figure 3.2.-~Four visuals: literature as developed in the detective story. 112 On the fourth day the children each received a file folder. They were asked to write four storylines for possible detective stories on the inside cover of the folder (Graves. 1983). After identifying four possible stories. they chose the one about which they wished to write and then began. By the end of the third writing session. all children had finished writing their stories. had been asked to make any changes they wished to improve the story. and had edited the finished product for the typist. When the children finished. they were permit- ted to engage in activities of their choice. providing such activities did not interfere with the children who were still writing. .All chil— dren finished except three who were absent. Upon their return. these three were provided with the time they needed to finish. It was apparent from the children's activity that they had had little or no experience with either revising or editing. Each of the pretest stories was typed by the researcher exactly as written by the child. No mistakes were corrected. although spelling errors were underlined to ease the children's work in editing. Using as an example a sample story from a four-week pilot study conducted before the commencement of this study. the entire group worked together to make revisions and to edit several paragraphs in that story. The typed pretest stories were then returned to the children. and they were encouraged to make any changes they desired or needed to make. The pretest stories were then retyped as corrected and returned to the children at the end of the study with their finished posttest stories. LAL_¥— fol ' rea< cl a and des cra 113 re tmen n Posttest After the writing of the pretest was completed. each session fol lowed the same pattern: the children listened to the researcher read aloud from the designated literature for the first half of the class time; the second half of each session was devoted to prewriting and writing activities. First. the children engaged in activities. as described below. designed to focus their attention on the authorks craft; second. they wrote their final story. The activities and mate- rials used during this time had already been refined during a four-week pilot study conducted in a fifth-grade classroom immediately before the beginning of this study. For three weeks following the writing of the pretest story. literature from the appropriate series was read aloud to each class for the first half of the class period. Both groups then participated in identical activities designed to focus the children's attention on how the author of the professional literature had developed his detective story. The questions in Appendix D. "From an Author's Point of View." were used to help the children focus on what the author. as a person/ writer like themselves. had done in his story to communicate to his audience what he wanted them to know. These questions were explored through discussion and activities. LBing prototypes of activities the children would later use individually to develop their own stories. the researcher had the children work in large groups and small groups to look at the way the professional author had developed the traits of a detective story in his writing. For example. the researcher and chi cha whi thi 114 : children in a large group identified the mental and physical characteristics given by the author to the detective and determined ‘ which of these were unique to the individual and which were demanded by the detective role. The children then. individually or in dyads. . repeated the process with a different character of their choice. In small groups the children also drew. on large pieces of newsprint. a 1 pencil sketch of a setting the author had verbally created in the novel to which they were listening. They subsequently presented their sketch to the class. described it verbally. indicated what they had not included and why. and asked for suggestions concerning other details that might be added. During another activity. the class as a large group decided on the major events in the novel. The following day. each child received a copy of the plot outline containing these major events in a format identical to one they would later use to develop their own plots. Then. individually. in dyads. triads. or small groups. they completed the plot outline by adding other events that provided the transitions between those already listed. Finally. to focus the children's attention on the author's conscious need to decide what information to disclose to the audience. and to accent the author's deliberateness in so doing. the children participated in an interview in which they were to assume the identity of the author and to respond to questions from a talk-show host (the researcher) con- cerning why they chose to disclose or not disclose particular infor— mation. whel i re wer lo mas 115 Since a number of the children had used the names of classmates when writing their pretest stories. and since this had resulted in frequent disruption and distraction from the writing task. the children were instructed not to use names of classmates in their second stories. To provide alternatives. an extensive list of possible character names was compiled by several of the children. written on newsprint with magic marker. and displayed in the room. After looking at the author's craft through the eyes of the professional author. the children began to design their own classical detective stories. ‘The first half of each class period was still devoted to reading aloud from the appropriate detective series. During the second half of the time. the children participated in activities developed to help them write their own stories. First. they partici— pated in extensive brainstorming on possible topics. Next. they brain- stormed specific settings in which detective stories might occur. Names of possible characters were already displayed around the room. Finally. the children were asked to use the t0pics and settings pre- viously brainstormed and orally to compose possible storylines. The researcher wrote these on the board as the children dictated. Follow— ing this large-group work. each child received a spiral-bound booklet (Appendix E). with a professionally cut coverstock cover. designed to help him/her think through and plan the story she/he planned to write. The professional-looking booklets were designed to permit each child to see the scope of the project in its entirety as well as to safeguard each chilcfls work against loss and damage. Upon receiving their hook' stow choo with Once the nea sto to ea< 116 booklets. the children were asked to write» on the inside cover. four storylines for different detective stories they might write. .After choosing the one about which they wished to write. each child shared it with the researcher. who checked it for characters. setting. and plot. Once the storyline was approved. the child was free to progress through the booklet at his/her own pace. The researcher met with each child a minimum of two times. once near the beginning of the writing process and again when the chilcfis story was completed. During the first conference. which was designed to establish a prototype for future conferences. the researcher asked each child to talk about his/her writing. Treating the child as the author. the researcher asked questions to clarify her understanding of what the child was doing and what the child intended to do. Ilia child requested help during a conference. the researcher used questions to clarify for herself and for the child what the child wanted to say to the reader. In the process of explaining aloud to the researcher. the child frequently decided by him/herself how to proceed- If the child insisted on specific suggestions. the researcher then provided three possible suggestions and asked the child to add anothen. As soon as such conversation stimulated an idea. most children abruptly returned to the text and proceeded to write with surprising intensity. indicat- ing that the child was very much in control of his/her writing even when temporary help was needed. The children in this study appeared to be unfamiliar with revision or editing. Therefore. as each child finished drafting a story. (hope she e exaci the ( fina chil rese ask par ire sh ch ‘ 117 Tstory. she/he was given a separate editing checksheet of things to do [(Appendix F). Since the researcher was to type the finished stories. lshe encouraged the children to edit carefully since she would type Eexactly what they had written in order to retain their meaning. After lthe child checked off each item. the researcher proofread the story a ; final time for editing purposes. circling any remaining errors. The child then corrected the circled items. brought the story to the a researcher. and went over the corrections with her. All joint editing was done in context; 1.6.. the child was i asked to read aloud a minimum of one sentence and frequently the entire paragraph in which the error or omission occurred. This oral reading 1 frequently prompted the children to make revisions in the text as well as editing changes. The more technical errors the child made. the more she/he read aloud to the researcher; frequently the child chose to change the text. Since the results of the pre- and posttest writing were to be shared by the children with their parents. special care was taken to provide a finished product in the posttest story. The researcher took extreme care. however. to make no suggestions pertain- ing to content. She did help children edit for omitted words. spelling and punctuation omissions or errors. and incorrect or missing tense markers. She also noted places where referents were uncertain. or where the child's meaning was unclear. In such cases. the child was told that the researcher. as a "reader." had difficulty understanding what the child. as author. meant. The child was asked to read the section aloud and to change it if she/he felt the necessity to do so. Usua If t aske 118 {Usually the child recognized the problem immediately and corrected it. (If the child felt the meaning was clear as it read. the researcher :asked the child to explain what was meant so that she could understand. §and then the two simply went on. In such cases. no changes were made. lIf the child recognized the problem but did not know how to change it, {she/he was asked to explain the problem and then asked to suggest a Epossible solution. This usually triggered an idea acceptable to the ‘!child. and she/he proceeded to make the necessary change. If more help I was needed. the researcher offered three possible suggestions. asked I the child to think of a fourth. and proceeded as described above. One of the factors to be measured in this study was the extent 1 to which the child's vocabulary reflected the richness of vocabulary found in the literature they heard read aloud. As the children became involved in their writing. they became intensely involved in communi- cating meaning. The intensity of the process of getting the meaning down on paper often seemed to preclude conscious attention to word choice. To help the children focus on word choice. each child was directed. on the editing checksheet (Appendix F). to replace ten words used in his/her story with ten more effective words. Any child who asked for clarification of these directions was told to choose rich and exciting words like the author of the novel had done. After submitting the edited copy. each child left the study and returned to his/her original classroom. Although almost all of the children finished their rough drafts before Christmas break. many still had editing to finish after they returned to school in January. During L this lens grou eve was 119 this time. the researcher continued to read each day. although the length of the oral reading time was decreased to ten minutes for each group. Ten minutes of oral reading seemed to provide enough time for every child to become involved in the text being read. Such involvement was deemed necessary by this researcher for the text to have an effect on a child's writing. All children met with the researcher for a total of 35 days. six of which were involved with orientation and the writing of pretest stories. During the treatment. all children heard literature read aloud for a minimum of 20 minutes and a maximum of 30 minutes per day. After the thirty-fifth day (lJL. during the eighth week of the study). children began completing their stories and returning to their normal <:lassrooms. During this week. the time spent reading aloud was reduced to 20 minutes per session so that the children could finish their writing. revising. and editing. .All children had finished draft- ing their stories by Wednesday of the week following the holiday break. After all children had finished drafting their stories. the length of time devoted to reading aloud was again reduced. this time to ten minutes per session. Since this study was designed to determine the effect of listening to literature on childrtwfls narrative writing. it was deemed important that no writing occur without exposure to that literature. Therefore. each child heard literature read aloud each day before she/he commenced writing on his/her story. .Although some chil- dren took longer to finish writing than others and thus heard more literaturtaread aloud. no child ayeh grote uihhoat haahjhg ]i;eratune LL-‘— were in A for fol by 120 read aload. After all posttest stories had been typed. the researcher presented each child with his/her pre- and posttest stories bound in a plastic cover. Compilation of the Data After the treatment was completed. pretest and posttest stories were typed for rating. Samples of the chilchwwfls stories may be found in Appendix G. Since each story was to be read and rated holistically for meaning and content. all stories were edited according to the following guidelines to assure that the raters would not be influenced by surface features of the writing: --Numerals in the text were changed to words when appropriate. --"A" was changed to "an" before a word beginning with a vowel. --Paragraphs were created if none were present. --Spelling was corrected. --Inappropriate choice of verb tense. if infrequent. was corrected. --When punctuation was missing at the end of a sentence. a period or semicolon was used. --A particular form of punctuation consistently misused was corrected or deleted. ~~Commas. if missing. were inserted as necessary in the following places: -—after introductory words such as yes. no. well. or oh; —-after introductory phrases and clauses; --to set off transposed words and phrases; --to set off appositives; --after parenthetical expressions such as moreover or however; --in compound sentences (if useful to clarify meaning); --with a series of nouns. verbs. or phrases. o is ic ti The raters of the children's writing were both enrolled at Michigan State University as Ph.D. students in Children's Literature and Language Arts. The first had earned her master's degree from the University of Southern California in Education with a remedial reading LL¥— foc for 121 focus. She had been an elementary teacher of grades three through six for 21 years. 1 year in Michigan and 20 years with the Department of Defense Dependent Schools in Japan. Germany. and Washington. DAL During five of those years. she served as a reading-improvement spe- cialist. The second rater had earned a B.A. in English and an M.A. in Speech at the University of Michigan and an Educational Specialist degree in Curriculum and Administration from Michigan State University. In addition to teaching grades 8 through 12 for 17 years. she served on a Michigan Department of Education committee to revise the state Writing Objectives K-12. She also served as the Vice-President and President for the Michigan Council of Teachers of English. made presen- tations on writing for professional conferences at the state and national levels. and conducted writing inservices at the secondary level. The two raters scored children's writing with the same instru- ment used to rate the professional “Literature. Each piece of writing was independently rated. with discrepancies resolved by a third rater. the researcher. Scoring was conducted according to accepted procedures in the field of holistic evaluation. 'These procedures were initially used on a large scale by the Educational Testing Service for National Assessment of Educational Progress UMullis. 1982). were refined and used extensively'by the Bay Area Writing Project (Myers. 1980). and have recently been summarized by White (l985h To assure an impartial reading of the pre- and posttests. all 96 stories were rated by the same twoievaluators as one set of papers. ‘The stories were randomly L,L.-; arra wri COP 122 arranged and read in the same order by each rater. At the conclusion of the rating. the evaluators indicated that they were not aware of having read two stories by each of 48 children. i.e.. 48 pre-post sets of writing. Intensive training of the raters was conducted using stories written by 30 fifth-grade children during a four—week pilot study conducted by the researcher immediately before the formal study. The rater training extended over a lZ—hour period on two consecutive days. After an initial reading. explanation. and discussion of the primary traits on the Literature Rating Form (Appendix A). the three raters read. rated. and discussed their ratings on several sample stories. This process of reading. rating. and discussing continued until the ratings between the raters reached a correlation of .90 when computed using the Pearson product—moment computational formula. Rating then began on the writing products of this study. Mullis (1980) reported that reading for a primary trait takes approximately one minute for papers by nine year olds and one and one- half minutes for papers by 13 year olds during scoring for the National Assessment of Educational Progress. She explained that this time will vary depending on factors involved. including the length of the papers to be read and the number of secondary traits to be scored. The raters for this study spent 20 to 30 minutes per paper read. This is compar— able to the time Mullis reported since the stories were relatively long and a total of 15 traits were evaluated for each paper. The total time necessary to score the stories exceeded 40 hours. An initial six—hour period was used to establish a routine for reading and to provide time for questions. After this initial period. the two raters continued scoring as their schedules permitted. Following White's (1985) recom- mendation. a four-hour retraining session using new sample papers was conducted after both raters had finished scoring half of the papers. Although White recommended that such retraining be conducted as fre- quently as every few hours to maintain fidelity to the standards origi- nally jointly agreed upon and to prevent reader drift up or down the established scale of judgment. only one retraining session was feasible during this study. The interrater reliability on the literary quality of the children's writing was computed using the sum of the ratings given by each of the two raters on traits 1 through 5 of Part I of the Litera- ture Rating Form (Appendix A) across 48 subjects. The Pearson product- moment correlation for the two raters was {ML Discrepancies on Part I were resolved if the totals of the two raters differed by more than five points. indicating an average difference of more than one point per item. Any time the sum of the ratings given by the two raters on the literary quality of writing differed by more than five points. the third independent evaluator rerated the traits where the scores were more than one point apart. and then the third rating was substituted for the most deviant score. Upon the advice of a statistical consultant. when the third rating fell nfldway between the two original ratings. the rating to be changed was randomly determined to prevent bias in either direction. Ten per BCC 124 percent of the items in Part I were rated by the third rater. which according to both Diederich (1974) and White (1985) is average for a holistic rating. For a reading of this complexity. however. where the same readers evaluated 15 traits for each paper. this appears to be an excel lent reading (Mullis. 1980). The two ratings were then averaged to obtain a score representing the literary quality of each chilcfls writing. After discrepancies were resolved. the Pearson product-moment correlation between the adjusted ratings on Part I was .874. The interrater reliability on the development of the genre characteristics in the chilchtwfls writing was computed using the sum of the ratings given by each rater on traits 6 through 15 of Part II of the Literature Rating Form (Appendix A) across 48 subjects. The Pearson product-moment correlation for the two raters on these traits was also.70. Discrepancies on Part II were resolved if the totals of the two raters differed by more than ten points. indicating an average difference of more than one point per item. Any time the sum of the ratings given by the two raters on the literary quality of writing differed by more than ten points. a third rater rated the traits where the rater scores were more than one point apart. and then the third rating was substituted for the most deviant rating. When the third rating fell midway between the two original ratings. the rating to be changed was randomly decided. Fifteen percent of the items on Part II were rated by the third rater. The two ratings were then averaged to obtain a score for each child representing the extent to which she/he 125 developed the traits of the genre in his/her story. After discrepan- cies were resolved. the Pearson product-moment correlation between the adjusted ratings on Part II was .91. Aiomistic Eating In addition to the two holistic scores. three other measures were obtained from the childrenhs writing--the number of T-units per story. the number of words per T-unit. and a weighted vocabulary score. T-units. determined according to Huntis definition. were counted in each chilcfls pretest and posttest story. Words per T-unit were determined by dividing the number of T-units by the number of total words written minus the number of words used in mazes. A maze was defined by Loban (1976) as "a series of words (or initial parts of words). or unattached fragments which do not constitute a communication unit and are not necessary to the communication unit" (p. 10). A vocabulary score was obtained for each child by adding together the points assigned to each word defined as a mature word choice. Words were assigned point values following the same procedure used for valuing words in the professional literature. However. in addition to the guidelines followed in assigning point values to uncom- mon words used in the professional literature. it was necessary to include several additional categories of words for the children's writing. Such categories were identified by Finn (1977) and are com- posed of words found in childrenhs writing which are rare in print but do not reflect maturity or sophistication. Instead. these words are ‘so informal that they are not likely to appear in print. 126 If used at ‘all by professional authors. they are used infrequently and judiciously and thus received a high number of points in the AHI corpus. In the .children's writing. however. these words may vary with the topic and/or ‘type of writing. are used very frequently. and represent an immature rather than a mature word choice. The categories identified in this study as representing immature word choice were (a) slang words. (b) words used with great frequency in dialogue. (c) strings of letters representing sounds and/or screams. and (d) words representing part of the popular culture of fifth-grade children. Since these words appeared to represent immaturity rather than maturity of vocabulary usage on the part of the children but did receive high values in the corpus. it was decided to retain them as part of the data but to assign nominal point values for their use. Because the words in the first two categories were used frequently in the children's writing. they were assigned a low point value of 500 points apiece; the words in the last two categories were used less frequently and were assigned 5.000 points each. Words from the children's writing that were included in these categories are listed in Table 3.8. deryieh Chapter III described the pilot study. design of the study. and the sample population. The criteria for the selection of the professional literature read aloud to the children and the manner in which that literature was analyzed were also explained in detail. The procedures and methodology used in the two treatments as well as in the 127 compilation of the data were carefully detailed. Chapter IV presents the results of the statistical analysis of those data. Table 3.8.--Low-frequency words in children's writing representing immature word choice. Words Beceivihg 599 Points Each . yes ya ok. OK. O.K. hi ‘ bye dude huh yea yeah wow gee gosh ‘ yucky ho guy guys kid dummy dumb darn goody Barfball goll yup yuck yippee yikes yah geez geez-o-peet Words Receiving 5.999 Poihts Each pfew uh or any variation uuuh or any variation um or any variation hm or any variation shhh or any variation letters representing screams letters representing sounds Raggedy Ann and variations Cabbage Patch Porsche Corvette Karate Judo Ty Ouan Do Scrabble Parchesi CHAPTER IV ANALYSIS OF THE EXPERIMENTAL RESULTS This chapter contains the analysis of the data obtained from the two stories written by the subjects of this study. one before the treatment and one during the treatment. The study was designed to answer four questions: 1. Does the literary quality of narrative literature which is read to and discussed with children affect the literary quality of their narrative writing when that writing is rated holistically? it Do children who hear literature in which the traits of the genre are well developed obtain higher scores. when rated on a primary trait analysis of their own writing within that genre. than children who hear literature in which those traits are less well developed? 3. .Are selected atomistic characteristics of the literature read aloud to children. ids. language fluency as measured by the number of T-units. syntactic complexity as measured by the number of words per Teunit. and vocabulary maturity as measured by a weighted vocabulary score. reflected in their writing? AL Does improvement in writing correlate with type of treat- ment for children with different reading abilities. as identified by a standardized reading test? Is the writing of children with high 128 lL 129 reading scores affected more by one treatment than the other? Is the writing of children with low reading scores affected more by one treatment than the other? 9veryieh In this study. the 48 subjects were randomly assigned. within reading ability groups determined by standardized reading test scores. to one of two treatments. 'The subjects in Treatment 1 heard chil- drenhs literature read aloud which was determined by qualified raters to be high in literary quality and in the development of the traits of the classical detective genre. 'The subjects in Treatment 2 heard childrtwfls literature read aloud which was rated lower in both of these aspects. The researcher administered the treatment to both groups. As described in Chapter III. two sets of data were collected for each child. One set of data was derived from the first. or pretest. story and was used as the covariate in a multivariate analysis of covariance (MANCOVAQ. A corresponding set of data was obtained from the second. or posttest. story and used as the dependent variable. Each set of data contained two holistic and three atomistic scores. The holistic scores were determined by trained readers who evaluated the children's writing using a primary trait analysis. The first score for each story was determined by averaging the ratings given by Rater l on traits 1 through 5 on Part I of the Literature Rating Form (Appendix A) with the ratings given by Rater 2 on these same measures. The averaged scores were then added to obtain a composite score which represented the literary quality of the child's 130 writing. 'The second score was obtained by following the procedure described above. but using traits 6 through 15 on Part II of the Literature Rating Form (Appendix A). This score represented the extent to which the child had developed the traits of the classical detective genre in his/her story. In addition to the holistic measures. both pretest and posttest sets of data also contained three atomistic measures: the number of T-units in the story. the number of words per T-unit. and a weighted vocabulary score representing the maturity of vocabulary used (Appendix J). To determine how the quality of literature read aloud affected the writing of readers determined by a standardized reading test to be of differing abilities. a 3 x 2 randomized block design was used as the basis for a multivariate analysis of covariance. Blocking was used to reduce the experimental error. and random assignment was used to equate the groups on all extraneous factors that might affect the outcome of the study. Since the host school district had administered the Gates- MacGinitie Reading Test. Level 0. Form 1 to their students in May of the preceding year. with an equivalent form administered to incoming children. the scores for this test were used to determine the groups for the 3 x 2 design. After the childrenhs names had been removed from their test scores. the test scores were divided into three groups according to high (Group 1). middlee(Group 2). and low (Group 3) reading comprehension scores (Table 3.1). As indicated in Chapter 111. these scores were skewed to the high end of the national normal 131 distribution scale. as might be expected for children from middle- to upper-middle—class backgrounds. Once the groups had been determined. members were randomly assigned. within groups. to treatments. 'The researcher administered the treatment to both groups. thus eliminating theielement of teacher variability. Separate MANCOVAs were conducted to analyze the holistic and the atomistic measures. 0 r ion of e 1 nd i i co es The analysis of data for this study was conducted to confirm or disprove hypotheses designed in light of a review of the best and most current theory and research available» Tukey (1977) suggested. however. that analysis is not enough. In explaining the need for exploration in addition to analysis. he employed an analogy particu— larly appropriate to this research: Even [in experiments where one line of data analysis was planned as part of the experiment or inquiry]. however. restricting one's self to the planned analysis—-failing to accompany it with exploration--1oses sight of the most interesting results too frequently to be comfortable. As all detective stories remind us. many of the circumstances surrounding a crime are accidental or misleading. Equally. many of the indications to be discerned in bodies of data are accidental or misleading. lt>accept all appearances as conclusive would be destructively foolish. either in crime detection or in data analy- sis. To fail to collect all appearances because some--or even most-~are only accidents would. however. be gross misfeasance deserving (and often receiving) appropriate punishment. (p. 3) In the process of exploring the data. this researcher discovered that the correlation between the children's reading ability. as measured by a standardized test. and the quality of the childrenks pretest writing. as determined holistically. was very low and not 132 significant when tested at the p = .05 level. Nor was there a corre- lation between the number of T-units written by the children and their reading scores. The relationship between the words per T—unit and vocabulary maturity was significant. however. 'The correlation between the Gates-MacGinitie reading comprehension scores and the pretest writing scores is shown below: Holistic rating: literary quality -- rxy = .19 (p = .10) (see also Table 4.1) Holistic rating: genre developnent -- rxy = .13 (p = .19) Number of T#units -- rxy = .03 (p = .42) Words per Teunit -- rXy = .33 (p = .01) Maturity of vocabulary -- rxy = .34 (p = .01) Because of the low correlation between the quality of the childrenks writing measured holistically and their tested reading ability. grouping the children by reading ability did not result in the smaller variance within cells which had been anticipated and which would have occurred had the writing scores of the children also been similar. The number of T-units written. the number of words per T-unit. and the maturity of vocabulary score were analyzed together in a sep- arate multivariate analysis of covariance. While there is a correla- tion between reading ability scores and both the words per T-unit measure and the maturity of vocabulary measure. there is no significant correlation between reading ability and the number of T-units written. This may. therefore. have decreased the possibility of finding existing differences for these measures. also. The discrepancy between reading and writing scores for the subjects of this research in no way affects the validity of the results of this study. It does mean. however. that because the 133 Table 4.1.--Comparison of standardized reading test scores (Gates- MacGinitie) with pretest writing scores (Literary Quality). Treatment 1 Treatment 2 Gates Pretest Gates Pretest Reading Writing Reading Writing Group 1--High Readers 99 16.5 93 11.5 93 16.5 88 19.5 93 8 81 18.5 93 9 78 14 84 15 76 7.5 78 16.5 73 10 78 22 73 10 76 19 73 10.5 Group 2--Above—Average Readers 73 13.5 71 18 73 12 66 23 73 12 64 13.5 71 9 62 10.5 66 17 60 10.5 66 11.5 60 17 62 13.5 58 12.5 58 12 58 12.5 A Group 3-—Average and Low Readers 55 15.5 55 11.5 53 12 55 7.5 53 13.5 51 9.5 49 16.5 51 9.5 47 23.5 49 16.5 47 10 . 44 6.5 42 14.5 44 16.5 20 5.5 29 12.5 ‘fi 134 within-cells variation on the pretest was great rather than small as expected. differences between groups. if they do in fact exist. became more difficult to find. n l sis o e o istic in s The first MANCOVA analyzed the scores from the two holistic ratings together. One score represented the literary quality of the chilchwwfis writing; the other represented the extent to which the children developed the traits of the genre in their stories. 'The results from the MANCOVA showed the early writing of the children (pretest) to be highly predictive of their later writing (posttest) with p <.001. The analysis also showed that after adjusting for this relationship. there was an interaction effect (p < .11) between treat- ments and reading groups. An interaction effect indicates that the effect of the treatment varies across groups; ix». in this study. the effect of the literature read aloud in the classroom varied with the reading ability (as determined by a standardized test) of the children. The results of this MANCOVA are discussed below. The discussion focuses first on the interaction between the literary quality of the literature read aloud and reading ability. second on the interaction between the development of genre traits in the literature read aloud and reading ability. and finally on a comparison of the interaction of both-~literary quality and genre development in the literature read aloud——with reading ability. 135 W The literary quality of the stories written by the children of differing reading abilities in the two treatment groups is represented by the cell means graphed in Figure 45L. The means. illustrated on the graph. indicate that the three reading groups in Treatment 1 who heard detective stories rated higher in literary quality wrote posttest stories judged to be higher in literary quality than their correspond- ing groups in Treatment 2 who heard detective stories rated lower in literary quality. The biggest difference between Treatment 1 and Treatment 2 occurred in the writing of the lowest readers (Group 3) in this study. 'The literary quality of their writing improved the most when they heard the high-quality literature. ‘That difference was bigger than the difference between treatments for Group 1 or Group 2. Both Groups 2 and 3 in Treatment 1. ixa. all the children determined by a standardized reading test to be in the sixth stanine or below nationally. not only did better than the corresponding groups in Treat- ment 2 but did almost as well as the best readers in Treatment 2. Just as taking a photograph from a different perspective will often clarify whether what one sees is really there. so looking at numbers from a different perspective can sometimes help determine whether an apparent difference is real or not. The MANCOVA indicated that there was a difference between the effects of Treatment 1 and Treatment 2 on the literary quality of the children's writing. The significance of this MANCOVA was not as high as desired. however; 1.90 there was more probability than this researcher desired that these 136 20.0 h 19.0 m 18.0 w 17.0 - 16.0 Group 1 Group 2 Group 3 Adjusted Posttest Means Treatment 1 Treatment 2 Group 1 20.3 618.6 Group 2 18.4 16.8 Group 3 18.3 16.1 Figure 4.l.--MANCOVA results: literary quality of children's writing. results were the result of chance. Therefore. to further explore the treatment effects indicated by the MANCOVA. the two-sample Median Test and the Mann-Whitney U Test were conducted using SPSS (Statistical Package for the Social Sciences. 1977L The Median Test combines the scores for the two treatment groups. determines the median. and creates a 2 x 2 contingency table using the number of cases exceeding and not exceeding the median to compute a chi-square statistic. The results of the Median Test. which uses the median rather than the mean as the measure of central tendency. support the results of the MANCOVA. Table 4.2 shows the comparison of the treatment groups both on the 137 Table 4.2.--Median Test: literary quality. Posttest Pretest Trt 1 Trt 2 Trt 1 Trt 2 Total Sample # Above Median 15 7 # Above Median 14 9 Median 18.0 Median 12.5 # Below Median 9 17 # Below Median 10 15 p < .03* p < .25 Group 1 # Above Median 5 2 # Above Median 2 Median 18.5 Median 14.5 # Below Median 3 6 # Below Median 6 Group 2 # Above Median 5 3 # Above Median 3 4 Median 17.3 Median 12.5 # Below Median 3 5 # Below Median 5 4 p < .62 p < 1.0 Group 3 # Above Median 7 1 # Above Median 5 3 Median 17.5 Median 12.3 # Below Median 1 7 # Below Median 3 5 p< .62 *Significant at the p < .05 level. 138 pre- and posttest. and indicates that the difference between the two was not significant before the treatment. but was significant at the .03 level after the treatment. The Median Test also reflects the results of the MANCOVA for the three reading groups. Except for Group 1. al'l reading groups in Treatment 1 show positive change. those in Treatment 2. negative change. Again reflecting the results of the MANCOVA. the Median Test found a significant difference in favor of Treatment 1 for the lowest readers in this study. p < .01. The Mann-Whitney U Test is more powerful than the Median Test because it uses the rank of each case rather than just its location relative to the median. It first combines the scores of both treat- ments. then ranks them in order of increasing size. The test statistic is created from the number of times a score from Treatment 1 precedes a score from Treatment 2. ‘The Mann-Whitney U Test also shows Treatment 1 to be superior to Treatment 2 on the posttest at the .02 level. It. too. indicates that for the lowest readers. the children in Treatment 1 achieved at a significantly higher level than the children in Treatment 2. Although the difference between treatments for Groups 1 and 2 is not significant at the .05 level. Table 4.3 indicates that the mean rank increased for all reading groups in Treatment 1 and decreased for all groups in Treatment 2. and supports the findings previously dis- cussed. 139 Table 4.3.--Mann-Whitney U Test results. 4* Lg Mean Rank Mean Rank Pretest Posttest Total sample: Treatment 1 26.33 29.10 Treatment 2 22.67 19.90 Significance .36 .02* Group 1: Treatment 1 9.63 10.00 Treatment 2 7.00 7.38 Significance .21 .34 Group 2: Treatment 1 7.31 8.94 Treatment 2 9.69 8.06 Significance -.32 .71 Group 3: Treatment 1 9.88 10.75 Treatment 2 7.13 6.25 Significance .25 .05* *Significant at the p < .05 level. Development of Genre Traits The development of the traits of the classical detective'genre by children of differing reading abilities in the two treatment groups is represented by the cell means graphed in Figure 4.2. The means. illustrated on the graph. indicate a strong interaction between the development of genre traits in the professional literature read aloud and the reading ability of the children. The children in the highest reading group did very well in developing the genre traits in their own writing. Apparently thoseo_ mo. v a use up ucmo_m_cm_m« «No. «mo. a_o. >u___nm30co Uo__muim m m N Eocoocm wo moocmoo Jo.m ON.N om.: v:_m>ih mo.— a... w_._ Lotto pumpemum mo.m _N.m sm.m co_um_>ma atmacmum _m.m mo.m .m.m some ooeocowm_a N©._ KN._ mm._ —m._ m—._ Nm._ LOLLm Ucmvcmum mm.m mm.m mo.m NN.: mm.m mN.: co_um_>mu Ucmvcmum ::.:_ m_._— _m.mw mu.:_ wo.m— mm.N_ cmmz N ucoEumoch O we s_o. aNO. «No. >6__mamaota eo_mmu-m m m m EOccoc» mo moocmoo #0.: m_.m mm.N o:_m>ih m_._ mn.~ wo.N Lotto ccmccmum mm.m No.m Nw.m cowum_>op ctmncmum _w.s em.m oo.e came mocataeeea mo.~ mm._ :m.. mm. mm._ mm._ totem ccmpcmum Nm.m m~.m .m.: mm.m m:.: Nn.: co_um_>mp ccmpcmum mo.w_ mm.m_ mo.m_ om.~_ .m..~ .m.m_ cmoz _ ucoEumock umouumOm “wooded “mouumOa “wooded amouomOm umouoca m a30co N a20co _ Quota .>u__mno >cmcouh_ii “mouiu ”mc_u_c3 m_:occ—_LUii.o.: o—nmh ..m>o. 00. v a 050 um ucmo.0.e0.m« «.0. «00. «.0. >0...0m20ca 00..m010 0 0 0 EOvoocm mo moocmoo 00.0 00.0 00.0 m:.m>-0 00.:. 00.00 ...0. cocco 0cm0emum 00.0: 00.00 00.0: co.um_>o0 0cmvcmum 00.00 00.00. 0_.00 cmoe mucmcm00.o 00.0. 00.0 00.00 00.0 00.0. 00.0 Lotto 0cmccmum 00.0: 00.0. 00.00. 00.:0 :0.00 00.0. co.um.>mn 0cm0cmum 00.00 00.00 00.00 00.00 0..00 00.00 cmmz 0 acoEumoch mu «00. «00. «.0. >0...nm00co 00..mui0 0 0 0 sovomcm mo moocmoo 00.0 00.0 00.0 o:.m>ih 00.0. 00.00 00.00 coLco 0cmvcwum :0.00 00.00 00..0 co.um.>ou 0cmvcmum 00..0 00.00 00.00 cmoE oococoww.0 2.0. 8;: 00.8 00.0 00:0 2.0 tote 285$ 00.:: 00.00 00.00 00.00 00.00 :0.0. co.um.>ov weavcmum 00.00. 00.00 00.00. 00.0: 00..0. 00.0: cmoz . peoEumock amouum00 umouocm “mouumom “mouocm umouumoa ummuocm 0 030cm 0 030cm . 030cc .mp.c:10 mo c00532i1 ummuiu "mc.u_L3 m_cmc0..;01|.0.: 0.0mk IIIIIIIIIIIIIIIIIIIIIIIIIIIIllll----—-———___.______ 152 As discussed earlier in this chapter, analyses Of the data reveal that there is no significant correlation for the children in this study, between reading comprehension, as measured by a standard- ized reading test, and either of the holistic measures of the chil- drenis writing, or the number of T-units written. Blocking the sUbJects by reading scores, therefore, did not diminish the within- the children in Treatment 2. The pre- and posttest stories written by the children were evaluated holistically/by two highly qualified, trained, independent raters using a primary trait scoring instrument. The criteria on which the chilchxflfls writing were judged were the same criteria used to evaluate the professional literature read to the children. Each series read to the children contained all of the literary traits, but devel— oped them to a greater or lesser degree. The raters participated in intensive training both in the philosophy and the procedures of primary trait scoring, a sophisticated form of holistic evaluation. They were instructed to respond to each piece of writing as a unified whole and, using Part I of the primary trait instrument (Appendix A), to evaluate the writerksfacility in developing particular literary characteris- tics. To assure a uniform rating of all stories, the writing samples, 48 pretest stories and 48 posttest stories, were randomly mixed and then evaluated by the raters, who thOUth them to be samples of writing by 96 different children. 'The scores given by the two raters on each trait were averaged. Those averages for the five traits were then added to obtain an overall score for literary quality. After the data had been prepared, they were submitted with the ratings on genre development to a multivariate analysis of covariance using the posttest scores as dependent variables and the pretest scores as covariates. ‘The results indicated an interaction effect due to differences between Treatment 1 and Treatment 2 for different reading groups. A comparison of the cell means showed that while all three 165 groups of readers hearing the highly rated literature did better than the corresponding groups hearing the less sophisticated literature, the effect was the greatest for the lowest readers. The lowest readers, those with less exposure to the written word, showed the most improve- ment in the literary quality of their writing. The best readers, those who had already read more, did not show as much improvement in the literary quality of their writing as did the readers with less experi- ence. The results of the MANCOVA are supported by the two-sample Median Test and by the Mann-Whitney U Test, both of which use the median rather than the mean as the measure of central tendency. Both tests found Treatment l significant over Treatment 2 on the posttest (p < .05), and both found a significant difference in favor of Treat- ment 1 for the lowest readers in the study, p < .01 and p < .05, respectively. A separate MANCOVA using the five scores for the traits of literary quality showed Treatment l to be superior to Treatment 2 at ther.04 level. 0f the five literary traits measured, four of them, plot development, characterization, setting, and style» improved more when the children heard the literature rated the most highly on these traits. Of these four, plot and character development were the most affected. In discussing plot, Wells (l913, 1929) in her now—classic book Ihe Technique of the Mystery story maintained that the writing of detective stories is both easy and difficult since the structure of the plot is always the same-~the problem and its solution “L 39%. Valin (l981) argued, therefore, that it is what a writer does within this 1; L 166 formula restriction that determines the quality of the story (p. l9). In this study, the literature read aloud in Treatment l was determined to be superior to that read in Treatment 2 in terms of plot develop- ment. It exceeded the basic expectations of adequate structure and exhibited originality, freshness, and vitality. It is this quality, the ability to go beyond the formula, that the children in Treatment l reflected in their writing. These children were also better able to create characters who were fully developed and convincingly real within their assigned roles than the children in Treatment 2. In a book entitled writing EOQQJQE Eiction, Koontz (l972) explained that the development of characters, not the plot or setting, is the most impor- tant element in a detective story because it is the characters, with their individual motivations, who hold the key to the puzzle~(p. 103). The children who heard the literature in which the characters were better developed created more fully developed characters in their own writing than did the children who heard the literature in which the characters were less well developed. The development of the mood of the story showed greater improvement in the writing of those children who heard the less sophis- ticated literature. In a classical detective story, the mood is created when closure is assured; closure is assured when the mental astuteness of the detective is established. The children in Treat- ment 2 also did better than those in Treatment 1 in establishing the mental ability of the detective. Since they did not, however, develop their characters as well as the children in Treatment l, it appears that 167 they may have established their main character as a stereotypical detective, including the dimension of intelligence without developing the stereotype into a well-rounded character. ‘This would have satis- fied the formula demand for closure and established the mood of the story, resulting in higher scores on mood in the ratings. In their development of rounded characters, it is quite possible that the children in Treatment 1 neglected to address specifically'or to establish the particular qualities of mental astuteness in the character of the detective. In failing to establish his/her mental astuteness, they would have failed to guarantee closure, which is dependent on this particular characteristic of the detective, and thus have failed to adequately establish the mood. The overall consistency of the results of this study in favor of Treatment 1 provides strong evidence that middle- to upper-middle- class fifth-grade children of all reading abilities who hear classical detective stories of high quality read aloud to them write detective stories of better literary quality than children who hear less sophis— ticated stories from this genre. The effect is the greatest for the lowest readers. Random assignment of the children in this study permits the generalization of these findings to all children from similar populations. Research Question 2 Do children who hear high-quality literature in which the traits of the genre are well developed obtain higher scores, when rated on a primary trait analysis of their own writing within that genre, than children who hear literature in which those traits are less well developed? '5: 1; 168 The answer to this question was determined by the same holistic rating procedure used to answer number one. As the raters evaluated the childrenhs writing for literary quality, they also rated it for development of the characteristics of the classical detective genre using Part II of the primary trait instrument (Appendix A). After the data had been prepared, they were submitted with the ratings on literary quality to aImJltivariate analysis.of'covariance using the posttest scores as dependent variables and the pretest scores as covariates. iAs noted above, the results indicate an inter- action effect due to differences between Treatment 1 and Treatment 2 in relation to reading ability. ‘The effect of Treatment l, hearing the literature rated higher in quality read aloud, was especially great on the writing of the highest reading group. A comparison of the cell means shows that the very good readers hearing the highly rated litera- ture developed the traits of the genre in their own writing to a much greater extent than any other group. Apparently, when these readers heard and discussed the literature in which the traits of the genre were well developed, they were able to detect the qualities that made it better and then to incorporate these qualities into their own writing. The mean for Group 2 in Treatment l was higher than the mean of its corresponding group in Treatment 2; it was, in fact, higher than the mean for Group l in Treatment 2. The mean for Group 3 in Treatment 1 also exceeded the means for both Groups 1 and 2 in Treatment 2. It did not exceed the mean of its corresponding group in Treatment 2, 169 however, since the lowest readers in Treatment 2 (those hearing the less highly rated literature) received higher scores on genre development than any group except the best readers in Treatment 1. They did not, however, come close to matching the performance of the good readers who heard the sophisticated literature. It appears that while hearing literature in which the traits of a genre are well devel- oped may help all readers develop those traits in their own writing, the effect is especially great for very good readers. 'These readers, who have had the most experience with the written word, were the most able to derive benefit from the sophisticated literature in terms of develOping the traits of the genre in their own writing. Low readers, although they do quite well when hearing the highly rated literature, receive even greater help when they hear less SOphisticated literature read aloud. For these readers, hearing literature of high quality positively affected the literary quality of their writing, but perhaps because they had had less experience with the written word, the com- plexity of the more sophisticated literature appeared to overshadow the underlying structure of the genre. While the better readers were able to detect the structure and benefit from its complex elaboration in the more sophisticated literature, poorer readers needed to hear a detec- tive story in which the treatment of the traits was less sophisticated if they were to develop these traits and integrate them into their own writing. The consistency of the results summarized above indicates that hearing literature in which the traits of the genre are well developed i . F I .3 *1: V 1 __ 170 affects the development of those traits in the children's own writing, but the nature of that effect differs depending on reading ability. There is strong evidence that hearing wel l-crafted classical detective stories read aloud does result in the writing of better classical detective stories by readers above the national average. ‘There is also evidence, however, that readers of average and below-average ability may be more successful in developing the traits of a genre in their own writing, at least in their initial attempts at such writing, if the literature they hear read aloud is less sophisticated. A separate MANCOVA using the ten scores for the traits of the classical detective genre shows Treatment l to be superior to Treatment 2 at the .005 level. The children who heard the more sophisticated literature did significantly better in developing the qualities of the genre in their own writing than the children who heard the less sophis- ticated literaturet ‘The MANCOVA also performs a univariate F test for each factor individually, thus comparing the effects of Treatment l and Treatment 2 on that item without the interaction of the other nine. Of the ten traits measured, the univariate F tests showed that the follow- ing were better developed by the children who heard the literature rated highest in development of genre traits: ~-the creation of a puzzle story in which a normal situation is disrupted and returned to normal by a major character who solves the problem through intelligence, using logic and deduction together with clues known and/or discovered (number 6); --the development of a setting that reflects the real world, but is removed from it to the extent that events and actions in the story are not subjected to the standards of plausibility and probability used in the real world (number 10); 171 -—the development and presentation of clues that are integral to the story, can be seen as contributing to the solution once that solution is known, yet are unobtrusive on a first reading (number 12): -—the development of a solution to the crime which is a logical outcome of the clues presented in the story when seen from the appropriate perspective (number l4); --the development of a story ending which involves a resolution of the disruption, including an explanation of the solution (number l5). The univariate F tests showed that the traits below were more affected by Treatment 2: --the adequate establishment of the detective as mentally astute (number 7); --the development of characters' actions which are consistent with their designated roles (number 8); --the development of a plot which centers on the detectiveks investigation, solution of the crime, explanation of the solution, and final resolution of the situation (number 9); —-the presentation of the crime as a background for the story and a reason for the deductive activity of the detective(s) but not as the focus of the story (number ll); —-the development of suspense, in spite of the formula restriction that things will turn out all right, by creating uncertainty about the fate of a character (number l3). The adjusted means, parameter coefficients, standard errors, and significance levels for all ten traits are reported in Tables 4.4 and 4.5. It is interesting to compare the first five traits, which were better developed in the writing of the children who heard the litera- ture rated higher in development of genre traits, with the second five, which were more affected by the literature in which the traits of the _, Mr. ,vViflmnnnllelwfln . i. . , flirt We“. linflidmlui Ii {in 172 genre were less well developed. The first five traits focus on underlying structure rather than detail, since they involve the devel— opment of elements vital to the cohesion of the story; they also appear to be more related to critical thinking abilities than the second five. The first five involve abstract thinking through the use of logic, four directly through the use of clues and their explanation, the fifth through the creation of a setting that must meet two sets of criteria. In developing the first five traits, the writer gives evidence of critical thinking ability by developing those traits in his/her writing in such a way as to maintain the interest of his/her audience while at the same time challenging that audience, itself, to think critically. The following are specific examples of critical thinking abilities (Cianciolo & Le Pere, 1969) as they are developed through the traits of the classical detective genre: l. Fact and opinion: The writer presents his/her material in such a way that the reader must decide what is fact, what is informa- tion, and what is opinion, as well as deciding which of these is relevant to the problem. 2. Cause and effect: In presenting his/her material, the writer cloaks that material in such a way that the cause (clues), while clear if the effect (solution) is known, is elusive if that effect is not known. 3. Using information to support or disprove a hypothesis: The writer intentionally creates a story that appears to have no solution, 173 but which, in the end, will be solved using the very information pro- vided in the story itself. 4. Interpretation of implied ideas: The writer leads the reader to accept a world, which by implication is similar to the real world, but in fact does not meet the standards upon which the real world is based; she/he may also attempt to lead the reader away from the actual solution by presenting material designed to fool the reader into accepting implied rather than stated evidence. The second set of five traits seems much different from the first. These five focus on aspects of the detective formula, those things that are necessary for a story to be defined as a detective story, but not necessarily important to the quality of that story. Many of these appear to involve a linear process of thinking that might be realized quite mechanically. The establishment of the detective's mental astuteness, for example, merely requires some statement to that effect; it would be enhanced by, but does not require, character development. None of these five traits necessitates the use of logic or evaluation necessary in higher levels of critical thinking as the first five do. Even the development of a plot containing the four aspects identified in trait nine might be accomplished through the use of sequence rather than causation. These observations seem quite in keeping with the characteristics of the literature read aloud to the two treatment groups. The Literary Series read to the children in Treatment l was determined by qualified readers to be significantly superior on all traits to the Popular Series read in Treatment 2. Its 174 development of genre traits, as well as its development of literary traits, was much more complex and sophisticated. The complexity and sophistication of the Literary Series seemed to be a factor in the intensity of interest the children exhib- ited in the novels read. In‘Treatment l the interest of the children seemed to remain high throughout the treatment period. In contrast, the interest of the children in Treatment 2 seemed to diminish and to waver; .As time went on, they exhibited less interest when the litera- ture was read aloud, acting almost as if they had heard it before, which in fact they had since all detective plots are the same and their series did not go as much beyond the basic formula structure to exhibit freshness, vitality, and originality as did the series read in Treat- ment 1. Research Question 3 Are identified atomistic characteristics of the literature read aloud to children, i.eu.‘language fluency as measured by the number of T-units, syntactic complexity as measured by the number of words per T-unit, and vocabulary maturity as measured by a weighted vocabulary score, reflected in their writing? A multivariate analysis of covariance, again using the posttest scores of the measures as dependent variables and the pretest measures as covariates, indicated that the covariance between the pre- and posttests was very high and that there were no significant differences between Treatment l and Treatment 2 on any of these measures. The literature read aloud to the children in Treatment 1 contained fewer T-units per hundred words and those T-units contained more words than the literature read aloud to the children in Treatment 2 (Tablee342h a; g 175 As explained in Chapter III, the measure of vocabulary quality is somewhat convoluted, with the literature read in Treatment 1 containing more mature vocabulary, and the literature read in Treatment 2 contain- ing somewhat less mature vocabulary, but containing more of it (Table 3.3). The children's writing did not show significant gains in any of these measures for either treatment. When considered together with the children‘s improvement in literary quality of writing, these results appear to be consistent with the findings of Hildyard and Olson (l982L Hildyard and Olson found that children attend to different things when they listen than they do when they read. When they listen, children attend to information necessary for the coherence of the main story theme and for the introduction of main events. When reading, however, children pay attention to all specified details, including incidental ones. Two boys in this study unwittingly attested to the validity of Hildyard and Olsonksfindings in this note they wrote during class and inadvertently left behind: "Bruce, let's write notes like the girls!" "O.K.ll" "Do you ever pay attention to the story?" "Noll" When the boys returned to the classroom to gather up their coats for recess, the researcher asked the boys what they meant by not paying attention. Although quite embarrassed, as might be expected, they candidly explained that they listened to the story but didnH:"pay attention to the details!‘ Further discussion revealed, however, that 176 the boys understood the story and remembered the information necessary to its coherence. It may be that the "not paying attention" to which they referred reflected what Hildyard and Olson measured. Perhaps the atomistic aspects of writing are included in the details to which listeners do not attend. Perhaps this is part of what the boys missed when they did not "pay attention." If this is so, it is possible that atomistic aspects of writing are more influenced, in the absence of intervention, by seeing a text than by hearing it. In a preliminary study, Eckhoff (1983) found that, as early as second grade, childrenus writing reflects the linguistic structures (including number of words per T-unit, format, and stylistic features) of the basal text to which they are assigned. It appears that if atomistic measures of writing are to be influenced by listening to literature read aloud, childrenks attention wil l have to be focused on them. ‘The teacher will need to highlight examples of the effective use of these aspects of writing by both professional and child authors and, in so doing, bring these aspects of writing into the chilcfls conscious awareness, thus enabling the child to hear them when she/he listens to literature. Research Qgestioh 4 Does improvement in writing correlate with type of treatment for children of different reading abilities, as identified by a standardized reading test, or is the writing of children with different reading abilities affected differently? Is the writing of children with low reading ability particularly affected when literature of high quality is read to them? The results of this study indicate that a difference does exist between the effects on childrewfls writing of hearing literature of high ii .L 177 quality and the effects of hearing less sophisticated literature. The nature of that effect, however, differs across reading abilities as measured by standardized tests. In terms of literary quality, the difference between Treatment 1 and Treatment 2 is the greatest for the lowest readers, although the cell means for literary quality of writing are higher for all groups in Treatment l than for the corresponding groups in Treatment 2. Improvement in developing the traits of the genre, however, was greatest for the best readers in Treatment l. Hearing the literature rated the highest in development of genre traits apparently enables good readers to realize more fully those traits in their own writing than if they hear less sophisticated literature. Since hearing the wel l-developed literature also results in significant improvement on those genre traits that involveaabstract thinking and higher-level cognitive skills, it appears that.thelcomplexity of this literaturelcomplements and/or stimulates higher-level thinking pro— cesses. It is interesting, however, that the lowest group of readers developed the genre traits better if they heard the less sophisticated literature» Perhaps these children found the genre characteristics easier to identify in the less sophisticated literature. (M'it may be that the structure of the detective story, as they perceived it, was more clearly reinforced by the less sophisticated literature and thus permitted them to write well in a style that was already familiar to them (Roskelly, l984L. Since it may be surmised that the lowest read- ers had less exposure to literature, it seems feasible that the less complex literature may have helped them identify and isolate the 178 narrative structure of the genre (Billman, l984L. The fact that these same readers developed the genre traits to a lesser degree but still quite well when hearing the more complicated literature suggests that these readers might benefit the most from initial exposure to the structure of a genre in relatively unsophisticated literature if they are going to write stories in that genre. Once that structure is recognized, however, literature of higher quality should be offered, both because of its positive influence on literary quality and because of its apparent stimulation of critical thinking abilities. Additional Findings In addition to the findings noted above, a two-tailed t-test indicated that the writing of all groups in the study improved sig- nificantly in literary quality at the .05 level or better (Table 4.6). While the findings of this study provide strong evidence that children who hear and discuss literature rated high in literary quality write stories of higher literary quality than those who hear and discuss literature rated lower in literary quality, both groups who heard literature read aloud and discussed that literature improved signifi- cantly in the literary quality of their writing. This reaffirms the importance of hearing literature in the classroom for childrenk; writing. The number of T-units written by all groups also increased; the difference between the pre- and posttest scores is significant at the .05 level or better for five of the six groups, meaning that over repeated trials, such improvement would occur by chance only five out 179 of lOO times. For Group 2 in Treatment 2, the significance level was p < .09 (Table 4.7). Although the children spent much more time on their second stories than they spent on their first ones, the children were permitted to write on both of their stories until they were fin— ished. The second time, they wrote much more. Unlike the initial stories, the second stories were written according to a structured teaching plan that asked each writer to design a cohesive story by specifically focusing on an overview of the story and then on charac- ters, plot, setting, and conclusion. In spite of the fact that the children were less interested in writing a second detective story, and in spite of considerable resistance to the task of preplanning, their second stories contained significantly more T-units, reflecting a greater number of ideas expressed, and hence an increased fluency with language than the initial stories. This corroborates the value of using a structured method in the classroom of helping children to develop an overview of their story as well as to design the specifics before they begin to write. A regression analysis indicated that before the treatment, the correlation between reading ability, as measured by the Gates- MacGinitie reading comprehension test, and the writing of the children, when measured holistically, was very low for the subjects in this study. This discrepancy in reading and writing measures is contrary to what would be expected in light of existing research, which has pre- dicted a strong relationship among all of the language arts (Chomsky, T972; Evanechko et a1., l975; Loban, l963, l976), and contrary to what 180 was anticipated before this research. It is important to recognize that the results of this study might have been very different had the children been grouped by writing ability, rather than by tested reading ability. Loban (1963, 1976) found that there is a high correlation between reading and writing. The discrepancy between his finding and the finding of the present study may be due in part to the fact that the writing of the children in Loban%s(l976) study was evaluated primarily for rhetorical and mechanical features, ixm, organization, syntax, vocabulary, sense of audience, and mechanics (pp. 25-26L According to Lloyd-Jones (l977), the rating of rhetorical features often appears to be holistic because rating scales seem to apply to the whole discourse. They are, in fact, however, "isolated from the con— text by the reader and scored separately-~an atomistic system tied to abstract categories associated with traditional rhetorics" (p. 35). In this study, the childrenwswriting was evaluated holistically. It was edited for spelling, punctuation, and minor usage transgressions before rating so that such mechanical errors would not interfere with the raters' judgments. It is this holistic rating that does not corre- late with the reading scores. ‘There is a correlation between the atomistic measures of writing, ixm, words per T-unit and quality of vocabulary, and the reading scores, indicating that children who score high on standardized reading tests have a command of both content and rhetorical features of writing, whereas children who score lower have stories to tell and the ability to tell them, butckanot effectively 181 command the rhetorical and/or mechanical skills necessary to communicate effectively in writing, especially'when those skills are measured out of context. The findings in this study showing lack of correlation between reading and writing scores also contradict the findings of Evanechko et al. (l975), who found that the number of T-units written by a child was the best predictor of the child's reading comprehension score. In this study, the number of T-units written by the child on the pretest had the lowest correlation with reading comprehension of all five writing measures examined. While the children in the research by Evanechko et al. had a 20-minute time limit and an assigned topic, the children in this study had no time limit and an assigned genre, but free choice of topic. In light of the results of this study together with Lobanus findings, it may be that assignment of a topic together with a rela— tively short period for writing provides an advantage for good readers, who appear to have greater command of the technical aspects of writing, and a lesser advantage for poorer readers who do not, resulting in a high correlation between reading comprehension scores and T-units in writing. Implications Since the children in this research were randomly assigned to treatments, the results of this study may be generalized to all children from similar populations. Lbing the classical detective genre as the vehicle for research, this study found strong evidence that the literary quality of childrenkanarrative writing improves when they main. Hummi..l,a .ufl n, l by hi . .1 L hear narrative writing of high literary quality, especially if those children have'scored low on a standardized reading test. Again using the detective genre, this study also provides strong evidence that when stories from a particular genre of childrenhs literature are read aloud to children, the traits of that genre are better developed in the children's writing, particularly the writing of good readers, when they hear high-quality literature in which the traits of the genre are well developed. While children with average and below-average reading abilities (as determined by a standardized reading test) also do well in developing genre traits in their own writing when they hear such literature, they do even better in developing those traits when the literature they hear is of good literary quality but is not quite as sophisticated. A separate finding, however, indicates the quality of the literature heard also affected the particular genre traits that were developed. Results indicated that all children who heard the more sophisticated literature (high, middle, and low readers) developed in their writing those traits that involved more extensive application of critical thinking abilities. It appears, therefore, that both literary quality and critical thinking abilities are stimulated by hearing high- quality literature read aloud. Poorer readers, however, develop the traits of the genre to a greater extent when they hear the less complex and sophisticated literature. ‘While a less sophisticated example might be used to introduce and help these readers become familiar with the traits and thus the underlying structure of a genre, 183 results of this study indicate that it is imperative, both for the literary quality of children's writing and for the development of critical thinking abilities through that writing, that teachers read aloud literature of excellent quality with its inherent sophistication. Findings from this research do support the hypothesis that children with lower ability in reading will benefit the most from hearing high quality literature read aloud in the classroom. ‘The children in this sample were from middle- to upper—middle-class homes, which traditionally have a high literacy orientation. It was assumed that children from such a population who do not score high on stan- dardized reading tests would still have the literacy background sup- portive of high achievement in all of the language arts. lflzwas hypothesized, therefore, that exposure to quality literature through hearing that literature read aloud in the classroom would result in more improvement for these children, who had less exposure to the written word, than for children with better reading ability and more exposure to the written word. Findings from the study show that while the literary quality of stories written by all three reading groups hearing the highly rated literature improved more than the correspond- ing groups in Treatment 2, the greatest amount of improvement was indeed demonstrated by the lowest readers. A second important finding regarding readers with lower ability indicates that they show marked improvement in developing the traits of the genre in their writing, more than any group except the highest readers in Treatment l, if they hear less sophisticated literature. 'This suggests that readers of 184 average reading ability and below need exposure to a variety of literature of varying degrees of sophistication, at least during their initial attempts to write in a genre. Since it is precisely as they attempt to write in a different style of discourse that they appear to lose control of rhetorical aspects of writing and to mix aspects of text construction appropriate to different types of discourse (Keech, 1984), exposure to high-quality literature is certainly important. If, however, children are attempting to write in a particular genre, shar- ing a less sophisticated example with them, at least initially, appears to be important. As the children become familiar with the narrative structure of the genre, sophistication of the literature could be increased, thus maintaining the quality of genre development, increas— ing the literary quality of their writing, and stimulating the develop- ment of critical thinking abilities through writing. The results of this study show that no significant gains in T-units or words per T-unit could be attributed to either treatment. It is possible, however, that significant differences on the atomistic measures in this study might have been found if the children had been able to participate effectively in the revision process. For many children, the intensity of the drafting process, of getting the infor- mation down and creating the story, seemed to preclude attention to how that story was presented. Since the content of writing was so impor- tant, form was rarely attended to until the telling was finished, if at all. It appears that lengthening t-units when that is appropriate and 185 better vocabulary choices are more likely to be accomplished during the revision or editing process. Although not affected by treatment, the number of T-units did increase for all groups in the study. The increase for five groups was significant at p < .05, for the sixth, Group 2 in Treatment 2, at p < .09. Although the children worked on their posttest for a much longer time than they did on their pretest stories, this finding is still surprising since the children were permitted to write on both their pretest and posttest stories until they were finished. The posttest was their second story in the same genre, and the children were not as excited about writing it as they were their first one. The fact that they had so much more to write in the posttest story indi- cates that a structured program to help children gain an overview of their story as well as to develop the story in detail before actual writing commences results in the expression of more ideas. When com- bined with the benefits of hearing high-quality literature read aloud, the result is increased fluency in use of language and increased liter- ary quality of narrative writing. As with the Teunit measures, the maturity of vocabulary used by the children during this study was not significantly affected by the quality of the literature they heard read aloud. It may be that maturity of vocabulary is an aspect of writing that develops only in classrooms where the use of effective vocabulary is expressly encour- aged. Research has indicated that hearing childrenhs literature read aloud, when combined with follow-up activities, does have an effect on 186 the reading vocabulary of children (Cohen, l968h. Such exposure also seems to affect childrtwfls speaking vocabulary, but again, only when oral activities accompany that listening (Strickland, 1971). Although unusual or antiquated vocabulary was frequently defined for therchil- dren in this study, no additional attempt was made to focus on vocabu- lary development. In a related study that examined the effects of reading childrenus‘literature aloud on the writing of third graders, Michener (1985) also failed to find gains in vocabulary. It may be that an increase in the use of mature vocabulary in writing, at least in the elementary school, requires discussion, including a specific focus on such use and an atmosphere that not only supports but inten- tionally encourages such growth. Establishing such an atmosphere might also increase the support of the peer group, thus making the use of mature and/or unique vocabulary acceptable in the classroom. ‘This researcher does not feel that the children in this study lacked the knowledge of or the ability to use mature vocabulary, nor does she feel that the children avoided using such vocabulary for reasons such as spelling difficulty. In seems rather that they, as individuals and as a group, did not recognize the power of effective vocabulary. Thus they simply did not realize the need to use it, nor had they developed the habit of doing so. This indicates that the use of mature vocabu- lary may develop with an increasing sense of authorship as well as the support of the classroom group. 187 Discnssion and Qbservations The analysis of the data obtained during this study provides strong support for the hypothesis that reading sophisticated literature of high quality aloud in the classroom has a positive effect on the narrative writing of'middle- to upper-class fifth-grade children of all reading abilities. The literary quality of writing improved signifi- cantly for all six groups in the study. It improved the most, however, for those children who heard the more sophisticated literature. It is important to note that both of the detective series used in this study were reviewed in professional journals indicating that both were well- crafted stories. Since series detective stories are not held in high regard by many professionals in the field of childrenhs literature, they are seldom reviewed in highly respected professional journals. Those that are reviewed constitute the best literature of its kind available» The series used in Treatment l received consistently higher reviews than the series read aloud in Treatment 2. In addition, the readers who rated the series using the primary trait instrument, both professionals in the field of children's literature, evaluated the series used in Treatment l very high as literature. Both argued, however, that that very quality, with its sophisticated style, it use of difficult and archaic vocabulary, and its historical English set- ting, made it too sophisticated for most fifth graders to read. They felt, based on their experience with children of this age, that the second series was much more appropriate. It is important to realize, however, that what children can understand and appreciate does not 188 necessarily correspond with what they are able or choose to read. It is the opinion of this researcher, after reading aloud from both series for an extended period of time, that the more sophisticated literature challenged the interest of the children to a greater extent than the Popular Series. ‘This was manifested in differences of behavior and response. Whereas the attention of the group hearing the sophisticated literature remained consistentthroughout the nine-week period, the attention of the children hearing the popular literature seemed to decrease. It appeared that they did not need to pay attention to know what was going to happen; the formula was clear and the outcome was known. They did not need to concentrate to follow the story. The complexity of the literary series, however, seemed to command the attention of the children to a greater extent. ‘The intricacy of the story development went beyond the confines of the formula and demanded continuing involvement if one was to maintain the continuity of the story. Their interest remained high as they continued to match wits with the author. The writing of the professional literature and the writing of the<:hildren were evaluated on how well they developed the traits of the classical detective story. Discussion of the stories with the children focused on these traits. Many of the children, however, wrotera much more general adventure-type story involving sequence but little ratiocination. Many of the mystery/detective stories with which children are familiar, both from television and from their own reading, are, in fact, adventure rather than detective stories g“. 189 (Billman, 1984; Cawelti, 1976). This researcher feels that the appeal of these high-adventure stories had a strong effect on the stories the children developed. The preoccupation with the exposure to such stories is particularly evident in the children's writing about chase scenes and in their frequent reference to expensive sports cars so prevalent in many television pseudo-detective series. It was difficult during the relatively short time of this study to narrow the childrenks focus from a general detective/adventure story to the much more spe— cific criteria of the classical detective genre. In a classroom it might be much more effective to have the children isolate the traits of the genre as they know it from their reading or from television, and then to compare the traits they have identified to the traits of the theoretical genre. Once the children recognize the similarities and differences, they could create their own stories using either set of traits or a combination of both. Another finding of this research is that the reading ability of these children as measured by a standardized test was not correlated with their writing when that writing was evaluated holistically. This simply means that in this study, low readers and high readers wrote equally good stories. These stories were rated (for content only) by two professional teachers, both high experienced in evaluating children's writing, both involved in advanced-degree programs in lan- guage arts and childrtwfls literature. The children wrote their stories as part of the regular classroom curriculum over a long period of time. Thus they produced language in a relatively normal environment, not as % ' 13:; 190 a part of a testing situation. It seems that many children who are identified by standardized tests as being poor readers actually have the ability to control language when they are in command of the language situation. This researcher feels that schools should seri- ously consider using writing, measured holistically for content not mechanics, in addition to adequate measures of reading in assessing children"; language competence. Donaldson (1978), in her book thldnen's Minds, lends support to this recommendation in her descrip- tion of the differences between the receptive and productive uses of language: "When you produce language you are in control; you need talk only about what you choose to talk about. . . . Listeners and readers are not in control in the same way" “L 74). The overpowering emphasis in society on teaching children to read may have suffered by its own emphasis on receptive language and its failure to recognize the impor- tance, for many children, of productive language in the goal of gaining total language competence. Since no correlation was found between the childrenhs writing and the reading scores used to determine the groups (blocks) for this study, the blocking effect, used to increase the possibility of finding differences that do actually exist, had very little effect. In spite of this, the results of the multivariate analysis were significant at the p < [)1 level; tests for the median strongly supported the findings of the MANCOVA. ‘This researcher feels that strong attention should be paid to the finding of no correlation between reading as measured by a reading comprehension test and the quality of writing as measured ‘11—”,— _ 191 holistically. Based on research studies such as Loban's, one would expect at least a moderate correlation for a sample such as the one in this study. That such a correlation was not found is a matter for further research. It may be that significant correlations do exist between reading as measured by a standardized test and other types of discourse. Perhaps such correlations would be found in samples from other socioeconomic populations. This finding does indicate, however, that educators must refrain from assuming that children who do well on reading tests will also write well, or that those who do poorly on such é tests will also write poorly. The findings of this research do not support such an assumption. The children in this study quickly became accustomed to the routine of listening for the first half of the period. It appeared that there was a difference, however, between the quality of listening as the reading began each period and the quality of listening as the reading progressed. Although the children were usually anxious to have the researcher begin reading, the noise and activity level were much higher at the beginning of the reading than they were after approxi- mately ten minutes had elapsed. Since children in the upper grades listen to literature read aloud relatively infrequently, this might indicate that children need more help in focusing their attention on the listening task than was included in the design of this study. After ten minutes, however, the literature appeared to have captured the children's attention. The room was quiet, the activity level low, the interest generally high. 'This condition persisted for :5: a 192 approximately 15 minutes, after which attention began to wane and other activity again increased. It may be, therefore, that there is a mini- mum amount of time that must be devoted to reading aloud to this age writer if optimum benefits for writing are to be derived. This study indicates that a minimum of ten minutes per session, when that reading is done daily, is necessary. The children in this study were interested in, even excited about, writing their first story. The assignment to write a second story in the same genre, however, did not receive as warm a reception. In spite of a district-wide emphasis on writing across the curriculum, the children did not complain about the daily story writing procedure. They did complain that they did not want to write another detective story. Although all of the children completed their second stories, the lack of excitement over writing in the same genre probably had an effect on the quality of the final stories. In a classroom where the writing of a pretest would not be necessary, the procedure used in this study, discussing literature from the perspective of the author to discover how the traits of a genre are realized, would result in even more significant results than these children demonstrated. Because the stories were typed by the researcher, the children had no need to recopy their stories. This benefited the childrenks writing in two ways. First, the children were free to write as much as they wished. Comments and questions from the children during the drafting process indicated that this freed many of them to write more extensively. For example, several of the children asked a second time . 11.1,... . llfluvxh é __ L 193 if they would have to recopy their stories. Assuring them that they would not have to do so, the researcher asked why they wanted to know. One child replied he wanted to know how much to write, and others nodded in agreement. Second, since the typing was to be done by the researcher without the presence of the authors, the children were alerted to the importance of accuracy in their final editing process. This was initially'impressed on the 49094 cm mN NH H; n N mm NH Z N m N N NH 3 Hm m N wH N. N. g m N mH «H 0 mm m N N NH OH Hm m N I H H 9. m N H 3 NH mm m N R H H mm N N ow H H HN. N N mm NN NH om N N N NH «H mm N N N mH HH ow N N mm mN mN mm N N N H «H a... N N «N H HH No N N N 3 NH mm H N 0H N.H m wN. H N HN N «H mm H N Nm NH HH 2. H N Nm mH ON mm H N mv mN mH Hm H N mN mH oH ms H N mN NH 0H ms H N N NN S 3 m H N ON 3 mm m H I m a N m H Hm NN «N we m H on N «H mm m H mm 0N N.H me m H N H 3 NH. m H NH 3 NH mm m H 3 H NH 2 N H HN «H «H Nw N H N N NH ms N H H N NH 3 N H mm H: H mm N H H HH m HN. N H Hm 0N NH mm N H 3 NH HH MN N H S «N N.H mm H H mm H NN E. H H on H m mm H H mm N 3 as H H mm N H em H H NH mN N.H S. H H NN H 0 mm H H in N NH 3 H H 212.95 3292.4 .3505 zoummsao macaw stems HAZE .Hmm.wemom\fimm.wmmm HNMQIOVDI‘QGN OZIMm