THE EFFECTS OF SENTENCE-COMBINING CURRICULA ON DESCRIPTIVE WRITING PERFORMANCE, READING ACHIEVEMENT, AND WRITING APPREHENSION: A STUDY WITH HIGH-RISK SECONDARY STUDENTS BY KAY WESTALL GONSOULIN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1993
ACKNOWLEDGEMENTS I would like to express my appreciation to Dr. Ruthellen Crews who guided me through the Ed.S. program and through a major portion of my doctoral study. She has been a true mentor. I am extremely grateful to Dr. H. Thompson Fillmer who graciously took over as chair of my committee at Dr. Crews' retirement. His expertise, guidance, and suggestions have been of immeasurable help. I appreciate his professionalism, his sincere interest in my work, and his friendship over these many years. I also wish to thank Dr. David Miller whose expertise in statistical matters has helped me greatly, and Dr. Edward Turner and Dr. Robert Wright whose longstanding friendships and professional advice have been most rewarding and supportive. For their computer help and support in preparing this manuscript, I would like to thank Don and Rhonda Royston. Their assistance was absolutely invaluable. To my husband's mother and stepfather, Sarah and Ted Shivik, I wish to express my thanks for their support and help. To my sisters, Marilyn and Vernice, I express my appreciation for their encouragement and support. Words cannot express my love and gratitude to husband, Hank, and our sons, Taylor and Derek, without whose love and unfailing ii
support this project would not have been completed. Finally, I wish to express my love and appreciation to my parents, Viola and James Westall, for it was in their dedication and sacrifice that this dream was born. This dissertation is dedicated in loving memory to my sister, Janice V. McNab, and her husband, James D. McNab, who were killed in the Amtrak derailment September 22, 1993. May I continue their tradition of excellence and caring in education and in the service of others. 111
TABLE OF CONTENTS ACKNOWLEDGMENTS ii ABSTRACT vi CHAPTERS I INTRODUCTION 1 Statement of the Problem 1 Scope of the Study 2 Significance of the Study 2 Assumptions 13 Definitions 14 Hypotheses 17 Procedures 18 II REVIEW OF LITERATURE 22 Effects of Syntactic Knowledge on Reading and/ or Writing Performance 22 Research on Experimental Programs to Teach Syntactic Knowledge Through Sentence-Combining Activities 30 Research on Apprehension Toward Writing 67 III METHODOLOGY 81 Setting of the Study 81 Instrumentation 83 Regular Reading Instruction 99 Experimental Writing Curricula 101 IV ANALYSIS OF THE DATA 116 Review of the Study 116 Findings Related to the Hypotheses 119 V SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 134 Summary 134 Implications and Conclusions 136 Recommendations for Future Research 140
APPENDIX. REFERENCES 143 150 BIOGRAPHICAL SKETCH 163
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE EFFECTS OF SENTENCE-COMBINING CURRICULA ON DESCRIPTIVE WRITING PERFORMANCE READING ACHIEVEMENT, AND WRITING APPREHENSION: A STUDY WITH HIGH-RISK SECONDARY STUDENTS By Kay Westall Gonsoulin December 1993 Chairperson: H. Thompson Fillmer Major Department: Instruction and Curriculum This study was undertaken to investigate the effects of instruction in sentence combining on the reading achievement, writing performance, and writing apprehension of 54 high-risk students ages 15-19. Students were randomly assigned by classes to one of three groups: sentence combining with nominal/modifying phrases at the sentence level and regular reading program, regular reading curriculum only, or sentence combining with nominal clauses and factors leading to coherent writing at the paragraph level and regular reading program. The nominal phrases group and the nominal clauses group received 24 sentence-combining lessons of 50 minutes each, 8 regular reading lessons of 50 minutes each, and 8 periods of
50 minutes each of selected reading over the 8-week treatment. The regular reading group served as the control group and received no sentence-combining training. For 4 out of 5 days they were assigned to instruction and practice in the Random House High Intensity reading system and 1 day out of 5 to selected reading. These effects were measured by scores from the Metropolitan Achievement Test, 1979 ed. Advanced Level 1, Reading Survey, and by scores of descriptive writing samples scored using the guidelines of the National Assessment of Educational Progress Writing Assessment, and by scores from the Writing Apprehension Survey. Students were tested before and after the sentencecombining curricula. Results were analyzed using the SAS package with analysis of covariance. The results of the study indicated a significant effect for 1 of the 13 writing factors: nominal phrases. The nominal/modifying group performed better than the nominal clause group or the control group. There were no significant treatment differences on the holistic writing factor nor on any of the reading factors. There were also no differences on the writing apprehension factor due to treatment. There was a treatment effect on the total numbers of words, or fluency factor, which was significant at .05 (actual p=<.003). The group using nominal clauses with factors leading to coherent writing moved from 12 6 words to 164 words per writing.
CHAPTER I INTRODUCTION As the requirements for graduation from high school have increased in the state of Florida, the drop-out rate, especially of high-risk or nontraditional students, has risen. Students in this category tend to drop out for many reasons; one of these reasons appears to be lack of academic success due to poor reading and writing skills and apprehension over courses which have even moderate writing requirements. Teachers concerned about this situation are searching for effective ways to assist students in acquiring the skills and the confidence they need to stay in school and graduate. Statement of the Problem The primary purpose of this study was to investigate the effects of instruction in sentence combining on the reading and writing achievement of high-risk secondary students. These effects were measured by scores from the reading section of the Metropolitan Achievement Test (MAT) Advanced Level 1, Form KS (Prescott et al., 1979), and by scores of descriptive writing samples scored using the
2 guidelines of the National Assessment of Educational Progress Writing Assessment (1980). In addition, the researcher sought to determine which of two types of sentence-combining instruction, (a) nominal clauses and factors leading to coherent writing or (b) nominal/modifying phrases, produces the better student writing performance and reading achievement scores. Finally, the researcher sought to determine if instruction in sentence combining had an effect on the writing apprehension of the students as measured by the Writing Apprehension Survey (Daly & Miller, 1975a) Scope of the Study Participants in this study consisted of 54 students enrolled in grades 10-12 in a remedial reading course in a secondary school in north central Florida. The students in the study scored in stanines 1, 2, or 3 on the reading section of the MAT (Prescott et al. 1979). The results of the study can only be generalized to students who are similarly identified as to reading ability. Because students in this study were not randomly assigned to the classes which were used for the experimental and control groups, the design of the study was quasi-experimental; therefore, causal inferences cannot be safely made. Significance of the Study Reading and writing courses for high-risk students have been viewed as places where these students can be drilled on
3 discrete skills or knowledge of terminology that they have not mastered. As the number of these students in high school has risen, textbook publishers and curriculum planners have maintained the more is better philosophy and have retained traditional methods of instruction, focusing instruction on English grammar for writing improvement and isolated skills in decoding and comprehension for reading improvement John Goodlad (198 3) noted that reading and language art classrooms could be characterized as, "preoccupied with recall of specific and isolated elements and information served up in workbook and guizlike formats" (p. 18) Applebee et al. (1990) noted in their report on the National Assessment of Educational Progress from 1988 that students in low-ability classes were more likely than those in highability classes to have teachers who gave exercises that focused on the mechanics of writing (p. 8) There are several possible reasons for the failure of current writing programs for high-risk students in high school. First, there is the lack of transfer from drill on grammar to the actual use of correct grammatical structures in writing. Students do not make the connection between isolated exercises and actual writing. Early in this century, Symonds (1931) found that mere repetition of grammatical rules or use of sentences containing grammatical errors to be corrected and the analysis of sentences with
4 errors were not as effective as having students choose the correct form without regard to grammatical rules. Symonds (1931) also found that having students recognize correct and incorrect forms was more effective than having students learn grammar rules. Many studies since Symonds, in fact, over 2 00 (Elley, Barham, Lamb, & Wyllie, 1976) have substantiated his findings, but the view that instruction in grammar improves writing persists. More recently Hillocks (1986) documented this persistent practice of grammatical study under the guise of writing improvement. Hillock's study showed that the study of grammar is irrelevant to composition improvement and steals instructional time from classroom activities more directly related to writing. Zemelman and Daniels (1988) chronicled the fact that many secondary teachers show little confidence that students can learn the proper use of mechanics from practicing writing. Zemelman and Daniels also noted that these teachers spend much time and money on teaching the subskills which would supposedly aid writing. These teachers utilize valuable time separately teaching spelling, grammar, usage, and mechanics which could be spent more wisely in the practice of writing (p. 9). In fact, the National Council of Teachers of English Task Force on Teaching English to the Disadvantaged (Corbin & Crosby, 19 65) documented the fact that it is common
5 practice in the English programs of many secondary schools to teach, "formal grammar concerned with abstractions and terminology ... to students who are least likely to succeed with it" (p. 138) The Task Force concluded that, "the time devoted to teaching formal grammar to disadvantaged students can be better spent in other areas of English instruction" (p. 138) Second, much of the instructional material is written at too low an interest level or too high a skill level for high-risk students. Materials appear aimed at either elementary school or college markets. Better material needs to be marketed for the high-risk secondary student. Third, high-risk secondary students lack exposure to advanced or mature syntactic models of writing resulting in lack of knowledge of such models. Extensive emphasis on mechanical errors prevents students from developing strategies for improving their writing skills and gaining confidence in themselves as writers. Hillock's (1986) findings from research emphasized that focusing on grammar and mechanics takes time away from composition. Zemelman and Daniel's (1988) also contended that the time spent in grammar drill would be spent much better in sentencecombining exercises. Sentence-combining exercises are those exercises in which students combine several shorter sentences into longer, more complex ones. Zemelman and
6 Daniels concluded that sentence combining is successful in improving composition for the following reason: Sentence combining works because it is a productive, not an analytic, activity; it works by helping students to tap into their oral language base, their latent knowledge of many complex sentence types, and then to practice transferring this oral knowledge to writing. (P27) Finally, as Shanahan said in his book Reading and Writing Together (1990), reading and writing should be integrated in instruction: Too often we have not understood how best to combine reading and writing because we simply have not appreciated the full breadth and power of literacy. We have been too willing to see reading and writing as discrete school activities with important but rather limited outcomes and values. However, reading and writing by their very nature extend well beyond the schoolhouse. They are a basic part of the fabric of society. Reading and writing have a role in our feelings of power and self-esteem; they are entailed in our views of fairness and justice; they are moral, social, aesthetic, and intellectual processes. Unless we fully appreciate the scope and power of literacy, we cannot hope to unite reading and writing in ways that will fulfill our dreams, (p. 3) Theoretical Significance Research into the relationships among the language arts has important implications for high-risk learners. Interest in what the relationships are among the language arts, especially between reading and writing, may yield valuable information that addresses the problems of high-risk learners such as poor reading skills and poor writing performance. These problems need to be further researched.
7 In the past few years, researchers have become interested in the relationship between performance in reading and writing. Researchers have suggested that reading and writing might be related in areas such as metacognition, vocabulary meaning, syntax, and organizational style ( Langer, 1986; Rubin & Hansen, 1984; Stotsky, 1983). Birnbaum (1981) found that good readers thought differently during revision than did poor readers. Good and poor readers looked for different kinds of information. Poor readers were more detail conscious during reading and writing, making many of their decisions at the sentence level instead of at the paragraph level. Shanahan and Lomax (1988) have identified what they consider the components of the model for the relationship between reading and writing. In their model, reading knowledge is made up of three parts or variables: word analysis, word meanings and text comprehension. Their model identifies the components of writing to be the following: spelling, vocabulary diversity, syntactic complexity, and story structure complexity (p. 199) In Shanahan and Lomax s (1988) interactive model information can move both directions from reading to writing and from writing to reading. In this model," the influence of writing upon reading takes place across discourse levels, with lower level writing knowledge contributing to reading knowledge at the next higher level of discourse" (p. 199)
8 Writing, then, has an effect on the element of reading directly above it in the discourse hierarchy and it has an impact on the higher level reading elements (p. 199). Shanahan and Lomax (1988) conclude their argument for an interactive model of reading and writing based on their research in the following statement: Reading influences writing and writing influences reading: theories of literacy development need to emphasize both of these characteristics. Similarly, these findings suggest that reading and writing should be taught in ways that maximize the possibility of using information drawn from both reading and writing. These results challenge the wisdom of instructional programs that provide students with limited opportunities to gain knowledge through writing, or that delay these opportunities until reading proficiency is well developed. Significant knowledge transfer takes place in both directions, even at relatively low levels of literacy attainment (p. 2 08) One of the factors that has been found to connect reading and writing performance is syntactic knowledge or maturity, that is, knowledge of the grammatical or sentence rules of the language (Langer, 1986; Rubin and Hansen, 1984; Shanhan, 1980; Stotsky, 1983). The rate of acquisition of syntactic structures varies widely according to variables such as sex, intelligence, experience, and socioeconomic status (Loban, 1970) In the final report of his 13-year study, Loban (1976) noted that students who were superior in one language process tended to be superior in the other language processes and that
9 students who were deficient in one language process tended to be low in the other language processes as well. One might assume that students such as those in this study who have been identified as poor readers would be poor in writing skills also. A specific syntactically-oriented writing approach, sentence combining, was mentioned by Stotsky (197 5) in her review of research on this strategy in which she stated, "explicit syntactic manipulation in writing as a means of improving reading comprehension may derive from the possibility that such exercises clarify both the meaning and the use of complex structures for children" (p. 33). In her review of 1983, Stotsky reiterated her view that sentencecombining activities are of value to children. According to Stotsky, these sentence-building exercises would help students understand syntactic relationships within sentences and would develop students' ability to synthesize larger numbers of elements within a total idea. Sentence-combining research has yielded promising results which might be applied to the instruction of highrisk students. O'Hare (1973) reported that sentencecombining instruction improved the syntactic maturity of 7th-grade students to a level commensurate with the syntactic maturity of the 12th-grade students who participated in Hunt's study (1965). Thus, similar instruction might result in achievement of more mature
10 scores with remedial students whose range of reading scores approximate those of the seventh-grade students in Hare s study (1973) Further support for syntactic programs was presented in the review of developments in the comprehension of text structures over the past 2 years by Pearson and Camperell (1981) who noted that work in linguistic connectives and sentence-combining techniques seems to suggest that attention to cohesion rather than atomization, or grammatical analysis, of sentence elements is more fruitful. These findings do not support the earlier work in transformational grammar in which researchers taught grammar terminology and sought to improve syntactic maturity by direct instruction in grammar and analysis of transformational patterns. The teaching of cohesive devices was suggested in the correlational study of the relationship between reading and writing performance by Hill (1982) She found five factors which were meaningful in measuring the writing performance of the 7thand 8th-grade students in her study. These five factors included the following: (a) writing mechanics (agreement of subject and verb, punctuation, spelling, and capitalization; (b) writing quality (holistic, paragraph coherence, and nominal clauses) ; (c) adverb clause/coordination; (d) nominal/modifying phrases; and (e) relative clauses.
11 Three of these five factors were significant predictors of the total reading achievement score. These factors were the following: (a) writing mechanics; (b) writing guality; and (c) nominal/modifying phrases. The data analyzed consisted of descriptive writing samples. As the studies previously mentioned had not supported mechanical aspects as being as helpful as cohesive devices in producing more mature syntactic patterns in writing, this researcher perceived the need for research which would test syntactic factors found to be a link between reading and writing without direct instruction in mechanics. In the implication section of her study, Hill (1982) called for experimental instructional programs which could be tested before and after the instruction to see if materials based on these factors would indeed improve the reading comprehension and writing performance of students. Hill (1982) also suggested that the focus of such instruction be first of all coherence and factors leading to coherence and then attention to mechanical aspects as used in revision. Thus, as sentence combining appears to have possibilities for increasing reading comprehension (Stotsky, 1975, 1983) and for improving written syntactic maturity (Hunt, 1965; Pearson & Camperell, 1981; Zemelman and Daniels, 1988) this study may aid in determining if
12 instruction in these linking factors can indeed improve reading comprehension and writing performance. Educational Significance There are several reasons why this study is educationally significant. First, there is a need for new instructional strategies for high-risk secondary students in both reading and writing. Traditional methods of isolated drill on discrete skills do not produce transfer to actual reading and writing tasks. Studies have shown that students work better when they attend to utilizing cohesive devices and synthesis rather than analysis of syntactic structures (Pearson & Camperell, 1981) These studies support the necessity of teaching students to write coherent structures and to use other strategies by which they may become successful writers. Second, there is a need for research in sentence combining at the secondary level. Many more studies have been done with elementary students. Perhaps, if more studies were conducted with secondary students, empirical data could be compiled to be analyzed as Neville and Searls did in their 1993 meta-analytic review of sentence-combining research. Unless more research is done, data will not be available for comparison. There is a particular lack of studies at the high school level with high-risk students. If sentence combining is beneficial in those formative years in elementary school before all of a child's language
13 abilities are developed as reported by Neville and Searls (1991), then perhaps the high-risk secondary learner whose literacy skills are also not fully developed could benefit. In fact, Evans (1986) states these are the very students he found in his studies who may benefit from sentence-combining exercises the most. Finally, there is a need to find writing activities to reduce writing apprehension, a form of communication disorder marked by reluctance to write or to take courses which are perceived as requiring much writing. It is no wonder that students have been apprehensive in courses that require them to constantly drill in skills and terminology in which they have already proven to be deficient. More flexible ways to approach writing are needed. This research may document a writing technique that may help to reduce student apprehension over writing. Assumptions In this study the following assumptions were made: 1. Scores on the Metropolitan Achievement Test, Reading Survey section, Advanced Level 1 (1979) represented each student's reading ability. 2. The descriptive writing samples elicited from participants represented each student's writing ability. 3. The Writing Apprehension Survey (Daly & Miller, 1975a) scores represented each student's apprehension about writing.
14 4. The writing instruction that students in the study received in their English classes did not include direct sentence-combining instruction. Definition of Terms For purpose of this study the following terms were used as defined: Reading refers to a student's performance on the reading tasks as measured by the Metropolitan Achievement Test, Reading Survey section, 1979 edition, collected before and after the treatment. Writing refers to a student's performance on two descriptive writing samples, collected before and after the treatment, scored according to the National Assessment of Educational Progress Writing Assessment guidelines (1980) Writing gualitv is an overall measure of a student's writing performance which included a holistic score, a measure of paragraph coherence, and the presence of modifying clauses. Clause refers to a group of words which contains a subject and a verb and expresses a complete thought. Dependent clauses depend on the independent clause for meaning and completeness. Phrase refers to a group of words, such as into the woods inside the room running down the street and to count the money.
15 High-risk students are those defined in this study as scoring in stanines 1,2, or 3 on the MAT, Reading Survey section, 1979 edition. These students are deficient in reading skills and for that reason often do not achieve well in school and suffer risk of failure and/or dropping out of school. Other terms found in the literature for these students have been, "disabled readers," "remedial readers," and "disadvantaged readers." Modifying clauses are clauses that modify nouns or occasionally, complete sentences including clauses of time, place and manner. Modifying phrases are restrictive and nonrestrictive phrases directly describing the nouns they modify, or appositives (some "of" phrases) Nominal clauses are clauses used as subjects, direct objects, subject complements or objects of prepositions. Nominal phrases are phrases used as subjects, direct objects, subject complements or objects of prepositions T-Unit is a term coined by Hunt (1965) to identify a main clause with all its attendant modifying words, phrases, and dependent clauses. For example, "The girl is holding a large jar with fireflies inside it that look like butterflies, and she is wondering whether she should change her mind and let them go," (Mullis & Mellon, 1980, p. 4). This example contains two "T-units" separated at the comma.
16 Syntactic knowledge is defined as the knowledge of the ways in which words are put together to form phrases, clauses, and sentences. Descriptive writing refers to writing which requires the writer to describe or tell about feelings, observations, or experiences. Sentence combining refers in writing to the expansion of sentence elements into gradually more complex structures by combining several simpler sentences into one or more mature sentence. Embedding refers to the combining of simple sentences into a more complex sentence, one with more clauses and/or phrases, by placing elements within the sentence. Open Exercises refers to sentence-combining exercises which may elicit several ways or alternatives in the process of combination. Signaled exercises refers to sentence-combining exercises which have only one solution which is signaled by a cue or a signal alerting the writer as to the manner in which the sentence should be combined. Holistic scoring refers to scoring which evaluates a piece of writing as to overall quality of writing performance by rating one set of papers against models from that set. This is also called "general impression" marking. Coherence refers to the way facts are ordered or arranged in writing.
17 Cohesion refers to the way facts are gathered together in writing. Modification refers to the process of describing by the addition of words, phrases, or clauses. Coordination refers to the combining of similar elements into pairs of a series. In sentence combining, two sentences can be combined by compounding the subjects and giving them a common predicate (verb or verb phrase with modifiers) Subordination refers to a way of reducing one sentence to a clause or a phrase so that it can be included as part of another sentence. Hypotheses The following hypotheses, stated in the null form, were tested in this study. Follow-up analyses were conducted when statistical significance was found. All hypotheses were tested at the .05 level of significance. Hypothesis I: After controlling for initial reading achievement as measured by the subscores and total reading score of the MAT, Reading Survey section (1979), there are no significant differences in reading achievement among subjects using two experimental writing curricula and those who are not. Hypothesis II: After controlling for initial performance in descriptive writing as measured by the sample ratings using the scoring procedures of the National
13 Assessment of Educational Progress Writing Assessment guidelines (1980), there are no significant differences in writing performance among subjects using two experimental writing curricula and those who are not. Hypothesis III: After controlling for initial apprehension as measured by the Writing Apprehension Survey (Daly & Miller, 1975a) there are no significant differences in writing apprehension among subjects using two experimental writing curricula and those who are not. Procedures Experimental Writing Curricula Two sentence-combining programs were written, one at the sentence level emphasizing nominal/modifying phrases and one concentrating on nominal clauses and factors leading to coherent writing at the paragraph level. The content, sequence, and amount of time devoted to each of the lessons was determined by the researcher after careful study of instructional programs used in sentence-combining research and books and articles by recognized experts in the field of composition, especially sentence combining. The Study Collection of the data The study sample consisted of 54 high-risk secondary students in grades 10-12 placed on the basis of low reading scores (stanines 1, 2, or 3 on the MAT, Reading Survey) in a
19 remedial reading course at a local public high school. Subjects were randomly assigned by classes to one of three conditions. Group 1, Experimental B, contained 18 subjects who received sentence-combining instruction with emphasis on nominal/modifying phrases at the sentence level in addition to the regular reading program. Group 2, the control group, contained 16 subjects and was instructed using a reading management system only. Group 3, Experimental A, contained 2 subjects and received sentence-combining instruction with nominal clauses and factors leading to coherent writing at the paragraph level in addition to the regular reading program. Scores on the MAT, Reading Survey section, Advanced Level 1 (1979 edition) were obtained for all subjects. The subscores and the total reading score were used to test for equivalency of the groups. They also served as covariates in the statistical analyses. Scores on the descriptive writing samples from the National Assessment of Educational Progress Writing Assessment (1980) "Describe Something" were collected for all subjects. A total of 13 scores derived from that analysis of the writing samples were used to test for equivalency of the groups and also served in the statistical analyses. Over an eight-week period, the two experimental groups completed 24 sentence-combining lessons, 8 lessons in the reading management system, and 8 periods of selected
20 reading. The control group completed 3 2 lessons in the reading management system and 8 periods of selected reading. At the end of the study all subjects were again administered the MAT, Reading Survey section (1979) another descriptive writing sample and the WAS (Daly & Miller, 1975a) Design and statistical analysis The quasi-experimental design used in this study was the nonrandomized control group, pretest-posttest design. Group membership (sentence combining with nominal phrases and reading program, reading program alone, and sentence combining with nominal clauses) served as the grouping factor. Pretest scores were used in analyses of covariance to partially adjust for nonequivalence of groups and to control initial knowledge or attitude. Separate analyses of covariance were conducted on the following dependent variables: reading achievement, writing performance, and apprehension about writing. The reading subscores and total reading scores were tabulated by the MAT publishers. The preand postdescriptive writing samples, entitled "Describe Something," from the NAEP Writing Assessment were analyzed syntactically by the researcher following the NAEP guidelines as outlined by Mullis and Mellon (1980) The writing apprehension scores were calculated by the researcher using the guidelines outlined by Daly and Miller (1975b) The data were analyzed using a packaged program from Statistical
21 Analysis System (SAS) (Helwig, 1978) When the null hypothesis was rejected, follow-up comparisons were calculated by the researcher.
CHAPTER II REVIEW OF THE LITERATURE The review of the literature includes the following sections: investigations on the relationships between reading and writing and knowledge of syntactic structures; research on experimental programs designed to teach syntactic knowledge through sentence-combining activities; and research on the apprehension of students about writing. Effects of Syntactic Knowledge on Reading and/or Writing Performance Research in cognitive psychology, psycholinguistics, and sociology has brought renewed interest in the interrelatedness of performance in the language arts. Of particular interest is the relationship between reading and writing performance. Some of the research focuses on the process whereby readers and writers become proficient. These researchers see the reader and writer as actively processing language and constructing meaning. Loban (1976) began his 13-year study in 1963 with a sample of 3 88 kindergarten children. Collecting information on performance in listening, speaking, reading, and writing, Loban followed these children through the 12th grade. Reading ability was measured by the Stanford Achievement Test. Writing ability was measured beginning in grade three by taking samples of writing about a picture. The writing samples were separated into five groups ranging from primitive to superior. Results revealed the superior group 22
23 in writing was also the highest group in reading achievement and teacher rating. Loban also noted that all writers ranked superior were reading above chronological age and all writers ranked primitive were reading below chronological age. Loban also found that the relationship between reading and writing performance became more prominent as the subjects grew older. Evanechko, Ollila, and Armstrong (1974) used samples of students' writing in their comparison of reading achievement on 8 reading subtests and 13 measures of syntactic maturity in 118 6th-grade students. The reading measure was the Bond-Balow-Hoyt New Developmental Reading Test, Intermediate Level. In the same week, students wrote a story on an assigned topic. A modified version of Botel and Granowsky's Formula for Measuring Syntactic Complexity was used to analyze the writing samples. The researchers suggested that both reading and writing require certain common language skills and that the presence of these skills should result in better performance in both reading and writing. The best combination of language measures to predict reading achievement was determined by a regression analysis. The regression analysis revealed that only 4 out of 13 language measures were significant predictors of reading achievement. Two predictors that were consistently first in the regression equation were the number of communication units,
24 a fluency of expression, and "two count structures," which is a measure complexity of sophistication. Kuntz (1975) compared the scores of 96 7th graders on the Gates-Mac Ginitie Reading Test on vocabulary, comprehension, and speed with their scores on the Falk Sentence Construction Test for completeness, pattern transformation, and precision transformation. Kuntz found a significant (p=<.00l) correlation between total syntactic attainment and total reading achievement. This led her to conclude there was a close and reliable relationship between reading achievement and written syntactic achievement. In another study with similar aims, Takahashi (1975) compared the comprehension of written syntactic structures by good and poor readers with their scores on a standardized reading test. Based on scores obtained on the Nelson-Denny Reading Test, she selected a total of 3 good and poor 9thgrade readers and 18 good 6th-grade readers. Scores on the Marcus Test of Sentence Meaning were used to measure syntactic knowledge. Takahashi reported significant differences (p=<.001) between the means of the two groups of 9th graders but no significant differences between those means and the 6th graders' means. She concluded that the 6th graders were processing sentence-by-sentence and the 9th graders were relying heavily on content and redundancy. Three studies were found to predict a student's reading score from writing performance. Lazdowski (1976) attempted
25 to predict a writer's reading level from a writing sample. He analyzed 338 compositions written by students in grades 7 -14. Standardized reading test scores ranged from below grade 3 to above grade 14 Academic or school grade levels were not considered in the study except in locating students within a wide range of reading ability. The writing samples were categorized according to reading ability grade levels without regard to the actual academic or school grade level. Readability was measured according to six readability formulas. Lazdowski analyzed samples for vocabulary load, sentence structure and density of ideas, and syntactic complexity. The study revealed a positive relationship between reading and writing ability. Based on the results of his experiment, Lazdowski developed a regression formula with which it is possible to predict, within one grade level, reading ability of a writer from a writing sample with a reliability of .87. Another regression formula for prediction, in this case reading achievement, was developed by D'Angelo (1981). A sample of 245 9th-grade students was randomly selected and categorized as possessing high, average, or low ability on the basis of the Otis-Lennon Mental Ability Test. Also administered were the Brown-Carlsen Listening Comprehension Test, the Listening to Remember Specification Test, and the Informative Writing Scale, Advanced Middle School Level.
26 After the administration and scoring of the various instruments, the researcher obtained correlation coefficients to determine the relationships between reading achievement and the independent variables: mental ability, listening memory, listening comprehension, and informative writing ability. The researcher used multiple regression to obtain predictive formulae, and a step-wise technique to obtain the best predictors. The results of the formulae were compared with actual reading scores to determine the effectiveness of each formula. All correlation coefficients for the total sample population were significant (p=<.01) with the Otis-Lennon Mental Ability Test having the largest correlation with reading: .67. However, none of the multiple regression formulae which were found to be most effective used the mental ability variable. Listening comprehension and listening memory were more effective predictors of reading ability than informative writing. The final study attempting to predict reading scores found by this researcher was conducted by Hartman (1984) Her purpose was to examine the predictive relationship of five qualitative and five quantitative measures of writing identified from previous research on the reading comprehension of 116 9th-grade students. Reading comprehension was measured by the Degrees of Reading Power (DRP) test. The raw scores from the DRP were used to place
27 subjects into high and low reading groups. The five qualitative variables rated holistically were organization of presentation, interpretation of meaning, emphasis of ideas, sentence agreement /control, and sentence structure/usage. The five quantitative variables, judged by the counting of syntactic structures, were total number of words, total number of T-units mean T-unit length, clauses per T-units, and subordination index. The assigned writing task was the retelling of a short narrative selection. Pearson Product-Moment correlations and step-wise regression analyses were used to analyze the data. Correlation coefficients indicated that significant (.05) relationships existed between DRP and seven of the ten writing variables for the total group. For the high readers, five variables were significantly related to DRP, and for the low readers, two writing variables were significantly related. Step-wise regression analyses produced the best equations for the prediction of reading comprehension from the writing sample. For the total group, emphasis of ideas and sentence structure/usage significantly (.05) predicted DRP. Sentence structure/usage and interpretation of meaning were significant predictors of DRP for high readers, while emphasis of ideas significantly predicted DRP for low readers. Birnbaum (1981) in her ethnographic study, investigated the reading and composing behaviors of four
28 4th-grade students and four 7th-grade students. Her selection of a small sample was intentional as she wanted to collect extensive data on each student. Each student was videotaped three times while selecting, reading silently, and discussing texts written in different modes and three times while composing silently in different modes and discussing the process. The students were audiotaped while reading and while composing aloud and observed in class for 3 5 hours. The students and their teachers and parents were interviewed concerning the students experiences with written language. Data were also collected from academic records. When reflectiveness was shown to be a characteristic of proficient reading and writing, the researcher administered the Matching Familiar Figures Test. Students rated more proficient in reading were also rated more proficient in writing and demonstrated more control over oral language. The less proficient writers used fewer criteria for selection of topic and often were slowed down with excessive attention to the surface features of the language. The findings suggest a model for development of proficiency in written language. Hill (1982) investigated the relationship between reading performance of 99 7th-grade and 8th-grade students on the Metropolitan Achievement Test, 1979 edition, reading comprehension scores and ratings on descriptive writing
29 samples scored by procedures of the National Assessment of Educational Progress Writing Assessment guidelines (NAEP, 1980) The following subscores of the MAT were collected: vocabulary, literal specific, literal global, inferential specific, inferential global, and evaluative. The descriptive writing samples were assessed to produce scores on four major gualities: holistic, cohesion, mechanics, and syntax. Factor analysis and multiple regression analysis were used to determine the relationship between writing performance and reading performance. The independent variables were 13 writing variables determined from the NAEP Writing Assessment scores. The dependent variables were the six reading subscores and a total reading score determined from the MAT. Five factors were found to be meaningful in measuring the writing performance of the subjects in the study. These five factors were the following: (a) writing mechanics (agreement of subject and verb, punctuation, and spelling) ; (b) writing guality, overall quality, paragraph coherence, and presence of nominal clauses; (c) adverb clause/coordination; (d) nominal/modifying phrases; and (e) relative clauses. Of these five factors, three were significant predictors of the total MAT reading score. These factors were writing mechanics, writing quality, and nominal/modifying phrases.
30 In summary, the studies reviewed in this section included samples from middle or junior high school and high school studies by Loban (1976) Evanecko et al. (1974) Kuntz (1975), and D'Angelo (1981) who found that a student's reading ability and achievement could be predicted from a writing sample. Takahashi (1975) found that the readingwriting relationship changes as students grow older. Hill (1982) found three factors-writing mechanics, writing quality, and nominal/modifying phrases-which were significant predictors of the total reading comprehension score. Robertson (1966) and Stoodt (1970) found that knowledge of connectives and conjunctions respectively were related to reading comprehension. There are some inconsistencies in the literature presently, but there is support for the notion that performance in reading and writing is related and that the relationship changes over time. Research on Experimental Programs Designed to Teach Syntactic Knowledge Through Sentence-Combining Activities Research on the interrelatedness of reading and writing performance indicates the existence of underlying common factors. One of these factors is syntactic knowledge. Interest in syntactic structures began with the research in generative-transformational grammar by Chomsky (1957) Chomsky theorized that there are two levels of language, the "surface structure" and the "deep structure." The surface
31 structure is a linear sequence of linguistic units which reflects one or more underlying propositions (a deep structure). According to Chomsky's theory, the deep structure propositions are transformed into surfacestructure sentences, based upon the language user's internalized system of grammatical rules. Although Chomsky's work provided the theoretical basis for sentence combining, an instructional method for enhancing syntactic fluency in written composition, it was Kellogg Hunt's first study (1965) that provided the impetus for much of the subsequent research in sentence combining. Hunt's development of the T-unit (a minimum terminable unitsmallest unit of meaning) as a measure of syntactic knowledge gave researchers an instrument for measuring syntactic growth from grade to grade and between ability groups. He was looking in his first study for developmental trends in the frequency of various grammatical structures written by 18 average students at each of the following three grade levels: 4, 8, and 12. Each student produced a 1,000 word writing sample; this writing was compared against professional adult writing. Hunt analyzed sentence length, clause length, the subordination ratio, kinds of subordinate clauses, and within clause structures. He found that the length of the T-unit was tied closely to maturity in writing performance. He also found that there was a limit to the amount of T-unit
32 expansion that occurred through the addition of subordinate clauses to any one sentence. Further syntactic growth was achieved by increasing the number of nonclause optional elements within one of the T-unit clauses. Words per T-unit was the best index of syntactic maturity; second was clause length; third was clauses per T-unit (the subordination ratio) ; and fourth was sentence length. Hunt's later work (1970) dealt with controlling the subject that the students in the study wrote about. He devised an instrument, a passage on aluminum, consisting of 32 short, choppy sentences of connected discourse that 50 students each from grades 4, 6, 8, 10, and 12 were asked to join together and rewrite in a better way. This later work confirmed results from his earlier study. By using a controlled instrument, Hunt was able to conclude that writers were influenced by their skill, not just by what they had to say. Syntactic maturity consisted mainly in the ability to make many embeddings per clause. Several studies were reviewed that built on the work of Hunt by using the Tunit to measure growth in writing after an experimental program in manipulating syntactic structures was conducted. Only one study was found in the literature that attempted to increase reading performance by teaching students to recognize syntactic patterns. Rinne (1967) attempted to increase the reading comprehension of 13 of 26 high school remedial English classes by teaching the
33 students in those classes to recognize sentence patterns in written communication. A Pattern Awareness Test (PAT) was written using transitive, intransitive, and linking verb structures in simple sentences. A Sentence Comprehension Test (SCT) was written to test literal comprehension in reading at the sentence level. These tests were administrated as pretests to all students in the 2 6 classes. The experimental groups, which were randomly selected, were given 5 weeks' training in pattern awareness supplementary to their regular curriculum. Control classes received no such training in grammar. The PAT and SCT posttests were given to experimental and control groups, and an analysis was completed to determine if the patterns awareness training improved sentence comprehension and if there was a correlation between the PAT and SCT. The correlation was significant ( .001) in favor of the experimental group. Reading comprehension as measured at the sentence level was not improved by the experimental treatment. Bateman and Zidonis (1964) studied the effects of a generative grammar upon the writing performance of 50 ninth graders in a university school in the Midwest. Two teachers taught these two classes and both groups received the regular curriculum. The experimental class studied, in addition to the regular curriculum, specially adapted materials from the area of generative grammar. Written compositions were collected from both classes during the
34 first three months of the first year and from the last three months of the second year. Both teachers met regularly with the investigators to standardize their writing assignments. The investigators developed an analytic instrument to assess the guality of the sentences in the samples used in the analysis. According to the generative grammar theory on which this study was based, there are two kinds of sentences in English: kernel and complex. Transformational rules are used to produce all sentences other than kernel sentences. The 46 transformational rules were of the following types: embedding, conjoining, deleting, and simple. The 46 grammatical structures were instrumental in the sentence evaluation techniques developed in this project, for they provided various scores for measuring the grammatical changes occurring over the two-year period in the students' writing. It was, so the researchers felt, the identification of these structures in the writing, not the students' awareness of ability to recall the labels, that was being measured. The Sentence Evaluation Techniques (SET) that were devised by the researchers consisted of three scores: the Structural Complexity Score (SCS) the Proportion of Well-Formed Sentences (PWF) and the Error Change Score (ECS) The SCS indicated the grammatical or structural richness of the sentences produced in the experiment, the PWF was the ratio of acceptable to unacceptable sentences, and the ECS denoted the frequency
35 and kinds of grammatical errors that happened in student writing. Statistical analysis failed to show a significant difference between the experimental and the control group. Mellon (19 69) developed a "transformational sentence combining" curriculum to correct methodological errors he felt had been made in the study by Bateman and Zidonis (1964). Mellon contended that it was the sentence-combining practice of the Bateman and Zidonis study, not the learning of grammatical rules, that accounted for the greater maturity in the students' free writing. Mellon' s subjects were 2 00 7th-grade students in urban, private, and suburban schools. Five control classes were taught traditional grammar. Two placebo classes studied no grammar at all but had extra lessons in literature. Five experimental classes were taught transformational grammar for 1 year during which they practiced sentence-combining exercises for 5 months. Results indicated that the writing produced by the experimental group showed significant increases in all 12 factors. O'Hare (1973) replicated Mellon' s curriculum but improved the methodology. O'Hare disposed of the transformational grammar terminology and used cue words for each sentence-combining transformation. O'Hare also controlled for teacher variable; he and another teacher both taught experimental and control groups. The sample consisted of 8 3 7th-grade students in 4 classes. Writing
36 was the same for both experimental and control groups with the exception that the experimental group received the sentence-combining writing activities. Randomly assigned to the experimental treatment, 41 students received one and a quarter hours per week of sentence-combining writing work. Fifty T-units per student from five different free compositions ranging over narrative, descriptive, and expository modes were collected from both pre and post writing samples. 1 Hare's year-end findings included the following: (a) significant (. 001) gains on all measured factors of syntactic maturity for the experimental groups (b) scores for the experimental group's syntactic maturity were similar to Hunt's 1965 12th-grade norms; (c) sex of student was not related to treatment effect; and (d) experimental students with low IQs significantly increased in syntactic maturity. Subsequent studies specifically designed to teach sentence-combining activities to improve written syntactic maturity and reading comprehension produced conflicting results. These studies may be divided into elementary, middle or junior high school, high school, and college level or adult. Fisher (1974) investigated the effects of selected exercises in sentence-combining and embedding based on the transformational grammar theory on the syntactic maturity and reading comprehension of 94 students in grades 5, 7,
37 and 9. At the beginning of the experiment and at the end, all students were asked to rewrite paragraphs consisting of kernel sentences, to complete cloze tests on 4th-, 8th-, and 12th-grade syntactic levels; and to complete a standardized reading test. The treatment consisted of sentence-combining exercises based on 12 transformations being completed by the students. The typical lesson consisted of several exercises, each consisting of two or more active, declarative sentences to be combined by the students. Later in the course, the students were taught to reverse the sentence-combining process so that they could separate a complex sentence into the kernel sentences from which it was composed. A multivariate analysis of covariance, covarying pretests, was used to determine if treatment, grade, and treatment-grade interaction had any effect on the total writing output and the total reading output of the experimental groups as compared to the control groups. An analysis of covariance was performed on each of the six writing factors and on each of the four reading factors. All of the experimental groups scored significantly better (.012) on the total reading output than did the control groups The researcher concluded that the sentence-combining practice contained in the experiment did enable students in all three grades to write more maturely than they did before
33 the experiment. This improvement was not dependent upon intelligence as all students did equally well. The sentence-combining curriculum did not help students to improve their reading comprehension. Another study with 3rd and 5th graders was conducted by Moore (1984) The researcher investigated whether sentencecombining instruction significantly influenced the writing effectiveness of the students. The instruments, which were not listed, reportedly measured syntactic fluency, quality of writing and proficiency in the components of writing. Analysis of variance was used to test 15 hypotheses. Third and 5th grade experimental students scored significantly higher in syntactic fluency than the control group. The 3rdgrade control group, however, scored higher on quality of writing. There was a significant interaction between teaching method and sex of the subjects. Isaacson (1985) sought to determine the potential syntactic development in written expression when students are given varying levels of adult assistance. The subjects were 15 poor writers randomly selected from students in a learning-disabilities resource program and 3 students who scored in the average range on the California Achievement Test and were not enrolled in remedial programs. Both groups were randomly assigned to the experimental group and to a no-assistance group.
39 Students were presented 10 relative clause problems. The instrument used was the written language portion of the Woodcock-Johnson Psychological Battery. Regression analysis revealed that accurate assessment of one's own performance and the ability to report necessary procedures were significantly related to performance even beyond the contribution of age and achievement. Piper (1985) investigated the effects of sentence combining and story expansion delivered by word-processing microcomputers on the written syntactic maturity of 6th graders. She attempted to identify a feasible method of computer-delivered writing instruction. The researcher trained 6th-grade language arts teachers word processing and sentence-combining techniques. She assigned intact groups to one of three treatments for one 50-minute lesson per week for 6 weeks: sentence combining, sentence expansion and microcomputers; sentence combining and computers; and a traditional delivery of sentence-combining lessons. Analysis of covariance and an analysis of writings for embeddings per T-unit and mean T-unit length revealed that the sentence combining and microcomputer delivery produced better results on embeddings per T-unit. Ledbetter (1988) investigated whether 3rd-grade students could learn to combine simple sentences into complex sentences and if sentence-combining sentences had an effect on total reading comprehension. One experimental
40 group of 4 5 students was taught sentence combining and reading comprehension by combining sentences in the dailyreading assignment. Another experimental group of 2 6 students were taught reading comprehension according to the teacher's manual and sentence combining with teachergenerated material separate from the reading material. The remaining 8 7 students formed the control group which studied only reading comprehension in the regular curriculum. Instruments used in the experiment were the Metropolitan Achievement Test, Primary 2, Form L, as pretest and as posttest, and two forms of a sentence-combining skills measure designed by the researcher, one form as a pretest and one as a posttest. The statistical analyses used were analysis of variance, analysis of covariance, and multivariate analyses of variance. These analyses revealed that the two experimental methods were more beneficial for the boys and that overall the experimental group with sentence combining with reading comprehension daily was the best for literal comprehension. A study by Young (1989) conducted with 233 3rd, 4th, and 5th graders combined sentence-combining research with that of free writing to determine if either method singly or the two combined produced the best effect on the writing of the students. The 6-week study focused on expository and narrative writing with the two experimental groups dealing with sentence combining and free writing and sentence
41 combining alone respectively and the control group with no intervention. The researcher used two measures to evaluate the experiment: an objective measure of syntactic maturity and a holistic rating of writing guality. The results revealed that the combination of sentence combining and free writing was the most effective in two areas: expository writing reguiring syntactic manipulation and narrative writing reguiring writing production. An elementary study, in which the researcher Laframboise (1990) sought to include word processing with sentence combining, was conducted with low-level 4th-grade students. The design was quasi-experimental with a pretestposttest hierarchical design. Students within 6 intact classes were randomly assigned to treatment. Dependent variables were (a) reading comprehension, measured by a traditional cloze test and a structural cloze test; and( b) syntactic fluency measured by the mean number of words per T-unit on a free writing sample. Pretest scores on the structural cloze test served as the covariate. The statistical analysis was multivariate analysis of covariance. The results showed that the differences were greater among type of task than between pretest and posttest among treatment groups. There were no significant differences between the three levels of treatment at the. 05 alpha level.
42 The last elementary study reviewed was by Bordonaro (1991) The researcher investigated the use of the John Collins Cumulative Writing Folder Program with sentence combining as a revision strategy to see if 5th graders would improve their writing in six areas: topic development, organization, supporting details, sentence structure, word choice, and mechanics. Both experimental and control groups used the Collins program; only the experimental group used sentence combining as a revision strategy. The writing samples from all groups were analytically scored by independent scorers. The overall writing performance of the experimental group showed significant improvement. The five variables that showed no significant difference were: organization, mechanics, supporting details, sentence structure and word choice. There were seven studies found in the literature that dealt specifically with middle or junior high school students. The first of these was conducted by Combs (1975) who tested the effect of sentence-combining practice as part of a language arts curriculum on the reading and writing performance of 100 7th graders in four classes. These classes, controlled for ability, sex, and teacher effect, followed a single curricular format excepting the sentencecombining exercises completed by the two experimental classes between pretest and posttest.
43 Effects of the sentence-combining treatment on students' writing performance were measured at pretest, posttest, and delayed posttest times. Effects on students' reading performance were measured by pretest and posttest only. A t-test comparison of control and experimental groups indicated that the experimental group wrote compositions at posttest that were (a) syntactically more mature than those of the control group, and (b) syntactically more mature than those they had written at posttest. The differences were significant at .001. A t-test comparison of judgment by teacher-raters showed that the quality of writing of the experimental group was not different from that of the control group at pretest, but significantly different at posttest. The same raters also judged the posttest writing of the experimental group significantly better than its pretest writing. No such differences were found between pretest and posttest writings of the control group. Results from the two reading measures indicated that the treatment produced no significant differences in reading rate nor comprehension scores on the Gates-MacGinitie Test. A t-test comparison of the posttest group means on the comprehension of the specially-constructed reading measure showed that the experimental group scored significantly (p=<.001) higher than the control group.
44 The researcher concluded that results from his experiment made a strong case for using sentence-combining practice as an integral part of a language arts curriculum. The study indicated that sentence-combining practice increases syntactic maturity and guality of writing, but its effects are lessened when instruction and practice are withdrawn (as shown on delayed posttest) The study also provided evidence that sentence-combining practice positively affects reading comprehension as measured by a specially-constructed reading measure. Simmons (1981) investigated whether sentence-combining activities had an effect on the reading and listening comprehension of 7th graders. She also investigated correlations among reading, listening, and mean T-unit length in writing and attitudes. The 12-week study involved 87 students in 4 classrooms. The experimental and control groups were each divided into regular and advanced language arts classes by reading ability. Each of the teachers taught an experimental and a control class. The treatment consisted of one and a half hours per week of open and closed, written and oral sentence-combining exercises as well as cloze exercises. Three measures were used before and after the treatment to measure its effects. Reading achievement was measured by the SRA Achievement Series, Level E, Forms 1 and 2. Listening comprehension was measured by the Stanford Achievement Test, Intermediate
45 Level II, Forms A and B. Syntactic fluency was measured by free writing in the narrative mode. A two-way analysis of variance yielded results showing no significant differences among reading, listening, and mean T-unit length in writing. The students did indicate they enjoyed the sentencecombining exercises, particularly the open and oral aspects of these exercises. Trivelli (1983) sought to determine whether traditional grammar instruction or a transformational-generative, sentence-combining program would improve the reading comprehension and written syntactic maturity of 8th-grade students. Reading comprehension was measured by the Stanford Achievement Test, a grade-level cloze test, and a difficult cloze test controlled for readability and syntactic maturity levels. Syntactic maturity was measured by mean words per T-unit calculated from a 300-word writing sample of three modes of discourse. About 3 hours of class time as well as 8 hours of homework were spent in cued and noncued sentence-combining of traditional grammar lessons. Campbell and Stanley's Solomon Four-Group Design was used for the study with random assignment of classes to one of the four treatment groups. The statistical procedure used to analyze the data was multiple linear regression. The researcher found nonsignificant differences in the scores from the Stanford Achievement Test and the mean words per T-unit. Results
46 indicated significance for written syntactic ability in favor of students using the sentence-combining method of instruction. Also, statistically significant correlations were found between written syntactic maturity and the Stanford Achievement Test in reading comprehension and the difficult cloze test. These correlations were not reported. Finally, correlations between the standardized and the cloze scores were nonsignificant. Neville and Searls (1985) elected to use a 10-week treatment time period for their study of the effect of sentence combining and kernel-identification practice on the syntactic component of 6th-graders' reading comprehension, as measured by a cloze instrument developed by the authors and by two subtests from the norm-referenced Test of Reading Comprehension (TORC) The experimental group completed 8 open sentence-combining exercises, 7 kernel-identification exercises, and 8 cloze exercises. Analysis of covariance was used and results revealed that gains made on immediate posttest scores on the cloze instrument were significant for the experimental group; gains on a 6-week delayed posttest on the cloze instrument were not significant. There was also no significant difference between experimental and comparison groups on the two TORC subtests. The researchers inferred that sentence combining and kernel-identification practice helped the experimental group to comprehend longer, syntactically more
47 complex sentences. The researchers concluded the following and suggested some direction for future research: Sentence combining has been found to be an effective means of improving writing performance; the research evidence for the effect of sentence combining on reading comprehension is not as definite. This study has confirmed earlier research that showed significantly different student performance on cloze tests but not on standardized reading tests. Future research with sentence combining should include criterion measures of both writing and reading, (p. 58) Spilton (1987) compared the effects of individualized language arts to the effects of sentence combining and to the effects of traditional grammar on the syntactic maturity and quality of writing of 8th graders in an English course over 12 weeks. The subjects were 105 students who were divided into three groups: individualized language arts, traditional grammar, and sentence-combining exercises. The design of the experiment was the control group pretest/posttest design. The quality of writing of the students was assessed by an analytic scale. The writing was also measured for T-unit length and clause length. The results showed that the language arts group was superior on T-unit length; the sentence-combining group was superior on clause length and the sentence-combining group was better than the grammar group on T-unit and clause length. There was no difference, however, between the three treatments on overall quality. Reading achievement and writing ability
48 were highly correlated although the correlations were not reported. Wilkinson (1990) studied the effects of sentencecombining practice on the reading and listening comprehension of 165 4th and 8th graders. The researcher used cued and uncued sentence-combining exercises she had developed herself and carried out the experiment for a period of 2 months. The researcher evaluated the study by using standardized and cloze tests with content and structure word deletions which were constructed from three paragraphs with controlled transformations. Univariate and multivariate analyses were used. The covariates were I.Q. and reading achievement. The results led the researcher to conclude that sentence-combining practice should be incorporated into the writing curriculum. The last middle school study reviewed was conducted by Jordan (1992) She examined the effects of sentencecombining instruction on the writing effectiveness of a cross-grouping of 6th-, 7th-, and 8th-grade students. The experimental group received sentence-combining instruction and the control group received no sentence-combining instruction. The researcher was interested if students' syntax would be affected by words per T-unit, words per clause, and clauses per T-unit after exposure to sentencecombining instruction. Analysis of covariance and t-test analyses were employed to test the data gathered on
49 syntactic fluency. Sentence-combining instruction had no significant effect on the syntactic fluency of middle school students when that fluency is measured by words per T-unit and clauses per T-unit. Sentence-combining instruction did have a significant effect when measured by words per clause. When middle school students' writing is compared to the writing of professionals, the students' writing mildly reflects maturity. There were six studies found in the literature which used high school students as subjects. Two of these studies attempted to improve syntactic maturity, writing maturity and reading ability and also included an attitude measure to assess students' attitudes toward sentence-combining instruction. Using the same research design at the same institution, Sullivan (1977) and Callaghan (1977/1978) used llth-grade and 9th-grade students respectively to measure the effects of a sentence-combining curriculum which included three experimental groups and a control group. One treatment group did 3 lessons; a second treatment group did 15 lessons; and a third treatment group did the same 15 lessons with oral recitation. The population in the Sullivan research involved four intact English classes for each of seven teachers from one urban and three suburban high schools. The population in the Callaghan research included 580 students in 3 6 classes in seven different schools; four urban, three suburban, and
50 two Catholic school teachers were involved in teaching the classes, each teacher teaching four classes. Students wrote four papers at both pretest and posttest in the narrative, descriptive, expository, and argumentative modes. They also completed two forms of the same reading test and a class attitude questionnaire at both pretest and posttest. In both research studies a random sample of pretest and posttest writing samples was used for data analysis. Results of the multivariate analysis of covariance, covarying on the pretest syntactic and reading measures, differed for the two studies. Sullivan reported significant differences in mean Tunit length, growth in noun substitutes, but no significant difference for final free modifiers nor for reading. There were also positive, significant changes in syntax, with those receiving 15 lessons making as much gain as those receiving 30. There were also positive, significant changes in attitude on the part of the experimental groups. There were, however, no significant differences in overall writing quality. Callaghan reported significant differences for the experimental groups for T-unit length and for final free modifiers. There were no significant gains for reading, nor were there significant differences in syntactic maturity. There were significant differences in student attitude toward sentence-combining practice and toward the revision
51 of composition between pretest and posttest favoring the experimental groups. Case study analysis indicated some syntactic differences in papers according to time written during the treatment or mode of writing. Some erosion of gains in syntactic maturity was apparent in follow-up papers. Howie (1979) investigated the effects of a fifteen-week sentence-combining curriculum using 91 9th graders from four classes in one surburban high school in Colorado. Students were randomly assigned by computer to their teachers and the hour of the class. Two classes were assigned to the experimental program which consisted of sentence-combining exercises taken from Marzano and DiStefano's Di-Comp (1977), O'Hare's Sentencecraft (1975), and Strong's A Composing Book (1973), consecutively. Before and after the experiment, writing tasks, the reading tests, and the attitude scales were administered to both groups. The discourse modes of description and exposition were elicited before and after the experiment. The reading test was a cloze instrument constructed on the Gray Oral reading passages, Forms A and B. The two attitude scales utilized a Likert Scale ranging from scores one through five of attitudes toward writing and reading. Twelve hypotheses dealt with writing, four with reading, and two with attitudes. The researcher found significant differences between groups in descriptive
52 compositions, favoring the experimental groups (.001) in the analysis of covariance statistical test. There was no significant difference in syntactic writing ability in expository composition, nor was there any significant difference in reading level or attitudes toward writing and reading. The researcher concluded that sentence-combining practice is effective in increasing syntactic maturity in the mode of description, but not in exposition. She also felt that further research was needed to investigate the effects of guality, audience and purpose, and attitudes on writing ability. MacNeill (1982) investigated the effect of Frank O' Hare's Sentencecraft sentence-combining program on the written syntactic skills, reading comprehension level and speed of reading of average ability 9th graders. The subjects were 140 students, 75 of whom were in the experimental group and 65 that were in the control group. The experimental group received 9 weeks of practice in combining kernel sentences into more complex and mature syntactic forms. For the remainder of the 16-week experimental period both groups wrote 6 argumentative compositions at 2-week intervals except for essay 2 which was written 3 weeks after pretest. The 6 compositions were scored counting total number of words, words per T-unit, words per clause and clauses per T-unit. The Davis Reading Test, composition and rate sections, was used for the
53 posttest at 16 weeks and then for delayed posttest 8 weeks later. There were no significant mean increases across the six writing occasions. The experimental treatment was not effective in improving students' level of comprehension. Flaherty (1985) sought to determine relationships between students' amount and kind of verbal interaction and their reading and writing growth. She used eight high school students in a basic reading class in a participant observation study. This study was a practitioner-designed, integrated approach to sentence combining and reading comprehension. Orientation sessions, interviews, a studentsurvey, video-and audio-taping, a forced choice rating of overall writing quality and reading comprehension testing were all used to collect evidence. The eight students' writing samples and evidence of amount and kind of interaction were analyzed and the relationships were noted. The researcher concluded that if individual basic reading students actively participate in large and small group discussions because they see value in them, and if instruction is based on emerging needs, then a positive relationship between their kind and amount of verbal interaction and their reading and writing can be seen. The last study reviewed which used high school students as subjects was conducted by Pendleton (1985). She sought to determine the effect of instruction in sentence combining and rhetorical arrangement on the organization and
54 overall quality of expository writing among selected 10th graders. Three classes received instruction in the experimental curriculum. One received the sentence-combining curriculum only. The second received the rhetorical arrangement only and the third received both sentence combining and rhetorical arrangement combined. Writing samples were adminstered monthly and were evaluated using holistic and primary trait scoring. Regression analysis was used to evaluate the results. All three treatment groups showed significant change in writing quality as determined by holistic evaluation. Only those groups receiving instruction in rhetorical arrangement or sentence combining and rhetorical arrangement showed a significant change in organizational quality by primary trait evaluation. Eleven studies were reviewed that dealt with college students or adults. Menendez (1978) studied whether weekly sentence-combining activities, practiced as part of a 1semester college composition course with a remedial curriculum would influence students' language performance in three areas: (a) writing performance as indicated by syntactic measures and holistic rating of writing quality; (b) punctuation skills; and (c) reading performance. The sample included 102 freshmen, selected on the basis of standardized test scores, who were assigned to one of the following treatments: (a) two intact experimental classes which met 3 times weekly and practiced sentence-combining in
55 addition to the regular curriculum; (b) one control class which met 3 times weekly and received extra writing practice in addition to the regular curriculum; and (c) 27 students in four classes of regular composition who served as a control group. Pretest and posttest treatment writing samples and punctuation tests were collected from all students. Pretests and posttests of reading comprehension and efficiency were administered to students in treatment conditions 1 and 2. Using T-unit analysis procedures, the researchers analyzed student writing for six factors of syntactic maturity. Differences in gains in mean scores for all treatment groups on syntactic maturity, writing quality, punctuation, and reading measures were registered on a oneway analysis of variance. Differences in mean gain scores between treatments 1 and 2 were also compared by t-test. Judged by the analysis of variance and t-test on each factor, the experimental group gains were significantly different on only two measures of syntactic maturity: clauses per T-unit and adjective clauses per 100 T-units; and on one measure of punctuation-error identification. The researcher found no significant differences in writing quality or reading performance of the experimental groups as compared with the control group.
56 Ledesma (1981) sought to determine whether sentence combining contributes to reading comprehension at three levels: literal, reasoning, and evaluative. The subjects were college freshmen enrolled in Developmental English I. They were randomly assigned to groups; the control group completed grammar lessons while the experimental group completed sentence-combining lessons. Comprehension scores of the Iowa Silent Reading Test and comprehension scores of the Constructed Passages written at 6th, 10th, and adult syntactic complexities were used for evaluation. Analysis of variance, nonparametric sign tests, and Pearson's r were used in the analysis of the data. There was a significant difference on the reasoning level for the sentence group. There were no significant differences on literal or evaluative levels. The more clauses and T-units students wrote in controlled composition, the higher the reading comprehension scores were on the ISRT. Lamm (1982) used a case study approach with four college students enrolled in a reading skills course. As a text the students used The Writer's Options: College Sentence Combining-Reading Skills Course A folder was kept on each student with notes on student progress and difficulties. Before and after the treatment each of the following instruments were used: the Stanford Achievement Test, Forms A and B; general cloze tests omitting structure words; and oral miscue analysis. The researcher concluded
57 that while sentence combining did provide a larger repertoire of structure words, it had relatively little effect on their ability to comprehend written syntax. Story (1985) used 71 college students taking Reading 100 to determine if sentence-combining techniques increased their reading comprehension. The researcher used the equivalent control design. The Stanford TASK, Level 2, was used before and after the treatment to evaluate the results which were analyzed using analysis of covariance. No significant differences were found. Porter (1985) conducted two studies with ESL freshhmen at a Texas university. The first study compared the sentence combining of foreign students with that of native freshmen. The analysis revealed significant differences between these two groups. These findings suggested changes in the sentence-combining lessons which were used in the second study. The researcher concluded from the results that sentence combining can be effective in teaching syntactic structures to foreign freshmen but that supplementary grammar should be a component of the lesson. Evans (1986) conducted a series of studies to explore three trends that had emerged from sentence-combining research: (a) students with lower abilities in sentence combining and reading comprehension tend to register higher gains in both areas as result of sentence-combining instruction; (b) the broader the range of any student's
58 sentence-combining strategies, the better he or she will do in reading comprehension; and (c) specific syntactic constructions or sentence-combining transformations are indigenous to identifiable chronological stages of development. The three studies were conducted with respectively 30 college juniors, 71 12th graders, and 30 6th graders. The design of each study was the pretest/posttest control group. The instrument used for analysis was the Descriptive Tests of Language. Multivariate analyses of variance were used to analyze the data. Sentence-combining instruction most influenced the reading and writing development of students with low abilities. Integration of traditional and transformational sentencecombining strategies influenced high students just as dramatically as exclusive instruction in sentence combining. Sentence combining most sensitively distinguished between and measured gains among the reading and writing skills of students who began such instruction with the lowest skills. Overall, the findings confirmed the bond between syntactic constructions in reading and writing. Brewer (1986) investigated the effects of two methods upon the writing of college freshmen. Three classes were taught by the researcher using a curriculum of sentence combining and writing apprehension-reducing practices in a writing workshop method. Three other classes were taught by the researcher using conventional practices used in English
59 composition classes. Data were obtained in four ways before and after the treatment: a Syntactic Maturity Test; a Writing Apprehension Test; a writing sample; and a post test only two-hour writing sample. The statistical analysis was completed using analysis of covariance. All students in the study reduced their writing apprehension. The high apprehensives in the treatment group wrote t-units which were significantly longer than those written by high apprehensives in the control group. Finally, on ratings of overall quality, holistic scores revealed that the students in the experimental group wrote essays at least equal in quality to those essays written by the control group. Whitt (1987) sought to determine if basic writing students ages 18-25 who were freshmen students enrolled in four classes of non-credit Basic Writing showed greater growth in syntactic fluency by the use of sentence combining or by the use of traditional methods of teaching grammar, mechanics, and sentence structure. Two teachers both taught a control and an experimental group. Both groups spent about 50% of the class on writing process. The balance of the time was spent on the practice of cued sentences and open sentences in the experimental group and on traditional skill exercises in the control group. Syntactic maturity was measured by T-unit length, clause length, and number of clauses per T-unit on preand posttests of (a) the Aluminum rewriting test; and (b) descriptive and narrative papers.
60 The experimental group using sentence combining showed significant differences on three factors of syntactic maturity on both measures when compared to the control group. Both the experimental and the control groups showed a significant increase in the number of words written, but no significant increase in syntactic maturity occurred in the control group. The results suggests that basic writing students carry over the sentence-combining techniques faster to the rewriting tasks of the Aluminum test than they do to their first draft, essay tasks. Taki El Din (1987) investigated the effectiveness of sentence-combining practice integrated in an English as foreign language course on the overall writing quality and syntactic maturity of Arab students at the United Arab Emirates University for 15 weeks. An experimental group of 63 students wrote sentence-combining exercises instead of correcting grammatical errors. A control group of 67 students used the regular EFL curriculum in composition with the correction of grammar. The instruments used were two preand two postwriting samples. Multivariate analyses of variance and t-tests were used as the statistical analyses. The writing samples were scored holistically and were analyzed for ten factors of syntactic maturity. The experimental group showed significant differences in number of words, number of clauses, number of words per T-unit, words per clause,
61 clauses per T-unit, and percentage of words in right free modifiers. Tracy (1990) investigated the effects of sentencecombining practice on syntactic maturity and writing quality in an English as a second language course with four classes of freshmen. The control group had no sentence-combining practice; however, the experimental group had sentencecombining practice and discussion 1 time per week. Both groups took pretests and posttests consisting of Davidson' s Test of the Ability to Subordinate and an essay which was analyzed by three indices of T-unit analysis: words per clause, words per T-unit, and clauses per unit. The final essays were holistically scored. This final holistic scoring was then statistically compared with the quantitative scores from the three indices of T-unit analysis. In their 1988 exploratory review, Searls and Neville summarized the sentence-combining studies from 1970-1985. They were interested in two basic issues: (a) whether the effectiveness of sentence-combining training was related to the grade level of the subjects; and (b) whether the effectiveness of sentence-combining training was related to the type of criterion measure used. In investigating these two issues the researchers found their analysis suggested four treatment variables that could have an influence on the results of the studies. These four treatment variables were
62 length of treatment, type of practice (massed vs distributed) who delivered the treatment, and where the treatment was delivered. As to the first of these two issues, sentence-combining training appeared to have the greatest effect when training was conducted on elementary grades. This is contrary to findings that sentence-combining training on measures of syntactic fluency in written composition has been shown to be consistently positive over a wide range of grade levels from elementary to college (Combs, 1976: Morenberg, Daiker & Kerek, 1978; O'Hare, 1973; Perron, 1976; Searls & Neville, 1982) The other issue, that of criterion measure, showed cloze tests to be more sensitive than standardized tests to the effects of sentence combining. The researchers suggested one reason for this might be that standardized reading tests, if appropriate to a subject's grade level, may not be written at high enough levels of syntactic complexity to reflect the effect of sentence-combining training. Further information showed that cloze tests may yield more significant results than standardized reading measures because of three factors: (a) what cloze tests measure; (b) the types of words deleted; and (c) the level of syntactic complexity of cloze tests (Fisher, 1973,1974; Froese & Kurushima, 1979; Neville & Searls, 1985). And as Neville
63 and Searls (1985) demonstrated in their cloze test, difficulty level of vocabulary can be held constant while the difficulty level of the syntax can be manipulated by increasing the number of words per T-unit through the use of more dependent clauses, nominals and adjectives. In terms of length of treatment, they found that 1-2 months of treatment was significantly (.01) better than 3-4 months. In addition, 5-9 months treatment was significantly (.05) better than 3-4 months. In terms of type of treatment, practice was defined as massed if the treatment lasted 3 months or less with a freguency of 4-5 times per week. Distributed practice was defined as treatment that lasted 4 or more months with a freguency of 1-3 times per week. Searls and Neville (1988) claimed that massed practice appeared to be more effective than distributed practice. All measures were administered immediately after sentencecombining training ended. They also reported that massed practice was better for short term recall and that distributed practice was better for long term retention. Only two studies they reviewed attempted to measure retention by means of delayed posttest with significant effects. Under the factor of who delivered the sentencecombining training, Searls and Neville found that in 17 studies treatment was administered by the experimenter; in
64 3 studies treatment was administered by the regular teacher; and in 4 studies the experimenter and the regular teacher worked together. Finally, Searls and Neville (1988) found that sentencecombining training appeared to be more effective when given outside the regular classroom situation. This situation was affected, however, by grade level. At the secondary level, the data indicated a decrease in positive results the longer the treatment lasted. There was no interaction between type of practice and grade level. Only at the secondary level did experimenter involvement appear to be effective. It was also at the secondary level that pull-out groups had the greatest effect. Searls and Neville's research (1988) about the benefits of sentence-combining instruction supports the earlier research of Schuster (1976) whose research found the following advantages of such instruction: (a) it acquaints students with resources of their language; (b) students write more mature sentences; (c) sentence-combining activities improve writing competence in general; (d) it helps students avoid writing fragments and run-ons (Those using such instruction wrote 34% fewer fragments and 15% fewer run-ons) ; (e) there was improvement in self-confidence as evidenced by the students writing longer essays (Students with the lowest IQ made the most dramatic increases-31% longer at posttest than at pretest) ; (f ) there was an
65 improvement in spelling; (g) lessons can be structured to include mastery of standard English; and (h) gains may be made in punctuation skills. Thus, Schuster's work also corroborates two of the key points in Charles Cooper's article, "An Outline for Writing Sentence-Combining Problems," (1973) that sentence-combining exercises along with an informal approach to correcting deviancy provide the traditional goals of grammar study: standard usage and control of written syntax. The other notable conclusion from Schuster's study is that students using these exercises experience greater fluency, that is production of longer essays. Less capable students traditionally write less often and when they write, write fewer words. Sentence combining then seems as summarized here by Sternglass (1980) in her article, "Sentence Combining and the Reading of Sentences" to be a means by which low-achieving students may gain confidence and build their writing skills: These activities (sentence-combining) seem particularly appropriate for basic writers who demonstrate a real lack of fluency in their writing. Since sentence-combining activities provide the basic semantic content for these basic writers, these writers are not held back by their lack of fluency in producing written text from engaging in the kinds of language processing activities that they also need to practice in learning to read more efficiently. . This proposed conceptual basis for the contribution made by practice in sentence combining may add one further dimension to our understanding of the process. It may at least help us move our students from focusing on words to attending to longer semantic units, (p. 3 28)
66 The review of literature on sentence-combining studies has included the original seminal studies which laid the groundwork for subsequent studies: Chomsky (1957), Hunt (1965, 1970), Bateman and Zidonis (1964), Rinne 1967), Mellon (1969), O'Hare (1973), and Fisher (1974). The review of other literature on sentence combining has included studies with special populations such as ESL students (Porter, 1985; Taki El Din, 1987), cross grouping (Jordan, 1992), and poor or remedial writers (Evans, 1986; Flaherty, 1985; Isaacson, 1985; Menendez 1978). Also reviewed were studies with specific strategies added to the teaching of sentence combining such as word processing (Laframboise, 1990; Piper, 1985), sentence combining as a revision strategy (Bordonaro, 1991) free writing (Young,1989) grammar (Ledesma, 1981; Spilton, 1987; Trivelli, 1983) and cloze procedure (Howie, 1979; Neville & Searls, 1985) Case study and participant observation research were reviwed (Flaherty, 1985; Lamm, 1982). Different types of discourse as a measure of sentence-combining improvement were reviewed (Callaghan, 1977/1978; MacNeill, 1982; Sullivan, 1977) And evaluation by both holistic and primary trait scoring was reviewed in Pendleton (1985). None of the studies reported gain in reading achievement from standardized tests. Further sentencecombining research may extend the type of curricular
67 activities that develop syntactic knowledge and improve reading and writing performance. Research on Apprehension Toward Writing The concept of "writing apprehension" was defined by Daly and Miller (1975a) as a "specific case of general communication apprehension Â— one's anxiety or fear about communicating in real or imagined communications" (p. 243) The research on writing apprehension may be divided into those studies which have investigated the nature, sources, or effects of writing apprehension and those studies which deal with writing programs or strategies designed to reduce or lower writing apprehension in its subjects. Some studies that were reviewed include both of these types. They will be discussed in the section which most closely matches the primary objectives of the study. To measure writing apprehension, Daly and Miller (1975a) developed a Likert-type scale format consisting of 63 items each with 5 possible responses. Items were so constructed that they dealt with some form of apprehension about writing. The valences of the items were random in an attempt to prevent any response bias due to the direction of items. The 63 -item instrument was completed by 164 undergraduates enrolled in basic composition courses and interpersonal communication courses in Spring 1974.
68 On the Daly and Miller instrument individuals scoring one standard deviation above the mean on writing apprehension were classified as high apprehensives, while those scoring one standard deviation below the mean, were classified as low apprehensives. The remaining subjects were classified as moderate in their anxiety. The comparison of this instrument with other similar instruments was considered desirable by the researchers. They also recommended that future research deal explicitly with the treatment of anxious subjects as with the effects of writing apprehension. Daly and Miller (1975b) furthered the investigation of writing apprehension through examination of students enrolled in remedial and regular sections of a basic composition course. The sample of 246 university undergraduates was selected for inclusion in the remedial program on the basis of a low Scholastic Aptitude Test (SAT) verbal score. The purpose of this study was to compare scores on the writing apprehension measure with SAT verbal scores as both relate to prediction of success expectations, willingness of students to voluntarily take additional courses in writing, and placement in remedial as opposed to regular basic composition courses. Students were asked to complete a two-page booklet which contained the now 26-item writing apprehension measure (Daly & Miller, 1975a) as well
69 as questionnaire items to produce the dependent measures and their SAT verbal scores. The dependent measures consisted of five items. One dealt with previous success in writing courses. Two items dealt with willingness to take additional courses in writing and the remaining two items dealt with expectations of success. It was suggested that there would be a low correlation between SAT verbal scores and writing apprehension scores. The obtained Pearson Product Moment correlation was .19 (p=<.01) which while significant accounted for little shared variance (less than 4%) The researchers initially predicted that there would be significant differences in the predictive power of the writing apprehension measure and the SAT verbal scores in relation to a number of predispositional variables. They found that writing apprehension correlated with these variables significantly more than did SAT verbal scores. Other findings were as follows: students voluntarily enrolled in advanced writing courses scored significantly lower than others on writing apprehension; individuals high in apprehension perceived their past experiences in writing as significantly less successful than did low apprehensives; and, most importantly, predispositions towards writing may not be effectively measured by many currently used aptitude tests. The authors concluded that the writing apprehension measure deals with attitudinal variables and provides a
70 valuable addition to our knowledge about students as writers. Subsequent research by Daly (1978) confirmed these initial findings and supported the research contention that low apprehensives scored significantly better than high apprehensives for all subtests of writing competency. In addition, the moderately apprehensive writers fell between the high and low apprehensives as to writing competency thus suggesting a continuum ranging from low to high apprehension. In addition to the studies pioneered by Daly and his associates, there were 11 other studies found in the literature in which researchers investigated the sources and effects of writing apprehension or attitude toward writing. All of these studies focused on entering college freshmen. Only those studies dealing with entering freshmen in remedial courses will be reviewed. The first of these studies was conducted by Stiles (1976) who analyzed the composing processes and attitudes toward writing of students in remedial English composition courses. The subjects were four writers from each of two 2year colleges (n=8) who had been characterized by their instructors as being seriously deficient in basic writing skills. Each subject met with the researcher 5 times. During 4 of these times, the subjects composed aloud in the presence of a tape recorder utilizing in turn the modes of
71 narration, description, exposition, and argumentation. During another session the subjects were interviewed by the researcher as to their previous experiences with and attitudes toward writing. The results of the study revealed that these students were preoccupied with the mechanics of writing, favored the expository mode and seemed to view the writing process as consisting of three parts, and were characterized by cognitive and linguistics deficiencies. However, these students still appeared to view writing in a favorable manner and to see the benefit of being able to write well. The greatest influences on their attitudes about writing were their families and their English instructors. Writing disability, so the researcher concluded, is a cumulative problem, one which needs to be addressed long before students reach the college level. Garcia (1977) conducted a study using 3 2 students at Arizona Western College. Instruments used in data gathering included the Writing Apprehension Survey (Daly & Miller, 1975a) an in-class essay assignment, and a rater-judged quality writing process. Statistical tests revealed significant differences on seven syntactic variables: infinitives, participles, gerunds, prepositional phrases, adjectival, adverbial, and nominal clauses. The following are the results of the study: (a) quality writing performance was dependent upon high and low levels of
72 writing apprehension; (b) attitude, as measured by the writing apprehension test, that a student holds toward the act of writing clearly affects both how he or she will write and how others will evaluate that writing; (c) lowapprehensive students demonstrated syntactic characteristics of mature writers more consistently than did highapprehensive students; and (d) syntactic encoding performance and guality writing performance are not limited to certain sex categories. Harvley-Felder (1978) explored the factors relating to writing apprehension in a study of 12 high writing apprehension students and 12 low apprehension students. Three factors were identified in the study and found to relate to writing apprehension: punishment, positive reinforcement, and communication-seeking behavior. The author found marked differences in these factors between the high and low apprehensive groups in preschool, elementary, junior high and high school, and home experiences. Powers, Cook, and Meyer (1979) also used the WAS with students admitted to a university under special conditions comparable to the remedial classification. The sample was 57 students enrolled in the reguired English composition course at a large midwestern university. Twenty-eight of these students had been admitted into a special program for students not meeting or minimally meeting entrance standards.
73 All subjects were asked on the first day to complete the 2 6-item WAS. The course was divided into nine sections meeting 2 days per week for 8 weeks. During the 8week period, students were required by individual instructors to write five to six essays of various types and lengths. On the last day of class all subjects again completed the WAS. For purposes of analysis the sample was divided into high apprehensives and low apprehensives. A dependent ttest was done over all subjects between the pretest and the posttest. Results indicated that writing apprehension scores were significantly higher (.05) at the conclusion of the treatment than at the beginning. The study indicated that compulsory writing increases writing apprehension. A case study exploring the sources of negative attitudes toward the writing of three skilled college students was conducted by Gay (1983). The study included interviews with the students concerning their writing histories and with their high school English teachers. In addition, information was gathered about their verbal abilities and their high school grades. Analysis of the data indicated that the three students held many misconceptions about the nature of writing that contributed to their negative attitudes and appeared to hinder the development of their writing abilities. For the unskilled writers in this study writing appeared to be the following: (a) outer-directed rather than inner-directed;
74 (b) written to please the teacher and earn a grade, not to please themselves; and (c) viewed in ways shaped by their previous teachers' attitudes. Stacks, Boozer, and Lally (1983) proposed the idea that written communication apprehension is a multidimensional construct rather than a unidimensional one as originally proposed by Daly and Miller (1975a). Stacks et al. presented a multidimensional instrument which contained many of the original statements of the Daly and Miller (1975a) and others. They reported a 6-f actor solution and argued that it better represents writing apprehension than the unidimensional instrument. They pointed to the work of Daly and Shamo (1976, 1978) who found that writing apprehension is predictive of occupational and academic choice. Citing the work of Faigley, Daly and Witte (1981), Stacks et al. noted that syntactic choice may be a significant predictor of writing apprehension. Stacks et al. (1983) further surmised that, "if it is and it can be demonstrated that low and high writing apprehensives differ in their syntactic choice, then pedagogical tools could be derived to impact on writing apprehension" (p. 4) Thus, the study was an exploration of writing competency (as measured by actual syntactic choice) in a business situation. The five variables predicting general writing apprehension came from five different qualities: existence, social perception, motion, time, and conditionality The
75 authors concluded that the results of this investigation support the multidimensional treatment of writing apprehension. They also concluded that the instrument predicted writing apprehension as related to different syntactic qualities, that it correlated highly with the factors obtained and that it provided diagnostic capabilities for business writing. In all cases the relationships between writing apprehension and individual factors were linear, suggesting that treatments should examine ways to either decrease or increase particular syntactic choice and use. Stacks et al. finally concluded the following about the value of sentence-combining exercises: Specifically, writers can be encouraged to engage in sentence-combining while transforming their notes and other prewriting materials into draft, and they can also be encouraged to exercise their new skill when they engage in revising and/or editing their drafts. In other words, once the writer learns the skill of sentence combining, the writer can use it to transform his or her own sentences into those of a complex and variable kind of structure. Because of the correlation between problems with syntax and writing apprehension, sentence-combining exercises would appear to be appropriate for writers with high apprehension, (p. 9) Finally, research efforts have also targeted teacher behavior or materials and strategies in the reduction of student writing apprehension. King (1979) attempted to assess both students and teachers attitudes toward writing
76 and to determine if these attitudes could be changed through teacher participation in the New Jersey Writing Project's 1977 Summer Writing Institute. Eighteen experimental teachers, seven control teachers, 1093 experimental students and 601 control students completed one or more test measurements. All teachers completed the Emig Teacher Attitude Scale. The Emig Student Attitude Scale was completed by 1093 experimental students in the fall; 952 completed the same form in the spring. The Emig Student Attitude Scale was completed by 601 control students in the fall and 449 control students responded to the same from in the spring. In addition, 2 63 control students completed the King Construct. Both the 52-item Emig Student Attitude Scale and the 50-item Emig Teacher Attitude Scale presented items in three categories: preferences for writing, perceptions of writing and processes of writing. Each scale consisted of items with five choices on a Likert-type scale. The King Construct Scale, based on George Kelly's psychology of personal constructs, presented 25 bi-polar constructs separated by a 5-point scale. The 25 bi-polar constructs represented student preferences for source, purpose, audience, and mode. Results showed that students in the experimental group exhibited a more positive attitude toward writing, reported writing more, and reported having a greater liking for what they wrote. The study concludes
77 that participation in the 1977 Summer Writing Institute of the New Jersey Writing Project led to a positive change in teachers' and students' attitudes toward writing. This study adds valuable information to the work of Daly and Miller (1975b) as it includes mode and audience that theirs did not. The next study on programs to reduce writing apprehension is interesting as it was not specifically a writing program but a reading program. Its goal was, however, the reduction of writing apprehension. Steidle (1977) investigated whether student attitudes toward reading were predictive of writing ability. He correlated scores on the Diederich Composition Rating Scale with measures of specific attitudes toward reading and general attitudes toward school for 920 students from grades 4, 5, 9, and 12. This was a subsample drawn from the fall 1974 and spring 1975 Virginia Educational Needs Assessment. The Virginia Affective Assessment Questionnaire (VAAQ) was developed by the Assessment Project staff in 1969-70 and 1974-76. Four forms of the VAAQ were developed: a primary form for kindergarten-6, an intermediate form for grades 79, and a secondary form for grades 10-12. At the conclusion of the study, data analysis showed a direct linear dependence between both specific and general attitude measures and writing quality. The author concluded that, while grade level of writer is important in the prediction
78 of writing quality, attitudes are more important and more precise in predicting that writing quality. In summary, the specific concept of "writing apprehension" was defined by Daly and Miller. Daly and Miller's development (1975a) of a self -report instrument called the Writing Apprehension Survey (WAS) gave researchers a tool by which to measure the amount of anxiety over writing which is present before and after various writing, and, in some cases, reading programs or strategies. This 26-item instrument with five Likert-type responses for each item measures both positive and negative attitudes toward writing. Through this instrument Daly and Miller identified those scoring one standard deviation above the mean as being high apprehensives and those scoring one standard deviation below as low apprehensives. Further work by Daly and Miller (1975b) correlated WAS scores with SAT verbal scores. In addition, Daly (1978) compared WAS scores with writing competency specifically in the areas of mechanics, grammar, and larger elements of fiction. Faigley, Daly, and Witte (1981) corroborated the Daly study (1978) as did Garcia (1977) when both found important syntactic factors which when present helped lower writing apprehension. Other studies also found noticeable differences in the writing competencies of high and low apprehensives. Book (1976) found differences in structure, language use, and
79 amount of information used while Gay (1983) and Selfe (1984) found differences in focus of inner-outer direction of writing, the audience written for, and previous attitudes toward writing. Other factors found to have an impact on writing apprehension were fear of punishment (HarvleyFelder, 1978) compulsory writing (Powers, Cook & Meyer, 1979) and reading ability (Steidle, 1977) In contrast to Daly and Miller's unidimensional view of the construct, writing apprehension, Stacks, Boozer, and Lally (1983) envisioned the construct as multidimensional and found a number of factors which influence writing apprehension. They developed their own writing apprehension instrument which built on the WAS (Daly & Miller, 1975a) and added 38 other statements covering a number of attitudes toward business communication. Writing apprehension research studies have shown that poor writers tend to be more apprehensive about their writing, select academic and occupational choices which appear to have lower writing requirements, and suffer from past negative experiences with writing. They also, so these studies have shown, focus more on the surface features of writing such as mechanics; they have more difficulty in the use of information and in the structuring of it during prewriting, drafting, and revising; and they suffer syntactic deficiencies which hinder both their reading and writing competencies. Thus, this research points the way to
30 writing curricula which are nonthreatening, which enable students to practice syntactic patterns without fear of failure, which utilize group discussion of sentence possibilities, and reward students for experimentation with language, not punish them for "errors."
CHAPTER III METHODOLOGY The purpose of this study was to determine which of two types of sentence-combining instruction produced the better student writing performance and reading achievement scores of high-risk secondary students. In addition, this researcher sought to determine if such a sentence-combining writing curriculum had an effect on the writing apprehension of the subjects. This chapter provides information on the research methodology, subjects, instrumentation, regular reading instruction, experimental writing curriculum, regular English instruction in writing, and procedures used by the researcher in conducting this study. Setting of the Study The school where the data were collected was located in a north-central Florida community of 83,000 people. It was made up of students from professional, academic, business, and "blue-collar" family backgrounds. It was a public high school with grades 9-12, a total student population at the time of the study of 1,936, and a faculty of 108. Student enrollment has fluctuated slightly over the years, but the 81
82 general pattern has been constant growth by about 4 students per year. In setting attendance guidelines, the local school board has adopted a policy of maintaining a black nonHispanic minority population at about the same percentage as the community (30%) with no more than a 5% variation either way. The white non-Hispanic population of the school during the study was 72.1% and the black non-Hispanic population was 24.6%. The remaining percentages were made up of Hispanic and other minority groups. Students were bused in order to achieve the racial balance reflected in the county population. The school draws its population from urban, suburban, and rural areas. Economically, many students came from affluent homes with 52.1% reporting an income of at least $30,000. Approximately 10% of the population came from an income level of $8,000 or less. The number of students who qualified for either free or reduced lunch was 242 or 12.5%. The school also participated in a federally-subsidized breakfast program. In terms of career goals, 7 5% of the students attended postsecondary school, and of the remaining 25%, 2.8% joined the military and 12.3% entered the job market. These figures were reported in a follow-up study conducted by the school board's Pupil Personnel and Guidance Department in
83 1980. This distribution has been virtually the same, within 5%, in all categories in all recent surveys. Academically, the student population served by the school in this study is widely diversified. Results from the Metropolitan Achievement Tests administered to all 10th graders during the school year followed a normal distribution. The subjects in the sample included a total of 54 10th, 11th and 12th graders in remedial reading courses. These students were placed in these classes because they scored at or below the third stanine on the Metropolitan Achievement Test, Reading section, 1979 edition. In this regard, they were homogeneously grouped by reading level or ability. The group was made up of 46 (85.19%) ethnic minority students, mostly black, and 8 (14.81%) white students. The students' ages ranged from 15-19 years of age. Grade level breakdown was Grade 10: 13%; Grade 11: 52%; and Grade 12: 3 5%. Instrumentation Six instruments were used to test the research hypotheses: the MAT, Reading Survey section (1979), was given before and after the treatment; a descriptive writing sample taken from the National Assessment of Educational Progress (1980) was collected before and after the treatment; and a writing apprehension measure, the Daly and Miller Apprehension Survey (1975a) was administered before and after the treatment.
34 Metropolitan Achievement Test. Reading Survey The students were given the Metropolitan Achievement Test, Reading Survey Section (1979 edition), Advanced Level I, Form KS, as a pretest and as a posttest. All subjects were given the same test on the same day in their regular reading classroom. Students were given directions according to the manual for administration. Time limits for the test were strictly followed. Students were given 3 5 minutes to complete the test as per the manual's instructions. The MAT Reading Survey section scores were machine scored by the test publisher (The Psychological Corporation) and class printouts were provided to the researcher. Reliability of the Reading Survey Section as reported in the administration manual was .94 and was calculated by the Kuder-Richardson 20 formula. Subscore reliabilities were not reported but were stated to be sufficiently high. The Reading Survey questions contains a total of 55 questions based on a series of graded reading passages. The scores include six subscores and one total reading score. The following are paraphrased descriptions of each reading subscore in the Teacher's Manual for Administering and Interpreting (1979) : Vocabulary 6 questions The pupil is required to use context clues to arrive at the meaning of a word as it used in the passage. All of the words tested have more than one meaning. (p. 50)
35 Literal Specific -21 questions The pupil is required to use information stated explicitly in the passage to answer questions related to details, sequence, character traits, and following directions, (p. 50) Literal Global -5 questions The pupil is required to use information based on the entire passage or a large section of it to answer questions related to the main idea of the passage, comparisons related to the main idea of the passage, comparisons, and contracts, and cause and effect. (p. 50) Inferential Specific -11 questions The pupil is required to use information implied in the passage to infer answers to questions related to details, sequence, character traits, and predicting outcomes, (p. 50) Inferential Global -7 questions The pupil is required to use information implied in all or a large section of the passage to infer answers to questions related to the main idea of the passage, comparisons and contrasts, and cause and effect, (p. 50) Evaluative -5 questions The pupil is required to make judgments about the contents of a passage to answer questions related to mood (overall feeling, emotional reactions) and criticism (reality vs fantasy, fact vs. fiction, validity, author's purpose, propaganda), (p. 51) Each of the six reading subscores and one total reading score determined by adding the six subscores provided the reading data points for this study. National Assessment Writing Exercise The students were asked to write on two topics, one before the study and one at the end of the study, which were similar to the topic "Describe Something" which was used in
36 the National Assessment of Educational Progress (NAEP) The "Describe Something" topic was used as a pre-study writing exercise to familiarize students with the type of descriptive writing that this exercise would involve. The following directions of this exercise were taken from The Third Assessment of Writing 1978-79 Released Exercise Set Everybody knows of something that is worth talking about. Maybe you know about a famous building like the Empire State Building in New York City or something like the Golden Gate Bridge in San Francisco. Or you might know a lot about the Mormon Tabernacle in Salt Lake City or the new sports stadium in Atlanta or St. Louis. Or you might be familiar with something from nature, like Niagara Falls, or a gigantic wheat field, a grove of orange trees, or a part of a wide, muddy river like the Mississippi. There is probably something you can describe. Choose something you know about. It may be something from around where you live, or something you have seen while traveling, or something you have studied in school. Think about it for awhile and then write a description of what it looks like so that it could be recognized by someone who has read your description. Name what you are describing and try to use your best writing. (NAEP, 1981, p. 323) The two topics which the students wrote at the beginning and end of this study were written by the investigator and were selected because they were intended to elicit the same type of descriptive writing from the students. Both topics were presented to the students at the preand postwriting sessions to ensure that the raters were not aware of which topic was the prewriting topic and which was the postwriting topic. Students were instructed to write on one topic at the prewriting session and to select the other topic at the postwriting session. All students
87 complied with this request. As a result, some students selected Topic A for the prewriting topic and some students selected Topic B. The same was true of the postwriting session. The following are the topics and instructions given to the students at the preand postwriting sessions: Directions: Pick one of the two topics-either A or B. Write on the space provided for your answer. You may want to write a few notes on the scratch paper provided and then put your final copy on this page. Topic A There are days which are special to us. We remember them because of the good things that happened on them. Pick one of those days and describe in detail what happened. Tell why that day was so special, or why it made you so happy. If you cannot remember such a day, tell what a special day would be like for you. What things would you like to do and experience? Explain how it would make you feel. Topic B There are many persons who affect us, our lives, and our relationships with other people. Describe one person who has meant a great deal to you. Include physical details and a description of his/her personality. Tell what this person has meant to you, and how he/she has influenced your life. The descriptive writing samples were collected for all 54 subjects on the same days and each student was given 45 minutes to write the assignment and was given 4 lines on page 1 and 25 lines on page 2 as suggested in the NAEP guidelines (NAEP, 1981, p. 323). The writing samples were taken during the same weeks that the MAT Reading Survey was
88 administered by the investigator for this study. These samples were collected before and after the treatment. Following the NAEP guidelines, the researcher evaluated the writing samples in four categories: holistic, cohesion, mechanics, and syntax. Mechanics includes scores for the number of errors for agreement, punctuation, spelling, and capitalization. Syntax includes scores on the number of nominal clauses, nominal phrases, relative clauses, modifying phrases, adverbial clauses, adverbial phrases, and inter-T-unit coordination. As a result, 13 writing scores were produced from the analysis of each writing sample. The following 13 scores provided the data points for the writing analysis: (a) holistic; (b) paragraph coherence; (c) agreement errors; d) punctuation errors; (e) spelling errors; (f) capitalization errors; (g) nominal clauses; (h) nominal phrases; (i) relative clauses; (j) modifying phrases; (k) adverbial clauses, 1) adverbial phrases, and m) inter-T-unit coordination. A description of each of the 13 writing scores follows. These procedures are based on the procedures detailed by the NAEP Scoring Guidelines from Results from the Third Assessment of Writing (NAEP, 1980) Holistic Score The holistic scoring procedure involved several steps. The investigator and one other person served as the raters.
89 They began by discussing the guidelines given and the sample anchor papers provided in the NAEP scoring guide. Examples of each of the four scoring categories were read and discussed. These examples were those provided in the NAEP released exercise set. A number of papers from the prestudy writing set were read and scored and discussed by both raters. When reasonable agreement was reached, the actual papers were scored in one sitting. There were periodic checks throughout the scoring session to maintain consistency. The inter-rater reliability was determined to be .85 for prewriting samples and .88 for the postwriting samples using the Pearson Product-Moment correlation procedure. The two scores were averaged for each student for each writing session. The range of scores for each sample on the holistic category was from a low of 1 to a high of 4. The following score point categories were used for the holistic scoring of the descriptive writing samples evaluated in this study. These categories were taken from the holistic scoring procedures of "Describe Something" as found in Results from the Third National Writing Assessment (1980) : Score of 4 These papers choose a single object and describe it with concrete, clear language. They contain considerable detail and substance, originality of language, and some sense of structure. There may be a few minor mechanical problems. They will often have focus, (p. 58)
90 Score of 3 These papers choose a single object and describe it clearly, though with less detail, originality, or focus than the 4 papers. There may be little sense of organization, but the object should be individualized and mechanical problems should be relatively minor (unless the paper is very strong) (p. 58) Score of 2 These papers do describe something but are thin, general, and often very short and/or confused, (p. 58) Score of 1 Papers scored as 1 are very brief, non-descriptive, and confused. They contain serious errors in syntax, diction, and mechanics, (p. 58) Paragraph Coherence Score The procedure for evaluating paragraph coherence was similar to that used in the holistic evaluation. Each of the two raters read and discussed the scoring criteria provided in the NAEP scoring guide. A number of papers from the October prestudy writing were scored and discussed. When agreement was reached, the actual papers were scored in one sitting. Again there were frequent checks for consistency throughout the scoring session. The inter-rater reliability score determined by Pearson Product-Moment correlation was .87 for the prewriting samples and .78 for the postwriting samples. The two scores were averaged for each student for each writing session which produced one paragraph coherence score for the prewriting sample and one for the postwriting sample. Again, the range of scores for
91 each sample on the paragraph coherence category was from a low of 1 to a high of 4. Descriptions of the four score-points taken from the NAEP standard cohesion-scoring scale (Mullis & Mellon, 1980) follow: Score of 4 Coherence. While there may be a sense of sections within the piece of writing, the sheer number and variety of cohesion strategies bind the details and sections into a wholeness. This sense of wholeness can be achieved by a saturation of syntactic repetition throughout the entire piece and/or by closure which retrospectively orders the entire piece and/ or by general statements which organize the whole piece, (p. 26) Score of 3 Cohesion. Details are both gathered and ordered. Cohesion is achieved in the ways illustrated briefly in the definition below. Cohesion does not necessarily lead to coherence to the successful binding of parts so that the sense of the whole discourse is greater than the sense of the parts. In pieces of writing that are cohesive rather than coherent, there are large sections of details which cohere but these sections stand apart as sections, (p. 26) Score of 2 Attempts at cohesion. There is evidence of gathering details but little or no evidence that these details are meaningfully ordered. In other words, very little seems lost if the details were arranged, (p. 26) Score of 1 Little or no evidence of cohesion. Basically, clauses and sentences are connected beyond pairings, (p. 26) Mechanics Scores The mechanics scores were composed of the numbers of errors which occurred in the following categories:
92 agreement, punctuation, spelling, and capitalization. As these errors were determined by reference to the rules listed in the NAEP scoring guide for syntax and mechanics and were counted by the investigator and checked by another scorer, there was no need to calculate inter-rater reliability for these scoring technigues (Mullis & Mellon, 1980) One score was recorded for each type of error made on each paper. The mechanical errors were described in the guidelines in the following manner (NAEP, 1980) : Agreement Errors A sentence where at least one of the following is present: subject/verb do not agree, pronoun/ antecedent do not agree; noun/modifier do not agree; subject /object pronoun misused; and/ or verb tense shift, (p. 95) Punctuation Errors Every error of commission and error of omission was scored for commas, dashes, guotation marks, semicolons, apostrophes, and end marks. The most informal rules of usage were used with the writer receiving the benefit of any doubt, (p. 96) Spelling Errors In addition to a misspelling, this category includes word division errors at the end of a line, two words written as one, one word written as two, superfluous plurals, and groups of distinguishable letters that do not make a legitimate word. (p. 96) Capitalization Errors A word is given a capitalization error score if the first word in a sentence is not capitalized, if a proper noun or adjective within a sentence is not capitalized, and if the pronoun "I" is not capitalized, (p. 96)
93 Syntactic Score A syntactic analysis was completed on each essay by following the procedures listed below: 1. Counting the number of T-units in each essay 2. Identifying the embeddings within each T-unit as to whether each was nominal, adjectival, or adverbial 3. Identifying the inter-T-unit conjoinings The syntactic categories that were analyzed are described below in descriptions taken from the NAEP scoring guides for describing syntax (Mullis & Mellon, 1980) Examples from that scoring guide are included for explanation. T-Unit A T-unit was defined as "one main clause with all the subordinate clauses attached to it" (Hunt, 1965, p. 20). The underlined portion of the following sentence is a T-unit: "The girl is holding a large jar with fireflies inside it that look like butterflies and she is wondering whether she should change her mind and let them go." (Mullis & Mellon, 1980, p. 4) Nominal clauses There are two types of nominal clauses; these are f active clauses and guestion clauses. These clauses may take the position that a nominal might take: subject, object, object of a preposition, subject complement, and appositive. The underlined constructions in the following examples are nominal clauses: "She told us that we would have to pay the entire bill (Factive nominal clause functioning as object of "told") "Some people never learn who their real friends are ." (Question nominal clause functioning as object of" learn")
94 Nominal Phrases Unlike nominal clauses, nominal phrases contain verbs in nonfinite (uninf lected) forms, but they function in the same nominal positions as nominal clauses. The forms of nominal phrases are the infinitive and the gerund. Example: infinitive : It is fun to go to the beach Example: gerund: Running the mile in nine minutes is exhilarating. Relative Clauses Relative clauses are sometimes referred to as adjectival clauses. There were three kinds considered here: restrictive relatives; nonrestrictive relatives; and adverbial clauses of time, place, or manner. Some examples are: "There are few people who know about this illness ." (restrictive relative) "Students don't like doing assignments that simply are busvwork (that relative) "He went to the store, which is on the corner of Third and Elm, to get the medicine" (nonrestrictive) "John and Mark got into a lot of trouble when they got home" (time) Modifying Phrases Phrases which modify or describe and are reduced relative clauses, that is clauses whose relativepronoun subjects and often whose verbs have been deleted, are called modifying phrases. Example: "The garden has three rose bushes blooming by the wall ." Adverbial Clauses These include adverbial clauses other than relative clauses of time, place and manner occurring in the semantic categories of reason, condition, or concession. Some examples are: "He is getting tired of school because he doesn't like doing his homework" (clause of reason) "The bus was late so we walked to school (clause of reason-purpose)
95 Adverbial Phrases Those phrases result from the reduction of adverbial clauses, and occur in the same semantic categories. Some examples are: Because of the rain we couldn't go the picnic." (reason) "We had to go to the bank to see if we could cash a check ." (purpose) Inter-T-Unit Coordination Whole T-units may be conjoined semantically by a coordinating conjunction ("and," "or," and "nor") as well as by connective words called conjunctive adverbs. Each of the 13 variables described above was tabulated for each of the 54 prewriting samples and the 54 postwriting samples. These scores provided the data points which were used in the statistical analysis of the writing. To correct for the varying lengths of the essays, each of the 11 of the 13 writing variables (holistic and paragraph coherence scores excluded) was converted following Mellon (1969) into a ratio of mechanical errors or syntactic constructions per 100 units. The holistic and paragraph coherence scores were assumed to be independent of the length of the essay and, therefore, not converted. The 1980 scoring for the descriptive passage was a holistic one. The scoring points were as described here. The other writing tasks were informative and persuasive. The researcher of this study selected the descriptive task as current research (Atwell, 1987; Britton, 197 5; Elbow, 1981; Moffett, 1983; Romano, 1987) discuss how personal writing and topics of interest to students enable them to connect with the world of writing more easily than informative/expository
96 or persuasive discourse. Because the high-risk students in this study were poor readers and writers, descriptive writing offered a more appropriate choice of discourse. These categories of discourse are listed in the 1990 (Applebee et al.) report of the 1984-88 assessment of the National Assessment of Educational Progress as personal/ imaginative writing, narrative writing, informative writing, and persuasive writing. The difference in the scoring in the 1984-88 scoring (Applebee, 1990) is that the more recent evaluation is accomplished by developing a primary trait scoring guide for each writing task. These guides typically define five levels of task accomplishment as compared with the four levels of holistic evaluation found in the 1980 scoring. The most recent scoring guide for levels of task accomplishment as follows: Levels of Task Accomplishment Score Elaborated Students providing elaborated responses went beyond the essential, reflecting a higher level of coherence and providing more detail to support the points made. Adeguate Students providing adeguate responses included the information and ideas necessary to accomplish the underlying task and were considered likely to be effective in achieving the desired purpose. Minimal Students writing at the minimal level recognized some or all of the elements needed to complete the task but did not manage these elements well enough to assure
97 that the purpose of the task would be achieved 1 Unsatisfactory Students who wrote papers judged as unsatisfactory provided very abbreviated, circular, or disjointed responses that did not even begin to address the writing task. Not rated A small percentage of the responses were blank, indecipherable, or completely off task, or contained a statement to the effect that the student did not know how to do the task; these responses were not rated. (Applebee, 1990, p. 105) A group of trained readers carried out the primary trait scoring over a period of several months. The fact that only twenty percent of the responses were rescored by a second reader to give an estimate of interreader reliability shows the amount of time consumed by such evaluation. The researcher of this study scored all of the essays for the eleven mechanical and syntactic factors counting all of the appropriate elements of each essay. This task took a considerable amount of time. In fact, two summers were devoted to analysis of the data and the scoring procedures. This may offer one reason why the 1984-88 assessment (Applebee et al., 1990) did not include these complicated procedures for assessing syntactic and mechanical factors. In addition, the scoring on the holistic and the coherence factors from the 1980 assessment are now subsumed under the primary trait scoring guides for each of the types of discourse. Although the previous assessment (1980) provided more detailed information on syntax and mechanics, as well
98 as the assessment of quality and coherence, the current assessment published in 1990 (Applebee et al.) appears to focus on the particular features of each type of discourse. Writing Apprehension Survey The students were given the Writing Apprehension Survey (WAS) during the week of pretesting on the reading and writing measures and again during the week of posttesting on the reading and writing measures. The WAS was developed and validated by Daly and Miller (1975a) To measure writing apprehension, Daly and Miller developed a Likert-type scale format consisting of 63 items each with 5 possible responses. Items were modeled after those in use in the measurement of communication apprehension (McCroskey, 1970) unwillingness to communicate (Heston & Paterline, 1974) receiver apprehension (Wheless, 1974) and general public speaking apprehension (McCroskey, 1970) Items were so constructed that they dealt with some form of apprehension about writing. The valences of the items were random in an attempt to avoid any response bias due to the direction of items. The 63-item instrument was completed by 164 undergraduates enrolled in basic composition courses and interpersonal communication courses in Spring 1974. Data from the Daly and Miller instrument were submitted to principal components factor analysis with orthogonal
99 rotation as determined by Kaiser's varimax criterion (Kaiser, 1958). An eigenvalue of 1.0 was set as termination of factor extraction. To retain a factor it had to have at least two items with satisfactory loadings of above .40. After a factor purification procedure was completed, 2 6 items, all with loadings above .60, were selected to compose the final instrument. The reliability of the instrument was obtained by a split-half technigue. The corrected reliability was .94. The test-retest reliability of the instrument over a week was .92. To establish validity, the researchers had 176 adults complete the 26-item instrument and then a one-way analysis of variance in writing reguirements with 3 levels of apprehension was used to test for the hypothesized differences. Individuals scoring one standard deviation above the mean on writing apprehension were classified as high apprehensives, while those scoring one standard deviation below the mean were classified as low apprehensive individuals. The remaining subjects were classified as moderate in apprehension. Regular Reading Instruction To evaluate differences in a reading program which includes a sentence-combining curriculum and one which does not, the researcher had all students in this study utilize the same learning program, the Random House High Intensity Learning System (HILS) a reading management system. The
100 HILS system covers skills in word study, vocabulary, comprehension, and study skills. In this program, students take an entry level diagnostic test, the Basic Test of Reading Comprehension (BTRC) which groups them according to reading ability. On the basis of this entry level test, students are categorized in one of the following groups: functionally illiterate, beginning reader, adequate reader, good reader, or very strong reader. These categories are used only for assigning materials. Students do not work in ability-identified groups. All work is individualized. Students receive a set of six or seven pretests over the skills previously mentioned. If the student passes a skill, the skill is recorded as mastered. Mastery level is set at 81%. If a skill is not passed or mastered (the student scored below 81%) the student is assigned work to help him learn the skill. The teacher instructs as needed to explain or give examples. After completing the assigned work, the student takes a posttest called a check-out test on that skill to prove mastery. If the student does not pass the check-out test, he is given more instruction and practice. This system was used 4 days a week for the reading only group with the 5th day of the week being reserved for selected recreational reading from books provided in reading level controlled kits published by Random House. This management system was used 1 day per week for the 2
101 experimental writing groups with the 5th day being the recreational reading as outlined above. Experimental Writing Curriculum Rationale As outlined in his review of sentence-combining instructional research, Lawlor (198 0) summarized several instructional implications of that research. First, he predicted that teachers utilizing a systematic sentencecombining program could expect their students to write more mature sentences. He cited Cooper (1973) who noted that sentence-combining instruction may accelerate the process by which a student acquires more mature syntactic processes. Second, Lawlor addressed the possibility that sentencecombining instruction in and of itself may not be able to accomplish that. Hence, Lawlor suggested that sentence combining be part of a total English program, not an isolated activity. A third point that Lawlor made was to reiterate the research findings that knowledge of formal grammar is not a guarantee of mature writing skills. The instructional implication was that sentence combining requires synthesis of sentence elements already provided thus giving students confidence. Finally, Lawlor noted the stimulating and motivational effect of sentence-combining instruction on students. When researchers used O 1 Hare's cuing system, Lawlor noted
102 sentence-combining exercises resembled puzzles or games and stimulated creativity. In general, the curriculum developed by this researcher followed the four conditions for designing sentencecombining instruction established by Mellon (1969) : 1. Instruction should be "a-rhetorical" ; i.e., students should be freed from the burden of generating their own content so that they can attend to syntax. 2 Instruction should present students with sentences whose structure is more mature than those the student is already producing. (Mellon criticized traditional grammar texts for their reliance on simple, immature sentences that are amenable to grammatical analysis but provide poor models for students.) 3 The ultimate outcome of instruction should be the "pseudo-production" of fully formed sentences with a predetermined structure. 4 The content of the sentences to be produced should be provided to the student at the onset of the exercise. Specifications Content One of the greatest problems faced by teachers of remedial, or "high-risk," students as they are designated in
103 this study, is that of motivating the students to work. Because of previous failure in traditional grammar study and writing programs that stress error hunting or the "red-ink" syndrome, high-risk students often exhibit negative feelings or perceptions about any writing activity. Any program that was to be successful with the high-risk students in this study had to be motivational. Therefore, this researcher sought descriptions of research and instructional programs that were geared to interest the high-risk learner. With the idea of motivation in mind, this researcher also sought a program that would fit the following description by Perron (1974) : The structured individualized lessons were successful not only in enhancing syntactic growth, but also in motivating students. The O'Hare system of cuing was easy for young students to grasp, and they seemed to enjoy working with stories, humorous sentences, and the problems that used students' names, (p. 171) Perron's study included semi-structured games and activities that were particularly useful in stimulating discussions and arguments on the relative merits of various syntactic operations. In fact, the students in the study seemed to prefer the oral discussion involved in choosing among several options provided in sentence combining. This researcher decided that any writing program would include opportunity for students to discuss the options when combining sentences.
104 An additional curriculum which greatly aided this researcher was the workbook written by Weinstein (1982) which contained 68 sentence-combining lessons entitled Sentence Combining Workbook Designed as a supplement to traditional writing texts, this workbook contains activities that allow students to create more detailed sentences by adding (a) single and multiple adjectives; (b) adjectival prepositional phrases; (c) adverbs; (d) adverbial prepositional phrases (e) noun, adverb, and adjective clauses; (f) participial, gerund, and infinitive phrases; (g) appositives and h) single word, phrase, and clause series. The workbook also contains a semester review and a comprehensive review of the exercises. The attraction of Weinstein' s workbook was that it was designed for remedial high school students who were described as being much like the students in the study and contained specific suggestions as to how these students should be introduced to sentence combining and how the lessons should be presented. The following suggestions by Weinstein appeared to support the research found by this researcher and reviewed in Chapter II and thus were incorporated into the experimental writing curriculum: 1. Do duplicate each section separately rather than as a bound book. With remedial students, a book with 75 pages of exercises is a threat. A single page with 5-8 sets of sentences is less so. 2. Do not hand these out as homework. At least for remedial classes, students need help as they work through the exercises. For sentence combining to be
105 effective, the students need oral feedback, whether it comes from you, from their peers, or from themselves. 3. Do correct the revised sentences in class, out loud. This allows students to discuss variants in writing. 4. Do prepare an introduction to each group of sections. That does not mean lengthy discussions of the grammatical rule involved. The most effective means is merely to write one additional sample, present it, and work it out with the class. Your example should include any signals used in the set. (p. 5) As Weinstein was aware of the motivational value of using students' names and topics of interest to them, he included the names of his students in his exercises. He found as Perron (1974) did that such exercises interested students more and kept them actively involved in finding different ways to combine the sentences. This researcher chose to also include this strategy as a feature of the experimental writing programs. Thus, the lessons contained actual students' names and items which would be of interest to them. Cues or Signals Because of the nature of the students who participated in this study, the researcher chose to follow the admonition of Weinstein (1982) and use only four basic signals for students to utilize in the combining of sentences. These signals were as follows, with examples given by Weinstein: 1. Underline words. Add this word or phrase to the first, base sentence. I have a dog The dog is brown.
106 I have a brown dog. 2. (,) (, and), Insert this before the next word or phrase. I have a dog. The dog is brown. I have a brown, furry dog. It is furry. ( ) 3. SOMETHING. Delete this from the revised sentence. SOMETHING is a dog's best defense. It is barking. Barking is a dog's best defense. 4. (WHO) (WHEN) (THAT). Include this in the revised sentence. XXXXXX. Delete this from the revised sentence. Tina is my sister. Tina is fifteen. (WHO) Tina who is fifteen is my sister Weinstein (1982) used these cues or signals to help the students generate the new, revised sentence. Most authorities recommend both signaled or cued and unsignaled or uncued exercises. To support this rationale, Lawlor (1980) stated, "Signaled exercises can provide focused practice on particular syntactic operations, and "Unsignaled exercises allow students to experiment with various combining strategies, selecting the one combination that seems to work best." Mellon (1979) recommended a ratio
107 of one unsignaled exercise to every four signaled exercises. This researcher used the cued exercises to present the week's strategies and then had the students practice the uncued type; the review lesson for the week included both cued and uncued exercises. Order or sequence of program The order in which the experimental writing program was to be presented was of importance because the researcher wished to present the material in a manner most conducive to the syntactic maturity of the students. Thus, the researcher looked at the seguencing of the major sentencecombining research projects that have been conducted. In addition, the researcher read authorities on composition and especially sentence combining. Mellon (1969) concluded that the development of syntactic maturity generally proceeds through three stages which are as follows: (a) elaboration through separate sentences; (b) elaboration through contained clauses; (c) elaboration through reduced clauses (e.g., prepositional phrases, absolutes) In his research with 7th-grade students, he used the following seguence in keeping with the above conclusions: (a) guest ion transformations and other single-sentence operations (passive transformations) ; (b) nominal embeddings, including f active and interrogative noun clauses, if inversions, infinitives, gerunds, and derived nouns; (c) relative-clause embeddings; (d) participial
108 phrases; (e) infinitives (as adjective modifiers) ; (f ) appositives; (g) prenominal adjectives and participles. O'Hare (1973) generally followed Mellon' s sequence for instruction. In his sentence-combining textbook, Sentencecraft (O'Hare, 1975) however, he made two changes. He included an introduction on coordinate predicate phrases and coordinate free modifiers. In addition, he included adverbial clauses and phrasal adverbial modifiers, most of which were free modifiers. Cooper's program (1973) included these items: (a) prenominal adjectives, participles, and compound adjectives; (b) postnominal prepositional phrases, participial phrases, and infinitives; (c) reduced-clause adjectival modifiers; (d) full relative clauses; (e) factive and interrogative noun clauses; (f) gerunds; and (g) infinitives. Lawlor (1980) presented his sequence as tentative and as needing more research. His sequence included the following: a) coordinate conjoining and adverbial-clause conjoining; b) adjectival embeddings consisting of restrictive relative clauses; c) phrasal adjectivals, single-word adjectivals, nominal embeddings consisting of full clauses, phrases, single words, free modifiers and cumulative sentences. Lawlor listed the four ways of combining information: coordination, subordination (adverbial -because) adjectival embedding operations, and nominal-embedding operations. Of
109 the four types of combining processes, he cited that research has shown that they are not equally effective in achieving the succinctness of expression characteristic of mature writing. It is surprising, therefore, that he began his instructional program with coordinate conjoining in light of the fact that several researchers have noted that the excessive use of coordinating conjunctions is a characteristic of immature writing (Hunt, 1965, 1970; O'Donnell et al., 1967). Adverbial conjoining, too, was shown to be an unreliable index of maturity. However, adjectival and nominal embeddings were found to be strong indicators of syntactic maturity. Hunt (1965) noted that these two types of sentence-combining operations increased dramatically in the writing of older students. Lawlor concluded what research has shown on the results of sentence-combining programs by stating the following, "Thus what differentiates older writers from younger writers is not the kind of structures they use, but the frequency with which they use certain structures" (p. 8) Most programs appeared to move from single adjectives to multiple adjectives, then to clauses and phrases. The work by Cooper (1973) and Weinstein (1982) seemed to offer the most logical order and follow the natural progression of syntactic development described by Mellon (1969)
110 Also, as the researcher wished to test factors proven to be links between reading achievement and writing performance as reported by Hill (1982), she separated the two curricula into one which dealt with nominal clauses and factors related to coherence at the paragraph level and the other which focused on nominal and modifying phrases at the sentence level. Both writing curricula began with single adjectives and then multiple adjectives in order to establish a base of knowledge of technigues in sentence combining and then focused on the different syntactic structures in the ensuing lessons. Amount of Instructional Time The amount of instructional time spent in sentencecombining strategies has been the subject of considerable debate. Lomax (1980) suggested that as little as five to ten minutes of sentence-combining work three times per week was sufficient. Lawlor (1980) stated that, "Thirty minutes of instructional time per week would seem the minimum time with 90 minutes the maximum." Mellon (1969) suggested 50 minutes per week in his study with 7th graders in which students completed two cued problems daily and two whole discourse exercises each week. O'Hare (1973) advocated 75 minutes per week supplemented by 3 minutes of homework weekly. The researcher of this study decided to allot three 50-minute sessions per week for 8 weeks to each of the two sentence-
Ill combining curricula in order to provide sufficient time for instruction, practice, and discussion of possible answers for the combining of the sentences. Regular English Program Writing Instruction Each student in this study was also concurrently enrolled in an English course at the 10th-, 11th-, or 12thgrade level. These English courses were part of the Florida Writing Enhancement Program. Teachers in this program taught no more than 100 students per day and were required to assign and grade one composition or piece of writing per student per week. Students in this research study were all taught the same curriculum at each grade in this writing program which was prescribed by county guidelines and state objectives. No instruction was given in sentence combining in the regular English program. Any growth in sentencecombining ability would be assumed by this researcher to thus be a result of the experimental writing program written by this researcher and not the result of the regular English program. Instruction and Data Collection As most sentence-combining research has focused on sentence combining or no sentence combining, this researcher wanted to test the hypothesis that one type of sentencecombining instruction might produce more mature writing than another type. Hence, the two sentence-combining
112 curricula each had as its own focus one of the three writing factors which Hill (1982) found were predictive of reading ability. This researcher excluded the third writing factor, writing mechanics, a factor which researchers have found should not be handled didactically but indirectly by example or as byproduct of the sentence-combining instruction. Thus, each of the two sentence-combining curricula gave students practice in proper punctuation and spelling by providing models which could be copied, learned, and assimilated. No grammatical terminology was taught to students. The types of combinings were identified but no attempt was made to have students learn or memorize grammatical terminology or rules of punctuation and capitalization. As both curricula handled the inclusion of mechanics in the same manner, they differed only in the types of synthetic factors presented. Group B: Sentence Combining with Nominal /Modifying Phrases Reading Program Group 1. The content of this group was nominal and modifying phrases taught at the sentence level. 2. During the first week of instruction, subjects in Group A were introduced to sentence-combining strategies. The researcher who was also their regular classroom teacher met with them Monday through Friday for 50 minutes per class session each day. On Monday subjects utilized the Random House High Intensity Reading Management System.
113 On Friday the subjects used the entire class period for selected reading. On Tuesday the researcher explained what sentence combining is, what its purposes are, and gave samples for subjects to see. The four basic cues or signals which the experimental subjects in Group B would use in the sentence-combining lessons were explained with examples. 3. Then sentences containing simple adjectives were introduced to be combined into a single sentence. The format of this lesson was followed throughout the rest of the study: (a) introduction of the sentence-combining strategy to be learned; (b) examples presented and discussed by the teacher; (c) a worksheet of ten sets of sentences to combine; (d) and discussion on how the combinings were done and alternate combinings. 4. On Wednesday sentences containing multiple adjectives were shown. Again, cues were given and explained. The format as described was followed. 5. On Thursday a review lesson on single and multiple adjectives was presented. Both cued and uncued sentences were given to aid students in being able handle the strategies on their own.
114 6. Each week thereafter during the study the syntactic factor of the week was presented, practiced, and discussed in this same format. Group C: Reading Program Only Group Group C served as the control group and received no sentence-combining instruction. Subjects in Group C received 4 days of 50 minutes each out of 5 days each week for the 8 weeks in instruction and practice in the Random House High Intensity Reading Management System and one day out of five days in a 50-minute period of selected reading. Group A: Sentence Combining with Nominal Clauses and Factors Leading to Coherence in Writing and Reading Program Group The content of Group A was nominal clauses and factors leading to coherence in writing. The introduction of sentence combining for this group followed the same procedures as Group B for the first week. Each week thereafter the syntactic factor for the week was presented, practiced, and discussed. For the first four weeks of the study the factors for Group A were combined into single sentences. In the 5th week of the study, instead of merely combining sets of sentences into single sentences, subjects were given approximately 20 sentences on the review day for the week which they were to combine into a paragraph. They were to combine sentences which seemed short and choppy or redundant. The three days of sentence combining of the last week of Group A's instruction (week 8) were devoted to
115 combining sets of sentences into a coherent paragraph of mature sentences. Chapter IV presents the analysis of the data.
CHAPTER IV ANALYSIS OF THE DATA Review of the Study This study was concerned with three questions. First, would high-risk secondary reading students trained in sentence combining perform significantly better on reading achievement and descriptive writing performance than those who were instructed only on a reading management system? Second, which type of sentence-combining curriculum, one using nominal clauses and coherent writing factors, or one using nominal/modifying phrases, produces the better student reading achievement scores and writing performance? Finally, would sentence-combining instruction have a positive effect on lowering the writing apprehension of high-risk secondary students? In order to answer these questions, the researcher randomly assigned three intact classes of remedial readers in the 10th, 11th and 12th grades to one of three instructional programs in reading. Students in Group 1 became Experimental B and studied the regular reading program 2 days per week and a sentence-combining curriculum emphasizing nominal/modifying phrases at the sentence level 3 days per week for 8 weeks. Students in Group 2 became 116
117 Control Group C and studied only the regular reading program 5 days per week for 8 weeks. Students in Group 3 became Experimental A and studied the regular reading program 2 days per week and a sentence-combining curriculum which emphasized nominal clauses and factors leading to coherent writing at the paragraph level 3 days per week for 8 weeks. All subjects in this study were given a pretest with the Metropolitan Achievement Test, Reading Survey (1979 ed.), Advanced I, Form KS, containing six subscores and a total reading score. Posttest scores were collected on this same measure at the end of the 8-week study. The reading subscores and total reading scores were tabulated by the test publishers. Descriptive writing samples, which were taken before and after the treatment, were analyzed for syntactic analysis following the National Assessment of Educational Progress Writing Assessment guidelines as outlined by Mullis and Mellon (1980, pp. 6-8). Scores and postscores on the Daly and Miller Writing Apprehension Survey (1975a) were also collected before and after the study. The reading achievement data, the descriptive writing data, and the writing apprehension data were analyzed using the nonrandomized control group, pretestposttest design. Group membership (type of reading instruction) served as the grouping factor. The Writing Apprehension Survey scores were calculated by the researcher using the guidelines outlined by Daly and Miller (1975a, p. 58)
118 Pretest scores were used in subsequent analyses of covariance to partially adjust for nonequivalence of the groups. Separate analyses of covariance were completed on the following dependent variables: reading achievement, writing performance in the descriptive mode and student apprehension about writing. Reading performance was further subdivided into six subscores and a total reading score. The reading subscores were: vocabulary; literal specific; literal global; inferential specific; inferential global; and evaluative. Therefore, seven separate analyses of covariance were completed on reading achievement. Writing performance on the descriptive samples was subdivided by the NAEP Writing Assessment into 13 writing categories: holistic; paragraph coherence; agreement; punctuation; spelling; capitalization; nominal clauses; nominal phrases; relative clauses; modifying phrases; adverbial clauses; adverbial phrases; and inter-T-unit coordination. Therefore, there were 13 separate analyses of covariance completed on writing performance. With the last analysis of covariance on writing apprehension, there were, in all, 21 separate analyses of covariance completed. Three hypotheses were tested. Statistical analyses were completed utilizing the Statistical Analysis System, (Helwig, 1978) When the null hypothesis was rejected, follow-up comparisons were
119 calculated by the researcher. Results from these statistical analyses are presented in this chapter. Findings Related to the Hypotheses An analysis of covariance was completed on each of the reading subscores and the total reading score. The descriptive statistics including the mean, standard deviation, and range for each of the six subscores and the total reading score of the MAT for each of the groups in the study on the pretest and the posttest are presented in the appendix. The following is a presentation of the hypothesis tested and a discussion of the findings. The following hypothesis, stated in the null form, was tested at the .05 level of confidence. The following hypothesis was tested for each of the reading variables. The hypothesis will be stated completely here and then referred to in subseguent discussions of each reading variable as Hypothesis I. Hypothesis I After controlling for initial reading achievement as measured on the subscores and the total reading score of the Metropolitan Achievement Test, Reading Survey (1979), there are no significant differences in reading achievement among subjects using two experimental writing curricula and those who are not.
120 Vocabulary The raw score on the pretest for vocabulary on the reading section of the MAT was used as the covariate. An analysis of variance test for the difference between these means was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable. Literal-Specific The raw score for literal specific on the pretest on the reading section of the MAT was used as the covariate. An analysis of variance test for initial differences was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there was not a significant treatment effect. Literal Global The raw score on the pretest for literal global on the reading section of the MAT was used as the covariate. An analysis of variance test for initial difference between these means was not significant. The test for interaction between the covariate and the levels of the treatment was not signifiant. The test for treatment effect was not significant. Therefore, the null hypothesis
121 was not rejected and it was concluded that there were no differences on the treatment on this variable. Inferential Specific The raw score on the pretest for inferential specific on the reading section of the MAT was used as the covariate. An analysis of variance test for initial differences between these means was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on this variable. Inferential Global The raw score on the pretest for inferential global on the reading section of the MAT was used as the covariate. The test for initial difference between the means was not significant. The test for interaction between the covariate and the levels of the treatment was significant, F (2, 48) =3. 47, p=<.03. There is an interaction present; the performance on this dependent variable by subjects in this study is dependent upon the level of the covariate that they began with as measured by this subscore. Figure 4-1 displays the relationship between the dependent variable and the levels of the covariate. The test for treatment effect was not significant. Therefore, the null hypothesis was rejected and it was concluded that the treatment effect depended upon the level of the covariate.
122 25 20H 15 10PRE Gil REGRESSION LINES EXPERIMENTAL B EXPERIMENTAL A cxafTRQt: Group RAW SCORES PRETEST GLOBAL INFERENTIAL INFORMATION Figure 4-1 Plot indicating interaction between the covariate and the levels of the treatment. Evaluative The raw score on the pretest for evaluative on the reading section of the MAT was used as the covariate. An analysis of variance test for initial difference was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not
123 rejected and it concluded that there were no differences on the treatment on this variable. Total reading The raw score on the pretest for total reading on the reading section of the MAT was used as the covariate. The test for initial difference between the means was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable. The following hypothesis was tested for each of the writing variables. The hypothesis will be stated completely here and then referred to in subsequent discussions of each writing variable as Hypothesis II. The descriptive statistics including the mean, standard deviation, and range for each of the 13 writing variables on the descriptive writing performance analyzed by NAEP guidelines for each of the groups in the study on the pretest and the posttest are presented in the appendix. Hypothesis II After controlling for initial performance in descriptive writing as measured by the sample ratings using the scoring procedures of the National Assessment of Educational Progress Writing Assessment guidelines (1980), there are no significant differences in writing performance
124 among subjects using two experimental writing curricula and those who are not. Holistic Score The raw score for holistic on the pretest on the writing sample of the NAEP was used as the covariate. An analysis of variance test for initial differences between the means was not significant. The test for interaction was not significant. The test for treatment was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable. Paragraph Coherence The raw score for paragraph coherence on the pretest on the writing sample of the NAEP was used as the covariate. The test for initial difference between the means was not significant. The test for interaction was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable. Agreement Errors The raw score for agreement errors on the pretest on the writing sample of the NAEP was used as the covariate. The test for initial difference was not significant. The test for interaction was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable.
125 Punctuation Errors The raw score for punctuation errors on the pretest on the writing sample of the NAEP was used as the covariate. The test for initial differences was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable. Spelling Errors The raw score for spelling errors on the pretest on the writing sample of the NAEP was used as the covariate. The test for initial difference between the means was not significant. The test for interaction between the covariate and the levels of treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable. Capitalization Errors The raw score for capitalization errors on the pretest on the writing sample of the NAEP was used for the covariate. The test for initial difference between the means was not significant. The test for interaction between the covariate and the levels of treatment was not significant. The test for treatment effect was not significant. Therefore, the null
126 hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable. Nominal Clauses The raw score for nominal clauses on the pretest on the writing sample on the NAEP was used for the covariate. The test for initial difference was significant, F (2, 51) =3. 13, p=.05. Therefore, the groups were different on this variable before treatment. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable. Nominal Phrases The raw score for nominal phrases on the pretest on the writing sample of the NAEP was used for the covariate. The test for initial difference was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The adjusted cell means and standard deviations for nominal phrases for all groups are presented in Table 4-1. The test for treatment effect was significant, F (3,50)=7.16, p=.001. Therefore, the null hypothesis was rejected and it was concluded that there was a significant treatment effect. Follow-up procedures using the Bonferroni technigue were completed with the error rate set at the .05 level of significance per family. The results of the follow-up comparisons, presented in Table 4-2 indicate that Group 1 or
127 Table 4-1 Adjusted Cell Means and Standard Deviations for Dependent Variable Nominal Phrases GROUP Variable: Nominal Phrases 8.804 (14.071) 0.000 (0.000) 0.000 (0.000) Table 4-2 Comparison of Three Methods for Nominal Phrases using the Bonferroni Procedure Comparisons Confidence Interval Decision Group B vs Group C (2.023, 16.015) Group B vs Group A (2.483, 15.870) Group C vs Group A (-6.532, 6.854) Reject H 1>2 Reject H 1>3 Do not Reject H o 2>3
128 Experimental B, students studying reading plus sentence combining with nominal/modifying phrases scored better thanboth the control group, Group C, which studied reading only and the other experimental group, Group A, that studied reading plus sentence combining with nominal clauses. Relative Clauses The raw score for relative clauses on the pretest on the writing sample of the NAEP was used for the covariate. The pretest means and (standard deviations) for the three groups for this variable were as follows: Group. 1:6.025, (11.117); Group 2 : 0.000 (0.000); Group 3: 23.672 (15.705). An analysis of variance test for the difference between these means was significant, F (2, 51) =6. 84, p=.05. Therefore, the groups were different on this variable before treatment. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, it was concluded that there were no differences on the treatment on this variable. Modifying Phrases The raw score for modifying phrases on the pretest on the writing sample of the NAEP was used for the covariate. The pretest means and (standard deviations) for the three groups for this variable were as follows: Group 1: 5.773 (7.905); Group 2: 2.821 (5.762); and Group 3: 0.000 (0.000). Analysis of variance test for initial difference between these means was significant, F (2,51)= 5.16, p=.009. Therefore, the groups were different
129 on this on this variable before treatment. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded there were no differences on the treatment on this variable. Adverbial Clauses The raw score for adverbial clauses on the pretest on the writing sample of the NAEP was used for the covariate. The test for initial difference was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded there were no differences on the treatment on this variable. Adverbial Phrases The raw score for adverbial phrases on the pretest on the writing sample of the NAEP was used for the covariate. The test for initial difference was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences in the treatment. T-Unit Coordination The raw score for T-unit coordination on the pretest on the writing sample of the NAEP was used for the covariate. The pretest means and (standard deviations) for the three groups for this variable
130 were as follows: Group 1: 41.997 (27.823); Group 2: 24.053 (18.745); Group 3: 29.651 (17.647). The test for initial difference was significant, F (2,51)=3.06, p=.05. Therefore, the groups were different on this variable before treatment. The test for interaction between the covariate and the levels of treatment was not significant. The test for treatment was not significant. Therefore, the null hypothesis was not rejected and it was concluded that there were no differences on the treatment on this variable. The following variables were not included in the original syntactic analysis as they were not reported by the NAEP assessment. Information concerning these variables was collected by this researcher as this information might produce an evaluation of fluency (number of words written) and maturity (number of T-units) Indeed, useful information was secured from the analyses of these items and, thus, is presented here as support for the research on the usefulness of sentence combining. Number of T-units The raw score on number of T-units on the pretest on the writing sample of the NAEP was used for the covariate. The test for initial difference was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The test for treatment was not significant at the preset level (.05). It was concluded that there were no significant differences on the treatment on this variable.
131 Total Number of Words Written The raw score for total number of words written on the pretest on the writing sample of the NAEP was used for the covariate. The test for initial difference between the means was not significant. The test for interaction between the covariate and the levels of the treatment was not significant. The adjusted cell means and standard deviations for total number of words written for all groups are presented in Table 4-3. The test for treatment was significant, F (3,50-6.25, p=.003. Therefore, the null hypothesis was rejected and it was concluded that there was a significant treatment effect. Follow-up procedures using the Bonferroni technique were completed with the error rate set at the .05 level of significance per family. The results of the follow-up comparisons, presented in Table 4-4 indicate that Group 3 or Experimental A, students who studied reading and a sentencecombining curriculum which emphasized nominal clauses and coherent writing factors, were more fluent (wrote more) than the control group or the other experimental group, Group B. Writing Apprehension The following hypothesis was tested for the variable, writing apprehension. The descriptive statistics including the mean, standard deviation, and range for this variable for each of the groups in the study on the pretest and the posttest are presented in the appendix.
132 Table 4-3 Adjusted Cell Means and Standard Deviations for Dependent Variable Total Number of Words Group Variable: Total # of Words 128.557 (56.088) 148.271 (44.639) 164.731 (25.754) Table 4-4 Comparisons of Three Methods for Total Number of Words Using the Bonferroni Procedure Comparison Confidence Interval Decision Group B vs Group C (-227.787, 281.479) Do not Reject H Q 1=2 Group B vs Group A (-383.539, -434.186) Reject H Q 1<3 Group C vs Group A (-147.600, -199.750) Reject H Q 2<3
133 Hypothesis III After controlling for initial apprehension as measured by the Writing Apprehension Survey (Daly & Miller, 1975a), there are no significant differences in writing apprehension among subjects using two experimental writing curricula and those who are not. The raw score on writing apprehension on the pretest on the Writing Apprehension Survey (WAS) was used for the covariate. The test for initial differences was not significant. The test for the interaction between the covariate and the levels of the treatment was not significant. The test for treatment effect was not significant. Therefore, it was concluded that there were no differences on this variable. Chapter V summarizes the findings of the study and discusses both theoretical and educational implications. Recommendations for future research are also made.
CHAPTER V SUMMARY, IMPLICATIONS, CONCLUSIONS AND RECOMMENDATIONS Summary This study examined the effects of sentence-combining instruction on the reading comprehension, descriptive writing ability and the writing apprehension of high-risk secondary students. The investigation sought to determine if students trained in sentence combining would perform significantly better on reading achievement and descriptive writing performance than those who were instructed only in a reading system. A second question investigated which type of sentence-combining instruction would be the more effective. Finally, the study sought to determine if the sentence-combining instruction would lower the writing apprehension of these students. The sample consisted of 54 high-risk secondary students enrolled in a remedial reading course in grades 10-12. Subjects were randomly assigned by classes to one of three groups: sentence combining with nominal/modifying phrases at the sentence level with reading; regular reading curriculum; or sentence combining with nominal clauses and factors leading to coherent writing at the paragraph level with reading. Reading scores from the Metropolitan Achievement 134
135 Test, descriptive writing samples, and scores on the Writing Apprehension Survey were collected for all subjects. Students assigned to Group 1 became Experimental B and studied the regular reading program 2 days per week and a sentence-combining curriculum emphasizing nominal/modifying phrases at the sentence level 3 days per week for 8 weeks. Students in Group 2 became Control Group C and studied only the regular reading program 5 days per week for 8 weeks. Students in Group 3 became Experimental A and studied the regular reading program 2 days per week and a sentencecombining curriculum emphasizing nominal clauses and factors leading to coherent writing at the paragraph level 3 days per week. Data were analyzed using the nonrandomized noneguivalent pretest posttest design. The reading section of the Metropolitan Achievement Test, the Descriptive Writing sample from the National Assessment of Educational Progress and the Writing Apprehension Survey served as the covariates. A comparison of groups revealed that Experimental B, the sentence-combining curriculum with nominal/modifying phrases scored better on nominal phrases (p=<.001) and that Experimental A, the sentence-combining curriculum with nominal clauses and factors leading to coherent writing at the paragraph level, scored better on fluency or total number of words (p=<.003).
136 Finally, there was an interaction between levels of the covariate, the pretest raw scores on the reading subscore, Global Inferential Information from the MAT, and the dependent variable, the posttest raw scores on the reading subscore, Global Inferential Information from the MAT. As the pretest scores increase, the gap becomes wider between the control group and the experimental groups. Implications and Conclusions None of the reading variables were significant. This finding supports previous research which reported no gains on standardized reading measures (Searls and Neville, 1988) Those researchers who reported gains utilized researchermade tests or cloze tests. The nominal phrases variable was significant at .05 (actual p=<.003). Group 1 or Experimental B, students studying reading and sentence combining with nominal/modifying phrases scored better than both the control group, Group 2 or C, and the other experimental group, A or 3 whose students studied reading and sentence combining with nominal clauses. The number of variables and covariates used in this study warrants some comment. The analysis of covariance was selected to increase the power of the statistical analysis as ANCOVA presents, in most situations, the smallest error term. Huck, Cormier, and Bounds (1974) cite the increase in power when multiple covariates are used (p. 145). The
137 conditions for this increase in power include the fact that (a) there is a high correlation between each covariate and the dependent variable and that (b) there is a low correlation between each pair of covariates. The first condition was true here as the covariate was the pretest and the dependent variable was the posttest for all variables. The fact that there were so many variables and covariates present might have produced at least one difference by chance. Therefore, the results of this experiment should be judged cautiously. There is the possibility of a Type I error, that is claiming there is a difference when it may not be present. On the reading subscore variable Global Inferential Information, there was an interaction present. The treatments were not equally effective for each group. The effectiveness of the treatment depended upon the placement of the students on the pretest or covariate. The students scoring the lowest on the covariate, the pretest, would have done better if they had been placed in Group 3 which was Experimental A, the group which studied reading and sentence combining with nominal clauses at the paragraph level. Students scoring in the middle and high ranges on the pretest would have done better in Group 1, Experimental B which studied reading and sentence combining with nominal/modifying phrases at the sentence level.
138 Students who scored the lowest on the covariate may need the connectedness of writing to be able to comprehend the main idea and the details of reading passages. As Evans (1986) found that students with the lowest abilities in sentence combining and reading comprehension tended to register higher gains in both areas as a result of sentence combining, so this research may spur researchers to use sentence combining with students' own text to create connected paragraphs. Connected writing rather than isolated drills may help these students see bridges between reading and writing. In fact, the next logical step to teaching students sentence-combining strategies may be to help them combine their own writing. On the total number of words written, a fluency measure, Group 3 or Experimental A students studying reading and sentence combining with nominal clauses at the paragraph level scored significantly better (p<.003) than Experimental B, the group studying reading and sentence combining with nominal/modifying phrases at the sentence level and the control group, Group 2 or C, which studied reading only. Group 3 moved from 12 6 words per writing at pretest to 164 words per writing at posttest. This finding of fluency may help investigators and teachers find ways to reduce apprehension as poor writing tend to not want to write. Perhaps sentence combining using the students' own writing may reduce the fear of writing.
139 Shanahan's (1990) comments about the role that reading and writing have in students' feelings of power and self-esteem have much to tell us about the way in which we teach reading and writing. They are not simply discrete skills isolated from the moral, social, aesthetic and intellectual processes. Reading and writing are integral parts of each student's path to literacy. Unless teachers teach reading and writing as related and interconnected, students will continue to receive disconnected learning. The finding about fluency in this study also supports the research of Whitt (1987) in which she found that sentence combining improves the syntactic fluency of basic students. It also supports the research of Evans (1986) which found that "sentence combining most sensitively distinguished between and measured gains among the reading and writing skills of students who began such instruction with the lowest skills" (p.l). The researcher of this current dissertation also utilized several factors noted to be helpful in improving syntactic maturity in the review of sentence-combining research from 1970-1985 by Searls and Nevile (1988) Factors they found related to significant improvement were the involvement of the researcher in the instructional process at the secondary level, massed practice of at least 2-3 times per week, and treatment of 1-2 months rather than 4-5 months. All of these factors were present here.
140 The one finding of Searls and Neville (1988) not supported here was that elementary studies were found to be more effective than secondary studies. One possible explanation may be that elementary and middle school students have not experienced as much failure as those students in the high-risk category in high school. Perhaps if more secondary studies were conducted, comparative data would be more plentiful and available to test this hypothesis further. Recommendations for Future Research Previous research studies as well as the current study have suggested that sentence-combining instruction can be effective in promoting the development of syntactic maturity in high-risk learners. The instruction should be an integral part of a well-rounded writing program which includes student-selected topics as well as syntactically-controlled sentences to be combined. Research should be conducted using the other modes of discourse: narration, exposition and argumentation. The instruction should include a combination of massed practice for 1-2 months followed by distributed practice for the remainder of the year with pretests, posttests and delayed posttests used to evaluate the effectiveness of the program. Several experimenters should be used to test for experimenter factor. The several experimenters should teach
141 both the experimental classes and the control classes. The students should be randomly assigned to classes before being assigned to treatment conditions. Students should be encouraged to produce text as well as combine text provided. Cued and uncued exercises should be used and time for discussion of possible answers should be built into the study. Exercises should be completed in the oral and written format. Strong (1986) observed that instruction should move from oral rehearsal to written transcription. By doing this he felt students would be transferring power from their primary language system (speaking) to their secondary language system (writing) Students should be taught to see language as a set of options, situation specific and flexible according to mode of discourse, audience and purpose, not as right and wrong answers. Students should explore different ways to combine the same sentences and discuss the advantages and disadvantages of each choice. The researcher should utilize both sentence-combining (synthesis) and analysis of combined sentences. Reading materials might be used which have embedded sentences By analyzing these reading materials, students could see the many embeddings and be able to comprehend more complex passages. Mechanics should be included in the study as a factor and students should look at ways to punctuate
142 combined sentences and discuss various ways to punctuate independent and dependent clauses. Thus, this research points the way to writing curricula which are non-threatening, which enable students to practice syntactic patterns without fear of failure, which utilize group discussion of sentence possibilities, and which reward students for experimentation with language, not punish them for errors.
APPENDIX DESCRIPTIVE STATISTICS METROPOLITAN READING SCORES PRE-TEST (GROUP 1)
Mean SD Min Val Max Val 144 METROPOLITAN READING SCORES POST-TEST (GROUP 1) Vocabulary
Mean SD Min Val 145 Max Val PRE EXPERIMENT CORRECTED WRITING SCORES (GROUP 2) Agreement
Mean SD Min Val 146 Max Val POST EXPERIMENT CORRECTED WRITING SCORES (GROUP 2) Agreement Punctuation Spelling Capitalization Nominal Clauses Nominal Phrases Relative Clauses Modifying Phrases Adverbial Clauses Adverbial Phrases T-Unit Coordination 16
147 PRE EXPERIMENT ACTUAL Holistic-K Holistic-L Paragraph Coher K Paragraph Coher L Actual Agreement Actual Punctuation Actual Spelling Actual Capital Actual Nominal CL Actual Nominal Phr Actual Relative CI Actual Modifying Phr Actual Adverb Cl Actual Adverb Phr Actual T-unit Coord Actual T-Units Actual Number-Words Actual Words/T-Unit PRE EXPERIMENT ACTUAL Holistic-K Holistic-L Paragraph Coh-K Paragraph Coh-L Actual Agreement Actual Punctuation Actual Spelling Actual Capital Actual Nominal Cl Actual Nominal Phr Actual Relative Cl Actual Modifying Phr Actual Adverb Cl Actual Adverb Phr Actual T-Unit Coord Actual T-Units Actual Number-Words Actual Words/T-Unit N
Mean SD Min Val Max Val 149 POST EXPERIMENT ACTUAL WRITING SCORES (GROUP 3) Holistic-K Holistic_L Paragraph Coh-K Paragraph Coh-L Actual Agreement Actual Punctuation Actual Spelling Actual Capitals Actual Nominal Cl Actual Nominal Phr Actual Relative Cl Actual Modify Phr Actual Adverb Cl Actual Adverb Phr Actual T-Unit Coord Actual T-Units Actual Number-Words Actual Words/T-Unit 20
REFERENCES Applebee, A.N. Langer, J. A. Jenkins, L.B., Mullis, I.V., & Foertsch, M.A. (1990). Learning to write in our nation's schools: Instruction and achievement in 1988 at grades 4. 8 and 12 The National Assessment of Educational Progress, Educational Testing Service (Grant No. G008720335) Washington, D.C., U.S. Department of Education. Atwell, N. (1987) In the middle: Writing, reading, and learning with adolescents Portsmouth, NJ: Boynton/Cook. Bateman, D., & Zidonis, F. (1966). The effect of a study of transformational grammar of the writing of ninth and tenth graders. Champaign, IL: National Council of Teachers of English. Birnbaum, J. (1981) A study of reading and writing behaviors of selected fourthand seventh-graders. (Doctoral dissertation, Rutgers University, The State University of New Jersey, 1980) Dissertation Abstracts International 42 152A. Book, V. (1976) Some effects of apprehension on writing performance. Speech presented at the annual meeting of the America Business Communication Association, San Diego, California, 24 pp. (ERIC Document Reproduction Service No. ED 132 595) Bordonaro, L.M. (1991) The effects of a revision technique (sentence combining) on urban fifth-grade students' writing mechanics of the writing process (Doctoral dissertation, The University of Massachusetts, 1991) Dissertation Abstracts International 51 3613A. Brewer, R. (1986) Sentence combining and the writing workshop: Effects of a combination of methods on the writing quality and writing of college freshmen. (Doctoral dissertation, University of MissouriColumbia, 1986) Dissertation Abstracts International 47, 771A. 150
151 Britton, J. (1975) The development of writing abilities (11-18) Urbana, IL: The National Council of Teachers of English. Bruno, B.J. (1981) The effect of sentence combining on the writing of ninth graders. (Doctoral dissertation, Arizona State University, 1980) Dissertation Ab stracts International 41 4352A. Buros, O. (1978). Metropolitan Achievement Test. The eighth mental measurement yearbook. (Vol. II, 12021205) Highland Park, NJ: The Gryphon Press. Callaghan, T. (1978) The effects of sentence-combining exercises on the syntactic maturity, quality of writing, reading ability, and attitudes of ninth-grade students (Doctoral dissertation, State University of New York at Buffalo, 1977) Dissertation Abstracts International 39 637A. Campbell, D., & Stanley, J. (1968). Experimental and guasiexperimental designs for research. Chicago, IL: Rand McNally. Chomsky, N. (1957). Syntactic structures The Hague: Mouton. Combs, W. (1975). Some further effects and implications of sentence-combining exercises for the secondary language arts curriculum (Doctoral dissertation, University of Minnesota, 1975) Dissertation Abstracts International 16, 1266A. Cooper, C. (1973). An outline for writing sentencecombining problems. English Journal 62 96-103; 108. Corbin, R. & Crosby, M. (1965). Language progra ms for the disadvantaged. National Council of Teachers of English Task Force on Teaching English to the Disadvantaged. Chicago, IL: NCTE. Couch, L.L. (1989) An examination of the effects of sentence imitation practice upon the syntactic features of college freshman writers (Doctoral dissertation, University of Missouri-Columbia, 1987) Dissertation Abstracts International 49 3245A. Daly, J. (1978) Writing apprehension and writing competency. Journal of Educational Research 7J2, 10-14.
152 Daly, J., & Miller, M. (1975). Daly-Miller Writing Apprehension Scale Austin, TX: The University of Texas at Austin. Daly, J., & Miller, M. (1975a). The empirical development of an instrument to measure writing apprehension. Research in the Teaching of English 9, 242-249. Daly, J., & Miller, M. (1975b). Further studies on writing apprehension: SAT scores, success expectations, willingness to take advanced courses and sex differences. Research in the Teaching of English 9, 250-256. Daly, J., & Miller, M. (1975c). Writing apprehension as a predictor of message intensity. Journal of Psychology 89 175-177. Daly, J., & Shamo, W. (1976). Writing apprehension and occupational choice. Journal of Occupational Psychology 49 55-56. Daly, J., & Shamo. (1978). Academic decisions as a function of writing apprehension. Research in the Teaching of English 12, 119-126. D'Angelo, J. (1981) Predicting reading achievement in a senior high school from intelligence, listening, and informative writing (Doctoral dissertation, University of Pittsburgh, 1977) Dissertation Abstracts International 38 2027A. de Beaugrande, R. (1985) Sentence combining and discourse processing: In search of a general theory. In D.A. Daiker, A. Kerek, & M. Morenberg (Eds.). Sentence b combining: A rhetorical perspective (pp. 61-75). Carbondale,IL: Southern Illinois University Press. Elbow, P. (1981) Writing with power New York: Oxford University Press. Elbow, P. (1985) The challenge for sentence combining. In D.A. Daiker, A. Kerek, & M. Morenberg (Eds.). Sentence combining: A rhetorical perspective (pp. 232-245) Carbondale, IL: Southern Illinois University Press. Elley, W.V., Barham, I.H., Lamb, H. & Wyllie, M. (1976). The role of grammar in a secondary school English curriculum. Research in the Teaching of English 10 5-21.
153 Evans, R. (1986). The effects of sentence-combining instruction on controlled and free writing and on scores for standardized tests of sentence structure and reading comprehension. (ERIC Document Reproduction Service No. ED 269 739) Evanechko, P., Ollila, L. & Armstrong, R. (1974). An investigation of the relationship between children's performance in written language and their reading ability. Research in the Teaching of English 8, 315326. Fagan, W.T. (1988) Concepts of reading and writing among low-literate adults. Reading Research and Instruction 27, 58-59. Faigley, L. Daly, J., & Witte, S.P. (1981). The role of writing apprehension in writing performance and writing competence. Journal of Educational Research 75, 16-21. Fisher, K. (1973). An investigation to determine if selected exercises in sentence combining can improve reading and writing (Doctoral dissertation, Indiana University, 1973) Dissertation Abstracts International 34 4556A. Flaherty, G.P. (1985) The speaking/reading/writing connection: Interaction in a basic reading classroom (Doctoral dissertation, University of Cincinnati, 1984) Dissertation Abstracts International 45., 2463A. Garcia, R. (1977). An investigation of relationships: Writing apprehension, syntactic performance, and writing quality (Doctoral dissertation, Arizona State University, 1977). Dissertation Abstracts International 39 4211A. Gay, P. (1983). How attitude interferes with performance of unskilled college writers (ERIC Document Reproduction Service No. ED 234 417) Goodlad, J. (1983) What some schools and classrooms teach. Educational Leadership 40 17-19. Hartman, D. M. (1984). An investigation into the predicted relationship of ten writing assessment variables to reading comprehension (Doctoral dissertation, The University of Connecticut, 1984) Dissertation Abstracts International 45 1353A. Harvley-Felder, Z. (1978) Some factors relating to writing apprehension: An exploratory study (Doctoral
154 Dissertation, The University of North Carolina at Chapel Hill, 1978) Dissertation Abstracts International 39 6503A. Hayes, I. (1985) An experimental study of sentence combining as a means of improving syntactic maturity, writing guality and grammatical fluency in the compositions of remedial high school students. (Doctoral dissertation, Columbia University Teachers College, 1984) Dissertation Abstracts International 45 2424A. Heise, L.L. (1988) The effects of writing on elementary school students' reading comprehension. (ERIC Document Reproduction Service No. 191 007) Helwig, J. (1978) SAS introductory guide. Gary, IN: SAS Institute, Inc. Hendrickson, C.A. (1981). The effect of sentence-combining technigues upon the development of syntactic coplexity and composition guality in high-school students. (Doctoral dissertation, Boston University School of Education, 1981) Dissertation Abstracts International 41 4957A. Heston, J. & Paterline, E. (1974) Unwillingness to communicate: Explication and scale development Paper presented at the Annual Convention of the International Communication Association, New Orleans, LA. Hill, S. (1982) Relationships between reading and writing performance: A correlational study of Metropolitan reading subscores (Doctoral dissertation, University of Florida, 1982) Dissertation Abstracts International 43 1862A. Hillocks, G. (1986) Research in the written composition: New directions for teaching Urbana, IL: National Conference on Research in English, Eric Clearinghouse on Reading and Communication Skills. Howie, K. (1977) A study: The effects of sentence combining: Practice on the writing ability and reading level of ninth-grade students (Doctoral dissertation, University of Colorado at Boulder, 1979) Dissertation Abstracts International 40 1980A. Huck, S., Cormier, W. and Bounds, W. (1974). Reading statistics and research New York, NY: Harper & Row.
155 Hunt,K. (1965). Grammatical structures written at three grade levels Research Report No. 3, Champaign, IL: National Council of Teachers of English. Hunt, K. & O'Donnell, R. (1970). An elementary school curriculum to develop better writing skills. Tallahassee, FL: Florida State University, U.S. Office of Education Grant No. 4-9-08-903-422-010. Isaacson, S.L. (1985). Assessing the potential syntax development of third and fourth grade writers (Doctoral dissertation, Arizona State University, 1985) Dissertation Abstracts International 46 956A. Jordan, Z.L. (1992). The effects of sentence-combining instruction on the syntactic fluency of the writing of a selected group of middle school students (Doctoral dissertation, Wayne State University, 1992) Dissertation Abstracts International 52 4206A. Kansas Institute of Research on Learning Disabilities. (1980) .Research Report No. 13. University of Kansas. In Poplin, M.S., Gray, R. Larsen, S., Banikowski, A., & Mehring, T. A comparison of components of written expression abilities in learning disabled and nonlearning disabled students at three grade levels. Learning Disability Quarterly 3, 46-53. Keppel, G. (1982) Analysis of variance. Design and analysis: A researcher's handbook 2nd edition, Englewood Cliffs, NJ: Prentice-Hall, Inc., 481-508. King, B. (1979) The modes of analyzing teacher and student attitudes toward writing: The Emig attitude scale and the King construct scale (Doctoral dissertation, Rutgers University, The State University of New Jersey, 1979) Dissertation Abstracts International 40 3934A. Kuntz, M. (1975). The relationship between written syntactic development and reading ability in seventh grade. Doctoral dissertation, University of Pittsburgh. (ERIC Document Reproduction Service No. ED 113 710) Laframboise, K.L. (1990) The effects of sentence-combining with and without processing on the reading comprehension and writing fluency of low-achieving fourth-grade students (Doctoral dissertation, University of South Florida, 1990) Dissertation Abstracts International 51 465A.
156 Lamm, J.S. (1982). Sentence combining: A case study of its effect on reading ability. (Doctoral dissertation, Indiana University of Pennsylvania, 1982) Dissertation Abstracts International 43 1958A. Langer, J. (1986). Children reading and writing Norwood, NJ: Ablex. Lawlor, J. (1980). Improving student writing through sentence combining: A literature review (Report No. SWRL-TN-2-80/05) Los Alamitos, CA: Southwest Regional Laboratory for Educational Research and Development. (ERIC Document Reproduction Service No. ED 192 356) Lazdowski, W. (197 6) Determining reading grade levels from analysis of written compositions (Doctoral dissertation, New Mexico State University, 1976) Dissertation Abstracts International 37 1504A. Ledbetter, E.J. (1988). The effect of the practice of combining sentences on the reading achievement of third grade students (Doctoral dissertation, University of South Carolina, 1987) Dissertation Abstracts International 49 775A. Ledesma, L.G. (1981) Sentence combining: Its role in comprehension at literal, reasoning, and evaluative levels and at three syntactic complexities (Doctoral dissertation, West Virginia University, 1980) Dissertation Abstracts International 41 3027A. Lerner, J.W. (1976) Children with learning disabilities (2nd. ed.). Boston, MA: Houghton-Mifflin. Loban, W. (1963). The language of elementary school children Research Report No. 1. Urbana,IL: National Council of Teachers of English. Loban, W. (1970) Stages, velocity and prediction of language development through grade twelve. Final Report, Office of Education, Bureau of Research, U.S. Department of Health, Education, and Welfare. Loban, W. (1976) Language development: Kindergarten through grade twelve Research report No. 18. Urbana,IL: National Council of Teachers of English. Lomax, R.G. (1983) A guide to multiple sample structural eguation modeling. Behavior Research Methods and Instrumentation 15 580-584.
157 MacNeill, T.B. (1982) The effect of sentence-combining practice on the development of reading comprehension and the written syntactic skills of ninth-grade students. (ERIC Document Reproduction Service No. ED 217 415) McCroskey, J. (1970) Measures of communication bound anxiety. Speech Monographs 37 269-277. Marzano, R. & diStefano, D. (1977). DiComp: A diagnostic system for teaching written composition Indian Rocks Beach, NJ: Wylie-Padol Publishing, Inc. Mavrogenes, N. & Galen, N. (1980). Improving the language facility of disabled secondary readers. National Council of Teachers of English, Urbana, 111. Research Foundation. (ERIC Reproduction Service No ED 193 612) Mellon, J. (1969). Transformational sentence combining: A method for enhancing the development of syntactic fluency in English composition (NCTE Research Report No. 10) Urbana, IL: National Council of Teachers of English. Mellon, J. (1979) Issues in the theory and practice of sentence combining: A twenty-year perspective. In D. Daiker, A. Kerek, & M. Morenberg. (Eds.) Sentence combining and the teaching of writing. Akron, OH:L and S Books. Menendez, D. (1978). The effect of sentence-combining practice on remedial college students' syntactic maturity, punctuation skills, and reading ability (Doctoral dissertation, Indiana University, 1978) Dissertation Abstracts International 39 6599A. Moffett, J. (1983) Teaching the universe of disourse Portsmouth, NJ: Boynton/Cook. Moore, M.M. (1984) The effects of supplementary sentence combining on writing effectiveness (Doctoral dissertation, East Texas State University, 1983) Dissertation Abstracts International 44 3310A. Mullis, I., & Mellon, J. (1980). Guidelines for describing three aspects of writing: Syntax, cohesion and mechanics. Denver, CO: National Assessment of Educational Progress. National Assessment of Educational Progress. (1975) Writing mechanics. 1969-74 Writing Report No. 05-W01, Denver, CO: NAEP.
158 National Assessment of Educational Progress. (1980) Writing Achievement. 1969-79 : Results from the third national writing assessment. Denver, CO: NAEP. Neville, D. & Searls, E. (1985). The effect of sentence combining and kernel-identification training on the syntactic component of reading comprehension. Research in the Teaching of English 19 37-61. Neville, D., & Searls, E. (1991). A meta-analytic review of the effect of sentence-combining on reading comprehension. Reading Research and Instruction 31 (1), 63-76. O'Donnell, R. Griffin, W. & Norris, R. (1967). Syntax of kindergarten and elementary school children: A transformational analysis (Research Report No. 8) Champaign, IL: National Council of Teachers of English. O'Hare, F. (1973). Sentence Combining: Improving student writing without formal grammar instruction NCTE Research Report No. 15. Urbana, IL: National Council of Teachers of English. O'Hare, F. (1975). Sentencecraft Langton, MA: Ginn & Company Pearson, R.D., & Camperell, K. (1981). Comprehension of text structures. In J. Guthrie (Ed.) Comprehension and teaching: Research reviews Newark, NJ: International Reading Association. Pendleton, M.E. (1986). The effect of sentence combining and rhetorical arrangement on the expository writing of tenth grade students (Doctoral dissertation, Boston University, 1985) Dissertation Abstracts International 46 3588A. Phillips, G. (1988) Reticence: Pathology of the normal speaker. Speech Monographs 35 39-49. Piper, K.L. (1985). An investigation of the effects of sentence combining and story expansion delivered by word processing microcomputers on the writing ability of sixth graders (Doctoral dissertation, Texas Tech University, 1984) Dissertation Abstracts International 45 2747A. Porter, D. (1973). Self report scales of communication apprehension and automatic arousal: A test of construct
159 validity Paper presented at the Annual Convention of the Speech Association, New York, NY. Porter, D.W. (1985) Sentence combining with students of English as a second language (Doctoral dissertation, The University of Texas at Austin, 1983). Dissertation Abstracts International 45 2018A. Powers, W., Cook, J., & Meyer, R. (1979). The effect of compulsory writing on writing apprehension. Research in the Teaching of English 13 225-230. Prescott, G., Balow, I., Hogan, T. & Farr, R. (1979). Metropolitan Achievement Test: Advanced 1. Form KS. New York, NY: Harcourt Brace Jovanovich. Rice, P.H. (1984). A study of the effectiveness of sentence combining in increasing the syntactic maturity of student writing at grades 7 through 11 (Doctoral dissertation, University of Cincinnati, 1983). Dissertation Abstracts International 44 2395A. Rinne, C. (1967) Improvement in reading comprehension through increased awareness of written syntactic patterns (Doctoral dissertation, Stanford University, 1967) Dissertation Abstracts International 28 2668A. Robertson, J. (1966) Pupil understanding of connectives in reading. Reading Research Ouaterly 3, 387-411,417. Romano, T. (1987) Clearing the way: Working with teenage writers Portsmouth, NH: Heineman. Rubin, A., & Hansen, J. (1984). Reading and writing: How are the first two "r's" related? Reading Education Report No. 51. Champaign, IL: Center for the Study of Reading. Schuster, E. (1976) Forward to basics through sentence combining Paper presented at the Annual Meeting of the Pennsylvania Council of Teachers of English (20th, Harrisburg, Pa) (ERIC Document Reproduction Service No. ED 133 774) Searls, E., & Neville, D. (1988). An exploratory review of sentence-combining research related to reading. Journal of Research and Development in Education 21, 1-15. Shanahan, T. (1980) The impact of writing instruction on learning to read. Reading World 19 (2), 357-368.
160 Shanahan, T. (1988) The reading-writing relationship: Seven instructional principles. The Reading Teacher 32 (3), 636-647. Shanahan, T. (Ed.) (1990). Reading and writing together: New perspectives for the classroom Norwood, NJ: Christopher-Gordon Publishers, Inc. Shanahan, T. & Lomax, R. (1988) A developmental comparison of three theoretical models of the reading-writing relationship. Research in the Teaching of English. 22 (2), 196-212. Simmons, S.S. (1981). Selected effects of sentence-combining exercises on the reading and listening comprehension and attitudes of seventh-grade students (Doctoral dissertation, Florida State University, 1981) Dissertation Abstracts International 42 2545A. Spilton, R. (1987) The effects of individualized language arts, sentence combining and traditional grammar on the syntactic maturity and guality of writing of a select group of eighth graders. (Doctoral dissertation, Georgia State University, 1987) Dissertation Abstracts International 47 4007A. Stacks, D. Boozer, R. & Lally, T. (1983). Syntactic language correlates of written communication apprehension Paper presented at the Annual Meeting of the American Business Communication Association, Houston, (ERIC Document Reproduction Service No ED 22 6 374) Steidle, E.F. (1977). An investigation of writing ability as a function of student attitude: General and specific. A paper presented at the Annual Meeting of the American Research Association, April, 1977. (ERIC Document Reproduction Service ED 139 023) Sternglass, M. (1980) Sentence combining and the reading of sentences. College Composition and Communication 31. 325-328. Stiles, E.R. (1976). A case study of remedial writers in selected two-year colleges in East Tennessee (Doctoral dissertation, The University of Tennessee, 197 6) Dissertation Abstracts International 37 7004A. Story, B.A. (1985) The effects of transformational sentence combining on the reading comprehension of college students (Doctoral dissertation, Auburn University, 1985) Dissertation Abstracts International 46 391A.
161 Stotsky,S. L. (1975). Sentence combining as a curricular activity: Its effects on written language development and reading comprehension. Research in the Teaching of English 9., 30-71. Stotsky, S.L. (1983) Research on reading/writing relationships: A synthesis and suggested directions. Language Arts 60 627-642. Strong, W. (1973) Sentence combining: A composing book New York, NY: Random House. Strong, W. (1986) Creative approaches to sentence combining Urbana, IL: ERIC Clearinghouse on Reading and Communication Skills and the National Council of Teachers of English. Sullivan, M. (1977) The effects of sentence-combining exercises on syntactic maturity, guality of writing, reading ability, and attitudes of students in grade eleven (Doctoral dissertation, State University of New York at Buffalo, 1977) Dissertation Abstracts International 39 1197A. Symonds, P. (1931). Practice versus grammar in the learning of correct English usage. Journal of Educational Psychology 22 82. Takahashi, B.L. (1975) Comprehension of written syntactic structures by good readers and slow readers Master's thesis, Rutgers University. (ERIC Document Reproduction Service No. ED 117 655) Taki El Din, S.R. (1987) The effectiveness of sentencecombining practice on Arab students* overall writing ability and syntactic maturity (Doctoral dissertation, Indiana University of Pennsylvania, 1987). Dissertation Abstracts International 47 3022A. Tracy, G.E. (1990). The effects of sentence-combining practice on syntactic maturity and writing guality in ESL in freshman composition (Doctoral dissertation, Oklahoma State University, 1990) Dissertation Abstracts International 50 2777A. Trivelli, F. (1983) Sentence combining: The effect on the syntactic ability and reading comprehension of eighth graders. (Doctoral dissertation, The University of Akron, 1983) Dissertation Abstracts International 44, 383A.
162 Weinstein, G. (1982) Sentence combining workbook Paper presented at the Annual Meeting of the California Association of Teachers of English (24th, San Francisco) (ERIC Reproduction Service No. ED 213 016) Whitt, L.M. (1987). The effects of sentence-combining instruction on the growth of syntactic fluency in adult basic writers (18-25) (Doctoral dissertation, University of Virginia, 1987) Dissertation Abstracts International 48 1139A. Wilkinson, P. A., & Patty, D. (1993). The effects of sentence combining on the reading comprehension of fourth-grade students. Research in the Teaching of English 22 (1) 104-125. Young, B.L. (1989). The effect of the combination of sentence-combining exercises and free writing opportunities on the writing ability of third-, fourthand fifth-grade students (Doctoral dissertation, The University of California, Riverside, 1988) Dissertation Abstracts International 49.2106A. Zemelman, S., & Daniels, H. (1988). A community of writers Portsmouth, NH: Heinemann.
BIOGRAPHICAL SKETCH Kay Westall Gonsoulin was born in Long Beach, California, on October 18, 1942. She graduated from Long Beach Polytechnic High School in 1960. She attended the University of California at Los Angeles and Long Beach City College before completing her studies at California State University at Long Beach where she received her B.A. in English in 1965, and her M.A. in English in 1969. From 1966 to 1969, she taught English at Long Beach Polytechnic High School. In 1969, she married Dr. Henry Gonsoulin, a dentist, and they moved to New Orleans, Louisiana, where she taught English from 1969-1971 at John McDonogh Senior High School. In 1971, she and her husband moved to Gainesville, Florida. In 1976, she received the Ed.S. in instruction and curriculum with a major in reading and language arts. She later began work on the Ph.D. in the same department and major. During her doctoral course work, she taught reading at the English Language Institute, an ESL program at the University of Florida. She has also taught English and reading from 1973 to the present at Buchholz High School where she has been chairperson of the Reading and Language Arts departments, member of the curriculum and steering committees, and a peer coach. She holds memberships in the 163
164 Alachua County Teachers of English, the Florida Council of Teachers of English, the National Council of Teachers of English, the Association for Supervision and Curriculum Development, the Conference on English Leadership, International Reading Association and Kappa Delta Pi, an honorary educational organization. She has been a reader, table leader and field tester for the writing portion of the College Level Academic Skills Test and the Florida Certification of Teachers Exam and a consultant for both the English and the Reading Subject Area exams of the FCTE. She and her husband have two sons: Taylor, age 21, a senior at Georgia Institute of Technology in mechanical engineering, and Derek, age 18, a freshman at the Florida School of the Arts at St. Johns River Community College majoring in fine arts. Currently, she is director of a 9th-grade interdisciplinary program designed to ease the transition of students from 8th grade to 9th grade. She directs 2 teachers in this program stressing team planning and teaching, cooperative learning, higher order thinking skills and performance assessment; provides reading support and instruction to the 9th-grade content area teachers; and teaches classes in critical thinking and study skills.
I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doptor of Philosophy H. Thompson Fillmer, Chair Professor of Instruction and Curriculum I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. M. David Miller Associate Professor of Foundations of Education I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. &L*J & ILUa^Lk^ Edward C. Turner Associate Professor of of Instruction and Curriculum I certify that I have read this study and that in my opinion it conforms to acceptable standards of r scholarly presentation and is fully adequate, /in scope/^no/ qfiality, as a dissertation for the degree of ja&ct^br of loberfc GVWrighl Associate Professbr of Instruction and Curriculum This dissertation was submitted to the Graduate Faculty of the College of Education and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December 1993 Dean, College of Educal Dean, Graduate School
UNIVERSITY OF FLORIDA 3 1262 08666 347 2