GRADE SIX STUDENTS UNDERSTANDING OF METAPHOR IN INFORMATIONAL TEXTbyLEIGH A. FAULKNERB.A., Mount Allison University, 1975B.Ed., Mount Allison University, 1975M.A. (Ed.), Saint Mary’s University, 1985A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF EDUCATIONinTHE FACULTY OF GRADUATE STUDIESDepartment of Language EducationTHE UNIVERSITY OF BRITISH COLUMBIAApril 1995We accept this thesis as conformingto the© Leigh A. Faulkner, 1995In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.Department of t jLi44eThe University of British ColumbiaVancouver, CanadaDate / / O / fDE-6 (2/88)ABSTRACTMetaphor research has become widespread. However, students’ understanding of metaphor ininformational text has received little study. With increased use of informational trade books in theclassroom, research in this area is needed. Fifty-five grade six students with Canadian English as theirfirst language participated in the study. Their understanding of metaphors in excerpts from threerecently published informational trade books was examined by the use of the reading think aloudtechnique and multiple choice activities. One think aloud was completed by each student in bothindividual and dyadic conditions. Multiple choice activities were completed individually after reading,but with the text available. The think-aloud protocols were examined using specific-trait analysis,holistic scoring, and miscue analysis. The multiple choice activities were scored against anticipatedadult-like understanding and the results were subjected to standard statistical tests. Level ofunderstanding of metaphors varied widely among students, with the overall average being about65%. Contrary to prediction, understanding was significantly higher in the individual conditioncompared to the dyadic condition. Although part of this difference could be attributed to differencesin passage difficulty, the anticipated scaffolding effect of reading with a partner was not found. Thereading think aloud was a rich source of information about both the meaning students constructedand the meaning-construction process. The study suggested that the think aloud could be used inthe classroom as an effective learning device, particularly in that it allowed less-capable readers toparticipate as equal partners in what might otherwise have been a frustrating reading task. Overall,there emerged a picture of students at various points along the path to full adult mastery of metaphor,with some students already demonstrating an adult level of understanding. Level of textunderstanding was consistent with level of metaphor understanding. The only metaphor-type effectidentified was for metaphors with copula-verb syntactic-frame structure. Abstractness of the words inthe metaphors did not affect meaning construction; however, conventionality of the metaphoricalexpressions did influence understanding.IIITABLE OF CONTENTSABSTRACT iiTABLE OF CONTENTS iiiLIST OF TABLES viiLIST OF FIGURES viiiACKNOWLEDGEMENT xCHAPTER ONE: OVERVIEWI. Statement of the Problem and Introduction 1II. Need for the Study 2III. Theoretical Framework and Assumptions 4IV. Research Questions 7V. Limitations of the Study 8VI. Definition of Terms 10CHAPTER TWO: LITERATURE REVIEWI. Introduction 14II. Social Interactionism 14Ill. Constructivism 17IV. Schema Theory 19V. Parallel Distributed Processing 22VI. Views of Metaphor 23i. Substitution View 24ii. Comparison View 25iii. Interaction View 26iv. Metaphor as Mapping 28VII. Metaphor Research 31i. Methodology 32ii. Child Focus 37a. Development of Metaphor Understanding 37b. Experience and Prior Knowledge 44ivc.Culture48iii Text Focus 49a. Context 49b. Explicitness and Similarities 51c. Ecological Validity 53d. Metaphor in Informational Text 54VIII. Summary 56CHAPTER THREE: RESEARCH DIALECTIC AND METHODOLOGYI. Introduction 59II. Research Proposal 67Ill. Pilot Study 71i. Research on Think Aloud 72ii. Materials 73iii. Procedures 73IV. Main Study 78i. Research Approach 78ii. Research Design and Instrumentation 79iii. Selection of Participants 80iv. Site Procedures 82v. Data Collection and Recording 83vi. Methodological Assumptions 86vii. Limitations 87V. Summary 88CHAPTER FOUR: FINDINGSI. Introduction 91II. Findings 92i. Major Question (a) 92a. Question One 92b. QuestionTwo 96c. Question Three 99d. Question Four 99ii. Major Question (b) 107e. Question Five 107Viii. Major Question (c) .116f. Question Six 116g. Question Seven 134Ill. Summary 139CHAPTER FIVE: DISCUSSION, CONCLUSIONS, APPLICATION, LIMITATIONS, AND RESEARCHIMPLICATIONSI. Introduction 140II. Discussion 140III. Conclusions 162IV. Application 163V. Limitations 167VI. Implications for Future Research 168VII. Concluding Remarks 170REFERENCES 172APPENDIX A Sample Letter to Publisher Seeking Permission to Photocopy 178APPENDIX B Copies of Multiple Choice Meaning Clarification Activities--Passages X, Y,andZ 179APPENDIX C Sample Individual Think-Aloud Protocol (Main Study) 185APPENDIX D Specific Traits and Definitions 187APPENDIX E Sample Specific-Trait Analysis (Sample Dyadic Think-Aloud Protocol, MainStudy) 194APPENDIX F Average Specific Traits Profile Graphs (Combined Categories) of Studentsin Extreme Groups by MCMCA Score 197APPENDIX G Holistic Rating Scale 198APPENDIX H MCMCA Scores--All Students, Individual and Dyadic Conditions 199viAPPENDIX I MCMCA Scores by Student, Passage, and Condition 201APPENDIX J Overall MCMCA Perlormance by Syntactic Frame and Similarity Type (AfterBroderick, 1992) 203APPENDIX K MCMCA Results by Item 205APPENDIX L MCMCA Metaphor Analysis (After Lakoff and Johnson, 1980, and Lakoff andTurner, 1989) 207APPENDIX M Holistic Scores--Dyadic Average 210APPENDIX N Holistic Scores--Individual Average 211APPENDIX 0 Individual MCMCA Scores and Level of Miscue--Comparison 213APPENDIX P Specific Traits Aggregate Summaries (Combined Categories)--ExtremeGroups 215APPENDIX Q Specific Traits Aggregate Percent Summaries (Combined Categories) --ExtremeGroups 217APPENDIX R Composition of Extreme Groups EGI4 and EGD4, with Passage Read 219APPENDIX S Passage Statistics 220VIILIST OF TABLESTable1 Multiple Choice Meaning Clarification Activities--Performance Statistics 932 Summary of wls Test for Normality of a Population--Passage by Condition 963 Changes in MCMCA Scores According to Level of Passage Difficulty 984 MCMCA Percentage Scores According to Metaphor Type (After Broderick, 1992).... 1015 MCMCA Scores--Sub-50% by Passage, Item, and Metaphor Type 1026 MCMCA Scores--80% and Above by Passage, Item, and Metaphor Type 1037 Extreme Groups--MCMCA Statistics 1178 Summary of Results--F-test for Population Variances Between and Within ExtremeGroups 1189 Number of Types of Specific Traits by Combined Category and Extreme Group 12110 Performance Summary of Students Appearing in Both High-Extreme Groups (EGI4and EGD4) 13011 Low-Extreme Individual Group (EGI1) Miscue Analysis Percent and Reading Level.... 13212 High-Extreme Individual Group (EGI4) Miscue Analysis Percent and Reading Level... 13313 Summary of Data from Table 3, Changes in MCMCA Scores According to Level ofPassage Difficulty 14614 Abstractness of Metaphor Key Words/Phrases--High- and Low-Extreme Items byMCMCA Score 149VIIILIST OF FIGURESFigure1 Distribution of MCMCA Scores 942 MCMCA Scores--Student Totals 943 Frequency of MCMCA Scores--Dyadic and Individual 974 Metaphor Features--High-Extreme MCMCA Items (After Lakoff, Johnson, andTurner) 1055 Metaphor Features--Low-Extreme MCMCA Items (After Lakoff, Johnson, andTurner) 1056 Metaphor Features--Highest Three MCMCA Items (After Lakoff, Johnson, andTurner) 1067 Metaphor Features--Highest Three MCMCA Items (After Lakoft, Johnson, andTurner) 1078 Specific Traits--Average Number and Average Percent--All Protocols 1099 Specific Traits--Average Number and Average Percent--Dyadic Protocols 10910 Specific Traits--Average Number and Average Percent--Individual Protocols 11011 Specific Traits--Average Number and Average Percent--Passage X Individual 11112 Specific Traits--Average Number and Average Percent--Passage Y Individual 11113 Specific Traits--Average Number and Average Percent--Passage Z Individual 11214 Holistic Averages--Extreme Group Individual 1st 11415 Holistic Averages--Extreme Group Individual 4th 11416 Specific Traits--Extreme Groups--Average, Mean, and Median 11817 Specific Traits--Average Number--Extreme Groups 11918 Specific Traits--Average Percent--Extreme Groups 12019 Holistic Score Average by Dyadic MCMCA Score 135ix20 Meaning Construction Average by Dyadic MCMCA Score 13621 Holistic Score Average by Individual MCMCA Score 13622 Meaning Construction Average by Individual MCMCA Score 13723 Holistic Average by Descriptor--Dyadic Protocols 138xACKNOWLEDGEM ENTI wish to express my sincere gratitude to Dr. Victor Froese for his guidance and supportthroughout my studies. I feel most fortunate to have had the opportunity to work under hissupervision. I wish also to thank the other members of my thesis committee, Dr. Robert Corny and Dr.Wendy Sutton. Each brought to the task unique qualities and expertise that I valued highly.There have been numerous colleagues--faculty and fellow graduate students-- within theDepartment of Language Education whose acceptance and friendship made me feel that I belongedand whose considered advice, when sought, helped me through difficult stages in the process. Toall, I express my thanks.To my wife and traveling companion, Edna, who thought from the very beginning that this wouldbe an exciting side trip to the big journey and who helped me keep my sense of perspective, Iacknowledge that it would not have been possible without your encouragement and love.1CHAPTER ONE: OVERVIEWI. Statement of the Problem and Introduction 1II. Need for the Study 2Ill. Theoretical Framework and Assumptions 4IV. Research Questions 7V. Limitations of the Study 8VI. Definition of Terms 10I. Statement of the Problem and IntroductionA growing body of research literature (e.g., Vosniadou, Ortony, Reynolds, and Wilson, 1984;Keil, 1986; Winner, 1988; Siltanen, 1989, 1990) focuses on children’s ability to understand metaphorand on factors affecting understanding. The importance of this research is emphasized by the keycontentions of Lakoff and Johnson (1980), Lakoff and Turner (1989), and Lakoff (1993) thatmetaphor is based on experience and is central to what we think and the way we think.From an educational perspective, there are bothersome gaps in the research. One suchgap--attention to ecological validity--is identified by Broderick (1992). In an examination of themetaphor types used as stimuli in tests of fifteen major studies with children, he challenges theecological validity of the studies on the ground that the metaphor types are different from themetaphorical language children normally encounter in printed stories. Broderick finds that the mostfrequently used stimulus types--the substitution and equation (copula-verb) types ofmetaphors--appear infrequently in children’s books. He gives no information about the types ofmetaphors children normally encounter in other types of text, such as informational text.A second gap in the literature is created by a general lack of attention to informational text.Studies of the importance of context in supporting metaphor interpretation have, for example, usedspecially created stories of differing lengths (e.g., Siltanen, 1989). This raises the question ofwhether the text structure generally associated with fiction (see, for example, Stein & Glenn, 1979)2might be influencing children’s responses, as indicated by Waggoner, Meese, and Palermo (1985).From their examination of the effect of story structure on children’s understanding of metaphor, theyconclude that even young children’s interpretation of metaphor may be affected by theirunderstanding of typical story structure. Their study sheds no light, however, on understanding ofmetaphor in informational text that, while perhaps exhibiting a strong narrative voice and certain storyelements such as, for example, conflict, may lack the “overarching framework that a reader can use toorganize and relate the information in the text” (Beck & McKeown, 1992, p. 135).As language arts programs broaden to include more trade books, both fiction and nonfiction,students are being exposed to much richer natural language than occurs in the traditional basalreading programs. While some of the newer reading series, such as the Impressions (1988) program,do contain authentic text (i.e., text not specifically written for classroom use) some of the series still inuse contain mainly specially written selections and/or selections that have been rewritten to suchcriteria that the language is flat and unnatural, lacking to a large degree the metaphorical structurestypical of natural language. (For a discussion of this in relation to one early reading series, see Luke,1988.)Although studies have looked at children’s comprehension of metaphor in fiction and inspecially constructed test-like text, children’s understanding of the metaphorical language theyencounter in nonfiction text, particularly informational text, remains largely unexamined. Additionally,it is not known in what way children’s level of understanding of metaphor is related to their level ofunderstanding of the text itself. This study proposes to investigate these two problems.II. Need for the StudyBroderick’s (1992) reminder of the importance of ecological validity is a helpful starting pointwhen considering children’s understanding of metaphor in informational text of the sort widelyavailable for use in the classroom. Without some awareness of children’s understanding of metaphoroccurring naturally in informational text, one is left with uncertainty as to whether specifically chosen3materials are within the reading ability of the students. On the other hand, once an awareness isreached, a basis for decision making can be established and informed adjustment of both curriculumand classroom practice can be made to ensure that children thoroughly understand the informationaltext they read and benefit from the increased richness of meaning provided by metaphor.This study proposes to address the problem of ecological validity identified by Broderick (1992>to the extent that it will use text that students might well encounter in their school program. By usingcoherent excerpts of such authentic text, the problem of adequacy of context will be addressed at thesame time.An additional measure of ecological validity will be provided by having students complete part ofthe work in pairs (dyads), thus drawing on current cooperative-learning practice, as influenced byVygotsky (1962, 1978) (see, for example, Johnson & Johnson, 1994). This will provide anopportunity to compare student understandings in the dyadic and individual conditions.Informational trade books are being used widely in both language arts and content areas of thecurriculum. An examination of such trade books reveals that much of the language is rich in metaphor.The use of analogy (a type of metaphoric language) as an advance organizer has been found tostimulate recall of text information under experimental conditions (Mayer and Bromage, 1980;Anderson and Pearson, 1984). Mayer and Bromage (1980) suggest that analogy works at the time oforiginal encoding of the information in memory as well as during recall by structuring the information ina coherent form. Given these crucial functions of metaphorical language, it becomes a pertinentquestion how children’s understanding of the metaphorical language is contributing to theirunderstanding of the informational text.There is, at present, no clear idea of how children’s understanding of metaphor relates to theirunderstanding of informational text. This study will attempt to determine whether such a relationshipexists and, if it does, something of its nature.An examination of the metaphor content of informational text was carried out as part of thisstudy. Selections of informational text appearing in authorized language arts anthologies and project4books of the Impressions program and basal readers were examined and were found to vary widely inthe amount of metaphor content, with the Impressions program material containing a higherproportion of metaphors than selections from the other basal reading programs. Randomly selectedinformational trade books also showed a high density of metaphor, much higher than the basalreaders. While these more-or-less informal findings raise questions about the appropriateness ofdifferent materials and whether students using a particular program are being provided with readingexperiences that will support successful reading beyond the specific program materials, theseconcerns will be left to others to investigate, Of particular interest so far as this study is concerned isthat students are likely to encounter natural, metaphor-rich language when they read informationaltrade books.Should a clear relationship between understanding of metaphor and understanding of text beidentified, this study will have served the purpose of drawing attention to the need for teachers to givecareful consideration to the metaphor content of informational text. It is hoped that the results of thisstudy would then provide teachers with an incentive and a means to examine this rich area oflanguage experience and student understanding.Ill. Theoretical Framework and AssumptionsThe view of metaphor which informs this study is that put forward by Lakoff and Johnson (1980),Lakoff and Turner (1989), and Lakoff (1993). According to this view, metaphors are conceptual innature and are a natural, inescapable part of our way of thinking. Their primary purpose is to extendunderstanding of one conceptual entity (target) by drawing upon knowledge of a second conceptualentity (source) by means of a mapping process. Drawing on schema theory, as put forth by Rumelhart(1980), according to which schemata are process-structures in which knowledge is held in anorganized manner and is rendered accessible in usable form, the metaphor process is held to consistof any or all of the following types of mapping from the source-domain schema to the target-domainschema: (a) slots, (b) relations, (c) properties, and (d) knowledge. These will be discussed in Chapter5Two. According to Lakoff, Johnson, and Turner, metaphors are based on solitary and culturallygrounded experience and have meaning because they are grounded in experience. This groundingis so powerful that metaphors govern both how and what we think.Schema theory and this view of metaphor are consistent with the further concept thatknowledge is constructed by the cognizing individual. The radical form of constructivism (so-called byits proponents) has been put forward by Ernst von Glasersfeld (1984, 1989). The more moderateform espoused by Jerome Bruner (1986, 1990) is more generally accepted. L. S. Vygotsky (1962,1978) sets forth the necessary conditions for knowledge construction to occur. He suggests that allindividual learning and development result from social interaction, a process in which language plays acritical mediational role. Constructivism and social interactionism can be seen as mutually informingconcepts that are helpful in describing how the individual comes to know.The radical form of constructivism, as informed by Vygotsky’s theory, has been adopted in thecurrent study; thus, this study will examine the meaning that children construct from their ownknowledge in the presence of both text and talk and will depart from the theoretical stance of thosestudies that conceptualize the process as one of comprehension of a meaning residing in the text.The focus, then, is children’s constructed understanding, not their comprehension in theconventional sense.Three particular factors have been found to influence outcomes of studies of children’sunderstanding of metaphor: (1) the context in which the metaphor is presented (Vosniadou, Ortony,Reynolds, & Wilson, 1984); (2) knowledge of story structure (Waggoner, Meese, & Palermo, 1985);and (3) types of metaphor (Broderick, 1992). Specifically, Vosniadou et al. found that when adequatecontext is provided, children’s understanding of metaphor increases. Waggoner et al., examiningchildren’s recall of metaphor in fiction, found that metaphor is more memorable if embedded at theoutcome node, according to story-grammar analysis, than if embedded at another location; andBroderick found that many studies of children’s understanding of metaphor used metaphor types notcommonly found in children’s books and likely to be of questionable familiarity to the children.6The current study addresses these three problematic areas in the following ways: first, all textsamples used are self-contained units of authentic text, from three to thirteen pages in length,preserving to a major extent the original context, including illustrations, photographs, and graphics;second, because informational text is used, possible influence of story structure is avoided; and,third, text samples have been drawn from authentic text, thus ensuring that the study’s metaphortypes are consistent with metaphor types children are exposed to in the course of reading readilyavailable informational trade books.The use of the think aloud as a data collection device, together with analysis of the resultingthink-aloud protocols, is supported by a growing literature (e.g., Dias, 1987; Baumann, Jones, &Seifert-Kessell, 1992; Ericsson & Simon, 1993). Dias contends that the think-aloud verbalization,while not necessarily “identical with” the thought process from which it arises, is “close to beingconcurrent” and is a useful reflection of the thought process (p. 9). Baumann et al. show thatthink-aloud instruction can improve student performance on metacognitive activities involved inreading, thereby indicating that thinking aloud can be linked in some non-trivial way to the readingprocess and can be taken to reveal something about the process itself and the meanings constructedduring reading. Ericsson and Simon provide a powerful theoretical rationale for the use of think-aloudprocedures in a variety of research situations.This study makes use of the think aloud to gather data. By using the concurrent think aloud, it ishoped that both the meaning-construction process and the meanings constructed during reading willbe uninfluenced by memory or recall processes or by possible effects of probes used to encourageretrospection.Based on the literature reviewed in Chapter Two and the methodology described in ChapterThree, seven key assumptions have been made about metaphor, constructivism, think alouds, socialinteraction, and the stability of specific traits. It is assumed:1. That metaphor is a cognitive structure rather than simply a linguistic one;2. That meaning making is a constructive process;73. That what we already know determines the possible meanings we can construct when weread;4. That thinking aloud while reading reveals something about the reading process and themeaning that is constructed;5. That grade six students are capable of doing think alouds successfully while reading;6. That working with another person will alter the meaning constructed during the readingexperience; and7. That the operation of the specific traits identified during the pilot study remains the same inthe main study.IV. Research QuestionsThree major questions underlie this study: (a) How well do grade six students understandmetaphor in the informational text they read? (b) How well do the same children understand themetaphor-bearing informational text itself? (c) How are understanding of text and understanding ofmetaphor related?For the purpose of this study, these three major questions are restated as seven specificquestions. The relationship between the major and specific questions is as follows:(a) How well do grade six students understand metaphor in the informational text they read?1. What is student level of understanding of metaphor as determined by performance on amultiple choice meaning clarification activity?2. What are the differences in understanding of metaphor by the same student in the individualand dyadic think-aloud conditions as determined by performance on multiple choice meaningclarification activities?3. What are the differences in understanding of metaphor in the same text for the individual anddyadic think-aloud conditions as determined by performance on a multiple choice meaning clarificationactivity?84. What are the differences in understanding of different types of metaphor as determined byperformance on multiple choice meaning clarification activities?(b) How well do the same children understand the metaphor-bearing informational text itself?5. What proportion of students achieve high and low scores on think-aloud tasks as determinedby holistic and specific-trait scoring and miscue analysis?(c) How are understanding of text and understanding of metaphor related?6. What are the features of the think alouds of students whose performance on a multiplechoice meaning clarification activity placed them in the first and fourth quartiles?7. What is the relationship between the level of students understanding of metaphor and theirunderstanding of text?V. Limitations of the StudyThis study was undertaken to examine how well grade six students understand metaphor in theinformational text they read, how well the same children understand the informational text itself, andhow understanding of text and understanding of metaphor are related.To make this study manageable in terms of time and cost, other potentially interesting questionswere not examined. It was not possible to investigate children’s understanding of metaphor in fictionor poetry. Research in these areas would need to take into account some of the special features ofthe two, such as the particular structure of story or poetry’s often very high density of metaphor, wheremeaning may be influenced by interactions among metaphors.This study did not set out to explore psychological aspects of metaphor or text understanding.While such processes are under debate and are of considerable interest, the purpose of this studywas to investigate the meaning students construct. Although the meaning-construction processesdiscussed in this study must surely have a psychological foundation, examination of the nature of thatfoundation was not the purpose here.Although student understanding of metaphor at all grade levels is of interest, it was necessary9to limit the study to one grade level only for the reasons mentioned above. Grade six students wereselected, in part because they are more skilled in metacognitive strategies than very young childrenand they have a rapidly growing world knowledge. When absent or weak, these two factors havebeen found to have a negative effect on reading understanding (Baker & Brown, 1984). It was hopedthat by using grade six students, the potentially negative influence of these two factors would be lessthan with younger children. Also, grade six students are unlikely to have fully developedmetacognitive processes. It has been found that, as readers mature, the metacognitive processesbecome automatic and drop below the level of consciousness, often going unreported duringthink-aloud tasks (Afflerbach & Johnston, 1984). Clearly, this study’s results can be taken as onlypossible suggestions of metaphor and text understanding of younger and older children.It was necessary to develop instruments and methods for this study. No adequate measureswere available for use under the conditions and for the purposes identified. Most studies havecreated texts and have structured metaphors in particular ways. Many studies have depended on theability of students to paraphrase metaphors. Some have taken for granted student ability to recall.Student understanding of text has generally been ignored. Most studies have been highly intrusivein that they have required students to use text in ways other than those with which they are familiarand in ways that emphasize the test-like nature of the activity. This study used methods and materialsas much as possible like those students would be familiar with from their classroom experience. Thespecific traits used to describe the think alouds, for example, emerged from the protocols themselvesand were not imposed upon them as would have been the case with previously developedinstruments. By adopting this approach, it was hoped to maintain the ecological validity of the materialand activity and to maintain a closer contact with the curriculum than would be possible if instrumentsand methods were borrowed from previous studies. The ecological validity was weakened to theextent that it was necessary for the students to work in small groups outside their classroom with astranger who clearly had a purpose beyond simply having them read and talk about passages of text.The test-like situation was not entirely avoided.10The issue of adequate context in the presentation of metaphors was addressed by usinghigh-quality colour photocopies of the front cover and initial pages of text of each book. In this way, itwas hoped to ensure that the metaphors were contextualized in a manner similar to the original.However, because the complete books were not available to the students and numbered fluorescentdots were affixed at points in each text for reference during the multiple choice activity, the originalcontext was somewhat altered.Researchers generally have gone to considerable effort to ensure that, where multiplepassages are used, they are equivalent. This study used passages that are different from each otherin subject matter and reading level. The books are representative of material classroom teachersmight well use in their programmes. The differences in the reading levels of the passages haveprovided data for additional comparative analysis.This study, approaching metaphor as a conceptual phenomenon rather than simply a linguisticone, assumes that the cultural contribution to one’s understanding of metaphor will pose specialproblems for non-native English speakers. Drawing subjects as it has from a native English-speakingpopulation, this study makes no attempt to examine the particular problems of English-as-a-second-language (ESL) learners, although such an examination is needed.VI. Definition of TermsIn this study, the key terms were operationally defined as follows:Metaphor: A metaphor is a conceptual entity, identifiable linguistically as a conjunction of terms,stated or implied, such that one term (the target) is understood in terms of some feature(s) of the other(the source). The purpose of metaphor is to create and/or enrich understanding.Individual think aloud: A process of speaking aloud what one is thinking. A reading think aloudcombines this with the oral reading of text in such a way that the text reading is interrupted by thespeaking aloud of the accompanying thoughts. It may also include either or both a speaking aloud ofthoughts as they occur during the reading and/or a retelling after the reading has been completed.11Dyadic think aloud: A think aloud produced by two participants.Think-aloud protocol: The verbatim transcription of the think aloud.Understanding: The meaning created by an individual as the result of an experience, such as,for example, reading. In this study, it is taken to mean performance on a multiple choice meaningclarification activity, performance as determined by holistic and specific trait scoring of a think aloud,and performance as determined by miscue analysis.Specific trait: One of various characteristics of student utterances produced during a thinkaloud, representing features of meaning construction, orientation toward task, or lack ofunderstanding.Specific-trait analysis: Identification of the specific traits in a think-aloud protocol.Specific-trait scoring: A profile of a think-aloud protocol, consisting of numbers and/orpercentages of specific traits identified during analysis.Holistic score: A numerical value assigned to a think-aloud protocol using a multi-item, six-point,Likert-type scale of descriptors.Informational text: Nonfiction whose main purpose is to provide factual information in a literaryway.Trade book: A book intended for sale to the general public. Such books find their way into theclassroom and school library, where they supplement the authorized and recommended texts.High score: A high comparative standing, as determined by rankings of scores obtained byboth holistic and specific-trait analysis of a think-aloud protocol, as well as by analysis of seriousmiscues made during reading. Serious miscues are defined as those that change the meaning of thetext. The miscue analysis is consistent with recommended practice for determination of level ofstudent comprehension of text, with a high score representing an independent, or easy, reading levelof less than 2% serious miscues (May, 1990, p. 388).Low score: A low comparative standing, as determined by rankings of scores obtained by bothholistic and specific-trait analysis of a think-aloud protocol, as well as by analysis of serious miscues12made during reading. Serious miscues are defined as those that change the meaning of the text.The miscue analysis is consistent with recommended practice for determination of level of studentcomprehension of text, with a low score representing a frustrational reading level of more than 5%serious miscues (May, 1990, p. 388).In preparation for the study of the specific questions identified above, a broad range of researchliterature and philosophical works was examined. The most pertinent pieces have been selected fordiscussion in Chapter Two.13CHAPTER TWO: LITERATURE REVIEWI. Introduction 14II. Social Interactionism 14Ill. Constructivism 17IV. Schema Theory 19V. Parallel Distributed Processing 22VI. Views of Metaphor 23i. Substitution View 24ii. Comparison View 25iii. Interaction View 26iv. Metaphor as Mapping 28VII. Metaphor Research 31i. Methodology 32ii. Child Focus 37a. Development of Metaphor Understanding 37b. Experience and Prior Knowledge 44c. Culture 48iii Text Focus 49a. Context 49b. Explicitness and Similarities 51c. Ecological Validity 53d. Metaphor in Informational Text 54VIII. Summary 5614I. IntroductionLythcott and Duschl (1990) identify the need for research conclusions to be based on acoherent worldview. They maintain that the worldview justifies the warrants upon which the researchconclusions must be based, the warrants being the assumptions that directly support theconclusions. Lythcott and Duschl suggest that the constructivist psychological worldview is capableof providing the necessary grounding.This study accepts the position of Lythcott and Duschl. This chapter will outline theconstructivist psychological worldview (hereafter called, simply, constructivism) and certain otherconcepts that interlock with it.The chapter will examine the following concepts from the perspectives indicated:1. Social interactionism, as a concept describing the general conditions under whichknowledge construction takes place;2. Constructivism, as a concept describing the specific manner in which the individualconstructs knowledge from experience;3. Schema theory, as a description of the manner in which knowledge is organized;4. Parallel distributed processing, as a concept describing how knowledge is processed;5. Metaphor, as a concept describing how knowledge, once constructed, feeds itself.This survey will attempt to show that the above concepts together form a simple, coherent heuristicdevice. A section reviewing metaphor research will follow the examination of the concepts. Thechapter will conclude with a summary of the central ideas.II. Social InteractionismThe concept of social interactionism is derived in large part from the work of L. S. Vygotsky.While the concept embraces all human learning, Vygotsky and those building on his work havefocused on language, in part because it provides a model case of the human developmental process,but also because, as Bruner (1986) says, language is an agent for altering the powers of15thought--giving thought a new means for explicating the world. In turn, language became therepository for new thoughts once achieved (p. 143). This examination of the concept will also focuson language, but for the additional reason that understanding of text is a central concern of this study.According to Vygotsky (1962, 1978), an individuals mental processes are social in origin--theresult of interactions of the individual with others or with cultural artifacts, such as language. Bruner(1986), a strong advocate of Vygotsky’s theories, says, “I agree with Vygotsky that there is at least onedeep parallel in all forms of knowledge acquisition--precisely the existence of a zone of proximaldevelopment and the procedures for aiding the learner to enter and progress across it’ (p. 78). Henotes that Vygotsky focuses on thought and language as two central aspects of knowledgeconstruction, with inner speech acting as a “regulatory process that, in Dewey’s famous words,provided a means for sorting one’s thoughts about the world” (p. 143). In Vygotsky’s terms, languagehas the power to shape thought and, in so doing, shape the individual’s construction of reality, while,at the same time, functioning as the “repository for new thoughts once achieved” (p. 143).Wells (1990a) also focuses on children’s relationships with text, which, in broad terms, hedefined elsewhere asany artifact that is constructed as a representation of meaning using a conventional symbolicsystem since, by virtue of its permanence and the symbolic mode in which it is created, such anartifact performs the essential function of allowing us to create an external, fixed representationof the sense we make of our experience so that we may reflect upon and manipulate it.(Wells,1990b, p. 378)This process of reflection and manipulation has the potential to be varied and rich. Wells (1990a)suggests that the written text can function as “an aid to the construction and interpretation ofmeaning,” a role served by “any form of symbolic representation that enables thought to be revisitedand revised” (p. 12). This “engagement with texts” is central to children’s developing literacy, which,in keeping with Vygotskian theory, “involves the orchestration of thinking, talking and doing, as well asreading and writing, in the construction and interpretation of meaning” (p. 13).The interpretation of meaning from text, according to Wells (1 990b), involves five modes:performative, functional, informational, re-creational, and epistemic (p. 374). While each has its place,16it is the epistemic mode whereby the writer’s “tentative and provisional attempt.. .to capture his or hercurrent understanding. ..may provoke further attempts at understanding as the writer or some otherreader interrogates the text in order to interpret its meaning” (p. 373). The “empowerment” resultingfrom the “transaction between the representation on the page and the representation in the head”contributes to levels of understanding--intellectual, moral, affective--that would otherwise be difficultor impossible to achieve” (p. 374).Vygotsky’s concept of the zone of proximal development (ZPD) is a key part of the concept ofsocial interactionism. In Vygotsky’s words,[T]he zone of proximal development.., is the distance between the actual developmental levelas determined by independent problem solving and the level of potential development asdetermined through problem solving under adult guidance or in collaboration with more capablepeers. (1978, p. 86)This ability to do more with assistance than is possible alone reveals what capacities are in the processof developing and provides an idea of what the child’s development picture will look like in the nearfuture (p. 87).The ZPD has become the focus of intense scrutiny and speculation by cognitive theorists, ashas the concept of cognitive scaffolding by means of language use. According to the latter concept,language can act as a support mechanism (scaffold) for performance within the ZPD, when providedby a more accomplished other--that is, one can do more with help than without help. The demandplaced upon the child by the need to learn to use language to establish its own context, for example,provides a clear opportunity for the process of scaffolding, which Bruner (1986> sees as “theprocedures for aiding the learner to enter and progress across” the zone of proximal development (p.78). By this process, the more proficient language user in effect provides an organizing framework ofaction or thought, within which the learner can fit the less-than-complete information that has beenlearned but which falls short of the full understanding or mastery that would mark an independentdevelopment level.The twin concepts of zone of proximal development and scaffolding (particularly languagescaffolding) are at the hub of social interactionism. Together they describe the conditions under17which individual learning and, thence, development, can occur. As will be seen later, this givessupport to the ideas that metaphor itself is experientially and culturally influenced and that metaphormeaning is not merely a linguistic given, but a cognitive construction of the individual, with an elementof commonality accounted for by social context.This study provided opportunity for each student to read a passage with a partner (dyadiccondition), then to complete a multiple choice activity individually. In this way it was hoped to gatherdata on the effect of cooperative reading on metaphor and passage understanding.Ill. ConstructivismConstructivism holds that all individuals are actively and continuously engaged in theconstruction of meaning from the flow of experience. This constructed meaning becomes knowledgefor the individual. The constructive process involves prior knowledge and current experience,linguistic and otherwise.The term constructivism encompasses a wide range of theoretical positions. At a simplelevel--as often used in education--it refers to either the undefined process a child goes through inlearning from an experience or the fact that what a child learns is, in some way, unique. At this level,there is no challenge to the existence of objective reality or to the capacity of humans to knowsomething verifiable about it. At this level, education functions on the premise that reality is waiting tobe discovered, but that individual extent of learning will vary.At the other extreme, so-called radical constructivism maintains that knowledge is simply thatwhich the individual has found to work. Radical constructivism considers knowledge to beunverifiable; rather, what maffers is that it be viable within the framework of the experience out ofwhich it arose (von Glasersfeld,1984, 1989) and that it possess a degree of stability, thus providingthe basis for choice of action within future experience. Knowledge, then, is the construction of abody of viable experience, that is, experience that fits one’s pursuit of goals, making the goalsattainable. This knowledge is built up “bit by bit in a succession of steps,” repeated often enough that18the “given experiential item has some kind of permanence” (1989, p. 442). Language provides partof the permanence, but “not even language enables a cognizing subject to get beyond theboundaries of subjectively constructed experiential reality” (p. 444).The roots of constructivism reach back at least to Giambattista Vico, whose major work, ScienzaNuova (1725), is currently undergoing enthusiastic critical examination (see, for example, Danesi,1993). Not only does Vico provide a constructivist explanation of human knowledge, hehypothesizes a significant role for metaphor in creating the surface level of mind, that part capable offorming, holding, and manipulating concepts.A particularly lucid statement of the challenges to, and implications of, the constructivist positionis given by Jerome Bruner (1986), who suggests that the cognizing individual applies preselectedsymbol systems, thus limiting the possible outcomes--the “possible worlds.” Of the challengeconfronting any theory of development, “its central technical concern,” he says, “will be to create inthe young an appreciation of the fact that many worlds are possible, that meaning and reality arecreated and not discovered, that negotiation is the art of constructing new meanings by whichindividuals can regu’ate their relations with each other” (p. 149). The cultivation of awareness ofpossibility--or, more exactly, awareness of multiple possibilities--is one of the values of metaphor. Inthis respect, metaphor can be seen to be a support for the type of theory of development favoured byBruner.Although not addressed by Louise Rosenblatt in such terms (see, for example, Rosenblatt,1978), the awareness of multiple possibilities fostered by metaphor would seem to be consistent withthe idea of an individually constructed aesthetic experience and may contribute to it. In this study,one facet of aesthetic experience--affective response--was identified as one of the specific traits ofthe think-aloud protocols and was considered to play a part in the meaning-construction process.Constructivism is consistent with the idea that metaphor is an active cognitive device, onecapable of further construction of knowledge from an individual’s prior knowledge and informationprovided by text. The relationship between meaning constructed for a text passage and metaphor19understanding is one focus of this study.The constructivist viewpoint entails an interest in the process of meaning making as well as inthe outcome. In a situation where the process is of interest, the role of memory may be a complicatingfactor. To avoid such a possibility, this study used the unprompted think aloud concurrent with thereading to gather data about the processes students employed and the meanings they constructed.To avoid the effect of reliance on recall of text during the multiple choice activity, students wereinstructed to refer to specific portions of the passage.Spivey (1987, 1990) identifies three specific constructive processes active during reading:organizing, selecting, and connecting information. In this study, these processes were evident inspecific traits identified in the think-aloud protocols. Specific traits revealed such strategies fororganizing information as paraphrasing, predicting, and evaluation of text as text, such selectingstrategies as inferencing and truth assessment, and such connecting strategies as comparing,contrasting, and linking information within the text and from the text with prior knowledge.IV. Schema TheorySchema theory relates in useful ways to both constructivism and metaphor. Anderson andPearson (1984) call a schema “an abstract knowledge structure...abstract in the sense that itsummarizes what is known about a variety of cases that differ in many particulars” (p. 259). Theschema is, then, a cognitive structure, to be identified in some measure, as noted by Rumelhart(1980), with the beta structures, frames, scripts, and plans suggested by other writers (p. 33). Theuse of the term schema itself, Rumelhart credits to F. C. Bartlett, whose 1932 work, Remembering: AStudy in Exierimental and Social Psychology, has influenced much that has occurred since in thefield of cognitive enquiry. To Bartlett,‘Schema’ refers to an active organisation of past reactions, or of past experiences, which mustalways be supposed to be operating in any well-adapted organic response. That is, wheneverthere is any order or regularity of behaviour, a particular response is possible only because it isrelated to other similar responses which have been serially organised, yet which operate, notsimply as individual members coming one after another, but as a unitary mass. (p. 201)20Rumelhart is clear that schemata represent knowledge at all levels of abstraction and are not tobe thought of as definitions. He considers them to be “recognition devices whose processing isaimed at the evaluation of their goodness of fit to the data being processed” (p. 41). He commentsthat “the fundamental processes of comprehension are taken to be analogous to hypothesis testing,goodness to fit, and parameter estimation” (p. 38), processes in which schemata play the central role,a role that he calls their “primary activity” (p. 39).Rumelhart suggests that schemata have variables related to different instantiations, that thevariables have certain constraints--knowledge about typical values of the variables and theirinterrelationships--and that they serve two purposes. These are, first, to “help in the identification ofvarious aspects of the situation with the variables of the schema,” and, second, to serve as defaultvalues that aid in the making of “initial ‘guesses’ for variables whose values have not yet beenobserved” (pp. 34-36). In other words, if only enough data are available to suggest a schema, and itseems to work to explain the given situation, then the knowledge held within the schema will comeinto play to fill in the slots for which new data have not been provided. Thus, a portion of the picture issufficient to suggest the whole picture, with the whole picture, in fact, being assumed on the basis ofthe incomplete information.Rumelhart also discusses accretion, tuning, and restructuring as different modes of learningpossible under schema theory (modes similar to Piaget’s processes of assimilation, accommodation,and disequilibration and equilibration, respectively). They are activated either conceptually--fromwhole to part--or by data action--from part to whole (pp. 41-42)--and result in new information beingadded to an existing schema (accretion), the schema being altered to take into account newinformation (tuning), or the creation of entirely new schemata (restructuring) (p. 52).There is a sense in which schemata are procedural. Rumelhart notes that they are “activecomputational devices capable of evaluating the quality of their own fit to the available data,” each withits own “network (or possibly a tree) of subschemata, each of which carries out its assigned task ofevaluating its goodness of fit whenever activated” (p. 39). He likens the structure of a schema and its21subschemata to a computer program, the intention of which is to carry out a set of procedures for apurpose (p. 40). Interestingly, Mac Cormac (1985) compares the brain to a computer, with the mindfunctioning as the software. It is but a short distance from this stage of mental model building toenvisage schemata as subprograms.With regard to reading, an important facet of schema handling is the capacity to draw inferences.To Spiro (1980), the past--that is, prior knowledge held in the form of schemata--is inferentiallyreconstructed, and it is the reader’s capacity to draw inferences when new material is encounteredthat makes text useful as a meaning-making device. Anderson and Pearson (1984) suggest fourkinds of inferences that can be identified in the reading process: first, those that have to do with thedecision to choose a specific schema in a particular situation; second, those that lead to theinstantiation of slots in the schema; third, those that have to do with choosing default values in theabsence of information; and, fourth, those involved in drawing conclusions in the face of missinginformation. Anderson and Pearson record that skill in drawing inferences seems to be age relatedand suggest that “growth in inference ability is really a difference in the growth of knowledge availablefor drawing inferences” (p. 271). This has clear significance for success in reading. It may also havesignificance for constructing meaning from metaphor, since metaphors must sometimes be thoughtover by connecting the immediate textual clues with prior knowledge in order to reach a conclusionrepresenting the most likely meaning, given the particular context.A key point is that schemata are cognitive structures, holding the knowledge one hasconstructed from previous experience, serving as models of knowledge against which newexperience is measured and classified and assigned significance. Further, schemata can be seen toprovide the raw material upon which the metaphor process works to produce its effects (Lakoff &Turner, 1989). Additionally, in this study, schema theory provided a useful way of thinking aboutconstruction of inferences and about relationships among elements of prior knowledge, both of whichwere revealed by students during their think alouds.22V. Parallel Distributed ProcessingParallel distributed processing (PDP), also termed connectionism, gives one possibleexplanation of how the brain processes knowledge. It provides a link between observablephenomena (such as human performance during experimentation) and their interpretation (in the formof experimental data and analyses), on the one side, and, on the other, examination of brain structureand function at the neurological level. In language reminiscent of schema theory, Chandler (1991)explains,Connectionist schemes. ..store information as patterns of excitory and inhibitory connectionsamong vast arrays of very simple, low-level processing units. The information content in thesystem exists not in isolatable segments of. ..memory but in the patterns and strengths of theactivated units....Based on experience (i.e., learning), a connectionist system will self-organizeinto patterns of associations among input features. (p. 240)A link between PDP and schema theory was at first only intuited, since schemata and PDP arehypothesized to operate at different levels of the cognitive structure--schemata at the macrolevel andPDP at the microlevel. Effort has been made to bring the two together (Rumelhart, Smolensky,McClelland, & Hinton, 1986).PDP suggests that in any event, such as the viewing and identification of an object, immediateneural input, stimulating numerous neurons, is processed according to certain criteria established byprevious experience and passed along by each to each of the next level neurons for furtherprocessing. This level will most likely be a so-called hidden level--one of possibly several between themore-observable input and output levels (McClelland & Rumelhart, 1986). At the level at which thetransformed neural input encounters the matching prototype, identification takes place. The numberof hidden levels of neurons may vary according to the complexity of the object or event, but the finalstage is always processing by the output-level neurons.The PDP model accounts for the speed of mental processing. According to earlierinformation-processing theory based on the operation of the digital computer, time-consumingextended sequential processing was required to accomplish complex tasks. On the other hand, thedistribution of processing of multiple stimuli across a neural network at each processing level allows23each neuron to contribute to the accumulated refined stimuli sent on to the next level, thus collapsinga multiple-stage process into one stage, with an attendant time saving (Norman, 1986).Chandler (1991) proposes a connectionist explanation of metaphor processing. Nouns andverbs, the key elements in many metaphors, he suggests, “are associated with conceptual prototypesembodied within a connectionist network,” with the patterns for nouns established between“microfeatures” and those for verbs among “temporal sequences of microrelations” (p. 253). Hecontends that both metaphorical and literal sentences “lead us to construct mental imagesconsiderably different from what most of us would consider the prototypical exemplars for the keyword” (p. 233). As a result, “there are no differences between on-line processing of metaphoric andliteral expressions” (p. 232). Similarly, Waggoner, Meese, and Palermo (1985) conclude from theirexperiment “that the process of comprehending metaphor is the same in kind as the process ofcomprehending literal language” (p. 1164), although their conclusion is based on the absence ofdifferences in performance on recall tasks involving metaphor and literal language.If PDP is, indeed, found to work in conjunction with schemata, it may provide an importantperspective from which to reexamine the cognitive function of metaphor, which, in turn, ishypothesized to be related to schemata (see, for example, Ortony, 1980, and Lakoff & Turner, 1989).VI. Views of MetaphorIn the twentieth century, metaphor has become a subject of considerable interest, an interestMark Johnson has called ‘metaphormania” (1981, ix). Views of the nature of metaphor haveundergone revision and new views have been developed.The language used to describe the structural elements of metaphor has varied somewhat overtime. For the standard form, A is B, A has been termed çpIg or tenor and B, vehicle. Lakoff andJohnson (1980) and Lakoff and Turner (1989) use target and source for the two parts, A and B,respectively. This thesis, drawing on the inspiration and ideas of Lakoff, Johnson, and Turner in anumber of ways, adopts their terminology.24The following sections present four views of metaphor. The first three--substitution,comparison, and interaction--provide a backdrop against which to examine the fourth--metaphor asmapping--the view that informs this study.i. Substitution ViewFor over two millennia, from the time of Plato and Aristotle to the present, metaphor has beenregarded with varying degrees of curiosity and misgiving by many who have given it close scrutiny.The curiosity and misgiving have been linked to the seemingly anomalous nature of metaphor, for,according to the substitution view, metaphor appears to be, at best, a verbal embellishment or, atworst, a misleading deviation from the sought-for clear expression presumed to reside in literallanguage (see, for example, Winner, 1988).The perceived deviance of metaphor results from the underlying belief that language is, at itsbase, literal, and that it is in literal expression that meaning resides. Where metaphor is involved, clarityof expression is considered to depend on the translation of the figurative language into literallanguage. According to this view, metaphor is an inexact substitute for literal language, and the exactmeaning is determined only by identifying the appropriate literal language to replace the metaphoricallanguage.The substitution view is linguistically based and depends on a two-value logic in whichpropositions are either literally true or false. Further, propositions are considered to be about anobjective reality that is independent of any observer; that is, the propositions and the objective realityupon which they rest are said to be mind-free--to have meaning in and of themselves. Thepropositions are to be considered true insofar as they accord with that objective reality. Metaphorsmust, then, be false because they make statements that cannot be verified by reference to objectivereality. They can be considered true only in the non-reality of the imagination.Considered according to the substitution view, a metaphor such as, for example, LIFE IS AJOURNEY is a rhetorical embellishment. In and of itself, it has no truth value--since it must be treated25as a literal statement—and therefore no meaning. A speaker or author of such a metaphor might bejudged to want merely to draw attention to life for the possible purpose of suggesting that the listeneror reader should pause and think about it. On the other hand, the speaker or author might be judgedto have meant to say something else entirely.ii. Comparison ViewThe original form of the comparison view of metaphor, traceable to Aristotle, holds that themetaphor process involves comparison of characteristics of the referents of the words that composethe two parts of the metaphor (Winner, 1988). The referents must have common characteristics, oftentaking the form of perceptual attributes. A true literal paraphrase must be possible for the metaphor.For the LIFE IS A JOURNEY metaphor, the paraphrase could be, for example, ‘Life is like a journeybecause they both have a beginning and an end and a limited number of other sharedcharacteristics.” This view treats metaphor as a form of simile, although not as explicit as the formusing ifl or .I. A. Richards (1936) offers a refinement of the comparison view of metaphor. According toRichards, metaphor is a means of comparing “thoughts,” and the “ground” of the comparison is the“common characteristics” of the metaphor’s two parts--the “tenor” and the “vehicle” (terms heintroduced). By focusing on thought, Richards moves the consideration of metaphor beyond thenarrow rhetorical base of the substitution view and the older form of the comparison view. What heproposes is an analogical model, though with awareness of the potential contribution of thedifferences, as well as the similarities, between tenor and vehicle in shaping the meaning of themetaphor.According to Richards’ comparison view, a metaphor involves a comparison between thecommon characteristics, which may be features or attributes of the entities, themselves, or semanticfeatures activated by the names of the entities. When a metaphor is analyzed, attributes and featurescommon to both vehicle and tenor are sought. These can form the basis for a meaningful26metaphorical comparison if there is sufficient difference in salience, that is, if one or more features orattributes are more dominant characteristics of the vehicle than of the tenor. Where there is sufficientsalience imbalance, the comparison will cause a distortion of the pertinent characteristics of the tenor,causing it to be viewed in a new way.Richards comparison view, like the old comparison view and the substitution view, is based onthe literal meaning view of language and the acceptance of an objective reality. To form a truemetaphor, the tenor and vehicle must have features or attributes in common and their existence mustbe verifiable through reference to objective reality. A literal paraphrase must be possible. For theLIFE IS A JOURNEY metaphor, a paraphrase might be, for example, Life is a journey because theyhave a number of common characteristics that allow me to think about life in a novel way.’iii. Interaction ViewThe interaction view is a recent attempt to avoid problems posed by the substitution andcomparison views. As the name suggests, the interaction view sees the meaning of metaphor as theoutcome of an interaction between the two parts of the metaphor, rather than resulting from asubstitution of literal language or a simple comparison of features or attributes. Of central importanceis the idea that metaphor involves the interaction of concepts.Max Black (1962) starts with Richards’ idea that the metaphor process involves the mind holdingtwo thoughts simultaneously, both thoughts being represented by a single word or phrase. Blackproposes that the interaction of the two parts of the metaphor results in a meaning that cannot bereduced to mere literal paraphrase of the expression or a simple statement of the sharedcharacteristics. He suggests that metaphor is a kind of filter that controls which of the culturally based,associated commonplaces of the two ideas enter the metaphorical interaction.The functions of source and target are different from their functions according to thesubstitution and comparison views. In the interaction view, the target imposes a filter that screens andlimits the characteristics and meaning that can be carried over from the source. This imposes a kind of27bidirectionality that is quite different from the unidirectionality of the substitution and comparisonviews. Accordingly, the source concept, as well as the target concept, is deemed to undergorestructuring. In the LIFE IS A JOURNEY metaphor, for example, the life concept sets up a filter thatpermits only certain characteristics of the journey concept to reach and act upon its characteristics,thereby altering them. The interaction produces a meaning for the metaphor that is too complex to bereduced to literal paraphrase.The nature of the filter has been given considerable attention by proponents of the interactionview. Way (1991) offers one description. She suggests a “dynamic type hierarchy” in which thesource and target are reclassified as types within a higher type (category). The characteristics of thetarget determine the nature of its membership in the higher type, and these parameters ofmembership become the filter through which the source’s characteristics, themselves havingundergone a similar process, are limited prior to transfer to the target.Apart from any descriptive and explicative value it may have, Way’s description, drawing, as itdoes, from information processing theory, is interesting as an example of the contributions tometaphor study being made by people from many academic disciplines and philosophical positions.Glucksberg and his associates (Gildea & Glucksberg, 1983; Glucksberg & Keysar, 1990;Glucksberg, 1991) have put forward a strong class-inclusion model to explain the function ofmetaphor. According to this view, a metaphor is a true statement, not a false literal statement, sincethe truth of the metaphor is apparent in the inclusion of both the target and source terms within a new(perhaps newly created> superordinate class initiated by the source term. In this view, no specialpsychological mechanisms are required to understand metaphor, and the meaning of a familiarmetaphor will be activated automatically as quickly as a literal meaning for a literal expression.Winner (1988) proposes that pragmatic elements need to be considered in any attempt toexplain how metaphor works. She accepts the interaction view; however, she rejects a semanticapproach to explaining metaphor, preferring, instead, a pragmatic approach that places “metaphoricityoutside the sentence--in the speaker’s use of the sentence, and hence in the speaker’s intentions”28(p. 23).An appealing feature of the interaction theory is that it leaves room for some kinds of metaphorsto function according to the comparison view. The interaction view (or some elaboration of it) seemsto be the currently dominant theory, with metaphor research generally using it as a starting point,although the comparison view provides a convenient and often-used platform from which to launchdescription of the metaphors themselves (Broderick, 1991; Evans & Gamble, 1988). Ortony (1980)contends that comparison, “the ability to perceive similarities and differences” (p. 363), is a centralcognitive process involved in much of the mind’s activity and is inescapable when metaphor isunderstood, whether according to the comparison or the interaction view. Such a position seems toentail the primacy of comparison, with metaphor the observable outcome of the process.iv. Metaphor as MappingLakoff and Johnson (1980), Lakoff and Turner (1989), and Lakoff (1993) put forward a view ofmetaphor that, while rejecting the idea of interaction between the two parts of the metaphor, doesshare certain features with Black’s view. Agreeing with Black, Lakoff, Johnson, and Turner holdmetaphor to be a conceptual entity firmly grounded in human experience, having the power toproduce meaning inexpressible by literal paraphrase. They maintain, however, that themetaphor-making process involves a one-way mapping from the source to the target--a process of“understanding and experiencing one kind of thing in terms of another” (Lakoff & Johnson, p. 5). Themapping may be of schema slots, relations, properties, or knowledge in the source domain (Lakoff &Turner, pp. 63-64). These mappings have the power to create new understanding (Lakoff &Johnson, p. 139) and to create similarities between conceptual domains where similarities did notpreviously exist (pp. 148, 150-153). These specific knowledge elements of the shared experienceon which the metaphor is grounded and the structural features of the source domain, as well as therelationships among those structural features, including the affective and orientational, togetherdefine experience--humans’ way of being in relation to the world. Further, against a cultural backdrop,29Lakoff, Johnson, and Turner portray metaphor as the fundamental way of interpreting experience andstructuring thinking and as a device to aid understanding--one without which our ways of thinking andour conceptions of reality would necessarily change (Lakoff & Johnson, pp. 3, 145; Lakoff & Turner,p. 51).Lakoff and Turner give these ideas a literary connection. They focus beyond the basicmetaphors in everyday life to the expression of these metaphors in our poetic existence. Accordingto Lakoff and Turner, these metaphors, except for certain idiosyncratic and image metaphors, aregrounded in a shared human experience that gives them immediacy of meaning, generalrecognizability, the power to connect us with each other through our language, and the power togenerate new meaning. They note the power of poetic metaphor to extend conventional metaphormeaning, to elaborate schemata in nonconventional ways, to question the limits of understandingexpressed via conventional metaphor, and the formation of composite metaphors whose power farexceeds that of ordinary metaphors (pp. 67-72). Conceptualized in this way, metaphors, because oftheir experiential basis within a particular cultural environment, can provide an understanding of bothhow and why we think as we do.Lakoff and Johnson identify three main types of metaphor: structural, orientational, andontological. Structural metaphors are those in which ‘one concept is metaphorically structured interms of another” (p. 14). The example used throughout this introduction to metaphors--LIFE IS AJOURNEY--is a structural metaphor. It functions by mapping elements of the knowledge structure ofthe conceptual domain journey onto the conceptual domain jj, restructuring it so as to highlightcertain features and to hide others. Mappings may include, for example, the following: traveler fromthe source domain onto person in the target domain; destinations onto purposes; routes onto meansof achieving purposes; distance traveled onto progress in life; crossroads onto choices; andpro4disions onto material resources (Lakoff & Turner, pp. 3-4).Orientational metaphors originate from our awareness of spatial relationships, such as“up—down, in-out, front-back, on-off, deep-shallow, central-peripheral” (Lakoff & Johnson, p. 14).30Examples of orientational metaphors include HAPPY IS UP and SAD IS DOWN, as in the commonmetaphorical expressions “My spirits r” and “I’m feeling down” (p. 15).The third type of metaphor--ontological metaphor--takes two forms: entity and substancemetaphors and container metaphors. The entity and substance metaphors are “ways of viewingevents, activities, emotions, ideas, etc., as entities and substances” (p. 25). THE MIND IS A MACHINEand THE MIND IS A BRITTLE OBJECT, as illustrated by the common metaphorical expressions,“We’re still trying to grind out the solution to this equation” and “The experience shattered him” (pp.27-28), respectively, are two such ontological metaphors. Container metaphors, on the other hand,originate in our experiences of land areas and visual perception. Common metaphorical expressionssuch as “There’s a lot of land in Kansas” and “That’s the center of my field of vision” (p. 30) areexamples. Ontological metaphors of these two types--entity and substance metaphors and containermetaphors--provide a powerful way of structuring our awareness of our social, emotional, andintellectual experience. Specifically, according to Lakoff and Johnson, “events and actions areconceptualized metaphorically as objects, activities as substances, states as containers” (p. 30).Lakoff and Turner distinguish between generic-level metaphors and specific-level metaphors.The former, such as PURPOSES ARE DESTINATIONS, EVENTS ARE ACTIONS, and STATES ARELOCATIONS (p. 52>, lack “fixed source and target domains” and have no “fixed list of entitiesspecified in the mapping” (p. 81). The latter, such as LIFE IS A JOURNEY, on the other hand, havefixed domains and lists of entities to be mapped (pp. 80-83), as indicated above.Lakoff and Johnson’s idea of the grounding of metaphor places emphasis on the culturalcontext in which the experiencing individual operates. They state,Cultural assumptions, values, and attitudes are not a conceptual overlay which we may or maynot place upon experience as we choose. It would be more correct to say that all experience iscultural through and through, that we experience our “world” in such a way that our culture isalready present in the very experience itself. (p. 57)One effect of culture is to provide coherence to the system of metaphors; that is, there will be a limitednumber of major conceptual metaphors, such as LIFE IS A JOURNEY, but a largernumber--sometimes a very large number--of metaphorical expressions of them, all consistent with31them and all mapping smaller features of the source conceptual domain onto the target conceptualdomain (see, for example, pp. 17-19, 52-55, and Lakofi & Turner, p. 51). For LIFE IS A JOURNEY,this coherence can be seen in such metaphorical expressions as “passing away,” referring to death,and ‘getting sidetracked,” referring to diversion from goals (Lakoff & Turner, pp. 2-3). For this reason,while there is variation in the way different cultures systematize their metaphors, there is a high degreeof coherence within each culture (Lakoff & Johnson, pp. 24, 118, 146). In this study, an attempt wasmade to avoid effects produced by diverse cultural backgrounds by choosing students from along-established, relatively stable, primarily unilingual cultural subgroup.The view of metaphor as mapping has been adopted in numerous research studies and hasbeen examined by philosophers of language. While not every aspect of this view is accepted by allwriters, three main ideas do seem to be accepted: first, that restructuring of knowledge in the targetconceptual domain takes place in line with knowledge in the source domain; second, that much(perhaps most) metaphor has a cultural basis; third, that metaphor strongly influences how we think.Some writers see a view of metaphor such as this to be a variation of the interaction view and, as such,subsumed under it (see, for example, Winner, 1988, and Way, 1991). While this point is notsignificant for this study, this position is specifically rejected by Lakoff and Turner (1989), who maintainthat the mapping is unidirectional from source to target.The ideas of Lakoff and Johnson (1980), Lakoff and Turner (1989), and Lakoff (1993) providethe theoretical orientation toward metaphor adopted in this study, with the single exception thattaxonomies developed by Broderick (1992) for the purpose of categorizing metaphor according tothe type of comparison and syntactic frame have been used to provide additional description ofstudent performance on the multiple choice activities.VII. Metaphor ResearchThe study of children’s understanding of metaphorical language is a relatively new area withinthe broader research field bringing together education and psychology. A key question that has32emerged is: How well do children understand the metaphors they read?This review considers the literature on metaphor research from three perspectives:Methodology, Child Focus, and Text Focus. This organization reflects three main concerns evident inthe literature and pertinent to this study. The divisions are, of course, not discrete and certain studieshave contributed in more than one area. Child Focus is reviewed under the following subheadings:Development of Understanding, Experience and Prior Knowledge, and Culture; and Text Focus isreviewed under the subheadings: Context, Explicitness and Similarities, Ecological Validity, andMetaphor in Informational Text.i. MethodologySeveral methodological problems have affected metaphor research. Vosniadou, Ortony,Reynolds, and Wilson (1984) identify three specific problems: that researchers have sometimesconfused children’s lack of background knowledge with failure to understand metaphors; thatmetaphors have often been presented to children with insufficient attention to context; and that therehas been an over-reliance on paraphrase as a measure of children’s understanding of metaphor (p.1589).Winner (1988) notes that some studies have used verbal measures, such as “asking children toparaphrase metaphors. ..to select the appropriate verbal paraphrase.. .or to judge whether sentencesmake sense or are anomalous,” while others have used nonverbal measures, such as having childrenpoint to pictures to indicate choice of meaning, having the children match words and visual stimuli,according to metaphoric similarity, having the children act out the metaphors using toys, and havingchildren match only nonverbal stimuli (p. 45). Based on her examination of studies using the abovemeans of testing children’s understanding, Winner observes that “[t]he measure used dramaticallyaffects the level of comprehension revealed. The greater the reliance on linguistic and metalinguisticabilities, the more likely is the measure to reveal a low level of understanding” (p. 45). Ortony (1980)agrees. He says, “It is well known that the skill to verbally explain one’s comprehension of something33develops later than the ability to actually comprehend” (p. 356).Vosniadou (1987) makes the same point, and she notes that “paraphrase and explanation arepoor measures of metaphor comprehension because they impose linguistic or metacognitivedemands well in excess of those required for metaphor comprehension alone” (p. 877). She goes onto suggest that multiple choice and enactment tasks are preferable to paraphrase tasks, “presumablybecause the former impose fewer linguistic and metacognitive demands than the latter’ (p. 877). Thechallenge to researchers, she says,is the development of better tasks to assess metaphor comprehension, tasks that do notconfound metaphor comprehension with lack of background knowledge, metalinguistic skill,ability to comprehend language out of context, and other variables that do not have much to dowith metaphoric competence per se. (p. 878)In a relatively early study examining children’s understanding of metaphor in narrative text,Cunningham (1976) presented grade six students with two doze passages, similar except that onewas non-metaphorical and the other heavily metaphorical. He found a significant difference in studentperformance on the doze tasks (t test, p<.OO1), from which he concluded that the presence ofmetaphor increased the reading difficulty of the passage compared to the no-metaphor passage.Cunningham’s methodology raises three specific concerns: First, the deletions were notparallel in the two texts for parts of speech; second, the scoring of the doze tests was done “against akey of exact deletions” (p. 367); third, the metalinguistic demands of the doze test were not taken intoaccount. Of the 25 blanks, only 11 required the same part of speech in both texts. The second blankin the no-metaphor passage, for example, required the possessive pronoun jj and the second blankin the metaphor passage required the noun mother. Both of these were strongly indicated bycontext, although the second required reading ahead to the next sentence for confirmation. Thesixteenth blank in the no-metaphor passage, on the other hand, required the verb while the sameblank in the metaphor passage required danced. Common usage made the former a likely choice,whereas the latter was highly arbitrary.The specific examples of metaphorical language chosen by Cunningham also raise questions.There is a mix of slang expressions, such as “a peach of a thing,” replacing “a great thing,” and34idiosyncratic expressions, such as “I got ivory faced,” replacing “I got real excited.” Such slangexpressions are likely to have a high degree of familiarity, perhaps on a level with the literal substitute,while the idiosyncratic expressions are entirely arbitrary and without contextual support.Results are further complicated in that each student had both forms of the doze test, varied onlyto the extent that the order in which the passages were stapled together was reversed in half thecases. Since a set time (30 minutes) was given, it is possible that students could have compared thetwo passages, thus arriving at an explanation of otherwise obscure phrases or even precise words, asituation that may well have favoured completion of the no-metaphor test. There is no indication,however, of the extent to which students referred back and forth between the passages.Vosniadou, Ortony, Reynolds, and Wilson (1984) undertook to examine factors affectingchildren’s understanding of metaphor in three experiments designed to avoid the methodologicalproblems mentioned above. Rather than depend on children’s ability to paraphrase, theexperimenters had the children enact their interpretation of the target statements, which consisted ofboth probable and improbable literal and metaphoric endings to specially constructed narratives, readto them by the experimenters. The three experiments indicated that close attention to the factorsidentified as confounding variables can result in a somewhat different picture of children’sunderstanding of metaphor, one that has the children able to understand metaphor at a younger agethan had hitherto been indicated by experimentation. A major finding of these experiments was that“complexity of the linguistic input is...an important variable in metaphor comprehension” (p. 1603).This serves as an important caution to those undertaking metaphor research.A possibly significant feature of the experiments unaccounted for by Vosniadou et al. is that themetaphors and literal targets were all placed at the ends of the narratives. Although this standardizesthe task in one respect, it may introduce a complicating factor, one identified by Waggoner, Meese,and Palermo (1985), who investigated children’s recall of metaphor within a story grammar structure.They contend that placement of the metaphors at the outcome position, according to story grammar,provides children with a structure they can “use not only to aid recall” but “to succeed in the more35difficult task of verbalizing the meaning of the metaphors” (p. 1163). They conclude that a statementin a story is easy or difficult to understand “in the context of that story and not as a function of whetherit [is] expressed in a literal or a metaphorical form” (p. 1164). Since Vosniadou et al. placed all thetargets at the outcome position, they may inadvertently have introduced an effect to be accounted forby story grammar.In her study of children’s production of metaphor--a branch of research that has paralleledstudies of metaphor understanding--Vosniadou (1987) adopts two necessary criteria for children’sutterances to be classified as metaphors: first, “the utterance must... be based on some perceptiblesimilarity between the two juxtaposed objects,” and, second, “the child must be aware that the twoobjects belong to different conventional categories” (p. 874). These criteria are based on thecomparison view of metaphor. Although they are useful for examining many metaphors, perhapsespecially those of young children, they fail to take into account that metaphors can create conceptualsimilarities where none hitherto existed (Lakoff & Johnson, 1980, p. 147). Interestingly, Vosniadou,herself, makes a similar point when she remarks that metaphor provides a way for children “tounderstand new phenomena, phenomena that are not quite similar to anything they haveexperienced before” (p. 822). She does not, however, reexamine the first of the two criteriamentioned above in light of this understanding of this special power of metaphor, nor does shereexamine her view of metaphor itself.Vosniadou has carried out research on occasion with Andrew Ortony (see for example,Vosniadou et al., 1984), and her view of metaphor is similar to his. Ortony (1980) claims what might betermed a pracmatic view of metaphor. He describes metaphor as any intentionally anomalousutterance for which the tension (the incompatibility between the two parts) “is, in principle, eliminable”(p. 352), thus embracing both the comparison and interaction views of metaphor. Ortony suggests acontext-sensitive activation mechanism” (p. 360) to link schemata and their subschemata in a searchfor “salient aspects” (similarities) between the two parts of the metaphor (p. 359). He suggests that“there do exist more probable and preferred values” for the variables of the schemata and that36“staying within reasonable limits” would be an important control factor in establishing links betweenthe schemata representing the two parts of the metaphor (p. 361). The idea of salience imbalancebetween salient features linking the two parts of the metaphor has been adopted by researchers (see,for example, Evans & Gamble, 1988, and Mate & Malicky, 1988). From the point of view adopted inthis study, such a position is seen as limiting in that it focuses on those metaphors whose source andtarget have identifiable preexisting similarities.Winner (1988) adopts the interaction view of metaphor (p. 23), although she elaborates upon itby proposing that metaphor understanding proceeds according to pragmatic principles, whereby thelistener must first detect nonliteral intent, then detect the relationship between sentence meaningand speaker meaning, and, finally, detect the speaker meaning itself (pp. 10-11). Such a processwould lead to a longer processing time for metaphors than for literal language, since the steps wouldhave to be performed sequentially. Research has attempted to detect time differences forunderstanding of literal and metaphoric language (see, for example, Gregory & Mergler, 1990).Findings have not been consistent, but the bulk of the evidence seems to suggest that there is nosignificant difference in processing time (Winner, p. 125). Winner reports this tentative conclusion asa refutation of a stage theory of metaphor-understanding development in children. In the currentstudy, differences in processing time are not investigated, since no attempt is made to addressquestions of psychological process.The literature reviewed above identified the following seven factors as having methodologicalimplications:1. Children’s lack of background knowledge;2. Effect of insufficient context;3. Over-reliance on paraphrase;4. Inappropriate stimulus sets and passages;5. Over-complexity of linguistic input;6. Effect of story grammar; and377. Operational definition of metaphor.In the present study, children’s background knowledge is of interest insofar as it influencesunderstanding of the metaphors, where that can be determined. Since the research interest is in theextent of understanding and such processes as are evident in the think alouds, no attempt is made todetermine level of background knowledge, to compensate for its lack, or to make fine judgmentsabout complex psychological processes. Numbers 2, 3, 4, and 6 are all taken into account by,respectively, providing lengthy passages, by having students complete multiple choice activities, byproviding authentic texts, and by using expository text. Number 5 is of interest as a factor influencingmetaphor and passage understanding, and the three passages used are of different reading levels.Number 7 is taken into account by defining metaphor broadly, as explained above.ii. Child Focusa) Development of Metaphor UnderstandinciA considerable amount of research has been undertaken in an attempt to determine the age orages at which children begin to demonstrate the ability to produce and to understand metaphor.Within this general area, one aspect has attracted particular interest, that is, whether there aredevelopmental stages in metaphor understanding. Part of the interest is related to the predictions ofPiagetian psychology that certain stages of cognitive development must be reached before a child iscapable of the complex recategorization hypothesized to take place when metaphors are produced orunderstood, according to comparison and interaction views of metaphor.In a study of poor and good readers in grades three and five, Seidenberg and Bernstein (1988)examined their data for effects of grade on interpretation of absurdities and ability to name criticalsemantic attributes, two elements they held to be similar in some respects to metaphor interpretation(p. 41). They found a grade effect for absurdities among the good readers and for ability to namecritical semantic attributes among poor readers, suggesting that these two elements follow adevelopmental pattern, at least for the groups studied. Correlations were also significant between38metaphor scores and scores on absurdities for the high-ability group and between scores onmetaphors and scores on synonyms for the low-ability group, which is taken to suggest “that factorsimplicated in the high and low group’s [sic] interpretation of metaphor may differ” (p. 43), suggestingeither a different developmental rate or route.Seidenberg and Bernstein conclude that “ability to name a critical corresponding attribute is nota sufficient condition for metaphor interpretation” (p. 44). Drawing on extensive literature in the areaof metacognition in relation to interpretation of literal statements, they suggest that the samemetacognitive strategies are necessary for children to “resolve comprehension failure for metaphoriclanguage” (p. 44). This conclusion may turn out to be supported by future research, but it raises otherissues, not least of which is the question of whether the processes for understanding literal andmetaphoric language are the same, a question to which there is not at present a clear answer.Seidenberg and Bernstein’s view that metaphor understanding is not a simple one-stepprocess is also held by Gregory and Mergler (1990). In a study of adults’ response time to the task ofclassifying true or false statements for literal truth, possible sense, or metaphoricity, they concludethat readers “make use of a variety of processing strategies” (p. 170) and that familiarity or unfamiliaritywith metaphors may, in itself, be a significant factor (p. 171). The linguistic demands of the task werehigh in that participants were required, among other things, to identify when metaphor-typestatements were false. This would suggest that at least some of the processing strategies aremetacognitive and, as such, developmental (see, for example, Brown, 1980). On the other hand, thestimulus sentences were presented in random order with no contextual support, raising a question asto whether the generally longer time required “to deny metaphoricity or possible sense than to affirmit’ (p. 167) would hold in a more typical reading situation.Siltanen (1989) details a four-stage developmental model of metaphor understanding based onthe types of conceptual categories six- to twelve-year olds create during a metaphor-understandingtask (p. 201), and she refers to a “developmental pattern through which children progress in acquiringadult metaphor comprehension” (p. 208). Consistent with this, Siltanen (1990) observes that “as39children [again, six- to twelve-year olds] develop cognitively and linguistically, they provide moreelaborate metaphor comprehension” (p. 15).Siltanen bases her work on Piagetian theory, according to which, she says, themetaphor-understanding process involves resolving cognitive disequilibrium set up by theincongruence between the two parts of the metaphor by “constructing a new joint category based onperceptual or conceptual grounds” (1989, p. 200). The view of metaphor understanding as areclassification process by which a superordinate category is found or created to subsume the twoconceptual categories represented by the two parts of the metaphor has been explored; however, ithas a serious weakness in that it fails to account for those metaphors in which no attributes are held incommon by the source and target. (For a possible way around part of this problem, see Glucksberg &Keysar, 1990.)A methodological weakness of Siltanen’s studies is that she required verbal interpretation of themetaphors. She defends this as being “an ecologically valid task,” that is, one similar to a real-life task(1990, p. 11). As was indicated above, however, this approach favours older children and fails toreveal the full extent of younger children’s knowledge.Using a very different research approach, Pearson (1990) asked three- and four-year-olds torepeat metaphorical, literal, and anomalous sentences, then compared the number of errorsproduced during the repetitions. According to Pearson, studies have shown this approach todifferentiate between what is meaningful and non-meaningful to adults and children alike, whilereducing the linguistic demands of such methods as paraphrase. Pearson found that the 52 childrenin her study produced very similar levels of error in the repetitions for the metaphorical and literalsentences but a significantly higher level for the anomalous sentences. From this, she concludes that“children are not strangers to metaphor. They are here seen processing the figurative sentences on apar with the literal” (p. 197). Also, she found no evidence for “a development with age in the capacityto repeat metaphors” (p. 197).In a study of recall of metaphor and literal statements by seven-, nine-, and eleven-year-olds,40Waggoner, Meese, and Palermo (1985) report a grade effect for recall of both metaphor and literalstatements embedded at the outcome node of a story schema and for the subsequent sentences.They conclude that “children as young as 7 years of age can and do correctly interpret metaphors”that are embedded in a story schema, suggesting that the story schema provides a structure thatreduces the “difficulty of the comprehension task so that it is within the child’s difficulty limit both torecall and to verbalize the meanings of the metaphors” (p. 1163). If Waggoner, Meese, and Palermoare correct, then predictable text structure, such as the so-called story grammar common to fiction(see, for example, Stein & Glenn, 1979), can become an aid to understanding metaphors undercertain conditions. This introduces a factor that must be taken into account in metaphor researchinvolving a recognizable story structure. In the present study, the effect of story structure is avoidedby using informational text.Kincade (1991) examined grade two and grade five children’s ability to recall explicit, implicit,and metaphor information in short narratives. She reports a grade level effect for recall of all threetypes of information. Of particular interest, however, is the finding that the performance of thechildren at both grade levels improved when probed recall was used, rather than free recall,suggesting “that the structured questions provided external organizational strategies and cues thatelicited information from the children’s memory” (p. 93). This finding is to be expected, given that“external organizational strategies” remove a considerable burden, especially from young childrenwho, as mentioned above, have been shown to know more than they can explain. Thus, whiledevelopmental differences are evident, they are not differences in kind, only in degree.Research findings are not consistent in the matter of the order of development ofunderstanding of abstract and concrete similarities involved in different types of metaphors. Based ontheir study, Silberstein, Gardner, Phelps, and Winner (1982), for example, contend that “childrenunderstand and produce perceptually based metaphors before conceptually grounded ones” (p.148). On the other hand, Broderick (1991), on the basis of a study in which he asked preschoolers tomatch stimulus pairs presented in physical or verbal form by the experimenter, reports that “children41are able to relate concepts in both concrete and abstract fashions from the earliest period of languagedevelopment” (p. 79). In a comment on the effect of methodology, Broderick observes, “Whenextraneous processing demands are eliminated, psychological/physical metaphor does not appear tobe a late development” (p. 77). The procedures of Silberstein et al., unlike those of Broderick, would,indeed, make linguistic demands (for example, choosing one of five possible endings to complete ametaphor) that may cause younger readers (as young as age eight in the study) difficulties that areunrelated to the metaphors themselves.Dent and Rosenberg (1990) also used physical objects to investigate children’s developmentof ability to understand visual metaphors and metaphoric similarity and to compare with adultperformance. They report that children “improve from ages 5 to 7 in comprehension of visualmetaphors, in which topic-vehicle interaction is explicit, and gradually from 5 to 10 in comprehendingmetaphoric similarity” (pp. 993-994). They report no stage-like development.In discussing their results, however, Dent and Rosenberg note that the types of questions theyasked may have directed the participants to provide certain kinds of responses, rather than others.They admit that their results “indicate complex relations between perceptual and discourse context”and that they “do not yet know in detail how discourse requests direct attention to objects, layouts,etc. that can be perceived” (p. 992).In a study comparing the metaphor understanding of verbal and picture stimuli of grade two andgrade five children, Kogan and Chadrow (1986) report a “large performance difference” in favour ofthe older children (p. 292). Additionally, they consider that there is no “generalized advantage foreither pictures or words” (p. 294), suggesting that metaphor understanding is not a function of onemodality more than another, although performance using pictures was enhanced when the pictureshad descriptive labels. While such a finding may be confusing if one looks only at perceptual,affective, and conceptual similarities between entities when explaining the metaphor process, it isconsistent with the metaphor as mapping of information view, according to which preexistingsimilarities are not required for the formation of a metaphor, but may instead result when the metaphor42is created. In the present study, the latter, broader view of metaphor avoids the limitations of anexisting-similarities-only approach.Vosniadou, Ortony, Reynolds, and Wilson (1984) examined the sources of difficulty youngchildren experience in understanding metaphor. By having preschool, grade one, and grade threechildren enact endings of stories, they were able to determine that understanding of concludingmetaphors was affected by the probability of the ending and the overall complexity of the linguisticinput. They contrast their findings with those of investigations based on Piagetian theory and taketheir own results as a refutation of those others. They take the fact that preschoolers understandmetaphor to some level under some conditions as a clear indication that, to understand metaphor,children do not have to reach “formal, or at least late concrete operational, thinking,” a prerequisite inPiagetian theory (p. 1603). They state that while “metaphor comprehension is a progressivedevelopment that starts quite early” (p. 1 601), “findings are inconsistent with the position that thedevelopment of metaphor understanding follows a clearly identified sequence of stages, which startswith literal responses first, and only later follows with more mature types of metaphoric understanding”(p. 1604).Vosniadou (1987) provides a useful picture of the metaphor-development question. From herinterpretation of the literature, she says thatin the most recent psychological literature, metaphor comprehension is conceptualized as acontinuous process that starts very early and develops gradually to encompass a greater varietyof metaphorical linguistic inputs... .The development of metaphor comprehension isconstrained primarily by limitations in children’s conceptual knowledge, linguistic skill, andinformation-processing ability.. ..What develops, according to this view, is the ability tounderstand more complex metaphorical inputs in a variety of linguistic and situational contexts.(p. 880)Implicit in much of the above research is the view that the processes for understanding literaland metaphorical language are essentially different, and further, that, of the two, literal language isbasic and metaphor is an add-on. This relative status of the two modes of language is denied by theview of metaphor proposed by Lakoff, Johnson, and Turner and by Vico, as described above; indeed,Vicos descriptions of the development of mind and language clearly portray metaphor as the basic43mode of thought, as well as of language. As yet, major research into the development of metaphorability has not been undertaken within a metaphor-first paradigm.Perhaps a major shift in the research paradigm is underway under the influence of theconnectionist (PDP) view of cognition. Such a shift would approach the two modes of language (literaland metaphorical> as essentially the same, a position described by Chandler (1991) in his explanationof a unified-process model, according to which literal and metaphorical language are processed in thesame way (p. 232). Chandler admits that “connectionism is still very limited in what cognitiveprocesses researchers can model well with it” (p. 239), but his reinterpretation of metaphors thatothers had interpreted in such a way as to support the literal-language-first view suggests thatconnectionism has the potential to inform research in significant ways.Vosniadou et al. (1984> provide significant incidental support for Chandler’s model. They notethat “children’s metaphoric thinking seems to be more flexible than generally believed” (p. 1604).This would be consistent with a model based on interconnections and interactions within and amongconceptual domains. Vosniadou et al. found, as well, that the children in their experiments had moredifficulty enacting metaphors involving nonliteral verbs than they did those involving nouns. Thiswould be consistent with a connectionist model in which the processing of verbs, as Chandler pointsout, introduces a temporal element not present with the processing of nouns (1991, p. 253).Keil (1986) studied the metaphor understanding of kindergarten, grade two, and grade fourchildren. The metaphors were constructed from paired semantic domains, such as animate-inanimateand idea-plant. From the children’s explanations of the metaphors, Keil determined that older childrenproduced more adult-like meanings, indicating a developmental aspect. The use of paraphrase mayhave biased the results, for reasons already mentioned; however, Keil found a bimodal distribution ofresponses for all grade levels, indicating that on a four-point ranking of responses (with 1 being low),children tended either to not understand or to understand at a high level.Keil sees his results as evidence that children’s metaphor understanding develops on adomain-by-domain basis. He says,44The results accord with recent research demonstrating that metaphor ability does not emerge asgeneral competence at a certain point in development. Children in the kindergarten age groupwere clearly proficient with certain groups of metaphors and not others. It is also evident that thedevelopment of metaphorical understanding is strongly influenced by structured conceptualdomains. Children’s comprehension of metaphors tends to emerge in a unified, all-or-none,manner within each of the conceptual domains considered; and metaphorical abilities amongterms within each domain are more closely related than those across domains. This pattern alsosupports other arguments that metaphors involve interactions between entire domains, not justisolated concepts. (p. 91)Keil’s interpretation that once children understand one metaphor linking two particularconceptual domains, then their likelihood of understanding other metaphors linking the two domainsis increased, is consistent with a connectionist explanation of the organization and processing ofinformation, according to which neural links to and in the proximity of prototypes become stronger withuse.The present study examines the understanding of metaphor and text by grade six students.The decision to use children of this age level was based in part on the literature reviewed here,indicating that while grade six students have achieved a high degree of independence in thedevelopmental aspects just outlined, they have not yet reached full adult ability for handling language.b) Experience and Prior KnowledcieWinner (1988) sees the development of children’s metaphor ability to be closely related to their“knowledge of the world” (pp. 62-63), a view shared by Vosniadou (1987; see, for example, p. 877).As mentioned above, Keil (1986) looked at prior knowledge as a factor in children’s metaphorunderstanding. He concludes that, although it is not possible to tell with certaintywhich of the two [knowledge] domains juxtaposed in a metaphor is more likely to contribute tothe younger children’s difficulties in comprehension.. much of the development of metaphorability is a consequence of broader developments in the conceptual domains that arejuxtaposed. (p. 94)Keil’s discussion of children’s prior knowledge in terms of domain knowledge, rather than singleconcept knowledge, is important in that it emphasizes the interrelatedness of knowledge. Keilconcludes that there is evidence “that the development of metaphorical understanding is stronglyinfluenced by structured conceptual domains” (p. 91), which he took as “support for the domain45interaction view of metaphor” (p. 94).Winner, Levy, Kaplan, and Rosenblatt (1988) provide support for Keil’s conclusion. Theysuggest that when metaphor is not understood, “the problem for children is not that they lack theability to discover the similitude between the two halves of the comparison, but rather that they oftenlack full enough knowledge of one of the halves of the comparison” (p. 58). They believe, with Keil,that “understanding metaphor is primarily a logical-analytical task” (p. 54), but they introduce a newaspect when they describe this task as involving a “match between two divergent aspects ofexperience” (p. 54). If the ground of metaphor can indeed be experience, in a broad sense, then thedefinition of knowledge must be equally broad and metaphor is seen clearly to be much more than amere linguistic device.Some studies have, however, focused on word-level knowledge. Seidenberg and Bernstein(1988) look specifically at word meanings. They report that for good and poor readers encounteringmetaphor understanding tasks, “there are significant group differences in performance. ..even whenthe metaphoric statements are context embedded,” findings that led the authors to the furtherconclusion that poor readers “have less sufficient knowledge of similar word meanings and multiplemeanings of words, and they pay less attention to the semantic content of sentences than high abilityreaders” (p. 43). For poor readers, knowledge of word meanings, as determined by the ability to namethe critical attribute linking the parts of the metaphor, is not enough in itself to ensure understandingof the metaphor. The authors note that their “findings suggest that abilities other than wordknowledge are implicated in metaphor comprehension. It appears that some poor readers may havethe knowledge but lack the ability to match the critical attributes even when the metaphor is contextembedded” (p. 44).Ortony (1980), referring to his earlier work (Ortony, 1975), suggests that “the comprehension ofmetaphor involves predicating of the topic [target] those salient aspects of the vehicle [source] thatare not incompatible with it’ (p. 359). Although Ortony describes the process in terms of the schematheory of knowledge and the activation of subschemata during metaphor understanding, his ideas46have often been applied more to semantic features of the target and source than to generalknowledge about their referents. From this has developed a salience imbalance model. This model,focusing on the comparison of semantic features, has shaped a considerable number of studies.Working from this perspective, for example, Evans and Gamble (1988> conclude that the “ground andthe individual’s knowledge about the topic and vehicle which forms it, and ability to access andintegrate the relevant knowledge, appear to be key factors” (p. 446) affecting metaphor interpretation.Similarly, Mate and Malicky (1988) looked at children’s existing knowledge by examining the semanticfeatures identified for the two parts of metaphors and establishing their salience imbalance, that is, thedegree to which a specific feature is more important to one part of the metaphor than to the other.From their study, Mate and Malicky suggest that “although salience imbalance is a significant feature ofthe metaphor comprehension process, subjects used a variety of strategies to obtain meaning formetaphorical statements” (p. 113).Extending the model, Kincade (1991) refers not only to “word knowledge of the topic andvehicle terms” but also to “familiarity with the subject-matter of the text” as factors in children’scomprehension of metaphors (p. 85). Winner (1988) adds the elements of pragmatic knowledge andknowledge about text function. She takes the position that the metaphor-understanding processinvolves the understanding of “both halves of [a] comparison” (p. 53), according to a three-step modelin which the reader must first detect nonliteral intent, then detect the relation between what has beensaid and what is meant, and finally detect unstated meaning (p. 54). To Broderick (1992), on the otherhand, prior knowledge includes conventionality (the commonness of the phrase). He concludes that“conventionality is a perplexing complication in theories of comprehension and continues to posechallenges to researchers constructing stimulus lists” (p. 187).Conventionality is a topic of central importance to Lakoff and Johnson (1980), Lakoff and Turner(1989), and Lakoff (1993). Lakoff and Turner observe, “Conventional metaphor...depends onconventional knowledge. In order to understand a target domain in terms of a source domain, onemust have appropriate knowledge of the source domain” (p. 60). They state, “In actuality we feel that47no metaphor can ever be comprehended or even adequately represented independently of itsexperiential basis” (p. 19). In their example LIFE IS A JOURNEY, they point out that the metaphoricalmeaning derived for life depends on the knowledge held about journeys. To understand themetaphor “is to have in mind, consciously or more likely unconsciously, a correspondence between atraveler and a person living the life, the road traveled and the ‘course’ of a lifetime, a starting point andthe time of birth, and so on” (p. 61). This process of bringing to bear conventional prior knowledge ofjourneys produces a “rich and varied” conceptualization of life (p. 61), not a simple one-to-onecorrespondence between semantic features of the source and target. The metaphorical mappingconsists of four types of knowledge: Slots, relations, properties, and specific knowledge in thesource domain are all available for mapping onto the target domain (pp. 63-64).The conventional knowledge--or domains of experience--is organized as “gestalts. ..naturalkinds of experience. ..[that] are a product of [o]ur bodies...[o]ur interactions with our physicalenvironment. ..[o]ur interactions with other people within our culture.. .products of human nature” (pp.117-118). In other words, our conceptual system is “a product of the kind of beings we are and theway we interact with our physical and cultural environments” (p. 119). The knowledge resulting fromthe interactions may be personal and private--the result of individual experience--or public andcommonplace--the result of knowledge sharing (Lakoff & Turner, p. 84). In the metaphor process,entire gestalts are activated, with each gestalt providing “a background for understanding thesentence in terms that make sense to us, that is, in terms of an experiential category of our culture”(p.168).Lakoff and Johnson maintain that metaphors “influence our experience and our actions” (p. 68).“Reason,” they state, “involves categorization, entailment, and inference. Imagination.. .involvesseeing one kind of thing in terms of another kind of thing,” and metaphor brings these together as“imaciinative rationality” (p. 193). Accordingly, we can understand our experience directly when weconceive of it as structured in coherent fashion by the gestalts that have emerged from our naturalinteractions, whereas experience is understood metaphorically “when we use a gestalt from one48domain of experience to structure experience in another domain” (p. 230).In the present study, students’ experience and background knowledge, be it of metaphors, oflanguage in general, or of the world, are understood to contribute in significant ways to theconstruction of meaning.c) CulturePritchard (1990) examined the reading strategies of grade eleven students in a cross-culturalstudy with culturally familiar and unfamiliar text. He reports that “the cultural background of the readerand the cultural perspective of the text” are two of the most important factors affecting the readingprocess and the understanding the reader constructs (p. 291). While Pritchard’s study did not look atmetaphor per Se, his conclusion concerning the importance of culture in the reading process mayrelate to the idea expressed by others (e.g., Lakoff & Johnston, 1980; Lakoff & Turner, 1989; Lakoff,1993) that the particularities of a culture have an effect on the metaphors that are created and sharedby members of the cultural group--a situation that would account in part for Pritchard’s results.In their discussion of the origin of orientational metaphors (those that depend on spatialorientation, such as “I’m feeling IJR” to indicate happiness [p. 58]) and ontological metaphors (such as“He did it in ten minutes” [p. 59], suggesting that time is a container), Lakoff and Johnson (1980)remark,Our physical and cultural experience provides many possible bases for spatialization metaphors.Which ones are chosen, and which ones are major, may vary from culture to culture. It is hard todistinguish the physical from the cultural basis of a metaphor, since the choice of the physicalbasis from among many possible ones has to do with cultural coherence. (p. 19)Similarly, structural metaphors, the third major group of metaphors, give rise to a large number ofmetaphorical expressions that reveal a mapping of the organizational structure of the source domainonto the target domain--the structural metaphor ARGUMENT IS WAR, for example, gives rise toexpressions that speak of verbally attacking, defending, counterattacking positions (p. 62).Lakoff and Johnson contend that “[c]ultural assumptions, values, and attitudes” (p. 57)determine how it is possible to conceptualize experience and that the conceptualization itself49depends on the metaphorical process in that the nonphysical is conceptualized in terms of thephysical, “that is, we conceptualize the less clearly delineated in terms of the more clearly delineated”(p. 59).Lakoff and Johnson maintain that ways of conceptualizing experience vary from culture toculture, partly because of the physical environment in which a culture develops and functions andpartly because of the way social relationships are defined. They state, “Since much of our social realityis understood in metaphorical terms, and since our conception of the physical world is partlymetaphorical, metaphor plays a very significant role in determining what is real for us” (p. 146). Thisposition is strongly supported by Lakoff and Turner (1989), who see our cognitive models beingacquired through direct experience and through culture (p. 66), and by Lakoff (1993), who says,“Metaphorical mappings vary in universality; some seem to be universal, others are widespread, andsome seem to be culture specific” (p. 245).What emerges, then, is a complex interaction of metaphor, experience, and conceptualization,with each shaping and being shaped by the others. This complex interaction may account for thecross-cultural differences Pritchard identifies in the reading behaviours of students. The potential forcultural influence on children’s understanding of metaphor was taken into account in the presentstudy by drawing participants from an identifiable, seemingly stable cultural subgroup.iii. Text Focusa) ContextWinner (1988) notes that studies that take context into account reach different conclusionsfrom those that do not. Metaphors presented in context, she says, “reveal the kinds of similarities thatchildren recognize” (p. 44). She cautions, however, that paraphrases out of context cannot be used“to conclude that children cannot generate relational interpretations of metaphors when called for” (p.44).In his review of the literature, Ortony (1980> identifies a “context-length by context-type50interaction’ (p. 359) in metaphor experiments done by Ortony, Schallert, Reynolds, and Antos(1978). He suggests that the context may cue the reader that an utterance is anomalous and requiresa nonliteral interpretation (p. 352), thus initiating a meaning search involving a “context-sensitivespreading activation mechanism” that begins with the main input schemata, then, under the influenceof the concepts active in the context at the time, activates related subschemata until a satisfactoryintersection of the schemata is arrived at, thus providing a set of features upon which to base ametaphorical comparison (pp. 360-361).Investigating the effects of context on children’s metaphor comprehension, Siltanen (1989)finds an age-context relationship. She says “that younger children.. .need different amounts ofcontext for easy, moderately difficult, and difficult metaphors than...older children” (p. 211). Supportfor Siltanen’s conclusion is provided by Chandler (1991), who maintains that “all metaphor usageseems to be highly sensitive to context effects for both linguistic and nonlinguistic contexts” (p. 237).From three experiments on children’s understanding of metaphor, Vosniadou, Ortony,Reynolds, and Wilson (1984) conclude that children must confront several sources of difficulty, one ofwhich is “the context in which the metaphorical language occurs” (p. 1595). They suggest “thatchildren use the linguistic and situational context to draw inferences” about the meaning of the text(p. 1603) and, where the context is insufficient and inferences turn out to be unsupportable, thatreaders must engage in inference revision, thus complicating the meaning-making process (p. 1603).Vosniadou (1987) concurs. She says that “it may be the case that metaphor comprehension isoriginally achieved only in situations where the already established context strongly leads toinferences that are inconsistent with a literal interpretation and consistent with the metaphor’s impliedmeaning” (p. 878). Kincade (1991) also agrees, identifying “adequate context support” as one of theconditions necessary for grade two and grade five children to understand abstract comparisons of thetype underlying certain metaphors (p. 94).Examining college students’ comprehension of metaphors, lnhoff, Lima, and Carroll (1984)found that metaphors were understood as fast as literal statements when contexts were long, but51more slowly when contexts were short (p. 563). From this, they conclude, “Conceptually supportivecontext is an important factor in the comprehension of metaphoric language” (p. 563). Also workingwith college students, Reynolds and Schwartz (1983) examined understanding ofcontext-dependent metaphors, those that depend on surrounding information to determine howthey will be understood (p. 451). They report “increased memorability for passages when theconcluding statement is expressed metaphorically rather than literally” (p. 455) and “greater recall ofthe metaphors than the literally equivalent statements” (p. 455), suggesting a mutually reinforcinginteraction between metaphor and context.Vosniadou (1987) claims that inaccurate results have been achieved in some studies because“the metaphorical sentences were often presented to children in the absence of a meaningfullinguistic or situational context” (p. 877). Related to this, Siltanen (1989) warns that “researcherscannot ignore metaphor difficulty as a critical factor in children’s metaphor comprehension” (p. 208).And, a basic question for the researcher, Ortony (1980) says, is “whether, given every opportunity, achild can show ability to comprehend metaphors” (p. 357).The present study attempts to address concerns for adequacy of context by using passages ofunedited, authentic text.b) Explicitness and SimilaritiesWhen a metaphorical statement is phrased in the simile form, that is, using ffl (or an equivalentphrase) to force the comparison, the resulting explicitness is considered by some researchers tomake the detection of similarities between the source and target easier. Placing her comments withinthe context of her three-step, pragmatic comprehension model (see Methodology section, above),Winner (1988) suggests that the presentation of metaphorical comparisons in simile form changes theunderstanding task for children, making it easier by reducing the reader’s processing requirements byat least one step--that is, by avoiding the need for the listener to first detect nonliteral intent (see, forexample, pp. 10-11 & 49).52Siltanen (1990) investigated relations among three factors: age, metaphor difficulty, andexplicitness. Her results are inconclusive and leave open the question of the degree to whichexplicitness contributes to metaphor understanding. She suggests that her results can be accountedfor by the children at a certain age making no distinction between the simile and metaphor forms of anexpression or focusing on “content words (nouns and verbs)” rather “than on function words(conjunctions and determiners)” (p. 17).Vosniadou, Ortony, Reynolds, and Wilson (1984) also investigated the effects of explicitness,with results somewhat different from those achieved by Siltanen. They report that “the insertion of‘like’ clearly helped the children enact the meaning of the metaphorical sentences” (p. 1600). Theysuggest two possible reasons for this: first, that the word jJ focuses attention on “perceptualsimilarity between the objects compared, or on the similarity of the actions in which the agentsengage,” and, second, ffl. focuses attention on “relational similarity,” that is, things behaving insimilar ways (p. 1600). The ability to enact does not mean, however, that children can necessarily giveverbal paraphrases of the metaphorical sentences, as required by Siltanen; this may account in part forthe differences in the results of the two studies.The question of the effect of explicitness, with the attendant questions of similarity of sourceand target and saliency of semantic features, arises from the comparison view of metaphor. While thisis not central to conceptualizing metaphor understanding according to the mapping view elaboratedby Lakoff and Johnson (1980), Lakoff and Turner (1989), and Lakoff (1993), it need not be excluded,since clearly some metaphors do involve comparison of existing features of the two parts of themetaphor. For this reason, although the question of the effect of explicitness on metaphorunderstanding is largely unanswered by the studies undertaken so far, it seems appropriate that thequestion should remain open.In the present study, no attempt is made to balance metaphors and similes. All metaphors arethose occurring naturally in the selected text passages.53c) EcoloQical ValidityAs has been pointed out throughout this review of the literature, the examination of children’smetaphor understanding has proceeded according to a variety of experimental paradigms and withthe use of a variety of metaphorical material. Broderick (1992) compares the metaphors occurring in53 popular children’s books and those used as stimuli in 15 major metaphor studies with children. Hesays, “In the study of children’s comprehension of figurative language to date. stimulus sets havebeen constructed largely without reference to what children are actually exposed to in everyday life”(p. 183). He notes that metaphoric equation (i.e., the use of a copula verb and noun phrases--e.g.,“that fellowa bad apple” [p.184]) and metaphoric substitution (i.e., “referring to an old referent ina new way--e.g., “their father, the wretch” [p. 184]) have been favoured in research but “are amongthe rarest forms in children’s literature” (p. 191).Broderick provides taxonomies for examining the ecological validity of stimulus sets ofmetaphorical expressions used in research. These are based on both syntactic frame (describing theform of the comparison) and similarity type (describing similarity relationships between the source andtarget) (pp. 184-187). As is readily evident, Broderick’s review focuses on studies based on thecomparison view of metaphor, and his taxonomies reflect his own attachment to the view, particularlyin his identification of the major taxonomies based on syntactic frames and comparison types. Thatnotwithstanding, his taxonomies provide a useful tool for analyzing aspects of the task faced bychildren when they make meaning of metaphors. Additionally, Broderick has made a significantcontribution to the field by drawing attention to the need for ecological validity in metaphor studies.The present study uses Broderick’s taxonomies to describe the metaphors occurring in thetarget passages and as one basis for comparing student understanding of the metaphors. This studyalso considers the ecological validity of the metaphors by using passages excerpted from authentictexts of the sort students are likely to encounter in their classrooms.54d) Metaphor in Informational TextMost studies of metaphor have used specially prepared sets of metaphors, specially preparedsets of sentences with metaphors embedded, or specially prepared stories. Very few studies haveused informational text, and of those that have, none could be found which used authentic text drawnfrom trade books. Consequently, very little is known about children’s understanding of metaphor ininformational text.In three studies, Pearson, Raphael, TePaske, and Hyser (1981) examined children’s recall ofmetaphor in expository text and the way metaphor affected recall of textual information. In the firststudy, two passages were adapted from an informational text, one as the literal form and the other asthe metaphorical form. Grade six and university students were tested for recall. The grade sixstudents recalled more of the metaphor structures than of the corresponding literal structures. In thesecond study, high- and low-reading ability grade three students performed the same type of testprocedure, but with inconclusive results. A followup investigation suggested that passage familiaritymight be a factor, that is, that the less familiar the passage, the more likely a metaphor was to bememorable within the passage. The third study, this time involving high- and low-reading ability gradethree and grade six students, produced ambiguous results, although there was evidence thatmetaphors were more memorable in the unfamiliar text.To examine the relationship between topic familiarity and metaphor understanding, Mercer(1985) gave grade seven students either a metaphor or literal text version of specially preparedfamiliar- and unfamiliar-topic expository passages. On oral free recall of target ideas, students recalledsignificantly more information conveyed by the metaphor targets of the unfamiliar topic text than of thefamiliar topic text. Also, interpretation of metaphor targets was significantly greater for the unfamiliartopic text (pp. 78-79). As Mercer acknowledges, the target metaphors were entirely of the similaritytype (comparison view), that is, based on some preexisting similarity between the source and target.Because other types of metaphors were not included, she cautions that generalizations may beinappropriate. Her text passages, although specially prepared for the study, were close to being55authentic informational text in that they were adaptations of trade books available at the time. Themetaphor target density was eight per passage, or one per 50 words or less--a density that may behigher than in most non-experimental texts, Each metaphor version had one metaphor target that wasmore properly a literal comparison (i.e., “weigh as much as a ten year old child” and “weigh as much asa large refrigerator”). It is unclear the extent to which this would alter the findings. It does, however,underscore the importance of a scheme, such as that offered by Broderick’s (1992) taxonomies ofliteral and metaphoric comparisons for establishing the metaphoricity of expressions.From a study of the effect of analogy (which makes use of the metaphorical process in anextended way--see, for example, Lakoff, 1993, p. 235) on children’s understanding of expositorytext, Vosniadou and Schommer (1988) report that explicit analogies assist five- and seven-year-oldchildren to “build representations of new concepts that are richer than the ones obtained frominformation contained in expository texts without analogies” (p. 535). The effect is age-related, whichis taken to indicate that expanding “world knowledge” permits the older children to “build theconstraints needed for more selective transfer” (pp. 534-535).Drawing on earlier work on advance organizers, Mayer and Bromage (1980) examined theeffects of concrete analogies on the recall of information. Participants were university students whoscored either high or low on the Scholastic Aptitude Test--Mathematics. None had had priorexperience with computers or computer programming. The target texts were simplified versions ofcertain commands in the BASIC computer language, one presented in a text-only format and theother in a text plus organizing-model format. Participants received both texts, but in different order.Subsequent tests indicated that the concrete analogies aided recall when presented before readingtook place. The effect was stronger for low-ability students. A second study looked at the effect oftest delay on recall, finding that the before group had a higher rate of recall than the after group. Oneconclusion reached by Mayer and Bromage is that analogical models should be more useful forunfamiliar information (p. 224), a conclusion consistent with Pearson, Raphael, TePaske, and Hyser’s(1981) study reported above.56Reynolds and Schwartz (1983) carried out a study comparing the effects of literal andmetaphorical concluding summarizing statements on passage recall. Results indicate that ‘undercertain circumstances and constraints, metaphors, like adjunct questions or advance organizers, canenhance learning from written text” (p. 455). Reynolds and Schwartz suggest that metaphors may actas “summary statements” that “serve to clarity some details of the text or to draw an analogy betweenan unknown event or procedure and a more familiar idea” (p. 458). The eight text passages createdfor the study were “short stories” of about 50 words in length (p. 452). Examination of the samplepassage provided by the authors shows it to have more in common with expository text than withfiction in that there is no structure typical of stories. On this basis, Reynolds and Schwartz’s commentscan be taken to apply to expository text and to relate to other studies reported in this section, at leasttentatively.The study reported here uses authentic informational text passages, each being the initialportion of a trade book readilyavailable for use in school. By choosing such material, it is hoped thatthe issue of ecological validity will be addressed to some degree and that findings, as preliminary asthey may be, will tell something about children’s reading of naturally occurring informational texts innaturally occurring situations.VIII. SummaryIn this chapter, five theories were examined to suggest ways in which they interlock to form asimple, coherent worldview capable of supplying the warrants upon which research into metaphor canrest. The following contentions were put forward and evidence offered in their support:1. That social interactionism sets general conditions under which knowledge construction cantake place--that is, when knowledge can be constructed;2. That constructivism describes a specific manner in which the individual can constructknowledge from experience--that is, J2 knowledge can be constructed from experience;3. That schema theory describes a model according to which knowledge can be organized--that57is, how knowledge can be stored;4. That parallel distributed processing describes how knowledge can be processed--that is,how knowledge can be systematized; and5. That views of metaphor describe how knowledge, once constructed, can feed itself--that is,how knowledge can become synergistic.In this chapter four major views of metaphor were examined:1. The substitution view, according to which metaphor is held to be a linguistic device by meansof which one word is substituted for another and according to which any metaphor is necessarilylogically false;2. The comparison view, according to which affention is drawn to dominant features of one partof the metaphor in such a way that the other part is endowed with the features;3. The interaction view, according to which the two parts of the metaphor are seen to influenceeach other; and4. The mapping view, according to which the meaning and/or structure of one domain ofexperience maps onto another, reshaping it.Metaphor research indicates that children, particularly the young, may comprehend metaphorbut be unable to explain its meaning. Research indicates that literal language is not necessarilyacquired by children prior to metaphorical language. Research into how the brain processesmetaphor is inconclusive. Recent enquiry into parallel distributed processing (PDP), however,suggests that broad segments of the neural network are engaged during concept formation,refinement, and activation and during comparison and/or mapping from one conceptual domain toanother.To avoid limitations and restrictions arising from a narrow conceptualization ofmetaphor--limitations and restrictions such as those imposed by the comparison view, for58example--this study defines metaphor broadly as a mapping process, as described in this chapter.Children’s background knowledge is of interest in this study insofar as it influences understanding ofthe metaphors, where that can be determined. Ensuring adequate context, use of methods otherthan paraphrase, use of authentic metaphorical passages, and avoidance of the effects of storystructure are all taken into account. These things are accomplished, respectively, by providinglengthy passages, by having students complete multiple choice activities, by providing authentictexts, and by using informational text passages. Complexity of linguistic input is of interest as a factorinfluencing metaphor and passage understanding; to this end the three passages used are ofdifferent degrees of reading difficulty.The present study examines the understanding of metaphor and text by grade six students,who, as a group, are generally acknowledged as having achieved a fairly high degree ofindependence in the developmental aspects of reading and metaphor understanding, although notthe full adult range and expertise.In the present study, students’ experience and background knowledge, be it of metaphors, oflanguage in general, or of the world, are considered to contribute in significant ways to theconstruction of meaning. No attempt is made to control for explicitness, according to the comparisonview. All metaphors are those occurring naturally in the selected text passages. Lastly, this studyconsiders the ecological validity of the metaphors by using passages excerpted from authentic textsof the sort students are likely to encounter in their classrooms.This study attempts to provide tentative suggestions about children’s understanding ofmetaphor in authentic informational text, an area that has not been examined. Further, it undertakesto use the ideas of Lakoff, Johnson, and Turner to inform a study to a degree not hitherto attempted,insofar as can be determined from the literature.59CHAPTER THREE: RESEARCH DIALECTIC AND METHODOLOGYI. Introduction 59II. Research Proposal 67Ill. Pilot Study 71i. Research on Think Aloud 72ii. Materials 73iii. Procedures 73IV. Main Study 78i. Research Approach 78ii. Research Design and Instrumentation 79iii. Selection of Participants 80iv. Site Procedures 82v. Data Collection and Recording 83vi. Methodological Assumptions 86vii. Limitations 87V. Summary 88I. IntroductionThis study arose from my long-standing interest in metaphor. As a student of English literature,studying and reading widely across genres and literary movements, I developed a strong sense of thepower of metaphor to capture the essence of the idea being represented and to raise it to a higherlevel of significance than possible with literal language alone, thus making the idea both moremeaningful and more memorable--something to be savoured, often to be shared with others, to becarried with me and used in my own life.As a poet, I had--and have--a very personal interest in how metaphor is constructed, how itfunctions on what might be considered the technical level, and how and why readers and listeners60respond to it affectively. In my own poetry, I found myself giving particular attention to the images Icreated, often through metaphor, and came to think of the most striking of these as power centresaround which meaning was organized and which, one might say, provided the gravity that gave thepoem weight.As a teacher, I became interested in the challenges and opportunities that metaphor presentedto my students. Although I was aware of how powerful metaphor could be to heighten meaning forsome of my students, to surprise, often to delight, I was also aware that not all students couldappreciate the power of metaphor or make it their own. For these students, metaphor was achallenge, not an opportunity.At the academic level, my first attempt to delineate the problem to my own satisfaction was, as itturned out, quite unsatisfactory. Although I knew there was a problem, I was not clear whether theproblem was one of the process of metaphor production, one of pure linguistic comprehension, oneof reading comprehension, or a problem of knowledge of metaphor. To compound problem, theresearchers were not clear, either, as to the origin of the children’s problems; indeed, researchersoften did not even agree on just what a metaphor was! The advantage was that there was muchinterest in metaphor at the experimental and philosophical levels of enquiry. The disadvantage wasthat one new to the field had to sort through diverse, persuasive theories and often contradictoryresearch findings, trying to make significant judgments with less than finely honed critical tools.I read widely, hoping to find a convincing philosophical perspective, a convincing explanation ofthe psychological processes of metaphor making and comprehension, and a convincing explanationof how and when these psychological processes developed in children.One of the books I read was George Lakoff and Mark Turner’s (1989) More Than Cool Reason:A Field Guide to Poetic Metaphor. Their ideas were strongly appealing, perhaps because theyharmonized with my own, hitherto poorly conceptualized and poorly articulated, ideas--ideas that had,nonetheless, been forming for at least twenty years. Lakoff and Turner confirmed for me thatmetaphor was much more than a simple linguistic embellishment. Metaphor was, as I had sensed, a61cognitive device. It was, they maintained, fundamental to our way of thinking, as well as to our way oftalking. It was ubiquitous. It was capable of creating new meaning, not just passing along existingmeaning. It had a cultural basis. It was, in a sense, a form of cultural shorthand.Ernst von Glasersfeld’s (1984) “An Introduction to Radical Constructivism” was also stronglyappealing. Again, it accorded with elements of my own thinking, elements expressed in somemeasure in my poetry. While many researchers were working from the constructivist perspective, fewadopted the extreme position put forward by von Glasersfeld. For him, constructivism did not simplydescribe the manipulation of information, although that was important; rather, it was a theory of howknowledge came to be and how the individual made meaning of the flow of imposing sensation. It wasa theory of being.Under the influence of socio-cultural theorist and researcher Gordon Wells, I came to L. S.Vygotsky’s (1962, 1978) socio-cultural theory of language. At first, it seemed to conflict withconstructivism. However, in time I discovered a workable balance between the internalness ofconstructivism and the externalness of social interactionism. I found that the two theories providedpowerful double lenses through which to examine the issues laid out so boldly by Lakoff and Johnson(1980), Lakoff and Turner (1989), and Lakoff (1993).I found that Jerome Bruner had adopted a constructivist-social interactionist stance in ActualMinds, Possible Worlds (1986) and Acts of Meaning (1990). For Bruner’s purposes, the extremestance of von Glasersfeld was unnecessary, and he maintained at least a tacit acknowledgement ofobjective reality. One of the powers of Bruner’s two books is the manner in which he makesaccessible Vygotsky’s ideas, especially the zone of proximal development and the social constructionof individual identity.My reading continued during all phases of my studies. I discovered articles and books thatexpanded my thinking. The book by Marcel Danesi (1993), Vico, Metaphor and the Origin ofLanguage, stands out, as does George Lakoff and Mark Johnson’s Metaphors We Live By (1980). Asecond article by Ernst von Glasersfeld (1989), “Facts and the Self From a Constructivist Point of62View,” gave texture to the striking outline of radical constructivism sketched in the 1984 bookchapter, “An Introduction to Radical Constructivism.” Indeed, I found the entire book, The InventedReality (1984>, edited by Paul Watzlawick, a fascinating exploration of the application of constructivistphilosophy across many academic disciplines and social contexts.Perhaps because of my background, I thought of the question of metaphor as a question ofpoetry and narrative. This stance was encouraged by the research literature. With this orientation, Iculled nine samples of metaphor from poetry and fiction in four grades two and three language artsanthologies and asked six grade three students to tell me what the text samples meant.I analyzed the metaphors to determine what critical knowledge (semantic features) would berequired to transfer from source to target to constitute understanding. I used this critical knowledge inmy analysis of the think-aloud protocols.On the premise that extreme cases might provide contrasting pictures of prior knowledge andmetacognitive processes, I identified five samples, two that had been understood by all students, onethat had been understood by five before probing and all six after probing, and two that had not beenunderstood by any of the students. In the case of the understood samples, two features wererevealing--that is, that the children had the prior knowledge of a cultural expression and that themetaphorical aspect could be ignored while still understanding the meaning of the text sample. Oneexample, “pitch black,” illustrates both features: It is a common cultural expression that the studentsreported hearing previously, and they understood it to mean “very, very dark” or “dark black” withoutknowing the meaning of the word pj itself.The two samples not understood by any of the students revealed three commoncharacteristics: a limited context, the presence of one extremely abstract term, and students’ lack ofprior knowledge.Again looking at extreme cases, I compared the think-aloud characteristics of the two studentswho understood the most text samples (6 of 9) with those of the two students who understood thefewest (3 of 9 and 2 of 9). The protocols of the first pair had a greater variety and complexity of63features considered indicative of understanding of the text (8), compared to the protocols of thesecond pair (4 and 3, respectively).From these analyses, I discovered that the students understood the text samples (notnecessarily the metaphors contained in the samples) without probing about 46% of the time. Idiscovered also that these children might well have the prior information necessary to compare thetwo parts of the metaphor (using the feature-comparison model), yet be unable, even under probing,to make the semantic connections that would constitute paraphrase of the metaphor (Ortony, 1980;Evans & Gamble, 1988), a condition noted among poor readers by Seidenberg and Bernstein (1988).This was entirely consistent with the ideas that a certain level of metacognitive capacity is neededwhen the meaning of text is not immediately clear and, further, that metacognitive capacity issomething unlikely to be fully developed by grade three. It was also consistent with the idea ofconventionality of metaphors--that some of them are learned because they are part of the ordinary,everyday way of talking about things (Lakoff & Johnson, 1980). I was forced to consider the questionof what constituted sufficient context for presentation of metaphors (Inhoff, Lima, & Carroll,1984;Vosniadou, Ortony, Reynolds, & Wilson, 1984; Winner, 1988).Methodologically, I discovered that the think aloud could be a useful device for getting at thesort of information that interested me but that it required some pre-teaching and familiarization, sincetalking in the presence of others usually entails talking the other, rather than to oneself. On a verysobering, practical side, I discovered that analysis of think-aloud protocols could be time-consumingand difficult.I felt it would be helpful to have a clearer idea of how proficient adult readers read text containingmetaphors. To this end, I selected a piece of fiction from a grade six language arts anthology,identifying in it six metaphor samples. I intended that the potential problem of insufficient contextwould be avoided by using the single text, rather than portions of several. I asked five students from adoctoral seminar, working individually, to read the passage aloud as if reading to a friend--a procedurereported by Ericsson and Simon (1993, p. xviii) to improve subject performance when compared to64control group subjects--and to stop at indicated points to explain the underlined portions of text, eachcontaining a metaphor. I audio-taped the think alouds, transcribed, and analyzed them.Expanding on the idea of critical knowledge used with the text samples for the grade threechildren, I constructed schematic webs (semantic-feature or salient-attribute charts) for the sixmetaphors. For example, for the metaphorical expression, ‘the herd was galloping away like thewind,” I identified the following shared features for “herd” and “wind”: movement, speed, force,uncontrollable. Reference to any of these features by the participants was taken to indicate the sort oflinkage underlying metaphor comprehension, according to the attribute-saliency form of thecomparison theory (see, for example, Ortony, 1980).The adult participants made numerous semantic connections. For the metaphorical expression(simile), “they swept like a brown flood” (referring to a herd of horses), participants mentioned six ofthe seven features I had identified as being major conceptual links, providing a total of elevenmentions.Results for four of the six samples showed one feature to be favoured above the others. In theexpression, “the herd was galloping away like the wind,” for example, speed was mentioned by allparticipants. This could be taken to suggest that the expression itself provided the clues necessary tomake this connection; on the other hand, it may be that a form of the expression was sufficientlyfamiliar to cause the participants to judge that speed is the important feature of the situation describedby the expression.Understanding of the metaphors was, as expected, 100%. There was little evidence ofmetacognitive activity, but this could be accounted for by the ease with which the adults read thetext--that is, the metacognitive processes were so automatic as to be invisible to the reader and so didnot occasion comment of the sort that could be taken as evidence of their presence and function.Context seemed sufficient for ease of reading and there was no evidence of focusing on key words tothe exclusion of other portions of text, as had occurred with the children when they created meaningaround puzzling portions of text. A major difference between the children and the adults was the65amount of unprompted elaboration provided in addition to the basic responses necessary to indicateunderstanding of the target samples, with the adults producing a much higher percentage.These two activities suggested that the think aloud could reveal considerable, potentiallysignificant information about the level of text understanding and the reading and thought processesengaged in by the reader. They also suggested that grade three students’ metaphor understandingmight be affected by their limited metacognitive abilities. The literature suggested that young childrenmight also lack the power to verbalize metaphor meaning, even when a meaning had beenconstructed from the reading (Vosniadou, Ortony, Reynolds, & Wilson, 1984).On the other hand, based on experience and my reading of the research literature, I felt thatgrade six students would form a more appropriate research group; specifically, they should possesshigher metacognitive and verbal abilities than grade three students, yet they should not be able toread with full adult proficiency, making it more natural for them “to report on processes which have[not] become ‘automatized” (Afflerbach & Johnston,1984, p314). This position is also supported byBrown (1980).Other factors imposed themselves at this point. The literature shows that children’s familiaritywith story structure (Bruner, 1990, p. 80, for example, says that humans have “an ‘innate’ and primitivepredisposition to narrative organization”) can influence memorability of metaphor, depending onwhere in the text the metaphor occurs (Waggoner, Meese, & Palermo, 1985). Other studies showthat vividness of metaphor can influence memorability of text for some readers (Pearson, Raphael,TePaske, & Hyser, 1981; Reynolds & Schwartz, 1983). As well, there seemed to be a certaindichotomy in asking children to read a story, something for which they would usually assume anaesthetic stance (Rosenblatt, 1978, 1991), then having them give paraphrases of portions of the text,treating it in an efferent manner.I found that the literature had very little to say about the problems and opportunities presentedto young readers by metaphor in informational text. It seemed that the topic should be ofconsiderable importance, especially since informational trade books are available in large numbers for66use in support of the curriculum. Once I decided that metaphor in informational text should be thefocus of the study, two of the potentially confounding factors mentioned above were eliminated--thatis, effects of story structure and effects of changing reader stance. The third factor (effect ofmetaphor on memory) could be controlled by making the text passage available throughout the entireactivity.As well, to address the potential problem that some children might be unable to verbalizemeaning, while actually understanding the metaphors, I decided to use a multiple choice activityfollowing the think aloud, a method Vosniadou (1987) identifies as a stronger method than requiringparaphrase. To increase the richness of information provided by the think alouds, I decided to ask thechildren to do a retell after completing the initial reading. This would be an extension of the thinkaloud and would require the students to give what they considered to be the main points, so as toreveal more of the meaning they had constructed during their reading.A factor that seems to influence understanding of metaphor is intimacy with culture. One of thegrade three children had English as her second language. She experienced difficulty with one of themore idiomatic expressions, although probing showed she had the requisite knowledge to make asemantic-feature comparison adequate to explain the two parts of the metaphor (see Ortony, 1980,and Winner, 1988). One of the adults also had English as her second language. After the thinkaloud, she commented that some of the expressions were difficult because she was not a nativeEnglish speaker and had not internalized all the expressions from childhood.These observations are consistent with the contentions of Lakoff and Johnson (1980), Lakoffand Turner (1989), and Lakoft (1993) that metaphor is strongly influenced by culture (see above,Chapter Two). On this basis, I decided that the meaning I would expect for any particular metaphor inthe study would be the anticipated meaning an adult, Canadian English speaker would be likely tomake.According to Vygotsky’s idea of the zone of proximal development, students should perform ata higher level when working in collaboration with more capable peers’ (1978, p. 86) [emphasis in67original]. Wells (1 990b) expresses this idea in the form of an apprenticeship model, whereby“collaborative talk about texts of varying kinds in the context of meaningful joint activities, undertakenwith the assistance of a more skilled co-participant,” can increase learning (p. 402). Wells (1 990a)suggests that withoutopportunities for this sort of talk about texts, students may learn to perform the more observableaspects of reading and writing and still fail to understand how these behaviors may empoweraction, thinking and feeling in their personal lives as well as in the tasks they are required toperform in the classroom. (p. 12)Moffett and Wagner (1991) carry the idea further. They describe partner reading, the processwhereby students take “turns sight-reading aloud to each other in a group of two to four” (p. 71).Although weaker readers are aided by the process, the benefits extend to enrichment of thecomprehension of all participants and increased awareness of their own response to the text (p. 71).Forman and Cazden (1985), studying the problem-solving interactions of pairs of fourth- andfifth-grade students, report that “by assuming complementary problem-solving roles, peers couldperform tasks together before they could perform them alone” (p. 343) [quoted in Cazden, 1988, p.130].Out of the foregoing, there arose the idea to examine whether two readers working together(i.e., in dyads) would produce more talk about the text than a lone reader. It seemed interesting toask, as well, whether students who read in the dyadic condition would show increased understandingof the metaphors on the planned multiple choice activity.II. Research ProposalI prepared a proposal entitled “Grade Six Students’ Understanding of Metaphor in InformationalText.’ In this, I noted that despite the increasing number of studies of children’s understanding ofmetaphor, informational text had received very little attention. Consequently, the problem remainedthat it was not known how well children understood the metaphorical language in informational text,nor was it known whether children’s level of understanding of metaphor was related to their level ofunderstanding of the text itself. One concern was central: The study should have ecological validity68by drawing material from actual informational books of the sort students were likely to encounter in theclassroom or choose to read on their own.The proposal set out three major questions around which the study was to be organized: (a)How well do grade six students understand metaphor in the informational text they read? (b) How welldo the same children understand the metaphor-bearing informational text itself? (c) How areunderstanding of text and understanding of metaphor related? These three major questions wererestated as seven specific questions. The relationship between the major and specific questions wasas follows:(a) How well do grade six students understand metaphor in the informational text they read?1. What is student level of understanding of metaphor as determined by performance on amultiple choice meaning clarification activity?2. What are the differences in understanding of metaphor by the same student in the individualand dyadic think-aloud conditions as determined by performance on multiple choice meaningclarification activities?3. What are the differences in understanding of metaphor in the same text for the individual anddyadic think-aloud conditions as determined by performance on a multiple choice meaning clarificationactivity?4. What are the differences in understanding of different types of metaphor as determined byperformance on multiple choice meaning clarification activities?(b) How well do the same children understand the metaphor-bearing informational text itself?5. What proportion of students achieve high and low scores on think-aloud tasks as determinedby holistic and specific-trait scoring?(c) How are understanding of text and understanding of metaphor related?6. What are the features of the think alouds of students whose performance on a multiplechoice meaning clarification activity placed them in the first and fourth quartiles?697. What is the relationship between the level of students’ understanding of metaphor and theirunderstanding of text?These questions were subsequently approved by my academic committee.I proposed to use authentic text from informational trade books, rather than either artificiallycontrived metaphor sets or text written for the purpose of the study.I examined the metaphor content of informational text in various language arts series that hadbeen authorized for use in British Columbia schools, copies of which are held by the LanguageEducation Research Centre of the Department of Language Education of The University of BritishColumbia. Specifically, I looked at all the informational text in the following:1. Spin Amonc the Stars, Grade 6, Network series (Nelson, 1991);2. Star Flicihts, Grade 6i, Network series (Nelson, 1984>;3. Handshakinqs, Grade 612, Expressways series (Gage, 1988);4. Lobstick, Grade 613, Expressways series (Gage, 1988);5. Nineteenth Moon, Grade 61, Unicorn series (McGraw-Hill Ryerson, 1985);6. All in Good Time, Grade 62, Unicorn series (McGraw-Hill Ryerson, 1985);7. Ride the Wave, Grade 6, Journeys series (Ginn, 1988);8. Wherever You Are, anthology and project book, Grade 61, Impressions series (Holt, Rinehart& Winston, 1987); and9. All Over the World, anthology and project book, Grade 62, Impressions series (Holt, Rinehart& Winston, 1987).I found that the amount of metaphor varied considerably, not only across series, but within themas well. Noticeable, however, was that informational text from Impressions generally showed a higheroccurrence of metaphor than the informational text of the other series.I examined five recently published informational trade books, randomly selected by ateacher-librarian from among publisher submissions to the National Council of Teachers of English70bibliographic project, Adventuring with Books. The five books were as follows:1. Monarch, by Kathryn Lasky, Harcourt Brace & Co., San Diego, 1993;2. The Search for the Right Whale, by Scott Kraus and Kenneth Mallory, Crown, New York,1993;3. Playful Slider, by Barbara Juster Esbensen, Little, Brown & Co., Toronto, 1993;4. Arctic Summer, by Downs Matthews, Simon & Schuster, New York, 1993; and5. Power Machines, by Ken Robbins, Henry Holt & Co., New York, 1993.The first three books all begin with metaphor-rich language. After the first few pages, thelanguage becomes less metaphoric. The fourth book begins and ends with an extendedpersonification of winter and, apart from this, has only occasional metaphors. The fifth book hasoccasional similes and strong comparisons constructed like similes, but metaphors are rare.In considering these observations, I felt that they raised questions about the appropriateness ofdifferent materials. I wondered, for example, whether students confined to a single language artsseries were being provided with reading experiences to support reading beyond the programmaterial. It seemed that students were increasingly likely to encounter metaphor when they readinformational trade books. Since there is a trend to use more such trade books in school, I felt thismade my central interest even more pertinent--that is, how well metaphorical language is beingunderstood and how its understanding relates to understanding of the text.In the proposal, the key terms were operationally defined as given in Chapter One, with thefollowing three exceptions:Understanding: The meaning created by an individual as the result of an experience, such as,for example, reading. In this study, it is taken to mean both performance on a multiple choice meaningclarification activity and performance as determined by holistic and specific-trait scoring of a thinkaloud.High score: A score, as determined by holistic and specific trait analysis of a think aloud,equivalent to a comprehension score of at least 90%. This is consistent with recommended practice71for determination of level of student comprehension of text using an Informal Reading Inventory andrepresents an independent reading level (May, 1990, p. 384).Low score: A score, as determined by holistic and specific trait analysis of a think aloud,equivalent to a comprehension score of less than 50%. This is consistent with recommended practicefor determination of level of student comprehension of text using an Informal Reading Inventory andrepresents afrustrational reading level (May, 1990, p.385).Ill. Pilot StudyI sought and received approval from the University Behavioural Sciences Screening Committeefor Research Involving Human Subjects to carry out a pilot study. The pilot study was done to testprocedures and materials prior to undertaking the main study and to develop a set of descriptors forthe understanding processes revealed by the students during their reading. It involved six grade sixstudents (three girls and three boys) from two combined grades six/seven classes in an urbanelementary school in the British Columbia lower mainland. The students were among the youngermembers of their classes, yet, because of promotion by age, which is standard in the elementaryschool system, these students were typical of the age of British Columbia grade six students,generally. At my request, students were chosen by their teachers to reflect a range of readingabilities. Corroborative data from school records were not sought, but subsequent performance bythe students supported the teachers’ choices. The students and their parents agreed to theparticipation.The school had a high English-as-a-second-language (ESL) population, although all studentswho participated in the pilot had English as their first language. The school had been operating for anumber of years and appeared to the visitor to provide a stable, yet vigorous educationalenvironment. At the time of the pilot, the school community was involved in intense self-examinationand long-range planning, as part of the school accreditation process. There was a blend of youngerand older teachers, led by an able administrative team of principal and vice-principal.72i. Research on Think AloudMy earlier experience using the think-aloud procedure with grade three students and adultssuggested it was a potentially rich source of information about readers understanding of text. Theliterature supports this. Waern (1980), while stating that “the think aloud comments cannot be seenas a direct reflection of all thought processes going on during reading.. .are incomplete.. .and reflectsome editing” (p. 128), does maintain that they reveal “the content of short-term [memory] store” (p.124).Afflerbach and Johnston (1984) give five advantages of think-aloud type procedures.Specifically, they say, oral reports have “one major advantage” in that “their validity relies on a differentset of assumptions from those of most other methods of investigating cognitive processes,” thusperforming “a valuable role in the collection of converging data sources”; they can “provide veridicialdescriptions of cognitive processes which otherwise could only be investigated indirectly”; they giveaccess to underlying “reasoning processes”; they may be the only available source of informationpermitting the “historical or genetic analysis of mental processes”; and they “allow an analysis of theaffective components of reading processes” (p. 308).Baumann, Seifert-Kessell, and Jones (1992) used the think aloud to examine students’monitoring of their own understanding while reading and found that they could increase the amountof comprehension monitoring students used by teaching the students think-aloud procedures.Baumann, Jones, and Seifert-Kessell (1993) specifically suggest that the think aloud “can be used topromote understanding of informational trade books or content area textbooks” (p. 192). They alsorecommend that “students participate in the social construction of think alouds” (p. 192), a suggestionin line with social-interaction theory. This agrees with my own position and serves to inform researchquestions two, three, five, and six, above.I decided also to have the students end each reading think aloud with a retell of the selection intheir own words. This is consistent with Afflerbach and Johnston (1984), who note that “to the extentthat concurrent and retrospective reports represent different data sources, they may be used as73multiple indicators” (p. 319).ii. MaterialsThe materials used for the pilot study were as follows:1. Two demonstration selections of informational text: ‘The Land of the Inuit” (Expresswaysprogram, Level 612, Handshakings student anthology) and “Hawaii” (Impressions program, Level 52,Knock at the Door student anthology); one copy of each per student and one for myself;2. Two practice selections of informational text: “Getting to Know Lake Ontario” (Impressionsprogram, Level 62, All Over the World student project book) and “The Invention of Levi’s”(Impressions program, Level 52, Knock at the Door student project book); one copy of each perstudent and one for myself;3. Informational text “Sharks” (Impressions program, Level 61, Wherever You Are); one copyper student and one for myself;4. Multiple Choice Meaning Clarification Activity forms “Sharks”; one copy per student; and5. Audio recorders, blank cassettes, and lapel microphones; one of each per student, except asecond cassette per individual for Session Three.The Multiple Choice Meaning Clarification Activity (MCMCA) consisted of fourteen multiplechoice items, each with one answer conforming to the meaning anticipated of an adult reader and twodistractors, as well as the option to write in a meaning if none of the presented choices agreed with themeaning constructed by the student while reading.iii. ProceduresI met with the students four times. The first meeting, held in the principal’s office, was informaland was intended to establish rapport with the students and to answer any questions they might have.Two of the pilot sessions were held in the ESL room and one in the multi-purpose/activity room.In the second meeting, Session 1, I followed a prepared script. I gave the students my reasons74for doing this study; shared with them some pertinent ideas about thinking, reading, and think aloud;demonstrated the individual think aloud procedure with the text “The Land of the Inuit”; provided thestudents with a sheet of simplified directions for doing the think aloud, explained the directions, andasked the students to refer to the directions as necessary throughout the following activity. Thestudents then completed a practice think aloud with the text “Getting to Know Lake Ontario.” Allportions of the session were audio taped and were later examined to determine the clarity of my ownpresentation and the quality of student talk, as well as to ensure that the students had handled thetape recorders without undue difficulty.Although I had intended the retell portion of the think aloud to follow an immediate playback ofthe tape, time restrictions prevented this procedure and the retell was not done. It was clear that I hadto revise my plan by eliminating the think-aloud replay. This was taken to be a simpler procedure. Itseemed to offer no disadvantage, since the students were urged to refer to the text throughout theretell.I concluded also that my think-aloud demonstration had been longer than necessary todemonstrate the procedure adequately. It was clear that this part of the plan needed revision.In Session 2, I gave feedback on the previous day’s practice think alouds, stressing such thingsas telling all thinking, not going too fast, and telling how something was known. I complimented thestudents for providing quite a lot of information--that it was a good first try. I confined mydemonstration of the dyadic think aloud, done with one of the students, to one page of the text“Hawaii.” My student partner worked easily with me, following my lead, taking her turn, and addingsignificantly to the talk.I changed the retell to immediately after the think aloud itself. As well, to my original script, Iadded two hints for working together: Listen to your partner and respect your partner’s ideas.I provided the students with a sheet of simplified directions for doing the think aloud in thedyadic condition, explained it, and asked them to refer to it as necessary during the practice activity.They completed a practice dyadic think aloud using the text “The Invention of Levi’s.” Dyads were75randomly assigned.The students indicated that they preferred working with a partner to working alone and thatunderstanding the think aloud would have been even easier if the dyadic part had been done first. Itook this into consideration when planning the main study.In Session 3, students were randomly assigned to complete either adyadic or individual thinkaloud using the text Sharks. Four students worked in dyads and the remaining two workedindividually. Upon completion of the reading think aloud, all students completed the MCMCAindividually as a combination think-aloud and paper task. The text was referred to so the task wouldnot be memory-dependent.Because of an undetected problem with the tape recorder, one of the dyads failed to recordtheir think aloud. This necessitated my return to the school to have them redo the activity--a notaltogether successful solution, since the repeated think aloud lacked the richness of language I hadanticipated, based on the practice sessions.Analysis of the pilot data consisted of three parts: analysis of the meanings constructed for themetaphors, according to the MCMCAs; analysis of the think-aloud protocols to determine the specifictraits exhibited (i.e., such features, inter alia, as paraphrase, restatement, inferencing, affectiveresponse, procedural); and holistic scoring, according to a scale devised for the purpose.The MCMCAs were scored against the anticipated meanings of adult native speakers. Eachitem had one anticipated adult-like response, one literal interpretation, and one tangential explanationfocusing on one part of the expression. Space was given for students to write in a meaning if none ofthe given choices agreed with the meaning they had constructed. If this option was chosen, thewrite-in explanation was analyzed to determine whether it could be restated as one of the givenmeanings or whether it could be considered an acceptable substitute for the anticipated adult-likeresponse. The students provided the anticipated adult-like responses 62 of a possible 84 times, or74% of the time. Individual scores ranged from 43% to 86%, with three of the six students scoring86%.76The 13 think-aloud protocols were examined (i.e., all target and practice protocols). Studentutterances apart from renderings of the text passage were analyzed to determine the thinkingprocesses involved. I took no advance categories to the task, allowing the categories to emerge fromthe protocols themselves. The specific traits identified gave one of potentially numerous possibleprofiles of the thinking processes used and meaning constructed during the reading. Individual anddyadic profiles based on these specific traits provided a means to compare the ways differentstudents completed the task. I noted that think alouds completed in the dyadic condition weregenerally longer and showed a wider range of specific traits than those completed in the individualcondition.Once I had analyzed the 13 protocols, I developed definitions for the specific traits andidentified examples. Two independent raters, using my definitions and examples, rated the fourprotocols for the target text “Sharks.” The three sets of ratings (from the two independent raters andmyseli were compared to establish inter-rater agreement. Two particular specific traits--Restatementand Paraphrase--were so similar as to make differentiation difficult.. These were subsequentlycollapsed into a single specific trait--Restatement/Paraphrase. Two other specific traits--Affirmation ofUnderstanding and Identifies Information not Previously Known--were also collapsed because ofconfusion during rating. Two-rater agreement--that is, agreement between at least two of the threeraters on every item--was calculated. Two-rater agreement between any two raters was 96.4%.Two-rater agreement between myself and at least one other rater was 94.8%.A six-point, Likert-type holistic rating scale was devised, using polar descriptors of qualitiesobserved in the protocols. Three students of a doctoral seminar, together with the instructor andmyself, used the scale to rate the four protocols produced during the reading of the target text“Sharks.” For each protocol, the five holistic ratings for each polar set were averaged, providing acomposite rating. The composite ratings were, in turn, averaged to provide an average rating for theentire protocol.The protocol that produced the greatest number of specific traits (117 on the initial think aloud77and 21 on the retell) and variety (12 of a possible 13 different traits on the initial think aloud and 3different traits on the retell> was given the highest average rating (5.1 of a possible 6) on the holisticrating scale by a wide margin. The protocol that produced the least variety of specific traits (4 of apossible 13) on the initial think aloud also received the lowest average rating (3.2 of a possible 6).With the other two protocols, the one that showed the greater number of specific traits (40 comparedto 30) and variety (8 compared to 5) also received a higher average ranking (4.0 compared to 3.6).The 14 metaphors used in the MCMCA were analyzed following Broderick’s (1992> scheme ofsyntactic frames and similarity types. No particular form of metaphor seemed to be more troublesomefor the students than another. One metaphor (“Pterodactyls glided the skies on huge, leatherywings”> of the two that were most difficult (anticipated adult response given by only one student) had astructure (Syntactic Frame: Descriptive Adjective phrase; Similarity Type: Disparate-kind) exactly thesame as one (“We see the imprints of their bat-type bodies...”> of the two that were understood by allsix students. The other metaphor (“In the wonderful cycle of the underwater world, everything floatsin its own balance”) understood by only one student had a structure (Syntactic Frame: DescriptiveVerb phrase; Similarity Type: Incompatible-kind) the same as a metaphor (“They weed out theweak...”) understood by five of the six students.Lakoff and Johnson (1980), Lakoff and Turner (1989), and Lakoff (1993) identify theconventionality of metaphor as central to its power, wide-spread use, and ease of understanding. At acertain point, the specific expressions of the underlying conventional metaphors become so familiarto speakers of the language that they are both produced and understood automatically. This processresults in numerous so-called dead metaphors (e.g., “He’s almost gone,’ used for a dying person,”[Lakoff & Turner, 1989, p. 129]) and idioms (e.g., “That’s still up in the air,” said of something that hasyet to be decided [Lakoff & Johnson, 1980, p. 137]). The confounding issue with such expressionsis that, although they may be entirely familiar to mature speakers, they may retain a degree of noveltyfor the less mature. Additionally, some children may well be exposed to and learn certain conventionalmetaphors and their common expressions earlier than other children in the same culture.78This commonness of particular metaphorical expressions complicates efforts to determinechildren’s proficiency in making meaning for metaphors, since it is difficult to ascertain without doubtwhether meaning is constructed on the occasion or is already held. In this pilot study, it was possibleto deduce that the metaphorical phrase “armour plating,” for example, was already known, since theanticipated adult meaning was identified by all students, although the text provided no contextualclues. Where context clues are sufficient to permit construction of meaning, provided adequate priorknowledge of metaphor source and target is also held, identification of actual source of meaning islikely to be complicated. To avoid the problem, I chose to focus on the constructed meaning itself andnot the source of meaning.The pilot study established that the think-aloud procedure would provide useful informationabout the thinking processes the readers engaged in and the meaning they constructed duringreading. It showed that specific-trait profiles of the protocols could be created, permitting comparison.Further, it revealed the extent of adult-like understanding of metaphors within authentic text.Difficulty was encountered in applying the concepts high score and low score asoperationalized. The decision was made to add miscue analysis as an additional means of analyzingthe think-aloud protocols. The operationalized definitions for understanding, high score, and lowscore were changed to the forms given in Chapter One, and Question 5 (see Research Proposalsection, above) was changed to the form given in the Research Questions section of Chapter One, totake into account the addition of miscue analysis.Based on the pilot results, I sought and received permission from my academic committee toundertake the main study.IV. Main Studyi. Research ApproachThe research approach was quasi-experimental. The study used three text passages ofdifferent levels of reading difficulty. Each reading think aloud was done in either individual or dyadic79condition. Each student completed two think alouds, one in each condition. Text passage,condition, and dyadic partner were randomly determined. No control group was used.ii. Research Design and InstrumentationThree target passages were used. These were of differing reading difficulty, as measured by anestablished readability formula. Each passage consisted of approximately six hundred words,comprising the first portion of a recently published informational trade book. Each text provided tentarget metaphors.The passages were selected from the National Council of Teachers of English (NCTE) project,Adventuring with Books--a major bibliographic undertaking currently housed at The University ofBritish Columbia (UBC), under the editorship of Dr. Wendy Sutton of the Department of LanguageEducation. An initial ranking by a project member had identified 66 books as being of the highestquality. I sorted these 66 books according to three criteria: probable interest to grade six students,reading difficulty judged appropriate for grade six students, and sufficient density of metaphor topermit a selection of ten metaphor samples for specific study in a passage consisting of approximatelythe first six hundred words of the book. This screening resulted in the selection of four books, one ofwhich I eliminated because it was a second title in a particular series of books by one publisher, and Ijudged it likely to be of less interest to students than its companion. The final selections were:1. Shadows of Night, by Barbara Bash, published by Sierra Club Publications, SanFrancisco,1 993;2. Frontier Home, by Raymond Bial, published by Houghton Mifflin, New York, 1993; and3. The Middle Ages, by Sarah Howarth, published by Viking, New York, 1993.The selected parts of the books were randomly designated Passage X, Passage Y, and Passage Z,respectively, for ease of reference.The three text samples were analyzed using the Flesch-Kincaid readability formula. Resultswere as follows:801. Passage X (Shadows of Night)--approximate grade level, 6;2. Passage Y (Frontier home)--approximate grade level, 11; and3. Passage Z (The Middle Ages) --approximate grade level, 8.I analyzed the thirty metaphors from the three selections according to Brodericks (1992)taxonomic scheme of syntactic frames and similarity types, as I had done in the pilot study.Additionally, I analyzed the metaphors according to Lakoff and Johnson’s (1980) and Lakoff andTurner’s (1989) description of type and features--whether, for example, the metaphor was a basicontological, a structural, or an orientational metaphor or was based on a basic conceptual metaphor,such as AN EVENT IS AN ACTION or the CONTAINER metaphor.I designed a ten-item Multiple Choice Meaning Clarification Activity (MCMCA) for each textpassage. Design was similar to the MCMCA used in the pilot study, that is, each item provided threepossible choices and the option to write in a meaning. The three item choices consisted of oneanticipated adult-like response and two distractors.Four pilot study text samples were chosen for demonstration and practice purposes (seeabove). “Hawaii” and “The land of the Inuit” were again used for demonstration, and “The Invention ofLevi’s” and “Getting to Know Lake Ontario” were used for dyadic and individual practice, respectively.All publishers were contacted for approval to make copies of the selections. In the case of thethree target passages, permission was sought to make high quality colour photocopies. A sampleletter is included in Appendix A.The holistic rating scale developed in the pilot study was adopted unchanged for use in themain study.For the main study, the key terms were operationally defined as for the pilot study, with theexception of changes to understanding, high score, and low score, as noted in the ResearchProposal section above.iii. Selection of ParticipantsI sought and received approval from the University Behavioural Sciences Screening Committee81for Research Involving Human Subjects to carry out the main study. Fifty-six grade six students fromone class in each of three public schools under the jurisdiction of a single school board in rural NovaScotia participated in the study. The 56 students (out of a potential 71) were those who agreed totake part and had parenljguardian permission to do so. One student was unable to complete the foursessions. Only data from the remaining 55 were used in the analysis. Anonymity of students andschools was maintained by coding all data.I chose to carry out the study in Nova Scotia because of my familiarity with the school system,with this particular school district, and with many of its teachers and administrators, and because of thecultural and linguistic stability of the area, a factor I considered to be of potential significance becauseof the culturally sensitive nature of metaphor understanding, according to Lakoff, Johnson, andTurner (see Chapter Two above).Initial contact was made with the superintendent of schools, who gave his approval for thestudy. The supervisor of curriculum took my request to the elementary principals, three of whomimmediately agreed to assist.The students who participated in the study were all native Canadian-English speakers. Theschools they attended and communities in which they lived were long-established and linguisticallystable. The traditional economic base of one of the communities had been fishing. This had beenjeopardized, however, by the decline of the Atlantic groundfishery. The community was struggling toadjust to a significant drop in standard of living for a large part of the population. The school wasaffected in that special measures had to be taken by teachers and principal to ensure that schoolactivities, whether co-curricular or extra-curricular, did not discriminate against children whose familieswere suffering financial hardship.The other communities were less dependent on the fishery. Theirs was a multi-facetedeconomy, influenced by agriculture, forestry, various service industries, and manufacturing, anchoredby one of Nova Scotia’s three tire manufacturing plants nearby. This plant provided a measure ofeconomic stability rare in the province.82The schools in all three communities were long-established. They provided a community focus,and school events and activities were generally well supported by students families. Communicationbetween the schools and their communities was supported by trust and a shared sense of purpose.The board’s policy of rotation of principals resulted in an infusion of new ideas and administrativestyle every few years. To implement their vision before the next rotation, principals had to capitalizeon the goodwill of the teachers and communities. This goodwill seemed to be forthcoming in thethree schools.Two of the schools were elementary, with grades primary to six. The third school had anelementary section and a junior-senior high school in the same complex. In the first two schools, thestudy participants were among the oldest group of students, while in the third they were among theoldest group in the &ementary, but approximately mid-age in the school complex.iv. Site ProceduresIn all schools, I met with groups of no more than six students at a time in a room separate fromthe regular classroom, such as the school library or resource room. In all cases the space providedadequate separation of students during their think alouds such that, during Sessions Three and Four,no student could hear another reading the same text passage.A script was prepared for each of the four sessions. The scripts for Sessions One and Two weresimilar to those prepared for Sessions Two and One, respectively, of the pilot study. The firstcontained information about my reason for doing the study and my ideas about thinking, reading, andthink aloud. The second reviewed the information shared during Session One. In each session, thequestion “Is this making sense to me?” was emphasized as a key question to ask oneself frequentlyduring reading, to answer, and to tell why or why not (adapted from Baumann, Jones, &Seifert-Kessell, 1993).In Session One, randomly assigned groups were introduced to the dyadic think aloud. Theywere given a demonstration, involving myself and a volunteer student, including demonstration of the83use of the recording equipment. The demonstration text “Hawaii” was used. The practice text was“The Invention of Levi’s.” The students were reminded to do a cooperative retell after completing theinitial think aloud. They were given individual copies of short, explicit directions to refer to whilecompleting the think aloud.In Session Two, randomly assigned, small groups were introduced to the individual think aloud.Proper use of the recording equipment was again demonstrated. The demonstration text “The Landof the Inuit” was used. The practice text was “Getting to Know Lake Ontario.” The students werereminded to do a retell after completing the initial think aloud. They were given individual copies ofshort, explicit directions to refer to while completing the think aloud.Session Three began with an overview of the session, followed by a review discussion ofthinking aloud and dyadic think aloud procedures. The students completed a dyadic think aloud witha randomly assigned target text (that is, Passage X, Y, or Z) and a randomly assigned partner. Again,the students were given individual copies of short, explicit directions to refer to while completing thethink aloud. An individual MCMCA was completed as a combined paper task and think-aloud activity,with each student working on a separate tape recorder.Session Four was conducted in the same manner as Session Three, except that the thinkalouds were done individually. The random draw was constructed to ensure that each student read adifferent passage from that read in Session Three.To avoid memory-dependence during the MCMCAs, subjects kept the text and were instructedto reread the specific portion of passage containing the metaphor related to each MCMCA item. Thelines of text containing the metaphors were identified with fluorescent yellow, adhesive dots, eachnumbered to correspond with the appropriate MCMCA item number. Copies of the three MCMCAsare included in Appendix B.v. Data Collection and RecordingFor the purpose of assuring anonymity of students, each was assigned a five part code such as,for example, Z2F4K. The first part could be X or , to indicate passage. The second part could be84either 1 or to indicate individual or dyadic condition, respectively. The third part indicated subjectgender and could be either E or M, to indicate female or male, respectively. The fourth part could be3, 4- or 5, to indicate one of the three schools, to which the numbers had been randomly assigned.The fifth part, a letter of the alphabet, was used to indicate randomly assigned position on a list ofparticipants by school. Thus, the example given above, Z2F4K, indicates that a female from school 4,the eleventh student on the randomized class list, completed a think aloud for Passage Z in the dyadiccondition.Data of two types were collected. One hundred ten MCMCAs (two produced by each student)provided information on the meanings students constructed for the metaphors within the context ofauthentic text. Responses were recorded on a spreadsheet, with the numeral 1 representing ananticipated adult-like response and the numeral Q representing any other response.The audio tapes of the think alouds were transcribed. An excerpt from a think-aloud protocol isincluded in Appendix C. In all, 55 individual and 28 dyadic protocols were produced (one of each perstudent). A fifty-sixth student contributed to a dyadic think aloud but, because of the start of a familyvacation, did not complete an individual think aloud. Data gathering had been scheduled to concludebefore the student left, but a one-day school cancellation due to weather forced an extension into asecond week. The student’s partner was counted.The protocols were analyzed using a modified version of the specific-traits inventory developedin the pilot study. The list of specific traits and definitions is included in Appendix D. This modifiedversion added three specific traits--[Evaluation of] Discussion, Procedural and Other/Miscellaneous--to the original thirteen. Sub-traits were added to account for those cases in which a speaker simplyagreed with an utterance of the previous speaker. Eight of the specific traits identified were similar tocategories and elements identified by Purves and Rippere (1968). (The specific traits were also similarin several respects to general comprehension strategies and local linguistic strategies developed byBlock, 1986, who had followed the ideas of extensive and reflexive modes used by Emig, 1971, andPerl, 1978 [cited in Block, 1986].) The specific traits were combined to form three overarching85categories--Meaning-Construction Processes (M/C), Neutral States (N/S), and Pre-MeaningConditions (P/M). The first, Meaning-Construction Processes, encompassed all those specific traitsthat revealed the students to be actively seeking to construct meaning and doing so successfully.Pre-Meaning Conditions were the opposite. Neutral States revealed the students to be orientingthemselves toward the task itself. This is consistent with the suggestions made by Afflerbach andJohnston (1984) that this process reduces the effect of inferencing and increases the accuracy of theratings. Because of the high two-rater agreement achieved in the pilot study, I felt justified incompleting the specific-trait analysis myself. Combining the specific traits into the three overarchingcategories produced 100% inter-rater agreement. Samples of analysis are included in Appendix E.Following Purves and Rippere (1968, p. 49), the specific traits were used to construct profiles oftypical think alouds for students, in this case for those whose perlormance on the MCMCAs placedthem in extreme groups roughly equivalent to the first or fourth quartiles (see Chapter Four, QuestionSix, for an explanation of the composition of the extreme groups). Appendix F gives average profilegraphs of the think alouds of students scoring in the extreme groups in both individual and dyadicconditions.The information resulting from the specific-trait analysis was further quantified by expressing thenumber of occurrences of each trait as a percentage of the total number of traits identified in theprotocol.All protocols were rated holistically by myself and three other adults who had been instructed inthe process. The resulting four scores for each pair of descriptors were averaged and used incalculating an overall average for the protocol. A sample holistic rating form is included in Appendix G.A miscue analysis was perlormed for each of the 55 protocols produced in the individualcondition. Dyadic protocols were not analyzed, since the interaction of readers made it unlikely that asingle student’s contribution could be ascertained with accuracy. While miscue analysis hasgenerated disagreement among researchers as to what miscues should be counted (see, forexample, McKenna, 1983), how the level of each type of miscue relates to understanding of text86(Englert & Semmel, 1981), and how to assign independent, instructional, and frustrational readinglevels (Lowell, 1970), the procedure has enough research support and currency of use to make itacceptable as one means among several for assessing student understanding of text.The opening three to five paragraphs of each passage were selected for miscue analysis,consisting of 269 words for Passage X, 280 words for Passage Y, and 256 words for Passage Z. Eachmiscue noted during the transcription from audio tape was analyzed. If the miscue was corrected bythe student or was of such a nature as to maintain the meaning of the text, it was not counted. If,however, the miscue altered the meaning of the text, it was considered as serious and was counted.Reading levels were assigned as follows: less than 2% serious miscues, easy reading level; 2% to 5%serious miscues, instructional reading level; over 5% serious miscues, frustrational reading level(following a scheme field tested in the Portland, Oregon, Public Schools, mentioned in May, 1990, p.388, and subsequently reported in Dean’s [1991] study, cited in May, 1994, p.410).The specific traits and holistic ratings, together with results of the miscue analysis, were used tosuggest possible high and low levels of text understanding.Student performances on MCMCAs after completing individual and dyadic think alouds werecompared. Overall performances on MCMCAs by same-text groups in the dyadic and individualconditions were compared.Performances on different types of metaphors were compared using Broderick’s (1992)taxonomy and Lakoff and Johnson’s (1980) and Lakoff and Turner’s (1989) types.vi. Methodological AssumptionsInter-rater agreement for the specific traits, as established in the pilot study, was 96.4%between any two of three raters. Two-rater agreement between myself and at least one other raterwas 94.8%. It was assumed that the specific traits had sufficient validity to be used in analyzing thethink-aloud protocols of the main study. Additionally, the combining of specific traits into three moregeneral categories reduced the amount of inferencing required during the rating process, thus87increasing overall agreement to 100%.The following methodological assumptions underlie this study:1. That ecological validity is enhanced by using maximally authentic text (Broderick, 1992);2. That the think aloud can provide information about meaning constructed during reading(Waern, 1980; Ericsson and Simon, 1993);3. That specific-trait analysis of think-aloud protocols can provide information about meaningconstructed during reading (Purves and Rippere, 1968; Dias, 1987);4. That miscue analysis has been established as a useful procedure for the roughdetermination of reader understanding of text (May, 1990, 1994);5. That the multiple choice has been established as a reliable form of testing for meaning; and6. That the text passages used in this study, in that they have been drawn from trade bookspublished by major North American publishers, are similar to informational texts being used in manyclassrooms.vii. LimitationsThis study was undertaken to examine how well grade six students understand metaphor in theinformational text they read, how well the same children understand the informational text itself, andhow understanding of text and understanding of metaphor are related.As stated in Chapter One, in order to make this study manageable in terms of time and cost,other potentially interesting questions were not examined. It was not possible to investigatechildren’s understanding of metaphor in fiction or poetry or to explore psychological aspects ofmetaphor or text understanding. The study was limited to grade six students, although a comparisonof students at different grade levels should be informative. This study’s results can be taken as onlypossible indications of metaphor and text understanding of younger and older children.It was necessary to develop instruments and methods for this study since no measures wereavailable for use under the conditions and for the purposes identified. This study used methods and88materials as much as possible like those familiar to students from their classroom experience. I hopedto maintain the ecological validity of the material and activity and to maintain a closer contact with thecurriculum than would be possible if instruments and methods were borrowed from previous studies.The test-like situation was not entirely avoided, however, and the ecological validity was weakened tothe extent that students worked in small groups outside their classroom for a purpose established by astranger.The contexts in which metaphors were presented were similar to the original texts; however, thecomplete books were not given to the students and numbered fluorescent dots were affixed atspecific points in each text, thus altering the appearance of the passages. The target passages wereof different levels of reading difficulty, according to a standard reading formula; consequently, certainanalyses involved a somewhat lower n than would have been the case had all passages been at thesame reading level.Although such an examination is needed, this study did not attempt to examine the potentialproblems metaphor presents to English-as-a-second-language (ESL) learners.V. SummaryThe three major and seven specific questions around which the study was organized are asfollows:(a) How well do grade six students understand metaphor in the informational text they read?1. What is student level of understanding of metaphor as determined by performance on amultiple choice meaning clarification activity?2. What are the differences in understanding of metaphor by the same student in the individualand dyadic think aloud conditions as determined by performance on multiple choice meaningclarification activities?3. What are the differences in understanding of metaphor in the same text for the individual anddyadic think aloud conditions as determined by performance on a multiple choice meaning clarification89activity?4. What are the differences in understanding of different types of metaphor as determined byperformance on multiple choice meaning clarification activities?(b) How well do the same children understand the metaphor-bearing informational text itself?5. What proportion of students achieve high and low scores on think-aloud tasks as determinedby holistic and specific-trait scoring and miscue analysis?(c) How are understanding of text and understanding of metaphor related?6. What are the features of the think alouds of students whose performance on a multiplechoice meaning clarification activity placed them in the first and fourth quartiles?7. What is the relationship between the level of students’ understanding of metaphor and theirunderstanding of text?The pilot study showed that the selected methods and procedures could provide potentiallyuseful information about grade six students’ understanding of metaphors and the text in which theyare embedded. Specific traits identified in the think-aloud protocols provided a way of describing thestudents’ meaning constructed during the reading. The multiple choice activities provided a means ofcomparing student understanding of metaphor with anticipated adult understanding.The study drew from a population of grade six students for whom Canadian English was theirfirst language. The target passages were drawn from recently published informational trade books.Reading think alouds were performed and recorded, and students completed multiple choiceactivities on the metaphors. Students completed both dyadic and individual think alouds.Scoring of the multiple choice activities gave numerical data on student understanding ofmetaphors. These were used in conjunction with type analysis of the metaphors to provideinformation on performance by type. Specific-trait analysis of the think-aloud protocols providednumerical data concerning understanding of text. From this information, think-aloud profiles wereconstructed. Holistic ratings provided numerical data on the protocols, as did miscue analysis. The90holistic ratings and miscue-analysis scores were used along with specific-trait analyses to suggestpossible high and low levels of text understanding.Assumptions were made about specific-trait analysis of think-aloud protocols, theappropriateness of miscue analysis as a device for determining understanding of text, the usefulnessof multiple choice activities in determining constructed meaning for metaphor, and the similarity of thetarget passages to texts being used today in classrooms.Limitations of generalizability because of the nature of the sample population and validitybecause of the originality of the instruments have been noted.91CHAPTER FOUR: FINDINGSI. Introduction 91II. Findings 92i. Major Question (a) 92a. Question One 92b. Question Two 96c. Question Three 99d. Question Four 99ii. Major Question (b) 107e. Question Five 107iii. Major Question (c) 116f. Question Six 116g. Question Seven 134Ill. Summary 139I. IntroductionThe examination of data involved both qualitative and quantitative analysis. In the case of themultiple choice activities, numerical data were gathered, permitting statistical examination according tostandard practices. Analysis of the think-aloud protocols, on the other hand, provided specific traitsdescriptive of aspects of the reading process as engaged in by these particular students. Thequantification of the occurrences of the specific traits provided data for standard statistical tests anddescriptions of the protocols, including percentage profiles. Holistic rating of the protocols wascarried out by myself and three other adults who had been instructed in the process. This Liked-typeholistic scale provided additional numerical data. Miscue analysis was performed on the 55 individualthink-aloud protocols.92II. Findingsi. Major Question (a)How well do grade six students understand metaphor in the informational text they read?a. Question OneWhat is student level of understanding of metaphor as determined by perlormance on a multiplechoice meaning clarification activity?Students completed a ten-item multiple choice meaning clarification activity (MCMCA)individually after each think aloud, whether individual or dyadic. The 110 MCMCAs so producedyielded 1100 item responses. The item responses were scored against anticipated adult-likeresponse, operationalized as the selection of the single non-distractor meaning provided for eachmultiple choice item (see Research Design and Instrumentation, Chapter Three, and Appendix B).From the raw scores, performance means, standard deviations, and percentages were calculated forthe whole set, as well as for each passage in both individual and dyadic conditions (six subgroups).Results are summarized in Table 1.According to these data, overall student understanding of metaphor matched anticipatedadult-like understanding for an average of 65% of the items. In the dyadic condition, averageunderstanding ranged from a low of 50% (Passage Y) to a high of 69% (Passage X). In the individualcondition, average understanding ranged from a low of 57% (Passage Y) to a high of 77% (PassageX).Student scores varied widely (see Appendix H). Single scores ranged from 1 to 10 (see Figure1) and combined scores (2 tests) ranged from 5 to 20 (see Figure 2). Thus, understanding ofmetaphor, as measured by these multiple choice activities, varied widely among these particularstudents. This finding may reflect a difference in ability, but it is also consistent with the view that thereis a developmental aspect to metaphor understanding, although it says nothing about the nature ofthat development.93Table 1Multiple Choice MeaninQ Clarification Activities--Performance StatisticsPassage Condition No. Students Meana Standard Deviation PercentbX Dyadic 18 6.944 2.100 69.4Individual 19 7.684 1.416 76.8Total 37 7.324 1.796 73.2Y Dyadic 18 5.000 2.223 50.0Individual 19 5.737 1.939 57.4Total 37 5.378 2.086 53.8Z Dyadic 19 6.421 1.677 64.2Individual 17 7.353 1.967 73.5Total 36 6.861 1.854 68.6All Passages 110 6.518 2.075 65.2aQut of 10.bflounded to nearest tenth.The distribution of all multiple choice (MCMCA) scores showed the general configuration of anormal distribution (see Figure 1). To test the degree to which the distribution conformed to thenormal distribution, the w/s test for normality of a population was performed. For a range of 9 andstandard deviation of 2.075, the studentized range statistic was =4.337. By interpolation from tabledvalues, for n=1 10, the critical values were 4.366 and 5.956, a=.05. Since the calculated value fellbelow the lower critical value, a significant difference between the sample frequency distribution andthe normal frequency distribution was indicated.U)Ca)4-,(I)‘4-0‘1)0Ez5 6 7 8 9 10 11 12 1314 151617 18 19 20Total Score302094CDN>.‘C.)Ca)a)U-100—1 2 3 4 5 6 7 8 9 10MCMCA ScoreFigure 1 Disfribuon of MCMCA Scores10-86420NCD CD CD•Z•”N Ih{1__ 111111.Figure 2 MCMCA Scores--Student Totals95This test compared the number of standard deviations encompassed by the range of scoreswith the number expected in a normal distribution for a given jj. For an n of 100, for example, theexpected number of standard deviations was 5.0 (Downie & Heath, 1983, p. 50), although, as thecritical values of the w/s test indicated, the number could vary from 4.31 to 5.90 (Kanji, 1993, p. 183).Because the distribution of MCMCA scores was neither bimodal nor seriously skewed, it was feltthat usual statistical procedures did not have to be abandoned.The w/s test for normality of a population was also performed for all three passages in bothindividual and dyadic conditions (3 passages x 2 conditions). All values of tell between therespective critical values, indicating no significant differences between the samples and a normalfrequency distribution. Results are summarized in Table 2.Since the samples were normally distributed, the Hartley test for equality of variances wasperformed for the six subgroups, with s ranging from 17 to 19, as shown in Table 1. This test isusually performed on same-size samples; however, it may be used for samples of approximately thesame size (Kanji, 1993, p. 64). The critical value of Fm (a=.05) for n - 1=15 and 6 groups was4.68. When n - 1=20, the critical value was 3.76. The calculation produced an E value of 2.463, wellbelow the critical value for either n. From this, it was indicated that there were no significantdifferences among the variances of the six samples.Findings. Pertaining to Question One, the findings were:1. Overall, students exhibited a 65% level of understanding of the metaphors; and2. There were wide variations in the scores for the three passages, with averages of 54% onPassage Y, 69% on Passage Z, and 73% on Passage X.96Table 2Summary of wls Test for Normality of a Population--Passage x ConditionPassage Condition Critical ValuesaX Individual 19 3.531 3.14 and 4.43Dyadic 18 3.810 3.10 and 4.37Y Individual 19 3.610 3.14 and 4.43Dyadic 18 3.149 3.10 and 4.37Z Individual 17 3.559 3.06 and 4.31Dyadic 19 3.578 3.14 and 4.43b. Question TwoWhat are the differences in understanding of metaphor by the same student in the individualand dyadic think-aloud conditions as determined by perlormance on multiple choice meaningclarification activities?A Pearson product-moment (Pearson ) correlation coefficient was calculated for the individualand dyadic scores (see Appendix H), producing a value, t=.362. For ct=.05 and n=55, the criticalvalue was 0.2616 (by interpolation). From this, it was concluded that there was a moderate positivecorrelation between the individual and dyadic scores.A t-test for dependent samples was also performed to compare the individual and dyadic scores(means of 6.909 and 6.127, respectively), with a result, t=2.51. With a=.05 and f=54, the criticalvalue was =2.006 (by interpolation). Since the t score exceeded the critical value, it was concludedthat there was a significant difference between the two sets of scores at the .05 level, with theindividual scores being significantly higher than the dyadic scores.97The student scores are presented in Table 3 in the order in which they were achieved andaccording to passage read--that is, dyadic score first, followed by individual score. When the scores inTable 3 for each student were summed and graphed, the resulting distribution showed somecharacteristics of a normal distribution (Figure 2).For the combined scores, the mean was 13.036, the variance was 11.199, and the standarddeviation was 3.346. 69.09% of the combined scores fell within ± one standard deviation of themean, with the first standard deviation below the mean contributing 38.18%. 92.73% of thecombined scores fell within ± two standard deviations of the mean. The w/s test for normality of apopulation produced a studentized value of=4.48. For n=55, the critical values were 3.90 and 5.43(ci =05), indicating no significant difference between the sample set and a normal distribution.When the individual and dyadic scores were graphed separately, their relative contributioncould be compared. Figure 3 presents the two sets of scores for the purpose of comparison. Thedifference revealed by the t-test for dependent samples is suggested by the distribution of the twosets relative to each other.20181614‘ 12— Dyadic1 0 Individualco co8(0 (064N N NN N201 2 3 4 5 6 7 8 9 10MCMCA ScoreFigure 3 Frequency of MCMCA Scores--Dyadic and Individual98Table 3Changes in MCMCA Scores According to Level of Passage DifficultyXtoY XtoZ YtoX YtoZ ZtoX ZtoYF3D 6,5 M3K blOb F3L4,7 M3G 7,9 F3B4,7 F3A7,5F3H 9,8 M4E 5,7a F4Q 7,9 F31 6,9 F3C 4,9 M3F 3,2M3N 7,6 M4M 8,7 F4X4,5 F3J 7,10 M3E 8,7a M4G 8,4F4A 9,7 M5C 7,6 M5E 5,6 F4B 1 ,7 F3M 8,8b F4S 7,6F4C 10,8 F5D 7,8a M5J 6,8 F4D 7,8 F4J 5,6 M5A5,gaM4L5,7a M5N8,10 F4F3,5 F4K5,10 M5K8,3M40 7,5 M4H 5,7 M4N 8,9M4T7,5 F412,3 M4P7,8F4V 2,4a F4R 4,b M4U 9,8aFSH 7,6 M5B 5,9 M4W 7,9F5l9,9b F5F1,8 M5G6,7F5M 6,6b M5Q 8,8b M5L 6,6bF5P 4,b F50 7,bNote. Level of text difficulty is judged according to the Flesch-Kincaid Readability Formula: PassageX, least difficult; Passage Z, more difficult; Passage Y, most difficult. Each entry consists of studentcode and scores in the order they were achieved.aOpposite to the order predicted by passage difficulty.bldentical scores on both passages.Findings. Pertaining to Question Two, the findings were:3. There was a significant positive correlation between students’ individual and dyadic scores;and4. Student understanding of metaphor was significantly higher after completing the think aloud99in the individual, as opposed to the dyadic, condition.c. Question ThreeWhat are the differences in understanding of metaphor in the same text for the individual anddyadic think-aloud conditions as determined by performance on a multiple choice meaning clarificationactivity?The three passages (X, Y, and Z) and the two conditions (individual and dyadic) provided sixdifferent sets of MCMCA scores (see Appendix I). Harmonic means were calculated for all groups andthese were used in performing the Tukey procedure (also known as the HSD--honestly significantdifference--test). This procedure revealed significant differences between subgroups involvingdifferent passages but did not reveal any significant differences between individual and dyadic scoresfor the same text passage.Finding. Pertaining to Question Three, the finding was:5. There was no significant difference in understanding of metaphor for the same passagebetween the dyadic and individual conditions.d. Question FourWhat are the differences in understandinc of different types of metaphor as determined byperformance on multiple choice meaning clarification activities?The Kuder-Richardson Formula 20 test was performed on the multiple choice meaningclarification activities to determine the degree of convergence of the items--that is, the degree towhich the items appeared to measure a single psychological construct. (Neither this test nor any othertest of internal consistency was reported in any of the studies examined for the present study.)Results for all three MCMCAs were in the range of 0.50.For certain test situations and purposes, this level would be considered seriously low. For a test1 00specially constructed for research, for example, a high degree of internal consistency is sought.However, the metaphor samples used in this study were not specially constructed, but occurrednaturally in the target texts. As well, the MCMCAs were short--consisting of 10 items each-- and testsof internal consistency normally return lower scores for short tests than for lengthier ones.In the construction of the MCMCAs, certain factors played a decisive role. For example, boththe choice of metaphor and the order were determined by the passages themselves. This wasintended to make it easier for the students to locate the metaphors in the text and to provide anadditional measure of ecological validity by keeping the task as similar as possible to an actualclassroom reading task.There is also the question of whether understanding of metaphor is a simple or a complexpsychological construct. If it is the former, then the low level of internal consistency would be aconcern; on the other hand, if metaphor understanding is not a simple psychological construct, onecould expect a low level of internal consistency, as evidenced by the results of these MCMCAs. Thisis consistent with the ideas of Broderick (1992), Lakoff and Johnson (1980), Lakoff and Turner(1989), and Lakoff (1993), as described above. It is the contention here that metaphor understandingis not a simple psychological construct, but rather a cluster of psychological elements and processes;thus, the low internal consistency of the multiple choice activities is to be expected.The 30 MCMCA metaphor samples were analyzed within the context of the text passagesaccording to the two ideas of metaphor described in Chapter Three: first, the syntactic-frame!comparison-type taxonomy put forward by Broderick (1992); second, the classification scheme usedby Lakoff and Johnson (1980), Lakoff and Turner (1989), and Lakoff (1993).The types of metaphors, as classified according to Broderick, are given in Appendix J. Usingthis information and that from Appendix K (MCMCA Results by Item), the percentages of adult-likeresponses to the individual items were calculated and are summarized in Table 4.101Table 4MCMCA Percentage Scores According to Metaøhor Type (After Broderick, 1992)Passagex Y zIncomp.a Disp.b Tot. lncomp. Disp. Tot. Incomp. Disp. Tot. TotalIS(lmplied)c 56.8 56.8 45.9 45.9 69.4 69.4 55.5iSd 81.1 81.1 32.4 32.4 56.8DS(VP) 80.2 80.2 64.9 64.9 69.9 69.9 71.2DS(Adj)f 60.8 73.0 66.9 80.6 47.2 63.9 65.9DS(NP) 86.5 86.5 86.5DS(PP)h 51.4 51.4 51.4Total 71.4 77.5 73,2 55.4 41.9 53.8i 71.0 47.2 68.6 65.2iN. Combined metaphors(not included above):1 x [DS(Adj)/lncomp. + DS(VP)/Disp.]--Passage Y, 48.6%1 x [DS(VP)/Disp. + lS(lmplied)/lncomp.]--Passage Y, 73.0%All percents rounded to nearest tenth.alncompatiblekind similarity type. bDisparatekind similarity type. cCopula verb (implied) syntacticframe. dCopula verb syntactic frame. eDescriptive (verb) phrase syntactic frame. fDescriptive(adjective) phrase syntactic frame. gDescriptive (noun) phrase syntactic frame. hDescriptive (presentparticiple) phrase syntactic frame. ‘Includes combined metaphors. JAIl scores.The extreme cases were identified at two levels--those with adult-like response below 50% andthose 80% and above. Eight metaphors were identified in each group. Those below 50%, along withtheir percentages for the dyadic and individual conditions and metaphor type, are given in Table 5 andthose 80% and above follow in Table 6.An examination of the types of metaphors represented in Tables 5 and 6 revealed that threeoccurred on both lists--that is, DS(Adj)/Incompatible-kind; lS(Implied)!lncompatible-kind; andDS(VP)!lncompatible-kind. For these three types, the overall adult-like responses were 67% (on 3items), 55% (on 6 items), and 71% (on 12 items), respectively, suggesting that there was nothingunique about the types themselves to account for the extreme scores on the individual items.102Scores--Sub-50% by Passaae. Item. and Metaohor TVDeItem %(Dyad) %(lndiv) %(Tota TypeX-4 44.4 47.4 46.0 DS(Adj)/lncompatible-kindY-1 38.9 57.9 48.6 Complex {[DS(Adj)/lncomp-Kind] + [DS(VP)/Disparate-kind]}Y-5 38.9 42.1 40.5 IS(l mplied)/l ncompatible-kindY-7 38.9 31.6 35.1 IS(lmplied)/lncompatible-kindY-9 38.9 26.3 32.4 IS/Disparate-kindY-10 44.4 52.6 48.6 DS(VP)/Incompatible-kindZ-3 42.1 52.9 47.2 DS(Adj)/Disparate-kindZ-4 31 .6 58.8 44.4 DS(VP)/Incompatible-kindNote. The type code used is that explained in Chapter Three and in the note to Table 4, above.All percents rounded to nearest tenth.Three types were without examples in either table--that is, IS/Disparate-kind;DS(NP)/Disparate-kind; and IS/Incompatible-kind. All, however, were represented by only a singleitem out of the 30 items of the three MCMCAs. With such meager representation, it was not possibleto draw meaningful conclusions.A further reduction of extreme scores was done to identify only the bottom and top three--thatis, those below 41% and above 86%. Two of those so identified were represented by single items,both the lowest (at 32%) and the highest (at 87%). The remaining two lowest-score items were both ofthe type, IS(lmplied)/lncompatible-kind; for the highest-score items, the remaining two were bothDS(VP)/lncompatible-kind. Again, since both of these types were represented in both lists, noconclusions as to their significance could be drawn.Table 5MCMCA103Scores--80% and Above by Passaae. Item. and MetaDhorTvneItem %(Dyad) %(lndM %(Total) TypeX-2 77.8 84.2 81.1 DS(VP)/lncompatible-kindX-5 83.3 89.5 86.5 DS(NP)/Disparate-kindX-6 83.3 78.9 81.1 IS/Incompatible-kindX-9 88.9 84.2 86.5 DS(VP)/lncompatible-kindZ-1 73.7 94.1 83.3 DS(VP)/Incompatible-kindZ-6 78.9 82.4 80.6 IS(lmplied)/lncompatible-kindZ-7 84.2 88.2 86.1 DS(VP)/lncompatible-kindZ-9 73.7 88.2 80.6 DS(Adj)/lncompatible-kindThe type code used is that explained in Chapter Three and in the note to Table 4, above.All percents rounded to nearest tenth.When all metaphor types were grouped according to syntactic frame only (excluding the twocombined metaphors identified above), IS (equation or copula-verb) types were shown to have had anadult-like response of 56% and DS (descriptive phrase) types, 69%. Grouping by similarity typeproduced the following results: Incompatible-kind, 66.8%; Disparate-kind, 60.6% (again, excludingthe two combined metaphors). For details, see Appendix J.A chi-square test for consistency was performed to examine the significance of the differencesbetween the IS and DS raw scores for the values given above. With =1 and a=.05, the criticalvalues were 98•10-5 and 5.02. The calculated value was 17.209. The calculated value exceeded theright critical value, indicating that the difference between the two sets of scores was significant.In his discussion of the ecological validity of the types of metaphors used in the stimulus sets ofa number of studies, Broderick (1992) points out that studies have depended largely on metaphorsTable 6MCMCA104constructed according to the IS syntactic frame, yet metaphors of this type were uncommon in theauthentic texts he examined. He maintains that some of the low rates of understanding shown bythese studies could be accounted for by this lack of ecological validity--that is, that children were beingasked to recognize types of metaphor with which they were largely unfamiliar.In the current study, IS syntactic-frame metaphors account for only 8 of 28 metaphors on thethree multiple choice activities (combined metaphors excluded). This low count, while higher thanthat reported by Broderick, does represent an imbalance between metaphors of the differentsyntactic-frame types. Additionally, the chi-square result seems to be further support for Broderickscontention that children find equation (copula-verb) type metaphors more difficult than metaphors ofthe descriptive-phrase type.A chi-square test performed on the raw scores for Incompatible- and Disparate-typecomparisons did not reveal significant differences, suggesting that students understood these twotypes equally well.The 30 metaphors used in the MCMCAs were also analyzed using the categories described byLakoff and Johnson (1980), Lakoff and Turner (1989), and Lakoff (1993). The results of this analysisare tabled in Appendix L.The data from the analysis were examined for potential relationships between the structure ofthe metaphors and student performance on them in the MCMCAs. Again, the eight high- andlow-extreme items were taken for the initial comparison. Figures 4 and 5 show the composition of themetaphors in each group.By comparing the two sets, it was found that three types of metaphor--orientational, ontological,and event/action--made up 20 of the 22 elements in the high-extreme set and 21 of the 24 elementsof the low-extreme set, with an actual overlap of 18 (ignoring any excess in one or the other set).Further, the average number of elements in each item of the high-extreme set was 2.75, while for thelow-extreme set the average was three. No significant difference between the sets was apparent.105543>C.)a)a)I.)2MCMCA ItemS Structure/MapOrientationalOntologicalEvent/ActionD Life/Journey0Figure 4 Metaphor Features--High-Extreme MCMCA Items (After Lakoff, Johnson and Turner)OrientationalOntologicalEvent/Action0 ContainerIZ AnalogyAnalogy(Compl)X-4 Y-1 Y-5 Y-7 Y-9 Y-1 0 Z-3 Z-4MCMCA ItemFigure 5 Metaphor Features--Low-Extreme MCMCA Items (After Lakoff, Johnson and Turner)106The three highest- and three lowest-percentage items were also examined. See Figures 6 and7. Each set of three was composed of eight elements, averaging 2.67 per item, with an overlap of fourelements in two types of metaphor--orientational and ontological. The remaining tour elements weredistributed among three types of metaphors, with only one--event/action--providing more than oneelement. While the sets appeared different, it seemed unlikely that those differences were sufficientto account for the wide range of scores.Findings. Pertaining to Question Four, the findings were:6. Student understanding of metaphors of the copula-verb (equation) syntactic-frame type wassignificantly lower than of the descriptive-phrase syntactic-frame type; and7. The Lakoff, Johnson, and Turner method of analyzing metaphors did not isolate differencesto account for the varying performance scores on the MCMCAs.543>C.)Ca)a)LL0X-9MCMCA ItemFigure 6 Metaphor Features--Highest Three MCMCA Items (After Lakoff, Johnson and Turner)Structure/MapOrientationalOntologicalEvent/Action1 07OrientationalOntologicalAnalogy(Compl)Figure 7 Metaphor Features--Lowest Three MCMCA Items (After Lakoff, Johnson and Turner)ii. Major Question (b)How well do the same children understand the metaphor-bearing informational text itself?e. Question FiveWhat proportion of students achieve high and low scores on think-aloud tasks as determined byholistic and specific-trait scoring and miscue analysis?All think alouds were transcribed from audio-tape. The resulting protocols (see sample,Appendix C) were printed and underwent specific-trait analysis using the specific traits developed inthe pilot study and refined during the analysis of the main study protocols (see Appendix D fordefinitions of all specific traits). A sample specific-trait analysis from the main study is given inAppendix E.As described in Chapter Three, the specific traits were combined to form three overarchingcategories: Meaning-Construction Processes (M/C), Neutral States (N/S), and Pre-MeaningConditions(P/M). This procedure is consistent with the idea put forward by Afflerbach and Johnston1MCMCA Item108(1984) that the identification of easily inferred processes will result in higher inter-rater agreement thanwhen identification of the processes requires a high degree of inferencing, as is necessary with theindividual specific traits. The pilot data on inter-rater agreement are consistent with this position, withtwo-way agreement on the specific traits (mid-90%) rising to all-rater agreement of 100% when thespecific traits are combined into the three overarching categories. These combined categoriesdifferentiate among the various stances-toward-task the students exhibit in their think alouds. TheMeaning-Construction Processes, as the name suggests, encompass all those specific traits thatreveal the students to be actively seeking to construct meaning and being successful in theendeavour. The Pre-Meaning Conditions are essentially the opposite, while Neutral States reveal thestudents to be orienting themselves toward the task itself.From the data, profiles were created showing both average number and average percent ofutterances for several sets of protocols. Presentation of the profiles by percent is advocated byPurves and Rippere (1968) as a useful way to detect patterns and make comparisons while avoidingthe distortions produced by large differences in the number of utterances. Since each can provideuseful information, however, both number and percent are used in this study.The profile for all protocols is shown in Figure 8. The total number of protocols contributing datawas 83, consisting of 55 individual and 28 dyadic. As can be seen from the figure, students producedmany more utterances (70%) whose purpose was to make meaning (MIC) than for other purposes,although orientation toward the task accounted for approximately one-fifth of the utterances.Figures 9 and 10 show similar information for the dyadic and individual protocols, respectively,as separate sets. Examination of Figures 9 and 10 reveals that students produced more utterances ineach category in the dyadic condition than in the individual condition, with the average total dyadicutterances being slightly over 267% of the average total individual utterances. A chi-square test wasperformed to determine whether the difference between the total average number of specific traits inthe dyadic and individual protocols was significant. With df=1 and a=.05, the critical values were98•1 o- and 5.02. The calculated value, 43.521, exceeded the upper critical value, indicating that109M/C N/S P/M Totalthe averages obtained under the two conditions were not consistent with each other.100800C)0 60g 40 Average#Average%>20CategoryFigure 8 Specific Traits--Average Number and Average Percent--All Protocols1401201 000C)0 800 Average#D 60 Average%C)La)>200M/C N/S P/M TotalCategoryFigure 9 Specific Traits--Average Number and Average Percent--Dyadic Protocols1101401201 000000I12 Average#00 Average%ci)0)Ici)>200CategoryFigure 1 0 Specific Traits--Average Number and Average Percent--Individual ProtocolsAs can be seen from Figures 9 and 10, the relative changes in percentages for the separatecategories are noticeable between dyadic and individual protocols, with Meaning Construction (MIC)contributing over 65% of the utterances in the dyadic protocols but over 76% in the individualprotocols. A second substantial shift is to be seen in the Neutral States (NIS) category, which makesup almost 27% in the dyadic condition but under 15% in the individual condition. As major as theseshifts appear to be, chi-square tests did not reveal significant differences, from which one can assumethat students were relatively consistent in their think-aloud strategies whether working alone or with apartner; in other words, the dyadic and individual think-aloud profiles were similar, apart from thesignificant difference in the total number of utterances in the two conditions.Percentages were also calculated for the combined categories of specific traits for the dyadicand individual groups for each of the three text passages. A chi-square calculation for the categorypercentages for the three dyadic sets revealed no significant differences (ct=.05). A similarM/C N/S P/M Totalchi-square calculation for the three individual sets was significant (cx=.05). The percentage profile111graphs for the three sets of individual protocols are given in Figures 11, 12, and 13.co0,C’.,M/C N/S P/M Total10080000Q Average#I :: Average%0CategoryFigure 1 1 Specific Traits--Average Number and Average Percent--Passage X Individual10080060V 40 D Average#Average/a4:200M/C N/S P/M TotalCategoryFigure 1 2 Specific Traits--Average Number and Average Percent--Passage Y Individual112100806000040a)a)> 20N N-4-r.TotalM/CD Average#Average%.. 1FZ_N/S P/MCategoryFigure 1 3 Specific Traits—Average Number and Average Percent--Passage Z IndividualThe most noticeable difference in the passage profiles is between the percentage for M/C andthe percentages for the other two categories for Passage X in the individual condition (Figure 11).Although the total number of specific traits is lower than for the other two passages (cf. Figures 12 and13), the traits are largely concentrated in the Meaning-Construction category. Since Passage X wasthe least difficult, it may be that the students could give more of their attention to the construction ofmeaning, rather than to other elements of the process. It may also be that when text is easy to read,students read in a more mature manner (i.e., with more of the behaviours attributed to proficientreaders). If the metacognitive load drops in such a situation, the processes may become automatic tothe reader and go unnoticed and unreported (Afflerbach & Johnston, 1984).Holistic scoring of the think-aloud protocols was carried out by four raters using the Likert-typescale described in Chapter Three (see sample form, Appendix G). For each protocol, the four scoresfor each pair of descriptors were averaged. Overall averages were calculated for the seven pairs ofdescriptors common to both the dyadic and individual think alouds. A chi-square test performed onthese overall average scores for the two sets (dyadic and individual) revealed no significant difference.113Appendices M and N give data on the holistic rating of dyadic and individual protocols,respectively. For the dyadic protocols, the average rating for seven items was 3.3. As stated above,the scale was a six-point, Likert-type scale (values, 1 to 6). The actual range of scores could be 5--thatis, 1 to 6. The average rating (3.3) was, therefore, slightly below the simple mean (3.5). Five of the 28dyadic protocols (18%) were rated at least five out of six. Six (21%) were rated less than two on thesix-point scale. For the 55 individual protocols, on the other hand, the average rating for the sameseven items was 3.1, again below the simple mean. Four protocols (7%) were rated at least five out ofsix, while 16 (29%) were rated less than two on the six-point scale. From the foregoing, it is evidentthat dyadic protocols were rated higher on the seven common pairs of descriptors, both on averageand at the extremes of the rating scale.Holistic averages for students whose MCMCA scores placed them in the individual and dyadicextreme groups as described below (see Question Six) were tabled and graphed. Holistic averages ofall eight pairs of descriptors for the low- and high-extreme dyadic groups were similar: for thelow-extreme dyadic group, 3.46, and for the high-extreme dyadic group, 3.60. On the other hand,holistic averages of the seven pairs of descriptors for the low- and high-extreme individual groupsappeared to differ considerably: for the low-extreme individual group, 1.88, and for the high extremeindividual group, 3.56 (see Figures 14 and 15). However, a chi-square test did not reveal thedifference to be significant (calculated value was 0.518; critical values were 98•1 u-S and 5.02; =1;ct=.05). This lack of significance will be examined in Chapter Five.Miscue analysis was carried out on the 55 protocols produced in the individual condition. Asreported in Chapter Three, the opening portion of each passage was used for the analysis, in whichonly serious miscues that changed the meaning of the text were counted. Three reading levels wereassigned: easy, where the serious miscues fell below 2%; instructional, where the miscues rangedfrom 2% and 5%; and frustrational level, where miscues exceeded 5% (see Chapter Three).:4o.C)LZZZEEZZZL2 Ave/Holistic- >< 0LI) LI)‘->.Studenta)3)C)4-,U)0I—ø z U cYz < sCI) CI) CI) r LI) LI) CI) )——— >. N — ‘- a)— >< N><>< ><>. N>Student• Ave/Holistic114U- < >CI) CI) t>-Figure 1 4 Holistic Averages--Extreme Group Individual 1 St65430Figure 1 5 Holistic Averages--Extreme Group Individual 4th115For the 55 protocols produced in the individual condition, miscue analysis gave the followingresults: Passage X--1 5 easy reading level, 3 instructional level, and 1 frustrational level; Passage Y--5easy, 5 instructional, and 9 frustrational level; Passage Z--4 easy, 5 instructional, and 8 frustrationallevel. Overall, students found the passages easy in 24 of the 55 cases; they found the passages tobe at their level of instruction in 13 additional cases; and they found the passages to be at theirfrustrational level in the remaining 18 cases.A Spearman rank correlation test was performed using the miscue percentages and theMCMCA scores obtained after reading in the individual condition. For a=.05 (two-sided), the criticalvalue was =1 .96. The calculated value was Z=- 4.21, indicating that the two sets were correlated witheach other; that is, there was a significant negative correlation between students’ individual MCMCAscores and level of serious miscues made while reading the passages. The comparison of individualMCMCA scores and percent of reading miscues is given in Appendix 0.Findings. Pertaining to Question Five, the findings were:8. Students produced significantly more analyzable utterances when working in dyads thanwhen working individually;9. More utterances (70%) were for the purpose of constructing meaning than for otherpurposes;10. Students were consistent in their approach to the think-aloud task whether workingdyadically or individually;11. When passage difficulty increased, the average percent of meaning-constructionutterances in the individual condition dropped significantly (a=.05);12. Holistic scoring revealed large but non-significant differences between the think-aloudprotocols of students who scored at the low and high extremes on the multiple choice meaningclarification activities in the individual condition; and13. Students’ scores on the individual MCMCAs showed a significant negative correlation with116the level of serious miscues made while reading the passages.iii. Major Question (c)How are understanding of text and understanding of metaphor related?f. Question SixWhat are the features of the think alouds of students whose per[ormance on a multiple choicemeaning clarification activity placed them in the first and fourth guartiles?The dyadic and individual MCMCA scores were ranked for the purpose of determining thescores that fell within the first and fourth quartiles of each set. The identification of exact quartilesproved difficult because in certain cases the boundary fell within same-rank groups that could not beseparated according to any meaningful criterion. As a result, clearly definable extreme groups wereestablished instead of quartiles, with low scores being designated as Group 1 and high scores asGroup 4. The extreme groups established by this method had the following numbers of members:Extreme Group Dyadic 1st (EGD1)--12; Extreme Group Dyadic 4th (EGD4)--14; Extreme GroupIndividual 1st (EGI1)--13; and Extreme Group Individual 4th (EGI4)--13.Comparative data for the extreme groups are given in Table 7. Using the data from Table 7 andthe grand mean (6.365) for all scores of the four extreme groups, the sum of squares between groupsand the sum of squares within groups were calculated and used in the calculation of an E-test for thetwo population variances. Results are given in Table 8. The critical value of E3,48 (.05) was 2.81 (byinterpolation). Since the calculated value was E =160.607, the variance in MCMCA scores betweenthe groups was taken to be significantly greater than the variance within the groups themselves,indicating that the individual groups had internal homogeneity but were significantly different fromeach other.To identify which of the groups differed significantly from the others, the data from Table 7 wereused in the calculation of the harmonic means for the pairs of groups, which in turn were used toperiorm the Tukey procedure (the HSD test--honestly significant difference test). Of the four relevant117Table 7Extreme Groups--MCMCA StatisticsEGD1 EGD4 EGI1 EGI4n 12 14 13 13total score (MCMCA) 36 120 54 121mean 3.00 8.571 4.154 9.308s2 1.455 0.571 0.975 0.231S 1 .206 0.756 0.987 0.481EGD1 -- Extreme Group Dyadic 1st; EGD4 -- Extreme Group Dyadic 4th; EGI1 -- Extreme GroupIndividual 1st; EGI4 -- Extreme Group Individual 4th.pairs, three showed significant differences--namely, EGD1 and EGI1, EGD1 and EGD4, and EGI1 andEGI4. That is, both low-extreme groups differed from each other, and each differed from thehigh-extreme group in the same condition. The fourth pairing, EGD4 and EGI4, showed no significantdifference--that is, there was no detectable significant difference between the students who scoredhighest in the dyadic condition and highest in the individual condition. This lack of differencebetween EGD4 and EGI4 is in contrast to the difference noted between students in the low-extremegroups and adds a new element to the finding of the I-test described above (Question Two) --that is,that students’ individual and dyadic scores differed significantly, indicating that it was the studentswho performed least well who also performed significantly differently in the two conditions. Thesimilarities between EGD4 and EGI4 will be examined in more detail below.The specific traits of the protocols of the four extreme groups were examined. Duplication waseliminated for those cases in which both dyadic partners had MCMCA scores in the same extremegroup; that is, the protocol count was not repeated for these dyads. In EGD1, two dyads wererepresented by both members. In EGD4, three dyads were so represented. As a result of eliminationof duplicates, these particular extreme groups contributed 10 and 11 sets of specific traits,118Table 8Summary of Results--F-test for Population Variances Between and Within Extreme GroupsSource SS df lyE FBetween 380.156 3 126.719 160.607*Within 37.895 48 0.789*R<05respectively, rather than 12 and 1 4--the number of individual students who contributed MCMCAscores. The average number of specific traits for the extreme groups, the mean number for the entireset, and the median number are shown in Figure 16.200U)4JI—C).4-C)C,,‘4-04-.Ez4-ID>0CiEGD1 EGD4 EGI1 EGI4 Mean MedianFigure 1 6 Specific Traits--Extreme Groups—Average, Mean, and Median119The combined categories of specific traits of the protocols for the students in the four extremegroups were aggregated and profiled. The aggregate summaries of combined specific-trait counts bycategory for the extreme groups are given in Appendix P. The aggregate summaries of percentagesof specific traits in combined categories for all extreme groups are given in Appendix Q. Figure 17shows the average number of specific traits in each combined category and the average total for thefour extreme groups. Figure 18 presents the average percent of combined categories for theextreme groups.200150- 100 EGD1EGD4a) EGI1flEGI4< 500CategoryFigure 1 7 Specific Traits--Average Number--Extreme GroupsAn examination of the specific traits constituting the combined categories for the four extremegroups revealed considerable similarity across the groups. Table 9 provides the pertinent data. Whenthe specific traits that designate agreement with a previous statement were removed, the remainingnumbers of specific traits were almost identical. This indicated that students were consistent in theirbasic approach to the task, in that they brought to bear the same repertoire of strategies even though,as indicated above (Question Five), individual readers produced a lower percentage ofM/C N/S P/M Total120meaning-construction utterances when the passage difficulty increased. An interesting differencewas between the number of types of specific traits within meaning-construction for the two individualextreme groups, with the lower group (EGI1) showing a wider range of basic specific traits--i 8--thanthe higher group (EGI4)--15.1008060EGD1EGD4ci) QEGI13) AnEJEGI4ci)>200M/C N/S P/MCategoryFigure 1 8 Specific Traits--Average Percent--Extreme GroupsAs Figure 17 shows, the low-extreme individual group (EGI1) produced fewest analyzableutterances. From this, one might surmise that, for students who performed lowest on tasks such asthis, there was a tendency to retreat from talk, possibly because of uncertainty about their own abilityto use talk effectively, or because it might reveal in them some inability to complete the tasksuccessfully, or perhaps because they were not convinced that talk had utility in this situation.The high-extreme individual group (EGI4), on the other hand, produced the highestpercentage of meaning-construction utterances (see Figure 18). This might be taken as an indicationof ability to focus on the task at hand, since the directions were to make meaning with the text passageand to talk about the meaning that was being made. It is known that as certain reading strategies,121Table 9Number of Types of Specific Traits by Combined Category and Extreme GroupExtreme Group Meaning Construction Neutral States Pre-Meaning TotalEGI4 18 4 6 28[15+3(]a [5+1(a)]EGI1 24 5 5 34[18 +6(a)] [4+1 (]EGD4 33 6 10 49[18 +15(a)] [4 +2(] [5 +5(]EGD1 29 6 9 44[17 +12(a)] [4+2(] [5÷4(]aBracketed values consist of, first, the basic number of types of specific traits and, second, followedby (a), the number of types of specific traits showing agreement with a previous utterance made byself or dyadic partner.including metacognitive strategies, are mastered by the reader, they drop below the threshold ofconsciousness and are performed automatically (Afflerbach & Johnston, 1984). In such a case, itwould not be surprising that the range of specific traits narrowed toward those likely to be mosteffective in demonstrating understanding.A closer examination of the protocols of these two extreme groups revealed that they varied inother ways than number and percent of meaning-construction specific traits. They differed also in thecomplexity of the meaning-construction process; specifically, EGI4 students were likely to userelatively sophisticated meaning-construction processes and to link specific traits in a chain ofcomment, whereas EGI1 students were more likely to use shorter, less complex utterances,frequently limited to a single specific trait.With respect to these differences, EGI4 students might raise a question about meaning not yet122made clear by the text. In protocol Xl F3C, for example, the student said, “I wonder how long?” afterreading that the brown bat had a long life span. Such questions indicating active involvement with thetext were rare in EGI1 protocols. The same student exhibited a combination of inferencing and theconscious use of a strategy to clarify meaning when, after reading that the female bat had entered anabandoned barn, she said, “So she goes to have her baby at the top of a barn, or maybe she’s justsleeping. Let’s find out.” She then read on to verify one or the other of her inferences. The samestudent also showed an awareness of her own knowledge and change in that knowledge when shecommented, “So, their mothers do leave them. I never thought they did.” Awareness of one’s ownknowledge was also exhibited in protocol Zi M3K, where the student concluded the think aloud withthe comment, “But I learned a few things, like what feudalism was and everything.” The same studenthad made his prior knowledge explicit at the beginning of the protocol when, considering the title andcover of the text, he said, “Well, from the title, it looks like it would be sort of knights and everything inthe story. It looks interesting. The Middle Ages is around the 1 lOOs, I think. That’s what...I think forsure it is. I’m not positive, but....” Such awareness of one’s own knowledge and change in it wasrelatively rare in the protocols of EGI1.Certain other features were exhibited fairly frequently by EGI4 protocols but rarely by those ofEGI1. Protocol X1M4N demonstrated an evaluation of the significance of information given when,after reading that the bat had flown nearly a hundred miles in two nights, the student commented,“That is ver.. .obviously very fast animals for something their size.” An evaluation of the logicalimplication of information given in text was shown in protocol Xl M5N, where the student concluded,“So, in three weeks they.. .they’re. .they must have a big growth spurt,” after reading about thechanges in the baby bats. The process of evaluation extended to consideration of the text, its style,perhaps, or its effectiveness. In protocol Xl F4Q, for example, the student, upon reading themeaning of the scientific name for the brown bat (translated as “mouse-eared” and “light-fleeing”),commented, “That’s a good way to put it!” In protocol Z1 M5B, the student evaluated theeffectiveness of text when he commented, “Chieftains.., right there’s a pretty interesting word,” then123went on to relate it to a more common word by suggesting, “I gather we get chief from [it],” asupposition that may well be correct since the word “chief” came into use in English a full century after“chieftain” (Webster’s Ninth New Collegiate Dictionary, 1986). Such complex evaluations of text andmeaning were lacking in the EGI1 protocols.A difference between the extreme groups was noticeable in the amount of predicting andverifying that was exhibited. In protocol Xl F4K (EGI4) for example, the student commented, “Okay,it’s what Ifigured,” having verified a prediction. In protocol Y1F5P (EGI1), the student made a strongopening set of predictions in response to the title. She said,Well, I think the Frontier Home will be about. ..uh.. .cabin maybe back in the woods. ..talkingabout the wilderness in the olden days, back when pioneers first came. Uh. . . it might beabout.. .the...the woods. ..the way they survived. It may be...uh.. it might be about Indians. No, Ithink it’ll be about the cabins that they stayed in. Yeah! That’s what I think it’s about. Urn. .111read it now.Despite the strong beginning, she failed either to verify the predictions or to use them to guide herreading. This student produced a lengthy protocol of 91 specific traits, of which 45% were MeaningConstruction and 45% were Neutral States, indicating that as many of her think-aloud utterances weredevoted to the demands of the task as to making meaning with the text. While this protocol was nottypical of the low-extreme group protocols, the failure to make connections among various parts of thetext was common.A student in EGI1 might misread a word and persist with the incorrect word, as in protocolYl F4V, where the student misread “villages” as “villas” and saw no incongruity with the resultingstatement that pioneers established “villas” in the “thick woodlands of Ohio and Kentucky.” In EGI4,on the other hand, a student was likely to check his or her understanding and prior knowledge, as inprotocol Xl M4N (EGI4), for example, where a complex evaluation of understanding took place whenthe reader, after encountering the scientific name for the brown bat, said, “Light-fleeing’. ..that’s. ..can understand that. Obviously, it doesn’t like light. Most bats don’t. And the scientific definitions areall in Latin. I knew that. So I understand that.”In EGI4 the monitoring of understanding might lead to a suggestion of a specific reading124strategy. In protocol X1M5N, the student interrupted his reading in mid-sentence to rephrase whathad just been read, “Okay.. so.. .they use their wings to cradle the baby.” He began to read again,only to discover that the meaning was not clear. As a result, he decided, “I’ll read that again,” thuschoosing a reading strategy commonly employed when understanding has broken down. In likemanner, in protocol Zi M3K (EGI4), the student suggested a specific reading strategy to clarify a point.In reference to the fall of the Roman Empire, he said, “Uh, I don’t know when exactly that was, but itmight tell if I read some more.” This student also showed an expectation that the text would providenecessary information--that is, that it would be complete. In EGI4 protocols, such fix-up strategieswere more common than in EGI1 protocols.The importance of prior knowledge was demonstrated by the protocols of both extreme groups.In EGI4, students routinely connected text with prior knowledge, sometimes identifying the source oftheir knowledge. In protocolZlF3J (EGI4), after reading that oaths were made in feudal society to aperson granting land, the student stated, “I know what an oath is because we studied that inclass.. .Oath of Allegiance. So it...it’s a...an oath is kind of a promise.” However, in a somewhat similarsituation in protocol Z1F4F (EGI1), the student concluded, “And this here tells us that Rome is inAmerica, I think,” but failed to make use of the available maps to clarify the misconception.The two extreme groups showed differences in the ways they accounted for context or failed toconsider it. In protocol Zi M5B (EGI4), the student, when reading that Rome withdrew its troops fromBritain in 442, contextualized the information first by identifying the numeral as a date and thenrecognizing that it was far in the past. He commented, “Hm...geez that’s a long time ago.” Students inEGI1 did not give context the same degree of attention. For example, in protocol Yl M40 (EGI1), afterreading the phrase, “deep woods abounding in shadows and mystery” incorrectly as “deep woodsabove in shadows and mysteries,” the student commented, “Mysteries! Hm! There’s mysteries.They are stories by. ..stories about when they.. .when the mystery. ..somebody killed somebody andeverything.” In the same protocol, the student encountered the word “displaced” and, failing toconsider the context in which the word was presented, concluded, “I don’t know what ‘displaced’125means. Kind of animals or something.” Similarly, in protocol Y1M5K (EGI1), the student, unable toread the word “necessities,” concluded that “it sounds like a disease.”Several protocols in both groups exhibited characteristic verbal patterns. In protocol Z1M5B(EGI4), for example, the student used specific patterns to initiate paraphrase of a section, generally(20 of 39 utterances) introducing his remarks with the phrases, “So, it seems like...,” “So, it lookslike...,” “So, it sounds like....” The paraphrases were usually accurate representations of the textmeaning (only six misinterpretations of text occurred). In protocol X1F4X (EGI1), on the other hand,although the think-aloud utterances generally began with the characteristic phrase, “So I know...,” thecompletion often consisted of little more than repetition of words just read, as when, for example, afterhaving read, “This is the story of the Little Brown Bat, one of the most common bats in North America,”the student said, “So, I know it’s one of the most common bats in North America.” Similarly, in protocolY1F3D (EGI1), the student, having read, “the sap could not flow into the branches,” simply restatedthe same sequence of words with a single word changed, “Okay.. .so that means that the sap couldnot run into the branches.”Having an established purpose for reading is widely accepted as an important contributor tosuccessful reading (Baker & Brown, 1984). In this study, the demonstrations of the think-aloudprocess, as well as the instructions to the students in the two practice sessions and the two targetsessions, emphasized that the purpose was to make meaning with what was being read. In a sense,this purpose was external to the passage being read. Additionally, the passages were clearlyinformational, yet to varying degrees had a narrative voice. For some students, these factors mayhave caused difficulty. Some students produced a relatively high number of neutral-states specifictraits, indicating orientation toward the task itself. This was often particularly noticeable with studentsfor whom the reading was at their frustrational level.An example of this is provided by protocol Yl F3D. It had 11.1% of serious miscues (frustrationallevel). Only 14 of its 36 specific traits were for meaning construction, 4 indicated pre-meaningconditions, and 18 were neutral-states utterances. In this protocol, there was a detectable sense of126confusion, and part of that seemed to be over purpose, as indicated by the student’s uncertaintywhen she encountered the extended caption under one of the pictures. This caption consisted of anexcerpt from a pioneer’s diary and was written in the first person, as opposed to the third person of themain text, was in smaller size typeface, was entirely indented from the left margin, and its authorshipwas credited, yet the student included the entire six-sentence excerpt within the main-text sentencespanning the break. In transition from the unfinished sentence to the excerpt, she said, “Anythingthey could be.. .out. . .Father meant us.. .oops.. .Father met us.. .two yoke of oxens and our ox wagon.”For this student, the purpose for reading seemed to be to say the words, although not necessarily allthe words.The retell of this protocol showed a similar lack of focus and a lack of attention to the significantmessages provided by the text. The student said, “I think this is a very good story. I enjoyed it.And. ..um...how they used to travel and stuff. And. let’s see! [inaudible portion]...the people livedtheir life. Okay. I’m done.”The low-extreme group protocols displayed certain features not evident in those of thehigh-extreme group. In protocolYlF5P (EGI1), the student found the reading very difficult. This wasevident in the number of failures to recognize and decode words. Also, the student showedfrustration with the task. At one point she said, “Oh, man! I never want to read another book in mylife!” The same student was disorganized in her approach to the task, often omitting large sections oftext and experiencing difficulty in keeping her place in the text. At one point, she said, “Oh... I think I’mdone that page. I donLt know. Where’s my place? I’ll just start that page ‘cause I don’t know where Iam.”Overall, the retells produced after the initial reading think aloud differed between the twoextreme groups. In each group, two protocols were without retells. The remaining 11 protocols ofEGI4 produced a total of 540 separate specific traits, 177 of which were for the purpose of retell. Theremaining 11 protocols of EGI1 produced 385 separate specific traits, 93 of which were for thepurpose of retell. The percent of specific traits devoted to retell varied between the two groups (33%127in EGI4 and 24% in EGI1), and the average number per protocol in EGI4 exceeded the average inEGI1 by a ratio of almost 2:1 (16.1 to 8.5). The retell was intended to provide an opportunity for thereexamination of the text, resulting perhaps in a statement of the gist of the passage. The strategiesstudents demonstrated in their retells were generally similar to those used during their initial thinkalouds, thus providing additional data for inclusion in the combined categories (see Appendix D).Interestingly, EGI1 protocols exhibited a broader range of specific traits than did EGI4protocols--12 and 9, respectively. For those in EGI4, the average number of different specific traitswas about two and one-half, while in EGI1 the average number of different specific traits was aboutthree and one-half. The reason for this is not entirely clear, but it may be that the proficient readers inEGI4 were simply showing the tendency to leave the more or less automatic processes unreported(Afflerbach & Johnston, 1984).The variety of specific traits used by students in EGI1 had the potential to contribute to meaningmaking; however, because the retells were generally quite short and often unfocused, the potentialwas not realized. For the retell of protocol Yl M4G (EGI1), for example, the student said,Urn.. .Frontier Home. I don’t know why they called it Frontier Home because. ..weIl, actually, theyreally talked about frontier life.. .and...um.. .uh.. .talked about what they did, and the tools, andthe covered wagon, and how much...l guess...not much of the.. .the home, but more of theland.While consisting of eight separate analyzable utterances of three different types (Evaluation of Text,Restatement/Paraphrase, and Other/Miscellaneous), the protocol did not capture the gist of the textpassage.For the retell in protocol Zi M3K (EGI4), on the other hand, the student said,Well, this was interesting. It told about feudalism and how the Roman Empire was.. .and hadmost of Europe in its power at one time and how the...what was it?. ..the Germanic tribes startedpressuring Rome and a new way of life started spreading, called feudalism. And the feudalists,they were usually poor and they.. .um. . .and were pov. ..they.. lived in poverty and theyprotected.. .they.. if they got... urn.. land or power from the king or queen, then they. ..that...they had to take an oath to protect the king or queen. And anybody who didn’t would be atraitor. So, it was interesting. But I learned a few things like what feudalism was and everything,so....This retell consisted of 10 separate analyzable utterances of three different types (Affective128Response, Restatement/Paraphrase, and Affirmation of Meaning). In contrast to the protocol fromEGI1, this protocol captured the gist of the passage to a significant degree.The exceptional cases in the two extreme groups provided interesting contrasts. In EGI4, forexample, protocol Xl M4W, produced by a student who scored 9 out of 10 on the MCMCA, had nocommentary with the initial reading of text and provided a retell consisting of 14Restatement/Paraphrase items that were entirely word-for-word excerpts from the original text. InEGI1, on the other hand, protocol Yl F5P, produced by a student who scored 4 out of 10 on theMCMCA, consisted of an 80-item commentary with the initial reading and a retell of 11 specific traits offive different types (Procedural, Restatement/Paraphrase, Other/Miscellaneous, Denial ofUnderstanding, and Evaluation of Text).For the two protocols under consideration (Xl M4W and Yl F5P), it is helpful to examine thedifferences in the actual reading of text. Protocol Xl M4W had a total of five miscues, only three ofwhich could be taken to indicate failure to recognize a word--”light-fleeling” for “light-fleeing,”“matternity” for “maternity,” and “particle” for “particular.” While the third is clearly a failure to recognizethe word, the second may be the result of attempting to sound out a word that is unfamiliar in print, andthe first may be merely a slip of the tongue. Since the text passage had over 600 words, thepercentage of serious miscues fell below 1%, well within the range of easy reading for this student.Protocol Yl F5P, as was indicated above, was produced by a student who experiencedconsiderable frustration with the task. In the first paragraph, after skipping the first two sentencesentirely, she made so many miscues in the remaining sentence that the portion was meaningless. Herthink-aloud commentary following the first paragraph reflected this lack of meaning. She said,Well, it sounds like they’re looking for...the people...are making [inaudible section]. They...they might.. .uh.. .be.. .uh... Oh, how do you say? They might be... Well, I know for sure that...the...for the cold winter...WeIl, I’ll read some more.The second and subsequent paragraphs were read with as much difficulty as the first, and thecommentary reflected the growing frustration.The same student, when completing the dyadic think aloud in the previous session, had left129most of the reading to her partner or had been coached through the reading of sections by the otherstudent. She had, however, contributed to the discussion of the text as an equal participant and withconsiderable maturity, showing that she possessed the prior knowledge necessary to make meaningwith the new text. As her opening inferences in the individual protocol indicated (see above), shealso possessed considerable prior knowledge about this topic. However, because she could notdecode enough words, she got little if anything from the text to link with her prior knowledge; thus,meaning construction was largely impossible. From the foregoing, one would judge that caution is inorder when attempting to link protocol complexity with MCMCA scores on an individual basis. Whilethe group differences may be strong, individual variation within the groups can be extreme.As reported above, when the harmonic means of the MCMCA scores of the four extremegroups were compared, EGD4 and EGI4, showed no significant difference; that is, there was nodetectable significant difference between the students who scored highest in the dyadic conditionand highest in the individual condition. It is possible that this similarity of groups resulted from a ceilingeffect, produced by either student reading ability or prior knowledge relating to the particularmetaphors used in this study. If the passages had been even more difficult, then, the groups mighthave shown a significant difference.An examination of these two groups revealed that 4 students appeared in both groups and 19appeared in one only (see Appendix R). Of particular interest was that for these two extreme groups(27 cases), Passage X was read 12 times, Passage Y was read 4 times, and Passage Z was read 11times. Passage X was one of the two passages read by each of the four students who appeared inboth sets. These occurrences of the passages are in the reverse order of the passage difficulty (seeChapter Three), suggesting that there was a relationship between passage difficulty and performanceby these students on the multiple choice activities, a relationship already noted for the whole group.The data on the four students who appeared in both high-extreme groups were tabulated forexamination (see Table 10). In the individual condition, student M4N produced an average number ofspecific traits (46 to the average of 44 for all individual protocols--see Figure 10). The other three130Table 10Performance Summary of Students Appearing in Both High-Extreme Groups (EGI4 and EGD4)uderM4N M5N F5l M3Klndii Dyad lndv Dyad lndiv Dyad mdiv DyadPassage X Z X Y Y X Z XMCMCAScore 9 8 10 8 9 9 10 10Specific Traits #M/C 36 51 58 158 51 151 74 153N/S 9 11 6 50 8 49 9 67P/M 1 16 1 8 21 6 7 16Total 46 78 65 216 80 206 90 236Specific Traits %M/C 78 65 89 73 64 73 82 65N/S 20 14 9 23 10 24 10 28P/M 2 21 2 4 26 3 8 7Holistic Scoreb 3.6 3.4 4.6 5.1 3.7 5.1 5.3 4.9Miscue%C 0.4 1.5 2.5 1.2Note. Code to Specific Traits: M!C--Meaning-Construction Processes, N/S--Neutral States,P/M--Pre-Meaning Conditions.aRounded to nearest whole number.bBased on the seven pairs of descriptors common to both individual and dyadic scales. Rounded tothe nearest tenth.cMiscue analysis not performed on dyadic protocols. Rounded to the nearest tenth.131students (viz., M5N, F51, and M3K) each produced a higher number of specific traits in the individualcondition. In the dyadic condition, student M4N and partner produced 78 specific traits, compared tothe average of 118 for all dyadic protocols (see Figure 9). Each of the dyadic protocols of the otherthree students had over 200 specific traits. The holistic scores were all above their respective groupaverage--3.1 for all individual protocols; 3.3 on the same seven items for all dyadic protocols. Three ofthe four individual protocols had less than 2% of serious miscues, indicating easy reading, and thefourth had 2.5%, indicating an instructional reading level that was close to easy. The individualprotocol of student F5l showed a much higher than average number of pre-meaning specific traits--21(26%) compared to 4 (9%) for all individual protocols (again, see Figure 10). Generally, then, apartfrom having MCMCA scores that were higher than average, these students produced more specifictraits than average, had higher holistic scores than average, and produced few serious miscues.The protocols of the two individual extreme groups (EGI1 and EGI4), each consisting of 13students, were compared for reading miscues. The results are summarized in Tables 11 (EGI1) and12 (EGI4). In EGI1, 11 of the 13 students read at the frustrational level, one read at the instructionallevel , and one read at the easy level. Passage X, rated easiest (Grade 6, Flesch-Kincaid), appearedonly once and was at the frustrational reading level for the student. Passage Y, rated hardest (Grade11), appeared nine times, with seven of those being at the frustrational reading level. Passage Z,rated at the Grade 8 level, appeared three times, all at the frustrational level.The results given in Table 12 are strikingly different. Passage X, rated easiest, appeared sixtimes, Passage Y, rated hardest, appeared twice, and Passage Z appeared five times. Eight of the 13readings were at the easy level and the remaining five were at the instructional level. Only one of thereadings approached the frustrational level, with a miscue rate of 4.7%.It is clear, then, that, as a group, students who scored lowest on the MCMCAs after readingindividually (EGI1) produced a higher percentage of miscues on their protocols than did students whoscored highest on the MCMCAs (EGI4).132Table 11Low-Extreme Individual Group (EGI1) Miscue Analysis Percent and Reading LevelPassagex Y zProtocol % Level % Level % LevelXl F4X 13.4 FrustrationalY1F3A 1.8 EasyYl F3D 1 1.1 FrustrationalY1M3F 8.9 FrustrationalY1M4G 2.1 InstructionalYl M40 14.3 FrustrationalYl M4T 6.1 FrustrationalYl F4V 10.7 FrustrationalYl M5K 10.7 FrustrationalYl F5P xa FrustrationalZ1 F4F 12.1 FrustrationalZ1 F41 24.6 FrustrationalZ1 F4R 5.1 FrustrationalaToo many miscues to count.133Table 12Hicih-Extreme Individual Group (EGI4) Miscue Analysis Percent and Reading LevelPassageX Y zProtocol % Level % Level % LevelX1F3C 3.0 InstructionalXl F4K 1.9 EasyXl M4N 0.4 EasyXl F4Q 0.4 EasyX1M4W 0.7 EasyX1M5N 1.5 EasyY1M5A 1.1 EasyYl F51 2.5 InstructionalZ1M3G 1.2 EasyZ1 F31 3.1 InstructionalZi F3J 4.7 InstructionalZ1M3K 1.2 EasyZi M5B 2.0 Instructional134Findings. Pertaining to Question Six, the findings were:14. With the exception of the two high-extreme groups (as determined by MCMCA scores), theunderstanding of metaphor by students in the extreme groups varied significantly;15. Students in the low-extreme individual group produced fewest analyzable utterances;16. Students in the high-extreme individual group produced the highest proportion ofmeaning-construction utterances; and17. Students who scored highest on the MCMCAs in the individual condition made fewermiscues when reading than did students who scored lowest on the MCMCAs in the individualcondition.g. Question SevenWhat is the relationship between the level of students’ understanding of metaphor and theirunderstanding of text?Analysis of students’ understanding of metaphor as measured by scores on the multiple choicemeaning clarification activities revealed an overall adult-like understanding of the 30 metaphors ofabout 65%, with significantly higher understanding in the individual condition than in the dyadiccondition (see Questions One and Two, above).Analysis of specific traits revealed that students produced more total utterances in the dyadiccondition than in the individual condition. Overall, more utterances (70%) were of themeaning-construction type than of other types (see Question Five, above).Although the difference was not statistically significant, the average dyadic holistic score washigher than the individual holistic score, when compared on the same seven pairs of descriptors--3.3and 3.1, respectively (see Question Five).Dyadic holistic average scores (all eight pairs of descriptors) and MCMCA scores achieved afterreading in the dyadic condition were graphed to show the average holistic score for students whoscored at each MCMCA value. With the single exception that students who achieved a score of 10135had a holistic average score one full point higher on the six-point scale than students with any otherMCMCA score, no significant pattern could be observed. See Figure 19.65() N. N.0) 4C’) C”) C”)>3 - -00(I)o Holistic Ave2 -0I01 2 3 4 5 6 7 8 9 10MCMCA ScoreFigure 1 9 Holistic Score Average by Dyadic MCMCA ScoreA similar graph was produced to show the relationship between MCMCA scores achieved afterreading in the dyadic condition and the average number of meaning-construction utterancesdetected in the dyadic think-aloud protocols by specific-trait analysis. The results are given in Figure20. Again, no significant overall pattern could be detected; however, as with the holistic scoringaverage, students who achieved a score of 10 on the MCMCA also produced a higher averagenumber of meaning-construction utterances in their think alouds.Individual holistic average scores (seven pairs of descriptors) and MCMCA scores achieved afterreading in the individual condition were graphed to show the average holistic score for students whoscored at each MCMCA value. There was a noticeable pattern, with students who scored from 6 to 10on the MCMCA achieving average holistic scores at least a full point above those who scored belowsix. See Figure 21.136S 50Ca)C)(ta)>4a)0C-)(I,C)4—,Cl)0I200 -C-flC-’? —EEzzzzLziiEzEzEEIEjjjEjjj:iiiinniM/C AverageHolistic Ave0-1 2 3 4 5 6 7 8 9 10MCMCA ScoreFigure 20 Meaning Construction Average by Dyadic MCMCA Score6543202 3 4 5 6 7 8 9 10MCMCA ScoreFigure 21 Holistic Score Average by Individual MCMCA Score137The relationship was examined between MCMCA scores achieved after reading in the individualcondition and the average number of meaning-construction utterances in the individual think-aloudprotocols. Results are given in Figure 22. As with the individual holistic averages, the individualaverages of meaning-construction utterances were higher for students whose MCMCA scores werefrom 6 to 10. Again, students who scored 10 on the MCMCA had the highest average number ofmeaning-construction utterances, although only fractionally higher than students with a score of 9,6050 -N. 1-3)co40 - -------- ...(, c,-7-30 —2 ....•.•..• /••..... ..N . —......—0) N •.. ./•,E20 ;:i }_ 0M/C Average100MCMCA ScoreFigure 22 Meaning Construction Average by Individual MCMCA ScoreAs reported above, the dyadic protocols and MCMCA scores did not show the same patterns asthe individual protocols and MCMCA scores. In a small number of cases both members of the dyadscored in the same extreme group (specifically, two dyads in EGD1 and three dyads in EGD4). Sincemost dyads, then, had members contributing MCMCA scores to different segments of the overall set,it is not surprising that their protocols tended to smooth out, rather than to separate into clearlydefined groups, as was the case with individuals. It would appear that the sharing process at work inthe dyadic think alouds masked individual differences, at least in those features measured by the138holistic rating scale and those identified by specific-trait analysis. This is supported by the averagescores for the eighth pair of descriptors on the holistic-rating scale, measuring the balance betweendyadic partners. As Figure 23 indicates, partners contributed relatively equally to the think alouds,averaging 5.1 out of 6 on the balance item, a score that is considerably higher than the average scoresfor the other pairs of descriptors.6-z//4.2- Q Averagec5 -. E- 2 -L - - oI— C)0C-)DescriptorFigure 23 Holistic Average by Descriptor--Dyadic ProtocolsFindings. Pertaining to Question Seven, the findings were:18. It was difficult to identify links between MCMCA scores and holistic and specific-traits scoresfor dyads;19. Students working in the individual condition were more likely to separate into groups withinwhich MCMCA scores and holistic scoring, as well as number of meaning-construction utterances,bore a positive relationship; and20. According to the criteria established for this study, when working individually, students with139a higher understanding of the target metaphors demonstrated a higher level of understanding of thetext passages in which the metaphors were presented.Ill. SummaryIn this chapter, the study data were examined in a variety of ways. Numerical data from themultiple choice activities were examined according to standard statistical practices. Specific-traitanalysis of the think-aloud protocols provided descriptions of the reading processes engaged in bythese students. The quantification of the occurrences of the specific traits provided data for standardstatistical tests and for aggregate and percentage profiles of the protocols. The Liken-type holisticscale provided numerical data for description of the protocols. Miscue analysis was performed on the55 individual think-aloud protocols.The various analyses produced a total of 20 findings in answer to the study’s seven questions.Taken together, these findings provide a picture of the understanding of the text passages and theunderstanding of the target metaphors by the study’s 55 participants. The findings will be discussedin the following chapter.140CHAPTER FIVE: DISCUSSION, CONCLUSIONS, APPLICATION, LIMITATIONS,AND RESEARCH IMPLICATIONSI. Introduction 140II. Discussion 140Ill. Conclusions 162IV. Application 163V. Limitations 167VI. Implications for Future Research 168VII. Concluding Remarks 170I. IntroductionIn this chapter, the findings from Chapter Four will be discussed. Connections with the researchliterature will be identified, and conclusions will be drawn. Application of the findings will be proposed,limitations will be noted, and implications for future research will be given. The chapter will end withbrief concluding remarks.II. DiscussionFindings Pertaining to Question OneFinding 1Overall, students exhibited a 65% level of understanding of the metaphors.Ones initial response to the overall level of understanding of metaphor by these students maybe one of surprise: Why do these students understand only about two-thirds of the metaphors?I think it is instructive to turn the observation around--that is, by grade six these students havealready reached two-thirds adult-level understanding of these metaphors when presented in thesecontexts. Looked at in a certain light, this can be considered a major accomplishment. For example,the texts used in this study were not written to conform to a formula determined beforehand as the141true indication of appropriate grade six level of understanding, whatever that might be. On thecontrary, these texts were authentic--that is, presented as trade books and intended by the authorsfor use by a fairly broad range of readers. There is little textual evidence that the authors andpublishers applied rigid criteria in marrying topic and text features. These were not written accordingto a formula, as is the case with some books, such as the so-called high-interest, low-vocabulary booksand certain textbooks. All of this notwithstanding, these students, as a group, still managed to reachtwo-thirds of the anticipated understanding of adult members of the culture (Lakoff & Johnson, 1980;Lakoff & Turner, 1989; Lakoff, 1993>.Finding 2There were wide variations in the scores for the three nassacies, with averages of 54% onPassage Y, 69% on Passage Z, and 73% on Passage X.Factors other than the metaphors themselves may have influenced the scores. In classroomreading events, teachers routinely take into account such factors as student maturity, intelligence,prior knowledge, overall reading ability, and test-taking ability. The first three, for example, might wellhave a direct influence on the understanding of metaphor itself, while the last two might have a directimpact on task performance.A point to be considered is that the text passages came from books written in other countries(USA and England). One would expect that the metaphors were representative of the stock readilyavailable and familiar to the authors; certainly, the authors could not have chosen the metaphors withthese particular students in mind. That notwithstanding, these students were able to identify theanticipated adult-like meaning a full 65% of the time over the three passages.The question, then, is whether these students had prior knowledge of these particularmetaphorical expressions, as predicted by Lakoff and Johnson (1980), Lakoff and Turner (1989), andLakoff (1993), or whether they were able to apply their understanding of the metaphorical process togeneral conceptual-domain knowledge to construct a meaning on demand. This study was not142designed to examine these particular aspects of meaning making, and further research is needed.This raises the further issue as to what extent an experimental task such as this actually activatesthe psychological mechanisms for making meaning via the metaphor process and to what extentexplicit prior knowledge of the specific metaphors is being tapped. If meaning for the metaphor mustbe created on demand, then the question of context, as proposed by Ortony, Schallert, Reynolds,and Antos (1978) and Siltanen (1989), may well be significant, since the surrounding text may imposeconstraints and suggest possibilities. This may, at least in part, account for the clear inverserelationship between MCMCA scores and passage difficulty (as established by the Flesch-Kincaidreadability formula). In other words, the surrounding text may have lacked the clear contextualfeatures necessary to provide support in activating the domain knowledge critical to the successfulcompletion of the mapping process.Considering the results, one might be tempted to decry the incursion of U.S. or British booksinto Canadian classrooms. On the basis of the material examined here, such a response is notjustified. The crucial issue, it seems, is not where the books are produced or the point of view fromwhich they are written, but rather how they can be used with and by our students. In this regard, itmust be noted that my use of the text passages in this study was narrow; in the classroom of athoughtful teacher, the texts would be used in a more coherent way, a way more likely to seempurposeful to the students.Findings Pertaining to Question TwoFinding 3There was a significant positive correlation between students’ individual and dyadic scores.This is an interesting finding in that it indicates a significant degree of commonality between theindividual and dyadic tasks, even though the passage difficulty changed from one task to the next,increasing for some students and decreasing for others. This suggests that metaphor understandingwas, indeed, an important part of what was being measured. It suggests, also, that although metaphor143understanding may not be a single psychological construct, there is a certain core common to thevarious metaphors.Finding 4Student understanding of metaphor was significantly higher after completing the think aloud inthe individual, as orposed to the dyadic, condition.This finding was a surprise. The higher individual scores suggest that something having to dowith the condition (individual or dyadic) influenced the outcomes. According to theories of socialinteractionism and cooperative learning, the students should have performed at a higher level in thedyadic condition. Below, I show that the direction of change in scores is generally consistent withexpectations based on passage difficulty (see also the discussion of Finding 11, below). This has theeffect, however, of suggesting that passage difficulty is a more powerful force than the sharing of atask, a conclusion that is hard to accept in light of the scaffolding effect associated with cooperativelearning, regardless to the relative abilities of the participants (Johnson & Johnson, 1994).One must ask whether something in the design of the study may have influenced theoutcomes--for example, the effect of practice, the effect of order, the effect of past experience withdyadic procedures, or the effect of individual preference for a particular type of learning task.Were the students perhaps unfamiliar with collaborative learning? It does not seem possible,since these students were from classrooms in which the whole-language, social-interactionistapproach was commonly practiced. The students were, for the most part, aware of themselves andothers as learners. They were accustomed to flexible grouping. With the exception of a very fewstudents, they appeared to be sensitive to the feelings of each other and presented to each other in asupportive manner.One possible explanation is that, in interacting with one another, students’ attention wasdiverted from the text. This seems unsupported by the quality of the think-aloud protocols. Whilethere are generally many more analyzable utterances in the dyadic protocols and a moderate shift in144focus toward specific traits within the combined category of Neutral States, there is no significantdifference in the relative percentages of analyzable ufferances in the three combined categoriescompared to the individual protocols.Another possible explanation is that the shift of focus from working with a partner to workingalone--that is, from dyadic reading to individual answering of the multiple choice activity--in some waypresented a conflict that the students did not resolve in the time provided. There is nothing in thedata to shed light on this.Yet another possibility was examined. Since dyadic think alouds had been done before theindividual think alouds (see Pilot Study, Procedures, Chapter Three), it was possible that theadditional practice might have affected the outcome in the following session. However, the studentshad practiced dyadic and individual think-aloud procedures in the first two sessions and, for the mostpart, seemed comfortable with the process, especially working with a partner. As well, one mightexpect the additional practice to affect the think-aloud process itself, resulting in more ease with theprocess and more of those features that mark successful think alouds. As noted, students werehighly consistent in their approach, whether working alone or with a partner, except for the amount oftalk per protocol, which was lower in the individual condition. In other words, there was nothing in theindividual think alouds, taken as a group, that would point to a higher level of understanding on theMCMCAs. Indeed, with regard to the think-aloud process, it was the opinion of a small number ofstudents that talking to oneself--or to a tape recorder--was an unusual thing to do, but no studentexpressed any surprise at being asked to talk to another student about what was being read. On thisbasis, if an influence of condition were to be found, one would expect it to be in favour of the dyadiccondition.As indicated above, it is possible that the results were complicated by the differences in thereading levels of the three passages. To take a narrow example (and one limited by the limitations ofreadability formulas), the Flesch-Kincaid Grade Level Formula produced a grade level score forPassage X of 6, for Passage Y, 11, and for Passage Z, 8. If the data from Table 3 (see Chapter Four,145Question Two) is examined for direction of change of scores, relating the rise or fall in each case to thechange in the grade levels of the passages, the results are largely predictable, as shown in Table 13.As can be seen from this table, only 7 of the 46 non-tie pairs of scores did not match direction ofchange consistent with passage difficulty level. A chi-square calculation using the data of Table 13,with df=1 and a=.05, gave a value of 0.249, within the critical values of 98•1 and 5.02, indicatingthat the change in direction of scores is within the expected range; that is, it is indicated that thedirection of change of scores in understanding of the metaphors is consistent with a difference inreading difficulty between each pair of passages compared.Again, the reason for the effect is not clear--that is, understanding of metaphor may drop whenpassage difficulty increases, but why? Is it because the context is such a significant determining factorfor the understanding of metaphor? The literature on context effect is not helpful here, focusingmainly on length of immediate context (see, for example, Ortony, Schallert, Reynolds, andAntos,1978, and Siltanen, 1989). Is it because, as the passage difficulty increases, reader attention isdrawn to the difficult parts in an effort to decode them? If such is the case, is less attention given tothose parts that do not immediately force themselves upon the reader’s consciousness as requiringparticular attention, as may be true of metaphor, especially if it is metaphor that has enoughcommonness that the reader may be aware of having heard it before and may be less concernedabout processing it to a deeper level than the clearly unknown surrounding text. This, of course, ishypothetical and needs examination beyond the level possible in this study.Whatever the specific dynamics enacted when reading difficulty increases, one couldreasonably expect the same factors to affect the results of authentic classroom reading tasks involvingthe understanding of metaphor.It might have been interesting to correlate MCMCA scores with data on student readingachievement and ability, as measured by recognized standardized tests, since the results would haveprovided yet another way to look at student performance in this study. However, recent scores werenot available on a student-by-student basis. While one would be inclined to predict that more-capable146Table 13Summary of Data from Table 3, Changes in MCMCA Scores According to Level of Passage Difficulty1st. Score 2nd. Score TcaIHigher(Expecteda) 15 24 39Higher (Not Expecteda) 2 5 7Total 17 29 46baBased on passage difficulty.bDoes not include 9 cases in which first and second scores were identical.readers would score higher on the multiple choice activity--and miscue analysis does provideprovisional support for this view--this study does not give conclusive evidence.Finding Pertaining to Question ThreeFinding 5There was no significant difference in understanding of metaphor for the same passagebetween the dyadic and individual conditions.Given the random assignment of students and passages, this finding is as unexpected asFinding 4. Social interactionism predicts higher scores in the dyadic condition (see, for example, theresearch support for cooperative learning in Johnson & Johnson, 1994); however, at the singlepassage level, students performed equally well in the individual condition. This suggests that whenpassage difficulty remains constant, condition (dyadic or individual) is unimportant. Taken togetherwith Finding 4, this is perplexing, since condition was found to have an (unanticipated) effect withchange to a less or a more difficult passage.147Findings Pertaining to Question FourFinding 6Student understanding of metaphors of the copula-verb (eciuation) syntactic-frame type wassknificantly lower than of the descriptive-phrase syntactic-frame type.This was the only metaphor-type effect identified in this study. Interestingly, it made nodifference whether the comparisons involved incompatible- or disparate-type source/targetcombinations (i.e., whether the source and target came from disparate categories, sharing physical orfunctional features, or from incompatible categories, between which there is conventionally no sharedfeature [Broderick, 1992, p. 187]).These findings suggest that for these students--and for certain metaphors--the significantelement in metaphor construction was not what was compared but how it was compared--that is, howthe metaphor was structured. This seems to suggest that the linguistic features of metaphor are animportant factor in their understanding. On the other hand, it may be seen from the opposite point ofview--that is, that the metaphor process remains intact but linguistic complexity reducesunderstanding. This need not be taken to support the idea that metaphor is primarily a linguisticdevice, as older views suggest, but simply that many and diverse factors affect themeaning-construction process. This would be consistent with the view of metaphor as mapping fromconceptual domain to conceptual domain.Finding 7The Lakoff, Johnson, and Turner method of analyzing metaphors did not isolate differences toaccount for the varying performance scores on the MCMCAs.The use of authentic text, as in this study, introduced metaphors of such diversity andcomplexity (refer to Appendix L) that analysis according to the ideas of Lakoff, Johnson, and Turnerwas not possible. In an experiment using specially designed metaphors, however, it should bepossible to control both the design and the complexity of the metaphors so as to make such analysis148possible.During the examination of Lakoff, Johnson, and Turner’s ideas of metaphor as a conceptualmapping (Lakoff & Johnson, 1980; Lakoff & Turner, 1989; Lakoff, 1993), two potentially relatedfactors were considered. The first was the degree of abstractness of the words involved in themetaphorical expression. This was felt to offer possible explanation for difficulty in understanding onthe premise that the more abstract the source and target words, the more complex would be themapping procedure. The second factor was the degree of conventionality or commonness of themetaphorical expression or major parts of it.With regard to abstractness, the eight items constituting the high-extreme set (items X-2, X-5,X-6, X-9, Z-1, Z-6, Z-7, and Z-9) and the eight items constituting the low-extreme set (items X-4, Y-1,Y-5, Y-7, Y-9, Y-10, Z-3, and Z-4) were examined for abstractness and concreteness (see Table 14).The degree of abstractness is very high throughout. Of the high set, only one of the 18 key words orphrases is concrete. In the low set, of the 20 key words, three are concrete. The degree ofabstractness does not seem to be a determining factor at the item level. Item Z-9, for example, in thehigh group (with 81% adult-like response), contains four abstract key words--more than any of theother 15 items. On the other hand, item Y-9, containing one abstract and one concrete key word, wasunderstood by only 32% of the students--the lowest percentage of all items.The second factor--degree of conventionality or commonness--appears to offer greaterexplanation of the differences between the extreme sets. Among the items in the high-extreme set,for example, common words, expressions, and concepts, such as “latches on,” “growing up,”“basket,” “fall,” “binding,” and “maternity,” contribute greatly to the structure of the metaphors.Among the items in the low-extreme set, on the other hand, less-common words, such as“abounding,” “fleeing,” “girdled,” “based,” and “protections,” are key words in several of themetaphors, and, in some cases, otherwise familiar words, such as “walks,” “life,” “comforts,”“interruption,” and “shook,” are used in conjunction with less familiar key words or in less familiar ways.149Table 14Abstractness of Metarthor Key Words/Phrases--High- and Low-Extreme Items by MCMCA ScoreExtreme Item Word/Phrase ObservationHigh X-2 sky abstractnight abstractdeepens abstractx5a maternity abstractcolony abstractX-6 forms (verb) abstractbasket concreteX9a latches on abstractZ-1 power abstractcrumbles (verb> abstractZ-6 fall abstractRoman Empire abstractz7a society abstractgrowing up abstractZ-9 oath abstractloyalty abstractbinding abstractlife abstractLow X-4 light abstractfleeing abstractY-1 woods concreteabounding abstract150(Table 14 continued)Extreme Item Word/Phrase Observationshadows abstractmystery abstracty.5b walks (noun) abstractlife abstracty7b civilization abstractcomforts (noun) abstractprotections abstractygb interruption abstracttrees concreteY-10 girdled abstract1-3 town concretebased abstractsociety abstractZ-4 raids abstractshook abstractpower abstractaHighest three items.bLowest three items.The impact of degree of conventionality--that is, the familiarity one would expect members ofour culture to possess--is consistent with the ideas of Lakoff and Johnson (1980), Lakoff and Turner(1989), and Lakoff (1993), as explained in Chapter Two. What is lacking, however, is a means to151measure the degree of conventionality. It is one matter to identify phrases (metaphorical andotherwise) that are conventional among adult members of a subculture, or possibly even in the largerculture, but it is much more difficult to predict the likelihood that children will have acquired the termsand conventionalized them at the personal level. This helps to explain why, for example, themetaphorical phrase “latches on” was understood well (87%) but “walks of life” showed a relatively lowlevel of understanding (41%). For an adult member of the subculture, both expressions may beentirely conventionalized; for a child, on the other hand, the expressions may not yet have reachedeven early stages of familiarization. It is unknown, for example, the degree to which the child’simmediate culture, including the home, provides linguistic practice with metaphorical expressions,leading to their conventionalization.Unknown, too, is the extent to which students’ past reading may have contributed to theprocess of conventionalizing the metaphors. One would expect that an avid reader would have hadmore opportunity to experience the conventional metaphors than a hesitant reader and to becomefamiliar with their use in a variety of texts. This would be consistent with the idea that there is a culturalbasis for metaphor, as proposed by Lakoff and Johnson (1980), Lakoff and Turner (1989), and Lakoff(1993).Findings Pertaining to Question FiveFinding 8Students produced significantly more analyzable utterances when working in dyads than whenworking individually.If, as social interactionist theory maintains, talk can function as a mediator between text andmeaning (Wells, 1991), then this finding should indicate an enhanced meaning-construction situationin the dyadic condition. This makes Finding 4 even more difficult to explain.From Figures 9 and 10 (see Chapter Four), it can be seen that the relative percentages ofspecific traits in the three combined categories in the dyadic and individual conditions vary. For152example, the percent of meaning-construction utterances is about 65% of all utterances in the dyadiccondition (Figure 9), but rises to about 76% in the individual condition. The neutral-states utterances(marking orientation toward task) account for about 15% of the utterances in the individual condition,but rise to about 27% in the dyadic condition. It seems reasonable to expect work with a partner toencourage the kind of task orientation talk found in the protocols of these students; indeed, onewould be surprised if there were little such talk. Even with this shift, the dyadic protocols averagedover twice as many meaning-construction utterances as individual protocols.All things being equal, then, the dyadic think aloud provides the opportunity for students toengage in high-quality talk, the sort Wells (1991) suggests will mediate learning. The challenge is toensure that other factors do not interfere with the mediational power of the talk.Finding 9More utterances (70%) were for the purpose of constructing meaning than for other purposes.This finding (see Figure 8, Chapter Four) confirms the value of the talk engaged in by thesestudents, giving additional weight to the remarks made in Finding 8. The think-aloud demonstrations,the practice sessions, and the directions given during each target session all emphasized makingmeaning with the text. This finding suggests that the students followed the directions well.Given the richness of the think alouds of these students, it was possible to examine not only themeanings they constructed, but in many cases to follow the processes the students engaged in asthey went about constructing meaning. This study thus supports the findings of others (e.g.,Afflerbach & Johnston, 1984; Ericsson & Simon, 1993). Clearly, the think aloud is a useful procedurefor the type of study undertaken here. It may well have considerable potential for application in theclassroom as a teaching tool and diagnostic device.Finding 10Students were consistent in their approach to the think-aloud task whether working dyadically or153individually.This finding suggests that the think aloud should be applicable in a variety of classroomsituations where the examination of text is called for, and, further, that the researcher or teachershould be confident that what is being revealed is dependable.This consistency may have resulted in part from the demonstration and practice of theprocedure. If this is so, it suggests that, with demonstration and frequent reinforcement by theclassroom teacher, the students should become quite proficient with the process and, as a result,much more aware of their own thinking processes and outcomes.On the other hand, there is a naturalness about the think-aloud process, especially when itinvolves students working together. This can be reassuring to the researcher or teacher who wishesto make use of the think aloud.Finding 11When passage difficulty increased, the average percent of meaning-construction utterances inthe individual condition dropped significantly (a=.05).As indicated above (Chapter Three and Finding 4), the three passages differed in readingdifficulty, with Passage X being least difficult and Passage Y being most difficult, based on theFlesch-Kincaid reading grade level formula. When the shift in meaning-construction utterances isexamined, it appears that as the text difficulty increases, the meaning-construction process suffers,and more noticeably when the student is reading individually, since the support offered by a partner islacking. This is consistent with reported observations that as reading difficulty increases, students relyon different strategies and exhibit different concepts of what constitutes the reading process (see, forexample, Baker & Brown 1984). This suggests that the easier the passage, the more the reader isable to concentrate on meaning construction and, further, that a capable reader is able to maintainfocus on meaning construction over a broader range of text difficulty.Mention of a few features of the three text passages will make the matter of varying difficulty154clearer (see also Appendix 5, Passage Statistics). Passage Y (Frontier Home, for example, focusingon a topic with the potential to be of considerable interest to young people (i.e., difficulties settlersfaced in the colonization of a frontier, in this case the American West), distanced itself through ahigh-demand vocabulary (e.g., abounding, clusters, ingenuity, interspersed, persisted, phenomenal,versatile, withered), long sentences and difficult sentence structure (e.g., sentences had an averagelength of 21.2 words, with the longest sentence having 36 words and only two sentences havingfewer than 12 words; six sentences were written in the passive voice; and sentence structure wasoften complex--e.g., “Jolted by every rock and root, they rode along for days or weeks, tree branchesswirling over their heads”; “Yet with opportunity came phenomenal difficulties, including threats fromthe weather, disease, wild animals, and sometimes from the Indians who were being displaced”), andpotentially unjustified assumptions regarding student prior knowledge of U.S. geography, history,and culture (e.g., of the Appalachian region, the Conestoga wagon, the states of Indiana, Kentucky,and Mississippi). The Dale-Chall Formula found 16.5% of the words to be unfamiliar, which issomewhat higher than Passage X, but lower than Passage Z.Passage Z (The Middle Acies) also presented textual difficulties, although less extreme thanPassage Y. It had a large number of words with three or more syllables (15.8%). It had fairlyspecialized vocabulary (e.g., chieftain, dominated, fealty, feudalism, Germanic, Goths, homage,Lombards, Syria, Vandals, villas). Interestingly, the word “society” caused readers considerabledifficulty. The passage had a few sentences with initial adverbial phrases and clauses that would makethe reading more difficult for some students (e.g., “At the beginning of the first century, the RomanEmpire dominated...”; “Although Roman customs continued for a while, a new way of runningsociety...”). For the most part, however, the sentences were in natural order, compensating for thefew more difficult cases, The Dale-Chall Formula found 19.5% of the words to be unfamiliar, highest ofthe three passages.Passage X (Shadows of Night: The Hidden World of the Little Brown Bat) had fewest difficulttext features. Seventy percent of the words had one syllable. Only 6% of the words had three or155more syllables. The sentences averaged 13.9 words in length, with 18 sentences having fewer than12 words and only 3 having more than 24 words. There was one Latin term, Myotis lucifugus, but itwas explained immediately. While a number of sentences began with adverbial phrases or clauses,these elements were generally fairly short. They were often simple expressions of time (e.g., “Atdawn...”; “Two months later...”; “During the first day...”) and were easily related to the following text.Most sentences, however, were in natural order, with the subject easily identified. The Dale-ChallFormula found 12.0% of the words to be unfamiliar.While any single feature of the sort mentioned here may not be enough to increase significantlythe difficulty of the reading task, it is likely that several such features in combination do change thenature of the reading task and, therefore, the meaning constructed with the text. This has implicationsfor research examining the meaning-construction process and for the classroom teacher faced withchoosing material for student use. In this study, the varying difficulty of the texts added an interestingadditional factor, but it did not interfere with the central aim, which was to determine level ofunderstanding of metaphor in authentic text. If the aim had been otherwise--to examine thepsychological processes involved in understanding metaphors, for example--the varying difficulty ofthe passages may well have complicated analysis significantly.Finding 12Holistic scoring revealed large but non-significant differences between the think-aloudprotocols of students who scored at the low and high extremes on the multiple choice meaningclarification activities in the individual condition.Although the differences did not reach the level of statistical significance, they reflected a cleardifference in quality to the raters (see Figures 14 and 15, Chapter Four). This perceived discrepancyis, perhaps, a reflection of the nature of the Liken-type scale used for the holistic rating; that is, thelowest value obtainable was one, and this designated a decidedly negative assessment of theprotocol (see Appendix G). If the scale were to run from zero to five, for example, the magnitude of156the differences would be increased, and the differences might well reach the level of statisticalsignificance.The usefulness of a measure like the holistic rating scale employed here perhaps needs to bereevaluated. Its relative ease of use and its more-or-less instant numbers’ make it attractive; yet, it isopen to effects of subjectivity on the part of the raters, especially when tiredness induced by lengthyrating sessions is involved.On the other hand, it could be countered that much of the assessment of student work is of justsuch a nature, where the teacher is judging quality against a set of descriptors not likely to be greatlydifferent from the descriptors used on this scale.Perhaps the solution is to use the results for exactly what they are--descriptive indicators ofidentifiable qualities--and avoid attaching too much weight to the numbers themselves.Finding 13Students’ scores on the individual MCMCAs showed a significant negative correlation with thelevel of serious miscues made while reading the passages.There has been considerable disagreement as to what constitutes instructional, andfrustrational levels of reading, based on miscue analysis, and whether the concepts are evenappropriate (see, for example, Lowell, 1970). That notwithstanding, it seemed that miscue analysiscould reveal enough about students’ text-handling skills to permit comparisons. I believe this isconfirmed by the data presented in Appendix 0.The method of counting only serious miscues may not find favour with all educators, some ofwhom may, for example, prefer to count all substitutions--meaningful as well as meaningless--asserious. In a sense, then, there is a degree of arbitrariness about such decisions as whether to assignlevels or where to set the boundaries if levels are assigned. That having been said, then, one mustask whether the parameters were adhered to and whether anything of interest was found. In the caseof the present study, I believe the inverse rank correlation between the MCMCA scores and level of157serious miscues (those that changed the meaning of the text) provides one more piece of evidencethat the students who demonstrated higher meaning-construction skills also understood themetaphors better than those students whose meaning-construction skills were weaker. As can beseen from Appendix 0, of the top 23 students in the individual condition (those scoring 8, 9, or 10 onthe MCMCA), only one read below the instructional level, and then only marginally. On the otherhand, among the bottom 20 (those who scored 2,3,4,5, or 6), 13 students read at the frustrationallevel.This is not to be construed as an endorsement for miscue analysis, per se. Rather, it is offeredwith the belief that where pooled information agrees, there may be something worth examining. Inthis case, the miscue analysis results are consistent with the results of both specific-trait scoring andholistic scoring.Findings Pertaining to Question SixFinding 14With the exception of the two high-extreme groups (as determined by MCMCA scores), theunderstanding of metaphor by students in the extreme groups varied significantly.The significant differences between the two low-extreme groups (EGI1 and EGD1) and theirrespective high-extreme groups (EGI4 and EGD4) are not surprising. It is perhaps not surprising thatthe two high-extreme groups did not vary significantly, since four students appeared in both groups,and there appeared to be a ceiling effect, with the top scores grouped at 9 and 10.It is surprising, however, that the two low-extreme groups differed significantly, given that sevenstudents--over half of each group--appeared in both groups. The statistical difference between thetwo groups is accounted for by the higher individual scores, but why the lowest performing studentsshould perform at a higher level when working alone, rather than with a partner, is unclear. It may bethat working with a partner is, for some reason, a distraction for these students. Whatever the cause,in this study, it was the very students who should have benefited most from the intended scaffolding158offered by dyadic work (Vygotsky, 1962, 1978; Wells, 1991) who, in fact, appear to have benefitedleast. Clearly, further examination of this effect is in order, given the widespread acceptance ofsocial-interaction ist theory.Although it was intended to avoid examining the relative performance of boys and girls (since todo so would require a study devoted solely to the one issue), the makeup of the groups of studentscommon to the two low-extreme groups and the two high-extreme groups was striking and surprising.Of the four students who appeared in both high-extreme groups, three were boys, while of the sevenstudents who appeared in both low-extreme groups, six were girls. I have no basis upon which tospeculate as to the cause of this phenomenon, but it would seem to merit further investigation.Finding 15Students in the low-extreme individual group produced fewest analyzable utterances.Taken in conjunction with Findings 8 and 14, this finding suggests that, for these students, talkabout the learning task may not be a familiar or comfortable strategy and may act as a distractor, ratherthan a scaffold. On the other hand, it may be that students with limited resources may have difficultyallocating those resources to more than one aspect of the task at the same time, that they become,essentially, single-strategy learners.Should the foregoing be true, then it would seem that part of classroom strategy instructionshould focus on how to bring to bear more than one strategy or skill at a time and how to do itconsciously. This may be a place where more teacher modeling of metacognitive strategies andspecific reading strategies is called for. And it may be, also, that the think aloud can play a significantpart in this process, for demonstration and practice and as an ongoing learning strategy. Despite thisstudy’s surprising results with dyadic think alouds, the literature on cooperative learning is substantialenough to suggest that, if properly used, dyadic work enhances learning (see, for example, Johnson& Johnson, 1994).In relation to the three findings (8, 14, and 15), it would seem that three issues may need further159examination: first, the nature and extent of the cognitive load of the think-aloud process forless-capable readers; second, whether the metacognitive demands are different for a reading thinkaloud than for a silent reading of text; and, third, whether the combined demands of engaging in thetwo processes simultaneously overload the less-capable student.Finding 16Students in the high-extreme individual group produced the highest proportion ofmeaning-construction utterances.This finding indicates that these students were very focused on the meaning-making process.Two interpretations of this are possible. Either the students were performing according to theinstructions they had been given--that is, to make meaning and to say what that meaning was--or theywere under-reporting their thought processes. One would expect these particular students to beproficient in following directions. Additionally, since these students were likely fairly mature readers,judging from the way they handled text and the sophistication of some of their reading processes, itmay be that some of the processes had become so automatic as to drop below ordinaryconsciousness (Anderson, 1980, and Ericsson & Simon, 1980, cited in Afflerbach & Johnston,1984). Specifically, for example, the on-the-go decisions about the task may have been sowell-established in the skills repertoire as to be spontaneous and, to the students, unnoteworthy.A second issue is somewhat more complex, that is, the degree to which admission of lack ofknowledge or acknowledgement that the meaning-making process has not been successful (theconditions identified by the Pre-meaning Conditions [P/M] group of specific traits) contributes,ultimately, to successful meaning making. In other words, when we do not know, is it important that webe aware of our lack of understanding? The answer to this question is of interest, not only to oneinvolved in the type of research reported in this study, but also to the teacher involved in readinginstruction.In this study, the P!M specific traits were not considered positive contributors to the1 60meaning-making process, although quantity of talk (i.e., regardless of purpose) was seen as anindication of engagement with task. There is a sense in which this approach entails a logicalinconsistency; however, for descriptive purposes, this approach seemed necessary.Finding 17Students who scored highest on the MCMCAs in the individual condition made fewer miscueswhen reading than did students who scored lowest on the MCMCAs in the individual condition.This finding is consistent with the overall tendency discussed under Finding 13 above. In thiscase, however, the differences are large (see Tables 11 and 12, Chapter Four), providing one morepiece of evidence of a link between metaphor understanding and text understanding, a link that is ofparticular interest in this study.Findings Pertaining to Question SevenFinding 18It was difficult to identify links between MCMCA scores and holistic and srecific-traits scores fordyads.Because of the random assignment of students to dyads, it was common for members to varyconsiderably in their MCMCA scores. Of the 10 dyads contributing to the low-extreme group (EGD1),only two had both members placing in the group--that is, eight members of EGI1 had partners whoscored outside the group. Similarly, of the 11 dyads contributing to the high-extreme group (EGD4),three had both members placing in the group--that is, eight members of EGD4 had partners whoscored outside the group. As a result, any description of the dyadic think-aloud protocols applied tomore than one quartile-like group, except in a minority of cases, therefore limiting the potential forinterpretation.Of interest in this regard, however, is the power of the dyad to mask certain aspects of theperformance of weaker partners. In some protocols, for example, where there was a clear difference in161reading ability between the partners, there was some evidence of a tendency for the more capablereader to compensate for the weaker partner by reading more, with the weaker reader still contributingas an equal partner in the discussion portion of the think aloud.On one hand, this makes analysis difficult; on the other hand, however, it suggests that thedyadic think aloud has the potential to benefit weaker readers by making them co-contributors to themeaning-making process.Finding 19Students working in the individual condition were more likely to separate into groups withinwhich MCMCA scores and holistic scoring, as well as number of meaning-construction utterances,bore a positive relationship.Although dyadic MCMCA scores did not relate clearly to specific-trait and holistic scores, theindividual MCMCA scores did so (see again Figures 21 and 22, Chapter Four). For the individualMCMCAs, the boundary between the upper and lower groups for both holistic scores and averagenumber of meaning-construction utterances was an MCMCA score of six. Interestingly, thiscorresponds closely to the grand mean for all MCMCA scores (6.52). This forms two fairly distinctgroups: the first, those who scored six or above on the individual MCMCA, who had a holistic averagescore above three, and who had an average number of meaning-construction utterances of 30 ormore; the second, all those who failed to reach these criteria.There were, of course, individual variations, and this profile cannot be taken as indicative of theperformance of any particular individual, nor should one of the values be taken as a predictor of eitherof the others. Nonetheless, this grouping extends beyond the extreme groups and gives anadditional glimpse of the overall picture of the relationship between MCMCA scores and think-aloudperformance.162Finding 20According to the criteria established for this study, when working individually, students with ahigher understanding of the target metaphors demonstrated a higher level of understanding of thetext passages in which the metaphors were presented.This finding should contribute to future research into children’s understanding of metaphors ininformational text by shifting the starting point to a higher level of assumptions than was possible forthis study. In future studies, it will be possible to focus on the dynamics of the single issue ofunderstanding of metaphor rather than to have a dual focus, as was the case here.It must be noted that this finding does not imply a causal relationship of any sort; it doessuggest that, on average, students who show more understanding of the metaphors also show moreunderstanding of the text. Similarly, it should not be concluded that a student who shows a higherunderstanding of the metaphors in a text will necessarily understand the text at a correspondinglyhigh level.Ill. ConclusionsBased on the findings examined in Chapter Four and above, a number of concepts, comprisingthe conclusions of this study, have been identified. They are as follows:1. Students understood about two-thirds of the metaphors;2. Student understanding was significantly higher in the individual condition compared to thedyadic condition;3. Overall text difficulty affected students’ ability to construct meaning for metaphors;4. The syntactic-frame structure of the metaphor affected students’ ability to constructmeaning;5. Abstractness of the words in the metaphor did not affect student ability to constructmeaning, but degree of conventionality of the metaphorical expression itself did have an effect;6. There was a positive correlation between understanding of text and understanding of the163metaphors embedded in the text;7. The think aloud revealed something of both the constructed meaning and the process ofmeaning construction; and8. Students who could not perform well on the multiple choice activities could still participate asequal partners in the dyadic think alouds.IV. ApplicationThe eight conclusions are considered to have implications for classroom practice.With regard to Conclusion 1, because there was a substantial percentage of metaphors notunderstood, and because other students may have more-or-less similar levels of understanding,teachers should find it worthwhile to give attention to metaphor in their teaching. This attention couldinvolve sensitizing students to metaphor, using the ideas of Lakoff, Johnson, and Turner to show theubiquity of metaphor. It could also involve drawing attention to specific examples of metaphor duringreading activities and the use of specific metaphors in teaching, not only to raise linguistic awareness,but to make prior knowledge explicit and to enhance learning of specific information. This would beconsistent with the approach to metaphor use in the classroom advocated by Pugh, Hicks, Davis, andVenstra (1992).With regard to Conclusion 2, when teachers use paired reading activities, they need to be awarethat careful monitoring is necessary to ensure that the anticipated outcomes are, in fact, beingachieved. Part of this process should be the careful enunciation of the purpose of the reading event.This study found indication that some students were uncertain as to their personal purpose forreading and were, as a result, tentative in their approach to the task. This is consistent with Baker andBrowns (1984> contention that a clearly established purpose for reading enhances the quality of thereading. The observations made in the analysis of the think-aloud protocols in this study can be takenas support for the position that teachers should aid and encourage students to set clear purposes forreading.1 64It stands to reason, as well, that a clearly defined purpose helps establish which cognitiveprocesses will need to be monitored during the reading and, in some situations, which metacognitivestrategies should be given particular attention. Given adequate modeling of the variousmetacognitive strategies by the teacher, and given practice of the sort that can be monitored by theteacher, by having the students engage in think alouds, for example, achievement should beenhanced.With regard to Conclusion 3, this study found that the level of students’ serious readingmiscues was correlated with understanding of metaphors in the text passage. Teachers need to beaware that miscue analysis can be a rough indicator of student understanding of metaphor, eitherpotentially or as the outcome of the reading of a particular text. Conversely, a simple multiple choicetest on metaphors in the passage may provide useful additional information about studentunderstanding, as judged from miscue analysis.This study found that students who performed at the high extreme individually and those whoperformed at the high extreme dyadically were likely to do equally well on the types of activities usedto measure understanding. For the most proficient readers, teachers will need to make their decisionto assign group or individual work on criteria other than simple test performance.This study found that students in the low-extreme group, in terms of understanding of targetmetaphors, were likely to produce less talk in their individual think alouds than students in thehigh-extreme group. It may be possible for teachers to use the reading think aloud as a roughindicator of student understanding of text without the necessity of engaging in overly detailedanalysis.This study found also that students in the high-extreme individual group were likely toconcentrate a higher percentage of their talk on meaning-construction processes than were studentsin other extreme groups. Teachers may wish to use the think-aloud skills of such students as modelsin their teaching of metacognitive skills to other students. Additionally, teachers need to ensure thattext is of the appropriate level of reading difficulty for the students using it in order to encourage1 65students to concentrate more of their attention to the process of meaning construction.This study also found that students understood more metaphors in passages that were easy toread than in those passages that were at their frustrational level. This is consistent with the findings ofDeans study (cited in May, 1994, p. 410) that understanding breaks down when reading becomestoo difficult. Teachers need to be aware that the potential benefit of metaphor-rich language may belost if students are not given reading material at their instructional or easy-reading level.With regard to Conclusion 4, this study found that these particular students experiencedsignificantly more difficulty with metaphors of the copula-verb type than those of thedescriptive-phrase type. This is consistent with the view held by Broderick (1992) that studentsencounter fewer cases of copula-verb metaphors in their reading and may not be as experienced withthem as with other kinds. Teachers need to be aware, then, that students may be understandingdescriptive-phrase type metaphors but experiencing difficulty making meaning for copula-verb typemetaphors.With regard to Conclusion 5, the importance of the conventionality of the metaphoricalexpressions underscores the need for rich, authentic language experiences in which students havethe opportunity to use language purposefully. It also underscores the need for the teacher to ensurethat the students become aware of the language environment created by their particular oralsubculture, since it is that which will, in large part, contribute to the conventionalization of the formsand expressions that the students need in order to become mature users of the language. Also,since the underlying metaphors are part of the broader culture of which the subculture is a part, whenstudents gain use of the language and thinking of the underlying metaphors, they become in factmembers of the wider culture. The work of Lakoff, Johnson, and Turner can provide teachers with astarting point into the process, and the ideas of Pugh, Hicks, Davis, and Venstra (1992) can providepractical guidance.This study showed that, at least in these particular metaphorical expressions, students were notaffected by the presence of abstract terms, nor were they aided by the presence of concrete terms.166This suggests that there is no reason for the teacher to be overly concerned about students’ ability touse and understand abstract terms, although this generalization must be respected as such, sinceindividual students will vary greatly in this, as in all things.With regard to Conclusion 6, teachers can make use of this relationship in their assessment ofstudents’ growth in understanding. By probing student understanding of metaphors during readingactivities, the teacher can form a general idea of the depth of understanding of the text, sincemetaphorical language in a text is a powerful element of it. Clearly, if the metaphors are notunderstood, part of the meaning is lost. It follows, then, that by helping students strengthen theirunderstanding of metaphor, teachers will contribute to their overall understanding of text. Part of thisprocess should involve making the process explicit through example, discussion, and generation andexamination of metaphors.With regard to Conclusion 7, teachers have traditionally listened to students read and listenedto their answers to questions, but one suspects that teachers spend little time examining students’talk for the purpose of detecting something about the process of thinking. This study has confirmedthe findings of others (e.g., Afflerbach & Johnston, 1984; Dias, 1985; Baumann, Jones, andSeifert-Kessel, 1993; Baumann, Seifert-Kessel, and Jones, 1992; Ericsson & Simon, 1993), that thethink aloud is a powerful assessment and learning device. For the teacher busily engaged in themyriad events of the classroom, the recorded think aloud, without the need for transcription, canprovide a source of informative data for examination outside class time. Based on the evidence of thisstudy, one would judge that the think aloud could take the place of some of the other, sometimesoverused, forms of assessment, especially since the understandings reached through think-aloudanalysis are so rich.The specific traits used in this study (see Appendix D), derived as they were from actual studenttalk, should be useful to the classroom teacher. Although the list is fairly extensive, certain of thespecific traits occur in student talk more often than others, and, with moderate practice, one can learnto identify them without undue effort. It seems, then, but a short and natural step to teach167identification of at least some of the specific traits to students, thereby giving them powerful, newcognitive and metacognitive strategies, as suggested by Baumann, Jones, & Seifert-Kessell (1993).With regard to Conclusion 8, the best dyadic think alouds were dynamic exchanges betweenpartners. The dyadic think aloud appears to be a procedure that encourages students to negotiateroles and to share responsibility for successful completion of the task. In at least these two ways, thedyadic think aloud conforms to the criteria for successful cooperative learning activities, as enunciatedby Johnson and Johnson (1994). Based on the results of this study, it seems to matter little whetherpartners differ widely in reading ability, for the dyadic think aloud encourages them to contribute to thetalk, thus becoming equal participants in the process. With appropriate teacher modeling and studentpractice, together with teacher monitoring and feedback, the dyadic think aloud has the potential tobecome a powerful learning device, with the added advantage of giving students a sense ofaccomplishment and belonging.V. LimitationsThis study was undertaken to examine how well grade six students understand metaphor in theinformational text they read, how well the same children understand the informational text itself, andhow understanding of text and understanding of metaphor are related. With these as the foci of thestudy, other potentially interesting questions had to be set aside. It was not possible to investigatechildren’s understanding of metaphor in fiction or poetry nor to explore psychological aspects ofmetaphor or text understanding.This study was limited to grade six students. A comparison with students at different gradelevels would be informative. This study’s results can be taken as only possible indications of metaphorand text understanding of younger and older children.This study was also limited to children of one small geographical area and one relativelyhomogeneous linguistic and cultural subgroup. Because culture is considered to be such a powerfulinfluence on metaphor acquisition (Lakoff & Johnson, 1980; Lakoff & Turner, 1989; and Lakoff,1 681993), one should exercise caution in applying the findings of this study to students of other or morediverse cultural backgrounds.This study attempted to maintain the ecological validity of the material and activity and tomaintain a closer contact with the curriculum than would have been possible if instruments andmethods had been borrowed from previous studies. The test-like situation was not entirely avoided,however, and the ecological validity was weakened to the extent that the texts were not connected totopics being studied in the classrooms and students worked in small groups outside their classroomsunder the direction of a stranger.The contexts in which metaphors were presented were similar to the original texts; however, thecomplete books were not given to the students and numbered fluorescent dots were affixed atspecific points in each text, thus slightly altering the appearance of the passages. The targetpassages were of different levels of reading difficulty, according to standard readability formulas;consequently, certain analyses involved a somewhat lower than would have been the case had allpassages been at the same reading level.Although such an examination is needed, this study did not attempt to examine the potentialproblems metaphor presents to English-as-a-second-language (ESL) learners.VI. Implications for Future ResearchThis study has raised several issues that merit further research. One has to do with researchmethodology, others have to do with children’s meaning-making process and with metaphor.Although this study used authentic text and attempted to make the data-gathering process asmuch as possible like an ordinary classroom reading event, the students were taken from theirclassrooms in small groups to work with an unfamiliar person for a purpose set entirely by that person.A less-intrusive study design is needed. An adaptation of the current study, using a modified design,could make an important contribution to metaphor research.This study provided information about the metaphor understanding of one group of students at1 69the grade six level. Using an adaptation of the current study design, further research involving bothyounger and older students would be useful in revealing something of the developmental aspect ofmetaphor understanding as it relates to the reading of authentic informational text.Given the cultural influence on both basic conceptual metaphors and specific metaphoricalexpressions (Lakoff & Johnson, 1980; Lakoff & Turner, 1989; Lakoff, 1993), research into theunderstanding of metaphor by non-native English speakers needs to be undertaken.The study found that these particular students perlormed significantly better on the multiplechoice activities after reading individually than after reading with a partner. At the outset, it wasexpected that should there be a significant difference, it would be in the opposite direction. Althoughthe reason for the difference was partially explained by the difference in difficulty of the texts, thematter was not resolved and needs further study.The study found that passage difficulty had a negative influence on understanding ofmetaphor. Although one would have been surprised to have found otherwise, it sheds no light onexactly what text features are implicated when metaphor understanding is reduced. Further, eventhough certain factors have been suggested by this study, there is no indication of how the factorswork or interact. Research is needed on both of these problems.The study found that the syntactic frame of the metaphors could have an influence onunderstanding, with a significant difference between copula-verb type and descriptive-phrase typemetaphors. Broderick (1992) has suggested that children encounter relatively few metaphors of thefirst type in their reading. This may be enough to explain the difference in levels of understanding ofthe types, but there is no conclusive evidence that this is so, and the matter would benefit from furtherresearch.The study found that the two high-extreme groups performed equally well. A ceiling effect mayhave been involved. This raises the question of what would happen to the groups if passage difficultyincreased even further: Would the groups finally perform significantly differently? The question isalso raised as to what can be learned about the nature of these students’ understanding of metaphor.170These questions need further study.This study suggested that the degree of conventionality, or commonness, of the specificmetaphor may be a factor in determining student understanding. Very little is known at present aboutthe effect of conventionality or the related topic of conventionality of the underlying metaphors, ofwhich the specific metaphors are examples. Research is needed in both of these areas.This study did not find a way to use the metaphor-as-mapping view as a predictive tool. Thestudy did suggest, however, that Lakoff, Johnson, and Turner’s view of metaphor as mapping may beuseful as a descriptive and diagnostic tool, Further examination of their ideas and further research areneeded if such a tool is to be developed.VII. Concluding RemarksAs I examined views of metaphor and the specific examples used in this study, I was struck bythe variety--of individual examples, certainly, but also of linguistic structure (as shown by Broderick)and conceptual structure (as shown by Lakoff, Johnson, and Turner). To say, as the older views ofmetaphor have, that it is simply a process of looking for preexisting similarity and comparing one thingto another, is, it seems to me, saying too little about what really happens when a metaphor is createdor understood--a point made strongly by Glucksberg and Keysar (1990). Further, it seems that toreduce metaphor analysis to semantic-feature analysis and comparison, as is sometimes done, fails totake into account essential features of metaphors and the metaphor process--features such as thesubtle transfer of meaning and structure from one conceptual domain to another, the affective imprintof the individual, the peculiar moulding of the metaphors and ways of thinking by the social subcultureto which the individual belongs, the linguistic customs and conventions of the wider language group,and the impact of the great cultural metaphors that provide the context in which social interaction takesplace and define the values and processes of the culture itself.At first, I was surprised and dismayed that Lakoff, Johnson, and Turner’s metaphor-as-mappingview did not yield a workable method to discriminate between the metaphors understood poorly and171those understood well by these students. Upon reflection, however, I realized that the richness ofthe mapping view precludes its reduction to a simple set of rules and procedures. As shown above(see Chapter Four, Question Four, and refer to Appendix L), it produces a wealth of information andleaves one with the sense that the possibilities for mapping among the elements and the possibilityfor prior experience to push one or another element to the fore during the meaning-making processare beyond simple explanation. As noted above, all this tends to make the mapping view less usefulas a predictive tool but of potential value as a descriptive tool and possibly as a diagnostic tool.If the multi-faceted promise of the metaphor-as-mapping view is realized through futureresearch, our knowledge of children’s understanding of metaphor will benefit greatly, and metaphorresearch itself will offer exciting new possibilities.172ReferencesAfflerbach, P., & Johnston, P. (1984). On the use of verbal reports in reading research. Journal ofReading Behavior, 16(4), 307-322.Anderson, R. C., & Pearson, P. D. (1984). A schema-theoretic view of basic processes in readingcomprehension. In P. D. Pearson (Ed.), Handbook of reading research (pp. 255-291). NewYork: Longman.Baker, L., & Brown, A. L. (1984). Cognitive monitoring in reading. In J. Flood (Ed.), Understandingreading comprehension (pp. 21-44). Newark, DE: International Reading Association.Bartlett, F. C. (1932). Remembering. Cambridge: University Press.Bash, B. (1993). Shadows of night. San Francisco, CA: Sierra Club Publications.Baumann, J. F., Jones, L. A., & Seifert-Kessell, N. (1993). Using think alouds to enhance children’scomprehension monitoring abilities. The Reading Teacher, 47(3), 184-1 93.Baumann, J. F., Seifert-Kessell, N., & Jones, L. A. (1992). Effect of think-aloud instruction onelementary students’ comprehension monitoring abilities. Journal of Reading Behavior, 24(2),143-1 72.Beck, I. L., & McKeown, M. G. (1992). Young students’ social studies learning: Going for depth. In M.S. Dreher &W. H. Slater (Eds.), Elementary school literacy: Critical issues (pp. 133-1 56).Norwood, NJ: Christopher-Gordon.Bial, R. (1993). Frontier home. New York: Houghton Mifflin.Black, M. (1962). Models and metanhors: Studies in language and philosophy. Ithaca, NY: CornellUniversity Press.Block, E. (1986). The comprehension strategies of second language learners. TESOL Quarterly,20(3), 463-494.Booth, J., Phenix, J., & Swartz, L. (Eds.). (1988). Impressions. Toronto, Ont.: Holt, Rinehart andWinston.Booth, J., Booth, D., Phenix, J., & Swartz, L. (Eds.). (1987). All over the world. Toronto, Ont.: Holt,Rinehart & Winston.Booth, J., Booth, D., Phenix, J., & Swartz, L. (Eds.). (1987). Knock at the door. Toronto, Ont.: Holt,Rinehart & Winston.Booth, J., Booth, D., Phenix, J., & Swartz, L. (Eds.). (1987). Wherever you are. Toronto, Ont.: Holt,Rinehart & Winston.Broderick, V. (1991). Young children’s comprehension of similarities underlying metaphor. Journalof Psycholinguistic Research, 20(2), 65-81.Broderick, V. (1992). Incidence of verbal comparisons in beginners’ books and in metaphorcomprehension research: A search for ecological validity. Journal of Child Language, 19,183-1 93.173Brown, A. L. (1980). Metacognitive development and reading. In P. J. Spiro, B.C. Bruce, &W. F.Brewer (Eds.), Theoretical issues in reading comprehension (pp. 453-481). Hillsdale, NJ:Lawrence Erlbaum Associates.Bruner, J. (1986). Actual minds, possible worlds. Cambridge, MA: Harvard University Press.Bruner, J. (1990). Acts of meaninci. Cambridge, MA: Harvard University Press.Cazden, C. B. (1988). Classroom discourse: The lanciuage of teaching and learning. Portsmouth,NH: Heinemann.Chandler, S. P. (1991). Metaphor comprehension: A connectionist approach to implications for themental lexicon. Metaphor and Symbolic Activity, 6(4), 227-258.Charlesworth, P., Jackson, M., Kneeshaw, D., & Stanton, 5. (1985). All in good time. Toronto, Ont.:McGraw-Hill Ryerson.Charlesworth, P., Jackson, M., Kneeshaw, D., & Stanton, 5. (1985). Nineteenth moon. Toronto,Ont.: McGraw-Hill Ryerson.Cunningham, J. W. (1976). Metaphor and reading comprehension. Journal of Reading Behavior,8(4), 363-368.Danesi, M. (1993). Vico, metaphor and the origin of language. Bloomington, IN: Indiana UniversityPress.Dent, C., & Rosenberg, L. (1990). Visual and verbal metaphors: Developmental interactions. ChNdDevelopment, 61, 983-994.Dias, P. X. (1987). Making sense of poetry: Patterns in the process. Ottawa, Ont.: The CanadianCouncil of Teachers of English.Downie, N. M., & Heath, R. W. (1983). Basic statistical methods (5th. ed.). New York: Harper & Row.Englert, C. S., & Semmel, M. I. (1981). The relationship of oral reading substitution miscues tocomprehension. The Reading Teacher, 35, 273-280.Ericsson, K. A., & Simon, H. A. (1993). Protocol analysis: Verbal reports as data (rev. ed.).Cambridge, MA: The MIT Press.Esbensen, B. J. (1993). Playful slider. Toronto: Little, Brown & Co.Evans, M. A., & Gamble, D. L. (1988). Attribute saliency and metaphor interpretation in school-agechildren. Journal of Child Language, 15, 435-449.Gildea, P., & Glucksberg, 5. (1983). On understanding metaphor: The role of context. Journal ofVerbal Learning and Verbal Behavior, 22, 577-590.von Glasersfeld, E. (1984). An introduction to radical constructivism. In P. Watzlawick (Ed.), Theinvented reality: How do we know what we believe we know? (pp. 17-40). New York: Norton.174von Glasersfeld, E. (1989). Facts and the self from a constructivist point of view. Poetics, 18,435-448.Glucksberg, S. (1991). Beyond literal meanings: The psychology of allusion. Psychological Science,2(3), 146-152.Glucksberg, S., & Keysar, B. (1990). Understanding metaphorical comparisons: Beyond similarity.Psychological Review, 97(1), 3-18.Gregory, M. E., & Mergler, N. L. (1990). Metaphor comprehension: In search of literal truth, possiblesense, and metaphoricity. Metaphor and Symbolic Activity, 5(3), 151 -1 73.Howarth, 5. (1993). The Middle Ages. New York: Viking.Inhoff, A. W., Lima, S. D., & Carroll, P. J. (1984). Contextual effects on metaphor comprehension inreading. Memory & Cognition, 12(6), 558-567.Johnson, D. W., & Johnson, R. T. (1994). Learning together and alone: Cooperative, competitive,and individualistic learning (4th. ed.). Boston: Allyn & Bacon.Johnson, M. (Ed.). (1981). Philosophical perspectives on metaphor. Minneapolis, MN: University ofMinnesota Press.Kanji, G. K. (1993). 100 statistical tests. London: Sage Publications Ltd.Keil, F. C. (1986). Conceptual domains and the acquisition of metaphor. Cognitive Development, 1,73-96.Kincade, K. M. (1991). Patterns in children’s ability to recall explicit, implicit and metaphoricalinformation. Journal of Research in Reading, 14(2), 81-98.Kogan, N., & Chadrow, M. (1986). Children’s comprehension of metaphor in the pictorial and verbalmodality. International Journal of Behavioral Development, 9, 285-295.Kraus, S., & Mallory, K. (1993). The search for the Right whale. New York: Crown.Lakoff, G. (1993). The contemporary theory of metaphor. In A. Ortony (Ed.), Metaphor and thought(pp. 202-251). Cambridge, MA: Cambridge University Press.Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago, IL: The University of ChicagoPress.Lakoff, G., & Turner, M. (1989). More than cool reason: Afield guide to poetic metaphor. Chicago, IL:The University of Chicago Press.Lasky, K. (1993). Monarch. San Diego: Harcourt Brace & Co.Lowell, P. E. (1970). Problems with identifying reading levels with informal reading inventories. In W.Durr (Ed.), Reading difficulties: Diagnosis, correction, and remediation (pp. 120-126). Newark,DE: International Reading Association.Luke, A. (1988). Literacy, textbooks and ideology: Postwar literacy instruction and the mythology ofDick and Jane. Barcombe, Lewes: The Falmer Press.175Lythcott, J., & Duschl, R. (1990). Qualitative research: From methods to conclusions. ScienceEducation, 74(4), 445-460.Mac Cormac, E. P. (1985). A cognitive theory of metaphor. Cambridge, MA: The MIT Press.Mate, Z., & Malicky, G. (1988). A semantic feature approach to metaphor comprehension: Whatchildren’s processing of metaphors reveals. Reading--Canada--Lecture, 6(2), 110-117.Matthews, D. (1993). Arctic summer. New York: Simon & Schuster.May, F. B. (1990). Reading as communication: An interactive approach (3rd. ed). Columbus, OH:Merrill.May, F. B. (1994). Reading as communication (4th. ed.). New York: Macmillan.Mayer, R. E., & Bromage, B. K. (1980). Different recall protocols for technical texts due to advanceorganizers. Journal of Educational Psychology, 72(2), 209-225.McClelland, J. L., & Rumelhart, D. E. (1986). A distributed model of human learning and memory. InJ. L. McClelland, D. E. Rumelhart, &the PDP Research Group (Eds.), Parallel distributedprocessing: Explorations in the microstructure of cognition. Volume 2: Psychological andbiolociical models (pp. 170-215). Cambridge, MA: The MIT Press.Mclnnes, J. (1984). Star flights. Scarborough, Ont.: Nelson.Mclnnes, J. (1991). Spin among the stars. Scarborough, Ont.: Nelson.McKenna, M. C. (1983). Informal reading inventories: A review of the issues. The Reading Teacher,36, 670-679.Mercer, K. L. (1985). An investigation of the effects of metaphor on seventh-grade students’comprehension of expository text. Unpublished master’s thesis, The University of BritishColumbia, Vancouver, B. C.Moffett, J., & Wagner, B. J. (1991). Student-centered reading activities. English Journal, 80(6),70-73.Norman, D. A. (1986). Reflections on cognition and parallel distributed processing. In J. L.McClelland, D. E. Rumelhart, &the PDP Research Group (Eds.), Parallel distributedprocessing: Explorations in the microstructure of cognition. Volume 2: Psychological andbiological models (pp. 531-546). Cambridge, MA: The MIT Press.Ortony, A. (1980). Metaphor. In R. J. Spiro, B.C. Bruce, &W. F. Brewer (Eds.), Theoretical issues inreading comprehension (pp. 349-365). Hillsdale, NJ: Lawrence Erlbaum Associates.Pearson, B. Z. (1990). The comprehension of metaphor by preschool children. Journal of ChildLanguage, 17, 185-203.Pearson, P. D., & Raphael, T. E. (1981). The function of metaphor in children’s recall of expositorypassages. Journal of Reading Behavior, 11(3), 249-261.Pritchard, R. (1990). The effects of cultural schemata on reading processing strategies. ReadingResearch Quarterly, 25(4), 273-295.176Pugh, S. L., Hicks, J. W., Davis, M., & Venstra, T. (1992). Bridging: A teacher’s guide to metaphoricalthinking. Urbana, IL: NCTE & Bloomington, IN: ERIC Clearinghouse on Reading andCommunication Skills.Purves, A. C., & Rippere, V. (1968). Elements of writing about a literary’ work: A study of response toliterature. Champaign, IL: NCTE.Reynolds, R, E., & Schwartz, R. M. (1983). Relation of metaphoric processing to comprehension andmemory. Journal of Educational Psychology, 75(3), 450-459.Richards, I. A. (1936). The philosophy of rhetoric. New York: Oxford University Press.Robbins, K. (1993). Power machines. New York: Henry Holt & Co.Rosenblatt, L. M. (1978). The reader, the text, the poem: The transactional theory of the literary work.Carbondale, IL.: Southern Illinois University Press.Rosenblatt, L. M. (1991). Literature--S.O.S.! Language Arts, 68, 444-448.Rumelhart, D. E. (1980). Schemata: The building blocks of cognition. In R. J. Spiro, B. C. Bruce, &W. F. Brewer (Eds.), Theoretical issues in reading comprehension (pp. 33-58). Hillsdale, NJ:Lawrence Erlbaum Associates.Rumelhart, D. E., Smolensky, P., McClelland, J. L., & Hinton, G. E. (1986). Schemata and sequentialthought processes in PDP models. In J. L. McClelland, D. E. Rumelhart, & the PDP ResearchGroup (Eds.), Parallel distributed processing: Explorations in the microstructure of cognition.Volume 2: Psychological and biological models (pp. 7-57). Cambridge, MA: The MIT Press.Seidenberg, P. L., & Bernstein, D. K. (1988). Metaphor comprehension and performance onmetaphor-related language tasks: A comparison of good and poor readers. RASE: Remedial &Special Education, 9(2), 39-45.Silberstein, L., Gardner, H., Phelps, E., & Winner, E. (1982). Autumn leaves and old photographs:The development of metaphor preferences. Journal of Experimental Child Psychology, 34,135-1 50.Siltanen, S. A. (1989). Effects of three levels of context on children’s metaphor comprehension.Journal of Genetic Psychology, 150(2), 197-215.Siltanen, S. A. (1990). Effects of explicitness on children’s metaphor comprehension. Metaphor andSymbolic Activity. 5(1), 1-20.Spiro, R. J. (1980). Constructive processes in prose comprehension and recall. In R. J. Spiro, B. C.Bruce, & W. F. Brewer (Eds.), Theoretical issues in reading. Hillsdale, NJ: Lawrence ErlbaumAssociates.Spivey, N. N. (1987). Construing constructivism: Reading research in the United States. Occasionalpaper No. 12. Poetics, 16, 169-192.Spivey, N. N. (1990). Transforming texts: Constructive processes in reading and writing. WrittenCommunication, 7(2), 256-287.1 77Stein, N. L., & Glenn, C. G. (1979). An analysis of story comprehension in elementary schoolchildren. In R. 0. Freedle (Ed.), New directions in discourse processing (pp. 53-120).Norwood, NJ: Ablex Publishing Corporation.Stevens, J. (1990). Intermediate statistics: A modern approach. Hillsdale, NJ: Erlbaum.Thorne, E. A., & Irwin, J. M. (1988). Handshakings. Toronto: Gage.Thorne, E. A., & Irwin, J. M. (1988). Lobstick. Toronto: Gage.Tuinman, J. (1988). Ride the wave. Toronto: Ginn.Vosniadou, S. (1987). Children and metaphors. Child Development, 58, 870-885.Vosniadou, S., Ortony, A., Reynolds, P. E., & Wilson, P. T. (1984). Sources of difficulty in the youngchild’s understanding of metaphorical language. Child Development, 55, 1588-1 606.Vygotsky, L. S. (1962). Thought and language. Cambridge, MA: The MIT Press.Vygotsky, L. S. (1978). Mind in society. Cambridge, MA: Harvard University Press.Waern, Y. (1980). Thinking aloud during reading: A descriptive model and its application.Scandinavian Journal of Psychology, 21, 123-132.Waggoner, J. E., Messe, M. J., & Palermo, D. 5. (1985). Grasping the meaning of metaphor: Storyrecall and comprehension. Child Development, 56, 1156-1166.Watzlawick, P. (1984). The invented reality: How do we know what we believe we know? New York:Norton.Way, E. C. (1991). Knowledge representation and metaphor. The Netherlands: Kluwer AcademicPublishers.Webster’s Ninth New Collegiate Dictionary. (1986). Markham, Ont.: Thomas Allen & Son.Wells, G. (1990a, May/June). Learning to be literate. The ATA Magazine, pp. 10-13.Wells, G. (1 990b). Talk about text: Where literacy is learned and taught. Curriculum Inquiry, 20(4),369-405.Wells, G. (1991). Talk for learning and teaching. Paper presented at the International Convention onLanguage and Literacy, University of East Anglia, England, April 6-10, 1991.Winner, E. (1988). The point of words: Children’s understanding of metaphor and irony. Cambridge,MA: Harvard University Press.Winner, E., Levy, J., Kaplan, J., & Rosenblatt, E. (1988). Children’s understanding of nonliterallanguage. Journal of Aesthetic Education, 22(1), 51-63.178APPENDIX ASample Letter to Publisher Seeking Permission to PhotocopyPermissions DepartmentHoughton Mifflin Company215 Park Avenue SouthNew York, New York 10003Dear Sir or Madam,I am a doctoral candidate in the Department of Language Education at The University of BritishColumbia. My dissertation will examine sixth-grade students’ understanding of metaphorical languagein informational trade books.Your publishing company has contributed books for inclusion in the NCTE publication, AdventuringWith Books (Eleventh Edition), being edited by Dr. Wendy Sutton of U.B.C.s Dept. of LanguageEducation. Dr. Sutton is interested in making the books widely known and in having the books make acontribution to research at the university level. As a member of my supervising committee, she hasencouraged me to look at the books as possible sources of text for my research, which I have done,isolating several books that would be appropriate.I am writing to you with two requests. The first is for permission to use one of your books -- FrontierHome, by Raymond Bial (© 1993) -- in my research. I would be using it with multiple groups of sixstudents each from three separate grade six classes in schools in Nova Scotia, Canada, during themonth of February, 1994. I wish to use the text in a form consistent with the way it occurs in thebooks, but, because of the cost of purchasing multiple copies of this and the other books I propose touse, my second request is for permission to colour photocopy the cover and pages 1 to 8 (six copiesof each).If you wish to verify the integrity of this request, you may contact Dr. Sutton or my advisor, Dr. VictorFroese, Head, Department of Language Education, U.B.C., at (604) 822-5235 (Dr. Froese) or (604)822-5229 (Dr. Sutton).Thank you for your consideration of my requests.Sincerely,Leigh Faulkner179APPENDIX BMultiple Choice Meaning Clarification Activities--Passages X, Y, and ZPassage XMultiple Choice Meaning Clarification Activity-- “Shadows of Night”Name:_____________________________Directions(a) Do this activity as a think aloud.(b) For each item below, look back to “Shadows of Night” and reread the part that’s marked with ayellow dot having the same number as the item.(c) Put a check mark in front of the ending that’s closest to your meaning of the underlined part as it’sused in “Shadows of Night”. If no ending is close to the meaning you’ve made, write in your ownmeaning.(d) When you’re ready, turn on your cassette recorder and begin.1. “Shadows of Night”______(a) moonlight makes shadows just the same as sunlight does.(b) it can be just as dark in shadows as it can at night.(c) bats are hidden by being darker than the night itself.My own meaning:2. “The evenina sky deepens into night”_(a) the sky is further away at night.(b) the sky fills with stars.(c) the sky gets darker.My own meaning:3. “Mouse-eared”(a) a bat’s ears look like a mouse’s ears.(b) a bat is a relative of a mouse.(c) a bat can hear the same things a mouse hears.My own meaning:4. “Light-fleeing”(a) a bat fears light.(b) a bat can fly faster because it doesn’t weigh much.(c) a bat disappears when the light disappears.My own meaning:1805. “Maternity colony”_____(a) a place whore mothers go to learn how to care for the young.(b) a group of females who will soon have babies.(c) a main settlement that other settlements grow from.My own meaning:6. “Forms a basket with her tail membrane”(a) removes part of her tail.___(b) curves her tail so it can hold something.(c) wraps her tail around her.My own meaning:7. “Cradles the baby”(a) rolls the baby up into a tight ball.(b) puts the baby in a special nest.(c) surrounds and protects the baby.My own meaning:8. “Silky hair”(a) hair that looks like smooth cloth.(b) a covering made of silk.(c) very, very soft hair.My own meaning:9. “It also latches on securely”(a) it lets go after a short time.(b) it takes a firm grip.(c) it has a hook for fastening.My own meaning:10. “The first attempts to fly around the barn are full of confusion”(a) their earliest flights are all mixed up.(b) they don’t know how to communicate with each other.(c) they fly around the barn because something is scaring them.My own meaning:181Passage YMultiple Choice Meaning Clarification Activity -- ‘Frontier Home”Name:_____________________________Directions(a) Do this activity as a think aloud.(b) For each item below, look back to “Frontier Home” and reread the part that’s marked with a yellowdot having the same number as the item.(c) Put a check mark in front of the ending that’s closest to your meaning of the underlined part as it’sused in “Frontier Home”. If no ending is close to the meaning you’ve made, write in your ownmeaning.(d) When you’re ready, turn on your cassette recorder and begin.1. “Deep woods abounding in shadows and mystery”______(a) the forest was made up of tall, dark trees of a kind the people didn’t know.(b) the forest had many things that were unknown to the people.(c) strange animals bounded off through the dark forest.My own meaning:2. “Before cold weather set in again”_(a) before they got to their new homes where the climate was colder.(b) before winter returned.(c) before dark when it got very cold in the wagons.My own meaning:3. “Occasional patches of snow persisted”_(a) they would sometimes have snowstorms that went on and on.(b) they would sometimes see small areas of snow that hadn’t yet melted.(c) they would sometimes find the road so full of snow they couldn’t go on.My own meaning:4. “Wagons . . . were transporting not only the pioneers’ fears but their expectations”(a) people were afraid their wagons would break down.(b) people depended on transport companies to carry their belongings.(c) people were both afraid and hopeful.My own meaning:1 825. “Came from all walks of life”_____(a) they had walked from community to community to find work.___(b) they came from all different kinds of jobs and backgrounds.(c) they had to walk because the wagons were too full.My own meaning:6. “Threats from the weather”(a) the weather reports called for rain or snow.(b) there were serious problems because of the weather.(c) the people didn’t have proper clothes for bad weather.My own meaning:7. “Civilization’s comforts and protections”(a) means when they travelled together they felt more comfortable.(b) means the things they had enjoyed and depended on in their old homes.(c) means the wagons they brought with them from home were very safe.My own meaning:8. “Occasionally breaking into a strange, open country”(a) going into an area that was free for them to take.(b) going into an area where they were strangers.(c) going into an area where there weren’t any forests.My own meaning:9. “Little more than an interruption to the trees”(a) a place filled with different kinds of trees.(b) a space before the forest started again.(c) a place where they wouldn’t cut down the trees.My own meaning:10. “Girdled”(a) put a tight belt on.(b) made a circle around.(c) got sap from the trees.My own meaning:183Passage ZMultiple Choice Meaning Clarification Activity -- “The Middle Ages”Name:____________________________Directions(a) Do this activity as a think aloud.(b) For each item below, look back to “The Middle Ages” and reread the part that’s marked with ayellow dot having the same number as the item.(c) Put a check mark in front of the ending that’s closest to your meaning of the underlined part as it’sused in “The Middle Ages”. If no ending is close to the meaning you’ve made, write in your ownmeaning.(d) When you’re ready, turn on your cassette recorder and begin.1. “An old power crumbles”_____(a) things break to pieces when they get old.___(b) power breaks things that are old.(c) something that had once been very strong breaks into pieces.My own meaning:2. “The sack of the great city of Rome”(a) a kind of pack that Romans carried things in.(b) the part of Rome where poor people lived.(c) the defeat of Rome and stealing of its riches.My own meaning:3. “Town-based society”(a) new towns built where old ones had been destroyed.(b) a society where people lived in the country but worked in town.(c) a society that depended on the organization of towns.My own meaning:4. “Raids by Germanic tribes shook the power of the Roman Empire”(a) attacks from outside encouraged people to form tribes.(b) attacks by outsiders weakened the government.(c) uprisings by soldiers took place against their own government.My own meaning:1 845. “Darinq raids put the Roman Empire under pressure”______(a) foreign soldiers were hired to protect Rome.(b) attacks from outside threatened society._(c) Rome found a new way to attack its enemies.My own meaning:6. “The fall of the Roman Empire”(a) the pulling down of the walls around the Roman Empire.(b) a special celebration held in autumn in the Roman Empire.(c) the destruction of the Roman Empire.My own meaning:7. “A new sort of society was cirowinci up’(a) many new cities were taking the place of old ones.(b) a different way of living was developing.(c) most of the people in the society were adults.My own meaning:8. “A special ceremony marked such occasions”(a) the people celebrated with dancing and singing.(b) a special list recorded the names of the people who attended.(c) a big event was held to give the agreement extra meaning.My own meaning:9. “The oath of loyalty was meant to be bindinci for life”(a) a person was never allowed to change his or her mind.(b) a kind of chain used to hold prisoners in dungeons.(c) a kind of cloth wrapped around their legs for protection.My own meaning:10. “Anyone who broke it would be regarded as a traitor”(a) anyone who didn’t do what they’d promised would be an enemy.(b) anyone who escaped from the chains would be hunted down.(c) anyone who lost their leg cloths would be in danger.My own meaning:185APPENDIX CSample Individual Think-Aloud Protocol (Main Study)[Note--Students own words in bold print; original text in plain print.]Shadows of NightCode X1F4JFrom the title, I guess it means (?) the people outside in the shadows that (?)The evening sky deepens into night. Birds return to their root... to their roasts (sic), and manyother animals quiet.. .Iy nesifle (sic) into their burrows. But some creatures are just waking up. Darkshapes flutter across the sky, suddenly turning, diving, swooping. Bats are beginning to hunt.This is the story of the Little Brown Bat, one of the most common bats in North America. For itssmall size, this species has a very long life span -- sometimes more than thirty years. Its scientific nameis Myot... myontis (sic)... These latin words mean “mouse-eared” and “light-feeling” (sic).” Well! Idon’t know what that word means... “mouse-eared and... I know that bats begin tohunt... hunt at night and they sleep during the day.It’s springtime. A female Little Brown Bat has reccontly (sic) emerred (sic) a ... hundred... ahibernating (sic) cave hidden deep in the woods. She has travelled near... nearly a hundred miles injust two nights. A baby is growing inside her, and she is looking for the right place to give birth. Well!Most people ... I guess have to find the right place to give birth to their baby. Atdawn, the female bat slips through the narrow opening at the top of the (sic) abandoned barn. Inside,she finds a maternity colony of hundreds of pregnant bats hanging from the rafters. She closes.., shechooses her spot amongs (sic) them and falls asleep.All other (sic) North America, female bats are gathering -- inside barns, churches (sic) steeples,attics, cellars, and chimleys (sic) -- often unnoticed by the humans living nearby. So... I guess thismeans that they could be living in my chimley and I wouldn’t know.Two months later, the baby bat is ready to be born. During the birth, the mother turns herself...turns herself... head up (which is upside down for a bat). She hangs from her numb claws.., from hernumb (sic) claws and forms a basket with her tail... The baby is... the baby is born feet first andclutches immediately to (sic) a foothold well... with its well-developed toes, squirming and pushing tohelp the rest of the (sic) body emerge. So I guess this means, it has trouble getting born.Um... As soon as the baby is born, the mother turns head down again and cradles the babyunder her wing to... to nurse. As it drinks the sweet, warm milk, she cuts the umbilical cord andgrooms the baby carefully. So... I guess it’s like.., when humans have babies, they haveto cut the (?) cord or whatever.186During the first day, the baby bat clings to the (sic) mother, struggling (sic) under her wing. Itsbody is covered with fine, skilky hair... silky hair, and its eyes are still closed. The baby grips itsmother’s fur tightly to prevent a fall. It also latches on securely to her nipple as it nurses. If the motheris distrupted (sic), she will carry her baby with her as she circles inside the ... barn. I guess thismeans she takes good care of the bat and makes sure nobody hurts her... or it orwhatever.Hundreds of other baby bats (called pups) ... I never knew they were called pups!...were born in the barn around the same time. When the mother bats go out to hunt at night, theyleave the pups clustered together tightly, hanging from the barn ceiling. Well! I didn’t know that.Well, I’m just surprised at that! Waiting for their mothers to return, they romp and twistle (sic)like kittens. To... like kittens.. .To a human, the heat and the den... and the denis, pungient (sic)smells (sic) of a bat roast (sic) to (sic) overpowering. But to a naked... but to a naked bat pup, the roast(sic) is... is wonderfully warm and comfort... and comftorting (sic). I didn’t know that!When a mother returns, she finds her baby in the midist (sic) of ... the screaming pups by callingout the (sic) listening for its particularly (sic) cry... cry. When she reaches her baby, she snif. . .shesniffs and licks it, making sure it is the right one. When they settle down... then they settle down tonurse. Well! It seems like she takes good care...In about three weeks, the young bats are ready to fly. Their pink skin is ... covered with soft,brown fur, and their wings have grown much larger. The first attempts to fly around the barn with... arecareful (sic) of coIl... colisions (sic). Jostling.., jostling, squeaking pups are everywhere. One youngbat manages to take off and fly to the other side without colliding with the others. But landing isharder. The pup doesn’t know what... how to flip around at the end of its flight, so it lands head upand has to turn around awkward (sic) on the wall. Hmm! Landing is pretty hard. Boy, Iwouldn’t want to be a bat!Well! I... kinda found out about bats. And it tells me... mostly, how babieswere born, how they grow, their skin, and how the mother makes... how the mothertells which baby is hers. And, I found out that it isn’t easy to fly in three weeks.And it’s kinda hard... when the room is all filled with bats. They have trouble...coming around... cause you’re scared that your gonna cry. And I... huh... feel thatthe bats take good care... just like our parents take care of us.187APPENDIX DSpecific Traits and DefinitionsMEANING-CONSTRUCTION PROCESSESThe processes named by the following specific traits are considered to involve meaning construction.In these processes the student is seen to be actively engaged with the text or the meaning-makingprocess, itself, either affectively or intellectually.AR ffective esponse-- expresses a personal response to the meaning of the text-- e.g., “Holy cow! That’s cheap!”; “That’s a lot!”-- expresses a personal position in relation to meaning of text-- e.g., “I don’t believe in that.”AR(a) -- agrees with partner’s or own AR statementRAR Retell -- affective flesponseRAR(a) -- agrees with AR statement in RetellTRP Restatement! Earaphrase-- repeats an idea in the same or substantially the same words (includes whole, recognizablephrases from the original)-- e.g., “...but there are still sharks...” (phrase appeared in the text)-- paraphrases by repeating an idea in different or substantially different words (onlynecessary identifying words carried over from the original)-- e.g., “...all the dinosaurs are gone...” (phrase did not appear in the text)-- states that the section of text just read is about a particular topic-- e.g., “that’s just talking about ... like the sharks and what they do...”-- provides the name for something previously named in the text-- e.g., “Yeah, they’re called the rivets...” (“rivets” had not been mentioned in theimmediately preceding portion of text, but had been mentioned earlier)-- draws attention to a section title, a picture or a !fl to get new information-- e.g., “This part is called...”188TRP(a) -- agrees with partner’s or own TRP statementRTRP Retell -- Restatement/EaraphraseRTRP(a) -- agrees with TRP statement in RetellMA Affirmation of Understanding-- affirms own understanding of text-- e.g., “Okay... so... I understand that...”-- states that the text is self-explanatory or that it makes sense-- e.g., “Okay... that paragraph makes sense to me...”-- mentions that information was or was not previously known-- e.g., “I never knew that...” or “I knew that...”-- mentions that a textual meaning has just become clear-- e.g., “Oh, I get it...” (the meaning had not been clear and had been under discussion>-- states that something has been learned (may or may not explain the information)-- e.g., “I learned quite a bit...”MA(a) -- agrees with partner’s or own correct MA statement or comments on own understanding orpronunciationRMA Retell -- Affirmation of UnderstandingRMA(a) -- agrees with MA statement in RetellTI Interpretation-- draws a correct inference or correctly predicts what the selection or a section of text will beabout, but must go beyond TRP-- e.g., “So, what he’s saying is we should... respect the animals...” (“respect” had not beenmentioned in the text)-- determines the meaning of a term through inference or analogy-- e.g., “So... it just goes up to here... up to your waist...” (the term being discussed was“waisthigh overalls”)TI(a) -- agrees with partner’s or own correct TI statement189RTI Retell-- InterpretationRTI(a) -- agrees with TI statement in RetellMAT Iruth Assessment-- considers the truth of information as possible but not certain-- e.g., That could be very possible...”-- offers an opinion or disagrees with text-- e.g., “I wouldn’t see why...” (this was said in relation to a statement made in the text>-- poses a question about information contained in the text or suggested by it (but aboutmeanings of terms contained in the text)-- e.g., ‘Well, the thing is... I wonder if any animals can catch diseases from other animals...’(the text had mentioned that sickly fish were eaten by sharks)-- remarks with uncertainty or disagreement about comment, suggestion or question posedby partner or by selfMAT(a) -- agrees with partner’s or own MAT statementRMAT Retell-- Iruth AssessmentMCL Compares/ontrasts/Links Portions of Text or Ideasexamines two or more portions of text to determine whether they present similar ordifferent information-- e.g., “I guess that explains the top paragraph...” (this was said about a later portion of text)-- states that an idea from one part of the text is the same as, explains, or is connected insome way to another idea in the text-- e.g., ‘That explains that sentence about ‘what has Toronto done in return’...”-- relates an idea from the text to self or another person- connects ideas or information from text with general or specific human experience-- e.g., “They don’t have them on my jeans...” (this refers to ‘rivets’, which had just beenmentioned in the text)-- answers a question by referring to specific information in the text-- e.g., ‘Well... it says it’s ‘untold’.. .“ (a question had been raised during discussion of thetext and “untold” was a correct answer to the question)190-- refers to a picture or map to support an ideaMCL(a) -- agrees with partner’s or own MCL statementRMCL Retell -- Corn pares/Contrasts!jinks Portions of Text or IdeasRMCL(a) -- agrees with MCL statement in RetellMPK Erior nowledge-- offers accurate information from memory-- e.g., “I found that there’s more than the great white shark... there’s mako shark andstuff...” (this information had come from a book read previously)-- gives from memory a correct meaninci of a word or conventional phrase-- e.g., “Adapt means... umrn... get used to...” (“adapt” had just been used withoutexplanation)MPK(a) -- makes a statement emphasizing an MPK statement-- e.g., “Really.”RMPK Retell -- EriorinowledgeRMPK(a) -- agrees with MPK statement in RetellMSK ource of Own nowledge-- tells how or where information was gained-- e.g., “I’ve seen a shark eat something...”MSK(a) -- agrees with partner’s or own correct MSK statementET Evaluation of Text qua Text-- comments on the style, textual features, effectiveness or appropriateness of the text itself-- e.g., “They’re not explaining it enough.”-- comments on the purpose of a portion of text-- e.g., “And that’s a niced little wrap up.”-- mentions a structural element of the genre (such as the ‘introduction’)191-- e.g., “... a big introduction, I guess.”-- suciciests that something does or does not belong in the text or that something potentiallyuseful has not been included-- e.g., “But it doesn’t tell us what kind of sharks its about.”-- sucigests when the text was written (for the purpose of commenting on the text)-- e.g., “Mmm... this must have been made a while ago.”ET(a) -- agrees with partner’s or own ET statementRET Retell -- Evaluation of TextRET(a) -- agrees with ET statement in RetellSRS Suggests a Specific Reading trategy-- suciciests a particular strategy in order to clarify text or to find information-- e.g., “So, let’s just read on and see...”SRS(a) -- agrees with partner’s or own SRS statementED Discussion-- remarks on the course or effectiveness of the dyadic discussion-- e.g., “I think that’s getting a little off topic, but it’s a good comparison.” (commenting on apersonal experience recounted by other member of the dyad)ED(a) agrees with partner’s or own ED statementNEUTRAL STATESThe states named by the following specific traits are considered to involve orientation toward the task,itself.EP Erocedural-- in dyad, asks question of partner to initiate turn change-- e.g., “Well, what happened?” or “Do you know?”-- remarks on, discusses or questions what to do or what has been done, how it should be192done, order, turn-taking, interpretation of directions or non-text features, or completion oftask-- e.g., “You read this.”-- remarks that there is nothing more to add to a point-- e.g., “...and I guess that was it.”-- uses a standard phrase to initiate turn-taking-- e.g., “Okay.”-- uses a standard phrase to end one section or begin another-- e.g., “Yeh” (where other’s dialogue is broken into units, thus encouraging other tocontinue)EP(a) -- agrees with partner’s or own EP statementREP Retell-- EroceduralREP(a) -- agrees with EP statement in RetellOM Other/Miscellaneous-- any meaning unit that does not belong under any specific trait given above, includingincomplete utterancesROM Retell -- Other/MiscellaneousPRE-MEANING CONDITIONSThe conditions named by the following specific traits are considered to represent understandinginconsistent with either the text or commonly-accepted information.MD Denial of Understanding-- admits text is not understood-- expresses doubt about own understanding of text, including the meaning orpronunciation of specific terms in text or raised in discussion-- e.g., “That part is still confusing to me...”; “I don’t get that part...”; “Hmm...wonder whatdredging is...”--sucjests that a question can’t be answered because information hasn’t been given193-- e.g., “I can’t answer that because I haven’t read the next sentence yet...”MD(a) -- agrees with an incorrect statement of meaning or pronunciation made by dyadic partner orcomments on own understanding or pronunciationRMD Retell -- Qenial of UnderstandingRMD(a) -- agrees with MD statement in RetellTM Misinterpretation-- draws an incorrect inference or incorrectly predicts what the selection or a section of thetext will be about-- e.g., “So... not very many people like the water... so nobody wants it, basically to bethere...” (the point of the text was that people did not look after their water resource)-- states an interpretation not supported by the text-- e.g., “So I guess that means that the water forms almost every city...” (the point was thatevery major city was situated on a body of water)-- incorrectly infers the meaning of a term used in the texte.g., “Probably just a little ship ‘cause ‘clipper’ is sort of a short word...” (this was an attemptto understand the term “clipper ship”, mentioned in the text)TM(a) -- agrees with partner’s or own incorrect TM statementRTM Retell -- MisinterpretationRTM(a) -- agrees with TM statement in RetellMPB Prior Beliefoffers inaccurate information from memory-- e.g., “I’m sure they called it denime...” (term “denim” had been used in the text)-- gives from memory an incorrect meaninci of a word or conventional phrase-- e.g., “It’s kind of like a bluey, greeny kind of...” (term “indigo” had been used in the text)MPB(a) -- agrees with partner’s or own MPB statement194APPENDIX ESample Soecific-Trait Analysis (Sample Dvadic Think-Aloud Protocol, Main Study)[Note--Students own words in bold print; original text in plain print.]Shadows of NightCode X2F3H & X2M3N (Dyad 3A)Si. (?)S2. and this is [student’s name]Si. doing a voice testing. Shadows of the (sic) Night You can talk you Know. (1)S2. The Hidden World of The Little Brown BatSi. So it’s probably about bats. (2)S2 The evening sky deepens into the (sic) night. The (sic) birds return to their roosts, and the (sic)man... and the (sic) many of (sic) the (sic) other animals quietly nestle into their burrows.Si. So... (3)S2. But some of (sic) the (sic) creatures are just waking up.Si. So that means that they’re...the birds are... the bats are going to... the birdsare going in trees and... (4)S2. They’re going back to their homes... (5)Si. And the little animals are going under ground. (6)// (laughter) They’re all goingasleep... (7)// oh they’re all waking up. (8) (laughter)S2. Dark shapes flutter across the sky, suddenly turning, diving, swooping. Bats are beginning tohunt.Si. So... bats are hunting in the night. (9) This is the story of the Little Brown Bat, one of themost common bats in North America. Uh... For its small size, these (sic) species ... very long lifespan... You can... (10)S2. It’s telling us... (11)Si .... come in any time now. (12) ii (laughter) (?) interrupting is the thing to do (?)(13) i/i guess so. (14)S2. It’s tells us about the Little Brown Bats. (15)Si. (laughs) Big Wings! (16) // --sometimes more than thirty years. Oh! Life Span! (17) // Ithought it was wing span. (18)// Oops!(19)//They have along life! (20) II Itsscientifics (sic) name is Myso... so...Mys... Myotis uckifus.(sic) These latin words mean“mouse-eared” and light-fleeing.” They could have just said that! (21)fi They didn’t195need to get (?) (22)S2. At (sic) springtime. A female Little Brown Bat has recently emerged from the (sic) hibernationcave hidden deep in the woods. She has travelled nearly a hundred miles in just two nights. Ababy is growing inside of (sic) her, and she is looking for the right place to give birth. At dawn,the female bat slips through the narrow opening in (sic) the top of the (sic) abandoned barn.Inside, she finds a matern...Si. maternityS2. colony of hundreds of pregnant bats (laughs) hanging... She chooses her spot among themand falls asleep.Si. Oh! That’s nice to know!(23)//Just what I wanted to know today. (24) //Hundreds of bats are pregnant in a barn. (25) ii (laughter) All over North America,female bats are gathering -- inside barns, church steeples, attics, cellars, and chimneys -- oftenunnoticed by the humans living nearby. We (?) hundreds of bats in our barn. (26) 1/Huh! I didn’t even notice them!(27)S2. During the birth (?),the baby is (?)...Si. What is (?) (28)S2. Oh! (29) Two months later, the baby bat is ready to be born. During the birth, the mother turnsherself head up...up (which is upside down for a bat). She hangs from her thum... thumb clawsand forms a basket with her tail... membrane.Si. Where are you? (30)S2. Right Here! (31) The baby is born feet first and clutches immediately to... for a foothold with a(sic) well-developed toes, squirming and pushing to help the rest of the (sic) body emerge.Si. Oh! Okay! (32) As soon as the baby is born, the mother turns head down again and cradlesthe baby under her wing... to nurse. As it drinks the sweet, warm milk, she cuts the umbilicalcord and grooms the baby carefully.S2. During the first day, the baby bat clings to (her) mother, snuggled under her wing. It... its body iscovered with fine, silky hair, and its eyes are still closed. The baby grips the (sic) mother’s full...fur tightly to prevent a fall. It also latches to (sic) securely to her nipple and (sic) nurses. When(sic) the mother is disturbed, she will carry her baby ... as she circles inside of (sic) the dark barn.Si. So a bat has a baby (33) // and they... put them under their wing. (34) Hundreds ofother baby bats (called pups) were born in the barn around the same time. When the motherbats go out to hunt at night, they leave the ba... the pups clustered together tightly, hangingfrom the barn ceiling. Waiting for their mothers to return, they romp and tussle like kittens. To ahuman, the heat and the dense, pungent smell of a bat roost is overpowering. But to a naked196bat pup, the roost is wonderfully warm and comforting. Hmm! (35)S2. When the (sic) mother returns, she finds her babies (sic) in the midst of... screaming pups bycalling out ... listening to (sic) a (sic) peculiar (sic) cry. When she reaches her baby, she sniffsand licks it, making sure that (sic) it’s (sic) the right one. Then she (sic) settles (sic) down tonurse.Si. They look funny! (36) II Oh! This is not.., not my child! (37) In about three weeks,the young bats are ready to fly. Their pink skin is now covered with soft, brown fur, and theirwings have grown much larger. The first attempts to fly around the barn are full of confusion.Jostling, squeaking pups are everywhere. One young bat manages to take off and fly to theother side without colliding with the others. But landing is harder. The pup doesn’t know howto flip around at the end of its flight, so it lands head up and... turns ... awkwardly to (sic) the wall.They fly around the ball, run in... barn.., run into each other (38) II and banginto walls. (39)// Okay! So... (40)S2. Now we have to do the retell? (41)Si. So... this story was about...other Little Brown Bats, (42)// how they (?) (43) iiPretty much... about their life and stuff. (44)// Like (?) things in the night. (45)S2. They live in barns and stuff. (46)Si. It’s quite interesting (?) 47)1/Okay! Lets... (48)Specific Trait Tally -- Code X2F3H & X2M3N (Dyad 3A)1. EP 17. TRP 33. TRP2. TRP 18. MA 34. TRP3. CM 19. CM 35. CM4. TRP 20. AR 36. AR5. TRP 21. ET 37. CM6. TRP 22. ET 38. TRP7. TRP 23. AR 39. TRP8. TRP 24. AR 40. EP9. TRP 25. TRP 41. EP10. EP 26. MSK 42. RTRP11. CM 27. MSK 43. ROM12. EP 28. CM 44. RTRP13. EP 29. CM 45. RTRP14. EP 30. EP 46. RTRP15. TRP 31. EP 47. PAR16. TM 32. EP 48. REP197DoLJ.0.APPENDIX FAverage Specific Traits Profile Graphs (Combined Categories) of Studentsin Extreme Groups by MCMCA ScoreIndividualMeaning Const Neutral States Pre-Meaning ConCombined CategoryDyadic—110080LE20I—(fi200(0Extreme Group Individual ist--Average Specific Traits ProfileMeaning Const Neutral States Pre-Meaning ConCombined CategoryExtreme Group Dyadic ist--Average Specific Traits Profile10080604020U)000Meaning Const Neutral States Pre-Meaning ConCombined CategoryExtreme Group Individual 4th--Average Specific Traits ProfileMeaning Const Neutral States Pre-Meaning ConCombined CategonjExtreme Group Dyadic 4th--Average Specific Traits Profile198APPENDIX GHolistic Ratinci ScaleDULLFEELING OF REMOTENESSNOT MUCH TALK‘POINTLESS” TALKLONG “DEAD” SPACESWITHOUT COMMENTARYOVER-DEPENDENCE ONA FEW COGNITIVESTRATEGI ESLOW-LEVEL COGNITIVESTRATEGI ESFOR DYADS:ONE OVERPOWERSTHE OTHERVIGOROUSSENSE OFENGAGEMENTLOTS OF TALK‘THOUGHTFUL” TALKCOMMENTARYTHROUGHOUTVARIETY OFCOGNITIVESTRATEGIESHIGH-LEVEL COGNITIVESTRATEGIESFOR DYADS:BAlANCE BETWEENPARTNERSHOLISTIC RATING SCALE --THINK-ALOUD PROTOCOLS Protocol IdentificationRater IdentificationDirections: Circle the number that best indicates the degree to which the protocol exhibits each set of constructs.1 2 3 4 5 61 2 3 4 5 61 2 3 4 5 61 2 3 4 5 61 2 3 4 5 61 2 3 4 5 61 2 3 4 5 61 2 3 4 5 6199APPENDIX HMCMCA Scores--All Students, Individual and Dyadic ConditionsScoreStudent lndMdual DyadicF3A 5 7F3B 7 4F3C 9 4P3D 5 6F3H 8 9P31 9 6F3J 10 7F3L 7 4F3M 8 8F4A 7 9F4B 7 1F4C 8 10F4D 8 7F4F 5 3F41 3 2F4J 6 5F4K 10 5F4Q 9 7F4R 4 4F4S 6 7P4V 4 2P4X 5 4P5D 8 7F5F 8 1F5H 6 7F51 9 9F5M 6 6F50 7 7F5P 4 4200(APPENDIX H continued)ScoreStudent IndMdual DyadicM3E 7 8M3F 2 3M3G 9 7M3K 10 10M3N 6 7M4E 7 5M4G 4 8M4H 7 5M4L 7 5M4M 7 8M4N 9 8M40 5 7M4P 8 7M4T 5 7M4U 8 9M4W 9 7M5A 9 5M5B 9 5M5C 6 7M5E 6 5M5G 7 6M5J 8 6M5K 3 8M5L 6 6M5N 10 8M5Q 8 8APPENDIX IMCMCA Scores by Student, Passace, and ConditionStudent X(Indiv) X(Dyadic) Y(Indiv) Y(Dyadic) Z(Indiv) Z(Dyadic)F3A 5 7F3B 7 4F3C 9 4F3D 6 5M3E 7 8M3F 2 3M3G 7 9F3H 9 8F31 6 9F3J 7 10M3K 10 10F3L 7 4F3M 8 8M3N 7 6F4A 9 7F4B 1 7F4C 10 8F4D 7 8M4E 5 7F4F 3 5M4G 4 8M4H 5 7F41 2 3F4J 6 5F4K 10 5M4L 5 7M4M 8 7M4N 9 8M40 7 5M4P 8 7201(APPENDIX I Continued)Student X(Indiv) X(Dyadic) Y(Indiv) Y(Dyadic) Z(Indiv) Z(Dyadic)F4Q 9 7F4R 4 4F4S 6 7M4T 7 5M4U 8 9F4V 2 4M4W 9 7F4X 5 4M5A 9 5M5B 5 9M5C 7 6F5D 7 8M5E 6 5F5F 1 8M5G 7 6F5H 7 6F51 9 9M5J 8 6M5K 3 8M5L 6 6F5M 6 6M5N 10 8F50 7 7F5P 4 4M5Q 8 8202203APPENDIX JOverall MCMCA Performance by Syntactic Frame and Similarity Type(After Broderick, 1992)lS(lmplied)/lncompatible-kind comparisonX-1 - 21/37; Y-5 - 15/37; Y-6 - 23/37; Y-7 - 13/37; Z-2 - 21/36; Z-6 - 29/36Total: 122/220; 55.5% (6 items)IS/Incompatible-kind comparisonX-6- 30/37Total: 30/37;81.1% (1 item)IS/Disparate-kind comparisonY-9- 12/37Total: 12/37; 32.4% (1 item)DS(VP)/lncompatible-kind comparisonX-2 - 30/37; X-7 - 27/37; X-9 - 32/37; Y-2 - 27/37; Y-3 - 27/37; Y-1 0 - 18/37; Z-1 - 20/36; Z-4 -16/36;Z-5 - 22/36; Z-7 - 31/36; Z-8 - 24/36; Z-1 0- 28/36Total: 312/438; 71.2% (12 items)DS (Adj)/l ncom patible-kind comparisonX-4- 17/37; X-10 - 28/37; Z-9 - 29/36Total: 74/110; 67.3% (3 items)DS(Adj)/Disparate-kind comparisonX-3- 27/37; X-8 - 27/37; Z-3- 17/36Total: 71/110; 64.5% (3 items)DS(NP)/Disparate-kind comparisonX-5- 32/37Total: 32/37; 86.5% (1 item)DS (PP)/Disparate-kind comparisonY-8- 19/37Total: 19/37; 51.4% (1 item)204(APPENDIX J continued)Combined{[DS(Adj)/I ncompatible-kind comparison] + [DS(VP)/Disparate-kind comparison]}Y-1- 18/37Total: 18/37; 48.6% (1 item){[DS (VP)/Disparate-ki nd comparison] + [IS(l mplied)/l ncompatible-kind comparison]}Y-4- 27/37Total: 27/37; 73.0% (1 item)Totals (excluding combined metaphors)IS: 164/294;55.8%DS: 508/732; 69.4%Incompatible-kind corn parison: 538/805; 66.8%Disparate-kind comparison: 134/221; 60.6%205APPENDIX KMCMCA Results by ItemPassage Xltem# 1 2 3 4 5 6 7 8 9 10Raw ScoreIndividual 13 16 15 9 17 15 13 15 16 17Dyadic 8 14 12 8 15 15 14 12 16 11Total 21 30 27 17 32 30 27 27 32 28Percent 56.8 81.1 73.0 46.0 86.5 81.1 73.0 73.0 86.5 75.7Passage Yltem# 1 2 3 4 5 6 7 8 9 10Raw ScoreIndividual 11 16 13 15 8 14 6 11 5 10Dyadic 7 11 14 12 7 9 7 8 7 8Total 18 27 27 27 15 23 13 19 12 18Percent 48.6 73.0 73.0 73.0 40.5 62.2 35.1 51.4 32.4 48.6Passage Zltem# 1 2 3 4 5 6 7 8 9 10Raw ScoreIndividual 16 10 9 10 11 14 15 12 15 13Dyadic 14 11 8 6 11 15 16 12 14 15Total 30 21 17 16 22 29 31 24 29 28Percent 83.3 58.3 47.2 44.4 61.1 80.6 86.1 66.7 80.6 77.8206(APPENDIX K continued)Totals — Passage x ConditionRaw Score PercentPassage XIndividual 146/190 76.8Dyadic 125/180 69.4Total 271/370 73.2Passage YIndividual 109/190 57.4Dyadic 90/180 50.0Total 199/370 53.8Passage ZIndividual 125/170 73.5Dyadic 122/1 90 64.2Total 247/360 68.6All Passages 717/1100 65.2Note. All percents rounded to the nearest tenth.207APPENDIX LMCMCA Metaphor Analysis(After Lakoff and Johnson, 1980, and Lakoft and Turner, 1989)Passage X[tem# 1 2 3 4 5 6 7 8 9 10MetaphorStructuralStructure MappingOrientationalOntological 2 2 1 1 1 2Event/ActionContainerLife/JourneySimileActualImpliedAnalogySimpleComplexTotal 3 3 1 2 2 2 2 1 2 3Additional FeaturesConventionalMetaphorSimilePhraseOtherMetonymySynecdoche208(APPENDIX L continued)Passage Yltem# 1 2 3 4 5 6 7 8 9 10MetaphorStructuralStructure MappingOrientationalOntological 2 1 1 2 2 2 3Event/ActionContainerLife/JourneySimileActualImpliedAnalogySimpleComplexTotal 4 2 2 3 3 3 3 3 2 2Additional FeaturesConventionalMetaphorSimilePhrase VOtherMetonymy VSynecdoche209[APPENDIX L continued)Passage ZItem# 1 2 3 4 5 6 7 8 9 10MetaphorStructuralStructure MappingOrientationalOntological 1 2 2 4 3 2 2 2 3Event/Action 1 1 1 1 1 1 1 2ContainerLife/JourneySimileActualImpliedAnalogySimpleComplexTotal 2 2 3 5 5 2 4 3 5 3Additional FeaturesConventionalMetaphorSimilePhraseOtherMetonymySynecdoche210APPENDIX MHolistic Scores--Dyadic AverageDyad SurrV7ltems Items AverageX2F3D/M3K 34.250 7 4.893X2F3H/M3N 18.750 7 2.679X2F4NM4T 20.750 7 2.964X2M4M/F4C 33.750 7 4.821X2M40/M4L 30.500 7 4.357X2F4V/M4E 22.750 7 3.250X2M5C/F5H 13.000 7 1.857X2F5D/F5P 39.000 7 5,571X2F51/F5M 36.000 7 5.143Y2M3G/F3L 19.000 7 2.714Y2F31/F3J 18.000 7 2.571Y2F4F/F4Q 21 .250 7 3.036Y2F4I/M4H 15.750 7 2.250Y2F4R/F4D 35.500 7 5.071Y2F4XIF4B 17.000 7 2.429Y2M5B/F5F 29.250 7 4.179Y2M5E/M5J 37.750 7 5.393Y2M5N/M5Q 35.500 7 5.071Z2F3AIF3M 12.250 7 1 .750Z2F3B/M3E 7.000 7 1.000Z2F3C/M3F 18.750 7 2.679Z2F4J/M4N 24.000 7 3.429Z2F4S/M4P 13.250 7 1.893Z2M4U/M4G 22.000 7 3.143Z2M4W/F4K 7.000 7 1 .000Z2M5A(+M5R) 28.250 7 4.036Z2M5G/M5L 14.000 7 2.000Z2M5K/F50 15.000 7 2.143211APPENDIX NHolistic Scores--Individual AveraceStudent SurrV7ltems Items AverageX1F3B 35.000 7 5.000X1F3C 27.000 7 3.857X1M3E 7.500 7 1.071X1F3L 21.750 7 3.107Xl F3M 7.500 7 1 .071X1F4J 24.000 7 3.429Xl F4K 12.500 7 1 .786X1M4N 25.250 7 3.607X1M4P 34.750 7 4.964X1F4Q 31.250 7 4.464X1M4U 27.750 7 3.964X1M4W 7.000 7 1.000X1F4X 17.500 7 2.500X1M5E 25.500 7 3.643X1M5G 11.750 7 1.679X1M5J 35.000 7 5.000X1M5L 33.750 7 4.821X1M5N 32.250 7 4.607X1F5O 17.500 7 2.500Y1F3A 8.500 7 1.214Y1F3D 13.500 7 1.929Y1M3F 7.750 7 1.107Y1F3H 29.750 7 4.250Y1M3N 22.000 7 3.143Y1F4A 29.500 7 4.214Y1F4C 26.750 7 3.821Y1M4G 8.750 7 1.250Y1M4L 24.750 7 3.536Y1M4O 27.500 7 3.929212(APPENDIX N continued>Student SurrV7ltems Items AverageY1F4S 27.250 7 3.893Y1M4T 9.250 7 1.321Y1F4V 11.000 7 1.571Y1M5A 33.000 7 4.714Y1F5H 20.250 7 2.893Y1F5I 26.000 7 3.714Y1M5K 18.500 7 2.643Y1F5M 26.000 7 3.714Y1F5P 14.500 7 2.071Z1M3G 19.000 7 2.714Z1F3I 26.750 7 3.821Z1F3J 15.250 7 2.179Z1M3K 37.000 7 5.286Z1F4B 23.500 7 3.357Z1F4D 27.250 7 3.893Z1M4E 16.000 7 2.286Z1 F4F 13.500 7 1 .929Z1 M4H 13.750 7 1 .964Z1F4I 9.500 7 1.357Z1M4M 36.500 7 5.214Z1F4R 11.250 7 1.607Z1M5B 32.000 7 4.571Z1M5C 9.250 7 1.321Zi F5D 28.250 7 4.036Z1F5F 21.500 7 3.071Z1M5Q 27.750 7 3.964213APPENDIX CIndividual MCMCA Scores and Level of Miscue--ComrarisonMiscue Rating LevelStudent MCMCA Score Miscue % Easy Instructional FrustrationalZ1M3K 10 1.2X1M5N 10 1.5Z1F3J 10 4.7 VX1F4K 10 1.9 VY1F5I 9 2.5X1M4N 9 0.4 VX1F4Q 9 0.4 VZ1M3G 9 1.2 VX1M4W 9 0.7 VZ1F3I 9 3.1 VY1M5A 9 1.1Z1M5B 9 2.0 VX1F3C 9 3.0 VY1F4C 8 0 VY1F3H 8 1.4 VX1M4U 8 4.8X1F3M 8 1.1 VZ1M5Q 8 3.1Z1F4D 8 0.8 VZ1F5D 8 3.1X1M4P 8 1.1 VX1M5J 8 0.4 VZ1F5F 8 5.9 VY1F4A 7 2.9 VX1M3E 7 1.1 VZ1M4M 7 1.6 VX1F5C 7 1.5 VX1M5G 7 0.4 VZ1M4E 7 15.6Z1M4H 7 7.8Y1M4L 7 9.6X1F3B 7 0.7 VX1F3L 7 1.5 VZ1F4B 7 7.0Y1F4S 6 7.1 VY1F5H 6 4.6 VY1M3N 6 3.9 V214(APPENDIX 0 continued)Miscue Rating LevelStudent MCMCA Score Miscue % Easy Instructional FrustrationalZ1M5C 6 7.0Y1F5M 6 1.8 1X1M5L 6 1.1 V1X1F4J 6 4.8 V1X1M5E 6 0.7Y1F3A 5 1.8 V1Y1M4O 5 14.3Y1M4T 5 6.1Y1F3D 5 11.1X1F4X 5 13.4 V1Z1F4F 5 12.1 V1Y1M4G 4 2.1 V1Z1F4R 4 5.1 V1Y1F5P 4 xaY1F4V 4 10.7 V’Y1M5K 3 10.7Z1F4I 3 24.6 V1Y1M3F 2 8.9xaToo many miscues to count.215APPENDIX PSpecific Traits Aggregate Summaries (Corn bined Categories) --Extreme GroupsGroup Student M/C N/S P/M TotalEGD 1 X2F5P 306 134 26 466Y2F4X/F4B 52 85 16 153Y2F4R 144 43 18 205Y2F3L 38 26 5 69Z2F3C/M3F 37 29 4 70Z2F3B 2 20 9 31Y2F4F 63 6 7 76X2F4V 32 16 7 55Y2F41 14 24 15 53Y2F5F 110 18 7 135Average 79.8 40.1 11.4 131.3EGD4 X2F4C/M4M 110 33 5 148X2M3K 153 67 16 236X2F51 151 49 6 206Z2M4U/M4G 59 38 13 110X2F4A 38 10 1 49X2F3H 28 19 1 48Y2M5Q/M5N 158 50 8 216Z2M5K 22 13 11 46Z2M4N 51 11 16 78Z2F3M 16 17 5 38Z2M3E 2 20 9 31Average 71.636 29.727 8.273 109.636EGI1 X1F4X 23 7 0 30Y1F3A 4 4 0 8Y1F3D 14 18 4 36Y1F4V 15 2 1 18Y1F5P 41 41 9 91Y1M3F 7 0 1 8Y1M4G 10 4 0 14Y1M4O 64 14 4 82Y1M4T 10 5 1 16Y1M5K 31 12 9 52Z1F4F 11 5 8 24Z1F4I 15 8 1 24Z1F4R 12 2 2 16Average 19.769 9.385 3.077 32.231216(APPENDIX P continued)Group Student M!C N/S PIM TotalEGI4 52 4 4 60X1F3CX1F4K 18 1 0 19X1F4Q 68 0 1 69X1M4N 36 9 1 46X1M4W 14 0 0 14X1M5N 58 6 1 65Y1F5I 51 8 21 80Y1M5A 71 3 8 82Z1F3I 31 4 0 35Z1F3J 27 4 2 33Z1M3G 25 9 2 36Z1M3K 74 9 7 90Zi M5B 46 56Average 43.923 52.69234.61574.154217APPENDIX QSpecific Traits Acjcirecate Percent Summaries (Combined Cateciories)--Extreme GroupsX2F5PY2F4XIF4BY2F4RY2F3LZ2F3C/M3FZ2F3BY2F4FX2F4VY2F41Y2F5FAverageX2F4C/M4MX2M3KX2F51Z2M4U/M4GX2F4AX2F3HY2M5Q/M5NZ2M5KZ2M4NZ2F3MZ2M3EAverageX1F4XY1F3AY1F3DYl F4VY1F5PY1M3FY1M4GY1M40Y1M4TY1M5KZ1F4FZ1F4IZ1F4RAverage65.66733.99070.24555.07252.8636.45282 .90658.18126.41781 .48060.77774.32464.83173.29953. 63677.55258.33373.15147. 82765.38342.1076.45265.34076.66650.00038.89083.33345.05687.50071 .42978.05262.50059.61445.83462.50175.00061 .33528.75555.55620.97637.68241 .42964.5177.89429.09245.28313.33330.54122.29628.38923.78734.54620.40939.58223.14928.26114.1 0244.73764,51727.11423.33450.00050.00011.11145.056028.57217.07431 .25023.07720.83433.33412.50029.1185.58010.4588.7817.2465.71529.0329.21012.72728.3015.1858.6823.3786.7802.91211.8182.04 12.0833.70423.91320.53913.15829.0327.5460011.1115.5569.8912.50004.8796.25017.30733.3344.16712.5009.547Group Student M/C% N/S% P/M%EGD1EGD4EGI1218(APPENDIX Q continued)X1F3CX1F4KX1F4QX1M4NX1M4WX1M5NY1F5IYl M5AZi F31Z1F3JZ1M3GZ1M3KZ1M5BAverage86.66894.73698.54978.263100.00089.23063.75086.58688.57181 .81869.44582.22182.14383.3586.6675.263019.56609.23010.0003.65911.42812.12225.00110.0005.3588.7586.66701.4492.17401.53826.2509.75706.0615.5567.77812.5007.884Group Student M!C% N/S% PIM%EG 14219APPENDIX PComposition of Extreme Groups EGI4 and EGD4, with Passage ReadEGI4 EGD4 PsageF3C XM3G ZF31 ZF3J ZF4K XF4Q XM4W XM5A YM5B ZM3K M3K Z/XM4N M4N X/ZP51 F51 YJXM5N M5N X(M3E ZF3H XF3M ZF4A XF4C XM4G ZM4M XM4U ZM5K ZM5Q Y220APPENDIX SPassage StatisticsPassageFeaturea X Y zNumber of Paragraphs 10 7 8Average Length (Sentences) 4.3 4.1 4.3Number of Sentences 43 29 35Average Length (Words) 13.9 21.2 15.2Longest Sentence Iords) 28 36 28Shortest Sentence ords) 2 9 5Passive Voice 3 6 3Short (<12 Words) 18 2 14Long (>24 Words) 3 8 4Number of Words 600 615 533Average Length (Letters) 4.38 4.75 4.60Average Length (Syllables) 1 .38 1 .52 1 .49Number of Little Words (1 Syllable) 425 416 347Percent 70.8 67.5 65.2Number of Big Words (3+ Syllables) 36 78 84Percent 6.0 12.7 15.8Number of Prepositions 72 84 74Flesch-Kincaid Grade Level 6 11 8aExclusive of titles, subtitles, and section headings.