Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

English as a second language (ESL) student self-concept : its relation to teacher perception and academic… Silver, Barbara Deborah 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_silver_barbara.pdf [ 8.78MB ]
JSON: 831-1.0086302.json
JSON-LD: 831-1.0086302-ld.json
RDF/XML (Pretty): 831-1.0086302-rdf.xml
RDF/JSON: 831-1.0086302-rdf.json
Turtle: 831-1.0086302-turtle.txt
N-Triples: 831-1.0086302-rdf-ntriples.txt
Original Record: 831-1.0086302-source.json
Full Text

Full Text

ENGLISH AS A SECOND LANGUAGE (ESL) STUDENT SELF-CONCEPT:ITS RELATION TO TEACHER PERCEPTION AND ACADEMIC ACHIEVEMENT -AN EXPLORATORY STUDYbyBARBARA DEBORAH SILVERB.A. (Psychology), McGill University, 1981A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF ARTSinTHE FACULTY OF GRADUATE STUDIES(Department of Educational Psychology and Special Education)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAAugust 1993© Barbara SilverIn 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.(SignatureDepartment of C A,V-t terliCGCCA-YU'l, The University of British ColumbiaVancouver, CanadaDate j cP^9 7)-/ /99.DE-6 (2/88)ABSTRACTThe purpose of this exploratory study as it applies to ESLstudents was two-fold: To examine the relationship between ESLstudent self-concept and teacher perception of student self-concept and to examine the extent to which student self-conceptand teacher perception of student self-concept relate to academicachievement in reading and mathematics. The Self-DescriptionQuestionnaire-1 (SDQ-i) was completed by 57 fifth- and sixth-gradestudents, ages 10-12 years and their teachers. The teachers'ratings of ESL student academic achievement were gathered alongwith student and school background information. Results ofdescriptive statistics, t-test comparisons, and correlationalanalyses indicate that the pattern of ESL students' self-conceptswere consistently lower than those for the normative sample,except in the area of math, with noted similarities in highest(parent relations) and lowest (physical appearance) self-conceptareas. Teacher perception ratings appeared more similar to theESL students' self ratings than to the normative sample. Teacherstended to rate ESL students' self-concepts higher than thestudents rated themselves. In contrast to previous research withthe SDQ-1, the strongest agreements between ESL students andteacher perceptions were in nonacademic areas. Correlationsbetween ESL students and academic achievements were mostlynonsignificant and negative, while correlations between teacherperceptions and academic achievement were mostly significant andpositive. Sex, ethnic group affiliation, school district, andbirth place differences were also noted. While conclusions remaini ispeculative, more research is needed in the area of languageproficiency, environmental factors, and cultural differences whichmay impact, not only the students' self-concept of themselves, butalso on teacher perceptions of ESL students' self-concept andtheir academic achievement.iiiTABLE OF CONTENTSABSTRACT^List of Tables^ viiiList of Figures ixAcknowledgements^ChapterA.B.I. INTRODUCTION^BACKGROUND TO THE PROBLEM^SELF ESTEEM LITERATURE1231.^Self-Concept and Teacher Perception^ 32.^Self-Concept and Academic Achievement^ 43.^Shavelson Model^ 5C. ENGLISH AS A SECOND LANGUAGE LITERATURE^ 7D. SELF-CONCEPT STUDIES WITH MINORITIES 8E. STATEMENT OF THE PROBLEM^ 9F. DEFINITION OF TERMS 101.^ESL^ 102.^Self-Concept^ 103.^Academic Self-Concept^ 114.^Academic Achievement 11G. SIGNIFICANCE OF THE STUDY 11H. SUMMARY OF CHAPTER I^ 13Chapter II. REVIEW OF THE LITERATURE^ 14A. SELF-CONCEPT^ 15B. SHAVELSON MODEL OF SELF-CONCEPT^ 17C. STUDIES IN SUPPORT OF THE SHAVELSON MODEL^ 201.^Structure of Self-Concept 21a. Multifaceted Hierarchy^ 21b. General Self-Concept 232.^Age and Sex Differences 253.^The Shavelson Model of Self-Concept andAcademic Achievement^ 31a.^Big Fish Little Pond Effect (BFLPE)...31ivb.^Internal/External Frame of Reference(I/E Model) ^ 36D.^SELF-CONCEPT AND TEACHER PERCEPTION^ 411. Theoretical Basis 412. Self-Other Agreement and SDQ Research^ 43E.^SELF-CONCEPT AND ACADEMIC ACHIEVEMENT^ 471. Self-Concept and Reading and Mathematics ^ 422. Self-Concept and Grades^ 51F.^ENGLISH AS A SECOND LANGUAGE STUDENTS ANDASSESSMENT^ 541. Language Proficiency^ 552. Second Language Teaching and AcademicAchievement^ 63a. Bilingual Programs^ 63b. ESL Program Evaluations^ 66c.^Schooling in Two Languages for LanguageMajority Students^ 68d.^Generalizations on Academic Achievementin a Second Language^ 693. ESL and Psychoeducational Assessment^ 714. ESL Students and Academic Achievement^ 725. Bilingualism and Cultural Identity^ 766. Testing and Placement Controversy 80G. SELF-CONCEPT AND MINORITY STUDIES^ 81H. SUMMARY OF CHAPTER II^ 84CHAPTER III. METHODOLOGY^ 86A. RESEARCH QUESTIONS^ 86B. SELECTION AND DESCRIPTION OF SUBJECTS 871.^Student Selection^ 872.^Student Description 893.^Student School Background^ 904.^The ESL Teachers^ 91C. INSTRUMENTATION^ 921. Self-Description Questionnaire-1 (SDQ-1)...922. Scoring^ 97D. DATA COLLECTION PROCEDURES^ 981.^Students^ 99V2. Teachers^ 993. Academic Achievement^ 101E.^DATA ANALYSIS 102CHAPTER IV.^RESULTS^ 104A.^DESCRIPTIVE STATISTICS^ 1041. Self-Concept Ratings 1042. ESL Self-Concept and NormativeComparisons^ 1053. Self-Concept and Teacher PerceptionComparisons 1084. Reading and Mathematics AcademicAchievement^ 109B.^PEARSON PRODUCT-MOMENT CORRELATION^ 1121. Correlations between ESL Self-ConceptScores^ 1132. Correlations between Teacher PerceptionScores 1163. Correlations between ESL Self-Concept andTeacher Perception^ 1194. Correlations between ESL Self-Concept andAcademic Achievement 1225. Correlations between Teacher Perception andAcademic Achievement^ 1246. Correlations between Reading and MathematicsAcademic Achievement^ 127C.^T-TESTS^ 1271. Sex Differences 1272. District Differences^ 1293. Birth Place^ 1304. Ethnicity 130CHAPTER V.^DISCUSSION AND CONCLUSIONS^ 132A.^DISCUSSION OF FINDINGS 1331. ESL Self-Concept^ 1332. Teacher Perception of ESL Self-Concept ^ 1373.^ESL Self-Concept and Teacher PerceptionRelationship^ 140vi4. ESL Self-Concept and AcademicAchievement^ 1485. Teacher Perception and AcademicAchievement 1516.^Other Significant Findings^ 153a. Sex Differences^ 153b. District Differences 154c. Birth Place Differences^ 156d. Ethnic Differences 157B. LIMITATIONS OF THE STUDY^ 159C. RECOMMENDATIONS FOR FURTHER RESEARCH^ 161REFERENCES^ 167APPENDIX A^LETTER TO TEACHERS AND PRINCIPALS^ 189APPENDIX B^PARENTAL/GUARDIAN AND STUDENT CONSENT FORMS ^ 192APPENDIX C^SCHOOL DEMOGRAPHIC QUESTIONNAIRE^ 197APPENDIX D^SELF-DESCRIPTION QUESTIONNAIRE-1 (SDQ-1) ^ 199APPENDIX E^TEACHER REPORT OF STUDENT BACKGROUND ANDGRADES^ 206APPENDIX F^DEFINITION OF SDQ-1 SCALES^ 208APPENDIX G^SDQ-1 SCORING AND PROFILE BOOKLET^ 212APPENDIX H^PEARSON PRODUCT MOMENT CORRELATIONS FOR ESLSTUDENT SELF-CONCEPT, TEACHER PERCEPTION ANDACADEMIC PERFORMANCE (24 x 24 Matrix) ^ 217viiLIST OF^TABLESTABLE 4-1 STUDENT, TEACHER, AND NORMATIVE COMPARISONS OFMEANS AND STANDARD DEVIATIONS ON THE SDQ-1^ 106TABLE 4-2 DISTRIBUTION OF READING AND MATHEMATICSACADEMIC ACHIEVEMENT SCORES^ 110TABLE 4-3 MEANS AND STANDARD DEVIATIONS FOR READING ANDMATHEMATICS ACADEMIC ACHIEVEMENT^ 111TABLE 4-4 PEARSON PRODUCT-MOMENT CORRELATIONS FOR ESLSTUDENT SELF-CONCEPT^ 114TABLE 4-5 PEARSON PRODUCT-MOMENT CORRELATIONS FORTEACHER PERCEPTION^ 117TABLE 4-6 PEARSON PRODUCT-MOMENT CORRELATIONS FOR ESLSTUDENT SELF-CONCEPT, AND TEACHER PERCEPTION ^ 120TABLE 4-7 CORRELATIONS BETWEEN ACADEMIC ACHIEVEMENT, ESLSELF-CONCEPT, AND TEACHER PERCEPTION ^ 123TABLE 4-8 MEANS, STANDARD DEVIATIONS, AND t-TEST VALUESFOR SIGNIFICANT DEMOGRAPHIC DIFFERENCESRELATED TO ESL SELF-CONCEPT AND TEACHERPERCEPTION^ 128vi i iLIST OF FIGURESFIGURE 2-1^STRUCTURE OF SELF-CONCEPT^ 19ixACKNOWLEDGEMENTSThe author wishes to acknowledge the committee members fortheir patience and support - Dr. David Whittaker for hisencouragement, Dr. Shelley Hymel for her careful critique, and Dr.John Allan for being there.The author is grateful to the teachers, principals, andstudents who participated in this study. Gratitude is alsoexpressed to the two school districts which allowed the study tobe conducted.In addition, the author would like to thank her friends andcolleagues for their tolerance and understanding. A special noteof gratitude goes out to Betty Ann Waddington for typing theinitial thesis draft and to Thomas Walker Arden for sharing hismother, Veronica Smith, who served as the author's nurturingcoach.x1I.^INTRODUCTIONThe development of a positive self-concept is assumed to beboth a goal and an outcome of the educational environment for allchildren:Schools can and do encourage young people to develop a senseof their own self-worth, to find personal challenge in theworld around them, to find satisfaction in their ownachievements, and to understand their own individuality...Their basic needs have not changed much across the centuries,whether those needs are expressed in elemental form in termsof food, shelter, or affection, or whether they are expressedas psychological needs for self-esteem and self-worth asstudents meet the challenges of everyday life... Thatyoungsters have always needed certain skills and bodies ofknowledge to operate successfully in society and to nourishtheir self-esteem is not in question... general agreementexists that the child's overall welfare should also beconsidered. Presentations to this Commission mentioned theimportance of children's social development, the developmentof their self-concepts, and the development of their moralstandards. Such educational goals are generally seen asnecessary to prepare students to live in a worldcharacterized by change and to help them adapt to this changein order to lead full and responsible lives.(The Report of the Royal Commission on Education, 1988, p. 70 ,77, & 78)While considerable attention in educational psychology hasfocused on the cognitive and instructional aspects of humanlearning, relatively few empirical studies have investigated theinfluence of affective factors in learning, achievement, andassessment in the area of ethnic minorities. Furthermore,although the importance of self-esteem has been an implicit andpersistent theme underlying multicultural and multilingualliterature in education, empirical research based on the dominantCanadian ethnic minorities is scarce.Recent interest in children's self-concept formation, itsdevelopment and relation to learning outcomes has resulted inissues such as the relationships among academic self-concept,inferred teacher perceptions of academic ability and specificschool achievement being addressed. However, the extent to whichthese variables are applicable to the learning outcomes ofchildren for whom English is a second language remains unanswered.A. BACKGROUND TO THE PROBLEMThe purpose of this investigation as it applies to English asa second language (ESL) students is two-fold:1. To examine the relationship between a student's self-concept and teacher perception of that student's self-concept; and2. To examine the extent to which student self-concept andteacher perception of student self-concept relate toacademic achievement in Reading and in Mathematics.Three bodies of knowledge need to be addressed, in order todeal with these issues:1. The self-concept or self-esteem literature;2. The ESL student assessment literature; and3.^Previous self-concept studies involving minorities.These areas will be discussed briefly in the followingsection and more explicitly in Chapter II.23B.^THE SELF-ESTEEM LITERATURE1.^Self-Concept and Teacher PerceptionControversy surrounds the relationship and use of self-concept and inferred ratings by teachers. Some theorists suggestthat self-concept is a "looking glass" reflection of perceptionsabout how one appears to others (Cooley, 1902; Kinch, 1963; Mead,1934; Shrauger & Schoeneman, 1979). This implies that self-perceptions and the perceptions of others should be stronglycorrelated and that changes in the perceptions of others will leadto changes in self-concept.Others suggest that self-reports and ratings by others arephenomenologically distinct and will agree only when the externalobserver knows the person well, observes a wide range ofbehaviors, has observed a broad enough sample of people to have anadequate frame of reference, and is able to make skillfulperceptions (Crandall, 1973; Marsh, Parker, & Smith, 1983; Marsh,Smith, & Barnes, 1984; Shavelson, Hubner, & Stanton, 1976; Wells &Marwell, 1976).The role of the teacher as significant other in the formationand development of children's school related self-concept appearsto have an effect on student attitude and success both in schooland in society:Schools also communicate subtle (and sometimes not so subtle)messages to ESL students regarding the value of their priorexperiences and the appropriateness of their language andculture within the Canadian context. Research suggests thatstudents who are valued by the wider society (and by theschools that inevitably tend o reflect that society) succeedto a greater extent than students whose backgrounds aredevalued.(Report of the External Review on the Vancouver SchoolBoard's ESL Program, 1989, p. 12)Furthermore, teacher ratings are probably the most importantsource of information about children in the school. Decisionsthat affect children are most often based on the evaluations andrecommendations of teachers in a variety of areas. These may beinfluenced by the teacher's perceptions of the student's self-concept.2.^Self-Concept and Academic AchievementThe importance of self-concept in academic achievement hasalso been of long debate (Burns, 1979; Hansford & Hattie, 1982;Marsh, 1986; Shavelson & Bolus, 1982; Wylie, 1979). Controversyhas focused around two major areas. Firstly, the causalrelationship between self-concept and achievement remainsuncertain (Calysn & Kenny, 1977; Scheirer & Kraut, 1979; Wylie,1979). Secondly, self-concept theorists continue to disagree asto the relative importance of global versus multidimensionaltheoretical models to define self-concept (Coopersmith, 1967;Harter, 1982, 1983, 1985; Marsh & Shavelson, 1983; Marx & Winnie,1978; Piers & Harris, 1969; Purkey, 1970; Shavelson et al., 1976;Soares & Soares, 1969).In the past, major consideration has been given to the roleof global self-concept and its relation to academic achievement.However, more recent research has offered support for themultidimensional and hierarchical nature of self-concept. Forexample, although the results of an extensive meta-analysis(Hansford & Hattie, 1982) suggested only a low positivecorrelation between self-concept and ability, the results of thestudy also reinforced findings that reported larger positive4correlations between self-concept of ability and achievementmeasures than those reported for global self-concept measures(Shavelson et al., 1976; Wylie, 1979).Several alternative multidimensional models of self-concepthave recently been put forth by theorists and investigators whoargue that a unidimensional or single-score approach may maskimportant distinctions that children make in various areas oftheir lives (e.g., Harter, 1982, 1983; Shavelson, Hubner, &Stanton, 1976). According to the model proposed by Harter (1982),the self is depicted as a profile of evaluative judgments acrossfive separate domains. According to the model proposed byShavelson et al.(1976) and later revised by Marsh and Shavelson(1985), academic achievement measures should be more highlycorrelated with academic self-concept than with general self-concept and academic achievement in particular areas should bemost highly correlated with self-concept in the same area; lesscorrelated with self-concept in other academic areas; and leastcorrelated with self-concepts in nonacademic areas. For thepurpose of the present study, the Shavelson model will be adopted.3.^The Shavelson ModelThe multidimensional and hierarchical model of self-conceptproposed by Shavelson et al. (1976) and revised by Marsh andShavelson (1985) will be adopted for the purpose of this study.Reasons for choosing this model include:1.^The multidimensional and hierarchical model appears toallow for a comparison of global and specific academic5self-concepts which may be obscured by the use of aglobal theoretical model alone.2. There appears to be strong support for this model in thecurrent self-esteem literature (Marsh, 1988a; Wylie,1989).3. The structure of this model differentiates among sevendomain specific areas. A second, more general level,defines self-concept in terms of academic English,academic Math, and nonacademic self-concepts. Generalself-concept is at the apex of the hierarchy. Recentpsychoeducational assessment research with ESL studentssuggests similar trends in the cognitive or IQ profileon the WISC-R (Cummins, 1984). The multi-dimensionaland hierarchical arrangement of self-concept may help toexplain the pattern of verbal-performance discrepancytypically found with ESL students (Lynn, Paglieri, &Chan, 1988; Tam, 1990).4.^The specificity of the model allows multiple andpossibly different self-concepts in various academic andnonacademic areas. It is suggested that themultilingual and multicultural child may have multiple"selves" and that choosing an identity may becomplicated by having grown-up in two cultures whosevalues are often very different (Lambert, 1981).Furthermore, it has been suggested by Lambert that ESLstudents who work out these conflicts:... retain pride in their home culture as well as inCanadian culture as a whole and feel able to identifywith both. Ideally, students will be able to see thestrengths and weaknesses of both cultures and use the6strengths of both as a foundation for choosing their ownvalues and identities. From the point of view oflanguage learning, these students are likely to bemotivated to develop their proficiency in both Englishand their Ll (first language).(Lambert, 1967 as cited in Cummins, 1981,p. 14-15).5. The possibility exists that this multifaceteddistinction may help teachers make more adequate self-concept judgements about their students withoutobscuring the results based on nonacademic inferences inareas about which teachers are generally less familiar.6. Also, findings derived from multidimensional self-concept ratings may yield significant and specificinformation about a student's needs, which, in turn, mayinfluence important programming decisions.C. ENGLISH AS A SECOND LANGUAGE LITERATUREA growing body of research over the last 20 years hascentered around the issues and policies related to second languageacquisition, testing, placement and counselling of Canadian ethnicminorities (Collier, 1987, 1989; Cummins, 1981, 1984, 1986;Samuda, 1975, 1985; Samuda & Crawford, 1980; Samuda, Kong,Cummins, Pascual-Leone, & Lewis, 1989). The use of psychologicaltests with immigrant minorities, the large proportion of ethnicminorities in special education classes and the assumptionsunderlying such testing and placement decisions have been of majorcontroversy (Samuda & Crawford, 1980; Samuda et al., 1989;Sattler, 1986; Tam, 1990).Studies suggest that depending on age of arrival, it takesfour to eight years, with an average of five, for limited English7proficient immigrant students to reach the national grade-levelnorms of native speakers. This occurs in all subject areas oflanguage and academic achievement as measured on standardizedtests (Collier, 1987, 1989; Cummins, 1981, 1984). Researchregarding language acquisition, proficiency, age relationships,and approximations to national grade level norms for Canadian-bornethnic minorities, however, appears to be sadly lacking although,according to 1988 survey by the External Review Team of VancouverSchool Board's ESL Programs, approximately 46.9 percent ofstudents within the Greater Vancouver area may be considered ESL.Recently, some of these factors have begun to be addressedparticularly in an attempt to clarify issues regarding thecognitive assessment of local Chinese immigrant students (Lynn,Paglieri & Chan, 1988; Tam, 1990).However, the importance of affective characteristics appears to begenerally overlooked in the ESL student assessment literature.D.^SELF -CONCEPT STUDIES WITH MINORITIESResearch with ethnic minorities and self-concept appears tobe scarce. Most investigations which have been reported havecentered around the ethnic identity of Afro-Americans in theUnited States. Findings are contradictory as to whethersignificant differences exist between the self-concepts of whitesand nonwhites at all levels of schooling (e.g., Goldman andMercer, 1976; Hansford & Hattie, 1982; Soares & Soares, 1969;Zirkel, 1971). Findings involving students from other minoritygroups are less numerous but also contradictory, Mexican-Americans8having received the most attention (Iheanacho, 1988; Wilkinson &Burke, 1985; Zirkel, 1971). These contradictions may be due todifferences of definitions, instruments and research designs. Inmany cases, the dominant culture has displayed possiblediscriminatory judgements directed toward the minority culture(Wilkinson & Burke, 1985).In Canada, self-concept as it relates to minorities is almostnonexistent (Akootie, 1984). The literature is dominated bystudies that concern themselves with examining the socialtransitions these groups face.E. STATEMENT OF THE PROBLEMTheoretical and empirical research suggests the importance ofself-esteem in academic achievement, specifically in secondlanguage acquisition. The literature also suggests a relationshipamong self-concept, teacher perception and academic achievementfor many students. However, studies with ethnic minorities whichinvestigate the relationship among and between these variablesappear to be limited at best. The questions to be explored thenare:1. What is the self-concept of the ESL student?2. What is the teacher perception of that ESL student'sself-concept?3.^What is the relationship between the ESL student's self-concept and teacher perceptions of that student's self-concept; and to what extent do they agree?91 04. What is the relationship between the ESL student's self-concept and the student's academic achievement inReading and in Mathematics; and to what extent do theyagree?5. What is the relationship between teacher perception ofthe ESL student's self-concept and the student'sacademic achievement in Reading and in Mathematics; andto what extent do they agree?E.^DEFINITION OF TERMS1. ESLEnglish as a second language, also multilingual ormulticultural, ethnic minority refers interchangeably to both theimmigrant and the Canadian-born student whose first language isnot the majority language, English, or who may have learnedanother language concurrently with English.2. Self-ConceptThe construct self-concept, broadly defined, is a person'sperceptions of him or herself. These perceptions are formedthrough one's experience with and interpretations of one'senvironment and may be influenced especially by reinforcements,evaluations by significant others, and one's attributions forone's own behavior. Self-concept is multifaceted andhierarchically organized. It is both descriptive and evaluativeand indistinguishable from self-esteem (Shavelson, Hubner, &Stanton, 1976).113. Academic Self-ConceptAcademic self-concept refers to the self perception of one'sacademic ability. It is separated into two distinct facets:verbal self-concept and math self-concept. A third facet,general-school self-concept also includes other school-relatedsubjects and combines with verbal and math self-concepts in theformulation of academic self-concept (Marsh & Shavelson, 1985). Aperson's definition of academic ability is believed to developover a series of learning experiences involving the judgments ofsignificant others concerning one's performance and capabilities.In turn, one's sense of academic self-worth helps to determine thedegree of enthusiasm and motivation a person invests in futureacademic tasks (Chapman & Boersma, 1980).4. Academic AchievementAcademic achievement refers to a measure or index of academicor school performance. It is usually associated with academicability and corresponds to a measure of learning outcome.G.^SIGNIFICANCE OF THE STUDYThe present exploratory study of ESL student self-concept hasinstructional, theoretical, and clinical relevance for educationalresearch and practice. It is especially significant because theinvestigation focuses on Canadian minority groups. Immigrantethnic minority representation has risen dramatically in BritishColumbia schools in the last decade. This trend is projected tocontinue for the 1990's, increasing the need for research in thisarea (Ashworth, Cummins, & Handscombe, 1989).Instructional research regarding teacher perception of ethnicself-concept may provide information that could facilitatecommunication and understanding between the ESL student and theteacher.Theoretically, the study may represent a potential test ofthe self-concept model proposed by Shavelson, Hubner, and Stanton(1976); Shavelson and Bolus (1982); and revised by Marsh andShavelson (1985). The present investigation uses the SelfDescription Questionnaire-1 (SDQ-1), a multidimensional self-concept instrument developed by Marsh (1988) and derived from thismodel. The current study reflects the need for research in thearea of minority self-concept and its relation to teacherperception and academic achievement.Clinically, this investigation may have implications forschool psychologists and counsellors. It could shed light on theimportance and interpretation of affective measures as part of apsychoeducational battery used for decisions regarding theassessment, placement and program evaluation of ESL students.This is important due to the current overrepresentation ofminorities in special education classes and the changingmulticultural/multilingual mosaic evidenced in British Columbiaschools (Report of the External Review on the Vancouver SchoolBoard's ESL Program, 1989; The Report of the Royal Commission onEducation, 1988).12H. SUMMARY OF CHAP TER IThe present study attempts to explore the issue of ESLstudent self-concept as it relates to inferred teacher perceptionof student self-concept and specific academic achievement. Thisinvestigation may provide insight to the extent these variablesare related and to the extent to which they relate to the learningoutcomes that contribute to the ESL student's school success.Chapter II reviews the literature pertaining to the researchquestions outlined in Chapter I.131 4II. ^REVIEW OF THE LITERATUREThis chapter deals with a review of the literature thatpertains to the following generalized problem statements:1. What is the relationship between the ESL student's self-concept and teacher perception of that student's self-concept?2. To what extent do the ESL student's self-concept andteacher perception of that student's self-concept relateto academic achievement in Reading and in Mathematics?In order to develop the literature review, the role andimportance of self-concept variables in school learning areconsidered. The theoretical model of self-concept conceived byShavelson and the subsequent development of the assessment tool,the Self Descriptive Questionnaire (SDQ), is briefly outlined andits use with various age groups is presented. Inferred ratings byteachers are discussed, as well as the relationship of self-concept and academic achievement. Current theory and research inthe area of bilingualism, second language acquisition, specialeducation, and the psychoeducational assessment of English as asecond language (ESL) students are considered. Previous studiesrelated to minority group self-concept are presented. The chapterconcludes with a brief summary of the research reviewed and astatement of the significance of the present investigation as itrelates to the areas of interest, specifically ESL student self-concept, its relation to teacher perception and academicachievement.A.^SELF CONCEPTIn the introduction to her 1989 review, Measures of Self-Concept, Wylie noted that:"...the flood of research directed toward phenomenological orconscious self-conceptions has continued, and self-esteem inparticular has been considered by both laypersons andprofessionals to be of great importance in accounting forhuman behaviour and to be a function of a very wide array ofvariables." (p. 2)The plethora of popular 'self-help' publications or numerousevaluations in the literature attempting to establish thepsychometric adequacy of measures of self-conception areindicative of the continued interest and importance placed onunderstanding the self-concept.The literature abounds with an array of terms that theoristshave constructed and employed to refer to the self-concept. Attimes, they refer to the same phenomena of self-concept and, attimes, to different phenomena. Terms and concepts such as self-love, self-confidence, self-respect, self-acceptance, self-evaluation, and self-worth are just some of the most commonlyemployed. In her review of measures of self-concept, Wylie (1974)attempted to devise a concept that would subsume the meanings ofthe above terms. She preferred to call it 'self-regard' as aninclusive term that would be less specific and less theory bound.However, the term self-esteem has dominated the literature as themost popular. Relative agreement has emerged that studies and15theories dealing with these concepts be referred to as the 'self-esteem literature'. On the basis of their review, Wells andMarwell (1976) also prefer the term self-esteem because theconcept is general enough to provide a type of common thread toencompass a diversity of approaches and styles. For the purposeof this discussion the term 'self-concept' has been chosen with anunderstanding that others may utilize different terminology.Self-theory is phenomenological in nature and is based on theprinciple that a person reacts to the world in a manner that isbased on his or her perception of it. An important feature of aperson's world is his or her own self, the self that he or sheperceives and experiences. The perceived self is what Fitts(1965) calls self-concept. There is a tendency to assign valuesto an individual's perception. It is the value that one places onthe perception of one's self that is called self-esteem. Wellsand Marwell (1976) argue that there is no universal definition ofself-esteem. Instead, there are differences in orientation andtheoretical emphasis. While some stress the behaviouralcomponents, others stress the cognitive ones. Some define self-concept in unidimensional terms and other in multidimensionalterms. Fitts (1965), Harter (1982, 1983, 1985), Shavelson et al.(1976), and Marsh and Shavelson (1985), for example, feel thatself-concept is complex and that a single scale is unwarranted andinadequate.For the purpose of the present study, the Shavelson model ofself-concept will be adopted. A brief explanation of thetheoretical basis underlying the Shavelson model, including adiscussion of its proposed multifaceted and hierarchical nature16and an outline of some of the studies conducted in support of themodel, will be presented in the following sections.B.^SHAVELSON MODEL OF SELF -CONCEPTHistorically, controversy and debate has surrounded most ofthe area composing the 'self-concept/self-esteem' literature.There has been uncertainty, not only about the relationshipbetween self-concept and achievement, but also about therelationship between self-concept and inferred ratings by others.Similarly, there has been disagreement about the directionality ofthese relationships, should causal relationships be indicated.There is also disagreement as to the relative importance of globalversus multidimensional models self-concept.Despite the controversy, reviewers of previous self-conceptstudies (Shavelson, Hubner & Stanton, 1976; Wylie, 1979; 1989)have attributed the general lack of consensus in the area of theself-esteem literature to the following: (a) imprecisedefinitions which lack a clear theoretical basis, (b) numerous andpsychometrically inferior instruments, and, (c) inappropriatemethodological procedures. Consequently, more recent researchershave sought to validate both the conceptual structure known asself-concept and the instruments designed to measure it (Byrne,1984; Marsh & Shavelson, 1985; Byrne & Shavelson, 1986; Shavelson& Marsh, 1986). Recent reviews of this research (Byrne, 1984;Marsh & Shavelson, 1985; Shavelson & Marsh, 1986) indicate thatself-concept cannot be adequately understood if itsmultidimensionality is ignored. In an attempt to remedy this17situation, Shavelson, Hubner, and Stanton (1976) reviewed specificcriteria for evaluating self-concept measures and proposed amultifaceted, hierarchical model which served as the basis for thepreadolescent Self-Description Questionnaire-1 (SDQ-1) used inthis study. The Self-Description Questionnaire-1 is an eightscale instrument intended to measure seven aspects of the self-concepts of preadolescent children (ages 7-13 years) as well astheir general sense of self-worth. The Self-DescriptionQuestionnaire-11 and the Self-Description Questionnaire-111 werelater developed to measure the self-concepts of older children andyoung adults.The Shavelson model (Shavelson, et al., 1976) and laterrevision by Marsh and Shavelson (1985) is represented in Figure 1.According to this model, self-concept, broadly defined, is aperson's perceptions regarding him or herself. These perceptionsare formed through experience with and interpretations of one'senvironment. They are especially influenced by the evaluations ofsignificant others, reinforcements, and attributions for one's ownbehaviour. Furthermore, self-concept is defined by seven majorfeatures:1. It is organized, or structured, into categories thatrelate to one another.2. It is multifaceted so that the particular facets reflecta self-referent category system adopted by a particularindividual and/or shared by a group.3.^It is hierarchical so that perceptions of personalbehaviour at the base move to inferences about self in18superordinate areas, and then to inferences aboutoneself in general.4.^General self-concept which is at the apex of the modelis stable but, as one descends the hierarchy self-concept becomes increasingly situation specific and, asa consequence, less stable.Figure 2-1Structure of Self-Concept 19Adapted from: Marsh, H.W. and Shavelson, R.J. (1985). Self-Concept: Itsmultifaceted, hierarchical structure. Educational Psychologist, 20, p. 114.205. Self-concept becomes increasingly distinct, ormultifaceted, with age as the individual moves frominfancy to adulthood.6. Self-concept is both descriptive ("I am happy") andevaluative ("I do well in mathematics").7.^It can be differentiated from other constructs such asacademic achievement.According to this revised, multifaceted and hierarchicalmodel, general self-concept (a higher-order factor) appears at theapex and is divided into academic and nonacademic self-concepts(also higher or second-order factors) at the next level. Academicself-concept is then broken down into two specific areas,Reading/English or Verbal and Mathematics. Finally, self-conceptsare broken down into seven first order factors - three particularsubject areas or factors (reading, math, general school) and fourspecific nonacademic areas or factors (physical ability, physicalappearance, peer relationships, and parent relationships)This model of self-concept implies that the closer to thebase of the hierarchy, the more situation specific becomes eachfacet of self-concept. Therefore, self-concept of academicability should be more closely related to academic achievementthan to ability in social and physical situations.C. STUDIES IN SUPPORT OF THE SHAVELSON MODELThe seven hypothesized factors, originally proposed in theShavelson model, have been replicated in more than a dozen factoranalytic studies of responses to the Self DescriptionQuestionnaire-1 (SDQ-1) by diverse populations of children andpreadolescents, lending strong support for the multidimensionalityand hierarchical nature of self-concept (Marsh, 1986a, 1987b;Marsh, Barnes, Cairns, & Tidman, 1984; Marsh, Relich & Smith,1983; Marsh & Smith, 1987; Marsh, Smith & Barnes, 1983, 1984,1985).1.^The Structure of the Self-Concept1.a.^Multifaceted HierarchyThe Shavelson model posits that self-concept is multifacetedand also hierarchically ordered. The factor structure was testedwith a single summary factor analysis performed on all 3,562responses to the SDQ-1 included in the normative data base.Responses to positively worded items from the original sevenfactors were used: eight positively worded items from each of theseven scales were divided into four item pairs and factor analyseswere performed on the 28 item pairs using SPSS program (See Marsh,Barnes, Cairns & Tidman, 1984; Marsh & O'Neill, 1984). Theresults clearly identified each of the seven SDQ-1 factors. Thelargest correlations occurred among the first three nonacademicfactors and between the General-School and the two other academicself-concept factors (Math and Reading), consistent with thehypothesized hierarchical ordering as proposed by Shavelson et al.(1976).Assumptions that self concept becomes more distinct with agewere tested in a series of exploratory factor analyses on a sampleof 658 students in grades two to five (Marsh, Barnes, Cairns, and21Tidman, 1984; Marsh & Hocevar, 1985; Marsh & Shavelson, 1985).Responses to the 28 item pairs were analyzed separately at eachgrade level. The results indicated seven factors which wereidentified at each grade level, and the factor loadings for eachvariable were consistently high on each factor it was designed tomeasure while low on the other factors. The factors wereespecially well-defined for grades four and five. The separationamong academic factors was less clear for grades two and three.At each grade level, highest correlations tended to be betweenGeneral-school and other academic factors and among the factorsmeasuring nonacademic dimensions. There was a consistent declinewith age in size of correlation among factors and the mediancorrelation among factors declined with increasing grade level(Marsh, Barnes, Cairns & Tidman, 1984). The Marsh et al.(1984)findings strongly support assumptions of a multifaceted,hierarchical self-concept, and subsequent interpretations based onthe SDQ scales by indicating that the scales measure distinctfactors that are related in theoretically defined ways.The Shavelson model and the SDQ-1 upon which it is basedassume a systematic hierarchical ordering of the self-conceptfacets which underlie the correlations among the first-orderfactors. However, data suggest that the higher order structureunderlying the SDQ factors may be more complicated than previouslythought. Shavelson hypothesized that the seven first-orderfactors would form two-second order factors, Nonacademic andAcademic. However, for example, Parent Relations highlycorrelated with some of the academic factors as well as with thenonacademic ones and, although Mathematics and Reading factors22were substantially correlated with the General-School factor, theywere nearly uncorrelated with each other.Four competing higher order factor models were developed totest alternative configurations of the higher order factorstructure underlying first-order factors to explain thehierarchical structure of the seven SDQ-1 factors. In separateanalyses of the data at each grade level, the second-orderacademic self-concept factors (Math Academic and Reading Academic)were posited to fit the data significantly better than any of theother second-order models (Marsh & Shavelson, 1985).Although the model is most consistent with the Shavelson etal. (1976) assumption that self-concept is hierarchically ordered,the particular form of this higher order structure appears morecomplicated than first proposed. Specifically, there seems to bea clear separation of Reading and Math self-concepts so that theycannot be incorporated into a general academic self-concept.1.b.^The General Self-ConceptThere are many different definitions of what constitutesgeneral self-concept. Largely, this construct is inferred and itis typically ill-defined and probably the basis of much of theconfusion surrounding the self-concept research (Marsh, 1986b;Marsh & Shavelson, 1985).Most early investigations regard self-concept as aunidimensional construct represented with a single, global scorecalled 'overall, total, or general self-concept' (Wylie, 1974,1979, 1989). The most common, an agglomerate self-concept, is avaguely defined total score for a broad cluster of diverse self-2 3report items which lack a coherent focus. Responses are oftenjust summed to form a total score. When a total score is used,considerable information is lost by averaging across scoresrepresenting reasonably independent facets of self-concept. Ifonly agglomerate self-concept is considered, themulitdimensionality of self-concept is ignored. Such a constructcannot be adequately characterized and is idiosyncratic to aparticular instrument. For example, Marsh and Smith (1982)suggest that this agglomerate use of self-concept is particularlydubious and probably led to many of the contradictory findingswhich abound in the self-concept research.A second use of the term general self-concept involves scalesthat are specifically designed to measure relativelyunidimensional constructs that are superordinate to specificfacets of self-concept. Items in such scales refer to a generalsense of self-worth or self-competence that could be applied indifferent areas rather than self-concept in specific facets. Thisis the approach used by Rosenberg (1965, 1979) and Harter (1982,1983) and the General-Self scales on the SDQ instruments developedfor use with older students.The third use includes the general self-concept that appearsat the apex of the Shavelson model and the general factor in thesecond-order factor analyses of the SDQ-1. Here, general self-concept refers to an inferred construct which is not directlymeasured (Marsh, 1986b; Marsh & Shavelson, 1985). Unlike theother two uses, this general self-concept cannot be tied to aspecific set of items but is an unobserved construct that, itself,is defined by unobserved constructs (i.e., it is a 'higher order24factor'). In this way, the General-Self infers a general, oroverall, positive self-perspective that is not specific to anyparticular facet of self-concept but could be applied to eachspecific facet of the self. It refers to a person's rating of himor herself as an effective, capable individual who is proud andsatisfied with the way he or she is. The original version of theSDQ-1 did not contain the General-Self scale. Its inclusion cameabout to provide a better basis of comparison between SDQ researchand other self-concept research.The use of the revised SDQ-1 by Marsh, Smith, and Barnes(1985) examined the responses of 559 fifth-grade students. Afactor analysis of the responses clearly identified each of theeight factors which the revised SDQ-1 was designed to measure.Although the General-Self factor was modestly correlated with eachof the other seven factors and more highly correlated withresponses to nonacademic than academic factors, it appearedreliably distinct from the other facets of self-concept.In the SDQ-1 manual (Marsh, 1988a) caution is advised ininterpretations based on the General-Self scale as neither thetheoretical nor the empirical basis for it is, as yet, wellestablished compared to the original seven facets. There is aneed for theoretical and empirical research to justify and defineoverall and general self-concept.2.^Age and Sex DifferencesThe examination of how self-concept varies with age and sexis frequently of concern to self-concept researchers. In her 1979review, Wylie summarized research prior to 1977 and concluded that25there was no convincing evidence for any age effect, positive ornegative, in overall self-concept within the age range of 6 to 50years. She also suggested that, although there was no evidencefor sex differences in overall self-concept at any age level, sexdifferences may be lost when items are summed to obtain a totalscore. Since then, however, other researchers have providedevidence of differences in self-concept related to both age andsex.Firstly, in the area of age related studies, Dusek andFlaherty (1981), in a longitudinal study of adolescent self-concept found no systematic age effects. Other research, however,suggests that self-concept may decline during preadolescence and,perhaps, early adolescence. It appears that the very positiveself-concepts of the youngest children are unrealistically highand become more realistic through additional life experiences.For example, Eshel and Klein (1981) in a cross-sectional study ofself-concepts in grades 1 to 4 found a sharp decline in generalself-concept with age. Other researchers have also reported asignificant age effect in self-perceptions of ability in differentareas. For example, Nicholls (1979) asked children, ages 6 to 12years, to rank their reading ability compared to their classmatesand found that rankings declined with age. Stipek and Stipek(1981) as well as Stipek and Tannatt (1984) found that children'sself-perceptions of their 'smartness' dropped between kindergartenand grade 3. Ruble, Boggiano, Feldman, and Loebl (1980) reportedthat self-ratings in a physical ability task were negativelycorrelated with age in grades 2 to 4. Meece, Parsons, Kaczela,Goff, and Futterman (1982) found that there was a steady decline26in mathematics self-concept between the junior high and highschool years and that the drop for females begins sooner and islarger. Marsh, Parker, and Barnes (1985) found that, for most ofthe SDQ-11 scales, high school student responses demonstrated adecline between grades 7 and 9, then levelled out, and increasedbetween grades 9 and 11. In an extensive longitudinal study,Bachman (1970) and Bachman, O'Malley, and Johnson (1978) alsofound that self-concepts may continue to increase during lateadolescence and early adulthood.Marsh (1985a) posited that very young children are egocentricand have consistently high and less differentiated self-conceptsin all areas. These self-concepts may be unrealistic andrelatively independent of any external criteria. As childrenbecome older, they incorporate more external information intotheir self-concepts so their self-concepts become more consistentwith external criteria. For most children this implies that self-concepts will decline with age in at least some areas, and thatacross a broad range of children, self-concepts will decline inall areas. As children incorporate more information about theiractual skills and abilities, as well as feedback from others,their self-concepts will also become more differentiated asposited by the Shavelson model and as observed in responses to theSDQ-1. This proposal is consistent with the decline inpreadolescent self-concepts with age, the increaseddifferentiation of self-facets with age, and the finding that asyoung children become older, their self-perceptions become morehighly correlated with performance, feedback, and other externalcriteria.27Dusek and Flaherty (1981) and Marsh, Barnes, Cairns, andTidman (1984) used exploratory factor analyses to demonstrate thatfactor structures derived from self-concept responses were similaracross age and sex. In the Marsh et al. (1984) study, students(N=658) were selected from grades 2 to 5. The researchers foundthat the effect of grade level was significant for all but theParent Relations scale where there was no age effect at all. Ingeneral, self-concepts were consistently high across the age rangeconsidered. For Peer Relations, self-concepts decreased fromgrades 2 to 4 but increased in grade 5. For each of the otherfive SDQ-1 factors and for all three total scores, self-conceptsdeclined moderately but consistently with increases in gradelevel.Secondly, in the area of sex related studies, Dusek andFlaherty (1981) reported differences in specific self-conceptsconsistent with sex stereotypes. Males had a higher self-conceptin masculinity, achievement, and leadership than females but lowerself-concepts in congeniality and sociability. In theirliterature review, Meece et al. (1982) reported that sexdifferences in math achievement and math self-concept are not aslarge in the elementary years. Generally, females do as well asmales on standardized tests of math achievement during elementaryand junior high school (Fennema, 1974; Sherman, 1980). However,some studies indicate that by early junior and late or senior highschool, females have lower levels of math achievement and self-concept. Meece et al. (1982) also indicate that by junior highschool and the high school years, females have lower math self-concepts than males. Furthermore, they suggest that as as28students go through high school, math self-concept declines butthat the decline starts sooner and is larger for females than formales. The authors imply that the decline in female math self-concept precedes the decline in math achievement and that thesocialization processes reflected in math self-concepts are onecause of the decline in achievement.Several Australian studies have found significant sexdifferences depending on age, the component of self-concept, andthe self-concept instrument (Marsh & Smith, 1982). Marsh, Relich,and Smith (1983) examined sex differences on the SDQ-1 forstudents in grades 5 and 6. They found that females had higherself-concepts in Reading and General-School and lower self-concepts in Physical Abilities, Mathematics, and PhysicalAppearance despite significant higher levels of math achievement.Similarly, in a study of 559 grade 5 students, Marsh, Smith, andBarnes (1985) found that females had significantly lower self-concepts than males in Physical Abilities, Physical Appearance,and Mathematics and significantly higher self-concepts in Readingdespite higher academic achievement as measured by objective testscores and teacher ratings of ability. When teacher ratings andtest scores were considered separately, teachers rated femalesreading ability to be higher and this sex difference was largerthan could be accounted for by differences in reading test scores.Being female also had a direct positive effect on math test scores(in that females scored higher), but not on teacher ratings ofmath ability. Similarly, being female had no direct effect onReading self-concept but had a negative effect on Mathematicsself-concept. In his dissertation, Relich (1983) also found that29sixth grade females had significantly lower Mathematics self-concepts than males, even though the females had higher levels ofmath achievement. In contrast, Marsh, Barnes, and Tidman (1984)found that sex differences did not vary across the age rangeconsidered in their study. Only moderate sex differences wereobserved for Physical Abilities (favouring males) and Reading(favouring females). The sex effects based on the normativesample (Marsh, 1985b) found the largest sex effects to beconsistent with previous research (i.e., Physical Abilities formales and Reading for females). No significant sex effects wereobserved on Parent Relations, General-School, and Total Academicself-concepts. Small effects, in favour of males, were observedon Physical Appearance, Peer Relations, Mathematics, and Total-Self self-concepts.Marsh, Parker, and Barnes (1985) conducted a large study ofadolescent responses to the SDQ-11 in grades 7 to 12 and foundsimilar sex differences to those observed in other SDQ research.In general, sex effects were small across all facets of self-concept and sex differences in specific facets were consistentwith sex stereotypes. Marsh, Byrne, and Shavelson (1989) examinedsex differences in senior school students using three differentself-concept instruments and found that, on all three scales,males had consistently higher math self-concepts whereas femaleshad significantly higher verbal self-concepts even aftercontrolling for school grades in English and in Mathematics.Stevenson and Newman (1986) also found that females had lower mathself-concepts. The researchers reported that grade 10 males had30more positive attitudes toward self in math than females, but thatfemales had more positive attitudes about reading.3.^The Shavelson Model of Self-Concept and AcademicAchievementResearch indicates that academic self-concepts are at leastmoderately correlated with corresponding levels of academicachievement, although the correlations almost never approach thereliabilities of the respective measures (Marsh, 1988a). However,although math and reading achievement may be significantlycorrelated with each other, Marsh and his colleagues suggest thatthe corresponding self-concepts - Academic Math and AcademicReading - are consistently uncorrelated across age and sex andacross academic and nonacademic settings. This implies thatacademic self-concepts reflect more than simple academicachievement. In order to explain these seemingly paradoxicalresults, two theoretical frame of reference models - the Big FishLittle Pond Effect (BFLPE) and the Internal/External (I/E) frameof reference model were proposed (Marsh, 1984b, 1984c, 1986d,1987; Marsh & Parker, 1984).3.a.^Big Fish Little Pond Effect (BFLPE Model)The BFLPE occurs if equally able students have loweracademic self-concepts when they compare themselves to more ablestudents and higher academic self-concepts when they comparethemselves with less able students (Marsh, 1984b, 1984c, 1987a;Marsh & Parker, 1984). The frame of reference model, designed toexplain the BFLPE, hypothesizes that students compare their own31academic ability, more or less accurately perceived, with theirperceptions of the academic ability of other students in theirimmediate reference group. Students then use this relativisticimpression as one basis for forming their own academic self-concepts. The BFLPE suggests, therefore, that average-abilitystudents will have higher academic self-concepts in low-abilityschools than in high-ability schools. For example, if an average-ability student attends a high-ability school, his level ofacademic ability will be below the average ability of otherstudents in that school, leading to an academic self-concept thatmay be below average. On the other hand, if an average-abilitystudent attends a low-ability school, the same level of academicability will be above the average of other students in that schooland may then lead to an academic self-concept that is aboveaverage. Therefore, the academic self-concepts of these studentswill depend on their objective academic ability and will also varywith the type of school they attend. According to this model,academic self-concept will be positively correlated withindividual achievement and variables relating to it (e.g., familySES) but negatively correlated with school-average achievement andvariables related to it (e.g., school-average SES/ability).For example, counter to popular assumption that disadvantagedchildren were likely to have lower self-concepts, several earlystudies in the United States found that students in lowsocioeconomic status (SES)/low-ability schools tended to havehigher self-concepts than students in high SES/high-abilityschools (Soares & Soares, 1969; Trowbridge, 1970, 1972). Thesestudies were based on school-average SES rather than individual32SES. Brookover and Passalacqua (1981) also reported that,although individual academic achievement was positively correlatedwith individual measures of academic self-concept, school-averagemeasures of academic achievement were negatively correlated withself-concept.In a more recent study, Marsh and Parker (1984) sampledsixth-grade classes selected from both high and low SES/abilityschools within the same city. The effects of both SES variablesand student academic ability on academic self-concepts and teacherinferred academic self-concepts were examined. The direct effectof academic ability on academic self-concept was positive, but thedirect effect of school-average academic ability/SES was negative.The findings suggest that being in a low-ability school may resultin a higher level of academic self-concept, even though it mayalso result in a somewhat lower level of academic achievement.In all three studies, the negative effect of school SES ontotal self-concept was statistically significant but small (r'sbetween -.07 and -.13), whereas the negative effect of school SESon academic self-concept was much stronger when family SES andstudent academic ability were controlled (-.36 controlled versus -.08 uncontrolled). Furthermore, despite the fact that teacherratings of student self-concepts were substantially correlatedwith student self-concepts, teachers in high-ability/SES schoolsjudged both the academic and nonacademic self-concepts of theirstudents to be higher (Marsh & Parker, 1984). The findingssuggest that teachers and students may be using different framesof reference with which to form their respective judgements. Aspredicted from the frame of reference model, students appeared to33be comparing themselves with other students in their own school,whereas teachers tended to use a broader, more absolute frame ofreference. Also, consistent with other SDQ-1 research, it appearsthat teachers less clearly differentiate between academic andnonacademic self-concepts than do their students.The frame of reference model used to explain the BFLPEpredicts that the size of the effect will vary according to thevariability of school-average ability. The Marsh and Parker(1984) study selected schools that appeared to be extreme in termsof school-average ability, perhaps increasing the size of theBFLPE in their study. In contrast, Bachman and O'Malley (1986)excluded all nonwhite students and predominantly nonwhite schoolsin a reanalysis of their earlier Youth in Transition data (Bachman& O'Malley, 1977). This exclusion may have decreased thevariability of school-average ability in their sample whencompared to a reanalysis of the same Youth in Transition data byMarsh (1987a).By ranking a group of students in terms of their academicachievement across the whole group and within their ownclassrooms, Rogers, Smith and Coleman (1978) found that thewithin-classroom rankings were more highly correlated with self-concept.Strang, Smith and Rogers (1978) compared the effects ofsegregation versus mainstreaming on the self-concepts ofacademically disadvantaged children. The results indicated thatlower self-concepts were reported for the academicallydisadvantaged group in the regular or mainstreamed classrooms.Schwarzer, Jerusalem, and Lange (1983) examined the self-concepts34of West German students who moved from nonselective primaryschools to secondary schools that were streamed on the basis ofacademic achievement. Although students chosen to enter the high-ability schools had substantially higher academic self-concepts atthe point of transition, academic self-concepts of both high andlow ability groups did not differ by the end of the first year inthe new schools. In a meta-analysis of studies on the effect ofhomogeneous ability grouping on self-concept, Kulik and Kulik(1982) found that high-ability students tended to have higherself-concepts in classes streamed according to similar abilitieswhen compared to students in homogeneous or unstreamed classes(see also Kulik, 1985; Marsh, 1984b). In a study of the careerdecisions of college males, Davis (1966) proposed a model similarto the BFLPE to explain why the academic quality of a college hadso little effect on career choice.In the longitudinal High School and Beyond study, Marsh(1988b) examined a variety of academic outcomes (e.g.,standardized examination performance, academic self-concept,selection of advanced course work, time spent on homework, qualityof academic effort, school grades, and post secondary attendance)in a nationally representative sample of students. The influenceof school average ability was not found to be positive for any ofthe 14 variables considered and the largest negative effectappeared to be on academic self-concept. Marsh concluded that theacademic outcomes produced by attending high-ability schools werenot even commensurate with the initial high ability levels ofstudents who attended these schools and that no academicadvantages of such schools were observed for the variables35considered. Marsh also suggested that it may be unjustified toassume that attending high-ability schools will necessarily resultin any academic advantages.Marsh (1984a) described a dynamic equilibrium model in whichacademic achievement, academic self-concept, and attributions forthe causes of academic success and failure are interwoven in anetwork of reciprocal interactions, such that a change in any onearea will produce changes in the others in order to reestablish anequilibrium. Thus, students moving from a low-ability school to ahigh-ability school might lower their academic self-concepts,improve their academic performance, change their academicattributions so as to protect their previous academic self-concepts, or use various combinations of these strategies.3.b.^Internal/External Frame of Reference (I/E Model)The Internal/External (I/E) frame of reference model (Marsh,1984, 1986d) describes the relationship between Reading andMathematics self-concepts and between these academic self-conceptsand verbal and math achievement. According to the model, studentsformulate their Reading and Math self-concepts in relation to bothexternal and internal comparisons or frames of reference.Achievement/ability measures in verbal and math areastypically correlate from .5 to .8. However, research based onresponses to the three SDQ instruments by various age groups, fromyoung children through to young adults and across the sexes, haveconsistently found little correlation between Reading and Mathself-concepts (Marsh, Parker, & Smith, 1983; Marsh, Smith, Barnes& Butler, 1983; Marsh & Groundwater-Smith, unpublished manuscript;36Marsh, Relich & Smith, 1983; Marsh et al., 1984; Marsh, Smith etal, 1985; Marsh, Parker & Barnes, 1985; Marsh, Barnes & Hocevar,1985; Marsh & O'Neill, 1984; Marsh, Richards & Barnes, 1986b).Marsh (1986d) proposed a theoretical model to explain thelack of correlation between verbal and math self-concepts and thisfinding also led to a revision of the Shavelson model (Marsh &Shavelson, 1985; Shavelson & Marsh, 1986). In the revised model,self-concepts in particular subject areas are believed to formAcademic Verbal and Academic Mathematics self-concepts.According to the Internal/External (I/E) frame of referencemodel, students formulate their Reading and Math self-concepts inrelation to both external and internal comparisons or frames ofreference. Internal comparisons occur when students compare theirown self-perceived abilities in different subjects (e.g., "I ambetter at math than at reading"). External comparisons operatewhen students compare their perceptions of their own abilitieswith the perceived abilities of other students within theimmediate context (e.g., "I am poor at both reading and math formy class"). In other words, although a student may accuratelyperceive him/herself to be below average in both math and readingskills (an external comparison), he/she may be relatively betterat math than at reading or other school subjects (an internalcomparison). Depending upon how these two components areweighted, the student may have an average or even above-averageMath self-concept despite his/her actual academic skills.Since math and reading abilities are compared with eachother, it is the difference between math and verbal skills thatcontributes to a higher self-concept in one area than in the37other. Marsh (1986d) suggests that the direct effect of readingachievement on Math self-concept and the direct effect of mathachievement on reading self-concept may then be significantlynegative. Subsequently, it is the joint operation of bothprocesses, depending on the relative strength of each, leads tothe near zero correlation between Reading and Math self-conceptsthat has been observed in the research with correlations ofpractical significant observed only for second and third gradestudents (Marsh 1986d).Although the I/E model does not require a near zerocorrelation between Reading and Math self-concepts, it doesrequire that the correlation be substantially less than thetypically large correlation between verbal and math achievementlevels. These findings (Marsh, 1986d) demonstrate a clearseparation between Math and Reading self-concepts as they are muchmore distinct than corresponding measures of academic achievementin the same two areas. The results also demonstrate that academicself-concepts in various areas may be affected by differentprocesses than are achievement measures.The research reviewed (Marsh 1984, 1986d) also questions theambiguous role and usefulness of general academic self-conceptwhich apparently cannot adequately reflect the diversity ofspecific academic self-concept facets. If the role of academicself-concept is to better understand the complexity of self in anacademic context, predict academic behaviours and accomplishments,provide outcome measures for academic interventions, and relateacademic self-concept to other constructs, then the specific38facets of academic self-concept may be more useful than a generalacademic facet.In his review, Marsh (1986d) reported that, in contrast tothe self-report data, a different pattern of results was observedin respect to inferred self-concepts based upon teacher and peerresponses and that there appeared to be no evidence that theinternal comparison process was operating. Academic self-conceptsinferred by teachers were highly correlated with objectivemeasures but they did not appear to accurately reflect therelativistic nature of self-concepts embodied in the internal andexternal comparison processes used by students in forming theirown self-concepts. This suggests that the external process maynot operate in the same way in the formation of self-conceptinferred by teachers and those based on the student's own self-reports. The findings also demonstrate that the formation of astudent's self-concept may be affected by different processes thanthose affecting the self-concepts inferred by significant others.The external frame of reference is similar to the processdescribed earlier to explain the BFLPE and is supported by theBFLPE studies. It is also believed that this process is used byexternal observers to infer the self-concepts of others. Althoughthe present I/E model emphasizes specific academic abilities andself-concepts, it may be likely that a similar process acts inother areas as well, such as in the formation of nonacademic self-concepts.Support for the I/E model and the SDQ-1 research haspractical implications for educators at all levels. If teachersare able to more accurately infer the academic self-concepts of39their students and better able to understand how they are formed,then their ability to provide positive reinforcement to studentsof all ability levels will be enhanced. Even though teachers seemto be able to infer self-concept in academic areas with at leastmoderate accuracy, there appears to be several biases in theirinferences. It seems unjustified to assume that academically weakstudents will necessarily have poor academic self-concepts in allsettings and in all subject areas. For example, Marsh (1988a)suggests that:1. Students in settings where other students are alsoacademically weak will have higher academic self-concepts than they would in settings where otherstudents are average or above average. Previous SDQresearch (Marsh, 1988a) suggests that teachers emphasizeabsolute measures of academic achievement in inferringacademic self-concepts of their students and largelyignore the particular setting which establishes theframe of reference for the student's own development oftheir self-concept.2. Inferred ratings by significant others (teachers andpeers) overemphasize the external comparison of studentacademic skills and underemphasize differences in skillsin particular academic areas. Therefore, a student whois weak in both math and verbal skills, but is strongerin one area than in the other, will tend to have muchlarger differences in his own self-report for Reading40and Math self-concept than is reflected in the self-concept inferred by teachers.3.^The model predicts that nearly everyone will feelreasonably good in at least some areas.D.^SELF-CONCEPT AND TEACHER PERCEPTION1.^Theoretical BasisSymbolic interactionists argue that self-concept emerges froma person's social interaction with others, that self-concept isbased on the ways others respond to the individual, and that aperson's perceptions of others responses reflect their actualresponses (Kinch, 1963). Symbolic interactionists argue thatsupport for the theory also requires a "congruence between (a)self-perceptions and others' actual perceptions of the person and,(b) perceived other-evaluations and actual other-evaluations"(Shrauger & Schoeneman, 1979, p.552).Research involving ratings by others may be formulated in avariety of ways. For example, external observers may be askedwhat they think or feel about a person. On the other hand,external observers may be asked to use their observations to inferwhat that person thinks about himself or herself (i.e., inferredself-concept). The second approach, which utilizes inferred self-concepts, is often used in self-concept research as self-conceptis based upon self-perceptions, whether accurate or not (Wells &Marwell, 1976). The distinction between these two types ofratings, however, is not always clear in self-concept research,and perhaps not to the external observers even when researchers41ask for inferred self-concepts. In the previous SDQ researchdescribed in this chapter and in the present study, ratings byothers refer to inferred self-concepts.There is also disagreement as to the relevance of inferredself-concept ratings even among researchers who agree that ratingsby others should be inferred self-concept ratings. Combs, Soper,and Courson (1963) argue that self-report measures of self-conceptare affected by sources of bias and that inferred ratings byexternal observers provide a more objective measure of self-concept. Others (e.g., Crandall, 1973; Marsh, Smith, Barnes &Butler, 1983; Shavelson et al., 1976) argue for the theoreticalseparation of self-concept based on a person's own self-report andinferred self-concepts based on the reports of others. Crandall(1973) suggests that ratings of others may be useful to validateor even to supplement self-report measures. Marsh argues thatratings by others are phenomenologically distinct from self-reports of self-concept and that the two will agree only if theexternal observer knows the subject well, observes a wide range ofbehaviors, has viewed a range of different subjects, and makesjudgements of the same specific characteristics as the subject.According to the Shavelson model (1976), self-concept isinfluenced by the evaluations of significant others but isconceived of as a different construct from self-concept inferredby external observers. Although there may be self-other agreementfor very specific self-concepts, particularly near the base of thehierarchy, the "correspondence between observer and the selfdecreases as one moves up the self-concept hierarchy" (p. 412).42Shrauger and Schoeneman (1979), based on an extensive reviewdesigned to test implications of Shavelson's (1976) theory,concluded that there is little consistent relationship betweenself-ratings and the ratings by others. Shrauger and Schoeneman(1979) reviewed studies that correlated self-reports withjudgements by others. However, the content of the self-reports intheir review was quite varied, no attempt was made to determine ifsome external observers (e.g., teachers, parents, peers) providedmore accurate assessments than others, and the distinctiveness ofdifferent components of self-concept was not considered whenmultiple characteristics were judged. Also, no distinction wasmade between studies that asked external observers to record theirown perceptions and those in which observers made inferred self-concept ratings.2.^Self-Other Agreement in SDQ ResearchA series of multitrait-multimethod studies by Marsh, Parker,and Smith (1983) and Marsh, Smith, and Barnes (1983, 1984)demonstrated significant agreement between multiple self-conceptsand inferred ratings by elementary school teachers and thestudents' peers. The average of the 56 convergent validitiesrepresenting self-other agreement in the above-mentioned studiesusing the SDQ-1 was .30 across all scales excluding the General-Self scale. Correlations in these studies were highest foracademic self-concepts and Physical Abilities and lowest forParent Relations and Physical Appearance. The results appear tosupport the convergent validity of the SDQ-1. Student-teacheragreement tended to be highest in the academic areas, where43teachers may most easily make relevant observations, and lowest onParent Relations, the area least observable for teachers.Self-other agreement differed markedly only in two areas:Parent Relations and Physical Appearance (Marsh, 1988a). Therelative lack of agreement on Parent Relations was expected sincethis is the area in which teachers and peers are least likely tohave an adequate basis for accurately inferring self-concepts.Lack of agreement on Physical Appearance was more surprising.Marsh and his colleagues (1988a) suggest that, perhaps, thestandards used by teachers as a basis for this inference aredifferent from those used by students. However, the studies notethat even student-peer agreement on this factor was among lowestof any of the scales, suggesting that students may be usingidiosyncratic standards in forming their own Physical Appearanceself-concepts and that these standards may not generalize even tothose that they use in making ratings about one of theirclassmates. Soares and Soares (1977, 1982) also demonstratedsignificant self-other agreement using multitrait-multimethodanalysis while providing evidence for the distinctiveness of thedifferent facets of self-concept.In another multitrait-multimethod study (Marsh, Barnes &Hocevar, 1985; Marsh & O'Neill, 1984), university studentsresponded to the SDQ-111. In addition to the self-ratings providedby the students, the person who knew each student best providedinferred self-concepts by also completing the SDQ-111 as if he orshe were the student. Over half of the students chose theirparents as the person who knew him or her best. Separate factoranalyses of both self-ratings and responses by significant others44identified the 13 dimensions of self-concept which the SDQ-111 isdesigned to measure. For each set of responses, internalconsistencies of all scales were high, whereas any correlationsamong factors were close to zero. Self-other agreement was quitehigh (mean r=.58) demonstrating that significant others are ableto accurately infer the multidimensional self-concepts of someonewhom they know well and supporting the validity of interpretationsbased on responses to the SDQ-111 for older students.In contrast to the results of self-other agreement studiesbased on either teachers or peers reports, significant others inthe SDQ-111 study were predominantly parents who accuratelyinferred self-concepts in academic and nonacademic areas.Agreement on the Parent Relations scale was particularly strong(r=.76) and agreement on Physical Appearance was substantial(r=.50) but still below average for all traits (r=.58). Theresearchers suggest that it is possible that respondents in theuniversity study were using internal standards that were moresimilar to those used by significant others. According to Marsh(1988a), possible explanations for such strong self-otheragreements in the SDQ-111 study include the following:1. Subjects were older and therefore knew themselves betteror based their self-responses on more objective andobservable criteria.2. Both subjects and significant others responded to thesame well-developed instrument.45463. Self-other agreement was for specific characteristicsrather than for broad, ambiguous characteristics or anoverall self-concept.4. The significant others knew the subjects better and in awider range of contexts than the observers in mostresearch.Only two of the SDQ-1 studies by Marsh, Smith, and Barnes(1984) included the General-Self scale. Self-other agreement onthis scale was the lowest of all the scales for self-conceptsinferred by peers and among the lowest for self-concepts inferredby teachers. Marsh et al. (1984) imply that this finding may beconsistent with the Shavelson's suggestion that self-otheragreement will be smallest for facets near the apex of thehierarchy where self-concept is less clearly tied to observablebehavior. Also consistent with this suggestion is that self-otheragreement was somewhat poorer for the General-School scale (meanof eight studies r=.333) than for the remaining Math and Verbalacademic scales (mean r=.37, respectively). The results of theSDQ-111 multitrait-multimethod study (Marsh et al., 1985) furthersupport these findings as the lowest self-other agreement was alsofor the General-Self and General-Academic scales.According to Wells and Marwell (1976), interpreting self-other agreement in terms of theory and previous research isdifficult because of the various types of inferred ratings used indifferent studies. The above-mentioned studies (Marsh et al.,1984, 1985; Marsh & O'Neill, 1984) appear consistent with theShavelson model, especially the prediction that self-otheragreement would be weaker for self-concepts close to the apex ofthe hierarchy. The significant others in these studies were askedto respond as if they actually were the subject which appears tobe the appropriate question to ask in order to determine theability of significant others to infer self-concepts and followsfrom the definition of self-concept as a person's self-perceptions.E.^SELF -CONCEPT AND ACADEMIC ACHIEVEMENTWylie (1979) noted that "many persons, especially educators,have unhesitatingly assumed that achievement and/or abilitymeasures will be strongly related to self-conceptions ofachievement and ability and to overall self-regard" (p. 355).According to the Shavelson model (Shavelson et al., 1976; Marshand Shavelson, 1985) and the SDQ-1 (Marsh, 1988a) upon which it isbased, self-concept is a multifaceted and hierarchically orderedconstruct. Therefore, the model predicts that academicachievement will be more positively correlated with academic self-concept than with nonacademic self-concept or overall self-concept, and that verbal and math achievement indicators will bemore highly correlated with academic self-concepts in matchingcontent areas than with other facets of self-concept.1.^Self-Concept and Reading and MathematicsHansford and Hattie (1982), in the most extensive meta-analysis of the achievement/self-concept relationship, found thatmeasures of ability correlated about .2 with measures of general47self-concept but about .4 with measures of academic self-concept.Similarly, Shavelson, and Bolus (1982) found that grades inEnglish, math and science were more highly correlated withmatching areas of self-concept than with general self-concept.Bachman (1970) reported that IQ correlated .46 with academic self-concept and .14 with general self-concept. Byrne (1984), in herreview of studies relating self-concept to academic achievement,found that nearly all studies report self-concept to be positivelycorrelated with academic self-concept and that many studies foundachievement to be more strongly correlated with academic self-concept than with general self-concept. These findings appear tosupport interpretations based on SDQ and indicate the need todistinguish among academic, nonacademic and general self-concepts.SDQ research has emphasized the distinctiveness of self-concept, particularly in verbal and math content areas, and it hasexamined extensively the relationships among reading/verbalachievement and Reading and Math self-concepts. A variety ofstudies examined the relationship between SDQ responses andacademic achievement (Marsh, Parker & Smith, 1983; Marsh, SmithBarnes & Butler, 1983; Marsh, Smith Barnes & Butler, 1983; Marsh,Relich & Smith, 1983; Marsh, Smith & Barnes, 1984; Marsh &Richards, 1986; Marsh & Gouvernet, in press). Academicachievement in these studies included both objective tests andteacher ratings of achievement, including verbal, math, andgeneral academic achievement indicators. Overall, correlationsbetween the SDQ scales and the academic achievement indicators inthese 11 research samples supported a dramatic distinction betweenacademic and nonacademic facets of self-concept and they48demonstrated the clear separation of Math and Reading self-concepts.Marsh, Relich, and Smith (1983), for example, demonstratedthat math achievement was substantially correlated with Math self-concept (r=.55), less correlated with self-concept in otheracademic areas (Reading r=.21; General-School, r=.43), and nearlyuncorrelated with self-concept in nonacademic areas. A similarpattern of correlations was observed for the other studies (Marsh,1988a).Tests of verbal achievement and teacher ratings of readingachievement were the most frequently used achievement indicatorsin SDQ research (e.g., Marsh, 1988a; Marsh, Parker & Smith, 1983;Marsh, Smith, Barnes & Butler, 1983). Of the 17 correlationsbetween Reading self-concept and verbal achievement indicatorsreported across the studies, all 17 were statisticallysignificant, ranging from .18 to .57 (median=r.40). However, themedian correlation between these same verbal achievement measuresand Math self-concept was.04 and only four of 17 correlations weresignificant. The 17 correlations between reading achievementindicators and General-School self-concept varied from .4 to .52(median r=.21) and 12 were statistically significant. In summary,these results indicate that reading achievement indicators aremost highly correlated with Reading self-concept, less correlatedwith the General-School scale and with Math self-concept, anduncorrelated or negatively correlated with self-concepts innonacademic areas.Math achievement indicators have been collected in fewerstudies (Marsh, 1988a). Nevertheless, the 13 correlations between49math achievement and Math self-concept varied from .17 to .66(median r=.32) and all were statistically significant. However,the 13 correlations between math achievement and Reading self-concept varied from -.01 to .36 (median r=.12) and only six werestatistically significant. The 13 correlations between mathachievement and General-School self-concept varied between -.02and .59 (median r=.26) and 11 were significant. In summary, mathachievement indicators were most highly correlated with Math self-concepts, less correlated with Reading self-concept, anduncorrelated or negatively correlated with self-concepts innonacademic areas.Marsh (1986d) reanalyzed the data on Verbal and Math self-concept of previous studies and their relationship using the SDQ-1, 11, and 111. The studies employed different reading andmathematics measures, including objective test scores, teacherratings, and school performance. Results of the 13 re-analysesindicated that correlations between indicators of verbal and mathachievement were substantial and ranged from .42 to .94.Correlations between measures of Reading/Verbal and Math self-concepts, on the other hand, were smaller and ranged from -.10 to.19. Also, three of the 13 correlation estimates, based onunweighted scores, were positive and ranged from .10 to .19whereas none of the other estimates, based on factor scores,reached statistical significance and ranged from -.10 to .20.Marsh (1988a) suggests that, perhaps, the different areas of self-concept may be more clearly differentiated by factor analyticallyderived scores than by unweighted scores.50The relationships between Reading/Verbal self-concept andverbal achievement and between Math self-concept and mathachievement were positive and statistically significant in all 13analyses (Marsh, 1986d). In contrast, the relationship betweenmath achievement and reading/verbal self-concept and betweenverbal achievement and math self-concept were all negative andstatistically significant except for one study (Marsh, Parker &Barnes, 1985) in which grade 11 and 12 students' achievement wasbased on self-assigned ability groupings reflecting studentinterest and future educational plans.Marsh, Bryne, and Shavelson (in press) tested the I/E modelwith grade 11 and 12 Canadian students (n=991) to three differentacademic self-concept instruments including the SDQ-111. Despitesubstantial correlations between school performance measures inmath and English (r=.51), correlations between Math and Verbalself-concepts from the various instruments varied from -.05 to .08and the correlation between total scores across the threeinstruments was zero.2.^Self-Concept and GradesChapman, Cullen, Boersma, and Maguire (1981) investigated theinterrelationships between general and academic self-concept,academic locus of control, and self-perceptions of future academicperformance in a sample of 376 elementary school children ingrades 3 to 6. The results showed that academic self-concept andexpectations of future performance correlated with report cardgrades. Similarly, Shavelson and Bolus (1982) found that gradesin English, mathematics, and science were more highly correlated51with matching areas of self-concept than with general self-concept. In his reanalysis of the Bachman and O'Malley (1986)findings, Marsh (1987a) provided further insight into thedistinction between academic ability and grade point average(GPA). As the schools in the study tended to grade on a curve,the distribution of grades was similar even when the students'actual abilities were not. Thus, equally able students tended tohave lower GPA's in high ability schools than in low abilityschools. Marsh found that ability test scores contributedstrongly to academic self-concept in addition to their indirecteffect through GPA.^Marsh also explained that this frame ofreference effect contributed to the BFLPE on academic self-conceptsuch that school-average ability negatively affected academicself-concept.The SDQ research has emphasized the distinctiveness of self-concepts in verbal and mathematical content areas, as well as thedistinction between academic and nonacademic facets of self-concept. Most of the studies have included both objective testsand teacher ratings of student ability or achievement (Marsh,Parker, & Smith, 1983; Marsh, Smith, Barnes, & Butler, 1983;Marsh, Relich, & Smith, 1983; Marsh Smith, & Barnes, 1984; Marsh &Richards, 1986; Marsh & Gouvernet, 1987). The present studyexamines the relationship between (1) the ESL student's self-concept and academic achievement and (2) teacher perceptions ofESL student's self-concept and academic achievement. In thepresent study, academic achievement is measured in terms of schoolperformance or grades in English and Math, based on the assumptionthat a student's marks or grades may best reflect a student's52academic achievement in school-related areas on a daily andcontinuing basis. Also, grades are the most frequent criteriaused by teachers to evaluate a student and to reflect dailyprogress and, therefore, may have the greatest impact in both thestudent's and teacher's formation of student's academic self-concept. Furthermore, teacher ratings of student achievement,although not formally assigned, most closely resemble grades. Inaddition, they have been the most frequently used achievementindicators in SDQ research.53F. ENGLISH AS A SECOND LANGUAGE STUDENTS  AND ASSESSMENTThis section reviews the literature as it pertains to secondlanguage learners in regard to theory and research in thefollowing areas: the development of language proficiency, secondlanguage teaching and academic achievement, psychoeducationalassessment, bilingualism, and cultural identity. In addition,these areas will be discussed as they relate to the controversysurrounding the testing and placement of ESL students. Althoughthis section is devoted to recent issues relating to ESL studentsand assessment, the degree to which these factors influence secondlanguage learners is beyond the scope of the present study untilthe primary questions regarding ESL student self-concept are firstexplored."Of the various procedures that school children undergo,perhaps the most critical in terms of their future areassessment and placement...placement decisions are made onthe basis of judgements about a student's level ofachievement and projections of future performance...Assessment and placement are themselves processes ofsocialization because they are based upon culturallydetermined perceptions, beliefs and values".The most important roles that a school undertakes - those ofdispenser of knowledge and skills, developer of humanresources and certifier of academic and technical competence- require it to make complex decisions about who gets whatand who goes where. These functions, in turn, serve tohighlight the initial importance of how a school that servesdifferent ethnocultural groups approaches assessment andplacement" (Samuda et al., 1989, p. 112-113).Furthermore, as Cummins (1984) suggests,"There has long been an unacknowledged relationship betweenbilingualism and special education as evidenced by thedisproportionate numbers of immigrant and minority languagechildren "deported" into special education classes andvocational streams in many countries" (p. 1).54The overrepresentation of ethnic minority students in specialeducation has led researchers to investigate the issues,assumptions, and policies underlying ESL assessment, placement,and counselling. In so doing, research findings over the last 20years have begun to shed light on the relationships between secondlanguage acquisition, teacher perceptions, and academicachievement.Unlike studies in the United States, little research in thepast has been conducted in Canada in regard to thedisproportionate numbers of minorities failing academically.Cummins (1984) indicates that surveys of the late 1960's and early1970's suggested that minority students, in Canada, out-performedtheir native English speaking counterparts. Recently, however,increased referrals for psychological and educational assessmentshave led to the reexamination of the theoretical assumptionsunderlying psychological assessment services for ESL students.For example, according to Cummins (1984), currently much ofassessment and pedagogy in bilingual special education "depends onimplicit theoretical assumptions that are logically inconsistentand contradicted by a considerable amount of research evidence"(p. 2). This has led to recent questions regarding thedevelopment of language proficiency, especially for secondlanguage learners.1.^Language ProficiencyAccording to Samuda and Crawford's (1970) survey of 34 schoolboards in Ontario, guidelines for the assessment of ESL studentsare often vague and perceived of as the most difficult part of the55school's identification and placement process. Cummins' (1980,1984) analyses of 428 teachers referral forms and psychologists'interpretations of assessment data in Western Canada reflect theimplicit assumptions about what constitutes language proficiencyand expose a variety of problems associated with the assessment ofESL students, particularly in reference to standardized tests.For example, educators have often focused on the acquisition ofEnglish as the primary goal of special programs for limitedEnglish proficient (LEP) students, assuming that the developmentof English proficiency would result in the student's eventualattainment of the academic skills needed to succeed in school in asecond language and at levels comparable to native Englishspeakers. Recently, research (e.g., Cummins 1979, 1981, 1984;Collier 1987, 1989) has begun to address the variables that mightinfluence this process, such as the length of time and the levelof second language (L2) proficiency required to achieve academicsuccess in a second language. Increasingly this research evidenceindicates that the optimal age question cannot be separated fromanother key variable in second language acquisition, cognitivedevelopment and proficiency in the first language (Cummins, 1981;Collier, 1989).According to McLaughlin (1984) and de Villiers and deVilliers (1978), first language (Li) acquisition may take aminimum of 12 years. From birth through age five, childrenacquire enormous amounts of Ll phonology, vocabulary, grammar,semantics, and pragmatics, but the process is not complete by thetime children reach school age. From ages six to 12 years,children still have to develop in the first language the complex56skills of reading and writing, in addition to the continuingacquisition of more complex rules of morphology and syntax,elaboration of speech acts, expansion of vocabulary, semanticdevelopment, and even some aspects of phonological development(McLaughlin, 1984; de Villiers & de Villiers, 1978). For schoolpurposes, language acquisition also must include the vocabularyand special uses of language for each subject area, such as themetalinguistic analysis of language in language arts classes andmany other learning strategies associated with the use of languagein each content area (Chamot & O'Malley, 1987; Heath, 1986).Second language acquisition research (Cummins, 1984; Collier,1989) has found that the process of Ll development has asignificant influence on the development of L2 proficiency. Oneimportant finding is that the lack of continuing Ll cognitivedevelopment during second language acquisition may lead to loweredproficiency levels in the second language and in cognitiveacademic growth. Lambert (1975, 1981, 1984) refers to this as'subtractive bilingualism', often developed by minority studentswho tend to experience academic difficulties. Cummins (1981b)describes this in terms of a lower threshold level in the firstlanguage, or 'limited bilingualism', with which negative cognitiveeffects may be associated. Furthermore, both Lambert and Cumminssuggest that special education programs may contribute tosubtractive or limited bilingualism and lower a student's academicself-concept by communicating that the minority student beeducated through his or her weaker L2.Several research reviews have identified groups of studentsexperiencing some negative cognitive effects of subtractive or57limited bilingualism (e.g., Cummins, 1981b, 1984; Dulay & Burt,1980; Duncan & De Avila, 1979; Skutnabb-Kangas, 1981). Beforepuberty, it appears that it does not matter when one beginsexposure to, or instruction in, a second language as long ascognitive development in the first language continues up throughage 12, the age by which first language acquisition is largelycompleted.Cummins (1980, 1981b) refers to a 'common underlyingproficiency', or interdependence, existing between a bilingualchild's two languages, even given widely varying surface features,with development of one language strongly aiding the developmentof the second one. That is to say, that although the surfacelanguage structures may be separate in either language, they seemto be interdependent and transferable from one language toanother. Therefore, experience in either language may promote thedevelopmental proficiency underlying both. The research findingsinvolving studies of bilingual education programs and FrenchImmersion, age on arrival studies, L1 literacy developmentprograms, and bilingual language use in the home appear to lendsupport for Cummin's theory of a common underlying proficiencybetween first language (L1) and second language (L2) skills(Collier, 1985, 1989; Cummins, 1980, 1981b; Cummins & Swain, 1986;Wells, 1981).Also, in terms of L2 acquisition, Cummins (1980, 1981a, 1984)distinguishes between the acquisition of basic interpersonalcommunicative skills or face-to-face language skills andcognitive/academic language proficiency. While basicinterpersonal communication skills reflects the surface structures58in English, is less cognitively demanding, and may be acquiredbetween two to three years of exposure to the new culture,cognitive/academic language proficiency is more linguistically andsemantically complex, reflecting the deeper structures of languageand usually taking five to seven years for language-minoritystudents to master in a second language.Language in school becomes increasingly complex and lessconnected to contextual cues as students move from one grade levelto the next. Cummins and Swain (1986) describe context-reduced,cognitively demanding school language as especially difficult tomaster. However, by the fourth grade, most school-relatedlanguage falls into this category. Nevertheless, failure byteachers and psychologists to take into account the developmentalrelationship between cognitive development, language proficiencyand academic achievement may result in incorrect interpretationsof ESL students academic difficulties and subsequent inappropriateacademic placements, especially when these students appear to havelittle difficulty understanding and communicating with teachersand peers in English.Collier (1989) cites a number of researchers who compared theperformance of subjects of different ages on language tasksassociated with school skills. When examining the optimum age forbeginning second language acquisition, most studies of both short-term and long-term gains found that students initiating secondlanguage acquisition between the ages of eight and twelve yearswere faster in the early acquisition of L2 skills. In addition,over several years' time, due to their cognitive maturity, theymaintained a greater cognitive advantage over younger children59initiating second language acquisition at four to seven years ofage.In order to address the optimal age question in regard tosecond language learners, Cummins (1981a) reexamined data from theRamsey and Wright (1974) study. Cummins (1981a) examined thelength of time needed for immigrants to acquire school languagewhen schooled exclusively in the second language after arrival.His study examined the achievement of 1,200 Canadian immigrants ingrades five, seven and nine. Cummins found that it took limitedEnglish proficient students five to seven years to reach native-speaker norms at the 50th percentile or 50th normal curveequivalent (NCE). Number of years of Ll schooling was notincluded as a variable. Cummins (1981a) found that length ofresidence was a significant variable on all tests of oral andwritten skills. Students being schooled only in the secondlanguage who were tested in the fifth, seventh and ninth gradeswere found to require a length of residence of five to seven yearsto reach the mean grade norms for native speakers in languageskills needed for school. Cummins found that older studentsperformed better than younger learners because they were morecognitively mature. This difference was lessened, however, whenyounger and older students were compared using norms appropriateto their age and grade. In addition, Cummins found that theeffect of length of residence and age on arrival variables seemedto diminish with time, especially after a length of residence offive years.In this same study, Cummins (1981a) found that immigrantstook approximately two to three years to reach proficiency in60basic communicative skills in English or the context-embedded,cognitively undemanding aspects of language. Yet, proficiency inbasic L2 skills did not correlate highly with the type of languageneeded for context-reduced, cognitively demanding language tasks,as measured on standardized tests. Nor did proficiency in basicskills correlate highly with informal measures designed to testthe thinking skills and more abstract thought required in theupper elementary grades and secondary school (Collier, 1987;Collier & Thomas, 1988; Gottlieb, 1985; Saville-Troike, 1984).Cummins (1981a) found that a period of five to seven years ofstudy in the second language is required to reach native speakerlevels in school-related language.In another study, Cummins, Swain, Nakajima, Handscombe,Green, and Tran (1984) examined age differences and the influenceof L1 development on L2 school language development. On measuresof L2 school skills, older students performed significantlybetter, whereas younger students outperformed older students incontext-embedded measures, or basic skills in English.Collier (1987; Collier & Thomas, 1988) conducted two studiesanalyzing the length of time required for 2,014 immigrants, whoseschooling was exclusively in English after arrival in the UnitedStates, to reach native-speaker norms on standardized achievementtest in reading, language arts, mathematics, science and socialstudies. Age of arrival ranged from four to 16 years and lengthof residence from two to six years. Over 75 different languageswere represented in the sample, which included 65% Asians and 20%Hispanics. The results of the studies showed that those studentsbelow age 12 years, who had had at least two years of formal61schooling in their first language before arriving in the UnitedStates, reached the 50th percentile or 50th normal curveequivalent on reading, language arts, science, and social studiestests in five to seven years. Evidence of transfer of contentknowledge in mathematics from first to second language wasdemonstrated by the students' high achievement on math scoresafter only two years of study in English. In contrast, youngstudents who had arrived between the ages of four and six yearsand who had received little or no schooling in their firstlanguage had not reached the 50th percentile or 50th normal curveequivalent within the first six years of length of residence andwere projected to reach it in seven to 10 years. Those studentswho arrived at ages 12 to 16 years also scored dramatically lowerthan students with an age of arrival (AOA) of eight to 11 years.After six years of schooling, all in the second language, they hadreached the 50th percentile or 50th normal curve equivalent onlyon standardized tests in mathematics.The studies appear to provide evidence that L2 proficiencyand academic achievement may not occur quickly and that theyinvolve a developmental process that takes a much longer time thanschool personnel have previously assumed. All three studies foundthat when schooled exclusively in the second language, studentsrequire a minimum of five years to reach the 50th normal curveequivalent on standardized tests. This appeared to be true evenfor the most advantaged students, that is, those who have a strongeducational background and who come from a middle or upper middle-class background. Adolescents with good cognitive development inthe first language, such as the adolescents in these studies,62reached high levels of proficiency in basic L2 skills in two tothree years with the possible exception of native-likepronunciation. As a result, their academic achievement laggedbehind that of native English-speaking peers. It appears that animportant key to successful second language acquisition andacademic achievement by adolescents may be uninterrupted academicinstruction during the acquisition of basic L2 skills.In summary, the research reviewed on language proficiency forsecond language learners highlights the critical interplay of ageon arrival, length of residence, and underlying developmentalproficiency in at least one language as key variables for thesuccessful acquisition of a second language. Overall, theresearch suggests that, depending on a child's age on arrival andlength of residence in the host country, along with previous firstlanguage development, it may take a minimum of two years toacquire basic communicative skills and up to seven or more yearsto reach the grade norms of native speakers in most subject areas,particularly if first language skills and concepts are not welldeveloped prior to introduction of the second language.2.^Second Language Teaching and Academic Achievement2.a.^Bilingual ProgramsIn her review of the research regarding language proficiency,second language programs, and academic achievement for language-minority students, Collier (1989) suggests that Li instructionthroughout the elementary school years, coupled with gradualintroduction of the second language, seems to produce a consistentpattern of greater academic achievement in the second language at63the end of four to seven years of schooling, even though the totalnumber of hours of instruction in the second language may bedramatically smaller when compared with schooling in the secondlanguage only. Also, the research findings support the view thatit takes four to seven years of dual language cognitive academicdevelopment for academic gains to be clearly seen, but that oncethose gains are achieved, students being schooled in bothlanguages are much more academically successful than their peersbeing schooled only in the second language.A number of cross-national and cross-cultural studies werecited by Collier (1989) in support of schooling in two languagesfor language minority students. A few examples will be brieflymentioned as they relate only broadly to the present discussion.Gale, McClay, Christie, and Harris (1981) reported that theresults of a longitudinal evaluation of an Australian bilingualprogram found that these students performed better on Englishlanguage measures after seven years of schooling than did theircounterparts schooled only in English . Malherbe (1978) foundthat, although students in South Africa experienced an initial lagwhen English was introduced, they reached grade-level norms on alltests of English by grade six. Skutnabb-Kangas (1979) found thatFinish nine to 11 year-old arrivals to Sweden, with several yearsof L1 schooling, achieved significantly higher than Finnishstudents who had little or no Ll schooling. In another studyreported by Skutnabb-Kangas, Finnish children who were schooled ina bilingual program in Sweden that permitted Ll cognitive academicdevelopment were also able to achieve at grade level by grade six.64In the United States, McConnell and Kendall (1987) found thatby grade five, bilingual-schooled immigrants to Washington statewere scoring at or above the 50th NCE in math, vocabulary, andEnglish reading. Similarly, Plante (1977) reported that Hispanicstudents from low-income families who participated in aConnecticut bilingual program were at or above national norms inEnglish and mathematics by the end of grade three, whereasstudents schooled only in English performed significantly lesswell. Tempes, Burnham, Pina, Campos, Matthews, Lear, and Herbert(1984) reported that Hispanic students in several bilingualprograms in California, where second language instruction wasintroduced in grade 3 while content area instruction continuedthroughout the elementary school years, reached national norms inEnglish reading by grades five or six, and in mathematics by gradethree or four. Krashen and Biber (1988) reported similar findingsfrom other California bilingual school studies in which Hispanicstudents from low-income families were able to reach the 50th NCEon English standardized tests of reading by grade six and inmathematics by grade three when Ll cognitive academic developmentmaterials were added to the curriculum. Troike (1978) reportedthat students in a French-English bilingual program in Minnesotawere at or above national norms in all content areas by the end offive years of schooling in both languages and that Hispanicstudents in a bilingual program in New Mexico reached nationalnorms in mathematics by the end of grade four and in Englishreading by the end of grade five. Medina and Valensuela de laGarza (1987) reported that Mexican American students in fourbilingual elementary schools in Arizona were scoring above65national norms on the California Achievement Test at the end ofthird grade on all subject area tests. Vorih and Rosier (1978),in an study of Navajo students, found that those in a bilingualprogram reached national norms by grade six, whereas studentsschooled only in English (L2) performed substantially lower. In alongitudinal study, Medina, Saldate, and Mishra (1985) reportedthat Hispanic students in Arizona who had attended a bilingualprogram at elementary school were still achieving at or abovenational norms in mathematics and in English reading by grade 12.In summary, according to Collier's (1989) review, studentswho are schooled in two languages, regardless of social class,generally appear to take from four to seven years to reachnational norms on standardized tests in reading, socials, andscience, whereas their performance may reach national norms in aslittle as two years in tests in mathematics and language arts.2.b.^ESL Programs EvaluationsVery few ESL program evaluations have been reported. A fewstudies of short-term gains have been summarized by Long (1983)who found that special L2 instruction does improve, to somedegree, students' performance on L2 tests, compared with that ofstudents who have 'natural' exposure to the language withoutspecial L2 instruction. However, there is a strong need forfurther research in this area.Collier (1987; Collier & Thomas, 1988) reported on the L2academic achievement of ESL students over a six-year period butthey did not evaluate the ESL program. There were no comparisongroups receiving only natural L2 exposure or bilingual instruction66and data was not available on the exact length of time that eachstudent remained in ESL classes. However, most students receivedone to two hours of ESL instruction daily for one to two years andwere immersed in the mainstream for the rest of their classes. Inthis situation, it took the eight to 11 year-old arrivals five toseven years to reach the 50th normal curve equivalent (NCE) on allthe standardized tests combined, and it was projected that thefour to seven year-old and adolescent arrivals would take seven to10 years to reach the 50th NCE.Saville-Troike (1984) followed 19 children, ages seven to 12years, through their first year of English (L2) acquisition,examining their achievement on many measures of oral and writtenEnglish and academic performance in all subject areas. She foundthat the two major factors that correlated significantly withtheir L2 academic achievement were development of Englishvocabulary and opportunity for continuing cognitive development inthe native language with peers and adults. The findings againappear to provide evidence for bilingual schooling.Gersten and Woodward (1985) summarized two evaluations ofstructured ESL immersion programs which provided instruction onlyin English through a sequential, step-by-step process. In oneprogram, Hispanic students scored slightly above national norms onlanguage arts, at the 47th NCE on mathematics, and at the 39th NCEon reading after four years of using highly structured DISTARmaterials. In the second structured immersion program, 16 Asianstudents placed in a special class reached the 58th NCE in readingand mathematics after five years.67In summary, there appears to be a great need for more studieson the long-term achievement of students being schooled entirelyin the second language. It may take as long as seven to 10 yearsfor nonnative speakers to reach the average level of performanceby native speakers on standardized tests, as found in the Collier(1987; Collier & Thomas, 1988) studies. In the bilingual programevaluations, comparison groups of students being schooledexclusively in the second language typically never reach the 50thNCE.In the lower mainland school districts of British Columbia,particularly where the ESL population is significantly large,either segregated and/or a combination of part-time pull-out ESLprograms have been instituted over the last decade. However,whether immersion or partial, all of these programs haveencouraged academic development in English only.2.c.^Schooling in Two Languages for Language MajorityStudentsOther studies that are thought to provide additional insightinto the question of how long it takes to master a second languageare evaluations of the Canadian immersion programs. Reviews ofthis research can be found in Cummins and Swain (1986), Genesee(1987), Larter and Cheng (1984), and Swain and Lapkin (1981).Early total immersion is the most widely implemented modelwhich typically provides all instruction in the second language(the minority language) for the first two years of schooling withgradual introduction of Ll by second or third grade. Typically,students reach national norms in tests in both languages and in68all subject areas by grade five and they continue to achieve abovenational norms throughout the rest of their schooling (Swain &Lapkin, 1981).Evaluations of early partial immersion programs, whichprovide balanced instruction in both languages from kindergartento Grade 12, have found that students' achievement sometimes lagsbehind that of the early total immersion students until grades sixor seven, perhaps because literacy training simultaneously in twolanguages causes confusion for students and it takes them longerto sort out the two language systems (Cummins & Swain, 1986).Late immersion students, who begin their immersion experiencein grades seven or eight and who have had sufficient L2preparation prior to the immersion experience, perform as well asearly total immersion students in measures of L2 proficiency, eventhough they have had approximately one fourth the number of hoursof L2 instruction. Overall, however, early total immersionstudents generally outperform students in all other types ofimmersion programs on attitudinal measures and on measures ofacademic achievement throughout their schooling (Genesee, 1987).2.d.^Generalizations on Academic Achievement in a SecondLanguageThe research reviewed on the length of time that it takes tobecome proficient in a second language for schooling purposes andto reach native-speaker norms in academic achievement has ledCollier (1989) to generalize about the relationships among thefollowing variables: first language acquisition, second languageacquisition, student age at the time of exposure to a second69language, academic achievement (as measured by standardized testsin all subject areas), membership in a language majority orlanguage minority community, and languages of instruction inschool. The pattern of generalizations may be summarized asfollows:1. Students schooled in two languages, regardless of socialclass background, generally take from four to sevenyears to reach national norms on standardized tests inreading, social studies, and science, and as little astwo years in mathematics and language arts.2. Immigrants arriving at ages eight to 12, with at leasttwo years of Ll schooling in their home country, takefive to seven years to reach the level of averageperformance by native speakers on L2 standardized testsin reading, social studies, and science and as little astwo years in mathematics and language arts when they areschooled in second language only programs after arrivalin the host country.3. Young arrivals, with no schooling in their firstlanguage, may take as long as seven to 10 years, orlonger, to reach the level of average performance bynative speakers on L2 standardized tests in reading,social studies, and science.4. Adolescent arrivals, regardless of academic background,who have not had L2 exposure and who unable to continueacademic work in their first language while acquiring asecond language may not have enough time left in high70school to make up the lost years of academic instructionand, without special assistance, these students maynever reach the 50th NCE or may drop out beforecompleting high school.5.^Consistent, uninterrupted cognitive academic developmentin all subjects throughout students' schooling appearsto be more important than the number of hours of L2instruction for successful academic achievement in asecond language.Furthermore, Collier (1989) indicates the need for furtherresearch with all ages of students acquiring a second language forschooling purposes and she suggests that, although mostcomparisons of student achievement in schools use the nationalnorms of standardized tests, these tests may not be the bestmeasures of second language proficiency for comparisons ofacademic achievement.3.^ESL Students and Psychoeducational AssessmentThe problems associated with the assessment of ESL studentshave stimulated a number of alternative evaluation procedures.One of these procedures is the use of translated versions of testswhich are normed, item- and factor-analyzed cross-culturally inorder to attempt to correct the difficulties inherent in directtranslations from one language to another. For example, theWechsler Intelligence Scale for Children (WISC-R) has been adaptedand standardized on various populations, including children fromHong Kong, Mexico, and Spain (Esquivel, 1985).71A number of studies have shown that Oriental populations havea different profile of performance on intelligence tests from thatof Caucasians in the United States and the United Kingdom (Lynn,1987). Lynn, Pagliari, and Chan (1988) suggest that Orientals,primarily Japanese and Hong Kong children, share a similar profileon tests of intelligence, in that they usually display highervisual-spatial scores, higher perceptual rates, and lower verbalscores when compared to Caucasians.Recently, Tam (1990) assessed the cognitive potential of 32Chinese immigrant students in both Cantonese (Li) and English (L2)and found that age on arrival and length of residence weresignificant predictive variables for the ESL student's verbalperformance. Other variables such as family SES, frequency ofspeaking Cantonese at home, gender, and previous English studywere also useful predictors of cognitive performance. Consistentwith the previous studies, the results found a similar profile ofhigh nonverbal and low verbal abilities on both the English andHong Kong cognitive measures. These findings help to furtherilluminate the relationships between some of the variablesassociated with language proficiency and cognitive ability which,in turn, continue to surround the controversy on the testing andplacement of language-minority students.4.^ESL Students and Academic AchievementThe relatively poor academic performance of many minoritylanguage students has been well documented (e.g., Coleman,Campbell, Hobson, McPartland, Mood, Weinfeld, & York,1966;Collier, 1989; Cummins, 1984; Shutnabb-Kangas & Toukomma, 1976;72Vernon,1982; Wong Fillmore, 1983). In addition, problems ofadjustment among minority language students have been observed byboth educators and social workers (Ashworth, 1988; Lambert, 1984).Researchers and theorists such as Ashworth (1988) and Cummins(1984), suggest that the difficulties these students face may notbe the bilingualism and psychoeducational 'deficits' per se, aspreviously assumed by some, but rather the 'cultural mismatch' orsocial and educational conditions in which minority studentsacquire their languages. Cummins (1984) strongly ascertains that,although psychoeducational influences may mediate or intervene inthe minority child's development, their effects can only beadequately understood within the context of more fundamentalsocial factors and their interaction with educational treatmentvariables. Furthermore, he suggests that,"these influences can, in principle, be changed with lessdifficulty than alleged cognitive deficits within thechild...and...The failure by educators and academics tocritically examine the implicit acceptance of middle-classdominant group values in the assessment and pedagogicalprocess has served to perpetuate the educational (andsocietal) status quo in which cultural and socioeconomicdifferences are frequently transformed into academicdeficits" (p.93).In the United States, studies and surveys on the academicachievement of students from various ethnic backgrounds tend toshow a different pattern of achievement for students of Asianorigin compared to other ethnic minority groups such as Mexican-Americans, Puerto Ricans, Portuguese, Blacks, and Native Americans(Coleman et al., 1966; Cummins, 1984; Vernon, 1982; Wong Fillmore,1983). To summarize, Hispanic, Native American, and Blackstudents have exhibited considerable educational disadvantages in73comparison to Anglo students, while Asian American students, onthe other hand, have tended to experience less disadvantagecompared to the other minority groups.In Canada, surveys conducted in Ontario, for example, havefound that minority students born in Canada tend to beoverrepresented in the high academic performance category, whileimmigrant minority students born outside of Canada were slightlyunderrepresented in comparison to their native-born English-speaking counterparts. Furthermore, although socioeconomicfactors appeared to significantly affect the English and Frenchfirst language groups, it seemed to have little effect for theCanadian-born minority students. However, both the Canadian-bornand the immigrant Chinese students demonstrated a high level ofacademic placement at all SES levels (Cummins, 1981a; 1984).Differences in achievement patterns between minority studentsled Wong Fillmore (1983) to conduct an ethnographic study, throughparental interviews and classroom observations, in order toinvestigate academic differences and their educationalimplications. The findings revealed that Mexican-American parentswanted their children to be happy and respected while Chineseparents wanted their children to be successful. Wong Fillmoreobserved that while the Mexican-American children tended to bevery socially mature and do better in less structured situationsand small groups, the Chinese children tended to orient themselvesto adults and to enjoy a highly structured classroom with clearlystated teacher expectations. According to Wong Fillmore, thedifferent peer orientation of the two groups was an importantfactor in accounting for the different achievement patterns. She74also relates these patterns of classroom behaviour to differencesin socialization between Chinese and Mexican-American homes.Specifically, she suggests that, for Chinese students, the use ofshame may be a powerful motivator in the home to promoteconformity to adult expectations and school success, whereas forMexican-American students, who may be given greater responsibilityand freedom in the home, there may be less tolerance for teacherdirected tasks and conformity to structure.More recently and locally in British Columbia, the findingsof a 1988 survey by the External Review Team on the VancouverSchool Boards' ESL Programs (Ashworth, Cummins, & Handscombe,1989) found that 46.9 percent of the total school population inVancouver schools were ESL and that that number was projected toincrease into the next decade. These students represented over 50countries, primarily Hong Kong, People's Republic of China,Vietnam, India, Figi, United States, and England. Also, aboutone-third of the ESL students were born outside Canada. In termsof ESL students' academic performance, the survey indicated thatapproximately one-half of the elementary ESL students werereported as being behind their age peers in English languagefacility and about one-third of the secondary ESL pupils werebehind their age peers in understanding and speaking English.Approximately two-fifths of the ESL students were behind inreading ability and about one-half were behind in writing ability.Furthermore, the committee suggested the need for changing servicedelivery and program models in order to meet the changing needswithin the schools. In summary, they proposed a more school-based, collaborative, and 'integrated setting' be adopted as soon75as possible in order to promote the development of languageproficiency, formal and informal interaction with peers, andcultural and group identity."The ideal is a program that supports ESL students' learningfor the entire day. Their instruction would take placemostly within an integrated setting and would segregate themfrom their English-speaking peers only for the time that isbilingual upgrading programs and programs providing groupidentity and support" (p.17).In addition, the committee outlined the importance of a teamapproach which promoted and included parents as participant co-educators along with the school, as well as the need for teacherin-service in order to better understand and individualizeinstruction to meet the growing population of ESL students andtheir needs.5.^Bilingualism and Cultural IdentityIn order to shed light on the changing cultural attitudes ofthe larger social environment as they relate to the development ofbilingualism and cultural identity for second language learners, abrief historical summary, adapted from Cummins (1984), outlinesthe changing socio-political attitudes directed towards ethnicminority groups.During the first half century, the prevailing attitude ofAnglo-conformity suggested that ethnic groups should give up theirown languages and cultures to assimilate into the dominant Britishculture. Bilingualism was regarded as a negative force in achild's development, causing confusion in a child's thinking.During this period of time, it was believed that bilingualchildren did poorly at school and that many experienced emotional76conflicts. Minority language children were made to feel that itwas necessary to reject their home culture in order to belong tothe majority culture and often these children ended up unable toidentify with either cultural group (Lambert, 1975). In 1971,after the Report of the Royal Commission on Bilingualism andBiculturalism, a shift in Canadian policy occurred, whereby thethe Canadian government adopted a policy of 'multiculturalismwithin a bilingual framework' which encouraged all ethnic groupsto enrich society by developing their cultures. Recognition wasgiven to all groups contributing to Canadian identity and thebenefits of linguistic diversity were encouraged. The RoyalCommission of 1971 recommended that the teaching of languagesother than English and French, and cultural subjects related tothem, be incorporated as options in public elementary schoolprograms wherever there was sufficient demand. However, thenumbers of students receiving heritage language instruction inpublic elementary schools and the types of programs continue tovary widely across provinces.Cummins (1981) and Lambert (1975, 1981, 1984) suggest thatthe patterns of bilingualism and cultural identity among minoritychildren are important to the understanding of the researchfindings on academic achievement. Furthermore, they suggest that"the past insensitivity of educators to these identity conflictshas contributed substantially to minority students' adjustmentproblems and academic difficulties" (Cummins, 1981, p. 16). Inthe past, in Canada, there has been a strong tendency to replaceLl with L2 (English) so that literacy skills in the first language(L1) were not developed. However, cultural identity and the77patterns of bilingualism children develop may be closely tied totheir attitudes towards the two languages and the two culturalgroups that speak these languages. With a strong need to belong,the process of choosing identity for minority language students,may be more complicated because the cultural milieux between homeand wider society may often display different values.Lambert (1975, 1981) presents four ways children may work outpossible conflicts between the language and culture of home andschool. These may be summarized as follows:1. Most commonly, minority children may reject homelanguage and culture and identify with Canadian home andculture. While this may lead to more rapid assimilationof English skills, it may also lead to familial discordand personal development problems.2. Minority children may reject the Canadian language andculture and identify with home language and culture. Byresisting assimilation, these children may associatemainly with their own ethnic group. They may experiencedifficulty learning a second language and exhibit poorschool performance.3. Minority children may be unable to identify comfortablywith either group. Home culture may often bediscredited. They may be unable or not allowed tointegrate.4. Minority children may identify with both languages andcultures such that they retain pride in both, evaluatethe strengths and weaknesses of both and choose their78own values and identity. These children may be morelikely to be motivated to develop proficiency in bothlanguages, have more potential for personal development,and contribute to Canadian society.As mentioned in the review of the literature on thedevelopment of language proficiency, Lambert (1975, 1981, 1984)suggests that 'subtractive bilingualism' may occur when L2replaces Ll and that it is often developed by minority studentswho tend to experience academic difficulties. For these students,proficiency in both languages may be less well developed thanamong natives in each. 'Additive bilingualism', on the otherhand, may occur when L1 remains dominant and L2 is added at nocost to L1. Similarly, Cummins (1976) 'threshold hypothesis'suggests that the levels of proficiency in two languages may be animportant intervening variables mediating the effects ofbilingualism on cognitive and academic development. Furthermore,Cummins suggests that for the ESL student, academic self-conceptmay often be low and that special education may often contributeto 'subtractive bilingualism' by communicating that the minoritylanguage student be educated through his or her weaker L2.In summary, the literature on second language learners tendsto suggest that positive academic achievement and positivecultural identity may be enhanced when the patterns ofbilingualism and cultural diversity are understood and when thedevelopment of language proficiency in both L1 and L2 arecelebrated and encouraged by the larger social, cultural andacademic environment surrounding the ESL student.79806.^Testing and Placement ControversyThe effect of inappropriate labelling of assessment onteacher expectations and children's self-images for minorities,and the overinclusion of minority children in special educationdue to incorrect identification has been well documented (e.g.,Coleman, et al., 1966; Collier, 1987, 1989; Cummins, 1984). Inthe past, studies carried out in the United States, Canada, andBritain have shown that teachers tend to have more negativeexpectations for minority students than for other groups. Forexample, Fram and Crawford (1972) found that kindergarten teachersviewed ESL students as less likely to demonstrate high levels ofacademic success and more likely to fail. In view of the factthat minority students frequently may be perceived of as 'lowachievers', Fram and Crawford suggest that they may alsoexperience less positive interactions with teachers than majoritygroup students. Other studies have also shown that teachers tendto use positive interactions more frequently with perceived highachievers than perceived low achievers (Good & Brophy, 1971;Kerman, Kimball and Martin, 1980; McDermott, 1978; Rist, 1970).Possible negative expectations for minority groups may bereinforced by the results of early identification measures whichmay not be appropriate for minority students and which may lead toteacher assumptions about students' proficiency in English and,inadvertently, affect the quality of instruction given to thesestudents (Cummins, 1984; Keogh & Daley, 1983). In discussingsimilar findings from the U.S. Commission on Civil Rights 1973report, the California State Department of Education (1982) statedthat, "The perceived status of students affects the interactionsbetween teachers and students and among the students themselves.In turn, student outcomes are affected" (as cited in Cummins,1984, p. 113). Furthermore, and consistent with the relatedliterature on the different pattern of academic achievementevidenced between minority groups, Cummins (1984) notes that,"...until recently, Asian-American students have had as lowstatus as any other minority group but have achieved wellacademically. During the past decade, their status appearsto have been elevated as a result of greater awareness on thepart of teachers and the general public of their relativelyhigh achievement" (p. 113).However, despite the controversial issues surrounding theidentification, testing, and placement of ESL students, anddespite a growing body of research on the interplay ofdifferential patterns of bilingualism, cognitive, and languagedevelopment and their relationship to academic achievement andteacher assumptions regarding second language learners, little hasbeen done to identify the self-concept of Canadian ethnicminorities and ESL students in particular.G.^SELF -CONCEPT AND MINORITY STUDIESMuch has been theorized about self-concept, and much of theresearch is conflicting, particularly when it applies to minoritygroups. Several early investigations have found that minoritygroups possess a poor self-concept on the basis of ethnic identityand socioeconomic status. Much of the research comes from theUnited States, where differences in Black and White subjects were,and still are, of paramount interest. Some have demonstrated a81significantly better self-concept for Blacks and other minoritygroups than for Whites (Goldman & Mercer, 1976; Hunt & Hardt,1969; Phillips, 1973). Studies by Hodgkins and Stankinas (1969),Williams and Byars (1968), and Zirkel (1971) found minority andBlack students to have a poor self-concept.Wylie (1974) criticized many of these early conflictingstudies because there was very little agreement as to theoperational definitions of self-concept. Furthermore, moststudies were carried out with instruments that had been used onlyonce or a few times, the result being that there were as manyinstruments as there were studies, making generalizations acrossthese studies virtually impossible.Iheanacho (1988) reviewed the 'minority self-conceptliterature' to analyze more recent findings which suggested thatminority groups, with similar background and ability levels, dohave higher self-concepts that Whites. The review found that themain obstacles minority adolescents face in their development ofself-concept are environmental factors, such as socioeconomic,educational, and political barriers, as well as their limitedaccess to mainstream society caused by a lack of opportunity topenetrate the barriers of discrimination.Wilkinson and Burke (1985) conducted a study to examine theimpact of the interaction of ethnicity, socioeconomic status, andself-concept on Mexican-American children. They found that self-concept of ability had a stronger effect than ethnic identity onchildren's academic performance. This finding was particularlyinteresting given the fact that self-concept of ability appearedto be influential regardless of ethnic identity or SES. Iheanacho82(1988) notes an early study by Bayton, McAlister, and Hamer (1956)which asked subjects to select a list of 85 adjectives thesubjects considered the most typical of upper class whiteAmericans, upper class black Americans, lower class whiteAmericans, and lower class white Americans. Certaincharacteristics were attributed to each class regardless of race,confirming that the low self-concept of minorities appears to bedue to family values, social and economic position, and not solelydue to their minority group affiliation. Iheanacho concludes thatfuture researchers should concentrate on finding the environmentalfactors that impact on self-concept as opposed to the culturalfactors. In terms of the interaction between race and SES, Wylie(1979) found that, "...some of the alleged influences ofracial/ethnic status are parallel precisely because most minorityracial/ethnic groups tend to occupy lower SES levels" (p.119).Research regarding Asian minorities, the topic of this study,although scarce and predominantly conducted in the United States,report that the SES disparities noted with Blacks, Hispanics, andNative Americans were not as great. These adolescents, many ofwhom included native-born or immigrant families from China, Japan,Korea, Cambodia, Hong Kong, Vietnam, Burma, Laos, Thailand,Malaysia, and the Philippines, are seen by some (e.g., Liu & Yu,1975) as having received better treatment and, consequently,higher income and educational levels. Liu and Yu postulate thatit is unlikely that the self-concept of the 'Asian American' willbe lower since environmental factors have remained in theirfavour.83In the Canadian context, research which specifically dealswith minorities and self-concept is almost nonexistent. Akoodie(1984) suggests that research efforts with minorities have beendirected to assess minority adjustment rather than self-conceptbecause many groups are in the midst of social transition and thephenomena of acculturation/assimilation has interested researchersmore.H. SUMMARY OF CHAP TER IIChapter II reviewed the literature relating to three bodiesof knowledge: self-esteem or self-concept, ESL student assessment,and minority studies. According to the research reviewed on theShavelson model of self-concept, as adopted in the present study,self-concept may be conceived of as multidimensional andhierarchical. Therefore, a student may have multiple self-perceptions and distinct self-concepts which are measurable andsituation-specific, as well as age and sex related. The researchindicates that teachers, as significant others, may often be ableto adequately infer the student's self-perceptions, particularlyin academic areas. However, the processes that students andteachers use to arrive at their perceptions are thought to bequite different. In addition, studies have shown that a student'sself-concept in specific areas appears to correlate with his orher academic achievements in related areas, although not asstrongly as external criteria might suggest.The ESL student assessment literature suggests that, forsecond language learners, there appears to be a relationship84between language proficiency, second language acquisition,bilingualism, cultural identity, cognitive profile, and academicsuccess. However, these issues, along with the controversiessurrounding the testing and placement of ESL students, continue tobe under debate. In addition, the literature relating to minorityself-concept continues to be scarce. The studies, which wereconducted primarily in the United States, tend to suggest thatenvironmental factors may influence self-concept more thancultural differences. In contrast, other research suggests thatcultural differences may greatly impact on the academicachievement of students for whom English is a second language.The interplay of the ESL student's self-concept with teacherperception and with academic achievement is the topic of thispresent exploratory study.85III:  METHODOLOGYThe purpose of this investigation as it applies to English asa second language (ESL) students is two-fold. First, it exploresthe relationship between an ESL student's self-concept and teacherperception of that student's self-concept. The study examinesboth student and teacher responses to the same self-conceptinstrument, the Self-Description Questionnaire-1. Second, itexamines the extent to which student self-concept and teacherperception of student self-concept relate to academic achievementin mathematics and in reading, as defined by the student's gradesand/or teacher ratings of student performance.A.^RESEARCH QUESTIONSThe current study is exploratory in nature and utilizes acorrelational design in order to investigate, more specifically,the following research questions:1. What is the self-concept of the ESL student?2. What is the teacher perception of that ESL student'sself-concept?3. What is the relationship between the ESL student's self-concept and teacher perceptions of that student's self-concept and to what extent do they agree?4. What is the relationship between the ESL student's self-concept and the student's academic achievement in86reading and in mathematics and to what extent do theyagree?5.^What is the relationship between teacher perception ofthe ESL student's self-concept and the student'sacademic achievement in reading and in mathematics andto what extent do they agree?B.^SELECTION AND DESCRIPTION OF SUBJECTSThe current study consisted of 57 fifth and sixth-gradestudents (28 male and 29 female) ranging in ages from 10 years 4months to 12 years 7 months (mean age = 10.86) and 21 teachers (4male and 17 female). The subjects were selected from two schooldistricts both situated in the lower mainland of British Columbia.A total of 5 schools participated in the study, 2 schools from theurban school district (34 students and 5 teachers) and 3 schoolsfrom the more suburban district (32 students and 16 teachers). Inthe suburban district, 3 of the 16 teachers were primaryrespondents who collaborated with the 13 other teachers on someitems; 2 of the 13 collaborating teachers taught as a team.1.^Student SelectionOriginally, the request for subject participation was sent toschools within the urban area only. Criteria for studentselection, at that time, was established as follows, basedprimarily on a review of the ESL assessment literature and theavailability of grade/age norms for the Self-DescriptionQuestionnaire-1:87881. Chinese immigrants, preferably of Hong Kong origin.2. Resident in Canada for approximately 5 years.3. Presently in grades 5 or 6.4. Aged 10 - 12 years; approximately 11 years old.5.^Presently mainstreamed into a regular classroom.However, due to a general lack of volunteer participation andextensive research previously conducted within this urban schooldistrict, the criteria for student selection was modified andextended to also include voluntary participation from a nearbysuburban school district. The modified criteria for studentselection were established as follows:1. Immigrants or Canadian-born ethnic minorities whosefirst language is other than English.2. Resident in Canada for any number of years.3. Presently in grades 5 or 6.4. Aged 10 - 12 years.5.^Presently mainstreamed into a regular classroom, in anESL setting, or receiving assistance on an ESL/ELC(English Language Centre) pull-out basis.After receiving school district permission to conduct thepresent study within the selected schools, initial telephonecontact was made with all school principals in order to elicitvolunteer cooperation from both principals as well as teachers.Letters describing the general purpose of the study and requestingparticipation and administration time were sent to the schoolprincipals and the respective teachers who agreed to participatein the study (see Appendix A). Letters to the students and theirparents as well as parental/guardian consent forms, in bothEnglish and Cantonese, were also sent to the schools (see AppendixB). The consent forms were then distributed by the schools. Onlystudents whose parents or guardians issued informed consent fortheir children to participate in this study were selected.2.^Student DescriptionOf the 57 students involved in the present study, 28 (49.1%)were male and 29 (50.9%) were female. Similarly, 29 (50.9%)students were enrolled in grade 5 and 28 (49.1%) in grade 6.Students ranged in ages from 10 years 4 months to 12 years 7months (mean age = 10.86). Thirty-four (59.6%) of the studentswere recruited from the two urban schools and 23 students (40.4%)were recruited from the three suburban schools. The majority ofstudents or 25 (43.9%) were recruited from one of the urbanschools; 15 of these students were enrolled in the same full-timeESL classroom. Twelve students (21.1%) were enrolled in regularclasses, 18 students (31.6%) were enrolled in full-time ESLclasses and 27 students (47.4%) attended either ESL or ELC pull-out programs. Two of the 18 full-time ESL students wereintegrated into regular math programs within the school. Whereasthe urban ESL students were enrolled in either regular, full-time,or pull-out programs, suburban ESL students recruited for thepresent study attended only ESL pull-out programs.Forty students (70.2%) were of Hong Kong origin (includingthree Canadian-born who spent less than six months in Canada89before returning to Hong Kong), six (10.5%) Canadian-born Chinese,three (5.3%) from Taiwan, three (5.3%) from Vietnam, two (3.5%)from Japan, two (3.5%) from Korea, and one (1.8%) from thePhilippines. Length of residence in Canada ranged from less than6 months to 12 years with the majority of students or 33 (57.8%)having been in Canada between 7 months and 2 years. Student's ageon arrival in Canada ranged from birth to 11 years with themajority of students or 30 (52.6%) having arrived in Canadabetween 9 years and 10 years of age.3.^Student School BackgroundAdditional school demographic information was collected inverbal communication with each school principal during a short, 10minute interview (see Appendix C). Background informationincluded data on the percentage of ESL students in the school,minority group representation, the number of actual ESL studentsreceiving service, ESL programs available in the school, school-average socioeconomic status, and school-average achievement(based on provincial government test scores). According to theprincipals' reports, 3 of the 5 schools consisted of over 65percent visual minority or ethnic group representation. All ofthe schools, except one suburban school, reported that ethnicgroup representation was predominantly of Chinese origin. The twourban school principals reported that the many of the ESL studentsappeared to be fluent in English. One of the urban schoolsconsisted of a large proportion of Canadian-born ESL students.Overall, a total of 10 to 25 percent of all students in the 5schools received some sort of ESL instruction. Based on real90estate housing values and principal's report, the socioeconomicstatus of the 5 schools was rated as follows: Two of the suburbanschools were rated as middle class and one as lower-middle toupper-middle class. One of the urban schools was rated as upper-middle class and one as a split between middle and lower class.School-average achievement was rated as average in two of theschools, as above average in two others, and as high in oneschool.In general, the principals indicated a changing ethnic mixeven within their own schools. Compared to the urban district,the suburban principals also reported an influx of more recent orsecond wave Chinese immigrants.4.^The ESL TeachersA total of twenty-one teachers participated in this study.Five teachers (all female) were recruited from the two urbanschools and 16 teachers (4 male and 12 female) were recruited fromthe three suburban schools. However, only three (1 male and 2female) of the 16 suburban teachers were primary respondents asthey were the teachers in charge of the ESL pull-out program andalso the teachers who subsequently answered the majority of theSDQ-1 items. The remaining 13 suburban teachers, two of whomshared a class, collaborated with the three primary ESL teacherson some of the SDQ-1 items. All of the teachers who volunteeredto participate in the present study were currently instructing theESL students also recruited at this time (N=57). Overall, theeight primary teachers in the five schools were responsible forresponding to an average of six (14%) SDQ-1 questionnaires each91with a minimum number of two (3.5%) and a maximum of fifteen(26.3%). A letter explaining the purpose of the investigation andrequesting participation was given to each teacher, along with theaforementioned package of information sent to the principals (seeAppendix A). The teachers in both school districts appearedwilling and eager to respond to questions concerning their ESLstudents. In general, all of the teachers appeared young orapproximately 25 to 35 years of age.C.^INSTRUMENTATION1.^Self-Description Questionnaire-1 (SDQ-1)The Self-Description Questionnaire-1 (SDQ-1) is an eightscale instrument designed to measure seven aspects of the self-concepts of preadolescent children, ages 7 to 13 years, as well astheir general sense of self-worth. This questionnaire wasadministered to all subjects, including both the ESL students andtheir teachers. Examples of both student and teacherquestionnaires are included in Appendix D.The SDQ was originally developed to measure self-concept infour nonacademic areas (Physical Ability, Physical Appearance,Peer Relations and Parent Relations) and three academic areas(Reading, Mathematics, and General-School which refers to allschool subjects). Recently, the scale was revised to include aGeneral-Self scale, a modification of the Rosenberg (1956, 1979)Self-Esteem Scale. The SDQ was also used to test specifichypotheses derived from Shavelson's hierarchical andmultidimensional model (Shavelson & Bolus, 1982; Shavelson,92Hubner, & Stanton, 1976) and subsequently led to a revision of themodel (Marsh, Byrne & Shavelson, 1988; Marsh & Shavelson, 1985;Shavelson & Marsh, 1986).The scale consists of a total of 76 items. Each of the 8individual SDQ-1 scales contains 8 positively-worded items. Adefinition of each individual scale and a list of the itemscomprising it, as stated in the SDQ-1 manual (Marsh, 1988), may befound in Appendix F. In completing the SDQ-1, respondents areasked to answer simple declarative statements such as, "I'm goodat mathematics" or "I make friends easily". Respondents maychoose from a 5-point Likert-type scale. Response options varyfrom Mostly False, False, Sometimes False/Sometimes True, MostlyTrue, True, or from 1 to 5, respectively. An additional 12negatively-worded items, intended to disrupt positive responsebiases, are not included in the calculation of self-conceptscores. Norms are based on the responses of 3,562 students fromNew South Wales, Australia in grades 2 to 6. The manual (Marsh,1988) also provides separate norms by sex and by grade level forgrades 2 to 4 and 5 to 6, respectively.Marsh (1988) reports relatively high internal consistencyreliability coefficients for the 8 individual SDQ-1 scales rangingfrom .80 to .92 (median =.86). Coefficient alphas, based on thetotal normative sample (N=3,562), are reported as follows:Physical Abilities/Sports, .83; Physical Appearance, .90; PeerRelations, .85; Parent Relations, .80; Reading, .89; Mathematics,.89; General School (All School Subjects), .86; General Self, .81(N=729). Coefficient alphas are also reported for the 3 total93scores as follows: Total Nonacademic, .91; Total Academic, .92;Total Self, .94.In order to measure the stability of self-concept responsesover time, Marsh, Smith, Barnes, and Butler (1983) examined SDQ-1responses based on two studies with students in grades 5 and 6(N=528) and one study with students in grade 4 (N=148) tested 6months apart within the same school year. Internal consistencyestimates for both the individual scales (M=.87) and the totalscales (M=.92) appeared high. Test-retest coefficients obtainedalso appeared high for both the individual scales (M=.61) and thetotal scales (M=.65) with the exception of the Parent Relationsscale at grade 4 (r=.27). The reliabilities of difference scores,reported for both individual scales (mean coefficient alpha =.74)and the total scores (mean coefficient alpha =.87), appeared tosupport the conclusion that the changes in self-concept aresystematic and not due to random fluctuations. Furthermore,correlations among the difference scores representing the SDQ-1scales were smaller than their reliabilities (mean=.24) and factoranalysis of the difference scores indicate that changes in self-concept appear to be multidimensional and specific to particulardimensions.Numerous exploratory and confirmatory analyses of the SDQdata have been reported lending strong support for the constructvalidity based on interpretations consistent with the Shavelson,Hubner, and Stanton (1976) model of self-concept and subsequentrevisions by Marsh and his colleagues (Marsh, Byrne, & Shavelson,1988; Marsh & Shavelson, 1985; Shavelson & Marsh, 1986). In herreview of analyses of data from a variety of studies using the94SDQ, Wylie (1989) reports that all items or item pairs for each ofthe 7 or 8 SDQ scales loaded relatively highly on a target factorand showed negligible loadings on the other 6 or 7 factors. Thisseries of findings appears to support the intendedmultidimensionality of the SDQ. In addition, Wylie indicates thatmore recent factor analyses defined an eighth factor correspondingto the General-Self scale and that subsequent factor correlationstend to correlate with each of the 7 original factors as well assupport the hierarchical model of self-concept proposed by theShavelson model and its revisions. Further research investigatingthe generalizability of the factor structures across age, sex, andnationality have also been explored.Responses to the SDQ-1, particularly in terms of the 3academic self-concept scales (Reading, Mathematics and General-School or All School Subjects), were found to be related to thefollowing variables: sex, age, socioeconomic status, academicachievement, teacher ratings of achievement and inferred self-concept, peer ratings of inferred self-concept, studentattributions for perceived causes of academic success and failure,responses to other self-concept instruments, and experimentalinterventions designed to enhance self-concept. For more details,refer to the manual (Marsh, 1988a) and to a review of the SDQinstrument (Wylie, 1989). As responses to the SDQ-1 appear to besystematically related to these external criteria in waysconsistent with theory, Marsh (1988a) and his colleagues arguethat strong support for the construct validity of the SDQ-1 may beindicated. However, in her most recent review, Wylie (1989)cautions "that virtually all the studies relating SDQ scores to95other variables in order to test theory-based predictions involveonly the three academic self-concept scales, Reading, Mathematics,and All School Subjects. Almost everything remains to be done toexplore the construct validity of the other four SDQ variables byusing them to test theory-based predictions." (p.73)The Self-Description Questionnaire-1 was selected for use inthe present investigation for several reasons. It is devised foruse with a preadolescent age group and most of the research behindthe instrument has been targeted in the 10 to 12 year age group ingrades 5 and 6 (Marsh, 1988a; Wylie, 1989). The instrument isrelatively new and measures the multiple dimensions of self-concept allowing for comparisons that might otherwise be obscuredby a more unidimensional test (Marsh, 1988a; Wylie, 1979, 1989).It appears to have a well-developed factor structure which hasbeen well researched. Furthermore, the SDQ-1 appears to measuredimensions, particularly academic-related ones, that are reliable,valid and based upon a strong theoretical model (Wylie, 1989).Although previous reviewers have criticized the poor quality ofinstruments used to measure self-concept, the development of theSDQ-1 appears as an attempt to remedy this situation (Wylie,1989). At present, little research seems to exist with respect toCanadian ethnic minorities and self-concept. At the same time, asthe mosaic of our Canadian culture continues to change withincreasing immigration from other countries, information aboutstudents for whom English is a second language becomes morecritical in order to understand their needs and how to programappropriately to meet them. This instrument provides informationon several aspects of self-concept which might help educators96begin to check possible assumptions they may have in regard totheir ESL students. Also, it is frequently assumed that teachersunderstand and are able to adequately assess their students self-concepts. The SDQ-1 may help to provide more objectiveinformation in terms of particular areas of ESL student self-concept and its relation to teacher perception and academicachievement.2.^ScoringResponses to the SDQ-1 may be scored using the SDQ-1 Scoringand Profile Booklet (see Appendix G). The booklet provides forthe calculation of individual scale raw scores, total scale rawscores, and optional control scores. A computer scoring programis also available from the author which calculates the raw scoredata, produces factor scores and z-scores, and relates othervariables to these scores. Parts of the computer program,relating to the raw score data for both students and teachers, toeach other, and to other variables, were utilized in the presentstudy. Refer to the manual (Marsh, 1988a) for a detaileddescription of scoring procedures.Responses to the 64 positively-worded items are scored asfollows: 1=False, 2=Mostly False, 3=Sometimes False/SometimesTrue, 4=Mostly True, and 5=True. There are 8 positively-wordeditems comprising each of the 8 individual scale scores. Responsesto the 12 negatively-worded items are reversed so that 1=True and5=False. The negatively-worded items are not calculated in theindividual scale or total scale raw scores. For each of the 897individual scales the lowest possible total raw score is 8 and thehighest possible total raw score is 40.The individual scale raw scores are used to calculate theTotal Academic, Total Nonacademic, and Total Self raw scores. Theindividual raw scale totals for Physical Abilities, PhysicalAppearance, Peer Relations, and Parent Relations are summed andthen divided by 4 to obtain the Total Nonacademic raw score.Similarly, the raw scale totals for Reading, Mathematics, andGeneral-School are summed and divided by 3 to obtain the TotalAcademic raw score. The Total Nonacademic and Total Academic rawscores are then summed and divided by 2 to obtain the Total-Selfraw score. When calculating total scale scores, results arerounded to the nearest whole number. For each total score, thelowest possible raw score is 8 and the highest is 40, with higherscores generally indicative of a more positive self-concept.Raw scores may then be converted to mid-interval percentilesand standard scores or non-normalized T-scores which have a meanof 50 and a standard deviation of 10. Normative comparisons arereported separately in the manual (Marsh, 1988a) for males andfemales in grades 2 to 4 and 5 to 6, respectively. Combined normtables are also available.D.^DATA COLLECTION PROCEDURESThe data was collected over a three-month period from April1990 to June 1990.98991. StudentsIn each of the 5 schools, all of the ESL students recruitedwere removed from their classes and given the Self-DescriptionQuestionnaire-1 at the same time in group session during a morningperiod. Administration of the SDQ-1 to the ESL students tookapproximately 25 minutes per school. To ensure standardizedpresentation and to circumvent potential reading problems, thestudent consent form or Request for Student Participation, thedirections for the SDQ-1 on the front cover of the questionnaire,and each item on the SDQ-1 were read aloud to the ESL students bythe present researcher (see Appendices B and D).During the initial instructions and prior to administrationof the questionnaire, the students were encouraged to ask thepresent researcher for clarification if they did not understandthe wording or intent of a particular statement. In addition,several more bilingual students acted as translators. Throughoutthe sessions the subjects were observed and all appeared tocomprehend. Students appeared to make responses at theappropriate times and they did not seem to make perseverativeresponses. On several occasions, students did ask for furtherexplanation. For example, one student asked for clarification asto the type of "Reading" referred to in the questionnaire andanother asked about the meaning of the word "athlete".2. TeachersAfter the ESL students in a particular school wereadministered the SDQ-1, the researcher personally handed the SDQ-1questionnaires to the teachers, one per each ESL student recruited100in their class. The teachers were given written directions tofollow and during this time they were encouraged to ask forclarification (see Appendix E).In the two urban schools, the five ESL students' homeroomteachers completed the SDQ-1 scales, one per ESL student sorecruited from their respective classes (i.e., 5 teachers and 34students). One of the urban teachers responded to 15questionnaires whereas the other three teachers responded to two,three, and five questionnaires respectively. The remaining urbanteacher responded to nine SDQ-1 questionnaires. In the threesuburban schools, however, due to an ESL Language Arts pull-outpolicy, the three ESL teachers responsible for the pull-outprogram decided, of their own accord, to collaborate with the 13regular or homeroom teachers to coordinate responses to some ofthe SDQ-1 items for each student recruited (i.e., 16 teachers and23 students). In two of the three suburban schools, the two ESLteachers each consulted with four classroom teachers on eight andsix students, respectively. The third ESL teacher, in theremaining suburban school, consulted with five other teachers onnine students; two of the classroom teachers shared a homeroomclass. For example, the suburban ESL teachers expressed concernsregarding their limited knowledge and experience in terms of anESL student's ability or performance in math-related areas as wellas in other subject areas, such as socials and science, which maynot involve ESL instruction directly. However, for the suburbanschools, only the three primary respondents, the ESL teachers,were included in the correlational comparisons.1 01In addition, many teachers from both school districtsexpressed concerns in terms of responding to some of thenonacademic SDQ-1 items, particularly those involving thestudent's parental relationships, a topic about which the teachersfelt less familiar. Nevertheless, all of the teachers reported togive their best effort on each item presented and no itemsappeared to be deleted.In addition, the teachers were asked to fill out a form foreach of their ESL students in regard to the student's backgroundhistory, if known (see Appendix E). Demographic information wasthus obtained in terms of the student's age, grade, type of ESLprogram, country of origin, ethnicity, home language, number ofyears in Canada, and age on arrival in Canada. The informationobtained may be used in subsequent or future studies.3.^Academic AchievementThe teachers were also requested to report the student's mostrecent grades or marks for both Mathematics and Reading orLanguage Arts as stated on the last report card (Spring session).If the ESL student did not receive report card marks, as in thesuburban district, the teachers were asked to rate the student'sperformance on a scale from A to D or 1 to 4. This 4-point scalewas later converted to an 8-point scale to include subtledifferences in the values, where A=1, B+=2, B=3, B-=4, C+5, C=6,C-=7, and D=8, as indicated by the teachers' ratings (see AppendixE) •Teacher responses to the SDQ-1 and gathering of thebackground information was estimated to take approximately 151 0 2minutes to complete per student. The teachers' data was picked upby the researcher at the school approximately the following week.E. DATA ANALYSISData was collected on eleven facets of self-concept (8individual scales and 3 total scales) in order to determine theself-concept of the ESL student, teacher perception of thatstudent's self-concept and whether, and to what degree, arelationship exists between the two. In addition, data wascollected on the student's academic achievement in both readingand mathematics in order to determine whether, and to what degree,a relationship exists between these various facets of self-conceptand academic achievement.Means, standard deviations, and ranges were calculated forall the variables. Pearson correlations were computed to describethe strength, including the magnitude and direction, of anyrelationships found among these variables. Normally, arelationship is considered statistically significant when thePearson Product-Moment Correlation Coefficient is at the alphalevel of .05 or lower. Alpha levels of .05, .01 and .001 wereused to determine the degree of statistical significance amongrelationships. However, given the relatively large number ofcorrelations computed for this study and the relatively smallsample size (N=57), a conservative approach was adopted in whichalpha levels of .05 were treated as 'trends'. T-tests were usedto determine if differences in means existed between variables.Analysis of the data was conducted using the computer program,SPSS-X (SPSS. Inc., 1983) in conjunction with the Scoring Programfor the SDQ (April 18, 1988).103IV: RESULTSThis chapter summarizes the results of the data analysesconducted and it describes these findings in terms of the researchquestions posed in Chapter Three. First, descriptive statisticsare presented which relate to self-concept ratings for bothstudent and teacher respondents. Academic achievement ratings andscore distributions for Reading and Mathematics are then reported.Second, Pearson correlation coefficients used to examine therelationship among these variables are discussed. In addition,t-test comparisons are presented to determine if significantdifferences exist between variables. Finally, other findings ofinterest are reported as they relate to the present investigation.A.^DESCRIPTIVE STATISTICS1.^Self-Concept RatingsThe Self-Description Questionnaire-1 (SDQ-1) was used tomeasure eleven facets of self-concept as follows: The eightindividual self-concept scales each consist of eight items andinclude four areas of nonacademic self-concept (Physical Abilities(PHYS), Physical Appearance (APPR), Peer Relations (PEER), andParent Relations (PRNT)), three areas of academic self-concept(Reading (READ), Mathematics (MATH), and General-School (SCHL)),and a General-Self (GENL) scale. The three total self-conceptscales (Total Nonacademic (NACD), Total Academic (ACD), and TotalSelf (TOTSLF)), each consist of scores based on individual scalescore averages. For each of the eight individual scales and the104105three total scales, the lowest possible raw score is 8 and thehighest possible raw score is 40. SDQ-1 responses were calculatedfor all 11 scales for both the 57 ESL students and the 57corresponding teacher perceptions involved in this study.As noted in Chapter III, the fifty-seven teacher ratingsrepresent perceptions based upon responses from the eight primaryESL teachers and the thirteen collaborating teachers. In order tosimplify the reporting of SDQ-1 findings, scale scoreabbreviations PHYS, APPR, PEER, PRNT, READ, MATH, SCHL, GENL,NACD, ACD, and TOTSLF refer to the ESL student self-ratings,whereas the abbreviations TPHYS, TAPPR, TPEER, TPRNT, TREAD,TMATH, TSCHL, TGENL, TNACD, TACD, and TTOTSLF refer to the teacherperception ratings.Table 4-1 shows the means, standard deviations, and ranges ofself-concept responses obtained on the SDQ-1 for the presentsample of 57 ESL students and their corresponding teacher ratingsor perceptions. Comparisons to the normative sample described inthe SDQ-1 manual (Marsh, 1988a) are also presented.2.^ESL Self -Concept and Normative ComparisonsIn general, mean self-concept scores for the ESL studentrelated closely to those for the normative sample on the SDQ-1.However, of the 11 possible scale score comparisons, mean ESLstudent self-concept scores tended to be consistently lower thanthose for the normative sample, except in the area of Mathematics(MATH) where the ESL student mean was higher. The mean score forESL student MATH self-concept was 32.56 (SD=6.84), whereas themean score for the normative sample was 28.78 (SD=8.83).Table 4-1Student, Teacher, and Normative Comparisons of Means and StandardDeviations on the SDO-1SDQ-1 Scale^Present^SampleMean^SD RangeNorm SampleMean^SDPHYS 27.21 6.30 13-39 32.65 6.28TPHYS 28.33 8.76 8-40APPR 25.91 5.30 9-40 27.49 8.48TAPPR 28.30* 6.02 16-40PEER 28.44 4.97 12-40 31.01 6.51TPEER 28.35 5.89 10-40PRNT 33.68 4.38 21-40 35.45 5.13TPRNT 28.35** 5.89 10-40READ 26.18 6.49 8-37 31.28 7.48TREAD 27.05 7.04 9-40MATH 32.56 6.84 15-40 28.78 8.83TMATH 32.54 6.53 10-40SCHL 27.26 4.78 17-37 28.55 7.05TSCHL 27.91 6.27 14-40GENL 28.90 4.85 12-40 32.99 5.60TGENL 29.75 4.31 21-40NACD 28.81 3.48 22-37a 31.77 4.96TNACD 28.46 4.18 19-38aACD 28.67 4.28 19-38a 29.53 6.17TACD 29.17 5.56 15-40aTOTSLF 28.74 3.33 21-37a 30.89 4.69TTOTSLF 28.82 4.16 20-38aNote.Present sample N=57; Normative sample N=3,563'T' refers to teacher ratings.Scale scores may range from 8 (lowest) to 40 (highest).a=Range scores rounded to nearest whole number.* Significant t-test differences computed at the .05 level.** Significant t-test differences computed at the .001 level.106107For both the ESL student and the normative sample, thehighest mean self-concept score was in the area of ParentRelations (PRNT: ESL M=33.68, SD=4.38; Norm M=35.45, SD=5.13) andlowest in the area of Physical Appearance (APPR: ESL M=25.91,SD=5.30; Norm M=27.49, SD=8.48). For the ESL student, the meanself-concept score for Reading was lower than the mean self-concept score for Mathematics (READ: ESL M=26.18, SD=6.49 andMATH: ESL M=32.56, SD=6.84). However, for the normative sample,the reverse was true such that the mean self-concept in Readingwas higher than the mean self-concept in Mathematics (READ: NormM=31.28, SD=7.48 and MATH: Norm M=28.78, SD=8.83).Overall, standard deviation scores for the ESL student tendedto vary from those for the normative sample. Consistently higherstandard deviation scores were reported for the normative samplewith the exception of self-concept ratings for Physical Abilitieswhich was very similar for both the ESL student and the normativesample (PHYS: ESL M=27.21, SD=6.30; Norm M=32.65, SD=6.28). Thehighest standard deviation score for the ESL student was inMathematics self-concept (MATH: ESL M=32.56, SD=6.84), followed byReading self-concept (READ: ESL: M=26.18, SD=6.49) and the lowestwas in Total-Self (TOTSLF: ESL M=28.74, SD=3.33). For thenormative sample, the highest standard deviation score was also inMathematics (MATH: Norm M=28.78, SD=8.83), followed by PhysicalAppearance (APPR: Norm M=27.49, SD=8.48). Similar to the presentESL sample, the lowest standard deviation score for the normativesample was in Total-Self self-concept (TOTSLF: Norm M=30.89,SD=4.69).1083.^ESL Self-Concept and Teacher Perception ComparisonsIn general, the mean teacher perception ratings were slightlyhigher than the ESL student self-concept ratings in most areas.However, for 3 of the 11 (27%) SDQ-1 scales, the range of scoresfor teacher ratings (TAPPR, TREAD, and TGENL) were narrower thanthose for the ESL student self-ratings. In all three cases, therange of teacher perception scores started at a higher level thanthose for the ESL student. Of the 11 SDQ-1 mean scale scorecomparisons between ESL student and teacher perceptions, only twowere statistically significant, Physical Appearance (APPR-TAPPR:t=-2.39, p<.05) and Parent Relations (PRNT-TPRNT: t=5.72, p<.001).For Physical Appearance, the mean ESL student self-ratingscore (APPR) was 25.91 (SD=5.30), whereas the mean teacherperception score (TAPPR) was 28.30 (SD=6.02). The difference of -2.386 (SD=7.53) between the two scores indicates a significantlylower student self-concept rating in the area of PhysicalAppearance. Also, for Physical Appearance, the range of scoresfor ESL self-concept (APPR: 9-40) was wider than for teacherperception ratings (TAPPR: 16-40).For Parent Relations the mean ESL student self-rating score(PRNT) was 33.69 (SD=4.38), whereas the mean teacher rating(TPRNT) was 28.35 (SD=5.89). The difference of 4.82 (SD=6.37)between the two scores indicates a significantly lower teacherperception rating in the area of Parent Relations. Also, forParent Relations, the range of scores for teacher ratings (TPRNT:10-40) was narrower than for the ESL student (PRNT: 21-40).Interestingly, for both the ESL student and teacherperception ratings in the area of Mathematics, both the mean self-109concept scores and the standard deviations scores were almostexactly the same ((MATH: M=32.56, SD=6.89; TMATH: M=32.54,SD=6.53). However, for Mathematics self-concept, the range ofscores for the ESL student (MATH: 15-40) was narrower than therange of scores for the teacher perception ratings (TMATH: 10-40).Similarly, for Physical Abilities, the range of scores for the ESLstudent (PHYS: 13-39) was narrower than for the teacher perceptionratings (TPHYS: 8-40).Also, mean self-concept ratings for both the ESL student andthe teacher perception were very similar in the areas of General-School or all school subjects (SCHL: M=27.26, SD=4.78; TSCHL:M=27.91, SD=6.27), Total Nonacademic self-concept (NACD: M=28.81,SD=3.48, TNACD: M=28.46, SD=4.18), and Total-Self self-concept(TOTSLF: M=28.74, SD=3.33; TTOTSLF: M=38.82, SD=4.16).Furthermore, for self-concept ratings in Mathematics, General-School, Total Nonacademic, and Total-Self areas, mean scorecomparisons between the ESL student and teacher perception ratingsappear more similar and consistent with each other than with theSDQ-1 normative sample.4.^Reading and Mathematics Academic AchievementReading and Mathematics academic achievement scores werecalculated for the 57 ESL students involved in this study based oneach student's most recent (spring) report card grades. If reportcard grades were not available, teachers were asked to rate thestudent's academic performance. Academic grades and teacherratings of achievement were calculated on an 8-point scale withpossible academic achievement scores ranging from a high of 1 to a110low of 8. Scores were rated as follows: 1=A, 2=B+, 3=B, 4=B-,5=C+, 6=C, 7=C-, 8=D.Table 4-2 shows the distribution of Reading and Mathematicsacademic achievement scores corresponding to point scale valuesand letter grades or ratings of achievement for the sample of 57ESL students involved in this study.Table 4-2Djstrihution of Reading and Mathematics Academic Achievement Scores RMARK^ MMARKScoreValuesFrequency Percent Cum.% Frequency Percent Cum.%1 A 6 10.5 10.5 24 42.1 42.12 B+ 1 1.8 12.3 --3 B 11 19.3 31.6 23 40.1 82.54 B- 1 1.8 33.3 --5 C+ 5 10.5 43.9 4 7.0 89.56 C 16 28.1 71.9 5 8.8 98.27 C- 10 17.5 89.5 1 1.8 100.08 D 6 10.5 100.0Note.N=57Score values range from 1 or A (highest) to 8 or D (lowest).RMARK refers to Reading mark or rating score.MMARK refers to Mathematics mark or rating score.A comparison of the academic achievement scores in Reading(RMARK) and Mathematics (MMARK) as presented in Table 4-2indicates differences in frequency distributions between the two.For example, 47 (82.5%) of the 57 ESL students obtainedMathematics achievement scores of 3 or better, corresponding toletter grades or ratings of A and B. However, only 18 students(31.6%) obtained similar achievement scores in Reading with the111majority, or 26 (45.6%), of the 57 ESL students obtaining Readingachievement scores between 6 and 7 or C and C-.Table 4-3 shows the means, standard deviations, and rangesobtained for both Reading (RMARK) and Mathematics (MMARK) asindicated by either the student's report card grades or teacherratings of the student's academic achievement.Table 4-3Means and Standard Deviations for Reading and Mathematics AcademicAchievement Academic^Mean^SD^Range^Mean^t-testAchievement DifferenceRMARK^5.07^2.15^1-82.44^9.01*MMARK 2.63^1.72^1-7Note.N=57Academic achievement scores range from a possible 1 (highest)to 8 (lowest).* Significant at the .001 levelOf the 57 students who participated in this study, Readingachievement scores (RMARK) ranged between a minimum of 1 and amaximum of 8, corresponding to letter grades or ratings of A to D.The average or mean Reading score for the sample was 5.07 whichindicates a letter grade or rating of C+.On the other hand, Mathematics achievement scores (MMARK)ranged between a minimum of 1 and a maximum of 7, corresponding toletter grades or ratings of A to C-, on a possible scale of 1 to 8or A to D. The sample mean score was 2.63, indicating a lettergrade or rating between B and B+.A t-test was conducted to determine the extent of differencebetween the means for Reading (RMARK) and Mathematics (MMARK)academic achievement scores. The resulting mean score differenceof 2.44 (SD=2.04) was highly significant (t=9.01,p<.001).B.^PEARSON PRODUCT MOMENT CORRELATIONSPearson correlation matrices were computed in order todetermine the relationships among and between the followingvariables: ESL self-concept, teacher perception of ESL self-concept, and academic achievement. Correlations were calculatedfor all 11 scales on the SDQ-1 to include both student and teacherratings for all 57 subjects involved in the present investigation.In addition, correlation coefficients were computed for academicachievement in Reading and in Mathematics as indicated by thestudent's report card grades or teacher ratings of studentachievement. Correlational directions involving academicachievement data (RMARK and MMARK) were reversed in order toparallel data collection on the SDQ-1.A 24 variable by 24 variable matrix generated 276 possiblePearson Product-Moment correlations, excluding each variablecorrelated with itself. Of these 276 correlations, 129 (47%) werefound to be significant, including 36 (13%) at p<.05, 17 (6%) atp<.01, and 76 (28%) at p<.001. The overall results are presentedin Appendix H. For clarity of presentation, the results of thePearson correlations are reported in the subsequent sections,corresponding to Tables 4-4 to 4-7 as follows: Table 4-4 refersto the correlations between the ESL self-concept scale scores on112113the SDQ-1 for the 57 ESL students involved in this study. Table4-5 presents the correlations between teacher perception scalescores for the same 57 ESL students. Table 4-6 displays thecorrelations between ESL self-concept and teacher perception scalescores. Table 4-7 refers to the correlations between ESL self-concept scale scores and academic achievement scores in Readingand in Mathematics. Table 4-7 also presents the correlationsbetween teacher perception scale scores and academic achievementscores in Reading and in Mathematics and it notes the correlationbetween academic scores in Reading and in Mathematics.1. Correlations between ESL Self-Concept ScoresA Pearson correlation matrix was computed between scalescores on the SDQ-1 in order to determine the relationshipsbetween various facets of self-concept for the ESL student. Theresults are presented in Table 4-4 and may be summarized asfollows:The correlations between ESL self-concept scale scoresgenerated 41 (75%) significant coefficients out of a possible 55,including 5 at p<.05, 6 at p<.01. and 30 at p<.001 (see Table 4-4). Correlations ranged from nonsignificant and near zero betweenPRNT and MATH (r=.0013) to highly significant and substantialbetween ACD and TOTSLF (r=.89, p<.001). Overall, for the ESLstudent, the strongest self-concept correlations were between thetotal scales themselves (TOTSLF and ACD: r=.89, p<.001; TOTSLF andNACD: r=.82, p<.001) and between NACD and PEER (r=.82, p<.001),followed by TOTSLF and APPR (r=.77, p<.001), NACD and SCHL (r=.76,Table 4-4Pearson Product-Moment Correlations for ESL Student Self-Concept PHYSPHYS^APPR1.00PEER PRNT READ MATH SCHL GENL NACD ACD TOTSLFAPPR .35**^1.00PEER .44***^.49*** 1.00PRNT -.15^.06 .26* 1.00READ .10 .42*** .38** .19 1.00MATH .08^.38** .05 .00 .06 1.00SCHL .27*^.52*** .21 .11 .43*** .33** 1.00GENL .23*^.64*** .46*** .25* .40*** .35** .56*** 1.00NACD .69***^.73*** .82*** .36** .40*** .20 .43*** .59*** 1.00ACD .19^.60*** .30* .14 .70* ** .68*** .76*** .60*** .47*** 1.00TOTSLF .49***^.77*** .62*** .28* .66*** .54*** .71*** .69*** .82*** .89*** 1.00Note. N=57 student ratingsCorrelations are rounded to two decimal places.* Significant 'trend' at the .05 level** Significant at the .01 level*** Significant at the .001 level115p<.001), TOTSLF and SCHL (r=.71, p<.001), TOTSLF and GENL (r=.69,p<.001), NACD and PHYS (r=.69, p<.001).In general, of the 8 correlations between individual self-concept scales and matching total self-concept scales, 7 out of 8,or the majority (88%), correlated substantially and significantly(PHYS-NACD: r=.69, p<.001; APPR-NACD: r=.73, p<.001; PEER-NACD:r=.82, p<.001; READ-ACD: r=.70, p<.001; MATH-ACD: r=.68, p<.001;SCHL-ACD: r=.68, p<.001; and GENL-TOTSLF: r=.69, p<.001,respectively) as expected from previous SDQ studies (Marsh,1988a). In fact, only self-concept in Parent Relations correlatedmoderately and less significantly (PRNT-NACD: r=.36, p<.01).Interestingly, ESL self-concept in Physical Appearance andPeer Relations also correlated significantly, although relativelyless strongly with Total Academic self-concept (APPR-ACD: r=.60,p<.001 and PEER-ACD: r=.30, p<.05, respectively). Similarly,General-School or all school subjects also correlated moderatelyand significantly with Total Nonacademic self-concept (SCHL-NACD:r=.43, p<.001).Also, of the 3 correlations between the Total self-conceptscales, both Total Nonacademic and Total Academic self-conceptscorrelated strongly with the Total-Self scale (NACD=TOTSLF: r=.82and ACD-TOTSLF: r=.89; p<.001, respectively) whereas TotalNonacademic and Total Academic self-concept scales correlatedsignificantly but only moderately (NACD-ACD: r=.47, p<.001). Asexpected from the review of previous SDQ research (Marsh, 1988a),the present correlation between READ an MATH self-concepts wasnonsignificant and near zero (r=.04).Interestingly, the weakest and the fewest correlations forthe ESL student were found between PRNT and the other individualself-concept scales (ranging from PRNT-MATH: r=.00 to PRNT-NACD:r=.36, p<.01). Also, one inverse and nonsignificant correlationwas found for ESL self-concept comparisons between PRNT and PHYS(r=-.15).2.^Correlations between Teacher Perception ScoresA Pearson correlation matrix was computed between scalescores on the SDQ-1 in order to determine the relationshipsbetween various facets of self-concept for teacher ratings of ESLstudent self-concept. The results are presented in Table 4-5 andmay be summarized as follows:The correlations between scale scores for teacher perceptionof ESL student self-concept generated a total of 47 (86%)significant coefficients out of a possible 55 correlations,including 9 at p<.05., 6 at p<.01, and 32 at p<.001. Thecorrelations ranged from nonsignificant and near zero betweenTPEER and TMATH (r=.04) to highly significant and substantialbetween TACD and SCHL (r=.94, p<.001). Overall, the number ofsubstantial correlations appears somewhat greater and strongeramong the teacher ratings than those reported for the ESLstudents. For teacher inferences, the highest self-conceptcorrelations were found among the total scales and in particularin relation to academic areas (TACD and SCHL: r=.94, p<.001,TTOTSLF and TSCHL: r=.89, p<.001; TTOTSLF and TACD: r=.89, p<.001;TACD and TREAD: r=.84, p<.001; TTREAD and TSCHL: r=.79, p<.001;TSCHL and TGENL: r=.79, p<.001; TACD and TGENL: r=.74, p<.001;116Table 4-5Pearson Product-Moment Correlations for Teacher PerceptionTPHYSTAPPRTPEERTPRNTTREADTMATHTSCHLTGENLTNACDTACDTTOTSLFTPHYS1.00.29*.25*-.*.24*.67***.26*.51***TAPPR1.00.48***.10.23*.29*.47***.70***.71***.39***.62***TPEER1.00.37**.19.04.30*.53***.77***.21.53***TPRNT1.00.38**.12.34**.44***.39***.33**.42***TREAD1.00.31**.79***.61***.30*.84***.71***TMATH1.00.59***.48***.32**.74***.66***TSCHL1.00.79***.53***.94***.89***TGENL1.00.70***.74***.85***TNACD1.00.45***.80***TACD1.00.89***TTOTSLF1.00Note. N=57 student ratings and 57 teacher ratingsCorrelations are rounded to two decimal places.* Significant 'trend' at the .05 level** Significant at the .01 level*** Significant at the .001 level118TTOTSLF and TREAD: r=.71, p<.001), followed by TNACD and TAPPR(r=.71, p<.001) and TNACD and TGENL (r=.70, p<.001).Similar to the findings for ESL self-concept, 7 out of 8, orthe majority (88%) of individual self-concept scales correlatedsignificantly and substantially with matching total self-conceptscales (TPHYS-TNACD: r=.67; TAPPR-TNACD: r=.71; TPEER-TNACD:r=.77; TREAD-TACD: r=.84; TMATH-TACD: r=.74; TSCHL-TACD: r=.94;and TGENL-TTOTSLF: r=.85; p<.001, respectively). Furthermore,similar to the findings for ESL student self-concept, teacherperception ratings for Parent Relations correlated significantlybut only moderately with Total Nonacademic self-concept (TPRNT-TNACD: r=.39, p<.001). Interestingly and unlike the findings forthe ESL student, teacher perception ratings for Parent Relationsalso correlated significantly, although moderately low, withteacher ratings for ESL student self-concept in Reading, General-School, and Total Academic self-concept areas (TPRNT-TREAD: r=.38,p<.01; TPRNT-TSCHL: r=.34, p<.01; and TPRNT-TACD: r=.33, p<.01).However, similar to the ESL student self-concept findings, teacherperception ratings for Parent Relations did not correlatesignificantly with teacher ratings of ESL student self-concept inMathematics (TPRNT-TMATH: r=.12). Also, similar to the ESLstudent self-rating, teacher perception of General-School or allschool subjects correlated significantly, although less strongly,with Total Nonacademic self-concept (TSCHL-TNACD: r=.53, p<.001).Of the 3 significant correlations between the total self-concept scales, strong correlations were found between TotalNonacademic and Total-Self scales and between Total Academic self-concept and Total-Self (TNACD-TTOTSLF: r=.80 and TACD-TTOTSLF:119r=.89; p<.001, respectively) and a moderate correlation wasindicated between Total Nonacademic and Total Academic self-concept scales (TNACD-TACD: r=.45, p<.001). These results alsoappear consistent with correlational findings for the ESL studentamong and between total scale scores.Unlike the near zero correlation finding between Reading andMath self-concepts for the ESL student, the correlation betweenTREAD and TMATH was significant but moderately low (r=.31, p<.01).The weakest correlations for teacher's perceptions were foundbetween the TPHYS scale and the other self-concept scales,although 5 correlations were significant at the p<.05 and 2 atp<.001. Similar to findings for the ESL student, one inverse andnonsignificant correlation was found between TPRNT and TPHYS (r=.-15).3. Correlations between ESL Self-Concept and TeacherPerceptionA Pearson correlation matrix was computed between ESL studentself-concept and teacher perception of ESL student self-concept onthe SDQ-1 in order to determine the relationships between variousfacets of self-other agreement, particularly in matching areas ofself-concept. The results are displayed in Table 4-6 and may besummarized and follows:In general, correlations between ESL student and teacherperception appeared to be fewer in number and weaker thancorrelations among scale scores measuring ESL student self-conceptand teacher perception ratings of ESL self-concept alone (see alsoTables 4-4 and 4-5). Of the 121 possible correlations betweenTable 4-6Pearson ProduQt-Moment Correlations for ESL Student Self-Concept and Teacher PerceptionPHYS APPR PEER PRNT READ MATH SCHL GENL NACD ACD TOTSLFTPHYS .52*** -.02 .19 .00 -.26* .13 .02 -.12 .30* -.06 .12TAPPR .02 .12 .00 .01 -.08 .10 .01 -.03 .06 .02 .04TPEER .14 .09 .46*** .01 .09 -.11 -.05 .05 .26* -.03 .12TPRNT -.21 -.13 .08 .11 .31** -.19 -.03 -.08 -.09 .04 -.02TREAD -.20 -.24* -.10 .07 .16 -.23 .04 -.02 -.20 -.03 -.12TMATH -.12 -.24 -.28* -.18 -.26* .29* -.19 -.20 -.30* -.05 -.19TSCHL -.14 -.28* -.15 -.02 -.05 -.03 -.08 -.09 -.23* -.07 -.16TGENL -.03 .03 .05 -.02 .12 .11 .10 .06 .01 .16 .11TNACD .27* .02 .29* .04 -.04 .01 -.01 -.08 .24* -.02 .11TACD -.19 -.30* -.21 -.05 -.05 .01 -.09 -.12 -.29* -.05 -.18TTOTSLF .01 -.19 .01 -.01 -.05 .01 -.06 -.12 -.07 -.05 -.07Note. N-57 student ratings and 57 teacher ratingsCorrelations are rounded to two decimal places.^Correlations in boldface refer to student-teacheragreement on matching self-concepts.Significant 'trend' at the .05 level** Significant at the .01 level*** Significant at the .001 level121student and teacher, only 18 (15%) were found to be significant.Of the 18 correlations, the majority or 15 were significant atp<.05 and over half of these or 9, correlated negatively. One ofthe 18 correlations was significant at p<.01 (READ-TPRNT) and 2(11%) were significant at <.001 (PHYS-TPHYS and PEER-TPEER).Furthermore, of the 11 possible correlations for student-teacheragreement in matching self-concept areas, significant correlationswere found on only 4 (36%) of the self-concept scales. Of these,half of the self-concept correlations were significant at p<.05(MATH-TMATH and NACD-TNCAD) and half were significant at p<.001(PHYS-TPHYS and PEER-TPEER).Overall, coefficients between ESL student and teacherperception in matching self-concept areas ranged betweenmoderately strong and significant in Physical Abilities (PHYS-TPHYS: r=.52, p<.001) and Peer Relations (PEER-TPEER: r=.46,p<.001), followed by moderately low and significant 'trends' inMathematics (MATH-TMATH: r=.29, p<.05) and Total Nonacademic self-concept (NACD-TNACD: r=.24, p<.05), to very low and nonsignificantin Parent Relations (PRNT-TPRNT: r=.11), Physical Appearance(APPR-TAPPR: r=.12), and Reading self-concept (READ-TREAD: r=.16),to negative and nonsignificant in total Academic self-concept(ACD-TACD: r=-.05), Total-Self (TOTSLF-TTOTSLF: r=-.07), andGeneral-School self-concept (SCHL-TSCHL: r=.-08). A negative butnonsignificant correlational 'trend' was also noted between ESLstudent self-concept of Physical Abilities and teacher perceptionof student Parent Relations (PHYS-TPRNT: r=-.21).Interestingly and surprisingly, ESL student and teacheragreement in matching self-concept areas appeared to be stronger1 22in nonacademic areas, particularly in terms of specific self-concepts of Physical Abilities and Peer Relations, and in terms ofoverall or Total Nonacademic self-concept. On the other hand,student-teacher agreement in Mathematics self-concept, althoughsignificant, was moderately low, and the correlation betweenstudent and teacher Reading self-concepts was nonsignificant andvery low.The majority of the 121 possible correlations or 103 (85%)were nonsignificant and 69 (57%) were inverse or negative. Also,of the 121 possible correlations, 27 (22%) were positive and nearzero while 35 (29%) were negative and near zero.4.^Correlations between ESL Self-Concept and AcademicAchievementCorrelational coefficients were computed in order todetermine the relationships between various facets of ESL self-concept on the SDQ-1 and student academic achievement in Readingand in Mathematics. Correlational directions involving academicachievement data were reversed in order to parallel datacollection on the SDQ-1. The results are reported in Table 4-7and may be summarized as follows:Correlations between ESL student self-concept and academicachievement in Reading (RMARK) and Mathematics (MMARK) were foundto be mostly nonsignificant and often negative. Of the 11possible correlations between ESL self-concept and Readingacademic achievement, none were found to be significant. Six(55%) of the 11 coefficients were negative and 8 (73%) were nearzero. Correlations between ESL self-concept and ReadingTable 4-7Correlations between Academic Achievement, ESL Self-Concept, and Teacher Perception Academic Achievement and ESL Self-ConceptPHYS^APPR^PEER^PRNT^READ^MATH^SCHL^GENL^NACD^ACD^TOTSLFRMARK^-.20^-.12^-.03^-.00^.05^.01^.03^.04^-.15^.04^-.05MMARK^-.18^-.34**^-.25^-.08^-.17 .12^-.14^-.16^-.32**^-.08^-.22Academic Achievement and Teacher Perception^TPHYS^TAPPR^TPEER^TPRNT^TREAD^TMATH^TSCHL^TGENL^TNACD^TACD^TTOTSLFRMARK^-.03^.16^.27*^.32**^.64***^.25*^.58* **^.48* **^.23*^.58* **^.51* **MMARK^.25*^.17 .15^-.21^.54***^.62***^.55***^.39***^.31**^.68***^.61***Note. N=57 student ratings and 57 teacher ratings of ESL student self-conceptCorrelations are rounded to two decimal places. Correlational directions involving academicachievement (RMARK and MMARK) are reversed to parallel data collection on the SDQ-1.RMARK-MMARK: r=.46, p<.001.* Significant 'trend' at the .05 level** Significant at the .01 level*** Significant at the .001 level1 24achievement ranged from negative and near significant (PHYS-RMARK:r=-.20) to zero and nonsignificant (PRNT-RMARK: r=.00). Notably,ESL Reading self-concept did not correlate significantly withReading achievement and, in fact, the correlation between the twowas near zero (READ-RMARK r=.05).Of the 11 possible correlations between ESL self-concept andMathematics, only 2 (18%) were found to be significant and bothwere negative and moderately low (APPR-MMARK: r=-.34, p<.01; NACD-MMARK: r=-.32, p<.01). Ten (91%) of the 11 correlations werenegative and 2 (18%) were near zero. Overall, correlationsbetween ESL self-concept and Mathematics achievement rangedbetween r=-.34, p<.01 (APPR-MMARK) to nonsignificant r=.08 (PRNT-MMARK and ACD-MMARK, respectively). The one positive correlationbetween ESL Mathematics self-concept and Mathematics academicachievement was also nonsignificant and very low (MATH-MMARK:r=.12). Although not quite significant at p<.05, a negativecorrelational trend was noted between ESL student Total-Self self-concept and academic achievement in Mathematics (TOTSLF-MMARK:r=.-22).5.^Correlations between Teacher Perception and AcademicAchievementCorrelational coefficients were computed in order todetermine the relationships between teacher ratings of variousfacets of ESL self-concept on the SDQ-1 and student academicachievement in Reading and in Mathematics. Correlationaldirections involving academic achievement data (RMARK and MMARK)were reversed in order to parallel data collection on the SDQ-1.1 25The results are also reported in Table 4-7 and may be summarizedas follows:In contrast to the correlations between ESL student self-concept and academic achievement, correlations between teacherperception of ESL self-concept and academic achievement in Reading(RMARK) and Mathematics (MMARK) were mostly significant andpositive. Overall, the highest correlation between teacherratings and academic achievement in Reading was with the specificarea of Reading self-concept (TREAD-RMARK: r=.64, p<.001), whereasthe highest correlation between teacher ratings and academicachievement in Mathematics was with Total Academic self-concept,followed closely with specific Mathematics self-concept (TACD-MMARK: r=.68 and TMATH-MMARK: r=.62; p<.001, respectively).Of the 11 correlations between teacher ratings and Readingachievement (RMARK), 9 (82%) were found to be significant. Three(33%) of the 9 correlations were significant at p<.05, 1 (11%) atp<.01, and the majority, or 5 (56%), were significant at p<.001.Only 1 (9%) of the 11 correlations was nonsignificant (TAPPR-RMARK: r=.16) and only 1 (9%) was nonsignificant and negative(TPHYS-RMARK: r=-.03). Coefficients ranged from moderately strongand highly significant (TREAD-RMARK) to negative, near zero, andnonsignificant (TAPPR-RMARK). Positive correlations ranged frommoderately strong and highly significant (TREAD-READ: r=.64,p=.001) to low and nonsignificant (TAPPR-RMARK: r=.16). Readingacademic achievement also correlated substantially andsignificantly with Reading, General-School, Total Academic, Total-Self, and General-Self self-concepts (TREAD-RMARK: r=.64, TSCHL-RMARK: r=.58, TACD-RMARK: r=.58, TTOTSLF-RMARK: r=.51, and TGENL-1 26RMARK: r=.48, respectively; p<.001), followed by significant butmoderately low correlations with Parent Relations, Peer Relations,Mathematics, and Total Nonacademic self-concepts (TPRNT-RMARK:r=.32 , p<.01; TPEER-RMARK: r=.27, p<.05; TMATH-RMARK: r=.25,p<.05; and TNACD-RMARK: r=.23, p<.05).Of the 11 correlations between teacher perception of ESLstudent self-concept and academic achievement in Mathematics, 8(73%) were found to be significant. Of these 8 correlations, 1(13%) was significant at p<.05, 1 (13%) at p<.01, and themajority, or 6 (75%), were significant at p<.001. Only onecorrelation was negative and almost a significant 'trend' at p<.05(TPRNT-MMARK: r=-.21). Coefficients ranged from substantial andhighly significant (TACD-MMARK: r=.68, p<.001) to low andnonsignificant (TPEER-MMARK: r=.15). Mathematics academicachievement also correlated substantially with teacher ratings ofESL student self-concepts in Mathematics, Total-Self, General-School, and Reading self-concept areas (TMATH-MMARK: r=.62,TTOTSLF-MMARK: r=.61, TSCHL-MMARK: r=.55, TREAD-MMARK: r=.54;p<.001, respectively), followed by significant but lowercorrelations with General-Self, Total Nonacademic, and PhysicalAbilities self-concepts areas (TGENL-MMARK: r=.39, p<.001; TNACD-MMARK: r=.31, p<.01; and TPHYS-MMARK: r=.25, p<.05). Nosignificant correlations were found between academic achievementin Mathematics and teacher perception of ESL student self-conceptsin Physical Appearance or Peer Relations (r=.17 and r=.15,respectively). A negative and almost significant 'trend' at p<.05was also noted between Mathematics achievement and teacherperception of Parent Relations (TPRNT-MMARK: r=-.21).1 276.^Correlations between Reading and Mathematics AcademicAchievementCorrelations were computed between academic achievementscores in Reading and in Mathematics in order to determine theirrelationship. Overall, the results reveal that ESL academicachievement in Reading and Mathematics correlated moderately andsignificantly with each other (RMARK-MMARK: r=.46, p<.001). Thisfinding is consistent with the results of previous studies (Marsh,1988a).C. T-TESTSAdditional t-test comparisons were conducted in order toexamine the differences between various demographic and backgroundvariables as they might relate to ESL student self-concept andteacher perception of ESL student self-concept in the presentinvestigation. Only significant findings of interest are reportedin Table 4-8 and may be explored further in subsequent or futureresearch. Overall, significant differences between means werefound for both student and teacher self-concept ratings, in termsof sex, district, birth place, and ethnic group comparisons. Theresults are summarized in the following sections.1. Sex DifferencesT-test comparisons indicated significant sex differencesbetween males (n=28) and females (n=29) in Physical Abilitiesself-concepts for both ESL student (t=4.23, p<.001) and teacherperception ratings (t=2.87, p<.01). Self-concept ratings by1 28Table 4-8Means, Standard Deviations, and t-Test Values for SignificantDemographic Differences Related to ESL Self-Concept and TeacherPerceptionVariable Scale^Group Mean SD t-test DF^ProbabilitySex:PHYS^male (n=28) 30.38 5.46 4.23*** 55 .000female (n=29) 24.17 5.58READ^male 24.32 7.18 - 2.19* 55 .033female 27.97 5.28TPHYS^male 31.50 5.43 2.87** 42.87 .006female 25.28 10.26TREAD^male 25.11 7.34 - 2.11* 55 .039female 28.93 6.30District:MATH^urban (n=34) 30.41 7.24 - 3.10** 55 .003suburban (n=23) 35.74 4.78TMATH^urban 30.29 6.84 - 3.76*** 54.85 .000suburban 35.87 4.35Birth Place:TMATH^Canada (n=6) 26.67 8.52 - 2.43* 55 .018other (n=51) 33.24 5.99TNACD^Canada 25.00 4.21 - 2.22* 55 .031other 28.87 4.02TTOTSLF^Canada 25.14 4.17 - 2.38* 55 .021other 29.25 3.98Bthnicity:PEER^H.K.Chinese (n=40) 27.45 4.56 - 2.40* 55 .020other (n=17) 30.76 5.24NACD^H.K.Chinese 28.23 3.00 -^1.98 55 .052other 30.17 4.18TMATH H.K.Chinese 34.15 5.08 2.55* 21.62 .018other 28.76 8.04TACD^H.K.Chinese 30.12 5.37 2.03* 55 .047other 26.94 5.49Note. N=57; H.K. refers to Hong Kong Chinese students* Significant at the .05 level** Significant at the .01 level*** Significant at the .001 level1 29students and by teachers were higher for males (PHYS: M=30.36,SD=5.46 and TPHYS: M=31.50, SD=5.43, respectively) than forfemales (PHYS: M=24.17, SD=5.58 and TPHYS: M=24.17; SD=5.58,respectively).T-test comparisons in Reading also indicated significant sexdifferences for both ESL students (t=2.19, p<.05) and teacherperception ratings (t=2.11, p<.05). Reading self-concept ratingsby ESL students and by teachers were higher for females (READ:M=27.97, SD=5.28 and TREAD: M=28.93, SD=6.30) than for males(READ: M=24.32, SD=7.18 and TREAD: M=25.11, SD=7.34). The presentfindings for both Physical Abilities and Reading self-conceptsappear to be consistent with sex stereotypes as reported in theliterature (e.g., Meece et al, 1982). No other significant sexdifferences were found for the other self-concept areas.2.^District DifferencesT-test comparisons between districts indicated significantdifferences between urban (n=34) and suburban (n=23) ESL studentsin Mathematics self-concept both for ESL student self-ratings(t=3.10, p<.01) and for teacher perception ratings (t=3.76,p<.001). For both student and teacher respondents, meanMathematics self-concept ratings were higher for the suburban ESLstudents (MATH: M=35.74, SD=4.78 and TMATH: M=35.87, SD=4.35,respectively) than for the urban ESL students (MATH: M=30.41,SD=7.24 and TMATH: M=30.29, SD=4.35, respectively).1303.Birth PlaceT-test comparisons between Canadian-born Chinese ESL students(n=6) and ESL students born in other countries (n=51) revealedsignificant differences in teacher perception of ESL student self-concept in the areas of Mathematics (TMATH: t=2.43, p<.05), TotalAcademic (TNACD: t=2.22, p<.05), and Total-Self (TTOTSLF: t=2.38,p<.05). For teacher perception ratings in these areas, meanscores were higher for the ESL students born in other countries(TMATH: M=33.24, SD=5.99; TNACD: M=28.87, SD=4.02; and TTOTSLF:M=29.25, SD=3.98, respectively) than for the Canadian-born ESLstudents (TMATH: M=26.67, SD=8.52; TNACD: M=25.00, SD=4.21; andTTOTSLF: M=25.14, SD=4.17, respectively). However, the numbers ofstudents represented in each sample are notably small anddisproportionate.4.^EthnicityT-test comparisons between the ESL students from Hong Kong(n=40) and ESL students from all other ethnic groups (n=17)revealed a significant difference in ESL student self-concept forPeer Relations (PEER: t=2.40, p<.05). A similar directional'trend', although not quite significant at p<.05, was noted in thearea of Total Nonacademic self-concept for the ESL students (NACD:t=1.98). For both Peer Relations and Total Nonacademic self-concepts, mean ratings were higher for the other ethnic groups(PEER: M=30.76, SD=5.24 and NACD: M=30.17, SD=4.18, respectively)than for the Hong Kong Chinese students (Peer: M=27.25, SD=4.56and NACD: M=28.23, SD=3.00, respectively).131For teacher ratings of ESL student self-concept, t-testcomparisons indicated significant differences between the studentsfrom Hong Kong and the students from other other ethnic groups inMathematics (TMATH: t=2.55, p<.05) and in Total Academic self-concepts (TACD: t=2.03, p<.05). For both Mathematics and TotalAcademic self-concept ratings by teachers, mean ratings werehigher for the Hong Kong Chinese ESL students (TMATH: M=34.15,SD=5.08 and TACD: M=30.12, SD=5.37, respectively) than for the ESLstudents from the other ethnic groups (TMATH: M=28.76, SD=8.04 andTACD: M=26.94, SD=5.49, respectively). Again, however, the smalland disproportionate numbers of students representing each sampleare noteworthy.132V: DISCUSS ION AND CONCLUSIONThe purpose of the present exploratory study as it applies toEnglish as a second language (ESL) students was two-fold:1. To examine the relationship between a student's self-concept and teacher perception of that student's self-concept; and,2. To examine the extent to which student self-concept andteacher perception of student self-concept relate toacademic achievement in Reading and in Mathematics.For this investigation, fifty-seven ESL students, ages 10 to12 years and currently in grades 5 and 6 were recruited from twoschool districts along with their corresponding teachers. Boththe ESL students and their teachers were administered the Self-Description Questionnaire-1. The ESL student's most recent reportcard grades or teacher ratings of student academic achievementwere collected. Additional background and demographicinformation, regarding the students and the schools, was gatheredfrom both teacher and principal reports. The ESL student self-concept data was then compared and contrasted with the inferredteacher ratings of ESL student self-concept. Similarly, both theESL student self-ratings and the teacher perception ratings werecompared and contrasted with the student's academic achievement inReading and in Mathematics. Statistical analyses utilized in the133present study included descriptive statistics, Pearson Product-Moment correlations, and t-test comparisons.This final chapter discusses the results of the present studyin terms of the five specific research questions explored. Itrelates the findings reported in Chapter IV to the relevantliterature on self-concept, the assessment of ESL students, andminority studies as reviewed in Chapter II. It also relates thepresent findings to current trends in education. Conclusions aredrawn as to the implications and generalizability of the resultsas they relate to students, teachers, and programs. The chapterconcludes with some of the possible limitations of the presentstudy and offers recommendations for further research.A.^DISCUSSION OF FINDINGSThis section deals individually with each of the fivespecific research questions posed regarding the nature of ESLstudent self-concept and teacher perception of ESL student self-concept, the relationship between the two, and the relationship ofboth to academic achievement in Reading and in Mathematics. Othersignificant findings of interest are also discussed as they relateto the present investigation.1.^ESL Self-ConceptQuestion One:What is the self-concept of the ESL student?134Descriptive analyses of responses on the Self-DescriptionQuestionnaire-1 (SDQ-1) suggest that, the standard deviationscores for the normative sample were consistently wider than forthe ESL students in this study. This finding suggests that thepresent sample was more homogeneous than the norm. Only thestandard deviation scores for Physical Abilities were found to besimilar for both. Overall, the pattern of self-concept for thishomogeneous group of ESL students appears consistently lower whencompared with the normative sample with the exception of ESL self-concept in Mathematics.However, for both the ESL students involved in this study andthe normative sample, the findings suggest a similar pattern ofhighest self-concepts in Parent Relations and lowest self-conceptsin Physical Appearance. Also, the majority of the correlationsbetween ESL self-concept scale scores were significant and all butone of the individual scale scores correlated substantially andsignificantly with the scale scores to which they were mostlogically related. In general, this pattern of high and low self-concepts and scale score correlations appear consistent withprevious findings by Marsh (1988a) and his colleagues as reviewedin the SDQ research studies in Chapter II. These findings tend tosuggest that, although not normed on the present population, theSDQ-1 may also be a valuable instrument to add to a test batterywhen attempting to assess the ESL student in order to shed lighton possible discrepancies in various areas of self-concept.Although the ESL students in this study appeared to have asomewhat lower self-concept than the students in the normativesample, this discrepancy may be partially explained by previous135studies investigating the relationships among SES, school-averageability, and self-concept (e.g., Marsh, 1988b, 1987a; Marsh &Parker, 1984). According to the background information gatheredfrom informal principals' reports, the majority of the ESLstudents in the present sample came from at least middle-classschools with at least average or better school-average academicperformance. Marsh's (1987a) BFLPE model postulates that studentsin higher SES/ability schools may have a lower self-concept,despite higher academic achievement, than students in lowerSES/ability schools because they tend to make relativisticcomparisons to other students within their particular school andsetting. The present finding also appears consistent with thelower self-concepts suggested by the literature for othersegregated and 'minority' groups within the umbrella of specialeducation.The present results may have implications as to the value ofservice delivery and program model (e.g., mainstreaming andintegration versus the use of separate or pull-out programs) andtheir respective effects on student self-concept, particularly inacademic areas but also in nonacademic areas. At present, theimplications of various specific program models have not beenfully explored in the research. Investigations appear only tohave created more questions than answers. The issues regardinghow best to teach the ESL student, as discussed in the literaturereview, continues to leave much room for speculation. However,recent surveys and evaluations (Collier, 1989; Cummins, 1984;Lambert, 1981), based on language proficiency and its relation toacademic achievement, theory, and research have tended to support136integrated and bilingual settings which may also leave room forthe development of cultural identity in two languages.Not surprisingly, the higher mean self-concept score observedfor the ESL students in Mathematics, coupled with a relativelylower mean self-concept score in Reading, appears reversedcompared to the normative group. This finding seems to beconsistent with the psychoeducational assessment literature (Lynn,Pagliari, & Chan, 1988; Tam, 1990) indicating a profile ofnonverbal or visual-spatial strengths for Orientals compared totheir verbal abilities and compared to the general population.Also, according to the literature (Cummins, 1984), patterns ofachievement, particularly the attainment of grade level norms,appear faster in math than for language/content areas for Orientalimmigrants. Although mathematical aptitude may often requireverbal ability, the ESL students in this study may have used adefinition of mathematics as computational (a strength) ratherthan word-problem oriented (a challenge).The Internal/External frame of reference model proposed byMarsh (Marsh, 1984, 1986d) and others may help to explain theconsistently higher mathematics self-concept observed for the ESLstudent compared to both the normative sample and the ESLstudent's own self-concepts in Reading and General-School areas.It may also help to explain why Math and Reading self-concepts arereversed for the ESL student compared to the normative sample.(i.e., "I am better at math compared to reading and other schoolsubjects and I am also better at math than other kids. But, I amhaving a really hard time with my reading compared to my math, andthe other kids are much better at reading than I am. Therefore, I137will rate my math self-concept as very high and my reading andother school subjects as very low.").Furthermore, the literature on language proficiency, age onarrival, and length of residence (e.g., Collier, 1989; Cummins,1984) indicates that the ESL student may require, on the average,three to seven years or more to reach grade norms when compared tonative speakers. Although beyond the scope of the present study,these variables may also have implications for the ESL student'sself-concept, particularly in specific academic areas such aslanguage arts, in which the student may feel less comfortable andproficient compared to native speakers. Again, loweredperceptions of language proficiency may help to lower thestudent's specific academic self-concept. Longitudinal studiesmight help to sort out if, indeed, the relationship betweenlanguage proficiency and self-concept changes over time such thatincreased proficiency results in higher language/verbal/readingself-concepts.2.^Teacher Perception of ESL Self-ConceptQuestion Two:What is the teacher perception of that ESL student's self-concept?In general, the mean scale score responses for teacherratings of student self-concept appeared higher and utilized awider standard deviation than those for the ESL students on theSDQ-1. Compared to the normative sample, however, teacher meanscale score ratings appeared lower and more in line with the ESL138student's self-perceptions than with those for the normativesample. These observations suggest that, while teachers mayperceive the ESL student's self-concepts to be somewhat higherthan they really are, they do, indeed, perceive subtle differencesin ESL self-concepts compared to those of the wider population,and that these differences are more consistent with the ESLstudent's own self-perceptions than with those of the normative orgeneral population.Also consistent with this explanation is the present findingthat t-test comparisons indicated only two significant differencesbetween teacher and student mean scale score responses. Teachersrated student self-concepts in Parent Relations to besignificantly lower compared to the ESL student's own ratings andthey perceived student self-concepts in Physical Appearance to besignificantly higher compared to the ESL student's self-ratings.That is, the teachers appeared to be more positive about thestudent's Physical Appearance while the students appeared to bemore positive about their Parent Relations than the teachers.Similar to the SDQ studies reviewed (Marsh, 1988a), correlationalcomparisons in the present study found no significantrelationships between student and teacher agreement in both ofthese areas. The present findings, therefore, appear consistentwith previously established patterns of self-otheragreement/differences. According to Marsh and others (e.g.,Marsh, 1988a; Marsh, Smith, & Barnes, 1983, 1984) self-conceptsevaluations by teachers may be difficult in both of these areas.For Parent Relations, teachers may have little or no directinformation upon which to base their judgements. In fact,139teachers in this study pointed out their concerns and discomfortto the present researcher when asked to answer the SDQ-1 questionsabout the student's Parent Relations as they felt this was an areawith which they were least familiar.Similarly, self-concepts in the area of Physical Appearanceappears to be a difficult area for both teachers and peers toevaluate. Marsh (1988a) has hypothesized that, although teacherstend to use more observable and objective criteria upon which tobase their evaluations, the formulation of preadolescent self-concept in Physical Appearance may be more idiosyncratic andcomplex than the Internal-External weighting process he proposed(Marsh, 1986a) to explain how preadolescents arrive at their ownparticular self-concepts in other areas. The present results tendto support the pattern reported in previous studies (e.g., Marsh,1988a) for both students and teachers.Comparison of score ranges between the ESL student andteacher perception ratings indicates that, for scores in PhysicalAppearance, Reading, and General-Self, teachers started rating ESLself-concepts higher than the ESL students rated themselves. Onall other scales (Physical Abilities, Peer Relations, ParentRelations, Mathematics, General-School, Total Nonacademic, andTotal Academic and Total-Self), teachers appeared to utilize awider range of scores in order to evaluate their perceptions ofESL student self-concepts. Comparison of score ranges andstandard deviations for student ratings and teacher perceptions,suggest that, in general, teachers tended to perceive moredifferences between ESL students. Again, this pattern of teacherratings may be explained by suppositions in the literature (Marsh,1401986d; 1988a) that inferred ratings by significant others may belinked to more observable and quantifiable data such as ability inphysical education and academic achievement scores on math tests.That is to say, the processes teachers use to arrive at theirperceptions may be different than the processes students use toarrive at their own self-perceptions.For teacher perceptions, self-concept correlations were verystrong between academic areas (Reading, Mathematics, and General-School). This finding makes sense and is consistent with theprevious research (Marsh, 1988a), particularly since theseacademic subjects are the areas where teachers have the mostopportunity to observe, compare, and evaluate their students.3.^The ESL Self-Concept and Teacher PerceptionRelationshipQuestion Three:What is the relationship between the ESL student's self-concept and teacher perceptions of that student's self-concept; and to what extent do they agree?In general, based on mean scale score comparisons and t-testsof significant differences between means, the presentinvestigation found that teacher perception of ESL student self-concept was similar, albeit somewhat higher, to the ESL student'sself-perceptions with two exceptions. The mean self-conceptrating for Physical Appearance was significantly lower for the ESLstudent, while the self-concept rating for Parent Relations wassignificantly lower for the teachers. As mentioned, these141findings concur with previous studies utilizing the SDQ-1 (e.g.,Marsh, 1986d, 1988a; Marsh, Smith, & Barnes, 1983, 1984) and withthe related SDQ literature as reviewed in Chapter II.Interestingly, a comparison of mean scale scores and standarddeviation scores in Mathematics self-concept indicates that theteacher perceptions were almost exactly the same as the ESLstudent's own self-perceptions. This finding may suggest that, atleast for self-concepts in Mathematics, students and teachers maybe utilizing a similar and more observable, quantifiable process.Also, compared to the normative sample, both the ESL student'smean scores in Mathematics and the teacher perception mean scoreswere higher, and the standard deviations lower, than the normativegroup. This finding suggests that the teachers also perceive theESL students to be more homogeneous, particularly in the area ofMathematics self-concept. However, unlike the near zerocorrelation found for the ESL students, for teachers, self-conceptscores in Reading and Mathematics correlate with each other. Thispattern of differences is consistent with previous SDQ research(Marsh, 1988a) and tends to support the suggestion (Marsh, 1986d,1988a) that students and teachers are using different processes inarriving at their individual self-concept perceptions and that theteachers, in particular, may be using more external, observable,and quantifiable criteria to formulate their perception ratings.Possible explanations for the significant discrepanciesbetween student and teacher self-concept ratings, especially interms of Reading and Mathematics, may be attributed to thedifferences in the processes used by students and teachers, and,perhaps, even by individuals within each group, to arrive at their142self-concept ratings. The frame of reference models proposed byMarsh and his colleagues (Marsh, 1986d, 1987a; Marsh & Parker,1984) attempt to explain the differences in processes that mightbe used by students and significant others such as teachers.According to the Internal-External frame of reference model(Marsh, 1986d), external raters, in particular, may tend to usemore external, observable, and quantifiable information upon whichto base their evaluations. Examples of external criteria might betest scores and comparisons with other students or with pastexperience. This is the same process used to explain the BFLPE(Marsh, 1987a; Marsh & Parker, 1984). Self-raters, on the otherhand, tend to make both internal and external comparisons andindividually weight their self-concept responses based on moreidiosyncratic criteria. According to this internal process,students compare their self-perceived ability in one area withtheir self-perceived ability in another area and use this internaland relativistic impression as a second basis for their self-concept ratings in each of the two areas. The I/E model alsopredicts the negative direct effect of achievement in one area(e.g., math) on self-concept in another area (e.g., reading). Forexample, a high Math self-concept will be more likely when mathskills are good (the external comparison) and when math skills arebetter than reading skills (the internal comparison). Thus, it isthe difference between math and reading skills which is predictiveof Math self-concept. Whereas high reading skills may actuallydetract from a high Math self-concept, low reading skills mayenhance high Math self-concept. Such an explanation might also be1 43the case for the present sample of ESL students, both in terms oftheir higher Math and their lower Reading self-concepts.Although there appears to be a large discrepancy betweenReading and Mathematics self-concepts for both the ESL student andteacher perceptions ratings, further t-tests comparisons betweenmeans for the ESL student may have helped to establish a level ofsignificant differences. Both the students and their teachersrated self-concepts in Reading and Mathematics areas similarly butdifferently compared to the normative group. For example, the ESLstudents rated self-concept in Math as the second highest andself-concept in Reading as the second lowest of all self-conceptratings. Similarly, the teachers rated ESL Math self-concept asthe highest and ESL Reading self-concept as the lowest of all theself-concept areas measured. The normative group, on the otherhand, rated reading self-concept higher than math self-concept.At first glance, it may appear as if the VE model may notadequately explain differences in the present sample compared toprevious self-other studies since both the ESL student and theteacher perceptions seem very similar for both Reading and Mathself-concepts. However, the lack of a significant correlationbetween self-other agreement for the present sample in the area ofReading self-concept, the distinct differences in math and readingacademic achievement scores, as well as the restricteddistribution of these scores, tends to suggest that, for thepresent ESL sample, the criteria for establishing externalcomparisons and the homogeneity for the present ESL sample may bemore definitive than for the general population. Furthermore, thepresent ESL sample represents primarily Oriental or Asian1 44'minority' group membership which may also increase the sample'shomogeneity.Pearson correlations explored the relationship between ESLstudent self-concept and teacher perception of student self-concept in order to examine the degree and direction of thisrelationship. Overall, correlations between ESL student andteacher perceptions on the SDQ-1 tended to be weaker and fewer innumber than correlations among self-concept scores for the ESLstudent and teacher perception ratings computed separately. Thisfinding might be expected as within measure ratings by the samerespondent tend to be higher than comparisons of between measureratings by different respondents. Also, the difference in overallcorrelational magnitude between the present study and previous SDQself-other investigations may be explained by the differentstatistical procedures applied. In much of the SDQ research,Marsh and his colleagues (Marsh, 1988a) used 'adjusted' scores,whereby both the student and teacher self-concept ratings were setwith similar means and standard deviations for comparisons. Wylie(1989) notes that the use of 'adjusted' scores tends to increasethe level of correlational significance between scores which mayhelp to explain the greater correlational magnitude reported inmany of the previous SDQ self-other studies compared to thepresent investigation. Overall, however, the patterns ofcorrelations between the ESL student self-concept ratings and theteacher perceptions of student self-concept, appear consistentwith previous research (Marsh, 1988a) and lend support to thevalidity and reliability of the SDQ-1 as well as themultidimensional theoretical structure upon which it is based.1 45In general, the lack of significant correlations betweenstudent and teacher, particularly in matching academic self-concept areas, appear, at first, to contradict Marsh's (1986d)prediction that student-teacher agreement will be strongest inacademic areas, the area most familiar to teachers. For the ESLstudents and teachers in this present sample, this seems to be thecase. However, the pattern of correlations, compared to theranking of highest and lowest means for the normative sample, aswell as the similar pattern of correlational scores averaged overeight self-other studies (Marsh, 1988a), tends to suggest that,overall, the SDQ does adequately assess ESL self-concept and thatoverall, student-teacher agreement is consistent, except for morelanguage-related or language-mediated areas such as Reading andGeneral-School or all school subjects.Perhaps for this sample of ESL students and their teachers,variables such as lack of proficiency with English, as well ascultural variables, may mediate the processes by which studentsand teachers arrive at their self-concept perceptions in even amore dramatic way than first proposed by Marsh and his colleagues(Marsh, 1986d; 1987a). At least for self-concepts in Reading andGeneral-School, it may be be suggested that the students andteachers in the present study used different processes toformulate their evaluations. The research on second languageacquisition, age on arrival, length of residence, and the time ittakes non-native speakers to reach grade norms (e.g., Collier,1989; Cummins, 1984) may help to explain why the ESL students inthis study rated their language-related self-concepts lower thanthe children in the normative sample and why, in keeping with the146I/E model, they may have rated their own abilities and,subsequently, their self-concepts in these areas relatively lowerthan one might anticipate from external criteria alone.Also, in contrast to the very strong correlations Marsh andhis colleagues (Marsh, 1988b) observed in self-other agreement foruniversity students, present study indicated a lack ofcorrelation in academic areas, particularly in Reading and inGeneral-School self-concept areas. The results of the presentstudy may be explained, in part, by the younger age of the currentsample of preadolescent students who may not, as yet, knowthemselves as well as the older university students knewthemselves.Furthermore, since the ESL students and their teachers agreedvery strongly in several important and specific nonacademic areas(i.e., Physical Abilities, Peer Relations and Total Nonacademicself-concepts), all of which may be based on external andobservable criteria, it appears that teachers may be better atevaluating the self-concepts of their ESL students in these areasthan one might at first expect. Although the ESL students andtheir teachers in the present study appear to agree in severalimportant nonacademic self-concept areas, a comparison of thecurrent findings with previous SDQ self-other research (Marsh,1988a) may also suggest that the ESL teachers in the present studymay not know their ESL students as well as other teachers knowtheir students. This explanation may be particularly indicated interms of academic self-concepts (i.e., Reading, General-School,and Total Academic self-concepts) where the present ESL student-teacher agreement was generally low and nonsignificant.^Apossible exception, however, may be found in the area ofMathematics self-concept where ESL student and teacher perceptionratings correlated significantly. Interestingly, Mathematics isan area which may also be termed more observable and quantifiablethan either Reading or General-School.Mary Ashworth (1988), in her national survey of ESLdifficulties, along with Lambert (1975, 1981) indicate that of allthe problems facing the ESL student perhaps the most significantmay be in nonacademic areas due to possible conflicts between homeand school. Similarly, Wong Fillmore (1983) suggests thatdifferences in home, culture, and socialization may effectstudents' academic achievement and motivation to succeed inschool. The results of the present study tend to suggest thatthis particular group of teachers do, in fact, appear tounderstand the ESL student's strengths and challenges in at leastin some nonacademic areas and that the direction of their sharedunderstanding is substantially related and in a similar directionto the ESL student's self-perceptions. Further research will beneeded to explore the depth of ESL teacher understanding and if,indeed, these nonacademic areas are problematic for the ESLstudent and whether interventions, and of what sort, may bebeneficial to enhance student self-concept in these particularareas. At this point, however, the SDQ-1 does appear to be abeneficial tool for assessing nonacademic as well as academicself-concepts for this group of children.The large number of negative correlations between ESL studentand teacher perceptions in various areas of self-concept suggestthat students and teachers are not using the same evaluation1471 48processes to formulate their self-concept perceptions and/or theyare not using them in the same way. This finding also supportsthe growing need for further research, such as local normativeevaluations and additional statistical procedures (e.g., factoranalyses) for both the student and the teacher data, as well asfurther correlational and, possibly, predictive evaluations.4.^ESL Self-Concept and Academic AchievementQuestion Four:What is the relationship between the ESL student's self-concept and the student's academic achievement in reading andin mathematics; and to what extent do they agree?Correlations between ESL student self-concept and academicachievement in Reading and in Mathematics were all nonsignificantin academic self-concept areas and mostly negative in nonacademicareas. A positive but near zero correlation was found betweenstudent self-concept in Reading and Reading achievement.Similarly, positive but near zero correlations were found betweenReading achievement scores and ESL self-concepts in other academicareas, including General-School, Mathematics, and total Academicself-concept.The correlations between ESL student self-concept inMathematics and Mathematics academic achievement wasnonsignificant and very weak. However, correlations between Mathachievement scores and the other academic areas (i.e., ReadingGeneral-School, and Total Academic self-concepts) were allnonsignificant. Although these findings were neither significant149nor causal, they may indicate the need for further research toexplore whether or not, for ESL students, math is perceived asunrelated to other academic school subjects and whether high mathscores may negatively affect self-concepts in other academic areaswhich is consistent with Marsh's (1986d) Internal-External frameof reference model.In general, the present findings are different from theliterature (Hansford & Hattie, 1982; Marsh, 1988a; Wylie, 1979,1989) which suggests that some of the facets of self-concept, thatis the ones associated with academics, should be at leastmoderately correlated with academic achievement. However, thecurrent findings may add support for the I/E frame of referencemodel (Marsh, 1986d) and the BFLPE (Marsh, 1987a; Marsh & Parker,1984) in that the ESL students in this study, unlike theirteachers, did not seem to be using the same processes to formulatetheir self-concepts and that, clearly, the ESL students' self-concepts were not formulated solely in relation to externalcriteria or marks.The present lack of significant correlations between academicachievement and matching academic self-concepts may be morecomplex as many of the students in this study never received anactual 'grade' per se on their report card. Furthermore, the lackof significant correlations between academic self-concepts andacademic achievement calls into question the implications ofanecdotal report cards and positively-worded comments. Althoughoften very encouraging, this type of reporting may not adequatelyconvey to the student where, in fact, he or she stands relative tohis or her other academic skills and in relation to other students1 50in the immediate environment or to a larger normative group.Reporting about students is a nebulous area in education and isbecoming more so for all students, teachers, and parents with theimpact of the Year 2000 and continuous progress. If externalcriteria are an important part of self-concept formation, for bothstudents and teachers, it may be more important to clarify gradesand scores than current trends suggest.Also, according to the Internal-External frame of referencemodel (Marsh, 1986d) and the BFLPE (Marsh, 1987a), it is uncertainas to with whom the ESL student may be comparing him or her self.It may be suggested that in schools where school-averageability/achievement is rated as average or better, such as themajority of schools from which the present sample of ESL studentswere drawn, the BFLPE may be very dramatic, particularly insegregated and pull-out classes. For example, the ESL student'smath abilities may seem very high in comparison to other students.On the other hand, the ESL student's reading abilities may appeareither 1) very poor in relation to the other students inmainstream classes, or 2) relatively better than other ESLstudents in segregated classes, or 3) relatively better than otherESL students in school populations where the percentage of ESL isalso very high. Further research is needed to explore theinterplay of these variables (e.g., individual studentability/achievement, type of class setting, and school populationdemographics) and their relation to the formation of ESL studentself-concepts.The nonsignificant correlations between academic self-concepts and academic achievement scores may also identify a151weakness in the present study which attempted to quantify studentmarks and or teacher ratings of student achievement. In addition,it suggests the need for further research which might utilizestandardized achievement tests for comparisons.5. Teacher Perception and Academic AchievementQuestion Five:What is the relationship between teacher perception of theESL student's self-concept and the student's academicachievement in reading and in mathematics; and to what extentdo they agree?Positive, substantial, and significant correlations werefound between teacher perception of ESL self-concept and academicachievement scores in both Reading and Mathematics. Inparticular, strong correlations were found between academicachievement ratings in Reading and in Mathematics and teacherperceptions of ESL student self-concept in matching academic areasand Total Academic self-concept. Substantial correlations werealso found between Reading and Math achievement scores and teacherratings of student self-concepts in General-School, Total-Self,and General-Self. Significant correlations were observed betweenMath achievement scores and teacher ratings of ESL student Readingself-concept and, to a lesser extent, between Reading achievementscores and teacher ratings of ESL student Mathematics self-concept. Significant but low correlations were found between Mathachievement scores and teacher ratings of Physical Abilities andbetween Reading achievement scores and teacher perceptions in Peer152Relations and Parent Relations. Similarly, both Reading and Mathachievement scores correlated significantly but moderately lowwith teacher ratings of the ESL student's Total Nonacademic self-concept.Overall, the significant findings between teacher perceptionratings and academic achievement appear in contrast to thegenerally low and nonsignificant correlations observed between theESL student self-ratings and academic achievement scores. Thepresent results also tend to suggest that the ESL teachers in thisstudy, like teachers in the previous investigations (Marsh, 1986d,1988a), may rely more on external and observable criteria, such asgrades and academic ratings, in order to base their evaluations ofstudent self-perceived ability and subsequent self-concept (anexternal comparison). The students themselves, on the other hand,may rely on different criteria, perhaps internal or external, or aweighted combination of both.Interestingly, the substantial correlations between theacademic achievement scores and teacher perception ratings inGeneral-Self and Total-Self areas may suggest that the processteachers use to evaluate more global aspects of student's self-concept may also be linked, to a large extent, on external,observable, and academic data. If not just simply a 'haloeffect', it appears that the teachers in the present study, unlikethe ESL students, may link positive academic achievement topositive self-concepts in other areas. The correlations betweenacademic achievement and teacher perception of ESL student self-concepts in nonacademic areas, such as Peer Relations and ParentRelations, may also suggest that the ESL teachers, unlike the1 53students themselves, relate student performance, academicabilities, and skills with the student's personal relationships tosignificant others, such as parents and friends. The extent towhich these significant others may actually be influential inhelping to formulate ESL student self-concept and academic skillsremains to be explored.Overall, the high correlations between academic achievementscores and teacher perception of academic self-concept ratings,are not surprising as it was the teachers themselves who issuedthe grades, rated the ESL student's abilities and skills, andprobably made comparisons to other students within their frame ofreference and experience. In addition, the majority of thesuburban ESL teachers chose to collaborate with the homeroomteachers who taught the ESL students in subject areas other thanlanguage arts (e.g., Math). The collaboration between teachersmay have confirmed, supported, and strengthened the teacherratings of academic achievement in subject areas in which theywere less certain. Furthermore, this possible reinforcement andconfirmation of a student's academic skills may also havestrengthen the correlational relationship between the teacher'sratings of academic achievement and perceived self-concept forthat ESL student.6. Other Significant Findings6.a. Sex DifferencesT-test comparisons revealed sex differences for both the ESLstudent self-ratings and the teacher perceptions of ESL studentself-concepts in terms of Physical Abilities and Reading. For1 54both the student and teacher ratings, self-concepts in PhysicalAbilities were found to be higher for boys than for girls, whereasself-concepts in Reading were found to be higher for girls thanfor boys. These findings are consistent with the sex stereotypesobserved in the literature (e.g., Meece et al., 1982) and alsowith the findings reviewed in the previous SDQ research (e.g.,Marsh, 1985b; Marsh, Relich, & Smith, 1983).^The present resultslend support for the differentiation of self-concept norms andseparate norm tables for males and females as proposed in the SDQ-1 manual (Marsh, 1988a).6.b.^District DifferencesT-test comparisons revealed district differences between theurban and suburban schools in terms of Mathematics self-concept asperceived by both the ESL student and inferred teacher ratings.Overall, the mean self-concept rating for Mathematics was higherfor the student and teacher ratings in the suburban districtschools than in the urban district schools.The present findings may be explained by differences inherentin program and service delivery models (e.g., full-time ESLclasses or integrated settings with pull-out programs for ESL ormathematics), teaching methods and styles, as well as possibledifferences in the background characteristics of the students andteachers and the particular philosophical, academic, and socialcontext of each district. Also, unlike the teachers in the urbandistrict, the suburban teachers collaborated on their responses tosome of the items on the SDQ-1, including questions concerningMathematics self-concept.155It is also uncertain whether or not the suburban studentswere more cognizant of their math achievement than the urbanstudents which might have influenced the internal and externalcomparison processes for both teachers and students. Furthermore,the numbers of students and teachers were not equal in bothdistricts which may also have influenced the present findings.Further research is needed in order to determine what factors wereoperating in each district and which contributed to the presentdifferences in results.However, since both the student and teacher ratings appear tobe in the same direction, it would appear that both of thesegroups tended to use a similar process to arrive at their self-concept evaluations. Furthermore, if inferred raters tend to usemore external comparisons (Marsh, 1986d, 1988a), one might assumethat, at least for Math self-concept in the suburban district,both the students and teachers may have, for some reason, usedbackground characteristics in a more objective, observable, andexternal process than the students and teachers in the urbandistrict.The data collected from the suburban schools appears toreflect a higher proportion of teacher respondents to students (16teachers to 23 students) compared to the urban schools (5 teachersto 34 students). This difference may be due to districtdifferences in policy and service delivery which may, in addition,influence how well teachers feel they know their ESL students indiverse or multidimensional areas. However, only the actual ESLteachers were used in the statistical computations which may notadequately reflect participation from the other teacher1 56collaborators who, although not officially counted as ESL teachersper se, may have helped to respond to some of the items. Thesedifferences in procedures, in addition to cultural differencesbased on ethnicity, length of time in Canada, and age on arrival,as well as differences in distinct program policies, may interactalone or in combination, to influence the present findings and,perhaps, obscure the results and interpretations upon which theyare based.6.c.^Birth Place DifferencesT-test comparisons found differences in teacher perceptionratings for self-concepts in Mathematics, Total Nonacademic, andTotal-Self areas between Canadian-born ESL students and all otherESL immigrant students involved in the present study. In allthree areas, the mean teacher self-concept ratings weresignificantly higher for the immigrant students than for theCanadian-born ESL students. These findings may suggest thatteachers perceive differences between immigrant students andCanadian-born ESL students.Possible reasons for these differences may be due to theactual differences in skill levels (e.g.,in mathematics) betweenthe two groups of students or a tendency for teachers toovercompensate and, therefore, rate their perceptions higher innonacademic areas with which they may feel less familiar,particularly in terms of immigrants' social and culturaldifferences. The Lambert (1975, 1981) research on the developmentof cultural identity may suggest that, for teacher perceptions inself-concept areas such as Total Nonacademic and Total-Self,1 57Canadian-born ESL students may appear, as a group, more sociallyand culturally like other Canadian-born students, whereasimmigrant ESL students may be perceived of as more homogeneous anddifferent from the larger reference group.Also, the large discrepancy between the numbers of Canadian-born students (n=6) and the immigrants, or all others (n=51),represented in the current sample is noteworthy and may havecontributed to the significant differences observed between thetwo groups. Nevertheless, Canadian-born ESL students may havesome idiosyncrasies which need to be explored separately. Furtherresearch is needed to examine if, indeed, significant differencesexist between the ESL students who immigrate to Canada from othercountries compared to the ESL students who are born here and how,and to what extent, teachers relate to these differences. Also,comparisons of these two groups to other Canadian-born English asa first language students may also need to be examined.6.d.^Ethnic DifferencesT-test comparisons between mean self-concept ratings revealedethnic group differences between Hong Kong Chinese students andall other students involved in the present study.^For studentratings, a significant difference was found in Peer Relations anda similar directional 'trend' was noted in Total Nonacademic self-concept areas, both of which were rated higher by the other ESLstudents than by the Chinese ESL students from Hong Kong. Forteacher perception ratings, significant differences were found inmean scores for Mathematics and Total Academic self-concepts, both1 58of which were rated higher for the Chinese ESL students from HongKong than for the other ESL students in this study.The pattern of findings for the ESL student perceptionssuggests that the level of socialization, acculturation, andassimilation may be different for Hong Kong Chinese students thanfor other ethnic students. Subsequently, the Hong Kong Chinesestudents may feel less comfortable and confident, particularly interms of peer relationships when compared to other ethnic groups.The pattern of findings for teacher perceptions suggests thatChinese Hong Kong students may be viewed as more able andconfident in academic areas, in general, and in terms ofMathematics, in particular. This finding is consistent with theliterature (Lynn, Pagliari, & Chan, 1988) advocating the visual-spatial strengths observed for Orientals, as well as with surveys(Cummins, 1984) relating higher patterns of academic achievementfor Asian-Americans and Chinese immigrants when compared to otherminority groups. It may also be questionable whether or notteachers may use some racial stereotyping when formulating theirself-concept evaluations. In addition, teachers may have maderelative comparisons when formulating self-concept perceptionsbased on differences between a student's math and readingachievement scores. Also, the majority of the Hong Kong studentsin this study came from middle to upper-middle class schools whichsuggests affordability for tutors. Again, the difference innumbers of subjects and the relatively small sample size isproblematic and should be considered carefully.There is need for further research in order to explore ethnicdifferences, not only in terms of abilities and skills, but alsoin terms of socialization and the formation of individual andcultural identity and how these factors relate to and, perhaps,predict specific self-concepts perceptions for both students andteachers.B. LIMITATIONS OF THE STUDYMany of the limitations of this study were outlined in theprevious discussion and may be summarized as follows:1. The present study was limited by a relatively small samplesize and the variability among its subjects, particularly in termsof ethnic background, level of language proficiency, length ofresidence in Canada, and age on arrival in Canada. Furthermore,the variation in type of program service delivery offered by twodifferent school districts and possible variations in policies,procedures, and instruction between the two also added to thealready large amount of variables inherent in measuring bothstudent and teacher ratings on 11 facets of self-concept.Therefore, the generalizability of results to other ESL studentsand their teachers may be difficult, and, although beyond thescope of the present study, these variables leave room for furtherexploration on many levels.2. As this study was exploratory in nature and used a relativelynew instrument which was normed on Australian students, Canadianand/or local norms were unavailable, making comparisons within theCanadian socio-cultural context speculative. Furthermore, thecurrent study did not include a control group matched on variables1 59160such as age and sex with the ESL student sample. Unfortunately,this was not possible for the present study due to thedifficulties recruiting students across two districts andconsidering differences in policies, procedures, and classroomsettings.3. This study was correlational, and although a correlationalstudy can effectively demonstrate relationships among specifiedvariables, causation cannot be determined because the direction ofthe relationships is unknown. The interpretation of thecorrelation can also be affected by mediating variables.4. Measurement issues also limit the interpretation of theresults. First, the academic achievement scores were reversed andhad to be adjusted in order to parallel the data collection on theSDQ-1. Second, the data was not factored and then compared to thenormative sample which would have made comparisons to thenormative sample and previous SDQ research studies more meaningfuland, perhaps, more consistent. Similarly, comparisons of factorscale scores and 'adjusted' scores between the ESL student andteacher perception ratings may also have revealed additionalinformation about the student-teacher self-concept relationship.Also, additional t-test comparisons, between the mean scale scoresfor the normative sample and the present sample as well as betweenthe reading and mathematics self-concept scale scores for thepresent ESL sample, might have revealed further significantcomparisons.161C.^RECOMMENDATIONS FOR FURTHER RESEARCHDue to the nature of exploratory research, this study mayhave produced more questions than it answered. Some of therecommendations for further research that arise are summarized asfollows:1. Canadian and local normative studies might shed more light onthe generalizability of the Self-Description Questionnaire-1 tothe general Canadian population and to other ethnic minoritygroups that might be included in the sample. Furthermore, theymight point out subtle differences across cultures that are not asyet apparent from the results of the present findings.2. A similar study might be replicated for use with 'regular'students and their teachers in a variety of program settings inorder to assess teacher perceptions of 'regular' students self-concepts and then compare those results with other groups ofstudents as a basis for normative comparisons.3. A similar study might be replicated in which othersignificant others, such as parents and peers, evaluate theirperceptions of ESL self-concept. These ratings might also becompared to the teachers' self-concept responses to investigatesimilarities and differences in student-other agreement which, inturn, might also lead to further understanding of the ESL student.4. Significant others might be asked to rate the ESL student,not only as they think the student thinks of him or herself (i.e.,an inferred rating), but also how the respondents believe thestudent should think about him or her self (i.e., an expected162rating). Similarly, students might be asked to rate themselves asthey feel the are and as they think others perceive them to be.Such investigations might uncover similarities and discrepanciesin actual and perceived self-concept ratings which might furtherenhance our understanding of how self-concepts are formed,perceived, and projected for both for self-raters and significantothers.5. Students and teachers might be asked to share their responseswith each other to learn more about each others' formulations,perceptions, and projections on self-concept issues. In turn,such a sharing might lead to the surfacing of unaddressed concernswhich might act as a focus for re-evaluation, possible counsellingissues, and/or programming needs particular to the ESL student.6. The self-concepts of high and low achievers might becompared, along with teacher perceptions for these groups, inorder to further investigate the interplay of achievement andself-concept formation.7.^Given the significant mean differences between male andfemale self-concept responses found in the present study (i.e.,higher self-concepts in Reading for females and higher self-concepts in Physical Ability for males), a future study might beconducted to investigate sex differences and ESL self-conceptformation for both the ESL students and the teacher perceptionratings. In general, the present results appear consistent withthe sex stereotypes as reviewed in the literature (Meece et al.,1982). Unfortunately, the relatively small sample size for thepresent study made further exploration of ESL sex differencesdifficult at this time.1 638. Student and teacher perceptions might be compared in high andlow SES/ability schools in order to shed light on the relationshipbetween ESL self-concept and school SES/ability.9. Attributional factors might be compared with self-concept,academic achievement, and perceptions by others. Comparisonsbetween and among these factors might enhance our understanding ofhow these factors relate to each other, and, to what extent theyeach relate and/or influence ESL student self-concept.10. A study might look at the differences and effects of ratingsarrived at through collaboration with others compared to singularratings of student self-concept when comparing ratings bysignificant others. Such a study might help to address whichmethod or style of response is a more accurate indicator of actualESL student self-concept.11. Further statistical analyses of the present data might addinformation on the differences between raw, factored, and'adjusted' scores for both the ESL students and their teachers andtheir relationship to both the present and normative sample. Sucha comparison might help to evaluate which procedure yields moresignificant and useful results.12. Predictive statistics such as multiple regression analysesmight add information as to some of the causal determinantsbetween ESL self-concept, teacher perception, and academicachievement.13. Also, some of the additional variables present in this studymight be used in subsequent and future studies with more'controlled' groups in order to help sort out some of the factorswhich might interplay with self-concept, inferred ratings, and1 64academic achievement. The nature of the 'control' group, matchedon variables particular to the sample, would then depend on theresearch questions posed. Some of the factors which have beenidentified in the literature on ESL student assessment as criticalto language proficiency (Collier, 1989; Cummins, 1984) mightinclude: age on arrival, length of residence in the host country,level of prior Ll literacy development, and the influence ofbilingual programs. Unfortunately, many of these factors weredifficult to control and investigate in the present study, due,for example, to difficulties inherent in recruiting subjects, theinvolvement of two districts with different policies, procedures,and classroom settings, and the relatively small sample size.14. Longitudinal studies are needed to sort out if, indeed, therelationship between language proficiency and self-concept changesover time, such that increased English proficiency results inhigher language, verbal, and reading self-concepts as the ESLstudents reach the grade norms of their native-speaking peers.For example, comparisons with mathematics self-concepts and otheracademic areas, as well as with teacher perceptions, might examinepossible changes with time, exposure, skill acquisition, andacculturation.15. As questions as to the most effective program and model ofservice delivery are paramount to educational practise, a studymight be conducted in order to investigate which setting,philosophy, and practise provides the ESL student with thegreatest support. A self-concept study might be of value as ameasure of program success.16516. There is also a need for research into what actually happenswhen students move from one academic setting to another where theaverage ability level is quite different. There is also a need toexplore individual characteristics that may determine how studentsreact to this potentially stressful transition.17. The use of standardized and objective tests and theirrelationship to ESL self-concept and teacher perception might becompared to the results obtained by grades. Such comparisonsmight help to determine the similarities and differences betweenvarious academic measures and their relationship to ESL self-concept.18. There appears to be a need for research in terms of theformation of physical appearance self-concept and how it isreported as teacher-student agreement in this area is one of thelowest for both the present study and the previous SDQ researchstudies (Marsh, 1988a).19. Intervention strategies might be applied in a test-retestsituation to evaluate their effectiveness in changing a student'sself-concept.20. A study might compare ESL self-concepts and teacherperceptions in situations where different methods of reportingacademic achievement are used (e.g., grades, tests and/oranecdotal reports) to investigate the self-concept and academicachievement relationship given different sources of academicinformation.21. There appears to be a need for both quantitative andqualitative research regarding ESL self-concept and perceptionsmade by others and how these relate to all areas (i.e., academicand nonacademic) in order to clarify further the interplay ofintrinsic and extrinsic factors in the formation and change ofself-concept.1 661 67REFERENCESAkoodie, M. (1973). Preschool children's attitudes towards colourand physical differences. Unpublished bachelor's thesis,Wilfrid Laurier University, Waterloo, Ontario.Akoodie, M. (1980). Immigrant students: A comparative assessment of ethnic identity, self-concept and locus of control.Unpublished doctoral dissertation, University of Toronto.Akoodie, M. (1984). Identity and self-concept in immigrantchildren. In Samuda, R. J., Beery, J., & Laferriere, M.(Eds.), Multiculturalism in Canada: Social and Educational Perspectives (pp. 253-265). Toronto, Ontario: Allyn andBacon.Alwin, D. F., & Jackson, D. J. (1981). Applications ofsimultaneous factor analysis to issues of factorialinvariance. In D. J. Jackson & E. F. Borgotta (Eds.), Factoranalysis and measurement in sociological research: Amultidimensional perspective (pp. 249-280). Beverly Hills,CA: Sage.Appel, R. (1979). The acquisition of Dutch by Turkish andMoroccan children in two different school models. Unpublishedresearch report, Institute for Developmental Psychology,Utrecht.Asher, J., & Garcia, G. (1969). The optimal age to learn aforeign language. Modern Language Journal, 2a, 334-341.Ashworth, M. (1988). Blessed with bilingual brains: Education ofimmigrant children with English as a second language.Vancouver: Pacific Education Press, University of BritishColumbia.Ashworth, M., Cummins, J., & Handscombe, J. (1989). Report of the External Review Team on the Vancouver School Board's ESL Programs. Vancouver: Vancouver School Board.Bachman, J. G. (1970). Youth in transition (Vol. 2): The impact of family background and intelligence on tenth-grade boys. AnnArbor: Institute for Social Research, The University ofMichigan.Bachman, J. G., & O'Malley, P. M. (1977). Self-esteem in youngmen: A longitudinal analysis of the impact of educational andoccupational attainment. Journal of Personality and Social Psychology, 35, 365-380.Bachman, J. G., & O'Malley, P. M. (1984). Black-white differencesin self-esteem: Are they due to differences in response style?American Journal of Sociology, 11, 624-639.168Bachman, J. G., & O'Malley, P. M. (1986). Self-concepts, self-esteem, and educational experiences: The frog pond revisited(again). Journal of Personality and Social Psychology, 50,33-46.Bachman, J. G., & O'Malley, P. M., & Johnston, J. (1978). Youth in transition (Vol. 4): Adolescence to adulthood - change andstability in the lives of young men. Ann Arbor: Institute forSocial Research, The University of Michigan.Barkley, K. L. (1985). Inter- and intra-group levels of esteemamong three ethnic groups. Journal of Negro Education, 54(1),56-70.Barona, A., & Barona, M. (1987). A model for the assessment oflimited English proficient students referred for specialeducation services. In S. H. Fradd & W. J. Tikunoff (Eds.),Bilingual education and bilingual special education (pp. 183-209). Boston: College-Hill Press.Bayton, J. A., McAlister, L. B., & Hamer, J. (1956). Race-classstereotypes. Journal of Negro Education, 2, 75-78.Benson, J., & Hocevar, D. (1985). The impact of item phrasing onthe validity of attitude scales for elementary schoolchildren. Journal of Educational Measurement, 22, 231-240.Bloom, B. S. (1976). Human characteristics and school learning.New York: McGraw-Hill.Bloom, B. S. (1977). Affective outcomes of school learning. PhiDelta Kappan, November, 193-198.Boersma, F. J., & Chapman, J. W. (1979). Student's perception of ability scale manual Edmonton, Canada: University of Alberta.British Columbia. Royal Commission on Education (1988). A legacy for learners. Victoria.British Columbia. Royal Commission on Education (1988).Commissioned papers: Volume 2. The learners of BritishColumbia. Victoria.British Columbia Ministry of Education (1986). Comparison of 1986/87 and 1985/86 enrollment, funding, and cost per student by district and program 3.30 English second language.Victoria: British Columbia Ministry of Education.British Columbia Ministry of Education (1989). Policy directions: A response to the Sullivan Royal Commission on education by the government of British Columbia. Vancouver: B.C. Ministryof Education.1 69Brookover, W. B., & Erikson, E. L. (1975). Sociology of education. Homewood, IL: Dorsey Press.Brookover, W. B., Erikson, E. L., & Joiner, L. M. (1967). Self-concept of ability and school achievement. III: Relationship of self-concept in high school (U.S. Office of Education,Cooperative Research Project No. 2831). East Lansing:Michigan State University, Office of Research andPublications.Brookover, W. B., LePere, J. M., Hamachek, D. E., Thomas, S., &Erikson, E. L. (1965). Self-concept of ability and school achieyement. II: Improving academic achievement throughstudent's self-concept enhancement (U.S. Office of EducationCooperative Research Project No. 1636). East Lansing:Michigan State University, Office of Research andPublications.Brookover, W. B., & Passalacqua, J. (1981). Comparison ofaggregate self-concepts for populations with differentreference groups. In M. Lynch, A. Norem-Hebeisen, & W. Gergen(Eds.), Self-concept: Advances in theory and research.Cambridge, MA: Ballinger.Brosnan, F.L. (1983). Overrepresentation of low socioeconomicminority students in special education programs in California.Learning Disabilities Ouarterly,, IL, 517-525.Burns, R. B. (1979). The self-concept: Theory, measurement, development and behaviour. London: Longman.Byrne, G. M. (1983). Investigating measures of self-concept.Measurement and Evaluation in Guidance,  152, 115-126.Byrne, B. M. (1984). The general/academic self-conceptnomological network: A review of construct validationresearch. Review of Educational Research, „5,A, 427-456.Byrne, B. M. (1988). The Self-Description Questionnaire-III:Testing for equivalent factorial validity across ability.Educational and Psychological Measurement, 48,  397-406.Byrne, B. M., & Shavelson, R. J. (1986). On the structure ofadolescent self-concept. Journal of Educational Psvcholoav,la, 474-481.Byrne, G. M., & Shavelson, R. J. (1987). Adolescent self-concept:Testing the assumption of equivalent structure across gender.American Educational Research Journal,  2AL, 365-385.Campbell, D. T., & Fiske, D. W. (1959). Convergent anddiscriminant validation by the multitrait-multimethod matrix.Psychological Bulletin, an, 81-105.170Canale, M., & Swain, M. (1979). Theoretical bases ofcommunicative approaches to second language teaching andtesting. Toronto: Ontario Ministry of Education.Caslyn, R. J., & Kenny, D. A. (1977). Self-concept of ability andperceived evaluation of others: Cause or effect of academicachievement. Journal of Educational Psychology, L2, 136-145.Chamot, A. U., & O'Malley, J. M. (1987). The cognitive academiclanguage learning approach: A bridge to the mainstream. TESOLOuarterly, 21, 227-249.Chapman, J. W., & Boersma, F. J. (1979). Affective variables inlearning disabilities. Journal of Educational Psychology, 11,250-258.Chapman, J. W., Cullen, J. L., Boersma, F. J., & Maguire, T. 0.(1981). Affective variables and school achievement: A studyof possible causal influences. Canadian Journal of Behavioural Science, 13, 181-192.Charlton, J. M. (1987). Self-concept and locus of control: Astudy of intermediate and secondary students with learningdisabilities Unpublished master's thesis, University ofBritish Columbia, Vancouver.Coleman, J. S., Campbell, E. Q., Hobson, C. J., McPartland, J.,Mood, A. M., Weinfeld, F. D., & York, R. L. (1966). Equalityof educational opportunity. Washington, DC: U.S. GovernmentPrinting Office.Collier, V. P. (1987). Age and rate of acquisition of secondlanguage for academic purposes. TESOL Quarterly, 21, 617-641.Collier, V. P. (1989). How Long? A synthesis of research onacademic achievement in a second language. TESOL Quarterly,23, 509-531.Collier, V. P., & Thomas, W. P. (1988). Acquisition of cognitive-academic second language proficiency: A six-year study.Paper presented at the annual meeting of the AmericanEducational Research Association: New Orleans.Combs, A., Soper, D., & Courson, C. (1963). The measurement ofself-concept and self-report. Educational and Psychological Measurement, 23, 493-500.Cooley, C. H. (1902). Human nature and the social order. NewYork: Scribners.Coopersmith, S. A. (1967). The antecedents of self-esteem. SanFrancisco: W. H. Freeman.171Covington, M. V. (1984). The motive of self-worth. In R. E. Ames& C. Ames (Eds.), Research on motivation in education (Vol. 1,pp. 78-114). Orlando, FL: Academic Press.Covington, M. V., & Beery, R. G. (1976). Self-worth and school learning. New York: Holt, Rinehart & Winston.Covington, M. V., & Omelich, C. L. (1979). Are causalattributions causal? A path analysis of the cognitive modelof achievement motivation. Journal of Personality and Social Psychology, 37, 1487-1504.Crandall, R. (1973). The measurement of self-esteem and relatedconstructs. In J. Robinson & P. Shaver (Eds.), Measures of social psychological attitudes (pp. 45-158). Ann Arbor:Institute for Social Research, The University of Michigan.Cummins, J. (1976). The influence of bilingualism on cognitivegrowth: A synthesis of research findings and explanatoryhypotheses. Working Papers on Bilingualism, 2, 1-43.Cummins, J. (1978). Educational implications of mother tonguemaintenance in minority-language groups. The Canadian Modern Language Review, 34, 395-416.Cummins, J. (1979). Cognitive/academic language proficiency,linguistic interdependence, the optimal age question, and someother matters. Working Papers on Bilingualism, 2, 1-43.Cummins, J. (1980). The entry and exit fallacy in bilingualeducation. NABE Journal, A, 25-59.Cummins, J. (1981a). Age on arrival and immigrant second languagelearning in Canada: A reassessment. Applied Linguistics, 11,132-149.Cummins, J. (1981b). The role of primary language development inpromoting educational success for language minority students.In Schooling and language minority students: A theoretical framework (pp. 3-49). Los Angeles: California StateUniversity, National Evaluation, Dissemination and AssessmentCenter.Cummins, J. (1981c). Bilingualism and minority-language children.Toronto, Ontario: Language and Literacy Series. OntarioInstitute for Studies in Education.Cummins, J. (1984). Bilingualism and special education: Issues in assessment and pedagogy. England: Multilingual Matters.Cummins, J., & Swain, M. (1986). Bilingualism in education. NewYork: Longman.1 72Cummins, J., Swain, M., Nakajima, K., Handscombe, J., Green, D., &Tran, C. (1984). Linguistic interdependence among Japaneseand Vietnamese immigrant students. In C. Rivera (Ed.),Communicative competence approaches to language proficiency assessment: Research and application (pp. 60-81). Clevedon,England: Multilingual Matters.Daly, J. A., Bell, R. A., & Korinek, J. (1987).Interrelationships among attitudes toward academic subjects.Contemporary Educational Psychology, 12,  147-155.Davis, J. A. (1966). The campus as a frog pond: An application oftheory of relative deprivation to career decisions for collegemen. American Journal of SociolomL, 72, 17-31.De Avila, E. A., & Duncan, S. E. (1980). The language minoritychild: A psychological, linguistic, and social analysis. InJ. E. Altis (Ed.), Georgetown University Round Table onLanguages and Linguistics 1980: Current issues in bilingual education (pp. 104-137). Washington, DC: GeorgetownUniversity Villiers, J. G. & de Villiers, P. A. (1978). Language acquisition. Cambridge, MA: Harvard University Press.Diaz, R. (1983). Thought and two languages: The impact ofbilingualism on cognitive development. Review of Research inEducation, 10, 23-54.Dulay, H., & Burt, M. (1980). The relative proficiency of limitedEnglish proficient students. In J. E. Alatis (Ed.),Georgetown University round table on languages and linguistics 1980: Current issues in bilingual education (pp. 181-200).Washington D.C.: Georgetown University Press.Duncan, S. E., & De Avila, E. A. (1979). Bilingualism andcognition: Some recent findings. NABE Journal, 1(1), 15-20.Dusek, J. B., & Flaherty, J. F. (1981). The development of self-concept during adolescent years. Monographs of the Society for Research in Child Development, 46(4, Serial No. 191).Ekstrand, L. (1976). Age and length of residence as variablesrelated to the adjustment of migrant children, with specialreference to second language learning. In G. Nickel (Ed.),Proceedings of the Fourth International Congress of AppliedLinguistics (Vol. 3, pp. 179-197). Stuttgart:Hochschulverlag.Ekstrand, L. H. (1977). Social and individual frame factors inL2 learning: Comparative aspects. In T. Skutnabb-Kangas(Ed.), papers from the first Nordic conference onbilingualism. Helsingfors: Universitetet.1 73Ekstrand, L. H. (1978). Bilingual and bicultural adaptation.Unpublished doctoral dissertation, University of Stockholm,Sweden.Ervin-Tripp, S. M. (1974). Is second language learning like thefirst? TESOL Ouarterly, 11, 111-127.Eshel, Y., & Klein, Z. (1981). Development of academic self-concept of lower-class and middle-class primary schoolchildren. Journal of Educational Psychology, 73, 287-293.Esquivel, G. B. (1985). Best practices in the assessment oflimited English proficient and bilingual children. In A.Thomas & J. Grimes (Eds.), Best practices in school vsvcholoctIL(pp. 207-213). Ohio: National Association of SchoolPsychologists.Fennema, E. (1974). Mathematics learning and the sexes: A review.Journal for Research in Mathematics Education,  a, 126-139.Fernandes, M. W., Michaels, W. B., & Smith, R. (1978). Thefactorial validity of three forms of the dimensions of self-concept measure. Educational and Psychological Measurement,38, 537-545.Fitts, W. H. (1965). Tennessee self-concept scale manual^LosAngeles: Western Psychological Services.Fleming, J. S., & Courtney, B. E. (1984). The dimensionality ofself-esteem: II: Hierarchical facet model for revisedmeasurement scales. Journal of Personality and Social Psychology, j., 404-421.Fleming, J. S., & Watts, W. A. (1980). The dimensionality ofself-esteem: Results for a college sample. Journal of Personality and Social Psychology, 39, 921-929.Fram, I., & Crawford, P. (1972). An examination of the relationship between sex, birth date, and English-as-a-secondlanguage and teacher's perceptions of academic success (Research Report No. ON00494). North York, Ontario: NorthYork Board of Education.Gale, K., McClay, D., Christie, M., & Harris, S. (1981). Academicachievement in the Milingimbi bilingual education program.TESOL Ouarterly, la, 297-314.Genesee, F. (1978). Is there an optimal age for starting secondlanguage instruction? McGill Journal of Education, 13, 145-154.Genesee, F. (1987). Learning through two languages: Studies of immersion and bilingual education. New York: Newbury House.1 74Gersten, R., & Woodward, J. (1985). A case for structuredimmersion. educational Leadership,^(1), 75-79.Goldman, R., & Mercer, B. (1976). Self-esteem and self-differentiation: A comparison between black and white childrenin follow-through and non- follow-through classes.Educational Research Ouarterly, 1(3), 43-49.Good, T. L., & Brophy, J. E. (1971). Analyzing classroominteraction: A more powerful alternative. Educational Technologv, 11, 36-40.Gottlieb, M. (1985, April). Communicative, language, and academicproficiencies of limited-English-proficient and native English-speaking elementary school students^Paper presentedat the annual meeting of the American Educational ResearchAssociation, Chicago. (ERIC Document Reproduction Service No.ED 260 593).Hakuta, K. (1986). Mirror of language: The debate onbilingualism. New York: Basic Books.Hakuta, K., & Snow, C. (1988). The role of research in policydecisions about bilingual education. NABE News, 9(3), 1, 18-21.Hansford, B. C., & Hattie, J. A. (1982). The relationship betweenself and achievement/performance measures. Review of Educational Research, 52, 123-142.Harter, S. (1982). The Perceived Competence Scale for Children.Child Development, 53, 87-97.Harter, S. (1983). Developmental perspectives on the self-system.In P. H. Mussen (ed.), Handbook of child psychology, (Vol. 4,4th ed., pp. 275-385). New York: Wiley.Harter, S. (1985). Processes underlying the construction,maintenance, and enhancement of the self-concept in children.In J. Suls & A. Greenwald (Eds.), Psychological perspectives on the self (Vol. 3, pp. 137-181). Hillsdale, New Jersey:Lawrence Erlbaum.Heath, S. B. (1986). Sociocultural contexts of languagedevelopment. In S. B. Heath (Ed.), Beyond language: Social and cultural factors in schooling language minority students (pp. 143-186). Los Angeles: California State University,National Evaluation, Dissemination and Assessment Center.Hernandez-Chavez, E., Burt, M., & Dulay, H. (1978). Languagedominance and proficiency testing: Some generalconsiderations. NABE Journal, 2, 41-54.175Hodgkins, B. J., & Stakenas, R. G. (1969). A study of self-concepts of Negro and white youth in segregated environments.Journal of Negro Education, 334 370-377.Hunt, D. E., & Hardt, R. H. (1969). The effects of upward boundprograms on the attitudes, motivation and academic achievementof Negro students. Journal of Social Issues, ., 122-124.Iheanacho, S. 0. (1988). Minority self-concept: A researchreview. Journal of Instructional Psychology, 15(1), 3-11.Jerusalem, M. (1984). Reference group, learning environment andself-evaluations: A dynamic multi-level analysis with latentvariables. In R. Schwarzer (Ed.), The self in anxiety, stress and depression, (pp. 61-73). North-Holland: Elsevier SciencePublishers.Keogh, B., & Daley, S. (1983). Early identification: Onecomponent of comprehensive services for at-risk children.Topics in Early Childhood Special Education,  3, 7-16.Kerman, S., Kimball, T., & Martin, M. (1980). Teacher expectations and student achievement. Downey, CA: Office ofthe Los Angeles County Superintendent of Schools.Kinch, J. W. (1963). A formalized theory of self-concept.American Journal of Sociology, 68, 481-486.Krashen, S. (1978). The monitor model for second languageacquisition. In R. C. Gingras (Ed.), Second language acquisition and for^ (pp. 1-26).Arlington: CAL Press.Krashen, S., & Biber, D. (1988). On course: Bilingual education's success in California. Sacramento: California Association forBilingual Education.Krashen, S. D., Scarcella, R. C., & Long, M. H. (1982). Child-adult differences in second language acquisition. Rowley, MA:Newbury House.Kulik, C. L. (1985). Effects of inter-class ability grouping on achievement and self-esteem. Paper presented at the 1985Annual Meeting of the American Psychological Association, LosAngeles.Kulik, C. L., & Kulik, J. A. (1982). Effects of ability groupingon secondary school students: A meta-analysis of evaluationfindings. American Educational Research Journal, 21, 799-806.Lambert, W. E. (1967). A social psychology of bilingualism.Journal of Social Issues, 21(2), 91-109.Lambert, W. E. (1975). Culture and language as factors inlearning and education. In A. Wolfgang (Ed.), Education of immigrant students: Issues and answers  (pp. 55-80). Toronto:The Ontario Institute for Studies in Education SymposiumSeries (5).Lambert, W. E. (1981). Bilingualism: Its nature and andsignificance. In Lambert, W. E., Snow, E. C., Goldfield, B.A., Chamot, A. U., & Chair, S. R. (Eds.), Bilingual education series: 10. Faces and facets of bilingualism. NationalClearinghouse for Bilingual Education: Center for AppliedLinguistics.Lambert, W. E. (1984). An overview of issues in immersioneducation. In California State Department of Education (Ed.),Studies on immersion education: A collection for UnitedStates educators (pp. 8-30). Sacramento: California StateDepartment of Education.Lambert, W. E., Snow, E. C., Goldfield, B. A., Chamot, A. U., &Chair, S. R. (1981). Bilingual education series: 10. Faces and facets of bilingualism. National Clearinghouse forBilingual Education: Center for Applied Linguistics.Larter, S., & Cheng, M. (1984). Bilingual education and bilingualism: A review of research literature. Toronto:Board of Education, Research Department.Liu, W. T., & Yu, E. S. (1975). Asian American youth. In R. L.Havighurst (Ed.), Youth: National society for study of education year book. Chicago, Ill.: National Society forStudy of Education.Long, M. H. (1983). Does second language instruction make adifference? A review of research. TESOL Ouarterly, 12, 359-382.Long, M. H. (1988). Maturational constraints on languagedevelopment. University of Hawaii Working Papers in ESL,7(1), 1-53.Lynn, R. (1987). The intelligence of the Mongoloids: Apsychometric, evolutionary, and neurological theory.Personality and Individual Differences,  $, 813.Lynn, R., Pagliari, C., & Chan, J. (1988). Intelligence in HongKong measured for Spearman's g and the visuospatial and verbalprimaries. Intelligence, 12, 423-433.Maccoby, E. E., & Jacklin, C. N. (1974). ,The psychology of sexdifferences. Stanford, CA: Stanford University Press.1 761 77Maclntyre, R. B. (1985). Techniques for identifying learning-impaired minority students. In R. J. Samuda & A. Wolfgang(Eds.), Intercultural counselling and assessment: Globalperspectives (pp. 155-164). Toronto: C. J. Hogrefe.Maldonado-Colon, E. (1986). Assessment: Considerations uponinterpreting data of linguistically/culturally differentstudents referred for disabilities or disorders. In A. C.Willig & H. F. Greenberg (Eds.), Bilingualism^ lrnindisabilities: Policy and practice for teachers andadministrators (pp. 69-77). New York: American Library.Malherbe, E. G. (1978). Bilingual education in the Republic ofSouth Africa. In B. Spolsky & R. Cooper (Eds.), Case studies in bilingual education (pp. 167-202). Rowley, MA: NewburyHouse.Marsh, H. W. (1984a). Relationships among dimensions of self-attribution, dimensions of self-concept, and academicachievements. Journal of Educational Psychology, 15, 1291-1308.Marsh, H. W. (1984b). Self-concept, social comparison and abilitygroupings: A reply to Kulik and Kulik. American Educational Research Journal, 21, 799-806.Marsh, H. W. (1984c). Self-concept: The application of a frame ofreference model to explain paradoxical results. Australian Journal of Psychology, 2a, 165-181.Marsh, H. W. (1985a). Age and sex effects in multiple dimensionsof preadolescent self-concept. Australian Journal of Psychology, 37, 197-204.Marsh, H. W. (1985b). The structure of masculinity/femininity: Anapplication of confirmatory factor analysis to higher-orderfactor structures and factorial invariance. Multivariate Behavioral Research, 20, 427-449.Marsh, H. W. (1986a). The bias of negatively worded items inrating scales for young children: A cognitive-developmentalphenomena. Developmental Psychology, 22, 37-49.Marsh, H. W. (1986b). Global self-esteem: Its relation to specificfacets of self-concept and their importance. Journal of Personality and Social Psychology, 51, 1224-1236.Marsh, H. W. (1986c). The self serving effect (bias?) in academicattributions: Its relation to academic achievement and self-concept. Journal of Educational Psychology, 211, 190-200.Marsh, H. W. (1986d). Verbal and math self-concepts: Aninternal/external frame of reference model. American Educational Research Journal, 22, 129-149.178Marsh, H. W. (1987a). The big-fish-little-pond effect on academicself-concept. Journal of Educational Psychology,  22, 280-295.Marsh, H. W. (1987b). The factorial invariance of responses bymales and females to a multidimensional self-conceptinstrument: Substantive and methodological issues.Multivariate Behavioral Research, 22, 457-480.Marsh, H. W. (1987c). The hierarchical structure of self-conceptand the application of hierarchical confirmatory factoranalysis. Journal of Educational Measurement, 24, 17-19.Marsh, H. W. (1988a). The Self-Description Questionnaire - 1 (SDO-1): Manual and research monograph. San Antonio, TX:Psychological Corporation.Marsh, H. W. (1988b). The frame of reference effects of school-average ability: The high school and beyond study. Manuscriptsubmitted for publication.Marsh, H. W. (1988c). Multidimensional self-concepts: Do v-1 .n. n •. iv -1 w •r• - • m ndifferent components of self? Manuscript submitted forpublication.Marsh, H. W. (1988d). Sex differences in the development of verbal and math constructs: The high school and beyond studyManuscript submitted for publication.Marsh, H. W. (1988e). Age and sex effects in multiple dimensions of self-concept: Preadolescence to early-adulthoodManuscript submitted for publication.Marsh, H. W., Antill, J. K., & Cunningham, J. D. (1987).Masculinity, femininity, and androgyny: Relations to self-esteem and social desirability. Journal of Personality, 55,661-685.Marsh, H. W., Barnes, J., Cairns, L., & Tidman, M. (1984). TheSelf-Description Questionnaire (SDQ): Age effects in thestructure and level of self-concept for preadolescentchildren. Journal of Educational Psychology, 76, 940-956.Marsh, H. W., Barnes, J., & Hocevar, D. (1985). Self-otheragreement on multidimensional self-concept ratings: Factoranalysis & multitrait-multimethod analysis. Journal of Personality and Social Psychology, A2, 1360-1377.Marsh, H. W., Byrne, B. M., & Shavelson, R. J. (in press). Amultifaceted academic self-concept: Its hierarchical structureand its relation to academic achievement. Journal of Educational Psychology.1 79Marsh, H. W., Cairns, L., Relich, J. D., Barnes, J., & Debus, R.(1984). The relationship between dimensions of self-attribution and dimensions of self-concept. Journal of Educational Psychology, J., 3-32.Marsh, H. W., & Gouvernet (in press). Multidimensional self-concepts and perceptions of control: Construct validation ofresponses by children. Journal of Educational Psychology.Marsh, H. W., & Hocevar, D. (1983). Confirmatory factor analysisof multitrait-multimethod matrices. Journal of Educational Measurement, 21, 231-248.Marsh, H. W., & Hocevar, D. (1985). The application ofconfirmatory factor analysis to the study of self-concept:First and higher order factor structures and their invarianceacross age groups. Psychological Bulletin, 22, 562-582.Marsh, H. W., & Holmes, I. W. (1990). Multidimensional self-concepts: Construct validation of responses by children.American Educational Research Journal,  1/(1), 89-117.Marsh, H. W., McDonald, R. P., & Balla, J. R. (1987). Goodness-of-fit indices in confirmatory factor analysis: The effect ofsample size and of over-fitting and under-fitting models.Manuscript submitted for publication.Marsh, H. W., & Myers, M. R. (1986). Masculinity, femininity, andandrogyny: A methodological and theoretical critique. Sex Roles, JA, 397-430.Marsh, H. W., & O'Neill, R. (1984). Self-DescriptionQuestionnaire-III (SDQ-III): The construct validity ofmultidimensional self-concept ratings by late-adolescents.Journal of Educational Measurement, 21, 153-174.Marsh, H. W., & Parker, J. W. (1984). Determinates of self-concepts: Is it better to be a relatively large fish in asmall pond even if you don't learn to swim as well? Journal of Personality and Social Psychology, 47, 213-231.Marsh, H. W., Parker, J., & Barnes, J. (1985). Multidimensionaladolescent self-concepts. Their relationship to age, sex, andacademic measures. American Educational Research Journal,  22,422-444.Marsh, H. W., Parker, J. W., & Smith, I. D. (1983). Preadolescentself-concept: Its relation to self-concept as inferred byteachers and to academic ability. British Journal of Educational Psychology, aa, 60-78.Marsh, H. W., Relich, J. D., & Smith, I. D. (1983). Self-concept:The construct validity of interpretations based upon the SDQ.Journal of Personality and Social Psychology, j, 173-187.Marsh, H. W., & Richards, G. (1988). Self-other agreement andself-other differences on multidimensional self-conceptratings. Manuscript submitted for publication.Marsh, H. W., & Richards, G. (in press). The Outward BoundBridging Course for low achieving high-school males: Effect onacademic achievement and multidimensional self-concepts.Australian Journal of Psychology.Marsh, H. W., Richards, G., & Barnes, J. (1986a).Multidimensional self-concepts: A long-term follow-up of theeffect of participation in an Outward Bound program.Personality and Social Psychology Bulletin, 12, 475-492.Marsh, H. W., Richards, G., & Barnes, J. (1986b).Multidimensional self-concepts: The effect of participation inan Outward Bound program. Journal of Personality and Social Psychology, la, 173-187.Marsh, H. W., & Shavelson, R. J. (1985). Self-concept: Itsmultifaceted, hierarchical structure. Educational Psychologist, 2a, 107-125.Marsh, H. W., & Smith, I. D. (1982). Multitrait-multimethodanalyses of two self-concept instruments. Journal of Educational Psychology, 2A, 430-440.Marsh, H. W., & Smith, I. D. (1987). A cross-national study ofthe structure and level of multidimensional self-concepts: Anapplication of confirmatory factor analysis. Australian Journal of Psychology, 12, 61-77.Marsh, H. W., Smith, I. D., & Barnes, J. (1983). Multitrait-multimethod analyses of the Self-Description Questionnaire:Student-teacher agreement on multidimensional ratings ofstudent self-concept. American Educational Research Journal,24, 333-357.Marsh, H. W., Smith, I. D., & Barnes, J. (1984). Multidimensionalself-concepts: Relationships with inferred self-concepts andacademic achievement. Australian Journal of Psychology,  an,367-386.Marsh, H. W., Smith, I. D., & Barnes, J. (1985). Multidimensionalself-concepts: Relationships with sex and academicachievement. Journal of Educational Psychology, 22, 581-596.Marsh, H. W., Smith, I. D., Barnes, J., & Butler, S. (1983).Self-concept: Reliability, dimensionality, validity, and themeasurement of change. Journal of Educational Psychology, 75,772-790.1 80181Marsh, H. W., Smith, I. D., Myers, M. R., & Owens, L., (in press).The transition from single-sex to coeducational high schools:Effects on multiple dimensions of self-concept and on academicachievement. American Educational Research Journal.Maruyama, G., Rubin, R. A., & Kingsbury, G. G. (1981). Self-esteem and education: Independent constructs or a commoncause. Journal of Personality and Social Psychology, 40, 962-975.Marx, R. W., & Winne, P. H. (1978). Construct interpretations ofthree self-concept inventories. American Educational ResearchJournal, la, 99-108.McClelland, D.C. (1965). Toward a theory of motive acquisition.American Psychologist, 21, 321-333.McConnell, B. B., & Kendall, J. R. (1987, April). Application ofthe cohort model to evaluate bilingual programs: The "BELEPS" program. paper presented at the annual meeting of theAmerican Educational Research Association, Washington, DC.McDermott, R. P. (1978). Relating and learning: An analysis oftwo classroom reading groups. In R. Shuy (Ed.), Linguistics and reading. Rowley, Mass.: Newbury House.McLaughlin, B. (1984). Second language acquisition in childhood: Vol. 1. Preschool children (2nd ed.). Hillsdale, NJ: LawrenceErlbaum.Mead, G. H. (1934). Mind, self and society. Chicago: Universityof Chicago Press.Medina, M., Saldate, S., & Mishra, S. (1985). The sustainingeffects of bilingual instruction: A follow-up study. Journalof Instructional Psychology, 12(3), 132-139.Medina, M., & Valenzuela de la Garza, J. (1987, April). Bilingual instruction and academic gains of Spanish-dominant Mexican-American students. Paper presented at the annual meeting ofthe American Educational Research Association, Washington,D.C.Meece, J. L., Parsons, J. E., Kaczala, C. M., Goff, S. B., &Futterman, R. (19482). Sex differences in math achievement:Toward a model of academic choice. Psychological Bulletin,21, 324-348.Michaels, W. B., Smit, R. A., & Michaels, J. J. (1975). Thefactorial validity of the Piers-Harris Children's Self-conceptScale for each of three samples of elementary, junior high andsenior high students in a large metropolitan school district.Educational and Psychological Measurement, 35, 405-414.1 8 2Miyamoto, S. F., & Dornbusch, S. (1956). A test of interactionisthypotheses of self-conception. American Journal of Sociology,j5, 399-403.Nicholls, J. (1976). What scale measures more than its namedenotes: The case of the Test Anxiety Scale for Children.Journal of Consulting and Clinical Psychology, 55, 91-95.Nicholls, J. (1979). Development of perception of attainment andcausal attributions for success and failure in reading.Journal of Educational Psychology, 71, 94-99.Oakland, T. (Ed.) (1977). Psychological and educational assessment of minority children. New York: Brunner/Mazel.Ogbu, J. (1978). Minority education and caste: The American system in cross-cultural perspective. New York: AcademicPress.Olmeda, E. (1981). Testing linguistic minorities. AmericanPsychologist, ln, 1078-1085.011er, J. W. (1978). The language factor in the evaluation ofbilingual education. In J. E. Altis (Ed.), Georgetown University round table on languages_and linguistics 1978 (pp.410-422). Washington, D.C.: Georgetown University Press.011er, J. W. (1979). Language tests at school: A pragmaticapproach. New York: London.Omark, D. R., and Erickson, J. G. (Eds.). (1983). The bilingual exceptional child. San Diego, California: College-Hill Press.Ovando, C. J., & Collier, V. P. (1985). Bilingual and ESL classrooms: Teaching in multicultural contexts. New York:McGraw-Hill.Phillips, A. S. (1973). Adolescence in Jamaica. Kingston,Jamaica: Jamaica Publishing House.Piers, E. V., & Harris, D. A. (1964). Age and other correlates ofself-concept in children. Journal of Educational Psychology,55, 91-95.Plante, A. J. (1977). The Connecticut "pairing" model proveseffective in bilingual/bicultural education. Phi Delta Kappan,^427.Purkey, W. W. (1970). Self-concept and school achievement.Englewood Cliffs, New Jersey: Prentice-Hall.Ramsey, C. A. & Wright, E. N. (1974). Age and second languagelearning. Journal of Social Psychology, 94, 115-121.1 83Relich, J. D. (1983). Attribution, affective variables andarithmetic achievement Unpublished dissertation. TheUniversity of Sydney, Australia.Rist, R. C. (1970). Student social class and teacherexpectations: The self-fulfilling prophecy in ghettoeducation. Harvard Educational Review, 40, 411-451.Rogers, C. M., Smith, M. D., & Coleman, J. M. (1978). Socialcomparison in the classroom: The relationship between academicachievement and self-concept. Journal of Educational Psychology, ja, 50-57.Rosenberg, M. (1965). Society and the adolescent child.Princeton, NJ: Princeton University Press.Rosenberg, M. (1979). Conceiving the self^New York: BasicBooks.Ruble, D. N., Boggiano, A. K., Feldman, N. S., & Loebl, J. H.(1980). Developmental analysis of the role of socialcomparison in self-evaluations. Developmental Psychology,105-115.Ryan, R. M., Connell, J. P., & Grolnick, W. S. (in press). Whenachievement is not intrinsically motivated: A theory andassessment of self-regulation in school. In A. K. Boggiano &T. S. Pittman (Eds.), Achievement and motivation: A social-developmental perspective. Samuda, R.J. (1984). Assessing the abilities of minority studentswithin a multiethnic milieu. In Samuda, R. J., Beery, J., &Laferriere, M. (Eds.) (1984). Multiculturalism in Canada: Social and educational perspectives (pp. 353-367) Toronto,Ontario: Allyn and Bacon, Inc.Samuda, R. J., & Crawford, D. H. (1980). Testing, assessment, counseling, and placement of ethnic minority students: Current methods in Ontario. Toronto: Ontario Institute for Studies inEducation.Samuda, R. J., Berry, J., & Laferriere, M. (Eds.) (1984).Multiculturalism in Canada: Social and educational perspectives. Toronto, Ontario: Allyn and Bacon, Inc.Samuda, R. J., & Wolfgang, A. (1985). Intercultural counsellingand assessment: Global perspectives. Toronto: C.J. Hogrefe.Samuda, R.J., & Kong, S. L. (Eds.) (1986). Multicultural education: Programs and methods. Kingston: InterculturalSocial Sciences Publications.1 84Samuda, R.J., Kong, S.L., Cummins, J., Pascual-Leone, J., & Lewis,J. (Eds.) (1989). Assessment and placement of minority students^Kingston/Toronto: Intercultural Social SciencePublications.Sattler, J. M. (1988). Assessment of children (3rd ed.). SanDiego: Jerome M. Sattler.Saville-Troike, M. (1984). What matters in second languagelearning for academic achievement? TESOL Ouarterlv, j., 199-219.Scheirer, M. A., & Kraut, R. E. (1979). Increasing educationalachievement via self-concept change. Review of Educational Research, Al, 131-150.Schwartzer, R., Jerusalem, J., & Lange, B. (1983). The change ofself-concept with respect to reference groups in school.Paper presented at the 1983 Annual Meeting of the AmericanEducational Research Association, Montreal.Selinger, H., Karshen, S., & Ladefoged, P. (1975). Maturationalconstraints in the acquisition of second languages. Language Sciences, 38, 20-22.Shapson, S. M., D'Oyley, V., & Lloyd, A. (Eds.) (1982).Bilingualism and multiculturalism in Canadian education.Vancouver: Centre for the Study of Curriculum and Instruction,University of British Columbia.Shapson, S. and D'Oyley, J. (Eds.) (1984). Bilingualism andmulticultural education: Canadian perspectives. Clevedon,Avon: Multicultural Matters 15.Shavelson, R. J., & Bolus, R. (1982). Self-concept: The interplayof theory and methods. Journal of Educational Psychology, 74,3-17.Shavelson, R. J., Hubner, J. J., & Stanton, G. C. (1976).Validation of construct interpretations. Review Educational Research, 46, 407-441.Shavelson, R. J., & Marsh, H. W. (1986). On the structure ofself-concept. In R. Schwarzer (Ed.). Anxiety and cognitions.Hillsdale, JN: Lawrence Erlbaum.Sherman, J. (1980). Mathematics, spatial visualization, andrelated factors: Changes in boys and girls, grades 8-11.Journal of Education Psychology, /2, 476-482.Shibutani, T. (1961). Society and personality. Englewood Cliffs,JN: Prentice-Hall.1 85Shrauger, J. S., & Schoeneman, T. J. (1979). Symbolicinteractionist view of self-concept: Through the glass darkly.Psychological Bulletin, an, 549-573.Skutnabb-Kangas, T. (1979). Language in the process of culturalassimilation and structural incorporation of linguistic minorities. Rosslyn, VA: National Clearinghouse for BilingualEducation.Skutnabb-Kangas, T. (1981). Bilingualism or not: The education ofminorities. Clevedon, England: Multilingual Matters.Skutnabb-Kangas, T., & Toukomaa, P. (1976). Teaching migrant children's mother tongue and learning the language of the host country in the context of the socio-cultural situation of the migrant family. Helsinki: Finnish National Commission forUNESCO.Smith, I. D., & Marsh, H. W. (1985). A cross-cultural study ofself-concept in primary school children. Collected Papers of the 1985 Annual Conference of the Australian Association for Research in Education, pp. 350-353.Snow, C., & Hoefnagel-Hohle, M. (1978). The critical period forlanguage acquisition: Evidence from second language learning.Child Development, Aa, 1114-1128.Snow, C. E. and Goldfied, B. A. (1981). Bilingual education andfirst language acquisition. In Lambert, W. E., Snow, E. C.,Goldfield, B. A., Chamot, A. U., & Chair, S. R. (Eds.),Bilingual education series: 10. Faces and facets of bilingualism. National Clearinghouse for Bilingual Education:Center for Applied Linguistics.Soares, A. T., & Soares, L. M., (1969). Self-perceptions ofculturally disadvantaged children. American Educational Research Journal, _E., 31-45.Soares, L. M., & Soares, A. T. (1977). The self-concept: Mini, maxi, multi? Paper presented at the 1977 Annual Meeting ofthe American Educational Research Association, New York.Soares, L. M., & Soares, A. T. (1982). Convergence and discrimination in academic self-concepts. Paper presented atthe 20th Congress of the International Association of AppliedPsychology, Edinburgh, Scotland.Stevenson, H. W., & Newman, R. S. (1986). Long-term prediction ofachievement and attitudes in mathematics and reading. Child Development, .52, 646-659.Stipek, D. J. (1981). Children's perceptions of their own and^186their classmates' ability. Journal of Educational Psychology,22, 404-410.Stipek, D. J. (1984). The development of achievement motivation.In R. E. Ames & C. Ames (Eds.), Research on motivation in education (Vol. 1, pp. 145-174). Orlando, FL: Academic Press.Stipek, D. J. & Tannatt, L. M. (1984). Children's judgments oftheir own and their peers' academic competence. Journal of Educational Psychology, 2L, 75-84.Strang, L., Smith, M. D. & Rogers,C. M. (1978). Socialcomparison, multiple reference groups and the self-concepts ofacademically handicapped children before and aftermainstreaming. Journal of Educational Psychology, la, 487-497.Swain, M., & Lapkin, S. (1981). Bilingual education in Ontario: Adecade of research. Toronto: Ministry of Education.Tam, S. (1990). Cognitive assessment of Chinese immigrant students in Cantonese and English. Unpublished master'sthesis, University of British Columbia, Vancouver.Tempes, F., Burnham, L., Pina, M., Campos, J., Matthews, S., Lear,E., & Herbert, C. (1984). Implementing theoretically soundprograms: Do they really work? Paper presented at the annualconference of the California Association for BilingualEducation, San Francisco.Troike, R. (1978). Research evidence for the effectiveness ofbilingual education. NABE Journal, ..(1), 13-24.Trowbridge, N. (1970). Effects of socio-economic class on self-concept of children. Psychology in the Schools, 7, 304-306.Trowbridge, N. (1972). Self-concept and socio-economic status inelementary school children. American Educational Research Journal, . 525-537.Tucker, J. A. (1980). Ethnic proportions in classes for thelearning disabled: Issues in nonbiased assessment. Journal of Special Education, 14, 93-105.Vernon, P. E. (1982). The abilities and achievements of Orientals in North America. New York: Academic Press.Vernon, P. E. (1980). Chinese immigrants and citizens in Canada.New Horizons, 21, 12-25.Vorih, L. & Rosier, P. (1978). Rock Point Community School: Anexample of a Navajo-English bilingual elementary schoolprogram. TESOL Quarterly, 12, 263-269.187Weiner, B. (1980). Human motivation. New York: Holt, Rinehart, &Winston.Wells, C. S. (1979). Describing children's linguistic developmentat home and at school. British Educational Research Journal,5, 75-89.Wells, L. E., & Marwell, G. (1976). Self-esteem: Its conceptualization and measurement. Beverly Hills, CA: SagePublications.Wilkinson, S. M., & Burke, J. P. (1984). Ethnicity, socioeconomicstatus, and self-concept: Effects on children's academicperformance. Journal of Instructional Psychology, 11(4), 203-210.Williams, R. L., & Byars, H. (1968). Negro self-esteem in atransitional society. Personnel and Guidance Journal, 47,120-125.Willig, A. C. (1986). Special education and the culturally andlinguistically different child: An overview of issues andchallenges. In A. C. Willig & H. F. Greenberg (Eds.),Bilingualism and learning disabilities: Policy and practice for teachers and administrators  (pp. 191-209). New York:American Library.Wilson, R. S. (1983). The Louisville twin study: Developmentalsynchronies in behavior. Child Development, 54, 298-316.Winne, P. H., Marx, R. W., & Taylor, T. D. (1977). A multitrait-multimethod study of three self-concept inventories. ChildDevelopment, 48, 893-901.Winne, P. H., Woodlands, M. J., & Wing, B. (1982). Comparabilityof self-concept among learning disabled, normal and giftedstudents. Journal of Learning Disabilities, .1.(8), 470-475.Wong Fillmore, L. (1983). The language learner as an individual:Implications of research on individual differences for the ESLteacher. In M. Clarke, & J. Handscombe (Eds.), On TESOL '82: Pacific perspectives on language learning and teaching.Washington, DC: TESOL.Wright, P., & Santa Cruz, R. (1983). Ethnic composition ofspecial education programs in California. LearningDisabilities Quarterly, ., 387-394.Wylie, R. C. (1974). The self-concept: A review of methodologicalconsiderations and measuring instruments (Rev. ed., Vol. 1).Lincoln: University of Nebraska Press.Wylie, R. C. (1979). The self-concept: Theory and research on selected topics (Vol. 2). Lincoln: University of NebraskaPress.Wylie, R. C. (1989). Measures of self-concept. Lincoln:University of Nebraska Press.Zirkel, P. (1971). Self-concept and the disadvantage of ethnicgroup membership. Review of Educational Research, 41, 211-225.1 88APPENDIX A: LETTER TO TEACHERS AND PRINCIPALS189Barbara SilvferGraduate StudentM.A. ProgramSchool PsychologyREQUEST FOR SUBJECT PARTICIPATION - TEACHER FORMDear Teacher:Your school has been requested to take part in a M.A. thesis projectto examine the relationship among the following parameters:1. ESL student's self-concept;2. Teacher perceptions of that student's self-concept; and3. The child's school performance or report card grades.As yet, little information is available in the area of self-conceptand Canadian ethnic minorities. We hope that this study may furtherour understanding of minority group self-concept and the dynamics ofthis professional interaction between teacher and student.In this study, we will be looking at ESL students who meet thefollowing criteria for selection:1. Chinese immigrants, preferably of Hong Kong origin.2. Resident in Canada for approximately 5 years.3. Presently in grades 5 or 6.4. Aged 10 - 12 years; approximately 11 years old.5. Presently mainstreamed into a regular classroom.As some of your present students meet these criteria, we are asking .for your participation in this research project - to contribute yourvaluable perceptions regarding your student's self-concepts. .The same self-concept questionnaires will be completed separately byboth the student and his/her teacher. This questionnaire will takethe teacher. approximately 10 - 15 minutes to complete per child.Finished questionnaires will then be given a code to maintain theanonymity and confidentiality of all concerned. Although you have theright to refuse to participate and you may withdraw from the study atany time, your involvement is crucial to the success of this project.On conclusion of the study, a complete summary will be given to eachparticipating teacher which will highlight the research findings.If you are willing to help in this request - and we hope that you willbe - please let your principal know so that we may schedule a meeting.And, if you have any further questions, please don't hesitate tocontact the researchers at UBC (228-5351).Again, thank you for your cooperation.Sincerely,D. Whittaker, Ph.D.Associate ProfessorDepartment of EducationalPsychology and Special Education190REQUEST FOR ADMINISTRATIVE AND SCHOOL TIME - SCHOOL PRINCIPAL FORMDear Principal:Your school has been requested to take part in a M.A thesis project toexamine the relationship among the following parameters:1. ESL student's self-concept;2. Teacher perceptions of that student's self-concept; and3. The child's school performance or report card grades.As yet, little information is available in the area of self-conceptand Canadian ethnic minorities. We hope that this study may furtherour understanding of minority group self-concept and the dynamics ofthis professional interaction between teacher and student..In order to properly select a representative sample of students, somesecretarial time will be needed to identify possible ESL students fromtheir cumulative school records. Criteria for student selectioninclude the following:1. Chinese immigrants, preferably of Hong Kong origin.2. Resident in Canada for approximately 5 years.3. Presently in grades 5 or 6.4. Aged 10 - 12 years; approximately 11 years old.5. Presently mainstreamed into a regular classroom.Should some of your present students indeed meet these criteria, wewill also be asking for the participation of their respectiveteachers, to contribute their valuable perceptions regarding theirstudent's self-concepts.The same self-concept questionnaires will be completed separately byboth the student and his/her teacher. The questionnaire will take thestudents approximately 15 - 20 minutes to complete; teachers maycomplete the same forms in approximately 10 - 15 minutes per child.The student data is to be collected once only at your school, in agroup testing of all consenting ESL students so identified. On thesame day, the researcher will also meet with the respective teachersto distribute the teacher questionnaires, to administer specificinstructions as to their completion, and to answer any questionsconcerning the study. Teacher questionnaires will be picked up at ascheduled time the following week.^It is hoped that all the data maybe collected at a convenient time for your school in February 1990, tobe completed by early March 1990.All finished questionnaires will be given a code to maintain anonymity.and confidentiality of all concerned. Although you have the right torefuse to participate and you may withdraw from the study at any time,yoUr willingness to allocate your time and energy is crucial to thesuccess of this project. On conclusion of the study, a completesummary, highlighting the research findings, will be made available tothe participating teachers and to the parents of the ESL students.If you have any questions, please don't hesitate to contact theresearchers at UBC (228-5351). We most greatly appreciate yourcooperation and we hope to be in touch with you soon to arrangefurther details. Thank you.Sincerely,191Barbara Sil Jer D. Whittaker, Ph.D.Graduate Student^ Associate ProfessorM.A. Program Department of EducationalSchool Psychology Psychology and Special EducationAPPENDIX B:^PARENTAL/GUARDIAN AND STUDENT CONSENT FORMS192193ESL Self-Concept:^Its Relation to Teacher Perception andAcademic AchievementDear Parent/Guardian:Your child has been selected to take part in a M.A. thesisproject to understand more about the self-concept ofchildren for whom English is a second language. Theseresults will be compared to the teacher's rating of yourchild's self-concept and to the child's school grades.The questionnaire your child will complete takesapproximately 15-20 minutes. Your child's schobl has givenus permission for this to be done at school.^I will sendthe results to UBC. Your child's name and school gradeswill be given a code number so that nobody will know to whomthey belong. Only the UBC researchers will know the resultsso they will not affect how your child is graded in school.Most children enjoy answering the questions and may learnmore about themselves. Both you and your child have theright to refuse to participate and, of course, you maywithdraw from the study at any time without affecting yourchild's class standing. However, your help is important tous.Please sign and complete the attached consent form andreturn it to your child's school as soon as possible. Asummary of the research findings will be available to you oncompletion of the study. If you have any questions, pleasecontact the researchers at UBC (228-5351).Thank you very much for your cooperation.Sincerely,Barbara SilverGraduate StudentM.A. Program School PsychologyD. Whittaker, PhD.Associate ProfessorUniversity of British ColumbiaDepartment of EducationalPsychology and Special Education228-5351194ESL Self-Concept:^Its Relation to Teacher Perception andAcademic AchievementREQUEST FOR SUBJECT PARTICIPATION - PARENT/GUARDIAN CONSENTI DO / DO NOT give permission for my child ^(circle one)^ (name of child)to participate in the M.A. thesis project and havehis/her school grades released to the UBC researchers.Signature of Parent or GuardianLETTER REQUESTING PARENT/GUARDIAN CONSENTIt^(IV 46c -k g4 4, 6^11tA^-^0 ±^ r. ,^j -,19-% ielj^471) :"Tf.^.11i; Ilk^14,^R-^g4t -1 Z. 11 ak:11-^% Et^AN6^i9 ti 137) )1.),tiL /44e, jYJ i J A 470 '4( fo (1,1^I^ ft. 11)^41.12-^.1)9 -4-3 a/13 f. 0- 3211^ITT^,^3 5'1 it pi If)1)-ki 5w) tt)/0 ,^f2,rrm16 ,01 4-$^fk_1 -f)^-Dv^44`z)^5)14- ,^k)blit flit^;) ep^, 4):^-0,^--A^-1•E.,f1b -t^, ;1. Cs, ,^1- t^D 1+211 4- t'-AL^,^s73^t^Y51 t- 011 4 54- g- 14lcilt -610- is13^19. -415i1111.AttAi- "Itk -t3(-T-11F.7.1 20. . 1, 1)^P.,tkt^-63SI 4 141- k-acts-Jars. 51(Vet 4)t..195196ESL Self-Concept:^Its Relation to Teacher Perception andAcademic AchievementREQUEST FOR SUBJECT PARTICIPATION - STUDENT FORM(to accompany the student questionnaire and to be read toeach student prior to testing)Dear Student:As you may know by now, you have been selected to take partin a M.A. thesis project to understand more about children'sself-concept or how you feel about yourself. When wefinish, I will send these questionnaires to UBC. Then Iwill give your name a code number so nobody will know it wasyou - we only want to see how children answer thesequestions, okay?I want you to remember that these questions have nothing todo with your schoolwork and will not count for your gradeson your report card. The questionnaire takes about 15-20minutes to answer. Most children enjoy doing it and maylearn more about themselves.^I'm sure you will too. Beforewe start, I want you to know that you don't have to do this.You may refuse to participate or withdraw at any timewithout affecting your class standing. However, your helpis important to us.^I would appreciate it if you wouldagree to work on this questionnaire with me. When you signand finish the questionnaire, we will know that you agree tohelp us. Okay?Thank you very much for your help.Sincerely,Barbara SilverGraduate StudentM.A. Program School PsychologyD. Whittaker, PhD.Associate ProfessorDepartment of EducationalPsychology and Special Education228-5351APPENDIX C:^SCHOOL DEMOGRAPHIC QUESTIONNAIRE197198Part 1:^SCHOOL BACKGROUND INFORMATION^(principal's verbal report)School Telephone District^Principal^ Testing Date^Teachers Grade^*students tested^^ Grade^*students^ Grade^*students^  Grade^*students^Grade^*students^ Grade^-*students^Total Sage of ESL Students in SchoolSchool Minority Groups ^Type and # of ESL Classes in School^# Students Receiving ESL Help^School Average SES^School Average Ability/Achievement^Part 2:^SUMMARY TEST INFORMATIONTotal * Students Tested^M^F^Mean Age^Gr.S ^Mean Age^Cr. 6^M^F^Mean AgeEthnic Group^ total^M^Xage^F^Xagetotal^II^Xage^F^Xage^ total^M^Xage^F^Xagetotal^M^Xage^F ^- Xage^ total^M^Xage^F^Xage APPENDIX D:^SELF-DESCRIPTION QUESTIONNAIRE-1^(SDQ-1)199ISELF-DESCRIPTIONQUESTIONNAIRE- IYour Name.^  Circle one: Boy GirlSchool.  Grade:^ Age.^Teacher:   Date-This is a chance to look at yourself. It is not a test. There are no right answers, and everyone will havedifferent answers. Be sure that your answers show how you feel about yourself. PLEASE DO NOTTALK ABOUT YOUR ANSWERS WITH ANYONE ELSE. We will keep your answers private and notshow them to anyone.When you are ready to begin, please read each sentence and choose an answer. (You may readquietly to yourself as I read aloud.) There are five possible answers for each question: "Truer "Falserand three answers in between. There are five boxes next to each sentence, one for each of theanswers. The answers are written at the top of the boxes. Choose your answer to a sentence andmake a check mark in the box under the answer you choose. DO NOT say your answer out loud ortalk about it with anyone else.Before you start, there are three examples below. A student, Bob, has already answered two of thesesentences to show you how to do it. In the third example you must choose your own answer and putin your own check mark.SOME-TIMESFALSE/SOME.MOSTLY TIMESEXAMPLES FALSE FALSE TRUE TRUEMOSTLY TRUE1. I like to read comic books  1 EDBob checked the box under the answer "Truer This means that he really likes to read comicbooks. If Bob did not like to read comic books very much, he would have answered "FALSE"or "MOSTLY FALSE:'2. In general, I am neat and tidy   2Bob answered "SOMETIMES FALSE, SOMETIMES TRUE" because he is not very neat, buthe is not very messy either.3. I like to watch TV • 3 I^I I^I 3For this sentence you have to choose the answer that is best for you. First you must decide ifthe sentence is "TRUE" or "FALSE;' or somewhere in between. If you really like to watch T.V.a lot, you would answer "TRUE" by making a check mark in the last box. If you hate watchingTV., you would answer "FALSE" by making a check mark in the first box. If your answer issomewhere in between, then you would choose one of the other three boxes.If you want to change an answer you have marked, you should cross out the check mark andput a new check mark in another box on the same line.For all the sentences be sure that your check mark is on the same line as the sentence you areanswering. You should have one answer and only one answer for each sentence. Do not leaveout any of the sentences. Once you have started, PLEASE DO NOT TALK. Turn over the pageand begin.200SOME-TIMESFALSE/SOME-MOSTLY TIMES MOSTLYFALSE FALSE TRUE^TRUE^TRUE1. I am good looking ^  1 1 11 1=== 1 .2. I'm good at all SCHOOL SUBJECTS^ 2 ED = = 1 1 1 1 23. I can run fast ^  3 1 1E:=1=== 34. I get good marks in READING ^ 4 I^11^1 == 1^1 45. My parents understand me  5 1 1 = 1^= = 56. I hate MATHEMATICS^  6 0 1 1 1^= = 67. I have lots of friends  7 1^1 I^ j= E ^1-1 78. I like the way I look ^  8 = 1^1 1^1^1 1^1 8=1 1910. I like to run and play hard ^  10 I. 1 = = = = 1011. I like READING ^ 11 1^1 = 1^11^11^11112. My parents are usually unhappy ordisappointed with what I do ^ 12 1^1 1^1 1^1 1^1 Q 1213. Work in mathematics is easy for me^ 13 Q 1^.11^11^11^1 13SOME-TIMESFALSE/SOME-MOSTLY TIMES MOSTLYFALSE FALSE TRUE^TRUE^TRUE14. I make friends easily ^  14 1^11^1 = 1^1 7-11415. I have a pleasant looking face ^ 15 1^11^11^"1=1^1516. I get good marks in all SCHOOL SUBJECTS ^ 16 1^1 1^1 i^1 = 1^1617. I hate sports and games ^  17 I^li^11^J1^11^1718. I'm good at READING 18 1^1 1^1^1^1 1^1819. I like my parents ^•  19 1^1 1^1 1^11^1 1^1920. I look forward to MATHEMATICS ^ 20 1^If^11^II^II^I 2021. Most kids have more friends than I do ^ 21 1-1 1^1 I.^1 = 1—.1 2122. I am a nice looking person ^ 22 1^11^1=1^1= 2223. I hate all SCHOOL SUBJECTS 231^11^11^II^1 1^1 2324. I enjoy sports and games ^ 24 r^1=1-71^II^1 2425. I am interested in READING 25 1^1=1^11= I^1 2526. My parents like me ^ 26 ED = = I 1 I 1 262019. I enjoy doing work in at SCHOOL SUBJECTS .... 9 [1 1^126. I have good muscles ^ 32f^1 31CH I31. I learn things quickly in all SCHOOL SUBJECTS .. 31 ni ^ 1^1I 32I 1 037361^11. 39SOME-TIMESFALSE!SOME-MOSTLY TIMES MOSTLYFALSE FALSE TRUE TRUE TRUE202C:::1^ED V28 =^ = 2829 ED30 = =^3027. I get good marks in MATHEMATICS^ 27 I^I28. I get along with kids easily ^29. I do lots of important things 30. I am ugly ^2933. I am dumb at reading 331^1 1^1 1^1 =3= 3334. If I have children of my own, I want tobring them up like my parents raised me ^35. I am interested in MATHEMATICS ^36. I am easy to like ^37. Overall, I am no good 37 r---138. Other kids think I am good looking ^ 3839. I am Interested in all SCHOOL SUBJECTS ^ 39SOME-TIMESFALSE!SOME-MOSTLY TIMES MOSTLYFALSE FALSE TRUE TRUE TRUE40. I am good at sports ^ 40 = Q = Q = 4041. I enjoy doing work in READING ^ 41 0 I^1^Cl 1^1 4142. My parents and I spend a lot of time together ^ 42 Q Q ELI 4243. I learn things quickly in MATHEMATICS ^ 43 I^144. Other kids want me to be their friend45. In general, I like being the way I am ^ 45 I^I46. I have a good looking body ^47. I am dumb in all SCHOOL SUBJECTS^48. I can run a long way without stopping ^49. Work in READING is easy for me ^50. My parents are easy to talk to 51. I like MATHEMATICS ^ 5152. I have more friends than most other kids ^ 52 E-1 I^134=^ 3435 = O I 1 = 3536 I-1^1^1^361^1C^O 1 1 4344 =LJD L.JO 441^ H 10 454546 Q = Q = 4647 E=1 ED = = 4748 ED46 ED50 50ED CD 0 51= 52I^I II^11^I1I^11^14849I^II^I53. Overall I have a lot to be proud of ^ 53 1^1 1^Q 1^1^5354. I'm better looking than most of my friends ^ 54 =55. I look forward to all SCHOOL SUBJECTS ^ 55 I56. I am a good athlete ^ 56 I^I57. I look forward to READING ^ 57 1---158. I get along well with my parents 5859. I'm good at MATHEMATICS ^ 5960. I am popular with kids of my own age^ 6061. I can't do anything right^ 61 ni62. I have nice features like nose, and eyes, and hair . 62 r763. Work in all SCHOOL SUBJECTS is easy for me ^ 63 1^1 =64. I'm good at throwing a ball ^ 64 I54I^I55I^I60I^I611II^ICH I O O 73O ED 74O E:=1 O 757665. I hate READING ^ 65 1-1 I^I66. My parents and I have a lot of fun together^ 66 1---1 I^167. I can do things as well as most other people ^ 671-168. I enjoy doing work in MATHEMATICS^ 68 = I^I I=69. Most other kids like me70. Other people think I am a good person^ 70 F-1 I^I71. I like all SCHOOL SUBJECTS ^ 71 ni I 1 Cl E O 7172. A lot of things about me are good ^ 72 r--173. I learn things quickly in READING ^ 73 I^I74. I'm as good as most other people ^ 74 FT I^I^I75. I am dumb at MATHEMATICS  75 ni I 176. When I do something, I do it well ^ 76 I^1 I^I I^ISOME.TIMESFALSE/SOME-MOSTLY TIMES MOSTLYFALSE FALSE TRUE^TRUE^TRUE62= 63I^I I^I 6465= 661 67I^I 691=H 1 1 1701^I I^I CD I 1 72I^IL^IO I^I=•^E-1 5771 58O I^1 59OI^1I^IBarbara Sil erGraduate StudentM.A. ProgramSchool Psychology204ESL Self-Concept:^Its Relation to Teacher Perception andAcademic AchievementDIRECTIONS FOR COMPLETION OF THE SDQ - TEACHER QUESTIONNAIRE(to accompany each teacher questionnaire and to be read toeach teacher before its completion)1. Please fill out the pertinent information on the frontcover of the Self-Description Questionnaire 1 regardingthe child you are evaluating. This includes thechild's name,. sex, school, grade, age, and birthdate.2. Please print your name and date the form in the spacesprovided. Completion of the questionnaire signifiesyour consent to participate in this study.Remember, finished questionnaires will be given a codeto maintain the anonymity and confidentiality of allconcerned. As participation is voluntary, you have theright to refuse to participate and you may withdrawfrom the study at any time.3. Please do not discuss your evaluations with anyoneelse, as it may invalidate the results.Then...4. PLEASE EVALUATE THE PUPIL'S SELF-CONCEPT, USING YOURPERCEPTIONS OF THE STUDENT'S OWN FEELINGS FOR EACH OFTHE FOLLOWING STATEMENTS.When you are finished, please seal all questionnairesin the envelope provided to ensure confidentiality.All questionnaires will be picked up from your schoolin one week onThank you for your cooperation.D. Whittaker, Ph.D.Associate ProfessorDepartment of EducationalPsychology and Special Education228-5351SELF-DESCRIPTIONQUESTIONNAIRE- IChild's Name! ^Circle one:^Boy^GirlSchool:^  Grade-^ Age:^Teacher  Date-Teacher's Signature: ^Birthdate:^This is a chance to look at yourself. It is not a test. There are no right answers, and everyone will havedifferent answers. Be sure that your answers show how you feel about yourself. PLEASE DO NOTTALK ABOUT YOUR ANSWERS WITH ANYONE ELSE. We will keep your answers private and notshow them to anyone.When you are ready to begin, please read each sentence and choose an answer. (You may readquietly to yourself as I read aloud.) There are five possible answers for each question: "True:* "False: .and three answers in between. There are five boxes next to each sentence, one for each of theanswers. The answers are written at the top of the boxes. Choose your answer to a sentence andmake a check mark in the box under the answer you choose. DO NOT say your answer out loud ortalk about it with anyone else.Before you start, there are three examples below. A student, Bob, has already answered two of thesesentences to show you how to do it. In the third example you must choose your own answer and putin your own check mark.EXAMPLESSOME.TIMESFALSE/SOME-MOSTLY TIMES MOSTLYFALSE FALSE TRUE^TRUE^TRUE1. I like to read comic books ^  1I^II^II^II^II lam( 1Bob checked the box under the answer "True:' This means that he really likes to read comicbooks. If Bob did not like to read comic books very much, he would have answered "FALSE"or "MOSTLY FALSE:'2. In general. I am neat and tidy ^ 2Bob answered "SOMETIMES FALSE, SOMETIMES TRUE:' because he is not very neat, buthe is not very messy either.3. I like to watch TV 31   I^II^II^II^I^3For this sentence you have to choose the answer that is best for you. First you must decide ifthe sentence is "TRUE:' or "FALSE' or somewhere in between. If you really like to watch TV.a lot, you would answer "TRUE" by making a check mark in the last box. If you hate watchingTV., you would answer "FALSE" by making a check mark in the first box. If your answer issomewhere in between, then you would choose one of the other three boxes.If you want to change an answer you have marked, you should cross out the check mark andput a new check mark in another box on the same line.For all the sentences be sure that your check mark is on the same line as the sentence you areanswering. You should have one answer and only one answer for each sentence. Do not leaveout any of the sentences. Once you have started, PLEASE DO NOT TALK. Turn over the pageand begin.205I^I 1^1 I^ yr^1^1 2APPENDIX E: TEACHER REPORT OF STUDENT BACKGROUND ANDGRADES206207Child's Name: ^Birthdate: ^  Age: ^Teacher's Name: ^School: ^  Grade: ^Type of Class (circle):^RegularESL (full-time)ESL / ELC / LAC^(pull-out)Canadian-born / Immigrant^(circle):Specify:^ Country of Origin ^Ethnic Group ^Languagets) Spoken at Home ^Number of Years in Canadaparents ^ childAge on Arrival in CanadaMarks on last Report Card: If not given, please rate (A, B, C, D)or(I, 2. 3. 4)Mathematics:Reading / orLanguage Arts: ^APPENDIX F:^DEFINITION OF SDQ-1 SCALES208The Physical Abilities scale measures a child's self-concept regarding hisor her abilities in physical activities, sports, and games. Items in thePhysical Abilities scale are:Item No.^Item^3.^I can run fast.^10. I like to run and play hard.24.^I enjoy sports and games.32. I have good muscles.40.^I am good at sports.48. I can run a long way without stopping.56.^I am a good athlete.64. I am good at throwing a ball.Figure 2. Physical Abilities ScaleItems of the Physical Appearance scale reflect a child's self-conceptregarding his or her physical attractiveness as compared with others, andthe perception of how others think he or she looks. Items in the PhysicalAppearance scale are:Item No.^Item^1.^I am good looking.8. I like the way I look.15.^I have a pleasant looking face.22. I am a nice looking person.38.^Other kids think I am good looking.46. I have a good looking body.54.^I am better looking than most of my friends.62. I have nice features like nose, and eyes, and hair.Figure 3. Physical Appearance ScaleThe Peer Relations scale measures the child's self-concept regarding hisor her popularity with peers, how easily the child makes friends, andwhether others want him or her as a friend. Items of the Peer Relationsscale are:Item No.^Item^7.^I have lots of friends.14. I make friends easily.28.^I get along with kids easily.36. I am easy to like.44.^Other kids want me to be their friend.52. I have more friends than most other kids.60.^I am popular with kids my own age.69. Most other kids like me.209Figure 4. Peer Relations ScaleThe Parent Relations scale reflects how well the child thinks he or she getsalong with his or her parents, how well the child likes his or her parents,and the extent to which the child experiences parental acceptance andapproval. Items of the Parent Relations scale are:Item No.^Item^5.^My parents understand me.19. I like my parents.26.^My parents like me.34. If I have children of my own, I want to bring themup like my parents raised me.42.^My parents and I spend a lot of time together.50. My parents are easy to talk to.58.^I get along well with my parents.66. My parents and I have a lot of fun together.'Figure 5. Parent Relations ScaleThe Reading scale reflects the child's self-concept regarding his or herability, enjoyment, and interest in reading. Items of the Reading scale are:Item No.^Item^4.^I get good marks in READING.11. I like READING.18.^I am good at READING.25. I am interested in READING.41.^1 enjoy doing work in READING.49. Work in READING is easy for me.57.^I look forward to READING.73. I learn things quickly in READING.Figure 6. Reading ScaleThe Mathematics scale measures the child's self-concept regarding his orher ability, enjoyment, and interest in mathematics. Items of the Mathematicsscale are:Item No.^Item^13.^Work in MATHEMATICS is easy for me.20. I look forward to MATHEMATICS.27.^I get good marks in MATHEMATICS.35. I am interested in MATHEMATICS.43.^I learn things quickly in MATHEMATICS.51. I like MATHEMATICS.59.^I am good at MATHEMATICS.68. I enjoy doing work in MATHEMATICS.210Figure 7. Mathematics ScaleThe General-School scale measures the child's self-concept regarding hisor her ability, enjoyment, and interest in school subjects. Items of theGeneral-School scale are:Item No.^Item^2.^am good at all SCHOOL SUBJECTS.9. enjoy doing work in all SCHOOL SUBJECTS.16.^get good marks in all SCHOOL SUBJECTS.31. learn things quickly in all SCHOOL SUBJECTS.39.^am interested in all SCHOOL SUBJECTS.55. look forward to all SCHOOL SUBJECTS.63.^Work in all SCHOOL SUBJECTS is easy for me.71. I like all SCHOOL SUBJECTS.Figure 8. General-School ScaleThe General-Self scale reflects the child's perception of himself or herselfas an affective, capable individual, proud of and satisfied with the way heor she is. Items of the General-Self scale are:Item No.^Item 29.^I do lots of important things.45. In general, I like being the way I am.53.^Overall I have a lot to be proud of.67. I can do things as well as most other people.70.^Other people think I am a good person.72. A lot of things about me are good.74.^I'm as good as most other people.76. When I do something, I do it well.Figure 9. General-Self Scale211APPENDIX G:^SDQ-1 SCORING AND PROFILE BOOKLET212SELF-DESCRIPTIONQUESTIONNAIRE-ISD eSCORING ANDPROFILE BOOKLETHERBERT W. MARSHCHILD'S NAME .DATE .SCHOOL .^TEACHER:SEX: ^ M ^ F AGE:^ GRADETHE PSYCHOLOGICAL CORPORATIONHARCOURT BRACE JOVANOVICH, INC.Copyright 0 1988 by The Psycholospcal Corporation. All rights reserved. No part of thispublication may be reproduced or transmitted In any form or by any means. electronicor mechanical, including Photocopy, recording, or any information storage and retrievalsystem, without permission in writing horn the Publisher. Printed in the United Statesof America.I 2 3050 701110^12 ABCDE^ 09869439--^•213Score Calculation and Summary214INDIVIDUAL SCALE SCORES:  For each scale, write the scores for the items listed in the blanks beside the item numbers. Sum theitem scores within each scale and write the total raw score in the blank provided below the item scores.PhysicalAbilitiesPhysicalAppearancePeerRelationsParentRelations Reading MathematicsGeneral-SchoolGeneral-SelfItem^(Mean)' Item^(Mean)' Item^(Mean)' Item^(Mean)' Item^(Mean)' Item^(Mean)' Item^(Mean)' Item^(Mean)'3 - (3.84) 1 - (3.53) 7 - (4.46) 5 - (4.38) 4 - (3.78) 13 - (3.52) 2 - (3.35) 29 - (3.74)10 - (4.14) 8 - (3.64) 14 - (4.01) 19 -- (4.80) 11 - (3.96) 20 - (3.23) 9 - (3.56) 45 - (4.35)24 - (4.66) 15 - (3.39) 28 - (4.10) 26 - (4.79) 18 - (3.95) 27 - (3.75) 16 - (3.42) 53 - (4.22)32 - (3.71) 22 - (3.43) 36 - (3.73) 34 - (4.30) 25 - (3.96) 35 - (3.64) 31 - (3.77) 67 - (4.22)40 - (4.28) 38 - (3.23) 44 - (3.98) 42 - (4.10) 41 - (3.87) 43 - (3.76) 39 - (3.79) 70 - (4.06)48 - (3.82) 46 - (3.42) 52 - (3.35) 50 - (4.27) 49 - (3.94) 51 - (3.62) 55 - (3.64) 72 - (4.05)56 - (3.89) 54 - (3.15) 60 - (3.98) 58 - (4.53) 57 - (3.82) 59 - (3.76) 63 - (3.40) 74 - (4.16)64 - (4.36) 62 - (3.70) 69 - (4.02) 66 - (4.34) 73 - (4.04) 68 - (3.51) 71 - (3.63) 76 - (4.19)RAWSCALETOTALSTOTAL NONACADEMIC: Copy the Raw Scale Totals for Physical Abilities, Physical Appearance, Peer Relations, and Parent Relationsinto the blanks provided below. Sum these scores and divide by 4 to get the Total Nonacademic raw score.4^= ^(Physical^(Physical^(Peer^(Parent^(Total)^ TOTAL NONACADEMICAbilities)^Appearance)^Relations)^Relations) Raw ScoreTOTAL ACADEMIC: Copy the Raw Scale Totals for Reading, Mathematics, and General-School into the blanks provided below.Sum these scores and divide by 3 to get the Total Academic raw score.3(Reading)^(Mathematics)^(General-School)^(Total)^ TOTAL ACADEMICRaw ScoreTOTAL SELF: Copy the Total Nonacademic and Total Academic raw scores into the blanks provided below. Sum thesescores and divide by 2 to get the Total Self raw score.2^= ^(Total Nonacademic)^(Total Academic)^(Total)^ TOTAL SELFRaw ScoreCONTROL SCORES (See Appendix A of the Manual for instructions on calculating Control raw scores.)Control^Control^Control^Control^Control^ControlScore 1 Score 2 Score 3 Score 4 Score 5 Score 6'Substitute the item mean for missing responses only if three or fewer responses are left blank.CONTROL SCOREST-Score ProfileDirections: Transfer the raw scores for the individual and total scales (and control scores) from page 2 to the spaces pro-vided below the profile. Then, convert the raw scores to percentile ranks and T scores using the tables in Appendices Aand B of the Manual. Record these values in the spaces provided and plot the T scores on the profile.I It 1, ctli70. cri c 470Cfici,L.01112/2TIP coll1 2 3 4 5 6 80- 75- 70- 65- 6C- 55. 5C- 45- 4C- 3.5- 3C- 25- 2C- 15- 1CflawScores:Percen-tiles:T Score: ^•General-Seit norms are not available for grades 2-4.Note: T scores falling in the shaded area (i.e.. T scores of 50 or above) represent above average self-concept: however, becauseof the skewed distribution of the scores, T scores above 50 are not readily interpretable.8075706560555045403530252015105215See Chapter 2 of the Manual for step-by-step directions for calculatingCONTROL SCORE 1:^ CONTROL SCORE 2:^ Control Scores.8Total^ TotalAbsolute  8^SignedValue^ ValueItem55^Item71^ Item35^Item7Control Score CalculationControl Score 1:^ Control Score 2:Inconsistency on Consistency onCorrelated Item Pairs^ Uncorrelated Item PairsItem3^Item48^ Item 10^Item2Item38^Item54^ Item15^Item58Control Score 3 - Noncontingent SummaryWrite the values of Control Score 2 and Control Score 1 in the appropri-ate blanks below. Subtract Control Score 1 from Control Score 2. Writethe result in the blank labeled Control Score 3.Control Score 2^Control Score 1^CONTROL SCORE 3Control Scores 4 and 5: Before entering the item values, reverse thedirection of the scores so that 1 = True, 2 = Mostly True, 3 SometimesFalse/Sometimes True, 4 = Mostly False, and 5 = False.Item44^Item69^ Item5^Item16^ x 8Itern17^ PhysicalItem41 - Item57^ Item43- Item19-^ = - ^ =   x 8 - ^ _Item43^Item59^ Item16^Item5^ Item21^ Peers^ x 8 ^ -Item48 Item56^ Item24 Item62^ Item33^ ReadItem22^Item46^ Item7^Item20   x 8 ^ -Item6^ MathItem19^Item26^ Item58 Item13   x 8 = ^ - ^Item23^ SchoolItem18^Item49^ Item68 Item7   x 8   - ^Item30^ Appearancettem9^Item71^ Item31 Item26^ x 8Item12^-^ParentsItem3 Item56^ Item54 Item19^ x 8Item65^ ReadItem14 Item28^ Item36^Item19^ x 8 = ^Item75^ MathItem50 Itern11^ Item66 Item2^ x 8 = -^Item27 Item59^ Item2^Item24^ Item47^ SchoolItem15^Item22^ Item38^Item10Item60^Item69^ Item52^Item24Remit^Item25^- ^Item13^Item26Item35^Item51^ Item9^Item64ttem2^Item16^ Item59^Item58CONTROL SCORE 4^CONTROL SCORE 5Negativity Bias Positivity BiasControl Score 6 - Individual Profile Variation:Calculate the standard deviation of the original seven scales (PhysicalAbilities. Physical Appearance. Peer Relations, Parent Relations.Reading. Math, and General-School).CONTROL SCORE 6216APPENDIX H: PEARSON PRODUCT MOMENT CORRELATIONS FOR ESLSTUDENT SELF-CONCEPT, TEACHER PERCEPTION,AND ACADEMIC ACHIEVEMENT (24 x 24 Matrix)217218Table H-1Pearson Product-Moment Correlations for ESL Student Self-Concept, Teacher Perception, and Academic Achievement PHYS^APPR^PEER^PRNT^READ^MATH^SCHL^GENLPHYS^1.00APPR .35**^1.00PEER^.44***^.49*** 1.00PRNT^-.15^.06^.26*^1.00READ .10 .42***^.38**^.19^1.00MATH^.08^.38**^.05 .00 .06^1.00SCHL .27*^.52* **^.21^.11^.43***^.33**^1.00GENL^.23*^.64***^.46***^.25*^.40***^.35**^.56*** 1.00NACD^.69***^.73***^.82***^.36**^.40***^.20^.43***^. 59***ACD .19^.60***^.30*^.14 .70* **^.68* **^.76***^.60***TOTSLF^.49***^.77***^.62***^.28*^.66***^.54***^.71***^.69***TPHYS^.52*** -.02^.19^.00^-.26*^.13^.02^- .12TAPPR^.02^.12 .00 .01^-.08 .10 .01^- .03TPEER^.14 .09^.46***^.01 .09^-.11^-.05 .05TPRNT^-.21^-.13 .08^.11^.31**^-.19^-.03^-.08TREAD^-.20^-.24*^-.10 .07 .16^-.23 .04^-.02TMATH^-.12^-.24^-.28*^-.18^-.26*^.29*^-.19^-.20TSCHL^-.14^-.28*^-.15^-.02^-.05^-.03^- .08^-.09TGENL^-.03 .03 .05^-.02 .12 .11 .10 .06TNACD^.27*^.02^.29*^.04^-.04^.01^-.01^-.08TACD^-.19^-.30*^-.21^-.05^-.05 .01^-.09^-.12TTOTSLF^.01^-.19 .01^-.01^-.05^.01^-.06^-.12RMARK^-.20^-.12^-.03^-.00^.05^.01^.03^.04MMARK^-.18^-.34**^-.25^-.08^-.17 .12^-.14^-.16Note. N=57 student ratings and 57 teacher ratingsCorrelations are rounded to two decimal places. Correlationaldirections involving academic achievement (RMARK and MMARK) arereversed to parallel data collection on the SDQ-1. Correlations inboldface refer to student-teacher agreement on matching self-concepts.* Significant 'trend' at the .05 level** Significant at the .01 level*** Significant at the .001 level219Table H-2Pearson Product-Moment Correlations for ESL Student Self-Concept, Teacher Perception, and Academic Achievement  (continued...)^NACD^ACD^TOTSLFPHYSAPPRPEERPRNTREADMATHSCHLGENLNACD^1.00ACD .47*** 1.00TOTSLF^.82***^.89*** 1.00TPHYS^.30*^-.06^.12TAPPR^.06 .02 .04TPEER^.26*^-.03^.12TPRNT^-.09 .04 -.02TREAD^-.20^-.03^-.12TMATH^-.30*^-.05^- .19TSCHL^-.23*^-.07^-.16TGENL^.01 .16 .11TNACD^.24*^-.02^.11TACD^-.29*^-.05^- .18TTOTSLF -.07 -.05^- .07RMARK^-.15^.04^-.05MMARK^-.32**^-.08^-.22Note. N=57 student ratings and 57 teacher ratingsCorrelations are rounded to two decimal places. Correlationaldirections involving academic achievement (RMARK and MMARK) arereversed to parallel data collection on the SDQ-1. Correlations inboldface refer to student-teacher agreement on matching self-concepts.* Significant 'trend' at the .05 level** Significant at the .01 level*** Significant at the .001 level220Table H-3Pearson Product-Moment Correlations for ESL Student Self-Concept, Teacher Perception, and Academic Achievement (continued...)TPHYS TAPPR TPEER TPRNT TREAD TMATH TSCHL TGENLPHYSAPPRPEERPRNTREADMATHSCHLGENLNACDACDTOTSLF.29* 1.00.25* .48*** 1.00-.15 .10 .37** 1.00.07 .23* .19 .38** 1.00.30 .29* .04 .12 .31**.29* .47*** .30* .34** .79***.24* .70*** .53*** .44*** .61***.67*** .71*** .77*** .39*** .30*.26* .39*** .21 .33** .84***.51*** .62*** .53*** .42*** .71***-.03 .16 .27* .32** .64***.25* .17 .15 -.21 .54***TPHYS^1.00TAPPRTPEERTPRNTTREADTMATHTSCHLTGENLTNACDTACDTTOTSLFRMARKMMARK1.00.59*** 1.00.48***^.79*** 1.00.32**^.53* **^.70***.74***^.94***^.74***.66***^.89***^.85***.25*^.58***^.48***.62***^.55***^.39***Note. N=57 student ratings and 57 teacher ratingsCorrelations are rounded to two decimal places. Correlationaldirections involving academic achievement (RMARK and MMARK) arereversed^to parallel data collection on the SDQ-1. Correlations inboldface^refer to student-teacher agreement on matching self-concepts.Significant 'trend' at the .05 level** Significant at the .01 level*** Significant at the .001 level221Table H-4Pearson Product-Moment Correlations for ESL Student Self-Concept, ieachriDiamis,Acjaievenant (continued. . .)TNACD^TACD^TTOTSLF RMARK^MMARKPHYSAPPRPEERPRNTREADMATHSCHLGENLNACDACDTOTSLFTPHYSTAPPRTPEERTPRNTTREADTMATHTSCHLTGENLTNACD^1.00TACD .45*** 1.00TTOTSLF^.80***^.89*** 1.00RMARK^.23*^.58***^.51*** 1.00MMARK .31**^.68***^.61***^.46*** 1.00Note. N=57 student ratings and 57 teacher ratingsCorrelations are rounded to two decimal places. Correlationaldirections involving academic achievement (RMARK and MMARK) arereversed to parallel data collection on the SDQ-1. Correlations inboldface refer to student-teacher agreement on matching self-concepts.* Significant 'trend' at the .05 level** Significant at the .01 level*** Significant at the .001 level


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items