USF Libraries
USF Digital Collections

Idiom comprehension in bilingual and monolingual adolescents

MISSING IMAGE

Material Information

Title:
Idiom comprehension in bilingual and monolingual adolescents
Physical Description:
Book
Language:
English
Creator:
Fusté-Herrmann, Belinda
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Spanish
Reading comprehension
Lexical depth
Comprehension monitoring
Polysemy
Dissertations, Academic -- Communication Sciences & Disorders -- Doctoral -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: A majority of Latino adolescents are reading below a proficient level, according to federal data, and there is a significant gap between overall reading proficiency of Latino and non-Latino, Caucasian adolescents. The purpose of this study was to investigate the linguistic underpinnings of Latino students' text comprehension. A positive relationship appears to exist between idiom comprehension and academic achievement, as well as idiom comprehension and reading comprehension, in typically developing, monolingual adolescents. Since reading comprehension and idiom comprehension share many of the same linguistic processes, idiom comprehension may provide a unique perspective for investigating Latino adolescents' reading comprehension.Using the Global Elaboration Model (GEM, Levorato, Nesi, & Cacciari, 2004) as the conceptual framework, the present study examined the relationship between idiom comprehension and reading comprehension with a population that had not been studied in this manner: bilingual (Spanish-English) adolescents in West Central Florida and their monolingual (English-only) peers. The GEM posits that idiom comprehension develops in tandem with other linguistic development requiring inferencing ability; and that idiom x comprehension ability can be predicted by reading comprehension ability. The present research design included the evaluation of idiomatic familiarity, semantic transparency, and contextual support, as well as three other linguistic measures: a) a reading comprehension task, b) an error detection task, and c) a synonym task.Results indicated that the three linguistic measures predicted 33% of the variance in idiom comprehension accuracy; and error detection was the strongest predictor of idiom comprehension accuracy. Furthermore, monolinguals outperformed bilinguals on all measures. The synonym task, a measure of lexical depth, best predicted language group membership. There was a three-way interaction among idiomatic familiarity, semantic transparency, and contextual support; and a three-way interaction among familiarity, transparency, and language group. Lastly, the three linguistic measures significantly predicted the bilinguals' amount of English experience, with qualitative differences emerging between sequential and simultaneous language learners. Findings lend support to the psychological reality of the GEM and provide insight into the linguistic foundations of reading comprehension in Spanish-English bilinguals.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2008.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Belinda Fusté-Herrmann.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 276 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001990962
oclc - 311509818
usfldc doi - E14-SFE0002352
usfldc handle - e14.2352
System ID:
SFS0026670:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001990962
003 fts
005 20090302133554.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 090302s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002352
035
(OCoLC)311509818
040
FHM
c FHM
049
FHMM
090
RC423 (Online)
1 100
Fust-Herrmann, Belinda.
0 245
Idiom comprehension in bilingual and monolingual adolescents
h [electronic resource] /
by Belinda Fust-Herrmann.
260
[Tampa, Fla] :
b University of South Florida,
2008.
500
Title from PDF of title page.
Document formatted into pages; contains 276 pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2008.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
520
ABSTRACT: A majority of Latino adolescents are reading below a proficient level, according to federal data, and there is a significant gap between overall reading proficiency of Latino and non-Latino, Caucasian adolescents. The purpose of this study was to investigate the linguistic underpinnings of Latino students' text comprehension. A positive relationship appears to exist between idiom comprehension and academic achievement, as well as idiom comprehension and reading comprehension, in typically developing, monolingual adolescents. Since reading comprehension and idiom comprehension share many of the same linguistic processes, idiom comprehension may provide a unique perspective for investigating Latino adolescents' reading comprehension.Using the Global Elaboration Model (GEM, Levorato, Nesi, & Cacciari, 2004) as the conceptual framework, the present study examined the relationship between idiom comprehension and reading comprehension with a population that had not been studied in this manner: bilingual (Spanish-English) adolescents in West Central Florida and their monolingual (English-only) peers. The GEM posits that idiom comprehension develops in tandem with other linguistic development requiring inferencing ability; and that idiom x comprehension ability can be predicted by reading comprehension ability. The present research design included the evaluation of idiomatic familiarity, semantic transparency, and contextual support, as well as three other linguistic measures: a) a reading comprehension task, b) an error detection task, and c) a synonym task.Results indicated that the three linguistic measures predicted 33% of the variance in idiom comprehension accuracy; and error detection was the strongest predictor of idiom comprehension accuracy. Furthermore, monolinguals outperformed bilinguals on all measures. The synonym task, a measure of lexical depth, best predicted language group membership. There was a three-way interaction among idiomatic familiarity, semantic transparency, and contextual support; and a three-way interaction among familiarity, transparency, and language group. Lastly, the three linguistic measures significantly predicted the bilinguals' amount of English experience, with qualitative differences emerging between sequential and simultaneous language learners. Findings lend support to the psychological reality of the GEM and provide insight into the linguistic foundations of reading comprehension in Spanish-English bilinguals.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor: Elaine R. Silliman, Ph.D.
653
Spanish
Reading comprehension
Lexical depth
Comprehension monitoring
Polysemy
690
Dissertations, Academic
z USF
x Communication Sciences & Disorders
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2352



PAGE 1

Idiom Comprehension In Bilingua l And Monolingual Adolescents by Belinda Fust-Herrmann A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Communicati on Sciences & Disorders College of Arts and Sciences University of South Florida Major Professor: Elaine R. Silliman, Ph.D. Judith Becker Bryant, Ph.D. Stefan Frisch, Ph.D. Lisa Lpez, Ph.D. Nathan Maxfield, Ph.D. Date of Approval: February 21, 2008 Keywords: spanish, reading comprehension, lexical depth, compre hension monitoring, polysemy, global elaboration model, figurative language Copyright 2008 Belinda Fust-Herrmann

PAGE 2

i Table of Contents List of Tables vi List of Figures viii Abstract ix Chapter One: Introduction 1 Idioms: Relevance, Comprehension Factors, and Models 4 The Pervasiveness of Idioms in Classrooms 5 Three Major Factors Affecting Idiom Comprehension: Semantic Transparency, Familiarity, a nd Context 6 Semantic Transparency 6 Familiarity 7 Context 8 Models of Idiom Comprehension 8 Early Hypotheses 9 Linguistic Representations Hypotheses 9 From Linguistic Hypotheses to Models of Language Processing 10 Development of Oral Idiom Comprehension: Monolingual and Cross-Linguistic Research 16 Typically Developing Children and Children with Linguistic/Cognitive Impairments 16 Typically Developing: Gradual Emergence 16 Idiom Comprehension in children with Linguistic/Cognitive Impairments 19 Idiom Comprehension in Second-Langua ge Learners 21 Adult Studies 22 Limitations of second language learner studies 24 Reading Comprehension, Idiom Comprehension, and The Global Elaboration Model 25 Idioms and Text Comprehension 25 Initial Studies 25 Longitudinal Investigation 26 Cross-sectional research on the Global Elaboration Model 28

PAGE 3

ii Shared Processes in Reading Comprehension and Idiom Comprehension 30 The C-I Model 30 Comprehension Monitoring and Error Detection 31 Processing Abilities Underlying Figurative Comprehension 33 The Model and Research Questions 34 Purpose 35 Principles 36 Principle 1 36 Principle 2 37 Principle 3 37 Principle 4 37 Research Questions 38 Chapter Two: Pilot Study 40 Method: Part 1 40 Participants 40 Materials: Development of the Pilot Idiom Measure 43 Familiarity 43 Semantic Transparency 43 Context 44 Procedure 44 Results: Statistical Analyses 46 Total Scores 46 Interactions 48 Main Effects 54 Results: Qualitative Analyses 56 Error Analysis 56 Item Analysis 58 Discussion 58 Method: Part 2 60 Participants: Adolescent Pilot Study on Familiarity Ratings 60 Materials: Development of the Familiarity Rating Form 61 Procedure 62 Data Analysis 63 Results 67 Discussion 71 Construction of the Final Idiom Compre hension Measure 71

PAGE 4

iii Balancing Items and Syllable Length 71 Balancing Familiarity Ratings 75 Balancing Response Accuracy 76 Chapter Three: Method 78 Experimental Design 78 Participants 78 Sample and School Characteristics 78 Sample Size and Characteristics 78 Sample School Characteristics 79 Recruitment 80 Inclusion Criteria 80 Procedure 81 Measures 81 Word Attack 81 Idiom Comprehension Measur e 82 Passage Comprehension 83 Comprehension Monitoring Task 83 Multiple Meanings Vocabulary: Synonyms 85 Student Language History Questionnaire 85 Administration and Scoring 86 Word Attack 87 Idiom Comprehension Measur e 87 Passage Comprehension 88 Monitoring Comprehension 88 Multiple Meanings Vocabulary: Synonyms 89 Student Language Questionnaire 90 Chapter Four: Results 91 Major Aims 91 Testing the Model: Aim 1 92 Collapsing Across Groups: Simultaneous Multiple Regression 92 Contribution of Language Group Variance: Simultaneous Multiple Regression 94 Language Group as Criterion Variable: Simultaneous Multiple Regression 96 Bilinguals Only: Simultaneous Multiple Regression 97 Monolinguals Only: Simultaneous Multiple Regression 99 Summary: Aim 1 Findings 101 Cross-Language Group Performance on Idiomatic Comprehension: Aim 2 101 Addressing Aim 2 104

PAGE 5

iv Interactions 105 Main Effects 109 Results of Additional Hypotheses 110 Summary: Aim 2 Findings 113 Effects of Within-Group Bilingual Proficiency: Aim 3 114 Descriptive Summary of the Language Experience Questionnaire 114 Quantitative Analysis of Language Experience 115 Descriptive Analysis of the Simultaneous and Sequential Learners 117 Summary: Aim 3 Findings 119 Overall Summary of Major Findings 119 Chapter Five: Discussion 122 Aim 1: Effects of Reading Co mprehension, Error Detection, & Synonym Performance on Idiom Comprehension 122 Error Detection 123 Reading Comprehension 123 Synonym Task 125 Summary 128 Aim 2: Idiom Comprehension Outcomes 129 The language experience hypothesis and beyond 129 The Global Elaboration Model 131 Control of prior knowledge 131 The relationship betwee n reading comprehension and idiom comprehension 132 Aim 3: Effects of Within-G roup Bilingual Proficiency on Performance 133 Potential Study Limitations 135 Taking English language proficiency into account 136 Inclusion of additi onal spoken language and cross-linguistic assessments of reading comprehension 137 Sample size and characteristics 138 Reliability of the idiom comprehension measure 138 Directions for Future Study 139 Conclusion 142 References 145 Appendices A R 159 Appendix A: USF Signage for Pilot St udy Participants 160 Appendix B: Non-significant Findings for the Pilot Study 162

PAGE 6

v Appendix C: Pilot Study Idiom Compre hension Measure 164 Appendix D: USF English Consent Forms 193 Appendix E: Pilot Study Synopsis 196 Appendix F: Polk County Public Schools 199 Appendix G: Polk County School Public Schools Consent Form (Pilot Study) 201 Appendix H: Familiarity Forms 204 Appendix I: New Idiom Comprehens ion Measure 209 Appendix J: Polk County Public School s Approval Letters 242 Appendix K: School District of Hillsborough County Approval 246 Appendix L: IRB Approved Consent Forms (English and Spanish) 248 Appendix M: Student Assent Form 253 Appendix N: Inclusion Questionnaire 256 Appendix O: Student Language Histor y Questionnaire 258 Appendix P: Error Detection Paradigm 264 Appendix Q: Percentage of Item Accuracy on the Synonym Task 271 Appendix R: Non-significant ANOVA and t-test Findings 274 Footnote 276 About the Author End Page

PAGE 7

vi List of Tables Table 2.1: USF Undergraduate Participant Information 42 Table 2.2: Total Scores for Idiom Accuracy as a Function of Language Group 46 Table 2.3: Accuracy Scores for Idiom Conditions as a Function of Language Group 47 Table 2.4: ANOVA Results for the Accuracy Scores on The Pilot Idiom Measure 48 Table 2.5: Error Analysis of All Item s in All Conditions on the Pilot Idiom Measure by Language Group and Error Type 56 Table 2.6: Familiarity Ratings in Descending Order by Idiom Category 64 Table 2.7: Syllable Counts for All Idioms 72 Table 2.8: Syllable, Familiarity, and Condition Match: Form A and Form B 74 Table 4.1: Means, Standard Deviat ions, and Intercorrelations for Participants (N= 62) Idiom Comprehension and Predictor Variables 93 Table 4.2: Simultaneous Regression Analysis Summary for Three Variables Predicting Idiom Comprehension 94 Table 4.3: Means, Standard Deviat ions, and Intercorrelations for Participants Idiom Comprehension and Predictor Variables Including Language Group 95 Table 4.4: Simultaneous Regression Analysis Summary for Four Variables Predicting Idiom Comprehension 95

PAGE 8

vii Table 4.5: Means, Standard Deviat ions, and Intercorrelations for Participants Language Group and Pr edictor Variables Including Idiom Comprehension 96 Table 4.6: Simultaneous Regression Analysis Summary for Four Variables Predicting Language Group 97 Table 4.7: Means, Standard Deviat ions, and Intercorrelations for Bilingual Participants Idiom Comprehension Accuracy and Three Linguistic Predictor Variables 98 Table 4.8: Simultaneous Regression Analysis Summary for Three Variables Predicting Bilinguals Idiom Compre hension Accuracy 99 Table 4.9: Means, Standard Deviat ions, and Intercorrelations for Monolingual Participants Idiom Comprehension Accuracy and Three Linguistic Predictor Variables 100 Table 4.10: Simultaneous Regression Analysis Summary for Three Variables Predicting Monolinguals Idiom Co mprehension Accuracy 100 Table 4.11: Means and Standard Deviations for the 40 Items on the Idiom Comprehension Measure 102 Table 4.12: ANOVA Results for th e Accuracy Scores on the Idiom Comprehension Measure 104 Table 4.13: Idiom Comprehension Scores for Monolinguals and Bilinguals in the Eight Conditions 106 Table 4.14: Error Analysis of All Items on the Idiom Comprehension Measure by Language Group (Monoli ngual (M), Bilingual (B)) and Error Type 112 Table 4.15: Means, Standard Deviat ions, and Intercorrelations for Bilingual Participants Questionnaire Results and Four Predictor Variables 117 Table 4.16: Simultaneous Regression Analysis Summary of Four Variables Predicting Questionnaire Results 117 Table 4.17: Performance Scores on All Four Measures for Simultaneous and Sequential Language Learners 118

PAGE 9

viii List of Figures Figure 1 Local and Global Coherence of an Idiomatic Expression 13 Figure 2 Construction-Integration Ph ases of Text Comprehension Based on Kintsch (1998) 31 Figure 3 The Four Principles of Literal and Figurative Reading Comprehension 36 Figure 4 The statistically significa nt three-way interactions among accuracy scores in the following conditions and each of their two levels: familiarity, tran sparency, and context, with language groups collapsed. 49 Figure 5 Confidence intervals of familiarity, semantic transparency, and context variables displayed in mean accuracy values and illustrated with error bars. 50 Figure 6 Confidence intervals of the familiarity and semantic transparency variables displayed in mean accuracy values and illustrated with error bars. 51 Figure 7 Confidence intervals of familiarity and context variables displayed in mean accuracy values and illustrated with error bars. 52 Figure 8 Confidence intervals of semantic transparency and context variables displayed in mean accuracy values and illustrated with error bars. 53 Figure 9 Boxplot of familiarity rati ngs (ranging from 1-3) as a function of familiar and unfamiliar idioms. 70 Figure 10 The statistically signifi cant three-way interaction among accuracy scores in the eight combinations of idiomatic conditions: familiarity, transparency, and context, with language groups collapsed. 107

PAGE 10

ix Idiom Comprehension in Bilingua l and Monolingual Adolescents Belinda Fust-Herrmann ABSTRACT A majority of Latino adolescents are read ing below a proficient level, according to federal data, and there is a significan t gap between overall r eading proficiency of Latino and non-Latino, Caucasian adolescen ts. The purpose of this study was to investigate the linguistic underpinnings of Latino students te xt comprehension. A positive relationship appears to exist betw een idiom comprehension and academic achievement, as well as idiom comprehensi on and reading comprehe nsion, in typically developing, monolingual adolescents. Si nce reading comprehension and idiom comprehension share many of the same lingui stic processes, idiom comprehension may provide a unique perspective for investigating Latino adolescents reading comprehension. Using the Global Elaboration Model (GEM Levorato, Nesi, & Cacciari, 2004) as the conceptual framework, the present study examined the relationship between idiom comprehension and reading comprehension with a population that had not been studied in this manner: bilingual (Spanish-English) adolescents in We st Central Florida and their monolingual (English-only) peers. The GEM pos its that idiom comprehension develops in tandem with other linguistic development re quiring inferencing abil ity; and that idiom

PAGE 11

x comprehension ability can be predicted by reading comprehension ability. The present research design included the evaluation of idiomatic familiarity, semantic transparency, and contextual support, as well as three other linguistic measures: a) a reading comprehension task, b) an error detection task, and c) a synonym task. Results indicated that the three linguisti c measures predicted 33% of the variance in idiom comprehension accuracy; and error detection was the str ongest predictor of idiom comprehension accuracy. Furthermore, monolinguals outperformed bilinguals on all measures. The synonym task, a measure of lexical depth, best predicted language group membership. There was a three-way interaction among idiomatic familiarity, semantic transparency, and contextual support; and a three-wa y interaction among familiarity, transparency, and language group. Lastly, the three linguistic measures significantly predicted the bi linguals amount of English ex perience, with qualitative differences emerging between sequential and simultaneous language learners. Findings lend support to the psychological reality of the GEM and provide insight into the linguistic foundations of reading compre hension in Spanish-English bilinguals.

PAGE 12

1 Chapter One Introduction The Latino population is the fastest growi ng population in the United States. This population is expected to increase from 35.3 million in 2000 to 60.4 by 2020 (Suro et al., 2005). Latinos now represent 19 percent of the U.S. school-age population, an increase from 12.7 percent from 1993 (Kohler & Lazar n, 2007). Latino English language learners (ELLs) comprise the largest group of ELLs (K oelsch, 2006). Federal data on the bilingual school-age population demonstr ate that a gap exists in English reading proficiency between Latino students and Ca ucasian, non-Latino students. Fo r example, results from the 2005 National Assessment of Educatio nal Progress (NAEP) (United States Department of Education (USDOE), National Center for Educational Statistics (NCES), 2006) showed a 25 point score gap at grad e 8 between Latino and Caucasian, non-Latino students. Although this gap has narrowed some what since 2003 (i.e., a 27 point score gap existed in 2003), the breadth of the gap remains, and continues to maintain itself in the 2007 Reading Report Card (USDOE, NCES The Nations Report Card, 2007). Furthermore, the Nations Report Card (U SDOE, NCES, 2007) showed a 21 point gap between Latinos and non-Latino Caucasians in grade 12 in 2005, up from a 20 point gap in 2002 and a 19 point gap in 1992. Nationw ide, according to the 2005 NAEP results,

PAGE 13

2 only 20 percent of Latinos in grade 12 are reading at a proficie nt level. In contrast, 43 percent of non-Latino Caucasians in grade 12 are reading at a prof icient level (USDOE, NCES, 2007). The proficient achievement level is described in part as being able to show an overall understanding of the text, including infere ntial as well as literal information (USDOE, NCES, The Nations Report Card, 2006, p. 29). In Florida, the 2007 NAEP reading scores indicated that only 26 percent of Latino students in the 8th grade were able to read at a pr oficient level (USDOE, NCES, Nations Report Card, 2007). One example of a critical reading activ ity on this assessment for grade 8 was to read a passage describing new immigrants experiences at Ellis Island during the 19th century. Following the passage, student s were to write a response to the following question: What two experiences might have c aused the new immigrants to say that they felt like cattle? This sample question underscores the necessity of students ability to make accurate literal and figurative inferences in order to achieve at the proficient level at grade 8. The reading achievement of Latino students whose first language is not English is correlated with diminished academic skills be ginning as early as grade 3 (Jimnez, 1994). Unfortunately, this negative relationship continues throughout the academic careers of these second language learners (Jimnez, 1994). Because of the No Child Left Behind Act (NCLB, 2002), many states (such as Flor ida) are requiring all students to pass a standardized reading compre hension measure as an exit requirement for high school graduation. Thus, negative corr elations between Latino students whose first language is not English and their reading achievement sugge sts that Latino adolescents may be at risk for academic failure and subsequent high school drop out. With this type of

PAGE 14

3 unresolved disparity in read ing achievement between Latinos and Caucasian non-Latinos, Latino adolescents may drop out of high school at a higher rate than their Caucasian nonLatino counterparts. In fact, in 2004, of all high school drop outs ages 16 to 24, 23.5 percent were Latino Americans compared to 6.8 percent who were Caucasian non-Latino (USDOE, NCES, 2006). Overall, only 53 percen t of Latinos in Florida (and nationwide) graduate from high school (Alliance for Education, 2007). Other NCES (USDOE, 2006) data show that the high school drop-out rate of Latino students born outside of the United States also remains higher ( 38.4 percent) than those Latinos who were born in the United States (first genera tion = 14.7 percent and second generation or higher = 13.7 percent). Taken together, these data are evidence of how low reading proficiency, when considered as the ability to in fer and integrate, puts the bi lingual adolescent population at risk for failing mandatory state assessment s, including those now required for high school graduation, and creates condi tions for not completing high school. Because of these factors, it is crucial to understand the la nguage processing skills necessary for these bilingual students to read mo re proficiently. One domain th at provides a unique vantage point for examining the underpinnings of te xt comprehension is idiom comprehension. Idioms, a type of non-literal, figurative language, such as spill the beans are pervasive in classroom discourse a nd academic text books (Nippold, 1991). In monolingual English-speaking children, a posi tive relationship appear s to exist between idiom comprehension and the level of read ing comprehension at age 9 years (Cain, Oakhill, & Lemmon, 2005). A similar re lationship was found between idiom comprehension and overall academic achievement in monolingual English-speaking adolescents (Nippold & Martin, 1989). A need currently exis ts to explore whether the

PAGE 15

4 same relationship holds between idiom co mprehension and reading proficiency in bilingual (Spanish-English) a dolescents. Furthermore, read ing comprehension and idiom comprehension appear to share similar cognitiv e-linguistic processes. Thus, insight into idiom comprehension may help to illu minate the underpinnings of reading comprehension as an inferential pr ocess in bilingual adolescents. In this chapter the research literatu re on idiom comprehension is reviewed. Firstly, idioms are defined and the factors th at affect idiom comprehension are discussed. Secondly, the theoretical frameworks for idiom comprehension in monolinguals are explored. Then, the development of idiom comprehension is reviewed in monolingual and cross-linguistic populations who are either typically-developing or cognitively/linguistica lly impaired, followed by an appraisal of the literature on adult bilinguals and idiom comprehension. Give n this background information, possible relationships are elaborated on betw een idiom comprehension and reading comprehension with a focus on shared cogni tive and linguistic underpinnings. Then the theoretical model developed for this study is presented. In the final section, three research questions associated wi th the study are outlined. Idioms: Relevance, Comprehension Factors, and Models Idioms are a subtype of the broader ca tegory of nonliteral, figurative language. Figurative language encompasses other nonlite ral forms such as similes, metaphors, sarcasm, irony, indirect reque sts, and hints (Holtgraves, 2005). An idiom is a meaning where the sums meaning is different from th at of the parts (Abkarian, Jones, & West, 1992; Johnson, Johnson, & Schlichting, 2004). Idio ms may be interp reted differently from other figurative language, however. For example, similes are easier to understand

PAGE 16

5 due to their inherent inclusion of the words like or as which act as cues (Gentner, Bowdle, Wolff, & Boronat, 2001). Examples of similes include as bright as the sun and slow like a turtle In addition, a metaphor (e.g., she is a snake) seems to be processed like an analogy, which is not always a possibl e solution for idiom comprehension since connections may be more opaque (Gentner et al., 2001). Furthermore, jokes and sarcasm are based on implicit meanings and more so on pragmatic variables (such as winking), as seen in children on the autism spectrum who have difficulty w ith this type of figurative language due to decreased pragmatic skills (Norbury, 2004). The Pervasiveness of Idioms in Classrooms Idioms are pervasive in most languages, but can be language specific or language general. For example, some idioms are histori cally traceable with translations in several languages, while others have developed from more colloquial pasts. For example, the Spanish idiom, no hay Moros en la costa literally translates to there are no Moors on the coast. Figuratively, this idiom translates to the coast is clear in English, but anyone who knows the history between the Spanish Moor s and Spanish Catholics can interpret a deeper meaning. Examples of North American English idioms include chip on your shoulder, back seat driver, and I wash my hands of it One study of the pervasiveness of idioms found that an idiom occurred in approximately 6.7% of all sentences in three fr equently used reading texts in grades 3-8 (Nippold, 1991). Frequency of idiom usage incr eased through the grades with a range of 6% at grade 3 to 9.7% by grade 8 (Ni ppold, 1991). Lazar, Warr-Leeper, Nicholson, and Johnson (1989) similarly investigated idiom fr equency in discourse used in kindergarten through grade 8 classrooms. Of 5400 teacher utterances, 11% contained at least one

PAGE 17

6 idiom. This frequent use of idioms may be detrimental for ch ildren with language learning or cognitive impairments or thos e acquiring English as a second language. Whether idioms are spoken or written, at least three factors impact on idiom comprehension. Three Major Factors Affecting Idiom Co mprehension: Semantic Transparency, Familiarity, and Context Semantic transparency. Semantic transparency refers to the relative correspondence of an idioms literal and fi gurative meanings (Nippold & Taylor, 1995). A transparent idioms meaning matches closely with the image conjured up by that idiom. For example, the idiom, a piece of cake, may conjure up an enjoyable task. In contrast, an opaque idiom conjur es up an image that is not helpful in interpretation. For example, beat around the bush as a literal image has little to do with that idioms meaning (i.e., avoiding a topic of discussion). Semantic transparency can be viewed on a continuum. One end reflects a more superfic ial, literal corres pondence and the opposite end reflects a deeper, more elusive and figur ative correspondence. Previous studies have concluded that transparent idioms are genera lly easier to decipher than opaque idioms (Nippold & Taylor, 1995; Norbury, 2004). Another way to discuss the transparency of idioms is in terms of their decomposition (Glucksberg, 2001), with a more decompositional idiom the meaning of each word adds up to the holistic meaning. Thus each semantic part is more meaningful than meaningless. Furthermore, idioms that are decompositional are able to be modified, such as he broke the ice, she breaks the ice, after the ice was broken, etc. These modifications are possible since each part of the idiom is meaningful (Sprenger, Levelt,

PAGE 18

7 & Kempen, 2006). For instance, break is associated with to end and ice is associated with tension. Noncompositional idioms cannot survive the same alterations. One example is the noncompositional idiom on the fly which cannot be decomposed into on the flied (Glucksberg, 2001) In addition, decomposition ranges along a continuum. More decompositional idioms are likened to tr ansparent idioms, and less decompositional idioms are equated with opaque idioms. Familiarity. The frequency with which an idiom occurs in a language is often defined as familiarity (Nippold & Taylor, 1995) ; however, frequency and familiarity are both moderated by culture. Familiarity is relative and depends on such factors as geographical location, linguist ic background (including di alect), culture, and age (Nippold & Rudinski, 1993). It appears that idiom comprehe nsion is easier when an idiom is more familiar to someone because less conceptual analysis is required (Qualls & Harris, 1999). Exposure may play an important role in idiom comprehension since having more experience with idioms may make those idioms mo re salient (Norbury, 2004). Ultimately, more frequently used idioms may be more familiar. Glucksberg (2001) described idioms as a secret language and a language owned by a culture that one has to be steeped in. In other words, idioms vary in frequency and familiarity depending on variables like demographic characteristics and cultural and linguistic identification. Ortony, Turner, a nd Larson-Shapiro (1985) formulated the experience hypothesis, which postulated that individuals idiom comprehension was dependent on their meaningful exposure to idioms. Later, Qualls and Harris (1999) expanded this hypothesis into the differential language experi ential hypothesis to explain social and regional effects on idiom compre hension. For example, Qualls and Harris

PAGE 19

8 (1999) found evidence for this hypothesis when investigating idiom comprehension in African Americans living in the southern pa rt of the United States. Membership in a particular linguistic and cultural community wa s seen as an importa nt variable in the familiarity of idioms (Qualls & Harris, 1999). Context. Contextual cues are imperative for comprehension of unfamiliar idioms in either the oral or written modality, partic ularly if idioms are more opaque in nature (Qualls, OBrien, Blood, & Hammer, 2003). Id ioms presented orally are typically accompanied by both linguistic cues (e.g., surr ounding words) and extralinguistic cues, such as intonation, stress, gestures, facial e xpressions, and social context. The ability to exploit context becomes even more important when extralinguistic cues are absent, such as in reading, where only linguistic contextu al cues are available. Context appears to facilitate idiom comprehension more in ol der elementary school-age children and beyond (Levorato & Cacciari, 1992). Younger children (4to 5-years-old) may have difficulty exploiting the surrounding linguistic cont ext (Levorato, Nesi, & Cacciari, 2004). Models of Idiom Comprehension In the last several decades many resear chers have speculated about how idioms are interpreted. During the 1970s and early 1980s several hypotheses we re put forth with a focus on how idioms are stored and accessed in the lexicon. Then, in the 1990s, a shift occurred in the research lite rature with a new focus on how idioms were linguistically processed. The first hypotheses are elaborat ed on briefly followed by a discussion of subsequent linguistic processi ng models of idiom comprehension, specifically the model for this proposal.

PAGE 20

9 Early hypotheses In 1973 Bobrow and Bell create d the Idiom List Hypothesis. A main assumption was that, when idioms ar e first encountered in spoken or written language, the listener or reader tries to in terpret the idiom litera lly. When the literal meaning fails to make sense, the listener/ reader then accesses a mental idiom list, described as a sort of mental idiom dicti onary, in order to determine the figurative meaning (Searle, 1979). Subsequently, Swinney and Cutler (1979) challenged the existe nce of a mental idiom list. Instead, they argued that idioms were considered as long words; that is, idioms were stored along side other words in the le xicon, not separately. Furthermore, Swinney and Cutler (1979) proposed that the meanings of idioms were processed simultaneously as figurative and literal. Through this process, the figurative and lite ral meanings compete and the most appropriate interpretation wins. As an extension of Swinney and Cutle rs view, Gibbs (1980) also described idioms as being stored as long words in th e lexicon. Gibbs (1980), however, refuted the competition theory in favor of the Direct A ccess Theory. As the theorys name implies, the meanings of idioms were posited to be accessed directly and immediately, by-passing the literal meaning. In other words, the litera l meaning was not the default meaning of all idiomatic language comprehension. Linguistic representations hypotheses. A shift in conceptual frameworks occurred in the late 1980s in idiom comprehension study. The ideas of separate lexicons and of idioms being stored as long words were further challenged. One conjecture was that idioms were constructed by constituents or linguistic parts (Cacciari & Tabossi, 1988;

PAGE 21

10 Tabossi & Cacciari, 1988). These constituent meanings (both literal and figurative) were simultaneously activated within one lexicon. A similar perspective concurrently emerged, the Idiom Decompositionality Hypothesis (Gibbs & Nayak, 1989). This hypothe sis focused on the si gnificance of each constituent of an idiomatic phrase to create a meaningful phrase. In other words, the emphasis shifted to part-whole relationships an emphasis that continued throughout the 1990s. From linguistic hypotheses to models of language processing The focus of the linguistic processing of idiomatic parts to create a meaningful whole was extended in the Composition Model (Gibbs, 1991, 1994; Tabossi & Zardon, 1995). In the Composition Model, idiom comprehension involved decompositional analysis at the semantic, syntactic, and lexical level, just like the an alysis that occurred when any other phrase was encountered. Thus, Gibbs (1991) conjectured that not all idioms were noncompositional (e.g., kick the bucket). Instead, many idioms were decomposable or analyzable into their component parts (e.g., raining cats and dogs ). Decomposition is now described as semantic transparency. Around the same time as the developm ent of the Composition Model (Gibbs, 1991, 1994; Tabossi & Zardon, 1995), Levorato a nd Cacciari (1992) and Levorato (1993) proposed the Global Elaboration Model. A premise of this model is that idiom comprehension develops in parallel with general cognitive and linguistic development through childhood. In other words, there is no idiom-specific process developed for idiom comprehension. However, an exception was hypothesized. Opaque idioms, whose meanings do not match the images they conjure in a one-to-one correspondence, were

PAGE 22

11 learned via rote memory. For instance, the meaning of kick the bucket (i.e., to die) does not correspond with an image of someone literally kicking a bucket. Thus, this model encompasses both linguistic processing (of tr ansparent idioms) and lexicalization (of opaque idioms). Levorato et al. (2004) expl ained idiom comprehensi on through their expanded model of semantic analysis, the Global El aboration Model. The degree of an idioms semantic analyzability is contingent on the re lationship between the literal meaning of the idioms constituents and the idioms figurative meaning (Levorato, Roch, & Nesi, 2007). Semantic analysis is accomplished by analyz ing an idioms constituents (i.e., linguistic parts) since an idioms constituents must be individually understood to create local coherence and then connected to generate global coherence. Unlike literal text comprehension, idiom comprehension required in terpretation of the c onstituents literal and figurative meanings. The outcome was that a logical semantic representation had to be constructed from contextually appropriate meanings. These semantic representations were then integrated and compared with the speakers/writers intended meaning as conveyed in the idiomatic expres sion (Levorato et al., 2004). The psychological reality of the Global Elaboration Model was tested through several studies with monoli ngual Italian-speaking or E nglish-speaking, school-age children (Cain et al., 2005; Levorato et al ., 2004; Levorato et al., 2007). Typical sample sizes have generally ranged from 22 to 101 participants. In these studies, there was a correlation between the ease of analysis (i.e., of analyzing constituents) and ease of comprehension. For instance, transparent idioms with a more direct relationship between the individual meanings of constituents and the overall figurative meaning were easier to

PAGE 23

12 comprehend. The Global Elaboration Model pos ited that two processes are used when interpreting unknown idioms: semantic analysis and inference from context. Semantic analysis is only beneficial for transparent idioms and inference generation can only occur if there is context present. Moreover, lo cal coherence occurs when each constituents appropriate meaning is accessed and understood. Global coherence results in one of two situations: when the meanings of local constituents directly corresponds with the figurative meaning (as in the case of transpar ent idioms) or when context and intended meaning are integrated with these constitu ent meanings to interpret the figurative meaning (in other words, opaque, unknown idioms). The process of interpreting an opaque idiom is illustrated in Figure 1. The figure illustrates the Global Elaboration

PAGE 24

13 Model at work. Figure 1. Local and global coherence of an idiomatic expression.

PAGE 25

14 It should be noted that one recent study did not find support for the Global Elaboration Model. Crutchley (2007) used th e Non-literal Comprehe nsion Subtest of the Assessment of Comprehension and Expressi on (ACE 6-11), which is made up of a forced-choice picture task and a writte n task, to analyze the responses of 789 monolingual English-speaking children, ages 6 11-years. The sample consisted of 121 6-year-olds, 136 7-year-olds, 136 8-year-olds, 133 9-year-olds, 145 10-year-olds, and 128 11-year-olds. Children were asked to choose one of four pictures that corresponded with a given idiomatic sentence for the first eight items. For the second set of eight items, children chose the correct interpretation of an idiomatic sentence from a set of four written choices (which the examiner also re ad aloud). The idiomatic expressions used were verb + particle constructions, such as look up or throw away (the particle portion is bolded). Evidence for the Global Elabor ation Model did not emerge since no developmental trend for the literal and then figurative interpretations of the items was found. Following the Global Elaboration Mode l, younger participan ts (6-year-olds) should have applied a literal strategy (inter preting idioms constituent by constituent) in idiom interpretation and the ol der participants (11-year-old s) should have exploited the context for more figurative comprehension. Crutchleys (2007) interpretation of the findings was that children were unprepared to tolerate vi olations of syntactic stru cture in the pursuit of an interpretation that prioritizes the semantics of individual words; rather, they seem to assume that the verb has a non-literal interpretation that is unavailable to them, and choose a distracter that seem s plausible in the context (p. 218). Instead, Crutchley (2007) offered a needs-only analysis (p. 218) hypothesis; that is children break down

PAGE 26

15 language chunks into their cons tituent parts only as needed. However, there are at least seven potential criticisms of this hypothesis. First, all of the verb + particle constructions may have been at least somewhat familiar to both the younger and older particip ants. If participants were familiar with these constructions they may be lexicalized and stored in the mental lexicon in a way similar to the lexicalization and storage of familiar opaque idioms (e.g., kick the bucket). To ascertain whether this process occurred, novel verb + particle constructions would need to be considered. Second, the need-only hypothesis is not at complete odds with the Global Elaboration Model as Crutchle y (2007) states. The Crutchley (2007) data demonstrated that children processed idioma tic language holistically (p. 219) when they were confronted with familiar, opaque idioms (Cain et al., 2005; Levorato et al., 2004). Third, Crutchley (2007) did find that pe rformance improved significantly across age groups, particularly in the written task where children lacked pictorial support. Fourth, as Crutchley points out, particle verbs are non-decomposable idioms and not syntactically frozen like some idioms, which are the type that ge nerally require more mature figurative language competence to interp ret. Perhaps particle verbs are easier to process and digest; thus, even the younger pa rticipants were able to avoid a literal interpretation route. Fifth, all of the items in the study were pr esented within a sup portive, short-story context. It may have been the case that ch ildren were biased to ward producing more figurative responses because of the presence of contextual cues. It w ould be interesting to run the same experiment with the items pl aced both inand out-of-context in order to

PAGE 27

16 assess whether semantic analysis would differ between the two conditions. A sixth criticism is that there were only seven items in each condition (i.e., seven picture tasks and seven written tasks), which may have led to unreliable results. It may be worthwhile to include more items in a future study. Finally, the participants reading compre hension abilities were not taken into consideration in the Crutchley (2007) design. The Global Elaboration Model posits that literal interpretation is preferred when te xt processing abilitie s are weak (Levorato, Roch, & Nesi, 2007, p. 491). Thus, children with poor text comprehension abilities would probably rely on literal interpretation, rath er than figurative in terpretation, of unknown idioms. Development of Oral Idiom Comprehension: Monolingual and CrossLinguistic Research Typically Developing Children and Children with Linguistic/Cognitive Impairments Idiom comprehension has been researched extensively with monolingual English, Italian and French speakers, primarily child ren. A developmental trajectory of idiom comprehension has been identified in th ese typically developi ng monolinguals. Typically developing: Gradual emergence. There appears to be a developmental trend, or gradual emergence, of idiom co mprehension in monolinguals (Levorato & Cacciari, 1995). However, the depth and breadth of idiom comprehension continues throughout adolescence and acr oss the lifespan (Nippold, Uhden, & Schwartz, 1997). Levorato and Cacciari (1995) found that Ita lian-speaking, monolingual children in grade two ( M = 7; 10 years) were more literally oriented than children in grade four ( M = 9; 11 years), who were more idiomatically orie nted. Young children typically interpreted

PAGE 28

17 idioms in a literal manner with a developmental trend towards more figurative comprehension. Levorato and Cacciari (1995) accounted for th e shift from more literal to more figurative interpretation as due to childrens initial processing of constituents in a bit by bit fashion, then developing the ability to infer figurative meaning holistically from written or spoken context. Thus, children appe ared to access the lite ral meanings of the local constituents of idioms without integr ating these meanings to create a holistic figurative meaning. Therefore, with further cognitive and linguistic development, local coherence eventually allowed for global representation of the text meaning in permitting children to exploit the linguistic contex t for more accurate and appropriate idiom comprehension. Similarly, Abkarian et al. (1992) found that, in a picture choice ta sk of oral idiom comprehension, English-speaking monolingua l 6-year-olds provided more figurative rationales for their choices than did thei r younger counterparts (3;6-6;0 years-old). Moreover, idioms were most ra pidly acquired between the ages of 7-to 11-years (Johnson et al., 2004). Interestingly, this is approximately the same age when a shift occurs in both speaking and writing from more oral (everyda y) language use to more literate language use in English-speaking monolinguals (Scott, 2002). Using mental imagery as a strategy to assess oral idiom comprehension (e.g., similar to a think-aloud process of on-line, verbalized problem-solving), Nippold and Duthie (2003) found that mental imagery fo r idioms followed a similar developmental trend as comprehension. They pr esented 40 preadolescents (mean age, 12; 3) with highly familiar idioms. Half of the idioms (10 idioms) were opaque and half (the remaining 10)

PAGE 29

18 were transparent. After giving examples of how to verbalize mental imagery of idioms, the participants were asked to describe th eir mental images in writing when presented with these idioms. Mental images were code d as irrelevant, litera l, or figurative. The responses of the 40 preadolescents were then compared with the responses of 40 adults (Mean Age = 27). The preadoles cents mental images tended to be less sophisticated, more literal, and reflective of only partial understanding. In contrast, adults mental images tended to be mo re figurative. Nippold and Duthie (2003) concluded from these two studies that the na ture of mental images may serve as a barometer of idiom comprehension depth. Moreove r, this developmental trend of increasingly sophisticated mental imagery mirrored the trend of more complexity in idiom comprehension development, from more literal interpretations to more figurative interpretations (Abkarian et al., 1992; Levorato & Cacciari, 1995). More recently, Caillies and Le Sourn Bissaoui (2006) found a developmental effect, in particular a grade effect, in idiom comprehension in French-speaking monolingual children. Findings indicated that de composable idiomatic expressions, those akin to transparent idioms, presented in context were unde rstood earlier than nondecomposable idioms or those more similar to opaque idioms. Specifically, monolingual French-speaking children did not understand decomposable idioms until they were in third kindergarten (ages 5;3 to 6;2). In cont rast, nondecomposable idioms were not understood until child ren were in the second grade (ages 7;6 to 9;2). Moreover, Caillies and Le Sourn Bissaoui (2006) conclude d that perhaps the fi gurative meaning of decomposable idioms might be interpreted from inferences drawn about the constituent

PAGE 30

19 word meanings; however, these inferences were less helpful in interpreting the figurative meaning of unfamiliar nondecomposable idioms. A final study (Chan & Marine llie, 2008) further supports this developmental trend in idiom comprehension. Native Englishspeaking preadolescents (grades 4 and 5; n = 20), young adolescents (grade 8; n = 20), older adolescen ts (grades 11 and 12; n = 20), and adults (college students; n = 20) defined 10 highly familiar idioms presented in isolation. There were signif icant age differences betwee n the adult group and the two younger groups, and between older adolescents and the two younger groups. Performance on idiom familiarity and idiom definitions improved with age. Two points emerge from these developm ental studies. First, non-decomposable idioms may be learned and lexicalized, depe nding on the frequency of exposure. Second, figurative competence appears to depend on academic experience and, potentially, the degree of semantic and pragmatic abilities that individual children have developed (Caillies & Le Sourn Bissaoui, 2006). Idiom comprehension in children wi th linguistic/cognitive impairments. Children with linguistic and/or cognitive impairments have distinctive profiles. In general, children with language impairments may have significant difficulty understanding idioms (Spector, 1992). Children with cognitive defic its also have difficu lty interpreting oral idioms (Ezell & Goldstein, 1991). Overall, ch ildren with linguistic and/or cognitive impairments typically interpret oral idioms literally, much like younger children (Norbury, 2004). For example, Ezell and Goldstein (1991) compared 22, 9-yea r-old children, who were classified with mild mental retarda tion (MR), with 22 typical ly developing 9-year-

PAGE 31

20 olds, and 22 younger children who were matc hed to the cognitively impaired group according to receptive vocabulary scores. A ll participants were English-speaking monolinguals. Even though children with cogni tive impairment preformed significantly better than the younger children in the study, they consisten tly tended to give literal responses. Norbury (2004) simultaneously investigated children with linguistic impairment, children with autism spectrum disorder (A SD), and children w ith both linguistic impairment and ASD. A total of 93 children between 8 and 15-years-old were classified into four groups based on three measures: a) three standardized language assessments used to examine expressive and receptive langu age ability, b) an autistic screening parent questionnaire, and c) a communi cation checklist to determine the existence of pragmatic impairment. The four resulting groups consisted of autistic spectrum with language impairment, autistic spectrum without langua ge impairment, language impairment only, and pragmatic impairment only. Norburys (2004) findings indicated th at all participants benefited from the use of context to compre hend unfamiliar oral idioms. Of importance, one of the most significant pr edictors of idiom understand ing was language ability; that is, those children with linguistic impairment (either with or without ASD) performed more poorly than those children without la nguage impairment. One limitation of this study involves the response format of the idio m test. Participants we re required to define and explain idioms, which was difficult for all participants, but perhaps created an even greater disadvantage for those children diagnos ed with expressive language impairment. In another study, Qualls, Lantz, Piet rzyk, Blood, and Hammer (2004) found that adolescents with a documented diagnosis of language-based learni ng disabilities (LBLD)

PAGE 32

21 in grade 8 ( n = 27) had more difficulty comprehending written idioms than their agematched and reading-ability-matched peers in grade 8 ( n = 21), who were also reading below grade level. Reading and language arts scores on the California Achievement Test (CAT) were obtained for each particip ant. The CAT assessed vocabulary (e.g., synonyms), reading comprehension (e.g., infe rence-making), language mechanics (e.g., editing skills), and language expression (e.g., coherent writing). Language ability (regardless of LBLD stat us) predicted more of the variance than did reading ability alone. In addition, a strong relationship emerged between idiom compre hension and reading ability as assessed by the Idiom Comprehens ion Test (ICT; Qualls & Harris, 1999). All studies investigating children with la nguage-based disorder s have collectively concluded that language impairment is one of the leading causes in idiom comprehension failure, as well as failure in other academica lly-related tasks, specifically tasks involving inference generation. Several studies have also demonstrated that children and adolescents with language-based learning di sabilities are typica lly unable to use contextual cues effectively to interpret idioms (Qualls et al ., 2004). Moreover, all of these findings suggest that idioms should be taught in an explicit manner to children with linguistic and cognitive deficits (Norbury, 2004). Idiom Comprehension in Second-Language Learners There has been minimal research on the or al and written idio m comprehension of bilingual children. The major ity of research has been c onducted with bilingual adults (Abel, 2003; Cooper, 1999; Laufer, 2000; Liont as, 2002). These studies have tended to search for insights to enhance idiom instruction.

PAGE 33

22 Adult studies. Cooper (1999) suggested that second-language learners use multiple strategies depending on the transpar ency/opaqueness, decompositionality, and/or familiarity of idioms. Cooper employed think-alouds to understand how adult secondlanguage learners processed idioms since th is methodology allowed for the evaluation of the usually silent processes involved in r eading comprehension. To give the idiomatic expressions context, Cooper in cluded more literal idioms (e.g., to see eye to eye ) and more oral idioms or slang (e.g., whats cooking? ). All were embedded in one to two sentences. The 18 participants, ages 17to 44-years-old, who were all learning English as a second language, included 8 Spanish-sp eakers, 3 Japanese-speakers, 5 Koreanspeakers, 1 Russian-speaker, and 1 Portuguese -speaker. As a group, there was an absence of correlation between the literal and figur ative meanings of opaque idioms, which seemed to be an obstacle in idiom comprehens ion. Idioms that were easier to interpret were reported to be more familiar. Cooper identified three strategies that the participan ts used at least 71% of the time: a) guessing from context, b) discussing an d analyzing the idiom, and c) referring to the literal meaning of an idiom. Approximately 29% of the time, the participants used additional strategies, including: a) requesting information, b) repeating or paraphrasing the idiom, c) using background knowledge, and d) referring to a similar idiom from their native language. A total of 57% of idioms were interpreted correctly. Major limitations of the study were that only qualitative and desc riptive statistics were employed, variables such as semantic transparency, familiarity, and context were not controlled, and the sample size was small and linguistically variable in thei r first languages.

PAGE 34

23 In another adult study, Li ontas (2002) created the Id iom Diffusion Model (IDM) to explain the idiom comprehension of 53 university students whose first language was English and who were second-language learne rs of Spanish, French, or German. The IDM contains a prediction phase similar to predictive inferencing, followed by a confirmation phase in the idiom comprehensi on of second-language le arners. Participants read second-language idioms in and out of context and then 1) wrote the idioms meanings, 2) identified the reading strategies used, the thought proces ses utilized, and the schema/image created while interpreting each idiom, and 3) identified their affective states when interpreting each idiom. Transfer of idiomatic knowledge was significantly affected by context, translation equivale ncy, degree of idiomatic opacity, lexical knowledge, syntactic arrangement, and literal meanings. The results supported the IDM; however, this model is not appropriate to investigate the idiom comprehension of children since a high level of metalinguistic aw areness is necessary to report ones own predictive inferencing and infe rence confirmation strategies. Next, Abel (2003) pointed out that earlier monolingual hypotheses of idiom comprehension (e.g., Swinney & Cutler, 1979) cen tered on the lexical level of activation rather than both the lexical and conceptual levels. Bilinguals appear to share a conceptual level of representation between their native and nonnative lexicons (e.g., Hernandez, Li, & MacWhinney, 2005); therefore, it may be that both languages are accessed at the conceptual level when the individual is faced with an unknown idiomatic expression. Abel (2003) introduced the Dual Idiom Repr esentation (DIR) model to address how 169 graduate and undergraduate native speakers of German appeared to store English nondecompositional idioms as idiom entries while decompositional idioms were

PAGE 35

24 represented by their constituents. Furthermore, results demonstrated that an idioms decompositional status determined its represen tational status (i.e., wh ether or not it was represented as a lexical entry), and an idioms frequency influenced the development of a lexical entry for a non-decompositional idiom; that is, the more frequent an idiom appeared in the language, the more likely a lexical entry for that idiom would be created. An assumption of the DIR is that second-language learners do not develop as many idiom entries as native speakers due to a lower frequency of encounter s with those idioms in the second language. Thus, when an idiom in the second language does not correspond to an idiom in the first language, second-language learners may rely more on constituent lexical entries. The overall premise of the DIR is similar to the Global Elaboration Model in that opaque idioms are typically lexi calized, and unknown tran sparent idioms are semantically analyzed. The Global Elaborat ion Model was chosen as the theoretical framework for the present study instead of the DIR since a) the DIR has only been utilized in one study on educated adults, and b) the research design of the current study does not assess idioms representational status. Limitations of second language learner studies Overall, a general limitation of these second language learner st udies is the lack of inferen tial statistical evidence. For instance, Cooper (1999), Liontas (2002), and Abel (2003) all utilized only descriptive statistics. Furthermore, factors known to impact on either oral or written idiom comprehension, such as semantic transparency, familiarity, and context, were not systematically controlled. A second limitation of these previous studies is their sample characteristics. All the samples consisted of adults with a cons iderable amount of edu cation. There have not

PAGE 36

25 been any studies conducted with bilingual childr en or adolescents, typically developing or with language/cognitive impairment. Furt hermore, none of the bilingual studies have investigated the relationshi p between idiom comprehensi on and text comprehension, a relationship that monolingual and cross-linguistic studies (e.g., Cain et al., 2005; Levorato et al., 2004) have suggested is strong. Reading Comprehension, Idiom Comprehension, and the Global Elaboration Model Idioms and Text Comprehension Initial studies. The Global Elaboration Model was used as the underlying rationale for one of the few idiom co mprehension studies involving reading comprehension. Levorato et al. (2004) investigated whether reading comprehension skills in monolingual Italian children predicted thei r idiom comprehension skills. Based on the models construct, the studys rationale was that, instead of semantically deconstructing an idiom into its individual parts, idiom co mprehension required children to integrate figurative meaning with contextual information (Levor ato et al., 2004). The models basic premise was that th e critical factor in acquiring and comprehending idioms concerned the ability to relate an idioms meaning to its surrounding social and linguistic contexts (Cain et al., 2005 ). The hypothesis tested was that reading comprehension skills would pr edict idiom comprehension skills. Results provided support for the hypothesis. Children with better readi ng comprehension abilities were more able to interpret idiomatic meani ngs that required inferencing in order to construct a global semantic representation.

PAGE 37

26 Longitudinal investigation A more recent study of idiom and reading comprehension in Italian-speaking children examined the predictiveness of reading comprehension in a longitudinal design. Levorat o et al. (2007) studied 6year-old first graders with various levels of reading co mprehension abilities (23 good comprehenders and 29 poor comprehenders) over eight m onths. To provide more evidence for their Global Elaboration Model, th e investigators analyzed ch ildrens comprehension of idiomatic and literal sentences at two distin ct times: in first grade and again in second grade (eight months later). The author s argued that this study made two new contributions to the literature. It was th e first longitudinal study to identify the developmental relationship between text and idiom comprehension, and to consider the role of literal sentence comprehension as a potential mediator be tween text and idiom comprehension. Text and idiom comprehensi on relies more on inferential capacity and comprehension monitoring (both higher-level pr ocessing skills necessary to attain global coherence) than did literal sentence compre hension (which, alone, is insufficient for accurate text and idiom comprehension). Thus, the Global Elaboration Model would predict that a) only the children who improved in text comprehension would improve in idiom comprehension, and b) literal sentence comprehension should play a lesser role in idiom comprehension than did text compre hension. The sentence comprehension task required participants to choose one of four pictures that depicted each sentence. It was not noted whether the sentences were read by the participants or read aloud by the investigators. Results indicated that, during the first phase of the study, skilled comprehenders preferred figurative interpretations of idiomatic expressions, while less-skilled

PAGE 38

27 comprehenders preferred literal interpretations. A multiple regression analysis demonstrated that text comprehension abiliti es accounted for approximately 32% of the variance in the idiom comprehension measure, whereas sentence comprehension did not explain any further variance. Thus, the authors concluded that the contribution of text comprehension ability explained a significan t amount about idiom co mprehension ability contrasted with sentence comprehension ability. In the second phase of the study eight months later, about half of the less-skilled comprehenders had improved their text co mprehension by 30 percent. There was no attrition of the less-skilled comprehende rs reported from phase one to phase two. Analyses of figurative vers us literal responses on the idiom comprehension task demonstrated that less-skilled comprehende rs, although they had improved from the first phase, still chose more literal answers than the skilled comprehenders. The authors posited that this pattern yielded evidence for shallower text processing, which is inadequate either for accurate global text or idiom comprehension. A multiple regression analysis showed that improvement in sentence comprehension played a role in childrens progress in idiom comprehension for the less-skilled comprehenders; however, the improvement in sentence comprehension wa s related to the improvement in text comprehension. Text comprehension was the mo st significant factor in improvement of idiom comprehension. There are a number of unresolved is sues with the longitudinal outcomes: 1) Unfortunately, the results of the se ntence comprehension test in the second phase were unstated; therefore, it is difficult to determine how the less skilled versus the

PAGE 39

28 more skilled comprehenders improved on this measure in comparison to the other measures. 2) Oral language ability was not assessed; thus, any conclusions regarding linguistic processing ability as shallow or deep are suspect. 3) Childrens decoding abilities, a skill that would supersed e independent text comprehension, were not tested. 4) In the second phase the formerly less-skilled comprehenders were divided into skilled and less-skilled groups again, depe nding on whether they improved their reading comprehension by 30%. No empirical rationale was given as to why 30% was used as a criterion; therefore, the selection of this percentage appears arbitrary. 5) Lastly, the number of items on the id iom measure was unreported. Hence, it is difficult to interpret fully the distribution of idiomatic, literal, and filler answers. Moreover, it was unclear how familiar particip ants were with the idioms utilized, and familiarity could be a confounding variab le. A study from Great Britain with monolingual English-speaking children addressed th is last limitation in particular (Cain et al., 2005). Cross-sectional research on the Global Elaboration Model Cain et al. (2005) investigated the relationship between idio m comprehension and reading comprehension based on the Global Elaboration Model initi ally developed by Levorato and Cacciari (1995). The idiom comprehension of 28 9and 10-year-old children with good ( n =14) and poor ( n =14) reading comprehension skills was compared. Children were matched on word reading ability and vocabulary knowledge scores from standardized measures. Transparent and opaque idioms were utilize d. An innovative addition to this line of

PAGE 40

29 research was the inclusion of both familiar and unfamiliar idioms that were translated European idioms. As mentioned previousl y, familiarity and exposure to idioms can confound idiom knowledge. To eliminate pr ior British cultural experience as a confounding variable, Cain et al. (2005) used the unfamiliar European idioms. Results demonstrated that children understood familiar idioms better than the unfamiliar ones, even when presented in context, a finding that supported the language experience hypothesis. Cain et al. (2005) also found evidence for the practicality of the Global Elaboration Model because the poor co mprehenders engaged in analyzing idioms constituent by constituent; while the more pr oficient readers reli ed on both local and global coherence, along with inferencing, to obtain meaning. In other words, good comprehenders were able to go beyond individual semantic analysis at the word level to accomplish two ends: they surpassed literal and inappropriate semantic meanings and drew inferences based on available contex t. The outcome was the integration of the appropriate semantic meanings and derivation of appropriate figurative meanings. Constituent by constituent analysis was not detr imental when the children were presented with transparent or decom positional idioms; however, piece by piece analysis led to literal and/or inappropriate analyses of opaque or non-decompositional idioms. The opaque idioms required use of textua l context to draw inferences. Cain et al. (2005) concluded, therefore, that idiom comprehension appeared related to levels of read ing comprehension. Although this study provided evidence in favor of the relationship between reading co mprehension and idiom comprehension while controlling for prior idiom knowledge, it did not asse ss other potential processing abilities that may be key in both reading comprehension and idiom comprehension. For

PAGE 41

30 instance, Levorato et al. (2004) suggested that future studies of idiom comprehension should identify the processing abilities that reading comprehension and idiom comprehension share. Shared Processes in Reading Comprehension and Idiom Comprehension Idiom and reading comprehension require similar conceptual understandings. These conceptual understandings include a well -developed theory of mind (the ability to attune interpretation to the speakers/write rs intended meaning), the application of background knowledge, and the knowledge that inferences must be generated. The Global Elaboration Model is based on a we ll recognized text comprehension model, Kintschs (1998) Construction -Integration (C-I) Model. Th e impetus for the C-I Model was the Discourse Model of Reading Compre hension (van Dijk & Kintsch, 1983). The C-I Model (Kintsch, 1998) is built on the importance of constructing a situational model (or a mental representation of a text) to create meaning. Th eoretically, the model can be applied to the comprehension of bot h oral and written discourse. The C-I model The C-I Model posits that a text is made up of many propositions, or units of linguistic meaning. To comprehend th e gist of a text, the reader must succeed in creating coherence, which is assembled through inference generation (e.g., combining known knowledge with incoming knowledge) and inference retrieval (e.g., accessing background knowledge via long term memory). Inferences also require theory of mind or perspective taking to understand the implic it meanings of others (both real and hypothetical) and the points of view of characters and authors. First, local coherence between propositions or constituents in the same sentence must be achieved. Then, global coherence from sentence to sentence across the text must

PAGE 42

31 be attained. Local and global coherence refer to the construction and integration phases, respectively (Kintsch, 1998). In tegration of meaning is nece ssary for text comprehension (see Figure 2). Figure 2. Construction-integration phases of text comprehension based on Kintsch (1998). Comprehension monitoring and error detection Making accurate meaning of text also requires an awareness of how well the text is understood or the ability to monitor comprehension (National Reading Panel, NRP, 2000). By monitoring comprehension, Outcome: Text comprehension Integrate with global coherence of intra-sentential propositions Integration Phase Obtain local coherence of inter-sentential propositions Construction Phase Phases Emerge from an Accurate Situational Model

PAGE 43

32 readers discern if they have understood the text or if they need to reprocess chunks of misunderstood text (Morrison, 2004). It is likely that the individua l reader must be able to monitor his/her comprehension of text in order to employ effective comprehension strategies (Morrison, 2004). Comprehension m onitoring is a type of cognitive monitoring and refers to students awareness of the degree to which they understand what they are reading (Morrison, 2004, p. 78). Morris on (2004) investigated the relationship between comprehension monitori ng in readers first (L1) and second languages (L2) at the university level in 52 advanced learners of French as the L2. In conducting the study, Morrison utilized an error de tection task, a technique us ed to manipulate a texts comprehensibility by purposefully embedding text errors in order to measure comprehension monitoring. Morrison (2004) found positive correlations between a) L1 reading proficiency and overall L1 error detection ( r = 0.60, p<0.01), b) L1 reading proficiency and L1 macro-level error detection ( r =0.54, p<0.01), and c) L1 reading proficiency and L1 micro-level error detection ( r =0.51, p<0.01). Similarly, Morrison found significant, positive correlations between L2 reading proficiency and error detection ability, as well as significant crosslinguistic correlations. Moreover, these findings suggested that the Morrison error detection task may be a reli able measure of comp rehension monitoring; and that comprehension monitoring is corr elated with reading comprehension ability. In the past, descriptive measures have been utilized to assess comprehension monitoring. For instance, one common past methodology had participants estimate how well they performed on a postreading comprehension measure. These estimates of performance, or confidence rati ngs, were then compared to the participants actual scores.

PAGE 44

33 Unlike descriptive tasks, the Morrison e rror detection paradigm assesses two known comprehension monitoring processe s: identifying an error and repairing an error, at an experimental level. Furthermore, in accord with Kintschs (1998) C-I Text Comprehension Model, the error detection pa radigm allows for error detection at the sentence or micro-level (i.e., the meaning-construction phase) as well as at the discourse or macro-level (i.e., the info rmation-integration phase). Processing abilities underlying figurative comprehension A set of specific processing abilities is required in figurative text comprehension, such as idiom comprehension and/or metaphor and proverb comprehension. According to Levorato and Cacciari (1995), the abilities involved incl ude: a) understanding each words multiple meanings, b) going beyond literal interpretati ons, c) using context to create a coherent figurative expression, and d) appreciating that what is said ma y not always coincide with what is meant. Furthermore, Levorato and Cacci ari (1995) refer to the attainment of these processing abilities as achieving figurative competence. Levorato et al. (2004) conjectured that difficulty in terpreting figurative meanings may be due to three factors: a) not being ab le to suppress the literal meanings of the idioms constituent words; b) having less ability to exploit contextual information to create a situation model; and c) the inability to make necessary inferences in order to choose the appropriate (figur ative) meaning. The developm ent of figurative competence is seen as emergent over time and nonlinear (Levorato & Cacciari, 1995). Furthermore, the same knowledge and processes (e.g., cognitive, linguistic, pragmatic) used to comprehend linguistic information in general are also used to comprehend idiomatic expressions. In sum, The cognitive skills necessary to understand

PAGE 45

34 figurative language are grounded in the capacity a child must possess to process a text (Levorato et al., 2004, p. 311). Levorato et al. (2004) outline d the four most relevant principles of reading comprehe nsion for idiom comprehension: 1) Application of inference generation a nd retrieval strategies to create local coherence at the word level and global cohe rence at the sentence level while considering contextual support. 2) Application of inhibitory strategies to suppress, or at least suspend, irrelevant constituent meanings in favor of relevant, figurative meanings. 3) Application of comprehension monitoring strategies to ensure accurate comprehension performance. 4) Application of establishing contextu ally specific and appropriate word meanings from various possible meanings. These four principles are the crux of the model underlying this study and are expanded on next. The Model and Research Questions The link between reading comprehension and idiom comprehension, as described by Cain et al. (2005) and Levor ato et al. (2004), provided no t only the motivation for the present study, but also its c onceptual framework. In sum, the evidence that reading comprehension may predict idiom comprehens ion lends support for incorporating the Kintsch (1998) C-I Model with the Global Elaboration Model. The studys purpose was to explore further the relationship betw een idiom and reading comprehension in adolescents who were Spanish-English bilingual s. In implementing this combined model, the innovative Cain et al. (2005) methodology was employed in an expanded manner

PAGE 46

35 with both monolingual (English-speaking) and bilingual (Spanish-English-speaking) adolescents as participants. The literature is notably devoid of i nvestigations on idiom comprehension as it relates to reading comprehension in Spanish-speaking bilingual adolescents in the United States. Purpose This studys purpose was to investigat e idiom comprehension in bilingual adolescents and their monolingual peers thr ough the systematic evaluation of each of Levorato et al.s (2004) four principles (see Figure 3). At the same time, the intent was to control systematically for the three variables of semantic transparency, familiarity, and context. This studys design went beyond prio r research on the Global Elaboration Model (e.g., Levorato et al., 2004; Le vorato et al., 2007) due to a) the focus on a bilingual sample, b) the investigation of cognitive-li nguistic processes shared by text and idiom comprehension, c) the assessment of decoding ab ility, d) the utilizati on of a statistically significant different skilledand less-skilled comprehenders groups, and e) the use of unfamiliar idioms similar to Cain et al. (2005). Unlike Cain et al. (2005), this study included a larger, diverse sample size ( N = 62) and a varied set of measurements.

PAGE 47

36 Figure 3. The four principles of literal and figurative reading comprehension (Adapted from: Levorato et al., 2004) Principles Each of the four principles was operationalized so th at it was assessed independently. Principle 1. Inference generation and retrieva l is essential to create local coherence at the word level a nd global coherence at the sent ence level combined with the Principle One Principle Two Principle Three Principle Four a) Inference making from single word level to sentence level d) Ability to choose contextually appropriate meaning from various possible meanings b) Ability to ignore contextually inappropriate meanings in favor of contextually appropriate ones c) Ability to monitor ones comprehension

PAGE 48

37 use of contextual support. This principle was assessed by measuring the participants ability to formulate inferences from the si ngle word level to the sentence level. This ability was tested through the administra tion of the reading comprehension passage subtest of the Woodcock Johnson III-Ac hievement (Woodcock, Mather, & McGrew, 2001; WJ III-ACH). Principle 2 Inhibitory strategies must be appl ied to suppress, or at least suspend, irrelevant constituent meanings in favor of relevant, figurative meanings This principle was assessed by measuring the participants ab ility to ignore inappr opriate, l iteral and contextually relevant, but inaccurate, figur ative meanings in favor of contextually appropriate and accurate figurative meanings. To meet this aim, a constructed multiple choice idiom measure systematically tested: a) familiar and unfamiliar idioms, with the unfamiliar idioms similar in form to those used in Cain et al. (2005) and b) transparent and opaque idioms, c) in and out of context. Principle 3 Comprehension monitoring strategies must be employed to maximize accurate comprehension performance. This th ird principle was assessed by measuring the participants ability to mon itor their comprehension at th e micro-level (the sentence level) and the macro-level (the paragraph le vel) using an error detection task derived from Morrison (2004). Principle 4. The ability to integrate contextua lly specific and appropriate word meanings from various possible definitions wa s assessed by measuring one part of lexical depth. Word knowledge can be described in at least two dimensions: breadth and depth. Lexical breadth refers to the shallow aspect of vocabulary size, or the number of words for which someone has at least some superficial level of knowledge (Qian, 1999, 2002).

PAGE 49

38 Depth of lexical knowledge refe rs to how well a word and its semantic relationships are known (Qian, 1999), such as knowledge of a words multiple meanings which are interconnected by a semantic network. Interconnected meanings, also referred to as polysemy (Nagy & Scott, 2000), comprise an important aspect of lexical depth. A synonym task from the Woodcock Johnson II I-ACH (Woodcock et al., 2001) Reading Vocabulary Subtest was sele cted for this purpose. The systematic measurement of the four principles in the model is further elaborated on in the Method chapter. Research Questions There were three questions related to the studys theoretical model: 1) To what extent does each of three of the linguistic va riables predict the criterion, idiom comprehension accuracy ? These variables were: a) reading comprehension, b) error detection, and c) knowledge of synonyms. It was hypothesized that performance on the three measures woul d strongly predict perf ormance on the idiom comprehension measure for both the bilingual and monolingual groups. 2) The second research question related to whether there were differences in idiomatic performance outcomes between the bilingual and monolingual language groups. The specific question concerned how th e performance outcomes of the bilingual adolescents would differ from the perf ormance outcomes of the control group (monolingual, English-speaking adolescents) given different levels of idiomatic familiarity, semantic transparency, and context. It was predicted that there would be an interaction among familiarity, semantic tr ansparency, context, and language group. A total of four sub-hypotheses were associated with this question.

PAGE 50

39 2a) Both monolinguals and bilinguals would perform less adequately on unfamiliar, opaque idioms. 2b) Monolinguals would perform better on familiar idioms based on the language experience hypothesis than would the bilinguals. 2c) All participants should perform be tter when given contextual support than without it; however, context would not bene fit less skilled compre henders as much as skilled comprehenders. 2d) Those participants with less adequate reading comprehension scores would choose more literal responses regardless of language group membership. 3) The third and final res earch question focused on the bilingual adolescents only. It was hypothesized that meani ngful differences would exis t within the bilingual group depending on age of acquisition (AOA) of E nglish or time spent in the United States, and, subsequently, amount of Spanish spoken on a daily basis. The question asked whether those bilingual students who were le ss linguistically assimilated (measured by amount of Spanish spoken, and thus less Englis h, daily) would perfor m in a significantly different manner from bilingual students who were more linguistically assimilated (spoke less Spanish, and thus more English, daily). This withingroup question required both quantitative and qualitative analyses of performance differences between the simultaneousand sequential-language-learners.

PAGE 51

40 Chapter Two Pilot Study The purpose of the pilot study was twofol d: a) to assess the validity of the constructed idiom measure and b) to comple te a preliminary analysis of performance differences on the idiom measure for two gr oups of undergraduate st udents: a bilingual group and a monolingual group. The pilot study cons isted of two parts. In the first part, monolingual (English-only) and bilingual undergraduate university students completed the pilot idiom measure, and th eir responses were statistica lly and qualitatively analyzed. During the second portion, monolingual (E nglish-only) adolescents completed a familiarity rating form, and their results were also quantitatively and qualitatively analyzed. The development of the idiom measure and the methodology employed in the undergraduate study is described first. Next, the development of the familiarity rating form and the methodology employed in the ad olescent study is explained. Following each of these descriptions, the anal yses of the pilot data are presented along with a discussion of their implications. Finally, the creation of the finalized idiom m easure is addressed. Method: Part 1 Participants For the quasi-experimental portion of the pi lot study, students at the University of South Florida with an unde rgraduate major in Communica tion Sciences and Disorders

PAGE 52

41 (CSD) were recruited through posted signs (see Appendix A) and emails sent by two of three participating professors. The principal i nvestigator also attend ed a third professors class and announced the need for participants. Students were given extra credit in return for their participation. Participants had to meet three criteria: a) be between the ages of 18to 35-yearsold; b) be an undergraduate student; and c) either be a monolingual (English-only) speaker or a self-reported bilingual speaker (speaker of English and at least one other language). These criteria were established in order to conduct stat istical comparisons between monolingual and bilingual participan ts with similar educational backgrounds. Furthermore, the age restrictions were in cluded to avoid any significant generational differences in idiom knowledge between th e undergraduates and the adolescents in the second pilot study. The sample consisted of 18 monolingual (English-speaking) and 18 bilingual students majoring in CSD. For the total group, there were 34 females and 2 males (both monolinguals), which was representative of the undergraduate population in the CSD Department at the University of South Florid a, Tampa. All participants were between 18 years and 11 months and 35 years and 2 mont hs old (see Table 2.1). The mean age of the monolinguals was 22; 4 ( SD = 2.8 years; range, 18;11 31;5), while the mean age of the bilingual students was 24; 5 ( SD = 5.06 years; 19; 11 35;2). The age span of the bilingual group was more variable; however, a t -test did not find a statistically significant difference in age between the two groups (see Appendix B for all non-significant t -test results). Of the 18 bilingua l students, 13 spoke English and Spanish and 5 spoke English along with Hindi, Arabic, Malayem, Creole and French, or

PAGE 53

42 Urdu (see Table 2.1). All of the bilingual studen ts had been in the United States for at least five years. Table 2.1 USF Undergraduate Participant Information Monolinguals_(Ages: M =22.36, SD =2.8; Range= 18.11-31.5)_________ Age (Years. Months) Gender Language __________________________________________________________ 18.11 Female English 20.8 Female English 20.8 Female English 20.10 Female English 21.2 Female English 21.3 Female English 21.5 Female English 21.6 Female English 21.6 Male English 21.9 Female English 22 Female English 22.3 Female English 23.5 Female English 23.9 Female English 24 Male English 24.10 Female English 31.5 Female English Bilinguals (Ages: M =24.45, SD =5.06; Range=19.11-35.2) ____________ Age (Years, Months) Gender Language (Other than English)_ 19.10 Female Spanish 20.4 Female Hindi 20.11 Female Arabic 21.1 Female Spanish 21.2 Female Spanish 21.3 Female Urdu 21.7 Female Malayem 21.7 Female Spanish 22.3 Female French/Creole 23 Female Spanish 23.3 Female Spanish 24.10 Female Spanish 27.1 Female Spanish 27.6 Female Spanish

PAGE 54

43 (Table 2.1 continued) 32 Female Spanish 33.9 Female Spanish 35.2 Female Spanish Note All bilingual participants were fluent in English. Materials: Development of the Pilot Idiom Measure The pilot idiom measure (see Appendix C), meant to be read silently and independently by each participant, tested th e familiarity, semantic transparency, and contextual strategies used for comprehendi ng idioms in a systematic manner through 96 multiple choice items. Possible selections were multiple choices in order to minimize oral language production, with 3 choices per idiom. Of the 3 answers, one was a literal but an inaccurate translation of the idiom, one was a figurative a nd correct translation of the idiom, and the third was figuratively related to the idiom but incorrect. Familiarity The 48 idioms were categorized into two levels of familiarity: 24 familiar and 24 unfamiliar. The familiar idioms had a higher frequency in American English, such as break the ice and, presumably, were more familiar than unfamiliar ones. These idioms were adapted from idiom meas ures utilized with monolingual, Englishspeaking children (e.g., Abrahamsen & Burke-Williams, 2004; Nippold, 1991; Nippold & Duthie, 2003; Norbury, 2004). The 24 unfamiliar idioms were translated European idioms, primarily adapted from Cain et al. (2005), such as to have salt in your pumpkin (meaning to be intelligent ) Most likely, these idioms had a lower frequency of occurrence in American English. Semantic transparency. The familiar and unfamiliar idiom groups were further subdivided into semantically transparen t and opaque categories. There were 12

PAGE 55

44 transparent familiar idioms, 12 opaque familiar idioms, 12 transparent unfamiliar idioms, and 12 opaque unfamiliar idioms. Those categorized as transparent had a more direct relationship between their literal and figurative meanings, such as to call it quits In contrast, with opaque idioms, a less direct re lationship existed between their literal and figurative meanings. For example, to pull someones leg is a more opaque idiomatic expression. Context. All 48 idioms were presented first ou t of context (in isolation) and then in short story contexts. Previous studies showed that comprehension of idiomatic expressions was facilitated by contextual support (Cacciari & Levorato, 1988; Nippold & Martin, 1989). For this studys purposes, unfamiliar idioms were those in which participants had to rely on contextual cues to interpret them. Therefore, unfamiliar idioms were testing the extent to which participants were able to take adva ntage of linguistic and social cues embedded in the short stories since reliance on familiarity alone in the unfamiliar idiom condition would lead to an erroneous interpretation. Procedure After completing consent forms (see A ppendix D), the undergraduate students completed the idiom measure. The measure was completed independently in three separate groups in the Language Laboratory of the CSD Department within two weeks of each other during the spring semester of 2006. Each student took approximately 15 to 20 minutes to complete the instrument. The fo llowing directions were presented to all participants orally: I am creating an idiom test for high school students and need to make certain that there are no unforeseen glitches. You will see each idiom, like spill the beans,

PAGE 56

45 appear twice on the test. The first time the idiom will appear alone and the second time the idiom will appear within context. It is very important that you work forward, and not go back and change your answers after reading the idiom a second time in context. Please read each idiom carefully and then choose the best definition of the idiom. You may not know some of the idioms, and may have to guess their meanings. The idioms may become progressively less familiar throughout the test. This task is comple tely voluntary, and if you wish to quit taking the test at any time, you are free to do so without any pe nalties or adverse effects on your grades. Please hand in th e test when you are done and thank you for participating. In addition to these oral instructions, the students were urged to read the printed instructions on the first page of the measure: Idioms are figurative or non-literal language like raining cats and dogs or bought the farm. I am creating an idiom test and need your help piloting this test before giving it to bilingual and monolingual high school students in the future. Their results will be compared to their reading and vocabulary scores to investigate any meaningful relationships. Please read each question care fully and then circle the best answer. There may be idioms that you do not know and will guess their meanings. It is important to work forward, and not to go back to change answers. If at any time you wish to stop completing this form you may do so without any consequences whatsoever. This is completely voluntary. If you have any quest ions feel free to ask me. I would like to thank you for participating.

PAGE 57

46 Upon completion of the task, each student received a card verifying participation along with a synopsis of the study and its purpo se (see Appendix E). All but one of the 36 students completed the entire te st. The data from the bilingua l student whose results were incomplete were not included in any anal ysis. To make the bi lingual and monolingual samples equivalent in number, the data fr om one monolingual part icipant were randomly chosen to be discarded as well. The final sample analyzed therefore consisted of 17 bilinguals and 17 monolinguals. Results: Statistical Analyses Total scores. The scores for each language group (bilingual or monolingual) were tallied for a total score, th ereby collapsing all the condi tions together. The maximum possible score for each participant was 96 (48 idioms, presented in and then out of context). For this analysis, re sponses were counted as either correct or incorrect, and a ttest was conducted to determine the mean differences between th e two language groups. The bilingual group had a mean score of 83.24 ( SD = 3.68) and the monolingual group had a mean of 85.71 ( SD = 2.76). Table 2.2 displays the de scriptive data (median, mode, and score ranges) for each language group. An independent t-test re vealed an observed t value of -2.21 and p=0.034, indicating a significant difference between the total mean scores of the two language groups. Furthe rmore, the estimated effect size of d = 0.76 was calculated. According to Cohens (1988) guidelines, this is a me dium to large effect size, suggesting that the magnitude of the mean difference in scores was meaningful. Table 2.2 Total Scores for Idiom Accuracy as a Function of Language Group ________________________________________________________________________ Monolinguals

PAGE 58

47 (Table 2.2 Continued) M SD Median Mode Range ________________________________________________________________________ 85.71 2.76 85 84 81-92 _______________________________________________________________________ Bilinguals M SD Median Mode Range ________________________________________________________________________ 83.24 3.68 84 78 78-89 ________________________________________________________________________ Next, to determine whether an effect existed for each variable when crossed with other variables [language group (bilingual or monolingual) x familiarity (familiar and unfamiliar) x semantic transparency (transparent and opaque) x context (with and without)], a four-way, repeated measures, analysis of variance (ANOVA) was conducted. Language group (bilingual or monolingual) was a between subjects factor and familiarity, semantic transparency, and context were with in subjects factors. Table 2.3 summarizes the descriptive data of the accuracy scores under each condition as a function of language group, and Table 2.4 summarizes the ANOVA results. Table 2.3 Accuracy Scores for Idiom Conditi ons as a Function of Language Group ________________________________________________________________________ Bilinguals Monolinguals Idiom Condition M SD M SD ________________________________________________________________________ Familiarity-Familiar 45.47 2.38 46.94 0.94 Familiarity-Unfamiliar 37.76 3.04 38.18 1.89 Transparency-Transparent 44.29 2.49 44.94 1.39 Transparency-Opaque 38.94 2.07 40.18 1.82 Context-In 46.59 1.29 47.59 0.60 Context-Out 36.65 3.01 37.53 2.17 ________________________________________________________________________

PAGE 59

48 Table 2.4 ANOVA Results for the Accuracy Sc ores on the Pilot Idiom Measure ________________________________________________________________________ Variable and source df MS F p ________________________________________________________________________ Familiarity (1, 32) 288.24 210.47 <.001* Familiarity x Group (1, 32) 1.19 0.87 0.36 Transparency (1, 32) 108.77 121.62 <.001* Transparency x Group (1, 32) 0.37 .41 0.53 Context (1, 32) 425.00 473.29 <.001* Context x Group (1, 32) 0.02 0.02 0.90 Familiarity x Transparency (1, 32) 74.13 77.63 <.001* Familiarity x Context (1, 32) 252.37 442.15 <.001* Transparency x Context (1, 32) 121.78 268.21 <.001* Familiarity x Transparency x Context (1, 32) 72.06 114.29 <.001* Familiarity x Transparency x Group (1, 32) 0.06 0.06 0.81 Familiarity x Context x Group (1, 32) 2.12 3.71 0.06 Transparency x Context x Group (1, 32) 0.94 2.07 0.16 Familiarity x Transparency x Context x Group (1, 32) 0.02 0.02 0.88 Statistically Significant Interactions. There was not a four-way intera ction that reached significance. There was one statistically significant three-way interaction among familiarity, transparency, and context, F(1, 32) = 114.29, MS= 72.06, p<0 .001, partial 2 = .78 with

PAGE 60

49 an observed power of 1.00. Post hoc testing us ing the Bonferroni procedure demonstrated a significant difference in performance due to context. Specifically, participants performed significantly worse under the fa miliar, opaque, out-of-context condition ( M =11), and far worse under the unfamiliar, opaque, out-of-context condition ( M =4.94) than in any of the in-context conditions (See Figure 4). Three Way Interaction: Familiarity x Transparency x Context0 2 4 6 8 10 12FTI FTO FOI FOO UTI UTO UOI UOO Mean Values Mean Values Figure 4. The statistically significant three-way interaction among accuracy scores in the following conditions and each of their two leve ls: familiarity, transparency, and context, with language groups collapsed. Note. The abbreviations used signify the following idiomatic conditions: FTI is familiar, transparent, incontext; FTO is familiar, transparent, out-of-context; FOI is familiar, opaque, in-context; FOO is familiar, opaque, out-of-context; UTI is unfa miliar, transparent, in-context; UTO is unfamiliar, transparent, out-ofcontext; UOI is unfamiliar, opaque, in-context; and UOO is unfamiliar, opaque, out-of-context. Furthermore, the confidence intervals (95th percentile) of each condition (e.g., Familiarity: familiar and unfamiliar) did not overl ap with each other, ensuring that they were each significantly different. Lastly, the most variable performance occurred under

PAGE 61

50 the no context and unfamiliar conditions, and the least amount of variability occurred in the in-context condition (see Fi gure 5 for the confidence intervals of these pair-wise comparisons). Figure 5. Confidence intervals of familiarity, semantic transparency, and context variables displayed in mean accuracy values and illustrated with error bars. Note. Values represent the mean of both langua ge groups collapsed together since group was not statistically significant. There were also three statistically si gnificant two-way interactions among the idiomatic variables. First, there was a significant interaction between familiarity and Opaque Transparen t Unfamiliar Familia r Out ofContext In Context 12.0 11.5 11.0 10.5 10.0 9.5 9.0 95% CI

PAGE 62

51 transparency, F(1, 32) = 77.63, MS = 74.13, p<0.001, and partial 2 = .71 with an observed power of 1.00. Post hoc testing using the Bonferroni adjustment demonstrated that all conditions were significantly different, with the weakest performance observed under the unfamiliar, opaque condition ( M = 8.34) and the best performance observed under the familiar, transparent condition ( M = 11.66) (see Figure 6 for the confidence intervals of each of these conditions). Figure 6. Confidence intervals of the familiarity and semantic transparency variables displayed in mean accuracy values and illustrated with error bars. Opaque Transparen t Unfamiliar Familia r 12.0 11.5 11.0 10.5 10.0 9.5 9.0 95% CI

PAGE 63

52 Note. Values represent the mean of both langua ge groups collapsed together since group was not statistically significant. Secondly, there was a significant interacti on between familiarity and context, F(1, 32) = 442.15, MS = 252.37, p<0.001, and partial 2 = .93 with an observed power of 1.00. Further post hoc testing using the Bonferroni procedure revealed a significant difference under the unfamiliar, out-of-context condition ( M = 7.23) (see Figure 7 for confidence intervals of these variables). Figure 7. Confidence intervals of familiarity and context variables displayed in mean accuracy values and illust rated with error bars. OutofContex t In-Context Unfamilia r Familiar 12.0 11.5 11.0 10.5 10.0 9.5 9.0 95% CI

PAGE 64

53 Note. Values represent the mean of both langua ge groups collapsed together since group was not statistically significant. Lastly, there was a significant interaction between transparency and context, F(1, 32) = 268.21, MS = 121.78, p< 0.001 and partial 2 = .89 with an observed power of 1.00. Again, post hoc testing using the Bonf erroni adjustment found a significant difference under the opaque c ondition, with performance decreasing when idioms were presented out-of-context (see Figure 8 for confidence intervals of these variables). Figure 8. Confidence intervals of semantic transp arency and context variables displayed in mean accuracy values and il lustrated with error bars. Out-of-Context In-Contex t Opaque Transparent 10.0 7.5 5.0 95% CI 12.0-

PAGE 65

54 Note. Values represent the mean of both langua ge groups collapsed together since group was not statistically significant. In addition, there were two other inte ractions approaching significance. The interaction of familiarity x context x group approached significance, F (1, 32)= 3.71, MS = 2.12, p = 0.06, partial 2 = .10 with an observed power of 0.46. Lastly, there was a three-way interaction with a trend towards significance invol ving transparency x context x group, F(1, 32) = 2.07, MS = 0.94, p = 0.16, a partial 2 = .78 with an observed power of 1.00. Taken together, these interactions suggest that, although there were no significant group interactions, the idiomatic conditions did interact significantly. Overall, participants performed better with familiar, tr ansparent idioms in context. Participants performed better on unfamiliar idioms when they were transparent and better on opaque idioms when they were familiar. Furthermore, participants performed better on both familiar and unfamiliar idioms when they were presented within context. Specifically, context was advantageous when interpreti ng unfamiliar idioms. Lastly, although context appeared to benefit participants under all id iomatic conditions, context was particularly helpful when given opaque idioms. Moreover, group interactions were approaching significance, suggesting that a larger sample size and greate r scrutiny of participants language experience may lead to significant results. Main effects. A main effect was found for the familiarity variable, F (1, 32) = 210.47, MS = 288.24, p<0.001, partial 2 = .87 with an observed power of 1.00. Both groups performed better ( M = 46.21) in the familiar idiom condition than in the unfamiliar idiom condition ( M = 37.97). Independently, the monolingual group had more

PAGE 66

55 correct answers ( M = 46.94) when interpreting familiar idioms versus unfamiliar idioms ( M = 38.18). Similarly, the bilingual group di d better on the familiar idioms ( M = 45.47) than on the unfamiliar idioms ( M = 37.76). The semantic transparency variable al so had a main effect, F (1,32)= 121.62, MS = 108.77, p <0.001, and partial 2 = 0.79 with an observed power of 1.00. Both groups performed better ( M = 44.62) in the transparent idiom condition then in the opaque idiom condition ( M = 39.56). The monolingual group produced more correct answers ( M = 44.94) with the transparent idioms c ontrasted with the opaque idioms ( M = 40.18). Similarly, the bilingual group did better with the transparent idioms ( M = 44.29) versus the opaque idioms ( M = 38.94). In terms of the context variable, there wa s a main effect for context, F(1, 32) = 473.286, MS = 425, p<0.001, partial 2 = 0.93 with an observed power of 1.00. Both groups performed better ( M = 47.18) in the within-context condition than the withoutcontext condition ( M = 37.09). The monolingual group provided more correct answers ( M = 47.59) with the idioms in context than with idioms out-of-context ( M = 37.53). Similarly, the bilingual group did better on the idioms presented in-context (M = 46.76) compared with the idioms out-of-context ( M = 36.65). These three significant main effects illustrated that the participants performed distinctively in each binary category of each idiomatic c ondition (familiarity, semantic transparency, and context). In other words, participants performed better with familiar rather than unfamiliar idioms, with transpar ent rather than opaque idioms, and with context rather than without. Th ese findings validated that each of the idiomatic conditions were systematically controlled. Overall, pa rticipants performed less well on unfamiliar

PAGE 67

56 ( M = 37.97) than familiar idioms ( M = 46.21), less well on opaque (M = 39.56) than transparent idioms ( M = 44.62), and less well on idioms out-of-context ( M = 37.09) than on idioms in-context ( M = 47.18). Results: Qualitative Analyses Since the statistical analyses only addresse d the data in a binary manner (whether accurate or inaccurate), a qualita tive error analysis was also carried out. Each incorrect answer was coded as being literal and incorrect or fi gurative and incorrect. Error analysis. Each of the four conditions (familiar transparent, familiar opaque, unfamiliar transparent, and unfamiliar opaque) inand out-of-context was analyzed. The total number of literal and figurative errors fo r each language group is illustrated in Table 2.5. Table 2.5 Error Analysis of All Items in All Conditions on the Pilot Idiom Measure By Language Group and Error Type ________________________________________________________________________ Condition M Literal Errors B Literal Errors M Figurative Errors B Figurative Errors Total ________________________________________________________________________ FTI 0 5 1 1 7 FTO 0 2 5 9 16 FOI 0 1 0 3 4 FOO 3 6 8 17 35* UTI 1 2 2 5 10 UTO 3 1 39 39 82 UOI 1 1 1 6 9 UOO 32 28 87 93 240 _______________________________________________________________________ One answer was left blank a nd was counted as an error.

PAGE 68

57 Note. The abbreviations used signify the following idiomatic conditions: FTI is familiar, transparent, incontext; FTO is familiar, transparent, out-of-context; FOI is familiar, opaque, in-context; FOO is familiar, opaque, out-of-context; UTI is unfa miliar, transparent, in-context; UTO is unfamiliar, transparent, out-ofcontext; UOI is unfamiliar, opaque, in-context; and UOO is unfamiliar, opaque, out-of-context. One example of an idiom that seven participants missed was the transparent, familiar idiom: take someone under ones wing The idiom short story context was: The more experienced pilot taught the ne wcomer, Jerry, how to fly the jet. He took Jerry under his wing. The choices were a) to give someone your seat on a plane, b) to offer someone guidance, and c) to teach someone to fly. Five of the participants who missed the correct answer b) to offer someone guidance chose c) to teach someone to fly, the literal response, and one bilingual and one monoli ngual chose the figurative, but incorrect response a) to give someone your seat on a plane. Taken as a whole, fewer errors were made when given familiar idioms (62 errors in total) rather than unfamiliar idioms (341 errors in total). Also, fewer errors were made with transparent idioms (115 e rrors in total) than opaque idioms (288 errors in total). Furthermore, more errors were made when idioms were presented without context (373 errors in total) rather than with in context (30 errors in total). Overall, the majority of errors occurr ed when idioms were presented out-ofcontext, particularly when they were both opaque and unfamiliar. Of interest, there were far fewer literal errors (104 erro rs in total) than figurative errors (317 errors in total). Lastly, the bilinguals had either equal or more errors excep t in two conditions: a) fewer

PAGE 69

58 literal errors on the unfamiliar, transparent idioms out-of-context and b) fewer literal errors on the unfamiliar, opaque idioms out-of-context. Item analysis. Lastly, an item analysis was conducted to detect any idiom scenarios that should be eliminated befo re going forward with the main study. The criterion was that any items presented in cont ext that 50% or more of the participants answered incorrectly were considered invalid questions. None of the items met this less than 50% criterion; that is, all of the items presented in context were correctly interpreted at least 50% of the time by all particip ants, both bilinguals and monolinguals. The item analysis was performed on idiom scenarios in-context instead of idiom scenarios out-of-contex t since those items out-of-context required either prior knowledge (e.g., the participant was familiar with the idio m already) or guessing (e.g., the participant was unfamiliar with the idiom and had to guess its meaning based on the three multiple choice responses without any supporting context). In summary, the qualitative analyses echoed the quantitative analyses. Participants perform distinctively under each idiomatic conditions two categories. Furthermore, according to the item analysis the items and their short story context appeared valid and the results paralleled results from past research. Namely, there was a pattern of heightened performance with fam iliar versus unfamiliar idioms, transparent versus opaque idioms, and context supports idiom comprehension in general. Discussion The independent t-test demonstrated th at there was a significant difference between the performances of the USF bilingu als and monolinguals on the idiom measure. However, when each variable was analyzed separately, there was no language group

PAGE 70

59 effect. Therefore, it appeared that there was an overall difference between the groups when total score (i.e., all 96 items under al l conditions crossed) was considered. As demonstrated through the one three-way and three two-way significant interactions, the three idiomatic conditions (familiarity, seman tic transparency, and context) interacted amongst each other. There was also a main effect for each of the three idiomatic conditions. Overall, all students perform ed better under the less taxing conditions: familiar, transparent idioms in context. Mo reover, these findings suggested that the idiomatic conditions were systematically co ntrolled, and each condition should remain in the main studys idiom comprehension measure (ICM). An interesting finding of the pilot data was that the USF monolinguals, although not significantly so, did perform better that the USF bilinguals in each condition (see Table 2.3 for descriptive data). Furthermore, there appears to be more variability among the bilinguals scores in gene ral than the monolinguals scores. Recruiting only SpanishEnglish bilinguals and using a questionnaire to explore participants language history and language experience during the main study shoul d group the bilinguals in a more refined manner (i.e., late versus early English lear ners, as well as highor low-use Spanish speakers). This grouping strategy aimed to a llow for exploration of any variability or patterns evident in the bilingu al sample of the main study. The interactions approaching significance confirmed that a difference may exist between the language groups given a more refined and larger sample. As for the qualitative analyses, the item analysis demonstrated that no question was missed more than 50% of the time by bilingual or monolingual participants. Furthermore, for most items missed within context (i.e., 20 figurative but incorrect

PAGE 71

60 responses and 10 literal and inco rrect responses) participants w ith errors tended to choose the figurative but incorrect m eaning over the literal meaning. This pattern demonstrated that the alternative figurative but incorrect meaning was challenging to at least some monolingual and bilingual participants with some college education. These analyses illustrated the importance of proceeding w ith the main study using a balanced and modified version of this measure with a larg er sample of adolescent bilingual (SpanishEnglish) and monolingual (English-only) participants. Method: Part 2 Participants: Adolescent Pilo t Study on Familiarity Ratings The second portion of the p ilot study was conducted at a public high school in a rural area of West Central Florida. The pa rticipating high school had a population of 1,633 students at the time of the study. During the 2005-2006 academic year, the student population consisted of 69% Caucasian, 22% Af rican American, 8% Latino and 1% other (e.g., Asian and Indian stude nts) (Polk County Public Schools, 2006). During that academic year enrollment consisted of 518 freshman, 491 sophomores, 358 juniors, and 266 seniors. The inclusion criteria for the adolescents specified that all participants had to be a) currently enrolled in hi gh school, b) between 13-18 yearsold, and c) a self-reported (and teacher-confirmed) native, monolingual E nglish speaker. Anyone who was receiving speech and/or language services for a speech and/or language impairment and was not a native, monolingual English-speaker was excluded from the study. The exclusionary criteria were necessary in order to obtain a sample of typically developing, monolingual English-speaking adolescents.

PAGE 72

61 The final sample consisted of 47 sophomo res. Four were African American and 43 were Caucasian. Of the 47, 18 were male a nd 29 were females. Furthermore, 40 of the 47 participants reported their dates of birth. Of the 40 partic ipants, their mean age was 15 years, 8 months old. Materials: Development of th e Familiarity Rating Form The investigator deemed it important to obtain familiarity ratings for the 48 idioms from a sample that would reflect the demographics of the sample for the main study. To rate the familiarity of the idioms, a rating form was constructed. A first step in devising the rating measure was to consult pa st literature for stra tegies on rating the familiarity of lexical items. Dale (1965) classi fied the extent of word knowledge into four categories: a) never heard it before, b) hear d it, but doesnt know what it means, c) recognizes it in context as having to do w ith _____, and d) knows it well. In a similar manner, Beck, McKeown, and Omanson (1987) described word knowledge as falling on a continuum of: a) no knowledge, b) general sense, c) narrow, context-bound knowledge, d) having knowledge of a word but not being ab le to recall it readily enough to use it in appropriate situations, and e) rich, decontextualized knowle dge of a words meaning, its relationship to other words, and its extension to metaphorical uses. These two paradigms for capturing the re lative nature of word knowledge were extended to conceptualize idiomatic knowle dge as representing a continuum of familiarity. Specifically, each of the 48 idioms (12 familiar transparent, 12 familiar opaque, 12 unfamiliar transparent, and 12 unf amiliar opaque) were listed without any contextual support, along with three columns labeled: a) kno w it, b) heard it, but dont know what it means, and c) never heard it before. This simplified continuum allowed

PAGE 73

62 participants to place a check mark in the co lumn that best described their knowledge of each of the 48 idioms. Each form was numbe red so that students remained anonymous. Procedure After obtaining approval from the Polk County Public Schools (see Appendix F), all participants completed a consent fo rm (see Appendix G). Monolingual Englishspeaking students were recruited through a r eading teachers four classes. The teacher, who distributed and collected the consent forms over a month-long period, announced the studys premise, that the participants would co mplete a short checklist and that there was a chance for one student to win two student movi e tickets in each of the four classes. The consent forms were signed by the participants parents and the particip ants also signed an assent. After giving verbal assent, the participating stud ents were given the following directions orally: Idioms are figurative or non-literal language like raining cats and dogs or bought the farm. I am creating an idiom test and need your help to decide which idioms on my list are familiar to you. When you receive your form, please read each idiom carefully and decide if you a) Know it, b) Heard it before, but do not know what it means, or c) Never hear d it before. Then, just place a check mark under the appropriate column. If at an y time you wish to stop completing this form, you may do so without any conse quences whatsoever. This is completely voluntary. If you have any questions feel free to ask me. After everyone has finished completing his/her form, I wi ll randomly choose a nu mber like the ones listed on your forms. The persons nu mber who corresponds with the number

PAGE 74

63 chosen will receive a pair of movie ticke ts. I would like to thank all of you for participating. The familiarity forms (see Appendix H) were completed independently in groups of approximately 30 students within each of the four classes. All participants completed the familiarity rating form within 10 to 15 minutes on May 12, 2006 (the students who were not participating worked on their class assignment instead). This method expedited the process and there was minimal disruption of the students and teachers schedules. As noted earlier, an incentive, a prize of 2 student movie tickets for each class (8 tickets in total) was raffled upon completion of the forms. As stated on the consent forms, all students who had turned in a consent form were included in the raffle, even those who were absent on May 12th. In each of the four classes, af ter all participants had completed the familiarity form, the students assigne d anonymous numbers were written on small index cards, the numbers were shuffled in a bag, and one randomly drawn number was chosen. The four students received two stude nt movie gift certif icates valued at approximately $11 for each pair. In addition, a small gift certificate of $25 to an office supply store was given to the principal as a token of appreciation for allowing the research to be conducted at his school. Data Analysis Data from each of the four classes were fi rst analyzed separately in order to detect any differences related to class membership. The familiarity ratings (know it; heard it, but dont know what it means; and never heard it be fore) were each assigned a point value. These values ranged from 1 point for a never heard it before response, to 2 points for a heard it, but dont know what it means response, and 3 points for a know it response.

PAGE 75

64 Then, the number of check marks in each column for each idiom for each class was calculated. Each columns tallies were multiplied by 1, 2, or 3, depending on the columns value. Next, each idioms row value was tallied. For example, there were eight participants in the third class. For the idiom blow off steam all eight participants marked the column know it, for a total of 24 points (8 partic ipants x 3 points each) for that particular idiom. After the totals for each class were calc ulated for each idiom, each idioms total value was converted into ratios by dividing the totals by the number of participants in each class. For example, the idiom hold ones head up scored a 19 for the class with 8 participants, so its ratio was a 2.38 (19/8). Tota ls were converted into ratios so that the point totals for each class could be compared regardless of the number of participants in each class. Next, each idioms totals for all classes combined were converted into a ratio by dividing by the total numbe r of participants ( N = 47). For example, the idiom hold ones head up received the following scores: 44, 37, 19 and 20 for a total score of 120. Thus, the ratio for this score was 2.55 (120/47). Tabl e 2.6 displays each idioms total familiarity rating across the 47 participants, in de scending order, per idiom category. Table 2.6 Familiarity Ratings in Descending Order by Idiom Category Familiar Transparent Idiom Familiarity Rating 8. Burning the candle at both ends 1.66 11. Take a shot in the dark 2.23

PAGE 76

65 (Table 2.6 continued) 1. Hold ones head up 2.55 7. Cry over spilled milk 2.60 2. Go by the book 2.75 5. Keep a straight face 2.81 9. Hold your tongue 2.85 12. The early bird catches the worm 2.85 4. Blow off steam 2.89 3. Take someone under ones wing 2.92 6. Right under your nose 2.92 10. Get off on the wrong foot 2.96 Familiar Opaque 20. Go to pot 1.32 21. Wet behind the ears 1.75 15. Paint the town red 1.85 22. Jump through hoops 2.19 24. To flip ones lid 2.34 14. Bring the house down 2.36 16. Have a soft spot in ones heart 2.43 23. Go cold turkey 2.45 19. At the drop of a hat 2.51 17. Chip off the old block 2.62

PAGE 77

66 (Table 2.6 continued) 18. Spill the beans 2.72 13. Beat around the bush 2.89 Unfamiliar Transparent 34. To fall into the apples 1.06 33. To fall down with four horseshoes up in the air 1.12 27. For a good hunger there is no hard bread 1.17 32. Its the water drop that makes the vase overflow 1.21 30. To try to make a hole out of water 1.30 36. To throw flowers to somebody 1.36 35. To cut a pear in two 1.47 28. To shoot sparrows with cannons 1.49 26. To run around like scalded pigs 1.49 31. To hold someones leg 1.53 29. To be drowning in a glass of water 1.57 25. To be caught between two fires 1.89 Unfamiliar Opaque 37. The turtle is shrouded 1.02 38. To eat the leaf 1.06 41. To have salt in your pumpkin 1.06 44. To pick up a log 1.06 45. To eat on the thumb 1.06 42. To whistle in your thumb 1.11

PAGE 78

67 (Table 2.6 continued) 48. To lay a rabbit on someone 1.13 39. To pet the horse first 1.17 40. To be at the green 1.23 47. Between dog and wolf 1.34 43. To put ones finger into ones eye 1.54 It was then determined, from a qualitati ve perspective, that any score, which equaled or fell below two points (i.e., hear d it, but dont know what it means) would be considered unfamiliar. In other words, for the particular idiom, most of the participants had marked it as heard it, but do not know what it means or never heard it before. Following a similar procedure, any score a bove two points was qual itatively considered as familiar. Results All of the unfamiliar idioms were rated as unfamiliar by all four classes, with ratings ranging from 1.02 to 1.89. All but four of the familiar idioms were rated as familiar by all four classes, with ratings ranging from 2.19 to 2.96. The four idioms that were rated as unfamiliar were: a) burning the candle at both ends (ratio=1.66), b) paint the town red (ratio=1.85), c) go to pot (ratio=1.32), and d) wet behind the ears (ratio=1.75). Interestingly, in comparing the responses of the 34 undergraduates from the first pilot study with the 47 adolescent s in the second pilot study for these four idioms, 10 of the undergraduates (five of whom were monolingual Englis h speakers) also did not know

PAGE 79

68 the meaning of burning the candle at both ends when presented out of context. Instead, they interpreted its meaning as to not be wasteful rather than the correct interpretation to work and/or play too hard without enough rest. Also, 12 undergraduates (four of whom were monolinguals) misinterpreted the idiom paint the town red when presented out of context as to make everyone in town mad rather than the correct interpretation to go out and celebrate. There were similar difficulties with go to pot out of context. Seven of the undergraduates (two of whom were monolingual) misinterpreted the idiom as meaning to put in the trashcan rather than to deteriorate. However, for wet behind the ears, all but one monolingual undergraduate correctly interpreted it, while the adolescent sample reported low familiarity (see Table 2.6, idiom #21). These differences in familiari ty may have been unforeseen generational and/or regional differences between the examiner and some participants. To eliminate any familiar idioms that were interpreted as too unfamiliar or too familiar, or any unfamiliar idioms that were rated as too familiar or too unfamiliar, the two idioms in each category with the highest rating and the lowest rating were eliminated as a way to control for ceiling and floor e ffects within each category. In the familiar transparent idiom category, burning the candle at both ends (familiarity rating = 1.66) and get off on the wrong foot (familiarity rating = 2.96) were eliminated. Go to pot (familiarity rating = 1.32) and beat around the bush (familiarity rating = 2.89) were both eliminated from the familiar opaque idiom category. In the unfamiliar idiom category, two transparent idioms were eliminated: to fall into the apples (familiarity rating=1.06) and to be caught between two fires (familiarity rating=1.89). Finally, two unfamiliar,

PAGE 80

69 opaque idioms were eliminated: the turtle is shrouded (familiarity rating=1.02) and to put a finger in ones eye (familiarity rating=1.54) After eliminating the eight idioms, a t-test was conducted to compare the familiarity ratings of the remaining 20 familiar idioms with the remaining familiarity ratings for the 20 unfamiliar idioms. The combined familiarity ratings for the transparent familiar and opaque familiar idioms had a mean of 2.53 (SD = 0.34). The combined familiarity ratings for transparent unfamiliar and opaque unfamiliar idioms had a mean of 1.26 ( SD = 0.17). The t-test results indicated a statistica lly significant difference ( p<0.001) between the familiarity ratings for the familiar and unfamiliar idioms with an observed t value of 14.95 and a critical t value of +/2.021. After eliminating these eight idioms, a box plot comparing the familiarity ratings for the familiar (including the 10 transpar ent and 10 opaque) and unfamiliar (including the 10 transparent and 10 opaque) idioms demo nstrated an absence of overlap between and the two familiarity categories (see Figure 9). (A previous box plot that included the eight idioms did demonstrate overlap.) This absence of overlap provided justification for the conclusion that the two familiarity cate gories represented local norms and were not arbitrary divisions based solely on previous research, including cross-linguistic studies.

PAGE 81

70 3 + | | | | + + 2.8 + | | | | | | | | | | | 2.6 + * | | + | | | | | | | 2.4 + | | | + + | | | | 2.2 + | | | | | | | 2 + | | | | | | | 1.8 + | | | | | 1.6 + | | | | | | 1.4 + + + | | | | | | | | + | 1.2 + * | | | | + + | | 1 + + + Familiar Idioms Unfamiliar Idioms Figure 9. Boxplot of familiarity ratings (ranging fr om 1-3) as a function of familiar and unfamiliar idioms.

PAGE 82

71 Discussion The adolescent pilot study confirmed the familiarity of the familiar (i.e., familiar, American) idioms and the unfamiliarity of the unfamiliar (i.e., unfamiliar, European) idioms. More importantly, through the adoles cent pilot study, local normative data on idiom familiarity were collected. Therefore, the labels familiar and unfamiliar were no longer arbitrary categories. Fu rthermore, the adolescent part icipants in the pilot study matched the participants in the main study demographically. Construction of the Final Idiom Comprehension Measure Balancing Items and Syllable Length To minimize memory retention due to re petition of the same idiom inand then out-of-context during the main study, the item s presented out-of-context differed from the items presented in-context. That is to say, in the main study, one participant received items #1-5 in-context and items #6-10 out-of-cont ext, while another pa rticipant received items #1-5 out-of-context and items #6-10 in co ntext so that #1 was balanced with item #6. In constructing the final idiom meas ure, each idioms syllable length was calculated. These syllable counts ranged from 3 to 11 syllables in length (see Table 2.7). The items were first matched in terms of each idioms syllable length (see Table 2.8). In other words, an opaque familiar idiom consisting of four syllables was matched with another opaque familiar idiom consisting of four syllables. The rationale for this procedure was to match the time it takes to read an idiom on Form A and Form B equivalent. This procedure al so allowed for balance in th e length of items #1-5 on one form and items #1-5 on the other form; hence, the idiom comprehension measure (ICM)

PAGE 83

72 consisted of Form A and Form B (see A ppendix I). The purpose of constructing two forms was to avoid confounding the measure by controlling and counterbalancing the order of the items presented in and out of context. Forms A and B also allowed for participants responses to be compared accu rately. The out-of-context items in Form A totaled 117 syllables, while the out-of-context idioms in Form B to taled 119 syllables. Table 2.7 Syllable Counts for All Idioms Idiom Syllable Counts 1. Hold ones head up 4 2. Go by the book 4 3. Take someone under ones wing 7 4. Blow off steam 3 5. Keep a straight face 4 6. Right under my nose 5 7. Cried over spilled milk 5 8. Hold your tongue 3 9. Take a shot in the dark 6 10. The early bird catches the worm 8 11. Bring the house down 4 12. Paint the town red 4 13. Have a soft spot in ones heart 7 14. Chip off the old block 5 15. Spill the beans 3

PAGE 84

73 (Table 2.7 continued) 16. At the drop of the hat 6 17. Wet behind the ears 5 18. Jump through hoops 3 19. Go cold turkey 4 20. To flip ones lid 4 21. To run around like scalded pigs 8 22. For a good hunger there is no hard bread 10 23. To shoot sparrows with cannons 7 24. To be drowning in a glass of water 10 25. To try to make a hole in water 9 26. To hold someones leg 5 27. Its the water drop that makes the vase overflow 11 28. To fall down with four horseshoes up in the air 11 29. To cut a pear in two 6 30. To throw flowers to somebody 8 31. To eat the leaf 4 32. To pet the horse first 5 33. To be at the green 5 34. To have salt in your pumpkin 7 35. To whistle in your thumb 6 36. To pick up a log 5 37. To eat on the thumb 5

PAGE 85

74 (Table 2.7 continued) 38. To play the donkey to get bran 8 39. Between dog and wolf 5 40. To lay a rabbit on someone 8 Table 2.8 Syllable, Familiarity, and Condition Match: Form A and Form B Form A: Syllable Count/Familiarity Rating Form B: Syllable Count/Familiarity Rating _____________________________________________________________________ Familiar Transparent 24 syllables total Familiar Transparent 25 syllables total 1. 4/2.55 2. 4/2.75 3. 7/2.92 10. 8/2.85 8. 3/2.85 4. 3/2.89 6. 5/2.92 9. 6/2.23 7. 5/2.60 5. 4/2.81 Familiar Opaque 22 syllables total Familiar Opaque 23 syllables total 11. 4/2.36 20. 4/2.34 12. 4/1.85 17. 5/1.75 16. 6/2.51 13. 7/2.43 14. 5/2.62 19. 4/2.45 18. 3/2.19 15. 3/2.72 Unfamiliar Transparent 43 syllables total Unfamiliar Transparent 41 syllables total 24. 10/1.57 22. 10/1.17 28. 11/1.12 27. 11/1/21

PAGE 86

75 (Table 2.8 continued) 25. 9/1.30 21. 8/1.49 23. 7/1.49 30. 7/1.36 29. 6/1.47 26. 5/1.53 Unfamiliar Opaque 28 syllables total Unfamiliar Opaque 30 syllables total 36. 5/1.06 37. 5/1.06 38. 8/1.23 40. 8/1.13 33. 5/1.23 39. 5/1.34 35. 6/1.11 34. 7/1.06 31. 4/1.06 32. 5/1.17 Note. Form A had a total of 117 syllables a nd Form B had a total of 119 syllables. Balancing Familiarity Ratings Next, Forms A and B were matched based on the previously described familiarity ratings (see Table 2.8). To ensure that the tw o forms were balanced in terms of syllable number and familiarity of the idioms, two separate t -tests were conducted. The first t -test examined the number of syllables in each fo rm. The 117 syllables in Form A and the 119 syllables in Form B were determined to not significantly differ (see Appendix B for a list of non-significant t -test values). The second t -test also confirmed that there were no significant differences between Form A and Form B in terms of familiarity ratings (see Appendix B). Therefore, the two forms were balanced in terms of syllable count and idiom familiarity. Moreover, Form A presented the same 20 idioms in-context as Form B did out-of-context; and Form B presented the same idioms in-context as Form A did outof-context.

PAGE 87

76 Balancing Response Accuracy Lastly, the accuracy totals of each idiom on each form were calculated. Since there were 34 participants in the first pilot study, the possible scores ranged from 0 (no one answered correctly) to 34 (all answered correctly). Both of the forms 40 questions accuracy totals were calculated. Form A had a mean accuracy of 29.58 (SD = 7.87). Form B had a mean accuracy of 29.75 (SD = 7.34). Not surprisingly, a t -test demonstrated that the two forms were not significantly different (see Appendix B), indicating that the level of difficulty of the items on Form A did not significantly differ from the level of difficulty of the items on Forms B. Two final modifications were conducted before going forth with the main study. First, the pronoun one/ones in the original measure was changed to you/your in the final measure in order to increase the level of readability within an adolescent population. Secondly, in the final analyses of the idiom measure two idioms were deemed to have transparent translations from English to Spanish: Its the water drop that makes the vase overflow and To be drowning in a glass of water Therefore, these two transparent, unfamiliar, Spanish-derived idioms were repl aced by two transparent, unfamiliar, Frenchderived idioms: To be a monkey on a branch and To put on the sails To be a monkey on a branch has 8 syllables and replaced the Spanish idiom Its the water drop that makes the vase overflow (13 syllables); and To put on the sails has 5 syllables and replaced the Spanish idiom To be drowning in a glass of water (10 syllables). Before replacement, Form A had 117 syllables and afterwards had 112 syllables; and Form B had 119 syllables, and afterwards had 114 syllables. The two forms were balanced in familiarity,

PAGE 88

77 semantic transparency, syllable length, and readability level when presented within context.

PAGE 89

78 Chapter Three Method Experimental Design This study utilized a mixed quasi-experim ental design with be tweenand withinsubject variables. The between-subject va riable was language group (Spanish/Englishspeaking bilingual or English-speaking monoli ngual) and there were six within-subjects variables: a) idiom familiarity (familiar and unfamiliar), b) idiomatic semantic transparency (transparent and opaque), c) idiom context (with a nd without short-story contexts), d) a reading comprehension ta sk, e) a comprehension monitoring (error detection) task, and f) a multiple meani ng (synonym) vocabulary task. The dependent variables included the scores from the idio m measure, the reading comprehension task, the comprehension monitoring task, and the multiple meanings vocabulary task. Participants Sample and School Characteristics Sample size and characteristics. The study was conducted in the spring semester of the 2006-2007 school year. Thirty-one high sc hool students for each language group were recruited (N = 62). There were 14 m onolingual males, 17 monolingual females, 12 bilingual males, and 19 bilingual females in the sample. All participants were Caucasian, except for two monolingual par ticipants (one male and one female) who were African American. In the monolingual group there were four 9th graders, 19 10th graders, four 11th

PAGE 90

79 graders, and four 12th graders. In the bilingual group there were five 9th graders, 15 10th graders, eight 11th graders, and three 12th graders. The participants ranged in age from 14years-old to 18; 6years-old (years; months). The monolingua ls ranged in age from 14; 10 18; 6 years old ( M = 16; 4, SD=0.97) and the bilinguals ranged in age from 14; 9 17; 8 years old ( M = 16; 8, SD= 1.03 ). There was not a significant difference in ages between the monolingual and bilingual langu age groups according to the results of an independent t test (see Appendix B). Sample school characteristics. Requests to conduct research in the Polk County Public Schools (PCPS) as well as in th e School District of Hillsborough County (SDHC) were approved (see Appendices J and K). The student populations of both school districts were diverse. For example, the PCPS student body was approximately 55 percent Caucasian, 23 percent African American, 20 percent Latino, and 2 percent other (Polk County Public Schools, 2007). Furthermore, there were more than 5,000 students whose primary language is other than English in the PCPS (Polk County Public Schools, 2007). The SDHC student body was approximately 44 percent Caucasian, 22 percent African American, 26 percent Latino, 2 percent As ian/Pacific Islander, 0.3 percent Native American, and about 6 percent other (School District of Hillsborough County, 2005). More than 10 percent of th e student population in the SC HD spoke a language other than English (School District of Hillsborough County, 2005). All participants enro lled in the study were current high school students within the public schools of either Polk or Hillsborough County. These school districts were located in west central Florida. The three participating schools were all in rural areas within the two counties. A total of eight monolingual a nd nine bilingual partic ipants were attending

PAGE 91

80 public schools in Polk County and 23 monolingual participants and 22 bilingual participants were attending a public school in Hillsborough County. All SDHC participants attended the same school. Students from two schools in PCPS participated, eight monolinguals and three bilinguals from one high school and six bilingual participants came fr om a second school. Recruitment To recruit participants, high school principals, Eng lish as a Second Language (ESOL) instructors, general education t eachers, and reading teachers assisted in identifying students who met the studys inclusio nary criteria. The pr incipal investigator visited the classrooms of part icipating teachers for three purposes: a) to explain the voluntary nature of the study; b) to distribute parent consent forms (see Appendix L) in English and Spanish; and c) to describe the incentives for part icipation. Specifically, upon completion of the study, five student part icipants from each school won two movie tickets (together worth approximately $15). Th e movie tickets were raffled following the same procedure used in the pilot study. The students were asked to return the pa rent consent forms within one week of disbursement if they chose to participate. In addition, each partic ipant also signed an assent form (see Appendix M) on the day he/s he participated in the study. The principal investigator distributed consent forms in at least three waves at each participating school in order to increase enrollment. Inclusion Criteria All participants had to meet four general inclusion criteria: a) have self-reported normal or aided hearing adequate for understand ing oral directions, b) have self-reported

PAGE 92

81 adequate vision (i.e., normal or corrected) to read at least 12 point font, c) not be receiving or eligible for speech and language or special education services (via self-report and teacher confirmation), and d) pass a nonw ord reading task (see Appendix N for the Inclusion Questionnaire). In addition to these four inclusion criter ia, the bilingual participants had to meet an ethnicity criterion and two educational cr iteria. Firstly, they had to be of Latino descent; that is, they must have been born in a Spanish-speaking country or have been born in the United States to parents/guardians/live-in family who were Latino and spoke Spanish. Secondly, they must have been enrolled full-time in U.S. schools since elementary school and had been exited from any English as a Second Language (ESOL) program. This information was discerned from the student language history questionnaire (see Appendix O). Previous bilingual rese arch (e.g., Kohnert & Bates, 2002; Kohnert, Bates, & Hernandez, 1999) indicated that students who spoke English as a second language and had four to eight years of Eng lish experience in English-only educational programs, have sufficient language skills to enable them to participate fully in experimental tasks in English (Windsor & Kohnert, 2004, p. 881). Procedure Measures Word attack. As one of four inclusion criterion, all participants had to pass the Word Attack Subtest from the Woodcock J ohnson III Test of Achievement (WJ-III ACH; Woodcock et al., 2001). This subtest measures each participants ability to read psuedowords that are linguisti cally similar to English word s. For instance, the words gusp, thrept, and malfreatsun are all pseudowords presented in the Word Attack subtest.

PAGE 93

82 The median reliability coefficient alphas for all age groups for the standard battery of the WJ-III ACH ranged from .81 to .94 with monolingual English speaking participants ranging from kindergarten to univers ity students (Woodcock et al., 2001). The rationale for using this subtest as an inclusion criterion was to exclude any participant who was unable to decode psuedow ords fluently and accurately, which might indicate reading difficulties th at would then influence text comprehension in a negative way. If a participant had not passed the Word A ttack Subtest, then he or she would have been excluded from the study and returned to cl ass. All participants passed this measure. Idiom comprehension measure. The development of the idiom comprehension measure was described in the previous ch apter, which focused on the pilot study. The methodology used to match Forms A and B (s ee Appendix I) is also discussed, including matching the forms based on idiom syllable length, item difficulty, and familiarity ratings. Several t -tests demonstrated that the forms were not significantly different from one another (see Appendix B). Results from the pilot study led to the conclusion that the idiom comprehension measure was a valid instrument. The final Forms A and B each presented fi ve familiar transparent idioms, five familiar opaque idioms, five unfamiliar transparent idioms, and five unfamiliar opaque idioms. Each idiom was presented in and out of context (i.e., a short story), for a total of 40 items per form. Possible selections were multiple choices in order to minimize oral language production, with three c hoices per idiom. Of the three answers, one was a literal but an inaccurate translation of the idiom, one was a figurative and correct translation of the idiom, and the third was figuratively re lated to the idiom but incorrect. To psuedo randomize the order of the items (e.g., so that all familiar, transparent idioms were not

PAGE 94

83 clustered), 10 forms of each Form A and Form B were created by randomizing the items on Microsoft Excel and then analyzing to ensu re that the categories were not clustered. This randomization should have prevented any order effects. Passage comprehension. The Passage Comprehension Subtest from the WJ-III ACH (Woodcock et al., 2001) wa s administered in English in order to assess English reading comprehension. A version of this su btest was also used in a similar manner by Proctor, Carlo, August, and Snow (2005) to as sess the English reading comprehension of school-age, Spanish-speaking children. The medi an reliability coefficient alphas for all age groups for the standard battery of the WJ-III ACH ranged from .81 to .94 with monolingual English speaking participants ra nging from kindergarten to university students (Woodcock et al., 2001). The WJ-III ACH Passage Comprehension Subt est, which is read silently, is a clozereading task, organized hierarchically from less to more complex passages. After reading the passage, the participant must orally fill-in the sentence with the appropriate lexical or syntactic choice. An analysis was conducted of th e figurative language used in the WJ-III ACH Passage Comprehension Subtest. There were no instances of idiomatic language use. Comprehension monitoring task. Morrisons (2004) erro r detection task was modified for this study as a measure of E nglish comprehension monitoring abilities in monolingual (English-only) and bilingual (Spanish-English) adolescents (see Appendix P). Five short stories about animals and ge ographic locations were chosen from an educational website ( http://www.educationworld.com) a nd each of these stories was divided into four parts each. E ach part was com prised of one to three sentences. After

PAGE 95

84 each part the participant was asked explic itly whether there was an error in the sentence(s) and, if so, to unde rline the error. There were a total of 20 questions, four questions for each of the five stories. There we re two errors per story (one microand one macro-error) for a total of 10 errors. The other 10 parts contained no errors. Consistent with Morrisons (2004) methodol ogy, deliberate errors were inserted at the microand macro-levels. A micro-level error involved a graphemic error in a word such as a misspelling (e.g., layd for laid ), an incorrect morphological ending (e.g., tallness for tallest ), or the use of an incorrect homophone (e.g., knight for night ) (Morrison, 2004). To detect a micro-level e rror, readers must comprehend accurately at the proposition level, as desc ribed in Kintschs (1998) mean ing-construction phase (see Figure 2). For example, this is a sentence f ound in the story about deserts with a microlevel error embedded: Parts of this desert will not see a single drop of reign this year. Kintschs integration phase is assessed by the reader making sense of the text as a whole, detecting any inconsis tencies or errors at the macro-level when propositions are integrated. An example of a macro-level error was the violation of the internal consistency of a text by including words that contradict information found in preceding or following sentences (Morrison, 2004). For example, after the re ader was told that Greece was about the size of the state of Alabama, the reader must identify the embedded macro-level error in this sentence: The United States, which is tiny when compared to Greece, has 12,300 miles of coastline. A readability level was also calculated using the Dale-Chall New Readability Formula (Chall & Dale, 1995). This formula estimated text difficulty based on the semantic and syntactic difficult y, which together correlate ( r =.92) with reading

PAGE 96

85 comprehension scores on cloze readings (Chall & Dale, 1995). First, two 100 word samples from the selected reading were assessed for the number of unfamiliar words using the Dale-Chall familiarity criteria. Then, the number of complete sentences was counted for each sample. Applying the form ula, a seventh grade reading level was calculated for the text in the short stor ies, including the set of directions. Multiple meanings vocabulary: Synonyms. The Reading Vocabulary Subtest from the WJ-III ACH (Woodcock et al., 2001) assesses the ability to read words and supply the words appropriate meanings. In the synony m task, a word similar in meaning to each written word must be selected. For example, in response to the item haul, any of these equivalent meanings would be synonymous: carry, pull, drag, or tote. If the response was move, take, or bring, it would be considered incorrect The subtest was administered according to manual instructions and a raw sc ore was obtained. Woodc ock et al. (2001) reported that the median reliability coefficien t alphas for all age groups for the standard battery of the WJ-III ACH ranged from .81 to .94 with monolingual English speaking participants ranging from kinderg arten to university students. Student language hi story questionnaire. All bilingual participants were interviewed by the principal investigator re garding their language histories and then completed the language history questionnaire. The questionnaire first consisted of questions related to each bilingual participan ts country of origin, age of acquisition of English and Spanish, educational background, and frequency of Spanish production and comprehension in the home environmen t. The remainder of the questions ( n= 20) independently answered by the participants were declarative statements followed by the

PAGE 97

86 same six-point likert scale, which ranged from never (0) to always (5). For example, one question was: I speak to my mother/g uardian in Spanish. The purpose of the language history questionnaire (see Appendix O) was to obtain a quantitative score for each bilingual participant by tallyi ng each partic ipants likert scale responses. Total scores were on a continuum with lower scores representing less daily use of the Spanish language and higher scores representing more daily use of the Spanish language. The relationship between the scores and performance on the other measures were analyzed. Administration and Scoring All participants were assessed indepe ndently on their school campuses, during school hours. Administration of all tasks took approximately 45 minutes for each participant to complete if he/she was monoli ngual to one hour if he/she was bilingual. To minimize attrition, each participant attended only one testing session. Furthermore, all measures were given in English, the participants academic language. After each participant met the aforem entioned inclusion criteria (see Appendix N), the principal investigator gave the following directions orally: I am studying students at your high school to learn more about how students read. I am studying reading comprehension by looking at four different tasks. We will be completing the four small tasks next. I will give you directions before we start each task. Your participation in this study is completely voluntary. If at any time you wish to stop and/or withdraw from th e study, please say so. If you need a break at anytime, please let me know that as well. Thank you for taking the time to participate in this study.

PAGE 98

87 Word attack. The standardized directions found in the WJ-III ACH (Woodcock et al., 2001) manual were utilized for administration of this subtest: I want you to read some words that are not real wo rds. Tell me how they sound. Following two sample items printed on a flip book, items were administered until the six highest-numbered items were failed, or until the page with the last item ha d been administered. There were 29 items in addition to the two sample items. If there was no response, the pr incipal investigator encouraged a response. If there was still no resp onse, the investigator pointed to the next word, as stated in the WJ-III ACH manual. To remain in the study, all participants obtained a grade equivalent score of at leas t grade 9, which meant that they all scored more than 27 points. Idiom comprehension measure. The following directions were given orally: Words and phrases can have several meani ngs. Read these phrases and circle the answer that means the same. Some phrases wi ll be in a short story, and some will be alone. There is only one answer for each question. If at any time you want to stop the task, you may do so without any consequences whatsoever. This is completely voluntary. If you have an y questions, feel free to ask me. Participants completed a randomized vers ion of either Form A or Form B. Individual assigned research numbers were written on each participants form in order to keep the data anonymous. The task was counterba lanced so that half of the bilinguals and half of the monolinguals completed Form A a nd the remaining halves completed Form B. Lastly, the presentation of Form A and Form B were counterbalanced so that the first bilingual participant r eceived Form A, the second bilingual participant rece ived Form B, the first monolingual particip ant received Form A, the s econd monolingual participant

PAGE 99

88 received Form B, and so on. Responses were counted as correct or incorrect, and incorrect responses were judge d as literal or figurative fo r a qualitative analysis. Passage comprehension. Each participant silently read passages printed on the flip book. Only one word responses were acceptable. There was no penalty for any mispronunciations caused by dialect or region al speech patterns (Woodcock et al., 2001). In accordance with the WJ-III ACH manual (Woodcock et al., 2001), the principal investigator tested by complete pages until the six lowest-numbered items administered were correct. Testing continued until the six highest-numbere d items administered were failed, or until the page with the last ite m had been administered. The WJ-III ACH scoring methods were used to score items as correct or incorrect such that responses were accepted as correct when they differed from th e manuals responses only in verb tense or number (singular/plural). Monitoring comprehension. The following explicit written instructions were given to find all errors in the reading passage first, and then to underline these errors: You will read five factual, short stories. Each story is about something different like an animal or a place. Your job is to look for errors in the stories. Some sentences in these stories may have errors and some may not. Some examples of errors are misspellings, incorrect verbs, a nd ideas that do not make sense with the rest of the story. For example, look at the errors underlined below: To make a peanut butter and jelly sand wich you need bread, peanut butter, and jelly. You will also need a nife to spread the peanut butter and jelly, as well as to kut the cake in two. Some people also prefer the crusts to be cut off. Either way,

PAGE 100

89 peanut butter and jelly sandwiches are messy, so you will not need a napkin. Peanut butter and jelly sandwiches is very popular. There are two misspelled words underlined: kut/cut and nife/knife. Also, there are two examples of ideas that do not go with the rest of the story. The story is about making a peanut butter and jelly sandwich, so cutting a cake does not fit with the storys main idea. The next error is the word not The reading first says that peanut butter and jelly sandwiches are messy, so it should say that you will need a napkin instead of saying you will not need a napkin. The last error is should say are since sandwiches is in the plural form meaning more than one. Now it is your turn to find these types of errors in each of the five stories below. After reading every one or two sentences in each story you will be asked if there is an error and, if there is, to underline it. Remember that not all of the sentences will have an error. Underlined errors that were not actual errors were not counted in terms of scores. Multiple meanings vocabulary: Synonyms. Administration and scoring followed the manual instructions (Woodcock et al., 2001): Read each of these words out loud and tell me another word that means the same. After giving each part icipant the two sample questions, the remaining 26 items were admi nistered. Testing continued until the four highest-numbered items administered were faile d, or until the last test item had been administered. Only one-word responses were accepted. If a two or more word response was produced, the follow-up request was for a one-word answer. To be counted as

PAGE 101

90 correct, the response must be identical to th e response given in the manual, and may only differ in verb tense or number (singular/ plural). Again, there was no penalty for any mispronunciations caused by dialect or regional speech patterns (Woodcock et al., 2001). Student language hi story questionnaire. Only the bilingual participants were interviewed with the student language history questionnaire. The principal investigator completed page one in conjunction with each bi lingual participant to ensure participants fit the bilingual inclusion criteria. Then, each participant completed the remainder of the questionnaire on his/her ow n. The following written di rections were given: Please read each statement carefully a nd circle the number/word that best describes your answer. If the question does not apply to you, please circle the number of the question. Each participants responses were added together for a total score in order to conduct a quantitative analysis. Participants were classified into one of two categories ba sed on their age of acquisition of English and Span ish: a) simultaneous or b) sequential. These categories were derived from the Fleege, MacKay, and Piske (2002) categories employed partly for estimating participants language dominance. Using this categoriza tion procedure, any statistically significant differences w ithin the bilingual group were explored.

PAGE 102

91 Chapter Four Results The conceptual framework of this stu dy is based on the global elaboration model (GEM) (Levorato et al., 2004). The GEM posits that semantic analysis is utilized to interpret unfamiliar transparent idioms but the surrounding context must be exploited to interpret unfamiliar opaque idioms. Young children rely heavily on semantic analysis (local coherence) to comprehend unknown idio ms, whether transparent or opaque. They gradually develop the ability to make in ferences and create global coherence by exploiting the surrounding cont ext. Idiom comprehension, li nguistic development, and reading comprehension appear to develop in tandem (Cain et al., 2005; Levorato et al., 2007). Thus, the speculation was that more-ski lled readers/comprehenders would be more adept than less-skilled readers/ comprehenders at creating the global coherence that results in accurate idiomatic interpretations. Major Aims Three major aims guided the investiga tion of the psychological reality of the GEM for a bilingual and monolingual adolescent population. The first aim related to the model illustrated in Figure 3. This model incorporated three underlying principles: (a) inferencing from the microto macroleve l; b) monitoring of co mprehension; and c) choosing appropriate constituent meanings from various possible meanings. The predictive value of these three principles fo r accurate idiom comprehension was tested.

PAGE 103

92 The three principles were operationalized as a) a reading comprehension task, b) an error detection task, and c) a synonym task to pr edict performance on th e idiom comprehension measure. The second aim focused on the interacti ons among language group (bilinguals and monolinguals) and performance on the idiom comprehension measure as a result of varying idiomatic familiarity, semantic transparency, and contextual support. The third aim was to investigate group differences with in the bilingual group. The results related to each of the three aims are presented se quentially followed by a summary of main findings. Testing the Model: Aim 1 The first aim of the study was to determin e the extent to which each of the three variables (a reading comprehension task, an error detection task, and a synonym task) predicted the criterion variable, idiom comp rehension accuracy. It was hypothesized that the performance on the three measures woul d predict the performance on the fourth measure, idiom comprehension, for both the bilingual and monolingual groups. Collapsing across groups: Simultaneous multiple regression. The first aim was to investigate the overall predictive power of each measure for idiom comprehension accuracy. To address the first aim directly, th e language groups were collapsed and then a multiple regression was conducted utilizing the scores on the reading comprehension, error detection, and synonym tasks as predictors for the scores on the fourth task (idiom comprehension). This multiple regression used the simultaneous method where all three predictor variables were weighted equally and ente red into a single model, simultaneously. The result was a statistically significant fit, F(3, 61) = 11.169, MS =

PAGE 104

93 59.499, p<.05. The Adjusted R Square for this model was .333, meaning that the model accounted for approximately 33% of the varian ce in the idiom comprehension scores. All four tasks were correlated with values ranging from 0.427 to 0. 533. These Pearson Correlations are pres ented in Table 4.1. Significant variables are displayed in Table 4.2. Beta values represent the change in th e outcome resulting from a unit change in the predictor (Field, 2000, p. 114). The beta valu e of the error detection task (= 0.351*) was statistically significant, indicating that this predictor variable ha d the greatest impact on the criterion variable (the idiom comprehension scores). Fu rthermore, the positive beta value demonstrated a positive relationship between the error detection task and the idiom comprehension measure. Table 4.1 Means, Standard Deviations, and In tercorrelations for Participants ( N = 62) Idiom Comprehension and Predictor Variables Variable M SD 1 2 3 Idiom comprehension measure 33.48 2.83 .427* .490* .533* Predictor variable 1. Synonym Task 15.79 2.27 -.532* .410* 2. Reading Comprehension Task 36.55 3.26 -.514* 3. Error Detection Task 7.90 2.23 -*p <.01

PAGE 105

94 Table 4.2 Simultaneous Regression Analysis Summary for Three Variables Predicting Idiom Comprehension Variable B SEB Synonym Task .207 .157 .166 Reading Comprehension Task .192 .116 .221 Error Detection Task .445 .157 .351* Note. R2 = .366 (N = 62, p < .01). *p<.01 Contribution of language group variance: Simultaneous multiple regression. A multiple regression was then conducted usi ng group as a fourth predictor of idiom comprehension. The purpose was to investigat e the predictive power of language group membership on idiom comprehension performance. Using the simultaneous method, a statistically significant m odel F(4, 57) = 8.236, MS = 44.636, p<.05 emerged. The Adjusted R Square for this model was .322; that is, the model accounted for approximately 32% of the variance in the id iom comprehension scores. Moreover, when compared to the previous three predictor analysis (Adjusted R Square = 33%), the amount of variance did not appear to cha nge when group was added as the fourth predictor. Group was negatively correlated with all other variables, with values ranging from -0.299 to -0. 494. Since the group vari able was binary (1= monolinguals, 2 = bilinguals) and previous anal yses had shown that monoli nguals performed significantly better on each of these four measures than did the bilinguals, this strong, negative correlation may indicate that, as group me mbership approached 1.0 (1.0 represented

PAGE 106

95 monolingual membership), the scores increa sed. The Pearson Correlations are presented in Table 4.3 and significant vari ables are displayed in Table 4.4. The error detection task had a significant beta ( = .348*) This indicated that the erro r detection variable still had the greatest impact on idiom comprehensi on performance, even with language group added as the fourth variable. Table 4.3 Means, Standard Deviations, and Inte rcorrelations for Participants Idiom Comprehension and Predictor Va riables Including Language Group Variable M SD 1 2 3 4 Idiom comprehension measure 33.48 2.83 -.299* .427* .490* .533* Predictor variable 1. Group 1.50 .50 --.494* -.319* -.394* 2. Synonym Task 15.79 2.27 -.532* .410* 3. Reading Comprehension Task 36.55 3.26 -.514* 4. Error Detection Task 7.90 2.23 -*p <.01 Table 4.4 Simultaneous Regression Analysis Summary for Four Variables Predicting Idiom Comprehension Variable B SEB Group -.066 .701 -.012 Synonym Task .201 .170 .161 Reading Comprehension Task .192 .117 .221

PAGE 107

96 (Table 4.4 continued) Error Detection Task .441 .163 .348* Note. R2 = .366 (N = 62, p < .01). *p<.01 Language group as criterion variable: Simultaneous multiple regression. An additional multiple regression was then performed which treated language group as the criterion variable and the ot her four variables (idiom co mprehension, error detection, passage comprehension, and the synonym task) as the four predictor variables. This analysis allowed examination of whether the four tasks could predict language group membership (monolingual or bilingual). A simultaneous multiple regression demonstrated that the model was significant, F(4, 57) = 5.777, MS = 1.118, p<.05. The Adjusted R Square for this model was .239. These four predictors accounted for approximately 24% of th e variance in the criterion variable (language group). The Pearson Correlations are presented in Table 4.5. Significant variables are displayed in Table 4.6. The synonym task had a significant beta value ( = -.405*), which implie d that this task had the gr eatest impact on the criterion variable (language group membership). Appendix Q displays synonym item accuracy for the bilingual and monolingual language groups. Table 4.5 Means, Standard Deviations, and Intercorre lations for Participants Language Group and Predictor Variables Including Idiom Comprehension Variable M SD 1 2 3 4 Language Group 1.50 .50 -.494* -.319* .394*-.299*

PAGE 108

97 (Table 4.5 continued) Predictor variable 1. Synonym Task 15.79 2.27 -.532* .410* .427* 2. Reading Comprehension Task 36.55 3.26 -.514* .490* 3. Error Detection Task 7.90 2.23 -.533* 4. Idiom comprehension measure 33.48 2.83 -*p <.01 Table 4.6 Simultaneous Regression Analysis Summary for Four Variables Predicting Language Group Variable B SEB Synonym Task -.090 .030 -.405* Reading Comprehension Task .003 .023 .022 Error Detection Task -.052 .032 -.232 Idiom Comprehension Task -.002 .025 -.013 Note. R2 = .288 (N = 62, p < .01). *p<.01 Bilinguals only: Simultane ous multiple regression. Lastly, two final simultaneous multiple regressions were conducted by analyzing each language group separately in order to determine if there were different significant variables predicting idiom comprehension accuracy. First, the data from the bilingual language group were entered into a simultaneous multiple regression that treated idiom comprehension as the criterion variable and the other three linguistic vari ables (error detection, passage comprehension,

PAGE 109

98 and the synonym task) as the pr edictor variables. This analysis allowed examination of whether the three tasks could predict idiom comprehension ac curacy for the bilinguals. A simultaneous multiple regression demonstrated that the model was significant, F(3, 30) = 4.749, MS = 35.356, p<.01. The Adjusted R Square for this model was .273. These three predictors accounted for approxima tely 27% of the variance in the criterion variable (idiom comprehension accuracy). The Pearson Correlations are presented in Table 4.7. Significant variables are displayed in Table 4.8. There were no significant beta values and all variables were positively correlated. In contra st with the other linguistic variables, the synonym task was not signifi cantly correlated with the other linguistic variables aside from the idiom comprehension measure. Table 4.7 Means, Standard Deviations, and Intercorre lations for Bilingual Participants Idiom Comprehension Accuracy and Thr ee Linguistic Predictor Variables Variable M SD 1 2 3 Idiom Comprehension 32.65 3.19 .265* .528* .487* Predictor variable 1. Synonym Task 14.68 1.90 -.288 .255 2. Reading Comprehension Task 35.52 3.05 -.534* 3. Error Detection Task 7.03 2.359 -*p <.01

PAGE 110

99 Table 4.8 Simultaneous Regression Analysis Summary for Three Variables Predicting Bilinguals Idiom Comprehension Accuracy Variable B SEB Synonym Task .157 .275 .093 Reading Comprehension Task .371 .196 .354 Error Detection Task .372 .252 .274 Note. R2 = .345 (N = 31, p < .01). *p<.01 Monolinguals only: Simultaneous multiple regression. The final simultaneous multiple regression analyzed only the data from the monolingual language gr oup. Their data were entered into a simultaneous multiple regres sion which treated idiom comprehension as the criterion variable and the other three linguistic variab les (error detection, passage comprehension, and the synonym task) as the pr edictor variables. Th is analysis allowed examination of whether the three tasks coul d predict idiom comprehension accuracy for the monolinguals. A simultaneous multiple regression demonstrated that the model was significant, F(3, 30) = 4.135, MS = 14.352, p<.05. The Adjusted R Square for this model was .239. These three predictors accounted for approxima tely 24% of the variance in the criterion variable (idiom comprehension accuracy). The Pearson Correlations are presented in Table 4.9. Significant variables are displaye d in Table 4.10. As with the bilingual-only analysis, there were no significant beta valu es. Unlike the bilingualonly analysis, though,

PAGE 111

100 all variables were positivel y and significantly correlated with the other linguistic variables, including the synonym task. Table 4.9 Means, Standard Deviations, and Intercorre lations for Monolingual Participants Idiom Comprehension Accuracy and Thr ee Linguistic Predictor Variables Variable M SD 1 2 3 Idiom Comprehension 34.32 2.14 .460** .330* .445* Predictor variable 1. Synonym Task 16.90 2.071 -.601**.301* 2. Reading Comprehension Task 37.58 3.18 -.346* 3. Error Detection Task 8.77 1.726 -*p <.05 **p <.01 Table 4.10 Simultaneous Regression Analysis Su mmary for Three Variables Predicting Monolinguals Idiom Comp rehension Accuracy Variable B SEB Synonym Task .371 .207 .360 Reading Comprehension Task -.002 .137 -.003 Error Detection Task .418 .212 .338 Note. R2 = .315 (N = 31, p < .01). *p<.01

PAGE 112

101 Summary: Aim 1 findings. It was evident from the exploration of Aim 1 that the idiom comprehension measure was significantly correlated with the other three measures (r = .427 .533), and that the thr ee measures were also significa ntly correlated with each other (r = .410 .532). The model tested in this first aim indicated that the three principles (operationalized through a r eading comprehension task, synonym task, and error detection task) predicted about 33% of the va riance in idiom comprehension scores. The variable making the most impact, the error detection task ( = .351*), was illuminated. Furthermore, the synonym task was the greate st predictor of language group membership. In addition, language group membership did not predict id iom comprehension performance, and idiom comprehension pe rformance did not predict language group membership. Lastly, more variance was accoun ted for on the three linguistic measures when the two language groups were collapsed. Cross-Language Group Performance on Idiomatic Comprehension: Aim 2 Before addressing the second aim, which analyzed the outcomes of the idiom comprehension measure, an item analysis was conducted using the Cronbach Alpha to measure the reliability of the idiom comprehension measure. Cronbach Alpha measures reliability in terms of the ratio of true score variance to observed score variance (Yu, 2007). First, the mean output presented in Table 4.11 demonstrates how difficult the items were. Since the items in this analysis we re considered either correct or incorrect, the means ranged from 0 to 1. A mean score of 1.0 indicated that all pa rticipants received a correct score for the item, s uggesting that the item was easie r. The mean scores for the idiom comprehension measure ranged from 0.452 (item #35, To whistle in your thumb ) to 1.000 (item #16, At the drop of a hat ).

PAGE 113

102 Secondly, the item correlation is a raw scor e. The more strongly the items were interrelated, the higher the consistency of th e test items (Yu, 2007). The Cronbach Alpha was .965, indicating strong internal consis tency among the test items on the idiom comprehension measure, which is one type of reliability. Moreover, due to the high interitem correlation, the idiom comprehension measure appeared to be reliable. Table 4.11 Means and Standard Deviations for the 40 Items on the Idiom Comprehension Measure Item M SD 1. Hold your head up 0.984 0.127 2. Go by the book 0.984 0.127 3. Take someone under your wing 0.984 0.127 4. Blow off steam 0.952 0.216 5. Keep a straight face 0.984 0.127 6. Right under my nose 0.984 0.127 7. Cry over spilled milk 0.984 0.127 8. Hold your tongue 0.919 0.275 9. Take a shot in the dark 0.968 0.178 10. The early bird catches the worm 0.952 0.216 11. Bring the house down 0.968 0.178 12. Paint the town red 0.903 0.298 13. Have a soft spot in your heart 0.790 0.410 14. Chip off the old block 0.968 0.178 15. Spill the beans 0.968 0.178

PAGE 114

103 (Table 4.11 continued) 16. At the drop of a hat 1.000 0.000 17. Wet behind the ears 0.887 0.319 18. Jump through hoops 0.919 0.275 19. Go cold turkey 0.919 0.275 20. To flip your lid 0.871 0.338 21. To run around like scalded pigs 0.871 0.338 22. For a good hunger there is no hard bread 0.919 0.275 23. To shoot sparrows with cannons 0.952 0.216 24. To try to make a hole in water 0.823 0.385 25. To hold someones leg 0.952 0.216 26. To fall down with four horseshoes in the air 0.774 0.422 27. To cut a pear in two 0.645 0.482 28. To throw flowers to somebody 0.758 0.432 29. To put on the sails 0.984 0.127 30. To be a monkey on a branch 0.935 0.248 31. To eat the leaf 0.839 0.371 32. To pet the horse first 0.855 0.355 33. To be at the green 0.774 0.422 34. To have salt in your pumpkin 0.597 0.495 35. To whistle in your thumb 0.452 0.502 36. To pick up a log 0.468 0.503 37. To eat on the thumb 0.468 0.503

PAGE 115

104 (Table 4.11 continued) 38. To play the donkey to get bran 0.516 0.504 39. Between dog and wolf 0.806 0.398 40. To lay a rabbit on someone 0.613 0.491 Addressing aim 2. The second aim of the study was to determine whether there were differences in performance outcomes on the idiom comprehension measure between the bilingual and monolingual language groups. Specifically, the research question posed was how the performance outcomes of the b ilingual adolescents would differ from the performance outcomes of the control group (m onolingual, English-speaking adolescents) on familiarity, semantic transparency, and cont ext. It was predicted that there would be an interaction among familiarity, semantic transparency, context, and language group. A four-way, repeated measures, mixed ANO VA with one between-subject variable (language group) and three within-subject va riables (familiarity, semantic transparency, and context, each with two levels) was conducte d in order to answer this question. Table 4.12 summarizes the ANOVA results. Table 4.12 ANOVA Results for the Accuracy Scores on the Idiom Comprehension Measure Variable/Source df MS F p partial 2 F (1, 60) 128.032 329.606 <.001* 0.846 F x G (1, 60) 1.161 2.990 0.089 0.047 T (1, 60) 72.782 253.274 <.001* 0.808 T x G (1, 60) 0.976 3.396 0.070 0.054 C (1, 60) 134.202 441.796 <.001* 0.880

PAGE 116

105 (Table 4.12 continued) C x G (1, 60) 0.073 0.239 0.627 0.004 F x T (1, 60) 28.073 72.926 <.001* 0.549 F x C (1, 60) 72.782 218.171 <.001* 0.784 T x C (1, 60) 39.516 104.478 <.001* 0.635 F x T x C (1, 60) 21.806 39.455 <.001* 0.397 F x T x G (1, 60) 2.331 6.054 0.017* 0.092 F x C x G (1, 60) 0.202 0.604 0.440 0.010 T x C x G (1, 60) 0.290 0.768 0.384 0.013 F x T x C x G (1, 60) 0.032 0.058 0.810 0.001 *Statistically Significant Note. The abbreviations signify the following idiomatic conditions: F = familiarity, T = semantic transparency, C = context, and G = group. Interactions. The overall mean difference of the language groups was significant at the .05 level (mean difference = .210, CI = .037 .382), which indicated a large amount of score overlap. However, there was not a four-way inter action that reached statistical significance; that is, language group did not contri bute to any statistically significant differences beyond chance levels once the eight combinations of idiomatic conditions were considered (familiarity: fa miliar and unfamiliar, semantic transparency: transparent and opaque, and context: in and out). Table 4.13 di splays the descriptive data for the accuracy scores for the eight combinati ons of idiomatic conditions as a function of language group. Although language group membership did not prove to be statistically significant, the descriptive data showed that overall, monolingual pa rticipants performed

PAGE 117

106 slightly better than bilingual participants in all idiomatic conditions except for the unfamiliar, opaque condition both inand out-of-context. Table 4.13 Idiom Comprehension Scores for Monolinguals and Bilinguals in the Eight Conditions Monolinguals Bilinguals Idiomatic Condition M SD M SD FTI 4.97 .18 4.84 .45 FTO 4.97 .18 4.58 .56 FOI 4.94 .25 4.58 .77 FOO 4.52 .72 4.16 .74 UTI 4.87 .43 4.55 .77 UTO 4.06 .85 3.71 .82 UOI 4.42 .77 4.48 .72 UOO 1.58 .81 1.74 .93 Note. The abbreviations used signify the following idiomatic conditions: FTI is familiar, transparent, incontext; FTO is familiar, transparent, out-of-context; FOI is familiar, opaque, in-context; FOO is familiar, opaque, out-of-context; UTI is unfa miliar, transparent, in-context; UTO is unfamiliar, transparent, out-ofcontext; UOI is unfamiliar, opaque, in-context; and UOO is unfamiliar, opaque, out-of-context. There was one statistically significant three-way interaction among familiarity, transparency, and context, F(1, 60) = 39.455, MS = 21.806, p<0.001, partial 2 = .3971 with an observed power of 1.00 (see Figure 10). An additional three-way interaction emerged among group, familiarity, and tr ansparency, F(1, 60) = 6.054, MS = 2.331, p =

PAGE 118

107 0.017, partial 2 = .092 with an observed power of .678. This interaction was the only interaction where group was significant. 3 Way Interaction: Familiarity x Transparency x Context0 1 2 3 4 5FTI FTO FOI FOO UTI UTO UOI UOO Mean Values Mean Values Figure 10. The statistically significant three-way interaction among accuracy scores in the eight combinations of idiomatic conditi ons: familiarity, transparency, and context, with language groups collapsed. Note. The abbreviations used signify the following idiomatic conditions: FTI is familiar, transparent, incontext; FTO is familiar, transparent, out-of-context; FOI is familiar, opaque, in-context; FOO is familiar, opaque, out-of-context; UTI is unfa miliar, transparent, in-context; UTO is unfamiliar, transparent, out-ofcontext; UOI is unfamiliar, opaque, in-context; and UOO is unfamiliar, opaque, out-of-context. Post hoc testing using the B onferroni procedure demonstrat ed that all participants performed significantly differently when idioms were familiar or unfamiliar, when idioms were transparent or opaque, and when idioms were presented in and out of context. For instance, overall performance was best in the familiar/transparent/in-context condition ( M = 4.903, CI = 4.815 4.991), and performance was worse under the unfamiliar/opaque/out-of-context condition ( M = 1.661, CI = 1.440 -1.882). Furthermore, participants always performed better with c ontextual support rather than without the

PAGE 119

108 support of context; however, performance decr eased more with opaque idioms (even in context) and when idioms were unfamiliar (S ee Table 4.13). In additi on, in terms of the significant three way interaction among group, familiarity, and transparency, pairwise comparisons demonstrated that the monolingua ls performed significantly better than the bilinguals in all combinations of familiarity and transparency, except for the unfamiliar, opaque idioms, where bilinguals outperformed the monolinguals, although results did not reach statistical significance. There were also three statistically si gnificant two-way interactions among the idiomatic variables. First, there was a significant interaction between familiarity and transparency, F(1, 60) = 72.926, MS = 28.073, p<0.001, and partial 2 = .5491 with an observed power of 1.00. Pairwise comparisons indicated that participants scored best on familiar, transparent idioms (M = 4.839, CI = 4.766 4.911). Their scores then progressively declined with the pattern of: familiar/opaque ( M = 4.548, CI = 4.414 4.683), to unfamiliar/transparent ( M = 4.298, CI = 4.154 4.443), to unfamiliar/opaque idioms ( M = 3.056, CI = 2.916 3.197). Secondly, there was a significant two-way interaction betwee n familiarity and context, F(1, 60) = 72.782, MS = 218.171, p<0.001, and partial 2 = .7841 with an observed power of 1.00. Pairwise comparisons rev ealed a slightly differe nt pattern in that performance was best in the familiar/in-context condition ( M = 4.831, CI = 4.724 4.937) and then scores began to decline with the unfamiliar/in-context ( M = 4.581, CI = 4.442 4.719) condition, followed by the familiar/out-of-context ( M = 4.556, CI = 4.441 4.672), and the unfami liar/out-of-context ( M = 2.774, CI = 2.636 2.912) conditions.

PAGE 120

109 Lastly, there was a significant two-way interaction between transparency and context, F(1, 60) = 104.478, MS = 39.516, p<0.001, and partial 2 = .6351 with an observed power of 1.00. Pairwise comparisons again confirmed that performance was best in the transparent/in-context condition ( M = 4.806, CI = 4.709 4.904). However, in terms of mean differences, the participants th en did better with opaque idioms that were in-context ( M = 4.605, CI = 4.483 4. 726) than with transparent idioms out-of-context ( M = 4.331, CI = 4.202 4.459). Contextual support appeared to assist in accurate idiomatic comprehension. Similar to all other po st hoc tests, mean scores declined when participants faced opaque idioms that were out-of-context (M = 3.000, CI = 2.858 3.142). Main effects. There were also three statistically significant main effects for each idiomatic variable. A main effect was found for the familiarity variable, F(1, 60) = 329.606, MS = 128.032, p<0.001, and partial 2 = .8461 with an observed power of 1.00. Participants comprehended idioms more accurately when they were familiar ( M = 4.694, CI = 4.600 4.787) rather than unfamiliar ( M = 3.677, CI = 3.566 3.789). Likewise, a main effect was found for the transpar ency variable, F(1, 60) = 253.274, MS = 72.782, p<0.001, and partial 2 = .8081 with an observed power of 1.00. Again performance was better when idioms were semantically transparent ( M = 4.569, CI = 4.476 4.661) rather than semantically opaque ( M = 3.802, CI = 3.698 3.907). Lastly, there was also a main effect for the context vari able, F(1, 60) = 441.796, MS = 134.202, p<0.001, and partial 2 = .8801 with an observed power of 1.00. Contextual support assisted participants comprehension ( M = 4.706, CI = 4.610 4.801) more so than no contextual support ( M = 3.665, CI = 3.562 3.768).

PAGE 121

110 Results of additional hypotheses. There were four additional hypotheses associated with the second aim. Firstly, it was hypothesized that both monolinguals and bilinguals would perform less well on unfam iliar, opaque idioms. This hypothesis was partially supported in that, when comparing mean differences, both the monolinguals ( M = 1.58, SD = .81) and the bilinguals (M = 1.74, SD = .93) performed less well in this condition, but only when the unfamiliar/opaque idioms were out-of-context. Performance on in-context, unfamiliar/opaque idioms was better (monolinguals: M = 4.42, SD = .77, bilinguals: M = 4.48, SD = .72). Secondly, it was hypothesized that monoli nguals would perform better than the bilinguals on familiar idioms based on the language experience hypothesis. This hypothesis was supported in terms of mean di fferences since the monolinguals performed better on the familiar idioms ( M = 19.39, SD = 0.92) than did the bilinguals ( M = 18.16, SD = 1.86), regardless of whether the idioms were transparent or opaque, or inor out-ofcontext. An independent ttest confirmed these descri ptive data, t (60) = 3.284, p < .01, d = 0.84, an effect size that indi cated that the magnitude of difference between the means of the two language groups was a large one. Interestingly, the bilinguals did perform better than the monolinguals on both the unfamiliar/opaque/in-context and unfamiliar/opaque/out-of-context idioms (See Table 4.13). Thirdly, it was predicted that all partic ipants would perform better when given contextual support; however, context would bene fit less-skilled reading comprehenders less. This hypothesis was analyzed by firs t transforming the continuous scores on the reading comprehension test in to categorical scores. Based on the scoring table of the Woodcock Passage Comprehension subtest (WJ -III ACH; Woodcock et al., 2001), scores

PAGE 122

111 were converted into Grade Equivalent Es timates. For example, a score of 36 on the subtest equals a grade equivale nt score of 10.1. These grade e quivalent scores were then compared to the participants actual grade le vels at the time of te sting. If a participant scored exactly at grade level or below, they were considered less-skilled comprehenders. Those with scores at grade level (e.g., 9.0 in grade 9) were considered less-skilled comprehenders since the study was conducted after January of the school year; therefore, to score on grade level, the participants needed a score slightly above their current grade level (such as 9.2 if they were in grade 9). All scores above grade level (grade .2 and above) were considered as indicating skilled comprehenders. Using this procedure, a total of 19 monolinguals a nd 13 bilinguals were categorized as skilled comprehenders, and 12 monolinguals and 18 bilinguals were categorized as less-skilled comprehenders. In sum, there were 32 s killed comprehenders and 30 less-skilled comprehenders. Descriptive data were also calculated for these two groups. The skilled comprehenders had a mean score of 39.13 (SD = 1.88) and the lessskilled comprehenders had a mean score of 33.80 (SD = 1.80). Moreover, a statistically significant difference, t (60) = 11.388, p < .01, existed between the two reading groups. Next, total scores on unfamiliar idioms (transparent and opaque), in context were calculated. Because of the language experien ce hypothesis, familiar idioms were not included in this analysis. Also, because th is analysis was measuring the potential common denominator of inferencing ability, onl y idioms incontext were included. An independent t-test was conducted where equal variances were not assumed (due to the Levenes Test for Equality of Variances). A statistically significant difference between

PAGE 123

112 the scores of the skilled ( M = 9.34, SD = .83) and less-skilled ( M = 8.90, SD = 1.30) comprehension groups on unfamiliar idioms in context was not found (see Appendix B for all non-significant ttest results). Thus, the null hypot hesis could not be rejected. The fourth and final hypot hesis connected with Aim 2 predicted that those participants with poorer reading comprehe nsion scores would choose more literal responses. Incorrect responses were categorized as figurative or lite ral in nature (see Table 4.14). The literal errors were tallied. Then, an inde pendent t-test was calculated (equal variances were assumed according to the Levenes test) using the number of literal errors and again, a significant difference between the less-skilled comprehenders ( M = 2.07, SD = 2.36) and the skilled comprehenders ( M = 1.53, SD = 1.44) did not emerge (see Appendix B). The null hypothesis could no t be rejected. However, when an ANOVA was conducted, which included both figurative an d literal errors, there was a significant difference between the reading groups in te rms of the number of figurative errors, F(1, 60) = 6.442, MS = 26.502, p<0.014, and partial 2 = .097 with an observed power of 0.704, with the less-skilled comprehenders ma king significantly more figurative errors. Table 4.14 Error Analysis of All Items on the Idio m Comprehension Measure by Language Group (Monolingual (M), Bilingual (B)) and Error Type Condition M Literal Errors B Literal Errors M Figurative Errors B Figurative Errors Total FTI 0 3 1 2 6 FTO 1 4 1 9 15 FOI 1 5 1 8 15 FOO 1 8 13 18 40

PAGE 124

113 (Table 4.14 continued) UTI 0 1 4 13 18 UTO 5 11 24 29 69 UOI 6 6 12 10 34 UOO 26 31 80 70 207 Note1. The abbreviations used signify the following idioma tic conditions: FTI is familiar, transparent, incontext; FTO is familiar, transparent, out-of-context; FOI is familiar, opaque, in-context; FOO is familiar, opaque, out-of-context; UTI is unfa miliar, transparent, in-context; UTO is unfamiliar, transparent, out-ofcontext; UOI is unfamiliar, opaque, in-context; and UOO is unfamiliar, opaque, out-of-context. Note2. Highlighted portion indicates the only conditions in which the bilinguals scores were higher than the monolingual scores. In addition, in terms of the error analysis, there were more figurative errors ( N = 294) than literal errors ( N = 109) for both groups combined. The bilingual group not only made more literal errors ( N = 69) than did the monolingual group ( N = 40); but also more figurative errors ( N = 159) than did the monolingual group ( N = 135). However, an ANOVA demonstrated that these differences in language groups were not significant for either the literal responses or the figura tive error responses (see Appendix R for nonsignificant ANOVA results). Summary: Aim 2 findings. A Cronbach Alpha (r = .965) demonstrated high internal consistency of the items on the idio m comprehension measure. In terms of group differences in performance on this measure, the monolinguals performed better than the bilinguals on the idiom comprehension meas ure when total scores were considered. However, there was not a statistically significant four-way interaction among group, familiarity, transparency, and context. Instead, there was a significant three-way

PAGE 125

114 interaction among the idiomatic conditions a nd a significant three-way interaction among group, familiarity, and transparency. In terms of the hypotheses associated with the second aim, four findings resulted: a) participants performed least well on opaque idioms, out of context; b) the monolinguals did perform better than the bi linguals on familiar idioms, overall; c) the performance difference between skilled and less-skilled comprehenders on unfamiliar idioms in context was not significant; and d) a significant difference did not emerge between lessskilled and skilled comprehende rs on the amount of literal responses. Effects of Within-Group B ilingual Proficiency: Aim 3 Because there may be meaningful differe nces in results within the bilingual group, depending on age of acquisition (AOA) of English or time spent in the United States, an additional aim concerned whethe r those bilingual students who were less linguistically assimilated would perform signi ficantly differently than those students who were more linguistically assimilated. To an swer this question, the language history questionnaire scores were inserted into a multiple regression formula with four predictors: idiom comprehension total sc ore, error detection score, reading comprehension score, and synonym score. A descriptive summary of the questionnaire responses follows first. Descriptive summary of the l anguage experience questionnaire. The language experience questionnaire asked the participan ts to self-report some demographic and qualitative information including: a) their count ries of origin, b) their families countries of origin, c) when they learned English and Spanish, d) if they ev er attended school in Spanish, and e) how long they had lived in the United States. Of the 31 total bilingual

PAGE 126

115 participants, 22 (approximately 71%) were bor n on the mainland of the United States. Of the other 29%, four were born in Puerto Rico, two were born in Cuba, two were born in Mxico, and one was born in Venezuela. Hence, the sample was not a recent immigrant sample, as a general rule. Of the 71% who were born in the United St ates, the majority of their families originated in Mxico (50%), followed by Puerto Rico (23%), the Dominican Republic (9%), Cuba (9%), and Columbia (9%). Fourteen of the bilingual par ticipants (45%) were reportedly simultaneous Spanish-English learners, and 17 (55%) were sequential language learners. Of the seque ntial language learners, all but one participant learned Spanish first and English second. This one participant learned English from birth, and Spanish beginning at age 2-years. The rest of the sequential Englis h-language-learners (n = 16) began learning English from a) 3-years-of-age (38%) or 4-years-of-age (6%) when they entered a preschool environment; b) from 5-years-of age (25%) when they entered kindergarten; c) from 6-yearsof-age when she entered first grade (6%); or d) from 8years-of-age (25%) when they moved to the U.S. and began grades 2 or 3. One simultaneous language learner who was born in the U.S. had lived in the Dominican Republic briefly and completed all of grade 5 and some of grade 6 th ere before returning to school in the U.S. All bili ngual participants who were bor n outside of the U.S. (29%) had been living in the U.S. and attending sc hool in English for a minimum of nine years at the time of the study. Quantitative analysis of language experience. Each of the 31 bilingual participants completed the questionnaire. The participants cumulative likert scores ranged from 13 79 total points ( M = 52.06, SD = 19.40). More points symbolized

PAGE 127

116 greater use of and exposure to Spanish on a da ily basis. To inves tigate the relationship between a students score on the questionnaire and that students scores on the four measures, a simultaneous multiple regression was conducted. The criterion (dependent) variable was questionnaire scores and the predictor (independen t) variables were the four measures (reading comprehension, synonym ta sk, idiom comprehension measure, and error detection task). Results showed that the model was significant, F(4, 26) = 3.109, MS = 912.747, p<.05. The Adjusted R Square for this model was .219, which meant that these four predictors accounted for approximately 21% of the variance in the criterion variable (questionnaire scores). Caution should be used, however, in interpreting these results since the model entered four independent va riables and the sample size was only 31. The Pearson Correlations are pres ented in Table 4.15. Significant variables are displayed in Table 4.16. Results demonstrated that there was a negative correlation between questionnaire scores and three of the measures (reading comprehension, error detection, and synonym task). This result may indicate that, as a st udents likert score increased, scores on these measures decreased. In contrast, idiom comprehension, unlike the other three variables, was not correlated with the questionnaire scores, positively or negatively. Finally, the beta values of idiom comprehension ( = .411*) and reading compre hension ( = -.472*) were both significant. The interpretation is that these two variables had the greatest impact on the criterion variable (questi onnaire score). Inte restingly, the idiom comprehension beta value was positive, i ndicating a positive rela tionship between the measure and the questionnaire, even though they were not correlated.

PAGE 128

117 Table 4.15 Means, Standard Deviations, and Interc orrelations for Bilingual Participants Questionnaire Results and F our Predictor Variables Variable M SD 1 2 3 4 Questionnaire Results 52.06 19.39 -.168 -.432** -.365* .000 Predictor variable 1. Synonym Task 14.68 1.90 -.288 .255 .265 2. Reading Comprehension Task 35.52 3.05 -.534** .528** 3. Error Detection Task 7.03 2.36 -.487** 4. Idiom comprehension measure 32.65 3.20 -*p <.05 **p <.01 Table 4.16 Simultaneous Regression Analysis Summary for Four Variables Predicting Questionnaire Results Variable B SEB Synonym Task -.673 1.739 -.066 Reading Comprehension Task -2.998 1.312 -.472* Error Detection Task -2.438 1.644 -.297 Idiom Comprehension Task 2.494 1.208 .411* Note. R2 = .324 (N = 62, p < .05). Descriptive analysis of the simu ltaneous and sequential learners. To determine if there were any differences between the simu ltaneous language learners and the sequential

PAGE 129

118 language learners, group means from the idiom comprehension measure, reading comprehension measure, error detection task, and synonym task were compared in a descriptive manner. The two gr oups performed similarly on all four measures (See Table 4.17); however, the simultaneous language learners ( n = 14) performed better than the sequential language learners ( n = 17) on the synonym task th e task that best predicted language group membership. Thus, it appears that the simultaneous language learners may have performed more like the mono linguals, but a larger sample size of simultaneous and sequential language learners would be needed to test this hypothesis. Table 4.17 Performance Scores on All Four Measures for Simultaneous and Sequential Language Learners Simultaneous Sequential Performance Measure M SD M SD Synonym Task 15 1.47 14.41 2.210 Reading Comprehension 35.36 2.21 35.65 3.673 Error Detection 7.00 2.18 7.06 2.561 Idiom Comprehension 32.79 2.39 32.29 3.837 Furthermore, half (n = 7) of the simultaneous langua ge learners were considered skilled comprehenders and half ( n = 7) were considered l ess-skilled comprehenders. Eleven of the sequential language learners we re considered lessskilled comprehenders and only six were considered skilled comprehenders. The simultaneous language learners had about the same number of literal (inaccurate) responses ( M = 2.21, SD =

PAGE 130

119 1.97) as the sequential language learners ( M = 2.24, SD = 2.71). Lastly, the simultaneous learners performed slightly better on the unfamiliar idioms in context ( M =9.21, SD =1.05) than the sequential language learners ( M =8.76, SD = 1.35). Summary: Aim 3 findings. The regression model using the four measures to predict questionnaire scores wa s significant, but low (Adjusted R Square = .219). More interestingly, all of the measures but the idiom comprehension measure were negatively correlated with the questionnaire likert scor es. This may mean that, as likert scores increased (indicating more Spanish use), sc ores on the three measures decreased, or, inversely, that as likert scores decrease d (more English use), scores on measures increased. In contrast to this pattern, the idiom comprehension measure scores were not correlated with the likert scores. However, th e beta values of idiom comprehension ( = .411*) and reading comprehension ( = .472*) both significantly impacted on the criterion variable (questi onnaire score), with the idiom comprehension beta value indicating a positive relationshi p between the measure and the questionnaire, even though they were not correlated. Overall Summary of Major Findings There were six findings associ ated with the first aim: 1) Performance on the synonym task was th e best predictor of group membership. 2) The idiom comprehension measure was sign ificantly correlated with the other three measures (error detection, reading comprehension, and synonym tasks). 3) The error detection, reading comprehens ion, and synonym tasks were significantly correlated with each other and predicted about 33% of the variance in the idiom comprehension scores.

PAGE 131

120 4) The error detection task accounted for the most variance in idiom comprehension scores. 5) Language group membership did not predic t performance on the idiom comprehension measure. 6) The idiom comprehension scores did not predict group membership. Next, there were six findings related to the second aim: 1) A statistically significant four-way interaction among language group, idiomatic familiarity, semantic transparency, and contextual support was not found. 2) There was a sta tistically significant three-way interaction among the idiomatic variables, a statistically significant three-way inte raction among language group, familiarity, and semantic transparency, and subsequent two-way interactions and main effects. 3) The lowest mean scores occurred with opaque idioms, out of context. 4) The monolinguals outperformed the bilinguals on familiar idioms. 5) There was not a significant differen ce between the lessskilled and skilled comprehenders on unfamiliar idioms in context. 6) The number of literal errors did not diffe rentiate the lessskilled from the skilled comprehenders. Finally, there were four results a ssociated with the third aim: 1) Of the 31 bilingual particip ants, 14 were simultaneous language learners and 17 were sequential language learners. 2) The four measures significantly predicted the questionnaire scores.

PAGE 132

121 3) Except for the idiom comprehension measure, all measures were negatively correlated with the questionnaire. 4) The beta values of idiom comprehension and reading comprehension both significantly impacted on the questionnaire score, with idiom comprehension having a positive relationship.

PAGE 133

122 Chapter 5 Discussion The present study is the first to inves tigate idiom comprehension in bilingual (Spanish-English) adolescents, while also ex ploring variables known to contribute to accurate text comprehension, namely, compre hension monitoring, inference generation, and lexical depth. Findings pr ovided new information on the effects of these linguistic variables on idiom comprehension. In addi tion, differences between the two language groups were found as well as po tential qualitative differences between the sequential and simultaneous bilinguals. Discussion focuses on the specific aims of the study and their relation to previous studies. Moreover, the relevance of the findings is discussed in terms of how they do or do not lend support to related models and hypotheses. Aim 1: Effects of Reading Comprehension, Error Detection, & Synonym Performance on Idiom Comprehension The first aim was to test the model crea ted for this study. Specifically, the purpose was to determine the extent to which each of three linguistic variables (error detection, reading comprehension, and synonym performa nce) predicted the criterion variable, idiom comprehension accuracy. Support for the model was found in that the three variables did explain 33% of the variance in idiom comprehension accuracy. Error detection was the most powerful predictor.

PAGE 134

123 Error detection. The error detection task measured comprehension monitoring ability at the microand macrolevels. The microlevel refers to each proposition within a sentence or phrase as well as the senten ce itself, while macrolevel refers to construction of meaning across a text. Thus, poorer performance on the unfamiliar idioms in context would point to a potential problem with comprehension monitoring. Error detection may have required inferencing abi lities similar to those tapped in the idiom comprehension measure. This similarity can be explained by the C onstruction-Integration (C-I) model (Kintsch, 1998), which was the underlying theoretical model for both error detection and idiom comprehension. To review briefly, the C-I model posits th at two phases exist in the process of reading comprehension. During the first phase, meaning is construc ted by recognizing each word and then activating e ach words meaning and all of its associated meanings in long-term memory. The second phase, integrat ion, requires the generation of inferences and activation of prior knowledge to form a coherent representation of the text. These two phases are both critical for detecting text vi olations at the local and global levels, as well as in interpreting ambiguous text, such as idioms. Thus, patterns of performance on both the idiom comprehension test and the erro r detection task provide support for the C-I model. In addition, Morrisons (2004) error detecti on paradigm, which was modified for this study, appeared to be a valid pred ictor of idiom comprehension ability. Reading comprehension. Reading comprehension contributed to explaining some variance in the idiom comprehension measure. Perhaps the variance accounted for is due to the fact that the reading comprehension task required the constr uction and integration of both the microand macrolevels of text. On the other hand, the reading

PAGE 135

124 comprehension task was a cloze procedure, which consisted of single, independent sentences that became more demanding as the passages increased in length and complexity. Although inferencing was required to fill in the blanks accurately, the ability to construct meaning at a local le vel was weighted more heavily than was integration of local coherence at a more global level across an expansive text. For example, the basal for the participan ts age group was the following item: Many freshwater turtles are good to eat. Snapping _______ are sold commercially in large numbers. Moreover, cloze procedures may not be the best way to assess reading comprehension when global coherence is the larger aim. These results are somewhat at odds with past research, wh ich concluded that reading comprehension abilities predicted id iom comprehension abilities (Cain et al., 2005; Levorato et al., 2004; 2007). This interp retation was not strong ly supported in the present study. A weaker version was supporte d; that is, a stro ng, positive correlation existed between idiom comprehension and r eading comprehension. It should be noted, though, that Cain et al. used a different standardized meas ure of reading comprehension that required answering comp rehension questions from the Gray Oral Reading TestFourth Edition (Weiderhold & Bryant, 2001) a nd Levorato et al. utili zed a standardized Italian reading measure without cloze procedur es. Neither of the measures administered in these two studies consisted of cloze tasks, one potential reason why the present study does not mirror their results. What is mo re, Cutting and Scarborough (2006) concluded that different cognitive processes may be tapped by varied reading comprehension measures in general; therefore, if a different measure of reading comprehension had been selected for the current study, it may have been more predictive of idiom comprehension.

PAGE 136

125 The positive relationship between idiom comprehension and reading comprehension in this study does suggest th at idiom comprehension may be a possible indicator of undetected language-based read ing problems in bilingual and monolingual adolescents. The rationale is that idiom comprehension is a vehicle for assessing dynamic interactions among semantic processing, syntactic processing, and inference generation. For example, as Nippold, Moran, and Schwartz (2001) demonstrated in their idiom comprehension study, nearly 24% of participants ( N = 50; mean age 12 years, 4 months) performed significantly below their peers in in ferencing ability and reading skills, even though their teachers had considered them to be progressing normally in reading comprehension before the study. As Nippold et al. (2001) suggest, idiom comprehension tasks could be administered as a screening measure to identify students who are having difficulty with idiom understanding, and, thus, potentially, have undetected reading difficulties. Synonym task. Entering all three measures into th e regression model revealed that the synonym task was the least powerful pr edictor of idiom comprehension accuracy; however, performance on the synonym task wa s the most powerful predictor of language group membership (bilingual or monolingua l). The synonym task was meant as a measure of lexical depth and, in particular, polysemy, or knowledge of a words multiple meanings. A words meaning is always colored by the social context of use, giving every word different shades of meaning (Nagy & Sc ott, 2000). Polysemy is important in idiom interpretation since an unfamiliar idiom consis ts of words that can have literal meanings as well as figurative meanings when in the co ntext of the idiom. Shades of meaning are seldom directly taught (Nagy & Scott, 2000) but are usually implied. Typically, more

PAGE 137

126 frequent words (e.g., think) have more shades of meaning (related and unrelated) than lower frequency words (Nagy & Scott, 2000). For example, the accuracy of responses decreased for all participants for more literate, less frequent terms, such as: tarry, stratagem, cogitate, capacious, upbraid, fallow, and evanescent. However, the accuracy of the bilingual group declined much sooner on the synonym subtest than did the re sponse accuracy of the monolingual group, beginning with the term amusing. For example, the accuracy percentage of the bilingual group was 39% for amusing compared with 71% for the monolingual group. Many in the bilingual group gave the synonyms fascinating, amazing, or exciting for the term amusing. Another commonly missed item among the bilinguals was the term residence. Many gave the synonym neighbor Oddly enough, many bilinguals did not provide the correct synonym for consume, which is a cognate in Spanish ( consumir ). Cognates are translation equivalents or wo rds that look and sound simila r and share the same meaning in two languages, such as: different/diferen te; rea/area. According to Snow and Kim (2007), cognate knowledge must be explicitly taught. To benefit from cognate knowledge, Spanish-speakers need to recognize similarities in orthography, a skill reserved fo r those literate in Sp anish and less available for those who are only orally proficient. Mo reover, Snow and Kim (2007) suggest that even knowledge of fully translatable cogna tes is not enough to solve most reading comprehension problems because those cogna tes will not occur frequently enough. In fact, Snow and Kim (2007) conclude that atte ntion to polysemous meaning is the key to exploiting cross-language semantic relations hips to enhance reading comprehension. A speculation is that polysemy may be an index of the extent to which bilingual

PAGE 138

127 adolescents knowledge of Spanish-Englis h semantic relationships have become integrated or remain overlapped or separated. It was evident from the studys results that monolinguals and bilinguals differed most in their performance on the syn onym measure, which accessed polysemous meaning. Furthermore, performance on the synonym and the reading comprehension tasks were strongly correlate d. If, as a group, the bilingual participants had less well developed lexical depth, then both their fam iliarity with synonyms and their reading comprehension might be affected. An impor tant factor is that the synonym measure administered (the Reading Vocabulary S ubtest from the WJ-III ACH) was not normed on bilinguals, but on monolingual English-speaker s. The task was also demanding since it required generation of synonyms without any linguistic context. Nagy (2007) suggests that the provision of linguistic cues, such as sentence order, which requires syntactic awareness, may aid in selecting th e appropriate polysemous meaning. Overall, the results of this study supp ort past research that focused on the importance of lexical depth in literate la nguage development for both monolinguals and bilinguals (McGregor, 2004; Ordez, Carlo, Snow, & McLaughlin, 2002; Stahl, 2003). Furthermore, results of studies that have e xplicitly addressed poly semy in bilinguals (e.g., August, Carlo, Dressler, & Snow, 2005) echo th e present studys findings in that the bilinguals consistently performed below E nglish monolingual peers on tasks of English polysemy. The emerging evidence suggests that knowledge of English vocabulary in adolescents is evidently to some extent determined by their distribution of time over their two languages: those w ho spent the most time talking English and the least time

PAGE 139

128 speaking their native language ended up with the best knowledge of English vocabulary (Snow & Kim, 2007, p. 133). Lexical depth, as it relates to reading and academic success, warrants specific attention due to the substantial cognitive co mplexity it adds to the process of word learning (Ordoez et al., 2002), and the implications that lexical depth may have for bilinguals academic success. It is known th at vocabulary knowledge strongly influences reading comprehension in monolinguals (Na gy, 2007). Likewise, Pr octor et al. (2005) found that English vocabulary knowledge was critical for improved English reading comprehension in native Spanish-speaking bi linguals. Snow and Kim (2007) discuss this issue in terms of large problem spaces. Learning as many English words as their English-only peers, not to mention developmen t of lexical depth, is a large problem space compared to learning letters, phonemes, and spelling patterns, which are incrementally smaller problem spaces. The eradication of th ese large problem spaces appears linked to intensive and robust vocabulary instruction in early childhood settings and throughout the elementary and secondary grades (Snow & Kim, 2007), as developing lexical depth is one of the keys to becoming truly literate. Summary. In summary, approximately 33% of the variance on idiom comprehension performance was accounted for by the other three measures. Nonetheless, about 2/3s of the variance was left unexplai ned. This leads to the speculation that either there may be one or more additional factors at work that were not measured in this study, or, alternately, the measures did not assess th e constructs they set out to assess (see Cutting & Scarborough 2006, on the wide variati on in the construct validity of reading comprehension measures). Another explanat ion for the unexplained variance may be

PAGE 140

129 methodological. Extensive variance left accounted for may be explained by measures that were so strongly correlated with one other, that disentangling the variables was difficult. Aim 2: Idiom Comprehension Outcomes. The second aim of the study was to dete rmine whether there were differences between the bilingual and m onolingual language groups in their performance outcomes on the idiom comprehension measure. Over all, the monolingual group outperformed the bilingual group. Specifically, th e monolinguals consistently performed better than the bilinguals on the familiar idioms. One interpretation of this finding is that meaningful experience with figurative expressions predic ts language group performance on familiar idiom comprehension and, further, that cultural and sociolingu istic factors, particularly home language, mediates idiom comprehension. In addition, analyses collectively showed that, with one exception, context faci litated accurate idiom comprehension more than any other variable. The exception occurr ed on familiar, transparent idioms where monolingual performance did not differ between idioms inor outof context. The results of this aim will be discussed in terms of how they relate to past research findings, models, and hypotheses. The language experience hypothesis and beyond. One possibility accounting for these findings is the language experience hypothesis (Ortony et al., 1985; Qualls & Harris, 1999). Results from this study, as well as others (Cain et al., 2005; Levorato et al., 2004, 2007); support this hypothesis grounded to con cepts about the frequency of input. A more robust explanation, though, may be th e salience of literal meanings for L2 individuals. In the literal sa lient resonant model (Cieslicka 2004), literal meanings are more prominent than figurative meanings for L2 learners, whereas the opposite holds for

PAGE 141

130 their monolingual counterparts. The Cieslick a (2004) model expands on Gioras (2003) hypothesis of graded salience, which conjec tures that more salient meanings (i.e., familiar/frequent meanings) are prioritized a nd accessed first, despite contextual bias. Cieslicka (2004) points out that L2 learners often encounter the literal meanings of L2 lexical items before they discover the figura tive meanings in fixed, conventional phrases, such as idioms. In the present study, it may be that the bilinguals (particularly those with less English experience) relied on single constituent meanings with unknown, transparent idioms leading them to an inaccurate, literal response. An advantage of the literal salient resonance model is that it goes beyond th e generality of the language experience hypothesis to explain potentially why the bili ngual participants performed less well on familiar idioms than did the monolinguals. This interpretation is also consistent with prior bilingual re search on adult L2 comprehension of idioms (Abel, 2003). Abels Dual Idiom Representation (DIR) model posits that second-language le arners do not develop as ma ny idiom entries as native speakers due to their lower frequency of enc ounters with these multiple meanings in the L2. As a result, when an idiom in the second language does not corres pond to an idiom in the first language, second langua ge learners may rely more on constituent lexical entries. The findings of the present study, particular ly the bilinguals overall performance on transparent idioms, also support the DIR model and Abels (2003) conclusions. Results also echoed one of Coopers (1999) findings. Adult bilinguals, when compared with monolinguals, chose more literal than figurative responses by relying on literal interpretations of unknown idioms.

PAGE 142

131 The Global Elaboration Model. It was evident in this study and in past studies (e.g., Cain et al, 2005) that context was the most facilitative fact or in unfamiliar idiom comprehension accuracy, a result that supports the basic tenet of the Global Elaboration Model (GEM). The GEM posits that exploitation of context seems to be the major factor associated with figurative la nguage competence. Previous research (Cain et al., 2005; Nippold & Rudinski, 1993; Ni ppold & Taylor, 1995) has also found that English speaking pre-adolescents were more likely to recognize familiar, transparent idioms than familiar, opaque idioms when presented without context. This pattern parallels the findings of the present study for both the monolingual and bilingual groups. Although there were not any results contrary to th e GEMs basic processes, the developmental theory behind the GEM was not asse ssed as part of the current study. Control of prior knowledge. The majority of the unfamiliar idioms were taken from Cain et al. (2005) who found that their participants performed better on familiar rather than unfamiliar idioms when these were presented out of context. The overall pattern of performance on unfamiliar idioms in the Cain et al. study was similar to the current findings, although the bilinguals perf ormed better than the monolinguals when idioms were both unfamiliar and opaque, inor out-of-context. More investigation is necessary to determine why the bilinguals perf ormed better in this instance; however, the use of unfamiliar idioms did appear to contro l for familiarity as well as prior knowledge. For this reason, presenting unfamiliar idioms in context may be an appropriate method for assessing the comprehension monitoring, lexical depth, and inferential skills of bilingual students. Since the unfamiliar idio ms were equally unfamiliar to monolinguals and bilinguals alike, this paradigm enab les continued comparis on of monolingual and

PAGE 143

132 bilingual development in the linguistic a nd reading comprehension domains with an equitable assessment procedure. The relationship between reading comprehension and idiom comprehension. The Cain et al. (2005) and Le vorato et al. (2004, 2007) studies are the only ones that have investigated the direct relationship be tween reading comprehension and idiom comprehension prior to the present study. In th eir studies a consiste nt finding was that idiom comprehension and reading compre hension were related. Another related hypothesis was that only those children with better reading comprehension would be able to go beyond the literal meaning of individual semantic cons tituents to comprehend the global, and, therefore, figurative, meaning of an idiomatic phrase. For example, Cain et al. (2005) found that poor comprehenders performed worse than good comprehenders on opaque idioms in context. In this study, even though the skilled comprehenders did perform better than the less-skilled comprehenders on the unfamiliar id ioms in context in absolute terms, the difference was not statistically significant. This finding may be due to a combination of three factors. These include: a) the small amount of items that were unfamiliar, in-context (only 10); b) the manner in which the particip ants were arbitrarily categorized as lessskilled and skilled on the grade equi valent scale of the Woodcock Passage Comprehension subtest (WJ-III, Woodcock et al., 2001); and c) the age of the participants in this study (adolescents) compared with the younger pre-adolescent samples in the two prior studies. A further point is that Cain et al (2005) and Levorato et al. (2004, 2007) categorized their participants a priori into the two skilled and unskilled reading

PAGE 144

133 categories. Since the participants were olde r in the presen t study, perhaps less-skilled comprehenders were more able to compensate for weaker inferencing skills since they had more exposure to reading over time. Th is finding would support a developmental trend in idiom competence; a trend that the GEM postulates is present in the development of idiom comprehension (L evorato et al., 2007). It should also be noted that support was not evident for Crutchleys (2007) hypothesis. In brief, Crutch ley (2007) hypothesized that ch ildren would parse chunks of language into constituent parts only if needed. This hypothesis predicts that participants would always choose more figurative meanings to explain idioms, instead of interpreting idioms word by word. Since even the les s-skilled comprehenders did not overlook semantic analysis (literal re sponses) in favor of contex tually plausible responses (figurative responses), Crutch leys (2007) hypothesis was not replicated. However, verb + particle constructions were not a focus in the current study as they were in the Crutchley (2007) study. Aim 3: Effects of Within-Group Bilingual Proficiency on Performance The third aim concerned whethe r those bilingual students who were less linguistically assimilated (spoke less E nglish and more Spanish) would perform differently than those students who were mo re linguistically assimilated (spoke more English and less Spanish). The four measures (the idiom comprehension, error detection, reading comprehension, and synonym tasks) di d significantly predict total questionnaire scores (Adjusted R Square = .219). The am ount of variability accounted for in the questionnaire does suggest that the questionnai re is a valid instrument to some degree. Furthermore, all measures, with the exception of the idiom comprehension measure, were

PAGE 145

134 negatively correlated with the questionnaire. Insufficient academic English language knowledge may explain higher likert scores and lower scores on the error detection, reading comprehension, and synonym measures for the bilingual language group overall. In comparison, when everyday use of Spanish decreases and academic English use increases, scores on academic English measures may increase. Despite the apparent face validity of th e questionnaire, it had several limitations. Firstly, likert scores may have lacked strong predictive validity since they were qualitative estimates of the participants use of Spanish. Secondl y, questionnaire results were not qualified by a more objective measur e of language use and proficiency. Thirdly, the questionnaire did not address formal, academic language assimilation as much as everyday conversational language use. This limitation may explain why performance outcomes for the bilinguals on th e synonym task and their questionnaire scores were not significantly correlated, even t hough the synonym task was the best indicator of language group status. Lastly, there were four pred ictor variables entered into the multiple regression conducted to predic t the criterion variable. Th e results of the regression, therefore, may be inflated since there were onl y 31 participants in the bilingual sample. One hypothesis to explain why the simultaneous bilinguals descriptively outperformed the sequential bilinguals is that the simultaneous bilinguals possibly had a more balanced and integrated lexicon than the sequential bilingua ls. This hypothesis is consistent with the bilingual model of lexical knowledge (Hernandez, Li, & MacWhinney, 2005). This model depicts develo ping lexical organization as influenced by simultaneous growth in lexical diversity a nd lexical depth. Weakly integrated L1 and L2 lexical systems may even cause interferen ce when these bilinguals are asked to define

PAGE 146

135 L2 words (Hernandez et al., 2005). On the other hand, interference from the L1 may signal an emerging convergence of the system s. Theoretically, the L2 of a sequential language learner would first be parasitic on the L1 until it gains enough internal and external resonance to compete with the L 1. However, if the L1 has never become entrenched (especially in terms of academic language), then it may not be supportive enough for the L2 to grow in terms of breadth and depth. Therefore, in this scenario, the sequential language learner may appear more like a monolingual with language impairment. Data from this study suggest that the simultaneous bilinguals may have had a more entrenched L2 lexicon than the sequent ial bilinguals. Furthermore, this pattern implied that learning English and Spanish simultaneously since birth, in this sample at least, may be more advantageous for performance on tasks of English lexical depth. In general, the simultaneous b ilinguals performed more like the monolinguals; however, a larger sample of simultaneous and sequential bilinguals is needed to investigate this finding further. Potential Study Limitations There are at least four pot ential limitations of the cu rrent study. Firstly, English language proficiency was not assessed in an objective manner. Secondly, there were only four linguistic measures give n, none of these measures was administered in Spanish, and all were normed on English speakers. Thir dly, the subgroups (less-skilled/skilled comprehenders; simultaneous/sequential bilingual s) were not grouped a priori. Lastly, the idiom comprehension measure was a newly cons tructed measure. Each of these potential limitations will be addressed at length next.

PAGE 147

136 Taking English language proficiency into account. Past research on English language learners (ELLs) has illuminated the f act that these students may appear to be orally fluent in their L2, at least for social interactions, but perf orm below grade or age level on academic tasks in L2 (Cummins, 2000). In fact, a gap of se veral years seems to exist between achievement of oral language proficiency and academic proficiency in ELLs second languages. Although an ELL ma y reach peer-appropr iate levels of conversational proficiency with in a couple of years of exposure to the second language, academic language proficiency may take signif icantly longer to master (generally 5-7 years). Furthermore, monoli ngual English-speaking children come to school with oral language proficiency for conversational purposes, they continue to develop academic language proficiency throughout the remainde r of their school years (Cummins, 2000). In the present study all bilinguals had been living in the United States for at least nine years; however, it is possible that the bilinguals had variable levels of experience with and mastery of academic English. In addi tion to collecting language history via selfreport (the language history questionnaire ), a quantitative measure of language proficiency may aid in a more refined categor ization of bilinguals, such as those with lowand high-proficiency in English. Unfortunately, available English language proficiency tests are not always valid. For example, Pray (2005) investigated how well three commonly used assessments (the La nguage Assessment Scales-Oral (De Avila & Duncan, 1991; LAS-O), the Woodcock-Muoz Language Survey (Woodcock & MuozSandoval, 2001; WMLS), and the IDEA Oral Language Proficiency Test (Ballard, Tighe, & Dalton, 1980; IPT) measured English oral-language profic iency in fourth and fifth grade children who were either native, non-Hispanic, English-speaking monolingual

PAGE 148

137 children (n = 20) or of Hispanic descent ( n = 20). None of the native English-speakers scored as fluent on the WMLS. In compar ison, on the IPT, 85% of all participants (monolinguals and bilinguals) scored as fl uent in English, while performance on the LAS-O indicated that all participants (m onolinguals and bilinguals) were fluent in English. Pray (2005) concluded that the WMLS items may be assessing academic language proficiency instead of oral language pr oficiency, a point that questions the tests construct validity. Similarly, Pray (2005) f ound that the IPT was geared more towards testing academic language proficiency and not oral language proficiency. Lastly, Pray challenges the LAS-O scoring methods and its in ter-rater reliability. The investigators in the Pray study and the independent company th at scored LAS-Os were at odds in how the measure was scored. Despite these misgivings, measures of English language proficiency may be informative in a research study (such as the one conducted) in order to categorize bilinguals based on their level of academic language proficiency. Based on Prays (2005) analysis, it would have also been advantageous to: a) asse ss the oral English proficiency of both the monolingual and bilingual partic ipants, and b) admini ster a descriptive measure, such as an oral expository samp le, to compare with the outcomes from the formal measure. Inclusion of additional s poken language and cross-li nguistic assessments of reading comprehension. Past research has demonstrated that words and expressions that have abstract or multiple meanings are difficult for students with spoken language disorders to interpret (Nippold, 1991). An inclusion criter ion for the present study was that participants could not be presently en rolled in speech-language therapy or special

PAGE 149

138 education. It would be valuable, therefore, to assess formally all participants spoken language ability to ensure that none had undetected language impairments. In addition, assessing reading comprehe nsion skills in Spanish would have provided for a rich cross-linguistic analysis. Testing bilingua ls in both of their languages is imperative to identify their strengths and n eeds appropriately. Due to time constraints, these options were not possible in the curren t study, but would improve future studies of this nature, particularly si nce the standardized measures used were normed on English speakers. Sample size and characteristics Although there were significant findings based on the current sample size, a larger bilingual sample would allow for more quantitative analyses of differences based on sequential or simultaneous language learning. Furthermore, if students were first sorted into skilled and less-skilled reading comprehenders based on average and below average scores (e.g., below and above the standard score of 85 on a reading comprehe nsion measure with a mean score of 100), then the two reading groups, e qual in number, could be comp ared in a more quantitative manner. One example of this approach is seen in the Cain et al. (2005) study. As it were, reading group membership in this study was determined in a more arbitrary manner (by age equivalency scores) rather than a priori. Reliability of the idiom comprehension measure. The reliability of the idiom comprehension measure was assessed through a Cronbach Alpha. High internal consistency (r = .965) was demonstrated. However, approximately half of the items (the familiar idioms) had a mean accuracy of .90 and above. This finding brings into question the difficulty level of the measure and the vali dity of the foils chos en. For instance, the

PAGE 150

139 selection of foils could bias choices, as well as the short st ories that were provided as context. On the other hand, the variability found in the standard deviations of these means indicated that even the easie st items were not easy for al l participants. Moreover, the means were lower and the standard deviations were more variable for all participants when the idioms were unfamiliar. One interpretation is that the unfamiliar idioms represented a more level playing field for both groups, since the monolinguals likely had more experience with the familiar American idioms. Directions for Future Study An important direction would be to in vestigate bilingual adolescents with and without detected language impairments. If idiom comprehension is strongly related to reading comprehension ability, then difficulty with idiom comprehension may also be a diagnostic indicator of language impair ment. A longitudinal study following these participants through their school-age years may rev eal how figurative understanding emerges over time in typical and atypical la nguage-learners with one or more languages. In addition, long-term assessment should include the development of linguistic variables, such as lexical depth, comprehension monito ring, and inference generation that appear correlated with figurative language development, as well as the effects of approaches to the teaching of reading comprehension. A second research strand could focus on the corpus of unfamiliar idioms used in the present study. This corpus could be expa nded to include more unfamiliar expressions and then normed on a larger group of particip ants. Furthermore, the current method of presentation (multiple choice) could be comp ared with another method, such as orally defining the idiomatic expressions. This co mparison would allow for critical review of

PAGE 151

140 the present methodology, namely addressing th e ceiling scores achieved in this study. Cain et al. (2005) found that the oral definiti on of idioms was more difficult than multiple choice. Furthermore, Chan and Marinellie ( 2008) found that adolescent s similar in age to the participants in this study defined familiar idioms with accuracy levels that were not significantly different from adults (colle ge students), but which were significantly different from younger preadoles cents (grades 4, 5, and 8). Another direction in this line of resear ch would be to expand the study to examine the variables in both languages, English and Spanish. These studies could assess three of the tasks (comprehension monitoring, r eading comprehension, and synonym knowledge) in Spanish as well as English for the b ilingual participants. Although many of the bilingual participants stated that they were not able to read in Spanish, assessment in just English is only partially revealing these stude nts potential. To obtai n a complete picture of bilinguals, assessments need to be a ttempted in both languages using conceptual scoring. Conceptual scoring involves a bili ngual examiner counting overlapping lexical representations (i.e., representations shared by both languages) once. Then, the examiner would allow for responses in either Spanish or English, called singlets (i.e., words represented by only one of the two languages), to be counted correct as well (Bedore, Pea, Garca, & Cortez, 2005; Pearson, Fern andez, & Oller, 1993). This method is supported by Grosjeans (1998) holistic view of bilinguals. In other words, conceptual scores do not punish bilinguals for dual langu age activation at the lexical, lemma, or conceptual levels; and do not reward i nhibiting one language to activate another. Proponents argue that this pr ovides a more naturalistic context for testing. Recruiting bilingual participants who are biliterate could make this next direction feasible.

PAGE 152

141 Lastly, neurol inguistic research on figurativ e language comprehension using neuroimaging tools has yielded promising findings. For inst ance, studying adults with brain damage and children with either brain damage or callosal agenesis has potentially isolated which neural regions are responsible for accurate interpretation of various kinds of figurative language. These include irony (Pexman & Glenwright, 2007), sarcasm (Shamey-Tsoory, Tomer, & Aharon-Peretz, 2005), metaphors (Ramachandran, 2005), and idioms (Rizzo, Sandrini, & Papagano, 2007). These studies have collectively demonstrated that lesions of the prefront al cortex, and possibly the ventromedial prefrontal cortex, disrupt accurate figurative language comprehension. It is also posited that corpus collosum development may be important for the emergence of nonliteral language comprehension since the growth of the corpus callosum coincides with figurative language maturation (Huber-O krainec, Blaser, & Dennis, 2005). For example, deficits in idiom comprehe nsion have been found in children with corpus callosum agenesis and hypoplas ia (Huber-Okrainec et al., 2005). The simultaneous development of the corpus callosum and idiom comprehension may further explain the emergent and developmental tr end of figurative language. Future studies using neuroimaging tools to assess in ter-hemispheric communication and idiom comprehension in monolingual and bilingual adolescents with language impairment may further reveal why these adolescents str uggle with figurative language and reading comprehension. Furthermore, inhibitory control, which is responsible for lit eral suppression and essential for idiom comprehensi on, is localized to the prefrontal cortex, and is closely linked to working memory (Pexman & Glen wright, 2007). Working memory functions

PAGE 153

142 have proven to be important for metaphor comprehension (Kintsch, 2000). Chiappe and Chiappe (2007) found that individuals with high working memory capacity interpreted metaphors with greater accuracy and speed than did individuals with low working memory capacity. Future idiom comprehensi on studies could include a working memory task and enter this into a re gression model along with the linguistic measures. A working memory measure may explain some of the previously unaccounted variability in idiom comprehension ability. Conclusion The overarching goal of this study was to a dd to the current bi lingual literature on the relationships between a linguistic dom ain (idiom comprehension) and reading comprehension. A strong relationship emer ged between reading comprehension and idiom comprehension, with comprehension monitoring as the strongest predictor of idiom comprehension. Furthermore, the best i ndicator of language group membership was performance on the synonym task, indicating th at bilingual students in particular need more rigorous and robust vocabulary instru ction to develop deeper knowledge of polysemy. Lexical depth and comprehension monitoring are both higher-order skills necessary for proficient text comprehension. High stakes state assessments place more emphasis on academic vocabulary knowledge as students progress through the gr ade levels (Alliance for Education, 2007). Reading in a second language is inherently cross-linguistic (Koda, 2007). In order to eliminate those large problem spaces (Snow & Kim, 2007), bilinguals need explicit instruction on how to buttre ss their language/literacy learni ng in their L2 by exploiting their first language. Unfortunately, bilinguals do not always have a strong base in their

PAGE 154

143 first language on which to build. Thus, in struction based on E nglish derivational morphology may build a deeper processing stance towards multiple meanings in English (see Calderon et al., 2005; Carlo, Au gust, & Snow, 2005; Snow & Kim, 2007). Latino ELLs are persistently over-represented in speci al education (Koelsch, 2006). As Wagner, Francis, and Morris (2005) cite, it is unclear whether limited language proficiency in Englis h is interfering with learning or is masking a learning disability, or leads to poor performance on assessments used for identification, which are not culturally and linguistica lly appropriate for that purpos e (p. 6). Because of these issues, recent findings from the National Li teracy Panel on language-minority youth (August & Shanahan, 2006) suggest that ELLs of ten do not reach the same level of textlevel literacy as their native English-speaki ng counterparts. Hence ELLs require explicit and intensive instruction in higher-order, text-level skills, such as making inferences and using prior knowledge, instead of focusing on smaller problem spaces alone. This study demonstrated a significant difference in highe r-order, text-level English abilities in the monolingual and bilingual participants, such as robust semantic knowledge, comprehension monitoring, and overall reading comprehension. The results also suggest that the study of idiom compre hension, because of its ability to provide insight into semantic depth and comprehe nsion monitoring, does offer a unique vantage point to investigate the underp innings of text comprehensio n. The GEM (Levorato et al., 2007; Levorato et al., 2004), de rived from Kintschs (1998) Construction-Integration model, was supported qualitatively. However, the findings on bilingual adolescents, in particular, go beyond the GEM because E nglish text comprehension and idiom comprehension in bilinguals appears to be mediated most powerfully by the vocabulary

PAGE 155

144 of academic language and comprehension monitoring. This study points to a new direction for the bilingual research. Future studies focused on the linguistic and reading domains of bilinguals need to investigate how more equitable measures of language knowledge, such as unfamiliar idioms, can detect language impairments. More sensitive instru ments can result in the type of tailored intervention that, potentially, might lead to increased graduation rate s. Most importantly, it is essential to unde rstand how the degree of integrati on of two lexicons in bilingual students impacts on their development of higher-order skills nece ssary for academic achievement. In sum, a major priority is the eradication of the large problem spaces (Snow & Kim, 2007) that currently contribute to the literacy gap between bilinguals and their monolingual counterparts.

PAGE 156

145 Reference List Abel, B. (2003). English idioms in the fi rst language and second language lexicon: A dual representation approach. Second Language Research, 19 329-338. Abkarian, G.G., Jones, A., & West, G. (1992). Young childrens idiom comprehension: Trying to get the picture. Journal of Speech and Hearing Research, 35 580-587. Abrahamsen, E.P. & Burke-Williams, D. (2004). Comprehension of idioms by children with learning disabilities: Metaphoric transparency and syntactic frozeness. Journal of Psycholinguistic Research, 33 203-215. Alliance for Education (2007, February). Urge nt but overlooked: The literacy crisis among adolescent English language learners. Alliance for Excellent Education, 18. August, D., Carlo, M., Dressler, C., & Snow, C. (2005). The critical role of vocabulary development for English language learners. Learning Disabilities Research & Practice, 20, 50-57. August, D. & Shanahan, T. (Eds.). (2006). Developing literacy in second-language learners: Report of the national literacy pa nel on language-minority children and youth. Mahweh, NJ: Lawrence Erlbaum. Ballard, W.S., Tighe, P.L., & Dalton, E.F. (1980). IDEA Oral Language Proficiency Test I. Brea, CA: Ballard & Tighe.

PAGE 157

146 Beck, I. L., McKeown, M.G., & Omanson, R.C. (1987). The effects and uses of diverse vocabulary instructional techniques. In M.G. Mckeown & M.E. Curtis (Eds.), The nature of vocabulary acquisition (pp. 147-163). Hillsdale, NJ: Lawrence Erlbaum. Bedore, L.M., Pea, E.D., Garca, M., & Cortez, C. (2005). Conceptual versus monolingual scoring: When doe s it make a difference? Language, Speech, and Hearing Services in Schools (36), 188-200. Bobrow, D. & Bell, S. (1973). On cat ching onto idiomatic expressions. Memory and Cognition, 1, 343-346. Cacciari, C., & Levorato, M.C. (1988). The effect of semantic analyzability of idioms in metalinguistic tasks. Metaphor and Symbol, 13 159-177. Cacciari, C., & Tabossi, P. (1988). The comprehension of idioms. Journal of Memory and Language, 27, 668-683. Caillies, S., & Le Sourn Bissa oui, S. (2006). Idiom comprehens ion in French children: A cock-and-bull story. European Journal of Developmental Psychology, 3 189-206. Cain, K., Oakhill, J., & Lemmon, K. (2005). Th e relationship between childrens reading comprehension level and their comprehension of idioms. Journal of Experimental Child Psychology, 90, 65-87. Caldern, M., August, D., Slavin, R., Duran, D., Madden, N., Cheung, A. (2005). Bringing words to life in classrooms w ith English-language learners. In E.H. Hiebert & M.L. Kamil (Eds.), Teaching and learning vocabulary: Bringing research to practice (pp. 115-136) Mahwah, NJ: Lawrence Erlbaum. Carlo, M.S., August, D.E., & Snow, C. ( 2005). Sustained vocabul ary-learning strategy instruction for English-language learners. In E.H. Hiebert & M.L. Kamil (Eds.),

PAGE 158

147 Teaching and learning vocabulary: Bringing research to practice (pp. 137-154) Mahwah, NJ: Lawrence Erlbaum. Chall, J.S., & Dale, E. (1995). Manual fo r the new Dale-Chall readability formula. In J.S. Chall & E. Dale, Readability revisited: The ne w Dale-Chall readability formula. Cambridge, Massachusetts: Brookline Books. Chan, Y.-L., & Marinellie, S.A. (2008). Definitions of idioms in preadolescents, adolescents, and adults. Journal of Psycholinguistic Research, 37 1-20. Chiappe, D.L., & Chiappe, P. (2007). The role of working memory in metaphor production and comprehension. Journal of Memory and Language, 56, 172-188. Cieslicka, A. (2004). Bilingual language users' sensitivity to semantic analyzability of L2 idioms: Testing the effect of idiom analy zability in L2 metali nguistic tasks. In: Arabski, J. (Ed.). Pragmatics and language learning (pp. 143-164) Krakw: Universitas. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd edition). Hillsdale, NJ: Lawrence Erlbaum. Cooper, T.C. (1999). Processing of idioms by L2 learners of English. TESOL Quarterly, 33, 233-267. Crutchley, A. (2007). Comprehension of idiomatic verb + particle constructions in 6to 11-year-old children. First Language, 27 203-226. Cummins, J. (2000). Language, power, and pedagogy: Bilingual children in the crossfire. Clevedon, England: Mu ltilingual Matters. Cutting, L.E., & Scarborough, H.S. (2006). Pr ediction of reading comprehension: Relative contributions of word reco gnition, language proficiency, and other

PAGE 159

148 cognitive skills can depend on how comprehension is measured. Scientific Studies of Reading, 10, 277-299. Dale, E. (1965). Vocabulary measurement: Techniques and major findings. Elementary English, 42, 82-88. De Avila, E.A. & Duncan, S. E. (1991). Language assessment scales. San Rafael, CA: Linguametrics. Ezell, H.K., & Goldstein, H. (1991). Comp arison of idiom comprehension of normal children and children with mental retardation. Journal of Speech and Hearing Research, 34 812-819. Field, A. (2000). Discovering statistics: Using SPSS for Windows. London, England: Sage Publications, Ltd. Fleege, J.E., Mackay, I.R.A., & Piske, T. (2002). Assessing bilingual dominance. Applied Psycholinguistics, 23, 567-598. Gentner, D., Bowdle, B., Wolff, P., & Borona t, C. (2001). Metaphor is like analogy. In D. Gentner, K.J. Holyoak, & B.N. Kokinov (Eds.), The analogical mind: Perspectives from cognitive science, (pp. 199-254). Cambridge, Massachusetts: The MIT Press. Gibbs, R.W. (1980). Spilling the beans on understanding and memory for idioms in conversation. Memory and Cognition, 8, 149-156. Gibbs, R.W. (1991). Semantic analyzability in childrens unders tanding of idioms. Journal of Speech and Hearing Research, 34, 613-620. Gibbs, R.W., Jr. (1994). The poetics of mind, figura tive thought, language, and understanding. Cambridge, Massachusetts: Camb ridge University Press.

PAGE 160

149 Gibbs, R.W. & Nayak, N. (1989). Psycholinguis tic studies on the syntactic behavior of idioms. Cognitive Psychology, 21 100-138. Giora, R. (2003). Recent trends in literal vs. figurative language research: The case of irony. Speech Pathology. Retrieved on September 29, 2007 from http://www.speechpathology.com/articles/ article_detail.a sp?article_id=51. Glucksberg, S. (2001). Understanding figurative language: From metaphors to idioms. Oxford: University Press. Grosjean, F. (1998). Studying bilinguals: Methodological and conceptual issues. Bilingualism: Language and Cognition, 1 131-149. Hernandez, A., Li, P., & MacWhinney, B. (2005). The emergence of competing modules in bilingualism. Trends in Cognitive Science, 9 220-225. Holtgraves, T. (2005). Context and the compre hension of nonliteral meanings. In H.L. Colston & A.N. Katz (Eds.), Figurative language comprehension: Social and cultural influences (pp. 73-98). Mahweh, NJ: Lawrence Erlbaum. Huber-Okrainec, J., Blaser, S., E., & Dennis, M., (2005). Idiom compre hension deficits in relation to corpus callosum agenesis and hypoplasia in children with spina bifida meningomyelocele. Brain and Language, 93 349-368. Jimnez, R.T. (1994). Understanding and promoting reading comprehension. Bilingual Research Journal, 18 99-119. Johnson, D.D., Johnson, B.V.H., & Schlic hting, K. (2004). Logology: Word and language play. In J.F. Baumann and E.J. Kameenui (Eds.), Vocabulary instruction: Research to practice (pp.179-200). New York: Guilford Press.

PAGE 161

150 Kintsch, W. (1998). Inferences. In Comprehension : A paradigm for cognition (188197). New York: Cambridge University Press. Kintsch, W. (2000). Metaphor comprehe nsion: A computational theory. Psychonomic Bulletin & Review, 7, 257-266. Koda, K. (2007). Reading and language lear ning: Crosslinguistic constraints on second language reading development. Language Learning, 57 (s1), 1-44. Koelsch, N. (2006, November). Improving literacy outcomes in English language learners in high school: Considerations fo r states and district s in developing a coherent policy framework. National High School Center 1-9. Kohler, A.D., & Lazarn, M. (2007). Hisp anic education in the United States. National Council of La Raza, Statistical Brief No. 8. Kohnert, K. & Bates, E. (2002). Balancing bilinguals II: Lexical comprehension and cognitive processing in childre n learning Spanish and English. Journal of Speech, Language, and Hearing Research, 45, 347-359. Kohnert, K., Bates, E., & Hernandez, A. E. (1999). Balancing bilinguals: Lexicalsemantic production and cognitive proces sing in children learning Spanish and English. Journal of Speech, Language, and Hearing Research, 42, 1400-1413. Laufer, B. (2000). Avoidance of idioms in a second language: The effect of L1-L2 degree of similarity. Studia Linguistica, 54 186-196. Lazar, R., Warr-Leeper, G., Nicholson, C ., Johnson, S. (1989). Elementary school teachers use of multiple meaning expressions. Language, Speech, and Hearing Services in Schools, 20, 420-430.

PAGE 162

151 Levorato, M.C. (1993). The acquisition of id ioms and the development of figurative competence. In C. Cacciari & P. Tabossi (Eds.). Idioms. Processing, Structure, and Interpretation (pp. 105-128). Hillsdale, NJ: Lawrence Erlbaum. Levorato, M.C., & Cacciari, C. (1992). Ch ildrens comprehension and production of idioms: The role of context and familiarity. Journal of Child Language, 19 415433. Levorato, M.C., & Cacciari, C. (1995). The effects of different tasks on the comprehension and production of idioms in children. Journal of Experimental Child Psychology, 60 261-283. Levorato, M.C., Nesi, B., & Cacciari, C. (2004). Reading comprehension and understanding idiomatic expressions: A developmental study. Brain and Language, 91, 303-314. Levorato, M.C., Roch, M., & Nesi, B. (2007) A longitudinal study of idiom and text comprehension. Journal of Child Language, 34, 473-494. Liontas, J. (2002). Context and idio m understanding in second languages. Eurosla Yearbook, 2, 155-185. McGregor, K. (2004). Developmental depende ncies between lexical semantics and reading. In K. Apel, B. Ehren, E. R. Silliman, & C. A. Stone (Eds.), Handbook of language and literacy: Development and disorders (pp. 302-317). NY: Guilford. Morrison, L. (2004). Comprehension monitoring in first and second language reading. The Canadian Modern Language Review, 61 77-106. Nagy, W. (2007). Metalinguistic awarene ss and the vocabulary-comprehension connection. In R.K. Wagner, A.E. Muse, & K.R. Tannenbaum (Eds.), In

PAGE 163

152 Vocabulary acquisition: Implica tions for reading comprehension (pp. 52-77). New York: Guilford Press. Nagy, W.E., & Scott, J. (2000). Vocabulary proc esses. In M.L. Kamil, P.B., Mosenthal, P.D., Pearson, & R. Barr (Eds.), Handbook of reading research: Volume III (pp. 269-284). Mahwah, New Jersey: Lawrence Erlbaum. National Reading Panel (2000). Teaching children to read: An evidence-based assessment of the scientific research liter ature on reading and its implications for reading instruction. Washington, DC: National Institute of Child Health and Human Development. Nippold, M.A. (1991). Evaluating and enhanc ing idiom comprehension in languagedisordered students. Language, Speech, and Hearing Services in Schools, 22 100106. Nippold, M.A., & Duthie, J.K. (2003). Mental imagery and idiom comprehension: A comparison of school-age children and adults. Journal of Speech, Language, and Hearing Research, 46 788-799. Nippold, M.A., & Martin, S.T. (1989). Idiom inte rpretation in isolation versus context: A developmental study with adolescents. Journal of Speech and Hearing Research, 32, 59-66. Nippold, M.A., Moran, C., & Schwarz, I. E. (2001). Idiom understanding in preadolescents: Synergy in action. American Journal of Speech-Language Pathology, 10, 169-179.

PAGE 164

153 Nippold, M.A., & Rudinski, M. (1993). Familiarity and transparency in idiom explanation: A developmental study with adolescents. Journal of Speech and Hearing Research, 32, 59-66. Nippold, M.A., & Taylor, C.L. (1995). Id iom understanding in youth: Further examination of familiarity and transparency. Journal of Speech and Hearing Research, 38 426-433. Nippold, M.A., Uhden, L.D., & Schwartz, I.E. (1997). Proverb explanation through the lifespan: A developmental study of adolescents and adults. Journal of Speech, Language, and Hearing Research, 40 245-253. Norbury, C.F. (2004). Factors supporting id iom comprehension in children with communication disorders. Journal of Speech, Language, and Hearing Research, 47, 1179-1193. Ordez, C. L., Carlo, M. S., Snow, C. E ., & McLaughlin, B. (2002). Depth and breadth of vocabulary in two languages: Which vocabulary skills transfer? Journal of Educational Psychology, 94 719-728. Ortony, A., Turner, T.J., & Larson-Shapir o, N. (1985). Cultural and instructional influences on figurative language co mprehension by inner city children. Research in the Teaching of English, 19 25-36. Pearson, B.Z., Fernandez, S.C., & Oller, D. K. (1993). Lexical development in bilingual infants and toddlers: Comp arison to monolingual norms. Language Learning, 43, 93-120.

PAGE 165

154 Pea, E., Bedore, L. M., & Rappazzo, C. (2003). Comparison of Spanish, English, and bilingual children's performance across semantic tasks. Language, Speech, and Hearing Services in Schools, 34, 5-16. Pexman, P.M., & Glenwright, M. (2007). How do typically developing children grasp the meaning of verbal irony? Journal of Neurolinguistics, 20, 178-196. Pierce, C.A., Block, R.A., & Aguinis, H. (2004). Cautionary note on reporting etasquared values from multifactor ANOVA designs. Educational and Psychological Measurement, 64, 916-924. Polk County Public Schools (2007). Polk C ounty Public Schools, Florida. Retrieved March 06, 2007, from http://www.pcsb.k12.fl.us/ Polk County Public Schools (2006). Polk C ounty Public Schools, Florida. Retrieved March 01, 2006, from http://www.pcsb.k12.fl.us/ Pray, L. (2005). How well do commonly used language instruments measure English oral-language proficiency? Bilingual Research Journal, 29 387-409. Proctor, C. P., Carlo, M., August, D., & Snow, C. (2005). Native Spanish-speaking children reading in English: Toward a model of comprehension. Journal of Educational Psychology, 97 246-256. Qian, D.D. (1999). Assessing the roles of de pth and breadth of vocabulary knowledge in reading comprehension. The Canadian Modern Language Review, 56 282 307. Qian, D.D. (2002). Investigating the rela tionship between vocabulary knowledge and academic reading performance: An assessment perspective. Language Learning, 52, 513-536.

PAGE 166

155 Qualls, C.D. & Harris, J.J. (1999). Effects of familiarity on Idiom Comprehension in African American and European American fifth graders. Language, Speech, and Hearing Services in Schools, 30 141-151. Qualls, C.D., Lantz, J.M., Pietrzyk, R.M., Blood, G.W., & Hammer, C.S. (2004). Comprehension of idioms in adolescents w ith language-based learning disabilities compared to their typically developing peers. Journal of Communication Disorders, 37 295-311. Qualls, C.D., OBrien, R.M., Blood, G.W ., & Hammer, C.S. (2003). Contextual variation, familiarity, academic literacy, and rural adolescents idiom knowledge. Language, Speech, and Hearing Services in Schools, 34, 69-79. Ramachandran, V.S. (2005). Grasping metaphors: Researchers tie brai n area to figures of speech. Presented at the American Psyc hological Society annual convention in Los Angeles, May 26-29. Rizzo, S., Sandrini, M., & Papagano, C. (2007). The dorsolateral prefrontal cortex in idiom interpretation: An rTMS study. Brain Research Bulletin, 71, 523-528. School District of Hillsborough County (2005) School District Da ta. Retrieved March 06, 2007, from http://www.firn.edu/doe /eias/flmove/hillsbor.htm. Scott, C.M. (2002). A fork in the road less traveled: Writing intervention based on language profile. In K.G. Butler & E.R. Silliman (Eds.), In Speaking, reading, and writing in children with language learning disabilities: New paradigms in research and practice (pp. 219-229) Mahwah, New Jersey: Lawrence Erlbaum. Searle, J. (1979) Metaphor. In A. Ortony (Ed.), Metaphor and thought (pp. 92-123). Cambridge, England: Cambridge University Press.

PAGE 167

156 Shamey-Tsoory, S., Tomer, R, & Aharon-Peretz J. (2005). The neuroanatomical basis of understanding sarcasm and its rela tionship to social cognition. Neuropsychology, 19, 288-300. Snow, C., & Kim, Y-S. (2007). Large problem spaces: The challenge of vocabulary for English language learners. In R.K. Wagner, A.E. Muse, & K.R. Tannenbaum (Eds.), In Vocabulary acquisition: Implicat ions for reading comprehension (pp. 123-139). New York: Guilford Press. Spector, C.C. (1992). Remediating humor comp rehension deficits in language-impaired students. Language, Speech, and Hearing Services in Schools, 23 20-27. Sprenger, S.A., Levelt, W.J.M., & Kemp en, G. (2006). Lexical access during the production of idiomatic phrases. Journal of Memory and Language, 54, 161-184. Stahl, S. (2003). Vocabulary & readabili ty: How knowing word meaning affects comprehension. Topics in Language Disorders, 23 241-247. Suro, R., Fry, R., Kochar, R., Passel, J., Tafoya, S., Benavidas, T.C., Seaborn, M., Luben, A.F. (2005). Hispanics: A people in motion (pp. 1-19). Pew Hispanic Center. Swinney, D.A., & Cutler, A. (1979). The access a nd processing of idiomatic expressions. Journal of Verbal Learni ng and Verbal Behavior, 18, 523-534. Tabossi, P., & Cacciari, C. (1988). Context effects in the comprehension of idioms. Cognitive Science Society (Eds.), In Proceedings of the 10th annual conference of the Cognitive Science Society. (pp. 90-96). NJ: Lawrence Erlbaum. Tabossi, P. & Zardon, F. (1995). The activation of idiomatic meaning. M. Everaert, E. J. van der Linden, A. Schenk, & R. Schreuder (Eds.), In Idioms: Structural and psychological perspectives. (pp. 273-282). Hillsdale, NJ: Lawrence Erlbaum.

PAGE 168

157 United States Department of Education (USDOE) (2006). National Center for Education Statistics (NCES). Dropout Ra tes in the United States, 2003 Retrieved July 21, 2006, from http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2006062 United States Department of Education (US DOE). (2006). National Center for Education Statistics, The nations report card: Reading 2005. Retrieved February 2, 2007 from, http://nces.ed.gov United States Department of Education (US DOE). (2007). National Center for Education Statistics, The nations report card: Reading 2007. Retrieved October 15, 2007 from, http://nces.ed.gov United States Department of Education (US DOE). (2007). National Center for Education Statistics, The nations report card: Tw elfth Grade Reading. Retrieved July 15, 2007 from, http://nces.ed.gov van Dijk, T.A., & Kintsch, W. (1983). Strategies of discourse comprehension. New York: Academia Press. Wagner, R.K., Francis, D.J., & Morris, R.D. (2005).Identifying Eng lish language learners with learning disabilities: Key challenges and possi ble approaches. Learning Disabilities Research & Practice, 20 6. Weiderhold, J.L., & Bryant, B.R. (2001). Gr ay Oral Reading Test-Fourth Edition. ProEd, Inc. Windsor, J., & Kohnert, K. (2004). The s earch for common ground: Part 1. Lexical performance by linguistically diverse learners. Journal of Speech, Language, and Hearing Research, 47, 877-890.

PAGE 169

158 Woodcock, R., Mather, N., & McGrew, K. (2001). Woodcock Johnson-III Test of Achievement (WJ-III ACH). Itasca, IL: Riverside Publishing, a Houghton Mifflin Publishing Co. Woodcock, R. & Muoz-Sandoval, A. (2001). Woodcock-Muoz language survey normative update comprehensive manual. Itasca, NY: Riverside. Yu, A. (2007). Cronbach Alpha: Educational assessment course. Retrieved on July 18, 2007 from, http://www.creative-wisdom.com/teaching/assessment/alpha.html

PAGE 170

159 Appendices A R

PAGE 171

160 Appendix A: USF Signage fo r Pilot Study Participants

PAGE 172

161 Needed: Bilingual or Monolingu al (English) Undergraduates ages 18-35 I need your help! I am piloting an idiom test and need undergraduate participants ages 18-35. Idioms are a type of figurative language, like kick the bucket The test will take 15-20 minutes of your time. If interested please email Be linda Fust-Herrmann ASAP: belinda.fuste@verizon.net b e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t B e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t b e l i n d a f u s t e @ v e r i z o n n e t

PAGE 173

162 Appendix B: Non-significant t -test Results

PAGE 174

163 Non-significant t-test Results t-test comparison groups p values t observed values USF undergraduate age differences between language groups 0.15 -1.49 Syllable count differences between Form A and Form B 0.89 -0.14 Total familiarity rating differences between Form A and Form B 0.95 0.06 Response accuracy total differences between Form A and Form B 0.92 -0.10 Scores on unfamiliar idioms in context for skilled and lessskilled comprehension groups 0.11 -1.618 Total literal responses for skilled and lessskilled comprehension groups .083 .282

PAGE 175

164 Appendix C: Pilot Study Id iom Comprehension Measure

PAGE 176

165 Idiom Comprehension Measure Number: _________ Date: ___________ Are you bilingual? Yes_____ No ____ Have you lived in the US more than 5 years ____ Less than 5 years _____ ? Idioms are figurative or non-literal language like raining cats and dogs or bought the farm. I am creating an idiom test and need your help piloting this test before giving it to bilingual and monolingual high school students in the future. Their results will be compared to their reading and vocabulary scores to investigate any meaningful relationships. Please read each question carefu lly and then circle the best answer. There may be idioms that you do not know and will guess their mean ings. It is important to work forward, and not to go back to change answers. If at any ti me you wish to stop completing this form you may do so without any consequences whatso ever. This is completely voluntary. If you have any questions feel free to ask me. I would like to thank you for participating. Familiar: Transparent: Out of Context 1. Hold ones head up a) To prop ones head up with his hand b) To be brave and/or proud c) To be angry and/or upset 2. Go by the book a) To admire a novels character b) To read a lot c) To follow the rules

PAGE 177

166 3. Take someone under ones wing a) To give someone your seat on a plane b) To offer someone guidance c) To teach someone to fly 4. Blow off steam a) To get rid of stress b) To ignore a pot of boiling water c) To ride a steam boat 5. Keep a straight face a) To laugh in someones face b) To have plastic surgery on your face c) To not smile 6. Right under my nose a) To find in an obvious, nearby place b) To treat someone unfair, or unkind c) To have a thin mustache under your nose 7. Crying over spilled milk a) To cry because the mi lk was split on the floor b) To cry over something that has already happened c) To complain about someones cooking

PAGE 178

167 8. Burning the candle at both ends a) To let a candles wick burn at the top and the bottom b) To work and/or play too hard without enough rest c) To not be wasteful 9. Hold your tongue a) To tell a lie b) To pinch your tongue between your fingers c) To keep quiet 10. Get off on the wrong foot a) To make a bad start b) To have a limp c) To follow someones lead 11. Take a shot in the dark a) To shoot a gun at night b) To be worse than expected c) To take a guess 12. The early bird catches the worm a) The one who arrives early will be successful b) Worms are only available in the morning c) The one who can keep a secret is trustworthy

PAGE 179

168 Familiar: Opaque: Out of Context 1. Beat around the bush a) To beat a bush with a stick b) To avoid a topic c) To win a race by the length of a bush 2. Bring the house down a) To make others applaud a spectacular performance b) To make a room full of people angry c) To tear down a house with a bulldozer 3. Paint the town red a) To make everyone mad in town b) To go out and celebrate c) To paint a big city, like New York, red 4. Have a soft spot in ones heart a) To have a pain in ones heart b) To have a heart murmur c) To be fond of something or someone 5. Chip off the old block a) To act or look like ones parent(s)

PAGE 180

169 b) To live on the same block as ones family c) To save a piece of bric k from a houses foundation 6. Spill the beans a) To lie to someone b) To tell a secret c) To drop a pot of freshly cooked beans 7. At the drop of a hat a) To do as soon as it is convenient b) To change into a uniform with a hat c) To do something immediately, without pressure 8. Go to pot a) To put in the trash can b) To deteriorate c) To go to the bathroom 9. Wet behind the ears a) To be inexperienced b) To be a good swimmer c) To comb your hair back behind your ears 10. Jump through hoops a) To be in the circus

PAGE 181

170 b) To do whatever one is told c) To be a good athlete 11. Go cold turkey a) To not heat up the turkey b) To know something really well c) To stop an addictiv e behavior immediately 12. To flip ones lid a) To open the hood b) To be ecstatic c) To be very angry

PAGE 182

171 Familiar: Transparent: In Context 1. Hold ones head up After Judys teacher notices her chea ting on an exam, Judy finds it hard to hold her head up. a) To prop ones head up with his hand b) To be brave and/or proud c) To be angry and/or upset 2. Go by the book Officer Knack is a nice guy, but he never lets a criminal get away with a crime. He goes by the book. a) To admire a novels character b) To read a lot c) To follow the rules 3. Take someone under ones wing The more experienced pilot taught the ne wcomer, Jerry, how to fly the jet. He took Jerry under his wing. a) To give someone your seat on a plane b) To offer someone guidance c) To teach someone to fly

PAGE 183

172 4. Blow off steam Alex had had a difficult week at work. He could not wait to blow off steam once Friday night arrived. a) To get rid of stress b) To ignore a pot of boiling water c) To ride a steam boat 5. Keep a straight face Barbara was an experienced practical joker, but after seeing Janes face it was hard to keep a straight face. a) To laugh in someones face b) To have plastic surgery on your face c) To not smile 6. Right under my nose Steve trusted all of his family and friends. Thats why it was so hard to accept that the thief was right under his nose. a) To find in an obvious, nearby place b) To treat someone unfair, or unkind c) To have a thin mustache under your nose

PAGE 184

173 7. Crying over spilled milk Reece had spent her last dime on ingredient s for her and Lindseys dinner. But when Lindsey accidentally knocked the pot of s oup onto the floor and began to weep, Reece said, There is no use crying over spilt milk . a) To cry because the mi lk was split on the floor b) To cry over something that has al ready happened and cannot be reversed c) To complain about someones cooking 8. Burning the candle at both ends Shirley believed in using every bit of her time in the day. She worked two jobs and went out every night until dawn. Her frie nds always told her that she was burning the candle at both ends. a) To let a candles wick burn at the top and the bottom b) To work and/or play too hard without enough rest c) To not be wasteful 9. Hold your tongue Chad knew that Bob had taken Sues bike. Bu t, when Sue asked Chad and Bob who took it, Chad held his tongue. a) To tell a lie b) To pinch your tongue between your fingers c) To keep quiet

PAGE 185

174 10. Get off on the wrong foot Susan wanted to be on the marching band at school more than anything, but she was late to her first practice and forgot her drum s ticks. The band leader thought that Susan had gotten off on the wrong foot. a) To make a bad start b) To have a limp c) To follow someones lead 11. Take a shot in the dark Steve did not have time to study for his exam. For the essay question he took a shot in the dark. a) To shoot a gun at night b) To be worse than expected c) To take a guess 12. The early bird catches the worm Martha packed her briefcase the night before her interview. She was prepared because she knew that the early bird catches the worm. a) The one who arrives early will be successful b) Worms are only available in the morning c) The one who can keep a secret is trustworthy

PAGE 186

175 Familiar: Opaque: In Context 1. Beat around the bush Mark failed his big science test. When Ma rks mom asked how biology class was going, Mark started telling her about his English project. But Marks mom knew something was wrong, and that he was just beating around the bush a) To beat a bush with a stick b) To avoid a topic c) To win a race by the length of a bush 2. Bring the house down Sara had practiced her trumpet solo for a w hole month. When her band finally played in the club, she blew her trumpet with so much enthusiasm that she brought the house down. a) To make others appla ud a spectacular performance b) To make a room full of people angry c) To tear down a house with a bulldozer 3. Paint the town red Bobby just graduated from New York Universi ty. To celebrate he and his friends went out and painted the town red. a) To make everyone mad in town b) To go out and celebrate c) To paint a big city, like New York, red

PAGE 187

176 4. Have a soft spot in ones heart Even though Jasmine was allergic to cats, she had a soft spot in her heart for the orange and black stray, and always let her in for a dish of milk. a) To have a pain in ones heart b) To have a heart murmur c) To be fond of something or someone 5. Chip off the old block Jose was a hard worker who had already b een successful in his career and bought a home by age 25. His family and neighbors said that he was a chip off the old block. a) To act or look like ones parent(s) b) To live on the same block as ones family c) To save a piece of bric k from a houses foundation 6. Spill the beans Sandra felt so guilty about what she had done to her little brother that she eventually spilled the beans about how his game boy got broken. a) To lie to someone b) To tell a secret c) To drop a pot of freshly cooked beans 7. At the drop of a hat Xavier really admired his grandmother. Anytime she would ask him to come visit he would do so at the drop of a hat. a) To do as soon as it is convenient

PAGE 188

177 b) To change into a uniform with a hat c) To do something immediately, without pressure 8. Go to pot Nell was so disappointed when she went back to her old neighborhood. Everything was so dirty and full of garbage and had really gone to pot. a) To put in the trash can b) To deteriorate c) To go to the bathroom 9. Wet behind the ears Jack watched his new teammates do the butterf ly back and forth in the swimming pool. He longed to be that good, but right now he was new to the team and a little wet behind the ears. a) To be inexperienced b) To be a good swimmer c) To comb your hair back behind your ears 10. Jump through hoops Nancy wanted to be a part of the group more than anything. For this reason, she was willing to jump through hoops to be accepted. a) To be in the circus b) To do whatever one is told c) To be a good athlete 11. Go cold turkey

PAGE 189

178 John had tried to quick smoking many times. This time he was not going to gradually stop smoking though, he was going to go cold turkey. a) To not heat up the turkey b) To know something really well c) To stop an addictiv e behavior immediately 12. To flip ones lid Tyrones parents were away for the weekend. He had promised not to invite anyone over to the house while they were gone. When hi s parents returned to see the house in shambles from a party, they flipped their lids. a) To open the hood b) To be ecstatic c) To be very angry

PAGE 190

179 Unfamiliar Idioms: Transparent: Out of Context 1. To be caught between two fires a) To be in the middle of flames b) To be in a hurry c) To be caught between two difficult choices 2. To run around like scalded pigs a) To rush about crazily b) To be even worse than anticipated c) To squeal a lot 3. For a good hunger there is no hard bread a) Hard bread is better when you are starving b) To bore someone c) Anything tastes good when you are hungry 4. To shoot sparrows with cannons a) To defeat the enemy without exhausting oneself b) To use excessive means to fulfill an objective c) To kill many birds at once 5. To be drowning in a glass of water a) To be upset over nothing b) To hit a snag

PAGE 191

180 c) To swallow too much water and choke 6. To try to make a hole in water a) To dive into the water b) To make a good impression c) To try to do something that is impossible 7. To hold someones leg a) To wait a while b) To bore someone with endless conversation c) To make someone fall down 8. Its the water drop that makes the vase overflow a) The last thing that happened that finally made you upset b) To exaggerate the situation c) To waste water 9. To fall down with four horseshoes up in the air a) To be embarrassed b) To fall flat on ones back c) To fall down while playing horseshoes 10. To fall into the apples a) To pass out b) To become rich c) To fall while picking fruit

PAGE 192

181 11. To cut a pear in two a) To split a snack b) To meet in the middle c) To argue about something small 12. To throw flowers to somebody a) To throw flowers during a parade b) To speak highly of someone c) To squander money

PAGE 193

182 Unfamiliar Idioms: Opaque: Out of Context 1. The turtle is shrouded a) The sky is foggy b) Someone is undercover c) To be selfish 2. To eat the leaf a) To be a vegetarian b) To be late to work c) To keep a secret 3. To pet the horse first a) To win a bet at the tr ack you have to arrive early b) Rushing into something leads to mistakes c) To get up early 4. To be at the green a) To be out of money b) To be a novice c) To be at the golf course 5. To have salt in your pumpkin a) To make something sour

PAGE 194

183 b) To be intelligent c) To be arrogant 6. To whistle in your thumb a) To be quiet b) To avoid talking about something c) To be unable to get what you want 7. To put ones finger into ones eye a) To have influence b) To poke oneself in the eye c) To be entirely mistaken 8. To pick up a log a) To fall down and hurt oneself b) To hurry up c) To gather wood for a fire 9. To eat on the thumb a) To grab a bite to eat b) To eat small appetizers c) To eat too much 10. To play the donkey to get bran a) To play a childs game

PAGE 195

184 b) To play dumb c) To get on someones nerves 11. Between dog and wolf a) At dusk b) A dog having wolf characteristics c) To be cruel 12. To lay a rabbit on someone a) To tell a lie b) To stand someone up c) To try to hold on to a fast animal

PAGE 196

185 Unfamiliar: Transparent In Context 1. To be caught between two fires June was a star tennis pl ayer at her high school. She ha d to make a decision quickly, because she was running out of time. She had to decide if she wanted to go the university that had a girls tennis team or the one that she had a sc holarship to attend. She was caught between two fires. a) To be in the middle of flames b) To be in a hurry c) To be caught between two difficult choices 2. To run around like scalded pigs The twins had waited until the day of the party to buy all the refreshments and decorations. At 6pm, an hour before the party, they were running around like scalded pigs. a) To rush about crazily b) To be even worse than anticipated c) To squeal a lot 3. For a good hunger there is no hard bread Jason had been hiking all day, and had forgotten to pack his lunch with him. By the time he made it home he was star ving. His mother said that a ll she was making for dinner was leftovers. Jason told her that for a good hunger there is no hard bread. a) Hard bread is better when you are starving b) To bore someone c) Anything tastes good when you are hungry 4. To shoot sparrows with cannons

PAGE 197

186 Matt wanted to win the contest more than a nything. He put up posters urging his peers to vote for him, promised prizes to those who did, and rented a bullhorn to remind people to vote for him the next day. Some people voted for him, but some thought that he was shooting sparrows with cannons. a) To defeat the enemy without exhausting oneself b) To use excessive means to fulfill an objective c) To kill many birds at once 5. To be drowning in a glass of water Julie had studied all night for her exam. When she received a B on it, she was hysterical. Her friends heard her complaints and told her that she was just drowning in a glass of water. a) To be upset over nothing b) To hit a snag c) To swallow too much water and choke 6. To try to make a hole in water Jeremy only had $2.35 dollars to spend. When he continued to believe that he could buy a train ticket and have enough money for l unch, his friends told him that he was trying to make a hole out of water. \a) To dive into the water b) To make a good impression c) To try to do something that is impossible 7. To hold someones leg Jill has a reputation for talking about her pet birds obsessively. When Terry was finally able to walk away from Jill at the party, she told Matt, Jill really knows how to hold someones leg. a) To wait a while

PAGE 198

187 b) To bore someone with endless conversation c) To make someone fall down 8. Its the water drop that makes the vase overflow Sam had been late to work several times and had left early almost every day. Stacey worked with Sam, and did not think that hi s behavior was fair. Sh e thought about talking to her boss but didnt want to get Sam in trouble. One day Sam was supposed to take Staceys place after her shift. When Sam came in to work so late that Stacey missed her bus home, it was the water drop that made the vase overflow. a) The last thing that happene d that finally made you upset b) To exaggerate the situation c) To waste water 9. To fall down with four horseshoes up in the air Sean tried to run home to get out of the co ld, but there was a big patch of ice on the pavement and he fell down with four horseshoes up in the air. a) To be embarrassed b) To fall flat on ones back c) To fall down while playing horseshoes 10. To fall into the apples When Sheila got the news over the phone, she was so surprised that she fell into the apples. a) To pass out b) To become rich c) To fall while picking fruit

PAGE 199

188 11. To cut a pear in two Jimmy wanted to go to the mall, but Sydney wanted to go to the movies. They decided to cut a pear in two and do both. a) To split a snack b) To meet in the middle c) To argue about something small 12. To throw flowers to somebody Rachel respected her teacher, and when someone asked her about Mr. Feder she threw flowers to him. a) To throw flowers during a parade b) To speak highly of someone c) To squander money

PAGE 200

189 Unfamiliar Idioms: Opaque: In Context 1. The turtle is shrouded Bill had a hard time driving down the mountain in the morning. It had been raining and the visibility was low because the turtle was shrouded. a) The sky is foggy b) Someone is undercover c) To be selfish 2. To eat the leaf Sandy told Gina not to tell anyone wh at she had said. Gina promised to eat the leaf. a) To be a vegetarian b) To be late to work c) To keep a secret 3. To pet the horse first Jacob had not waited for the paint to dry befo re loading in the furniture. He ruined the new paint job. His mother said, Thats what happens when you pet the horse first. a) To win a bet at the trac k you have to arrive early b) Rushing into something leads to mistakes c) To get up early 4. To be at the green Lindsey went to the bank and was surprised th at she was not able to withdraw any money from the ATM. She did not realize that she was at the green

PAGE 201

190 a) To be out of money b) To be a novice c) To be at the golf course 5. To have salt in your pumpkin Ginny had passed all her exams and had gotte n onto the honor role. Her teachers and classmates all knew that she had salt in her pumpkin. a) To make something sour b) To be intelligent c) To be arrogant 6. To whistle in your thumb Leslie wanted a new car more than a nything, but without a paycheck she was whistling in her thumb. a) To be quiet b) To avoid talking about something c) To be unable to get what you want 7. To put ones finger into ones eye When Saras boss accused her of leaving early, Sara protested. Sara showed her boss her timecard to prove that she hadnt left and told him that he put his finger in his eye. a) To have influence b) To poke oneself in the eye c) To be entirely mistaken 8. To pick up a log

PAGE 202

191 Karen did not realize that the temperatur e had dropped and the sidewalk had frozen. When she tried to run across the street she picked up a log. a) To fall down and hurt oneself b) To hurry up c) To gather wood for a fire 9. To eat on the thumb Thomas was in a hurry to get to school a nd missed breakfast. On the way to the bus he ate on the thumb. a) To grab a bite to eat b) To eat small appetizers c) To eat too much 10. To play the donkey to get bran Max was the only one home after school. Wh en his mom came home and asked who had eaten all the cake, Max played the donkey to get bran. a) To play a childs game b) To play dumb c) To get on someones nerves 11. Between dog and wolf Zoe was supposed to be home before dark. Her parents wee pleased when she arrived between dog and wolf. a) At dusk b) A dog having wolf characteristics c) To be cruel

PAGE 203

192 12. To lay a rabbit on someone Philip waited on Stanley for nearly an hour at th e park before he rea lized that Stanley had laid a rabbit on him. a) To tell a lie b) To stand someone up c) To try to hold on to a fast animal

PAGE 204

193 Appendix D: USF English Consent Forms

PAGE 205

194

PAGE 206

195

PAGE 207

196 Appendix E: Pilot Study Synopsis

PAGE 208

197

PAGE 209

198

PAGE 210

199 Appendix F: Polk County Public Schools Approval

PAGE 211

200 Ms. Fuste-Herrmann, The Polk County Public Schools Research Review Board has conditionally approved your "Idiom Comprehension and Reading Comprehension" research proposal. Final approval will be granted upon satisfacto ry completion of the following: Documentation of final IRB appr oval from your university Please submit this documentation to my attention at the office of Assessment, Accountability, and Evaluat ion as soon as it becomes available. Martha Santiago, Director of ESOL, will be your district contact. Please contact her before beginning your pr oject and keep her aware of your progress. A copy of your final research report must be s ubmitted to her office and my office upon competition. If you have any questions, or if I can be of any further assistance, please contact me at the phone number or email address below. Thanks, Morgan Platt Polk County Public Schools Evaluation & Research, Senior Coordinator Assessment, Accountab ility & Evaluation (863)534-0736 morgan.platt@polk-fl.net

PAGE 212

201 Appendix G: Polk County Public Sc hools Consent Form (Pilot Study)

PAGE 213

202

PAGE 214

203

PAGE 215

204 Appendix H: Familiarity Forms

PAGE 216

205 Number: ____ Date: _____ Familiarity Rating Form Idioms are figurative or non-literal language like raining cats and dogs or bought the farm. As you remember, I am creating an idiom test and need your help to decide which idioms on my list are familiar to you. Please read each idiom carefully and decide if you a) Know it, b) Heard it before, but do not know what it means, or c) Never heard it before. Then, just place a check mark under the appropriate column. If at any time you wish to stop completing this form you may do so without any consequences whatsoever. This is completely voluntary. If you have any questions feel free to ask me. Thank you for participating. Idiom Know it Heard it, but dont know what it means Never heard it before 1. Hold ones head up 2. Go by the book 3. Take someone under ones wing 4. Blow off steam 5. Keep a straight face 6. Right under my nose 7. Cried over spilled milk 8. Burning the candle at both ends 9. Hold your tongue

PAGE 217

206 10. Get off on the wrong foot 11. Take a shot in the dark 12. The early bird catches the worm 13. Beat around the bush 14. Bring the house down 15. Paint the town red 16. Have a soft spot in ones heart 17. Chip off the old block 18. Spill the beans 19. At the drop of a hat 20. Go to pot 21. Wet behind the ears 22. Jump through hoops 23. Go cold turkey 24. To flip ones lid

PAGE 218

207 25. To be caught between two fires 26. To run around like scalded pigs 27. For a good hunger there is no hard bread 28. To shoot sparrows with cannons 29. To be drowning in a glass of water 30. To try to make a hole in water 31. To hold someones leg 32. Its the water drop that makes the vase overflow 33. To fall down with four horseshoes up in the air 34. To fall into the apples 35. To cut a pear in two 36. To throw flowers to somebody 37. The turtle is shrouded 38. To eat the leaf 39. To pet the horse first

PAGE 219

208 40. To be at the green 41. To have salt in your pumpkin 42. To whistle in your thumb 43. To put ones finger into ones eye 44. To pick up a log 45. To eat on the thumb 46. To play the donkey to get bran 47. Between dog and wolf 48.To lay a rabbit on someone Do you have any comments about this task or idioms in general? Thank you for your help!

PAGE 220

209 Appendix I: New Idiom Comprehension Measure

PAGE 221

210 Idiom Comprehension Measure: Form A Number: _________ Date: ___________ Words and phrases can have several meanin gs. Read these phrases and circle the answer that means the same. Some phrases will be in a short story, and some will be alone. There is only one answer for each question. Familiar: Transparent: Out of Context 1. Hold your head up a) To prop your head up with your hand b) To be brave and/or proud c) To be angry and/or upset 2. Take someone under your wing a) To give someone your seat on a plane b) To offer someone guidance c) To teach someone to fly 3. Right under my nose a) To find in an obvious, nearby place b) To treat someone unfair, or unkind c) To have a thin mustache under your nose

PAGE 222

211 4. Cry over spilled milk a) To cry because the mi lk was split on the floor b) To cry over something that has already happened c) To complain about someones cooking 5. Hold your tongue a) To tell a lie b) To pinch your tongue between your fingers c) To keep quiet

PAGE 223

212 Familiar: Opaque: Out of Context 1. Bring the house down d) To make others applaud a spectacular performance e) To make a room full of people angry f) To tear down a house with a bulldozer 2. Paint the town red d) To make everyone mad in town e) To go out and celebrate f) To paint a big city, like New York, red 3. Chip off the old block a) To act or look like your parent(s) b) To live on the same block as your family c) To save a piece of bric k from a houses foundation 4. At the drop of a hat a) To do as soon as it is convenient b) To change into a uniform with a hat c) To do something immediately, without pressure 5. Jump through hoops a) To be in the circus

PAGE 224

213 b) To do whatever you are told c) To be a good athlete

PAGE 225

214 Familiar: Transparent: In Context 1. Go by the book Officer Knack is a nice guy, but he never lets a criminal get away with a crime. He goes by the book. a) To admire a novels character b) To read a lot c) To follow the rules 2. Blow off steam Alex had had a difficult week at work. He could not wait to blow off steam once Friday night arrived. a) To get rid of stress b) To ignore a pot of boiling water c) To ride a steam boat 3. Keep a straight face Barbara was an experienced practical joker, but after seeing Janes face it was hard to keep a straight face. a) To laugh in someones face b) To have plastic surgery on your face c) To not smile 4. Take a shot in the dark Steve did not have time to study for his exam. For the essay question he took a shot in the dark. a) To shoot a gun at night

PAGE 226

215 b) To be worse than expected c) To take a guess 5. The early bird catches the worm Martha packed her briefcase the night before her interview. She was prepared because she knew that the early bird catches the worm. a) The one who arrives early will be successful b) Worms are only available in the morning c) The one who can keep a secret is trustworthy

PAGE 227

216 Familiar: Opaque: In Context 1. Have a soft spot in your heart Even though Jasmine was allergic to cats, she had a soft spot in her heart for the orange and black stray, and always let her in for a dish of milk. a) To have a pain in your heart b) To have a heart murmur c) To be fond of something or someone 2. Spill the beans Sandra felt so guilty about what she had done to her little brother that she eventually spilled the beans about how his game boy got broken. a) To lie to someone b) To tell a secret c) To drop a pot of freshly cooked beans 3. Wet behind the ears Jack watched his new teammates do the butterf ly back and forth in the swimming pool. He longed to be that good, but right now he was new to the team and a little wet behind the ears. a) To be inexperienced b) To be a good swimmer c) To comb your hair back behind your ears 4. Go cold turkey John had tried to quick smoking many times. This time he was not going to gradually stop smoking though, he was going to go cold turkey.

PAGE 228

217 a) To not heat up the turkey b) To know something really well c) To stop an addictiv e behavior immediately 5. To flip your lid Tyrons parents were away for the weekend. He had promised not to invite anyone over to the house while they were gone. When hi s parents returned to see the house in shambles from a party, they flipped their lids. a) To open the hood b) To be ecstatic c) To be very angry

PAGE 229

218 Unfamiliar Idioms: Transparent: Out of Context 1. To shoot sparrows with cannons a) To defeat the enemy without exhausting yourself b) To use excessive means to fulfill an objective c) To kill many birds at once 2. To put on the sails a) To take a trip by sea b) To hit a snag c) To leave 3. To try to make a hole in water a) To dive into the water b) To make a good impression c) To try to do something that is impossible 4. To fall down with four horseshoes up in the air a) To be embarrassed b) To fall flat on your back c) To fall down while playing horseshoes 5. To cut a pear in two a) To split a snack b) To meet in the middle

PAGE 230

219 c) To argue about something small

PAGE 231

220 Unfamiliar Idioms: Opaque: Out of Context 1. To eat the leaf a) To be a vegetarian b) To be late to work c) To keep a secret 2. To be at the green a) To be out of money b) To be a novice c) To be at the golf course 3. To whistle in your thumb a) To be quiet b) To avoid talking about something c) To be unable to get what you want 4. To pick up a log a) To fall down and hurt yourself b) To hurry up c) To gather wood for a fire 5. To play the donkey to get bran a) To play a childs game

PAGE 232

221 b) To play dumb c) To get on someones nerves

PAGE 233

222 Unfamiliar: Transparent In Context 1. To run around like scalded pigs The twins had waited until the day of the party to buy all the refreshments and decorations. At 6pm, an hour before the party, they were running around like scalded pigs. a) To rush about crazily b) To be even worse than anticipated c) To squeal a lot 2. For a good hunger there is no hard bread Jason had been hiking all day, and had forgotten to pack his lunch with him. By the time he made it home he was star ving. His mother said that a ll she was making for dinner was leftovers. Jason told her that for a good hunger there is no hard bread. a) Hard bread is better when you are starving b) To bore someone c) Anything tastes good when you are hungry 3. To hold someones leg Jill has a reputation for talking about her pet birds obsessively. When Terry was finally able to walk away from Jill at the party, she told Matt, Jill really knows how to hold someones leg. a) To wait a while b) To bore someone with endless conversation c) To make someone fall down 4. To be a monkey on a branch

PAGE 234

223 Sams friend, John, had talked Sam into moving away from his small hometown to New York City to room with him. John always talked about how wonde rful it was to live there. Sam moved there, but hated it. He could not stand the traffic and the small apartment. He told John, You grew up in a big city and that is why you are a monkey on the branch ! Sam decided to move back home. a) To feel at home b) To exaggerate the situation c) To act like a monkey 5. To throw flowers to somebody Rachel respected her teacher, and when someone asked her about Mr. Feder she threw flowers to him. a) To throw flowers during a parade b) To speak highly of someone c) To squander money

PAGE 235

224 Unfamiliar Idioms: Opaque: In Context 1. To pet the horse first Jacob had not waited for the paint to dry befo re loading in the furniture. He ruined the new paint job. His mother said, Thats what happens when you pet the horse first. a) To win a bet at the trac k you have to arrive early b) Rushing into something leads to mistakes c) To get up early 2. To have salt in your pumpkin Ginny had passed all her exams and had gotte n onto the honor role. Her teachers and classmates all knew that she had salt in her pumpkin. a) To make something sour b) To be intelligent c) To be arrogant 3. To eat on the thumb Thomas was in a hurry to get to school a nd missed breakfast. On the way to the bus he ate on the thumb. a) To grab a bite to eat b) To eat small appetizers c) To eat too much 4. Between dog and wolf Zoe was supposed to be home before dark. Her parents were pleased when she arrived between dog and wolf. a) At dusk

PAGE 236

225 b) A dog having wolf characteristics c) To be cruel 5. To lay a rabbit on someone Philip waited on Stanley for nearly an hour at th e park before he rea lized that Stanley had laid a rabbit on him. a) To tell a lie b) To stand someone up c) To try to hold on to a fast animal

PAGE 237

226 Idiom Comprehension Measure: Form B Number: _________ Date: ___________ Words and phrases can have several meanin gs. Read these phrases and circle the answer that means the same. Some phrases will be in a short story, and some will be alone. There is only one answer for each question. Familiar: Transparent: Out of Context 1. Go by the book a) To admire a novels character b) To read a lot c) To follow the rules 2. Blow off steam a) To get rid of stress b) To ignore a pot of boiling water c) To ride a steam boat 3. Keep a straight face a) To laugh in someones face b) To have plastic surgery on your face

PAGE 238

227 c) To not smile 4. Take a shot in the dark a) To shoot a gun at night b) To be worse than expected c) To take a guess 5. The early bird catches the worm a) The one who arrives early will be successful b) Worms are only available in the morning c) The one who can keep a secret is trustworthy

PAGE 239

228 Familiar: Opaque: Out of Context 1. Have a soft spot in your heart a) To have a pain in your heart b) To have a heart murmur c) To be fond of something or someone 2. Spill the beans a) To lie to someone b) To tell a secret c) To drop a pot of freshly cooked beans 3. Wet behind the ears a) To be inexperienced b) To be a good swimmer c) To comb your hair back behind your ears 4. Go cold turkey a) To not heat up the turkey b) To know something really well c) To stop an addictiv e behavior immediately 5. To flip your lid a) To open the hood b) To be ecstatic

PAGE 240

229 c) To be very angry

PAGE 241

230 Familiar: Transparent: In Context 1. Hold your head up After Judys teacher notices her chea ting on an exam, Judy finds it hard to hold her head up. a) To prop your head up with your hand b) To be brave and/or proud c) To be angry and/or upset 2. Take someone under your wing The more experienced pilot taught the ne wcomer, Jerry, how to fly the jet. He took Jerry under his wing. a) To give someone your seat on a plane b) To offer someone guidance c) To teach someone to fly 3. Right under my nose Steve trusted all of his family and friends. Thats why it was so hard to accept that the thief was right under his nose. a) To find in an obvious, nearby place b) To treat someone unfair, or unkind c) To have a thin mustache under your nose 4. Cry over spilled milk Reece had spent her last dime on ingredient s for her and Lindseys dinner. But when Lindsey accidentally knocked the pot of s oup onto the floor and began to weep, Reece said, There is no use crying over spilt milk .

PAGE 242

231 a) To cry because the mi lk was split on the floor b) To cry over something that has al ready happened and cannot be reversed c) To complain about someones cooking 5. Hold your tongue Chad knew that Bob had taken Sues bike. Bu t, when Sue asked Chad and Bob who took it, Chad held his tongue. a) To tell a lie b) To pinch your tongue between your fingers c) To keep quiet

PAGE 243

232 Familiar: Opaque: In Context 1. Bring the house down Sara had practiced her trumpet solo for a w hole month. When her band finally played in the club, she blew her trumpet with so much enthusiasm that she brought the house down. a) To make others appla ud a spectacular performance b) To make a room full of people angry c) To tear down a house with a bulldozer 2. Paint the town red Bobby just graduated from New York Universi ty. To celebrate he and his friends went out and painted the town red. a) To make everyone mad in town b) To go out and celebrate c) To paint a big city, like New York, red 3. Chip off the old block Jose was a hard worker who had already b een successful in his career and bought a home by age 25. His family and neighbors said that he was a chip off the old block. a) To act or look like your parent(s) b) To live on the same block as your family c) To save a piece of bric k from a houses foundation 4. At the drop of a hat Xavier really admired his grandmother. Anytime she would ask him to come visit he would do so at the drop of a hat.

PAGE 244

233 a) To do as soon as it is convenient b) To change into a uniform with a hat c) To do something immediately, without pressure 5. Jump through hoops Nancy wanted to be a part of the group more than anything. For this reason, she was willing to jump through hoops to be accepted. a) To be in the circus b) To do whatever you are told c) To be a good athlete

PAGE 245

234 Unfamiliar Idioms: Transparent: Out of Context 1. To run around like scalded pigs a) To rush about crazily b) To be even worse than anticipated c) To squeal a lot 2. For a good hunger there is no hard bread a) Hard bread is better when you are starving b) To bore someone c) Anything tastes good when you are hungry 3. To hold someones leg a) To wait a while b) To bore someone with endless conversation c) To make someone fall down 4. To be a monkey on a branch a) To feel at home b) To exaggerate the situation c) To act like a monkey 5. To throw flowers to somebody a) To throw flowers during a parade b) To speak highly of someone

PAGE 246

235 c) To squander money

PAGE 247

236 Unfamiliar Idioms: Opaque: Out of Context 1. To pet the horse first a) To win a bet at the tr ack you have to arrive early b) Rushing into something leads to mistakes c) To get up early 2. To have salt in your pumpkin a) To make something sour b) To be intelligent c) To be arrogant 3. To eat on the thumb a) To grab a bite to eat b) To eat small appetizers c) To eat too much 4. Between dog and wolf a) At dusk b) A dog having wolf characteristics c) To be cruel 5. To lay a rabbit on someone a) To tell a lie b) To stand someone up

PAGE 248

237 c) To try to hold on to a fast animal

PAGE 249

238 Unfamiliar: Transparent In Context 1. To shoot sparrows with cannons Matt wanted to win the contest more than a nything. He put up posters urging his peers to vote for him, promised prizes to those who did, and rented a bullhorn to remind people to vote for him the next day. Some people voted for him, but some thought that he was shooting sparrows with cannons. a) To defeat the enemy without exhausting oneself b) To use excessive means to fulfill an objective c) To kill many birds at once 2. To put on the sails Casey had planned to see the new movie at th e theater all week. Her favorite actors were all in it. But, when she arrived on Saturday an hour before show time and saw the long line wrapped all the way around the building and down the street, she put on the sails and decided to try again another day. a) To take a trip by sea b) To hit a snag c) To leave 3. To try to make a hole in water Jeremy only had $2.35 dollars to spend. When he continued to believe that he could buy a train ticket and have enough money for l unch, his friends told him that he was trying to make a hole out of water. a) To dive into the water b) To make a good impression c) To try to do something that is impossible 4. To fall down with four horseshoes up in the air

PAGE 250

239 Sean tried to run home to get out of the co ld, but there was a big patch of ice on the pavement and he fell down with four horseshoes up in the air. a) To be embarrassed b) To fall flat on your back c) To fall down while playing horseshoes 5. To cut a pear in two Jimmy wanted to go to the mall, but Sydney wanted to go to the movies. They decided to cut a pear in two and do both. a) To split a snack b) To meet in the middle c) To argue about something small

PAGE 251

240 Unfamiliar Idioms: Opaque: In Context 1. To eat the leaf Sandy told Gina not to tell anyone wh at she had said. Gina promised to eat the leaf. a) To be a vegetarian b) To be late to work c) To keep a secret 2. To be at the green Lindsey went to the bank and was surprised th at she was not able to withdraw any money from the ATM. She did not realize that she was at the green a) To be out of money b) To be a novice c) To be at the golf course 3. To whistle in your thumb Leslie wanted a new car more than a nything, but without a paycheck she was whistling in her thumb. a) To be quiet b) To avoid talking about something c) To be unable to get what you want 4. To pick up a log Karen did not realize that the temperatur e had dropped and the sidewalk had frozen. When she tried to run across the street she picked up a log. a) To fall down and hurt oneself

PAGE 252

241 b) To hurry up c) To gather wood for a fire 5. To play the donkey to get bran Max was the only one home after school. Wh en his mom came home and asked who had eaten all the cake, Max played the donkey to get bran. a) To play a childs game b) To play dumb c) To get on someones nerves

PAGE 253

242 Appendix J: Polk County Pub lic Schools Approval Letters

PAGE 254

243

PAGE 255

244

PAGE 256

245

PAGE 257

246 Appendix K: School District of Hillsborough County Approval

PAGE 258

247

PAGE 259

248 Appendix L: IRB Approved Consent Forms (English and Spanish)

PAGE 260

249

PAGE 261

250

PAGE 262

251

PAGE 263

252

PAGE 264

253 Appendix M: Student Assent Form

PAGE 265

254

PAGE 266

255

PAGE 267

256 Appendix N: Inclusion Questionnaire

PAGE 268

257 Inclusion Questionnaire Participant Number: _______ Date of Birth: ______ Grade Level: ______ Date of Study: _______ Criteria Checklist Response Do you have normal or aided hearing? Do you have normal or corrected vision? Are you receiving or are you eligible for speech and language services? Do you only speak English? Are you of Hispanic descent and/or speak Spanish? WJ-III Nonword Subtest Score: _____________

PAGE 269

258 Appendix O: Student Langua ge History Questionnaire

PAGE 270

259 Bilingual Language History Questionnaire Students Name: ________________ Date: _________________ 1. When did you first begin to learn English? 2. When did you first begin to learn Spanish? 3. Were you born in the United States? 4. If not, where were you born? 5. Have you ever attended school in Spanish? 6. If you were born outside of the U.S.: What was the last grade completed in your native country? 7. How many years have you liv ed in the United States? 8. What languages do you: Speak: ___________________ Understand: _______________ Read: ____________________ Write: ____________________

PAGE 271

260 Language Proficiency Rating Scale Participant #___________ Please read each statement carefully and circ le the number/word that best describes your answer. If the question does not apply to you, please circle the number of the question. 1. My mother/guardian speaks Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 2. I speak to my mother/guardian in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 3. My father/guardian speaks Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 4. I speak to my father/guardian in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always

PAGE 272

261 5. My brother(s)/sister(s) speak(s) Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 6. I speak Spanish to my brother(s)/sister(s). 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 7. Most of my family members speak Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 8. I speak to most of my family members in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 9. My neighbors speak Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 10. I speak to my neighbors in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always

PAGE 273

262 11. My friends speak Spanish to me outside of school or on the phone. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 12. I speak to my friends in Spanish outside of school or on the phone. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 13. I speak Spanish at school. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 14. I watch television in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 15. My family watches television in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always

PAGE 274

263 16. I read in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 17. I write in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 18. I listen to music sung in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 19. I email/text message/instant message in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always 20. People email/text message/instant message me in Spanish. 0 1 2 3 4 5 Never Rarely Sometimes Often Mostly Always

PAGE 275

264 Appendix P: Error Detection Paradigm

PAGE 276

265 Participant Number: ________ You will read five factual, s hort stories. Each story is about something different like an animal or a place. Your job is to look for e rrors in the stories. Some sentences in these stories may have errors and some may not. So me examples of errors are misspellings, incorrect verbs, and ideas that do not make sens e with the rest of th e story. For example, look at the errors underlined below: To make a peanut butter and jelly sand wich you need bread, peanut butter, and jelly. You will also need a nife to spread the peanut butter and jelly, as well as to kut the cake in two. Some people also prefer the crusts to be cut off. Either way, peanut butter and jelly sandwiches are messy, so you will not need a napkin. Peanut butter and jelly sandwiches is very popular. There are two misspelled words underlined: kut /cut and nife/knife. Also, there are two examples of ideas that do not go with the rest of the stor y. The story is about making a peanut butter and jelly sa ndwich, so cutting a cake does not fit with the storys main idea. The next error is the word not The reading first says that peanut butter and jelly sandwiches are messy, so it should say that you will need a napkin instead of saying you will not need a napkin. The last error is should say are since sandwiches is in the plural form, meaning more than one. Now it is your turn to find these types of errors in each of the five stories below. After reading every one or two sentences in each story you will be asked if there is an error and, if there is, to underline it. Remember that not all of the sentences will have an error.

PAGE 277

266 Turkeys 1. A turkey raised for food weighs twice as much as a wild turk ey. Wild turkeys can fly, but turkeys raised for food are too light to fly. Is there an error in the above senten ces? If so, please underline it. 2. Wild turkeys eat food such as acorns, seeds, insects, and berries. Is there an error in the above senten ce? If so, please underline it. 3. A female turkey lays about 18 eggs at a time and chicks hatch in one month. Is there an error in the above senten ce? If so, please underline it. 4. The skin on a wild turkey's throat can ch ange color. It changes from gray to shades of red, white, and blue wen the turkey is in danger. Is there an error in the above senten ces? If so, please underline it.

PAGE 278

267 Greece 1. The country of Greece is about the size of the state of Alabama. Is there an error in the above senten ce? If so, please underline it. 2. In spite of its small size, Greece has about 8,500 miles of coastline. Is there an error in the above senten ce? If so, please underline it. 3. The United States, which is tiny when compared to Greece, has 12,300 miles of coastline. Is there an error in the above senten ce? If so, please underline it. 4. Greece have a lot of coastline because it has more than 3,000 islands. Is there an error in the above senten ce? If so, please underline it.

PAGE 279

268 Flamingos 1. Flamingos build a nest by making piles of mud. The mother and father flamingo take turns sitting on the mother's egg. Is there an error in the above senten ces? If so, please underline it. 2. A flamingo's color comes from the shrimp and other creatures it eats. A flamingo can look for food in deep water becau se its legs are so short. Is there an error in the above senten ces? If so, please underline it. 3. Flamingos eat by sucking up water and mud. They pump the water and mud out of their bills and trap small water creatures inside their mouths. Is there an error in the above senten ces? If so, please underline it. 4. Flamingos must run a few step to gain the speed they need to fly. Is there an error in the above senten ce? If so, please underline it.

PAGE 280

269 Deserts 1. Deserts can be dry places, but no desert is as wet as the Atacama Desert in northern Chile. Is there an error in the above senten ce? If so, please underline it. 2. Parts of this desert will not see a single drop of reign this year. Is there an error in the above senten ce? If so, please underline it. 3. At one time, Arica, the largest city in nor thern Chile, did not see rain for 14 years. Is there an error in the above senten ce? If so, please underline it. 4. As a matter of fact, some part s of this desert have not seen rain in 400 years! Did you have an idea that any place on Earth could be that dry? Is there an error in the above senten ces? If so, please underline it.

PAGE 281

270 Cheetahs 1. The cheetah is the world's fastest land anim al. It can reach speeds of 70 miles an hour in just 3 seconds. Is there an error in the above senten ces? If so, please underline it. 2. You can tell a cheetah from a leopard by looki ng at its face. Cheetahs have black lines that run from their eyes to their mouths. But it is hard to catch a glimpse of them since cheetahs are so slow. Is there an error in the above senten ces? If so, please underline it. 3. Cheetahs feed on animals such as deer, rabb its, birds, and lizards. Sometimes they eat fruit like watermelon. Is there an error in the above senten ces? If so, please underline it. 4. In the wild, most cheetahs live only 10 to 15 year. Is there an error in the above senten ce? If so, please underline it.

PAGE 282

271 Appendix Q: Percentage of Item Accuracy on the Synonym Task

PAGE 283

272 Percentage of Item Accuracy on the Synonym Task Item Bilinguals Monolinguals Both Groups Combined Puppy 100% 100% 100% Hop 100% 100% 100% Small 100% 100% 100% Pal 100% 100% 100% Ill 94% 100% 97% Lady 94% 90% 97% Repair 94% 100% 97% Difficult 97% 100% 98.5% Exhausted 100% 97% 98.5% Hit 65% 77% 71% Final 94% 94% 94% Entire 100% 97% 98.5% Amusing 39% 71% 55% Blaze 39% 65% 52% Restrain 55% 81% 68% Incinerate 29% 55% 42% Haul 45% 77% 61% Consume 61% 77% 69% Residence 35% 71% 53% Tarry 3% 13% 8% Stratagem 0% 3% 1.5%

PAGE 284

273 (Appendix Q continued) Cogitate 3% 6% 4.5% Capacious 10% 6% 8% Upbraid 0% 3% 1.5% Fallow 0% 3% 1.5% Evanescent 0% 3% 1.5%

PAGE 285

274 Appendix R: Nonsignificant ANOVA and t-test Findings

PAGE 286

275 Nonsignificant ANOVA and t-test Findings Age Differences: observed t (1, 60) = 0.9311, p = .64 Group Differences in Literal: (F (1, 60) = 3.229, MSE = 11.758, p = .077) Figurative Responses: (F (1, 60) = 2.324, MSE = 10.081, p = .133)

PAGE 287

276 Footnote 1 Caution was used in interpreting partial et a squared as a measure of effect size since, unlike classical eta, partial eta holds other variables constant while measuring the strength of the variable of interest. This procedure can inflate effect sizes, making them appear larger than they actually are (P ierce, Block, & Aguinis, 2004). Bedore, Pea, Garca, and Cortez (2005) cite that one benefit of partial eta is that it is independent of the magnitude or number of other effects. Pres ently, guidelines for interpreting partial eta squared are absent from the literature (Bedore et al., 2005); however, Pea, Bedore, and Rappazzo (2003) adopted guidelines derived from correlation analyses. According to these guidelines, effect sizes between .80 1.00 are considered very large; effect sizes between .50 .80 are considered large; effect sizes betw een .25 .50 are considered moderate; effect sizes between .10 .25 are considered sm all; and effect sizes less than .10 are considered negligible (Be dore et al., 2005; Pe a et al., 2003).

PAGE 288

About the Author Belinda Fust-Herrmann received a Bachelors Degree in Spanish from Appalachian State University in 1997 and a Master of Arts Degree in Speech-Language Pathology from The University of North Caro lina at Greensboro in 2001. She practiced speech-language pathology for two years before beginning the Ph.D. program in the Department of Communication Sciences and Disorders at the Un iversity of South Florida, Tampa. While in the Ph.D. program at the Un iversity of South Florida, Ms. FustHerrmann taught undergraduate and gr aduate courses including Phonology, Multiculturalism, and Language Learning in the School Age Years; and supervised speech-language pathology graduate students during their Diagnostic Practicum. She has presented at annual state and national American Speech, Language, & Hearing Association (ASHA) conventions in 2003, 2004, and 2006. Ms. Fust-Herrmann was first author of a pub lication in a 2006 Learning Disabilities Research & Practice and was awarded the Latino Graduate Fe llowship from 2004 to 2007.