Understanding DLLs’ reading development by exploring child- and text-level factors outside the SVR framework

Understanding DLLs’ reading development by exploring child- and text-level factors outside the SVR framework

First Author: Amy Pratt -- University of California, Irvine
Keywords: English Language Learners (ELL), Reading, Emergent literacy, Reader characteristics, Reader-text interactions
Abstract / Summary: 

Though many Dual Language Learners (DLLs) become proficient readers in English, most do not (NAEP, 2019). The proposed symposium begins by using latent profile analysis to establish that, as early as PreK, DLLs are overrepresented in the “risk for reading failure” profile. Subsequent papers explore both child- and text-level factors that may contribute to variability within and across DLL readers from Kindergarten to fifth grade. Paper 2 investigates how kindergarteners’ metalinguistic skills, such as morphological awareness, predict spelling and word reading in second grade. Paper 3 investigates the extent to which working memory contributes to third graders’ reading comprehension in Spanish and English. Paper 4 explores the effect of text type and difficulty on comprehension among DLLs through 5th grade. Together, this work underscores that reading comprehension is not a unitary construct; rather, complex interactions that vary across readers and texts should be considered in future reading research with DLLs.

Symposium Papers: 

Examining relationships between early learning profiles and DLL status

First Author/Chair:Jill Pentimonti -- American Institutes for Research
Additional authors/chairs: 
Virginia Buysse

Purpose: Creating early learning profiles at pre-k entry can be used to typify groups of children who differ in terms of various dimensions of learning. The results are useful for measuring the gains that children make in pre-k to plan and evaluate instructional interventions. In the present work, we used person-centered methods to examine whether there may be reliable profiles that characterize children’s early language/literacy skills, and the extent to which profile membership may be associated with Dual Language Learner (DLL) status.

Methods: Participants were 400 children enrolled in the New York City Pre-K for All program. Children’s early language/literacy skills were screened in the fall of their pre-k year with the Preschool Early Literacy Indicator which measures the following skills: (a) alphabet knowledge, (b) vocabulary, (c) listening comprehension, and (d) phonological awareness. We used Latent Profile Analysis (LPA) to explore profiles of PreK children in regard to their early language/literacy skills. LPA-derived models were tested for 2- through 5-group solutions.

Results: Accounting for multiple model fit indices, the 5-group model was determined to be the best fit for our data. Next, a multinomial logistic regression was conducted to understand relationship between profile membership and children’s DLL status. A significant relation was found between profile membership and children’s DLL status, such that children who were DLLs were more likely to be placed in Profile 1 (in which scores were low across all dimensions of language/literacy) than Profiles 2-5.

Conclusion: DLLs are at greater risk to begin Pre-K with low skills across multiple dimensions of emergent literacy in English. Quality Pre-K skills instruction is needed to close the gap for this population.

From contenyou to continue: A cross-linguistic model of spelling development in Spanish-English DLLs

First Author/Chair:Kathleen Durant -- Kent State University
Additional authors/chairs: 
Linda Jarmulowicz; Leigh Harrell-Williams

Purpose: This study proposed a developmental model of the predictive relationships between oral language based metalinguistic skills and single word reading and spelling for sequential Spanish-English bilingual students.

Method: 71 Spanish- and English-speaking Dual Language Learners (DLLs) participated in this longitudinal study. A structural equation model investigated the differential predictive relationships between Spanish and English Phonological Awareness (S-PA, E-PA), English Morphological Awareness (MA) and later literacy skills (single word reading and spelling) during the developmental period from kindergarten to 2nd grade for ELLs.

Results: Spanish Phonological Awareness (S-PA) at the end of kindergarten was a significant direct longitudinal predictor of 2nd grade MA and indirect predictor of spelling. English Phonological Awareness (E-PA) at the end of kindergarten was a significant direct longitudinal predictor of 2nd grade English word reading and spelling. Within the model, S-PA was the only significant longitudinal predictor of MA, and E-PA was the most significant predictor of spelling skill, overall.

Conclusion: These results can inform language and literacy assessment procedures for identifying Spanish-speaking DLLs at risk for poor spelling outcomes by demonstrating how L1 and L2 PA differentially support accurate spelling development across early elementary school.

Examining working memory effects on reading comprehension for Spanish-English DLLs

First Author/Chair:Ryan Grimm -- SRI International
Additional authors/chairs: 
Emily Solari; H. Lee Swanson

Purpose: The Simple View of Reading (SVR; Gough & Tunmer, 1986) depicts reading comprehension as a function of two broad subcomponent skills, decoding and linguistic comprehension; each of which have distinct subskills that are predictive of overall reading ability. Cognitive abilities, not delineated in the SVR, have also been shown to contribute to reading comprehension (Aaron, Joshi, Gooden, & Bentum, 2008). There is empirical evidence that the SVR applies to bilingual Spanish- and English-speaking students (Hoover & Gough, 1990), but few studies include cognitive abilities. The present study investigates whether working memory mediates the relationships between subcomponent skills and reading comprehension in both Spanish and English.

Method: 430 Spanish-English Dual Language Learners (DLLs) third-graders who received English-only instruction. Structural equation modeling was used to explore potential mediation. Additionally, cross-language contributions of the subcomponent skills to reading comprehension were also explored.

Results: Mediation results varied by language. In English, working memory did not mediate word reading and reading comprehension, but it partially mediated oral language and reading comprehension. In Spanish, working memory did not mediate either relationship between word reading or oral language and reading comprehension. Regarding cross-language findings, English word reading and oral language both predicted Spanish reading comprehension, but Spanish word reading and oral language did not predict English reading comprehension.

Conclusions: Mediation effects occurred in English, which likely reflects the English-only instruction. Working memory has been shown to impact young students’ word reading and reading comprehension, and this study extends these findings to oral language skills for bilingual Spanish-speaking students.

Effect of text type and text difficulty on DLLs’ reading comprehension

First Author/Chair:Amy Pratt -- University of California, Irvine
Additional authors/chairs: 
Ashley Adams

Purpose: The purpose of the proposed work is to explore how features of text influence the reading comprehension of dual language learners (DLLs). The RAND Reading Study Group Report (Snow, 2002) conceptualizes reading comprehension as the combination of three elements: the reader, the text, and the purpose for reading. Given the complexity of reader-level variables among DLLs, little extant research has explored how text type and text difficulty influence DLLs’ reading comprehension.

Method: Participants included 56 Spanish-English DLLs (mean age = 108 months; SD =11.5 months) who were identified by their teachers as poor comprehenders in English. In addition to a battery of standardized reading and language measures, children read six multi-chapter experimental texts (3 narrative and 3 expository) and answered comprehension questions at the conclusion of each chapter. Texts were coded for difficulty using Flesch Kincaid Grade Level readability scales.

Results: Hierarchical linear models with three levels (classroom, child, text) revealed that both text type and text difficulty significantly predicted children’s English reading comprehension. On average, children performed 23% better on narrative texts. Additionally, for each 1 point increase in difficulty, children’s comprehension performance decreased by an average of 4.4%.

Conclusion: These results highlight the significant role of text in reading comprehension among DLLs, supporting work with monolinguals that suggests that reading comprehension is not a unitary construct. Rather, attention should be given to complex reader-by text-by purpose interactions.

Understanding DLLs’ Reading Development by Exploring Child- and Text-level Factors Outside the SVR Framework

First Author/Chair:DISCUSSANT Maria Adelaida Restrepo -- Arizona State University