Spanish-English within and across languages predictions of reading risk based on English proficiency and language of instruction

• In Event: Identifying and Supporting Multilingual Learners Across Systems and Classrooms

Abstract

Introduction and Objectives
Reading is an important skill for educational success and the early identification of struggling readers is crucial to provide timely and effective intervention. Emergent bilinguals are often over-identified as struggling reader because their second-language skills are still developing and most reading screening instruments are developed using an English monolingual calibration sample and paradigm, preventing accurate measurement of emergent bilinguals’ skills, which are likely unevenly distributed across languages. Misidentifications are costly and result in bilingual struggling readers not receiving the support they need, or they can result in the over-identification of learning disabilities (Rodríguez & Rodríguez, 2017).

Methods
We addressed this inequity by developing a free universal reading screener for kindergarteners, first, and second graders in California’s public education system, Multitudes, that has been grounded in multilingual development and measurement. We will report on a longitudinal study (n ~ 1,000) predicting Spanish-English bilingual children’s reading risk at the end of the school year using a combination of comparable, yet language-specific English and Spanish tasks assessing (pre-)reading skills, such as, phonological awareness, decoding, and oral language at the beginning of the year. This will allow us to (a) explore cross-linguistic screening models—that is, for example, screening in Spanish to predict English reading risk for those in English instructional programs; (b) develop multilingual screening protocols that combine students’ scores on both Spanish and English measures to predict risk in either or both languages; (c) explore the differences in the prediction models for children based on their English proficiency and language of instruction, and (d) compare the above approaches to the current status quo, monolingual English or Spanish screening protocols.

Results
A recently concluded pilot study (n = 308) suggests that Spanish screening of bilingual kindergarteners and first graders in English instructional programs (AUC = .74-.86) is superior to English screening (AUC = .68-.70). Results of the full analysis, including results for the multilingual screening protocol disaggregated by English language proficiency as measured by the ELPAC and language of instruction (i.e. dual language versus English-only), will be available by the time of the conference.

Significance
Our inclusive approach to reading screening seeks to explore the accuracy of models that consider bilingual measurement models. These analyses will propel the field forward in modelling the nuances of the predictive value of measures in each language to reading outcome measures in Spanish and English while accounting for English language proficiency and language of instruction.

Authors

• Julian M. Siebert, University of California - San Francisco

• Lillian Durán, University of Oregon

• Monica Pilar Zegers, University of California - San Francisco

• Francesca Pei, University of California - San Francisco

• Marilu Tempini, University of California - San Francisco