Using Home Language Surveys to Identify English Learners: Methods to Reduce Misclassification

• In Event: Poster Session 03
  In Poster Session: PS 03 Topic - Methods, History, Theory

Abstract

All states recommend, if not require, the use of a home language survey (HLS) as the first step in identifying English Learners (ELs) in need of language support services (National Research Council, 2011). Bailey and Kelly (2011) note that HLS’s vary considerably, that there is little guidance on how best to account for multiple languages and multiple interlocutors in the home, and they recommend validation studies be done that examine hit rates of HLS’s (i.e., false positives/negatives). Indeed, home language environments are complex - often with more than two adults and more than two languages used in the home (Verdon et al., 2014). However, in our experience, most HLS’s only ask about the language practices of two parents/adults. Thus, we asked: (1) is there a difference in classifying students’ home language background from HLS’s when using language use data in the home involving three vs. just two interlocutors?, and (2) for students whose home language classification changes depending on the number of interlocutors used, in what ways do classifications change?

Analyses were conducted using n = 180 students from an ongoing study examining the experiences of students in kindergarten through third grade enrolled in Spanish-English Dual Language programs across three public school districts. Students’ parents completed an HLS at study enrollment. The HLS added a third interlocutor (i.e., other adult in the home) to an existing set of four questions (Castro et al., 2014) on a 5-point scale (1=All Spanish, 5=All English) that asked: what language (English/Spanish) the student’s mother/P1 typically uses when speaking to the student; what language the student’s father/P2 typically uses when speaking to the student; and what language the student typically uses when speaking to each parent.

Analyses revealed that 32% (n = 57) of the sample indicated that a third adult lived in the home. When classifying student home language based on three interlocutors, the distributions were Mostly Spanish (n = 58, 32%), Approximately Equal Spanish and English (n = 34, 19%), and Mostly English (n = 88, 49%). There were 7 students (4%, of total sample; 12% of those with 3 interlocutors) whose home language group changed when using two interlocutors, which is comparable to and/or above documented rates of misclassification for medical misdiagnoses of 5% (Singh et al., 2014). Importantly, using two rather than three interlocutors in HLS’s appears to most commonly underestimate the presence of Spanish in the home environment (Table 1).

These results have non-trivial implications for identifying students classified as ELs in school systems and for their subsequent academic trajectories. Students identified as ELs have higher language arts and math achievement compared to students who were on the cusp of receiving language services but were not identified as ELs (Shin, 2018). Other studies have found negative effects of being classified as an EL on academic achievement (Umansky, 2018). Improving the precision of HLS’s is needed to ensure that students growing up in bilingual homes have the opportunity to receive language supports that promote positive academic outcomes.

Authors

• Doré R. LaForett, Child Trends - North Carolina
  Presenting Author

• Adam Winsler, George Mason University
  Non-Presenting Author

• Ximena Franco-Jenkins, University of North Carolina at Chapel Hill
  Non-Presenting Author

• M. Taylor Simpson, University of North Carolina at Chapel Hill
  Non-Presenting Author