Subgrouping English language learners based on English as a foreign language reading measures: A multilevel latent profile analysis

Subgrouping English language learners based on English as a foreign language reading measures: A multilevel latent profile analysis

First Author: Florina Erbeli -- Texas A&M University
Keywords: English Language Learners (ELL), Latent profile analysis
Abstract / Summary: 

Purpose: The goal of the present study was to examine the heterogeneity and prevalence of reading profiles of English as a foreign language students on two levels, the child-level and the school-level, and to investigate child-level and school-level predictors to predict student’s reading profile membership.

Method: The sample comprised of 672 L1 Slovenian seventh graders (Mage = 12.52; SDage = 0.39). A battery of EFL reading tests was administered to the students.

Results: A series of multilevel latent profile analyses with predictors on child- and school-level was carried out. The model with 5 reading profiles on child-level and 2 profiles on school-level fit the data best (LL = -8084.225, Entropy = 0.923, Average Latent Class Probabilities = 0.928-0.989). Of the final 5 reading profiles on child-level, two are worth mentioning: poor readers with word recognition problems and poor readers low on word recognition and language related skills. School-level solution represented 2 profiles: schools where students had a high probability of being an above average student on EFL reading measures and schools where students had a high probability of being an average or below average student on EFL reading measures. Finally, the model with predictors showed that all child-level and two school-level predictors predicted being a poor reader rather than an average or above average reader.

Conclusions: Results showed that interventions particularly in word recognition skills plus a support of contextual factors on child- and school-level should be taken in consideration in an EFL reading instruction.