Assessing Early Literacy With Hispanic Preschoolers: The Factor Structure of the Phonological Awareness Literacy Screening—Español

Abstract

For two decades, it has been recommended that assessment of literacy for preschool children be conducted in a child's primary language. However, only a few literacy assessments have been validated with a preschool, Spanish-speaking population. The purpose of the present study was to test the latent structure of the Phonological Awareness Literacy Screening for Preschool (PALS-PreK) Español with a sample of Spanish-speaking children in the southwestern United States. Children who could recognize at least 16 uppercase and nine lowercase letters were included in the analyses. Subscales of the PreK Español included Name Writing, Letter/Sound Association, Print Concepts, and Rhyme Awareness. Confirmatory factor analysis resulted in a two-factor model of alphabet knowledge (upper and lowercase letter recognition, letter sounds) and print and phonological awareness (name writing, print and rhyme awareness, letter sounds) with letter sounds loading on both factors. We found that the structure of PALS-PreK Español is similar to the English version and discussed implications.

Internal Structure and Item Characteristics of the Phonological Awareness Literacy Screening in Spanish for Preschool

“…The LS task loaded on both factors. Yaden, Marx, Cimetta, Alkhadim, and Cutshaw (2017) replicated the work of Townsend and Konold using a short, draft version of PALS español PreK. They too found that the same two correlated factors fit better than a single-factor model.…”

“…Nevertheless, his work championed the bifactor model, and it is consistent with our results. Considering that Yaden et al (2017) found support for a model with two correlated factors when using task-level item parcels but we did not when conducting an item-level analysis, the analytic procedure may be affecting the results. It is difficult to know for sure as Yaden et al used a shorter, preliminary version of the measure.…”

“…Another reason for not finding more support for the two-tier model that represented the simple view of early literacy may be attributed to the analytic procedures and the way the relationship between the two dimensions is represented. Prior factor analytic work with PALS assessment involved item parcels at the task level, which prevented the use of either a bifactor or testlet model (see Townsend & Konold, 2010; Yaden et al, 2017). Huang (2014) used item parcels formed from pairs of items, and this method enabled him to fit correlated factor models and a bifactor model to the data.…”

“…The simple view of reading entails two traits (Oral Language and Code-Related Skills), which have been shown to be highly correlated (Dickinson et al, 2003; Kendeou et al, 2009; Lonigan et al, 2000; NICHD Early Child Care Research Network, 2005). This relationship among traits seems to be adequately captured as either a model with correlated factors (Townsend & Konold, 2010; Yaden et al, 2017) or a model with a single general dimension and multiple specific dimension that reflect either domain-specific (Huang, 2014) or task-specific shifts in the construct. Thus, the main difference in these types of models may be more a matter of measurement utility than theoretical import.…”