Examining the Early Language and Literacy Trajectories of Dual Language Learners

• In Event: Communicating Psychometric Information to Diverse Audiences: From Early Childhood Teachers to Policy Makers

Abstract

1. Objectives
Our objective is to describe the statistical aspects of the early developmental trajectories in language and literacy development (LLD) for children with different language backgrounds, including (a) differences between dual language learners (DLLs) and their monolingual English-speaking peers and (b) differences within groups of children who speak the same home language. This project responds to the systemic need for information about DLLs’ early language and literacy trajectories (August, Shanahan, & Escamilla 2009; CECER-DLL, 2011; NASEM, 2017).

2. Theoretical framework
A theory in the developmental literature states that the results from longitudinal studies have the potential to inform instructional design in ways that will increase educational effectiveness (e.g., Lesaux et al., 2007; Hammer et al., 2014; Hoff, 2013). The idea is that longitudinal research can produce information about developmental trajectories, such as the sequence and rate of development, or document the positive impact of quality education. (Castro, 2014; Goldenberg, 2008). The current study produces longitudinal research on developmental trajectories that may support instructional enhancements.

3. Methods
We will use multilevel growth modeling and longitudinal IRT approaches (each accounting for cross-classified random effects and time-varying covariates) to describe trajectories of LLD for children with different home languages and sociocultural backgrounds. Multilevel modeling, where IRT proficiencies are treated as continuous outcomes, is a common method for analyzing the type of nested educational data in this study (Grady & Beretvas, 2010; Jeon & Rabe-Hesketh, 2012; Luo & Kwok, 2012). We will also model the data using a longitudinal IRT approach, which may produce more accurate results (Choi & Wilson, 2016; Pastor & Beretvas, 2006; Segawa, 2005)

4. Data sources
The longitudinal data consist of approximately 1.53 million DRDP assessments of 651,771 children aged 1-72 months collected between 2015 and 2019. The DRDP is an authentic, observational, formative assessment that measures multiple domains of child development, though this proposal focuses on the LLD domain. In addition to the DRDP scores, which are Rasch IRT scores from a partial credit model, the dataset contains 16 child-level demographics (e.g., home language, ethnicity/race) and a smaller number of site-level variables. Children may be assessed in English or their home language.

5. Results
Preliminary results from a four level (assessment, student, teacher, agency) piecewise longitudinal HLM show that that DLLs who speak Vietnamese at home have different language trajectories than DLLs from Spanish-speaking homes and English Monolingual children. The results show proof in concept of different language trajectories.

6. Scholarly significance
Although early LLD is an emerging area of the DLL research, the question of how do young DLLs develop language and literacy skills in English is a basic gap in the literature in need of research. This study meets that need; the proximate expected outcome of this research is a greater statistical understanding of LLD development for young linguistically diverse children.

Author

• Joshua Sussman, University of California - Berkeley