Co-Principal Investigator: Cabell, Sonia Q.

Purpose: Early language development is critical to young children's school readiness. Strong language skills support children's academic achievement in reading, math, and broader academic areas. Substantial numbers of young children enter kindergarten with weak language skills, placing them at-risk for disabilities and ongoing school difficulties. Despite this, all preschool programs do not provide children with sufficient opportunities to develop their language skills. Too many classrooms provide low-quality instructional and environmental support for children's language development. However, the quality of the language environment in preschool classrooms can be improved to lead to better child outcomes. This project, with diverse settings, robust measures, multiple measurement waves and sophisticated analytic modeling, improves the scale and scope of prior work to support better understanding of teachers' linguistic interactions with children. Researchers will focus on three quality facets of linguistic features of teacher talk across five distinct interaction settings, and how they each predict children's language skill growth. The team will identify the most important teacher linguistic behaviors and the instructional contexts most likely to facilitate children's language development. These findings will ultimately support the design of more effective future professional development and instruction in early childhood settings.

Project Activities: Researchers will code previously collected videos and transcribe and code audio files from authentic preschool classroom contexts, in connection with multiple directly assessed measures of children's language skills collected simultaneously. Via these preliminary analyses the team will establish a model representing the variability in the ways teachers verbally interact with children and how they and children spend time in the classroom. Researchers will investigate how these linguistic interactions and time-use patterns predict children's language development and how this is shaped by child and teacher characteristics such as dual language status, linguistic variations, race, and ethnicity.

Products: Anticipated products include the completed dataset that will ultimately be made available for other researchers to use in novel analyses. Primary research products will include conference presentations, peer-reviewed publications, social media and website postings and graphical displays of key findings will be tailored for dissemination to various audiences including researchers, educators, and early childhood policymakers.

Structured Abstract

Setting: Locations for this project include preschool classrooms from Florida, Georgia, and Alabama representing urban, rural, and suburban locales in 29 different counties.

Sample: The primary project sample includes 583 children ages three to five (M = 49.75 months), and their teachers, enrolled in 86 preschool and childcare classrooms across the three main sectors of private, Head Start and public prekindergarten settings. Sites were targeted that serve diverse populations and families from lower socioeconomic strata; in classrooms serving three-year-olds, four-year-olds, or a mix of both age groups. Children included White (43%), African American (45%) and other races/ethnicities (12%). Approximately 14% were identified as dual language learners (DLL). Teachers' education levels varied widely, and teachers represent varied races and ethnicities, and years of experience.

Factors: The primary factors of child language skill being explored will include three linguistic facets representing the quality of teacher-talk and teacher-child interaction contexts, and frequency and duration spent in focal instructional settings. Moderators will include children's initial language skill, DLL status, race, ethnicity, dialect use, linguistic match, age, and classroom composition.

Research Design and Methods: Researchers will carry out secondary data analyses and new coding of video and audio recordings from a study that was funded by the Spencer Foundation. The prior project primarily investigated teacher and classroom characteristics predicting language environments and child outcomes, focusing on teachers' knowledge and on diversity in ECE sites (e.g., across age groups, and site types). Researchers collected data at three waves during a school year on teacher practices and child language to compose one of the largest samples of naturalistic language data from both teachers and children. The project team will examine associations between children's language skill growth and three quality facets of teacher talk composed from coded data. The design takes advantage of a large archival dataset with three waves of each type of data. Extended classroom videos, audio-recordings of naturalistic teacher-, child- and dyadic language samples, and standardized assessments of children's language were collected. New video coding will target frequency of key instructional contexts (e.g., content instruction, book reading, and conversations). Substantially expanded and new transcription and coding of selected audio recordings will include linguistic features of both teacher and child language.

Control Condition: Due to the nature of this research there is no control condition.

Key Measures: For child outcomes, variables include standardized and naturalistic measures of semantics and syntax measured three times across the school year. At each wave of data collection children were assessed on five standardized language measures, including the Expressive One-Word Picture Vocabulary Test, the Receptive One-Word Picture Vocabulary Test, and three subtests of the Comprehensive Evaluation of Language Fundamentals-Preschool 2nd Edition (the receptive Sentence Structure and Concepts and Following Directions subtests and the Word Structure subtest). To assess teacher linguistic behavior, researchers will code teacher talk data to create variables representing teachers' language modeling and interactions with children in group and individual settings across the year.

Data Analytic Strategy: The research team will use multilevel structural equation modeling to identify the best characterization of teacher's linguistic behavior with respect to three theoretically-guided quality facets of quantity, diversity and complexity, responsiveness, and support, across contexts that may be more or less facilitative of language quality. The research team will conduct multi-level growth modeling analyses to investigate the unique, joint, and interactive associations of linguistic and contextual features with the growth of children's language, as represented by both standardized and natural language sample measures along with moderators representing child and classroom characteristics.