Reliability and Validity of Dynamic and Processing-based Assessments for Language in Diverse Bilingual School-age Children

点击翻译按钮获取中文摘要

Project Summary/Abstract Children with developmental language disorder (DLD) demonstrate deficits in language compared to unaffected peers from similar language-learning environments. Conventional assessment methods for DLD capture a child’s current skill level in one language; in the U.S., such assessments are available only in English or Spanish. These methods fail sequential bilingual children, who speak a minority language at home and are subsequently exposed to English, because they confound language-learning ability with prior language- learning experience. Assessments that index language-learning ability are sorely needed, especially for sequential bilingual children whose home language is not Spanish. This project evaluates two novel approaches to language assessment for sequential bilingual children: dynamic assessment and processing-based assessment. Dynamic assessment methods offer language teaching during the assessment and evaluate the child’s response. Processing-based methods assess underlying skills that may drive language learning. This project rigorously examines the psychometric properties of dynamic and processing-based assessments within a diverse group of sequential bilingual children entering English-based schooling. We will recruit 165 children aged 4 years, 10 months to 6 years, 2 months who speak any non-English language at home. The first study timepoint will occur as children enter the school year and include an experimental assessment battery with dynamic assessments at the narrative and morpheme levels and both linguistic and non-linguistic processing- based assessments. The assessment battery will be repeated 1 month later. Aim 1 establishes psychometric properties of the experimental assessment battery, including internal consistency and test-retest reliability, factor structure, and concurrent validity. Aims 2 and 3 then use a longitudinal design to examine predictive validity. Children will be followed over two academic years, with re-assessment every 4 months (totaling 5 growth timepoints following the initial assessment and test-retest timepoint). At each growth timepoint, children will complete an English language sample and parents and teachers will be interviewed to assess the language environment and presence of concerns regarding the child’s language. We will construct models of English growth that account for child-external factors (e.g., language environment, SES) and then determine how well the experimental dynamic and processing-based assessment battery predicts this growth (Aim 2). We will then identify a group of children at high risk of DLD based on the combination of poor English growth and parent or teacher concern regarding language development. Aim 3 will examine the sensitivity and specificity of the initial assessment battery for predicting DLD. Project results will demonstrate whether multiple dynamic assessment and processing-based assessment tools can reliably and validly measure language-learning ability across variable language-learning environments. For millions of children who speak home languages other than English, this project represents a critical step in validating language assessment tools.