Dynamic Assessment of Spanish-English Speaking Children for Developmental Language Disorder

Abstract

Background and Objectives: Assessment is the gateway to speech-language pathology (SLP) services. For Spanish-English bilingual children, there are few assessments with high classification accuracy, and those that exist may have barriers to use for monolingual English-speaking clinicians. One under-explored route is dynamic assessment, any assessment into which teaching is incorporated with a focus is on the child’s learning potential. The current project was a prospective feasibility study that updated a previously published dynamic assessment tool by Roseberry and Connell (1991) for the identification of Developmental Language Disorder (DLD), a lifelong difficulty with oral and written language whose origin is not attributable to another disorder or brain injury, in Spanish-English children, ages 4;0 – 5;11. The aims of this diagnostic accuracy study were to (1) create a replicable dynamic assessment, and (2) to understand which elements of scoring yielded the highest classification accuracy.Methods: Children were eligible if they were Spanish-English bilingual, ages 4;0 – 5;11, had no history of neurologic or other impairments that would exclude DLD. Data was collected from a convenience sample from October 2023 – April 2024. For the dynamic assessment, children were taught a novel morpheme (/-əb/, partial object), as morphology is a hallmark deficit in DLD, through a scripted storybook that incorporated statistical learning principles. Following teaching, the child’s modifiability was rated, and they were administered an immediate, one-hour, and one-day posttest. Children’s language status was determined by the Bilingual English-Spanish Oral Screener (BESOS; Peña et al., 2008, in development). Results: The assessment had high fidelity and scoring had high interrater reliability, demonstrating its replicability. To determine diagnostic accuracy, Receiver Operator Characteristic (ROC) curves with Bayesian logistic regression and Random Forest inputs determined the Area Under the Curve (AUC), which was used to compare scoring elements across 12 typically developing (TD) and 6 DLD children. Combining the modifiability rating scales and teaching phase score yielded the highest AUC (.83) with good sensitivity (83%) and specificity (100%). Discussion: This was an improvement on the original dynamic assessment by Roseberry and Connell (1991) and replications of it. Conclusions are not broadly generalizable due to small sample size and because there was low interrater reliability of the modifiability scales. Future investigations should focus on increasing reliability and investigating other parameters of the dynamic assessment to optimize utility and classification accuracy. After future study, this dynamic assessment may be used to identify DLD in Spanish-English bilingual preschoolers for entry to SLP services. Registration: Not registered. This abstract was formatted in adherence to the STARD for abstracts guidelines (STARD #1 -2; Cohen et al., 2017).