Academic impacts of language delay on oral and written discourse

Univ Of Maryland, College Park, College Park MD

Investigators

Abstract

In response to PAR‐24‐045, we will “leverage existing data and create open and shared data resources to aid in identifying patterns and predictors of developmental outcomes in late talking children, and exploring potential underlying (its) mechanisms, risk factors, and sequelae” by using archival resources and TalkBank computational software to track written language delay as students transition through school age. Specifically, we will track linguistic microstructure development in written language samples from students from grades 6 to 11, with and without DLD. Roughly 20‐30% of late talking (LT) children will continue to lag behind typically developing (TD) children throughout development and be later labeled as Developmentally Language Disordered (DLD). Their deficits become apparent in writing across all academic genres but little data exist to aid us in developing specific syntactic/grammatical intervention targets in the written mode. Prior research has addressed limited aspects of language microstructure, such as sample length or lexical diversity, that provide little guidance on how to develop individualized, focused interventions. Aim 1 will track writing skills in a carefully stratified, large corpus of persuasive essays (PERSUADE; Crossley, et al., 2022) from the later elementary through secondary school years, and focus on development of written language proficiency in native English language‐learning students with DLD and their TD peers. PERSUADE is an immense open‐ access collection (~25,000 samples) of student writing annotated for demographic features such as sex, race, home language, SES, and disability status. We will further annotate this corpus to identify syntactic, lexical and spelling features of the samples on an utterance‐by‐utterance basis. We will identify specific, actionable microstructure profiles that distinguish typical student writing performance from that of students who continue to be identified as language‐ learning disordered. Results will further inform the extremely sparse literature on either typical or disordered written language trajectories over the school years from grades 6‐11. Aim 2 will produce a free, open‐access software deliverable, a “writing report card” (AcadEval) available for use by the wider research, clinical and teaching community for evaluation and intervention planning purposes. AcadEval will generate individualized profiles of student strengths as well as intervention needs (benchmarked to our statistical analyses) that can be used to augment current, holistic rubrics and other performance data for benchmarking student writing progress. We will also produce a newly annotated corpus for data‐sharing by other researchers and new open‐access computational resources (software) for research, educational and clinical purposes. We will enhance diverse perspectives in both research focus and application: We continue ongoing work to distinguish English language variation (e.g., use of African‐American English and other variants) from disorder using PERSUADE and its unique demographic features. Finally, our lab recruits highly diverse students for whom our research approach provides flexibility to engage in research that removes many traditional barriers to participation.