Language and communication are central to children?s healthy development and well-being throughout the lifespan; thus, early identification of a language disorder is crucial to ensure the best possible outcomes for young children. However, accurate identification of language disorders in African American (AA) children from low-income households or those who speak non-mainstream American English (NMAE) has been quite challenging because the linguistic characteristics of language deprivation (e.g., smaller lexicon or vocabulary or poor grammatical development) and sociocultural dialectal features often overlap with patterns in language disorders. Moreover, low income AA children often do not fare well on standardized language assessments, but there is a paucity of empirical psychometric information for scores obtained from this population. It is unclear whether their low performances reflect their true ability, mismatch between their native language variety and language forms presented on the assessment, or measurement artifacts presented in the assessment including lack of reliability and validity or item bias. This project will evaluate the psychometric properties of a widely used standardized language assessment, the Clinical Evaluation of Language Fundamentals- Preschool 2nd Edition (CELF-P2). Specifically, the project will explore the internal structure of the CELF-P2?s Core Language subtests? item scores for impoverished children and will investigate potential factors (i.e., item features and child characteristics) that may affect item functioning.
The Specific Aims of this projects are:
Specific Aim1 : to test the hypothesized unidimensional measurement model of three Core Language Subtests of the CELF-P2, Sentence Structure, Expressive Vocabulary, and Word Structure;
Specific Aim2 : to investigate factors such as item features and child characteristics that might affect item functioning on the Core Language Subtests. The main focus of the Specific Aim1 will be to evaluate whether the inter-item structure of each of the three subtests parallel the scoring structure of the assessment (i.e., one-factor or unidimensional) or whether alternative measurement model(s) better explains the inter-item associations (e.g., multiple factors such as multidimensional or bi-factor model). The main focus of the Specific Aim2 will be to evaluate item parameters (difficulties and discriminations) change as a function of item features (e.g., item contents that overlap with AA NMAE dialect) and child characteristics (e.g., dialect density, risk of language disorder, or aggressive behavior) via explanatory item response theory model.
Language assessments that produce reliable and valid score for all populations, including socioeconomically, culturally, and linguistically diverse populations, is crucial to accurate identification of language disorders and for ensuring the best possible outcomes for young children. This project will evaluate psychometric properties of individual item scores on a widely used standardized language assessment for low-income African American children and those who speak non-mainstream American English. Furthermore, the project will investigate item features and person characteristics that may influence variations in item functioning.
