Tallying Reference Errors in Narratives: Integrative Language Function, Impairment, and Fetal Alcohol Spectrum Disorders

Tallying Reference Errors in Narratives: Integrative Language Function, Impairment, and Fetal Alcohol Spectrum Disorders John C. Thorne Chair of the Supervisory Committee: Professor Truman E. Coggins Department of Speech and Hearing Sciences This study investigates the construct validity of a new measure of Integrative Language functioning, Tallying Reference Errors In Narratives (TREIN), by examining the association between previously existing CNS impairment and Expressive Language functioning and elevated outcomes on the TREIN measure “rate of Nominal Reference Errors” (rNRE). The rNRE is a measure of referential cohesion errors in noun phrases. Study participants included 155 elementary school aged children, 75 of whom had been identified with CNS impairments during a clinical assessment of suspected Fetal Alcohol Spectrum Disorders (FASD). Data came from existing clinical and research data including oral narratives elicited using a wordless picture book. Referential cohesion in the narratives was analyzed blind to participants’ previous assessment results, diagnoses, age or gender. Statistical analysis of group level performance in terms of means, correlations, and performance distribution were conducted to reveal any existing relationships between narrative performance and other clinical measures of CNS impairment and language functioning. Results support the validity of the rNRE as a measure of Integrative Language functioning by demonstrating that an elevated rNRE, 1) is associated with previously identified CNS impairment, 2) is more common in children with FASD than their typically developing peers, and 3) may be found in children whether or not impairments are apparent on clinical assessments of Expressive Language function. Exploration of the clinical utility of TREIN measures based on Nominal Reference Errors and those based on cohesive errors in pronoun phrases was also conducted and indicated that tallies of Nominal Reference Errors like rNRE have more clinical potential in this age range than measures of pronoun errors. A strong developmental trend in mastery of nominal reference seen in the typically developing participants was absent in the FASD participant group, indicating an increasing clinical utility for identifying impaired children in the upper elementary school years. Result support further development of the TREIN and point to a need to better understand its performance in populations of typically developing children and children with a variety of CNS impairments, including those associated with prenatal alcohol exposure. Table of