PURPOSE: The purpose of this study was to assess the diagnostic accuracy of the Nonword Repetition Test (NRT; Dollaghan & Campbell, 1998) using a sample of 4- and 5-year-olds with and without specific language impairment (SLI) and to evaluate its feasibility for use in universal screening. METHOD: The NRT was administered to 29 children with SLI and 47 age-matched children with typical development. Diagnostic accuracy was computed using alternative scoring methods, which treated out-of-inventory phonemes either as errors or as unscorable. To estimate accuracy in a universal screening context, the probability of identifying a child at risk for language impairment was computed using the prevalence of SLI (7%) as the base rate. RESULTS: Diagnostic accuracy was acceptable using both scoring methods. The resulting likelihood ratios (LR+ = 22.66, 19.43; LR- = .05, .05) were similar to those reported for older children. The probability of accurate detection of children with SLI in the general population increased from 7% to 61%. However, this value suggests that many false positives could be expected. CONCLUSION: The NRT yielded results similar to those reported for older children. However, despite its strengths, the NRT is not sufficient for screening the general population of 4- and 5-year-olds.
PURPOSE: The purpose of this study was to assess the diagnostic accuracy of the Nonword Repetition Test (NRT; Dollaghan & Campbell, 1998) using a sample of 4- and 5-year-olds with and without specific language impairment (SLI) and to evaluate its feasibility for use in universal screening. METHOD: The NRT was administered to 29 children with SLI and 47 age-matched children with typical development. Diagnostic accuracy was computed using alternative scoring methods, which treated out-of-inventory phonemes either as errors or as unscorable. To estimate accuracy in a universal screening context, the probability of identifying a child at risk for language impairment was computed using the prevalence of SLI (7%) as the base rate. RESULTS: Diagnostic accuracy was acceptable using both scoring methods. The resulting likelihood ratios (LR+ = 22.66, 19.43; LR- = .05, .05) were similar to those reported for older children. The probability of accurate detection of children with SLI in the general population increased from 7% to 61%. However, this value suggests that many false positives could be expected. CONCLUSION: The NRT yielded results similar to those reported for older children. However, despite its strengths, the NRT is not sufficient for screening the general population of 4- and 5-year-olds.
Authors: Lisa D Gresch; Virginia A Marchman; Elizabeth C Loi; Anne Fernald; Heidi M Feldman Journal: J Speech Lang Hear Res Date: 2018-05-17 Impact factor: 2.297