Combining Nonword and Sentence Repetition to Identify Developmental Language Disorder

Nonword repetition and sentence repetition have been useful in identification of developmental language disorder (DLD). Each task has demonstrated good sensitivity and specificity, though their diagnostic accuracy has varied across children's ages, specific measures, and languages. These tasks tap into different underlying weaknesses associated with DLD, but few studies have examined the diagnostic utility of combining them.

The present study compares the diagnostic accuracy of each task-nonword repetition and sentence repetition-to that of combining the two tasks in order to evaluate their utility in DLD identification.

Participants were 196 monolingual Vietnamese speakers, ages 4-6, living in the northern region of Vietnam. As part of a larger study, participants were divided into typical development (TD: n = 168) and DLD (n = 28). We used logistic regression with ROC curves to empirically derive cutpoints and calculate diagnostic accuracy measures for each task separately and in combination.

Nonword repetition and sentence repetition each showed adequate sensitivity (NWR: 89.3%; SR: 82.1%) but unacceptable specificity (NWR: 41.1%; SR: 76.2%). The combination of tasks showed adequate sensitivity (85.7%) and specificity (81.6%) as well as informative positive (4.65) and negative (0.18) likelihood ratios.

Nonword and sentence repetition may tap different underlying weaknesses and together provide useful diagnostic information to accurately identify DLD in Vietnamese monolingual children, ages 4 to 6. The next step is to translate research findings to practice by further validating the combined model and automate procedures for ease of scoring and interpretation.

What is already known on this subject Nonword repetition and sentence repetition have contributed to the accurate identification of Developmental Language Disorder (DLD). However, the diagnostic accuracy of each task varies across ages and languages, and their combined effect is understudied. Because each task measures different weaknesses associated with DLD, combining them could improve identification accuracy. What does this study add to existing knowledge This study examined whether combining nonword and sentence repetition improves DLD identification. Within a group of 196 4- to 6-year-old children in Vietnam, we found that the combination of tasks was better at identifying DLD than each task alone. What are the potential or actual clinical implications of this work? Combining nonword and sentence repetition holds promise for accurately and quickly identifying DLD. Next steps are to further validate this combination of tasks and to translate research findings into clinical practice.