Related Experiment Video
Updated: Oct 21, 2026
Measuring Statistical Learning Across Modalities and Domains in School-Aged Children Via an Online Platform and Neuroimaging Techniques
Published on: June 30, 2020
Sentence Completion Expressive Language Task-Preschool: Preliminary Evidence of Validity and Feasibility for the
Karla N Washington1,2,3,4,5, Leslie E Kokotek5,6, Mekibib Altaye7,8
1Department of Speech-Language Pathology, University of Toronto, Ontario, Canada.
Purpose:
This preliminary investigation occurred across two studies and examined validity, discriminant accuracy, and feasibility of the Sentence Completion Expressive Language Task-Preschool (SCELT-P) for characterizing language production in a neuroimaging environment for typically developing (TD) preschoolers and those with developmental language disorder (DLD).
Method:
Two separate studies were completed. Study 1 (validity) included 82 preschoolers (50% female; 69 TD, 13 DLD) aged 4;0 (years;months) to 5;11 (Mage = 4;7). All children were assessed using standardized language measures (e.g., Clinical Evaluation of Language Fundamentals Preschool-Second Edition), a play-based language sample (analyzed using Developmental Sentence Scoring), and the 30-item SCELT-P (outside the neuroimaging environment), later streamlined to 20 items (SCELT-P Revised [SCELT-PR]). Study 2 (feasibility) included 14 preschoolers (Mage = 4;9, 43% female; nine TD, five with DLD) that were assessed using the SCELT-P (inside the neuroimaging environment) and functional magnetic resonance imaging using a sparse acquisition approach with a "listen" (passive) and "listen + speak" (active) condition implemented using DirectRT software.
Results:
In Study 1, the SCELT-P and SCELT-PR demonstrated high reliability, including interrater (intraclass correlation coefficient = .99), test-retest (r = .86 and .80, respectively), and internal consistency (Cronbach's α > .86 and .83, respectively), and similar strong criterion validity (r = .53-.84). Area under the curve analyses (.91) revealed both versions also evidenced fair sensitivity (.88 and .87) and specificity (.85) rates with a cutoff score of 15.5 and 9.5, respectively. In Study 2, 89% of TD preschoolers and 60% of preschoolers with DLD successfully completed the SCELT-P/R inside the neuroimaging environment, with activation maps illustrating relatively similar neurobiological characteristics across preschoolers in some brain regions.
Conclusion:
The SCELT-P, including its streamlined version, the SCELT-PR, holds promise as a reliable and valid task for including 4- to 5-year-olds in the neuroimaging environment to inform neural and cognitive systems involved in language development.
Supplemental Material:
https://doi.org/10.23641/asha.34543209.
