Download: Hendricks & Adlof (2017)
This study investigated the accuracy of using modified scoring procedures on standardized tests in accurately differentiating between typically developing and language impaired nonmainstream dialect speakers.
Background: Previous research demonstrates the challenge of identifying language impairment (LI) in children who speak non-mainstream American English (NMAE) dialects due a variety of factors. For example, many features that are considered indicators of LI in speakers of mainstream American English (MAE) are part of typical speech in NMAE dialects. This has led to over-identification of LI in NMAE speakers when using standardized assessments, illustrating the need for improved assessment procedures for NMAE speakers. Modified scoring procedures have been proposed as one solution to bias towards NMAE speakers on standardized assessments.
Hypothesis: The first objective of the current study was to compare the association between frequency of NMAE use and a measure of language ability according to two standardized assessments—the CELF-4 and the DELV-Screening Test, designed specifically for NMAE speakers. The researchers hypothesized that frequency of NMAE use would correlate with language score according to the CELF-4 and DELV-S. The second objective was to evaluate whether proposed modified scoring procedures for NMAE speakers produced improved diagnostic accuracy of the CELF-4 for those speakers.
Methodology and Participants: The study recruited participants from a larger project investigating language and reading disorders; therefore, the sample has a larger percentage of children with LI than the general population. Participants included monolingual second-graders (n=299) from a school district in South Carolina and spoke a range of dialects: African American English (AAE), mainstream American English (MAE), and Southern White English. Participants completed the DELV-S and the CELF-4. Scoring modifications were then applied to the CELF-4 results for AAE speakers only. To explore diagnostic accuracy of the CELF-4 scoring procedures (both modified and unmodified), the researchers considered children identified as “high risk” for LI according to the DELV-S as “true positives.”
Conclusion: After controlling for SES, researchers found a moderate positive correlation between use of NMAE and score on both the DELV-S and CELF-4, even after the application of modified scoring procedures for NMAE speakers on the CELF-4. Modified scoring procedures produced a slight increase in scores on the CELF-4 and in identification of true LI (LR+) in NMAE speakers. But it also produced significant decrease in accurate identification of typically developing NMAE speaking children (LR-). Modified scoring for AAE speakers on the CELF-4 did not result in accuracy acceptable according to identified standards.
Relevance to the field: This study provides further evidence for bias against NMAE speakers in standardized assessments as use is correlated with increased identification of LI. Although scoring modification procedures may serve to slightly reduce the number of NMAE speakers who are over-identified with LI, this also suggests that it may lead to significant under-identification. As a result, additional research is required before scoring modifications can be applied with confidence. This study further demonstrates that standardized assessments are not acceptable for accurately diagnosing LI in NMAE speaking children. Standardized assessments should be used for descriptive purposes only and other assessment procedures should be used including collecting a detailed case history and language sampling.
Hendricks, A.E. & Adlof, S.M. (2017). Language assessment with children who speak nonmainstream dialects: Examining the effects of scoring modifications in norm-referenced assessment. Language, Speech & Hearing Services in Schools. 48(3):168-182. doi: 10.1044/2017_LSHSS-16-0060.