Augmentative and Alternative Communication Generative Language Intervention: A Randomized Controlled Trial.

The trial was cut short by the COVID-19 pandemic. The plan was 60 children, 30 in each group; only 40 completed the baseline assessment, and the analysed groups ended at 29 and 19. No measurement in the study reached the conventional 80% power threshold, so the moderate-to-large effects rest on a small sample and the authors' own table shows they were underpowered. After the pandemic, eight children who were not assigned to intervention spent four months in the control condition and then received intervention. They had more time with the device before starting than the other children, and they are counted in both groups in the growth models. The authors report their gains were similar to those of the immediate intervention group. Five children were dismissed when the pandemic shut the study sites in March 2020 and a further seven did not complete the study for other reasons, including two whose speech improved enough that aided communication was no longer needed. The maintenance comparison rests on only 6 control children, so that time point is very fragile. The main outcome is a new, study-specific measure of aided grammar, and agreement between coders on its word-order component was only moderate-to-good, although agreement on the total score and on the number of relevant symbols was excellent. Intervention and measurement sessions were sometimes run by the same person, the authors note, and 30 different examiners delivered sessions across the study. Sessions took place in university clinics and labs with researcher-built materials, and the authors state that interventions like this still need field testing in everyday clinical and educational settings.

Funded by the National Institutes of Health (National Institute on Deafness and Other Communication Disorders, grant R01DC016321), awarded to Cathy Binger and Jennifer Kent-Walsh as principal investigators. The intervention is based on the research team's own published developmental model, and the authors write their own conclusions. No commercial funding or industry involvement is reported.

Do not read the moderate-to-large effect sizes as settled proof that this intervention works. The study was small and underpowered, with no measurement reaching 80% power, and the group difference was significant only at Months 1, 3 and 4 on one-sided tests. It was not significant at Month 2 or at the three-month maintenance check.