What You Need is What You Get: Adapting Word Prediction of Augmentative and Alternative Communication Aids to Youth Language

Abstract

This paper focuses on the development of prediction models for Augmentative and Alternative Communication (AAC) that are tailored specifically to child and young users. A new native-speaking corpus for child and adolescent languages is introduced, which allows the training of models that better fit the needs of this user category. The experiments conducted on real texts show that adapting the prediction models for children and young users leads to an improvement in keystroke savings, which in turn reduces the number of predicted words to be displayed. This opens up new possibilities for rethinking the virtual keyboard organization and interaction styles, with the aim of reducing cognitive load during text entry.