From Static Scaffolds to Adaptive Language Support: A Framework for Accessible Communication

Assistive language technologies increasingly offer multimodal supports for reading and communication, yet these supports are generally presented in predefined formats that do not account for differences in individual abilities or changing task demands. This paper outlines a framework for personalizing language assistance through the selection of linguistic and visual scaffolds. The proposed system combines a user profile with ongoing behavioral feedback to select among four presentation formats, ranging from unmodified text to sentence segmentation with pictograms and keyword labels. An optional neural component incorporates calibrated EEG features to investigate whether cognitive-state information can further improve adaptation. We propose a staged evaluation encompassing content validation, behavioral personalization, and the incremental contribution of neural measurements. By comparing fixed, behavior-adaptive, and neuroadaptive interfaces, the project aims to determine when additional support is beneficial and whether EEG provides meaningful advantages beyond observable user performance.