The University of St Andrews has a story about how SACHI researcher Per Ola Kristensson and his collaborator Keith Vertanen used crowdsourcing, Twitter, and other online sources to create better statistical language models for AAC devices. These devices enable users with communication difficulties to speak via a predictive keyboard interface. The new language models are by far the largest that have been built so far and they provide a 5-12% reduction in the average number of keystrokes that users have to type to communicate.
The research paper is open access and you can read it for free in the Association for Computational Linguistics’s digital library.
Reference:
Vertanen, K. and Kristensson, P.O. 2011. The imagination of crowds: conversational AAC language modeling using crowdsourcing and large data sources. In Proceedings of the ACL Conference on Empirical Methods in Natural Language Processing (EMNLP 2011). ACL: 700-711.