In this video from TED Talks, he explains how this happened and talks about some of the results - you can hear his son's speech develop from 'guga' to 'wa'der' for water.
But, it seems that was not enough, because his team then took the tools they developed for this project and applied them to other patterns of interaction, especially examples from mass media communications.
His TEDTalks profile page also provides some useful inks to papers and labs where this work is developed.
Can we finally put the 'poverty of stimulus' argument to bed now thanks to tools such as these, or are we going to have to hear more on language acquisition devices?
Acknowledgment: Thanks to eldon for bringing this topic & video to my attention, Find more on her blog.