For a colleague working on speech recognition who is unfamiliar with the connectionist work in this area: I would be grateful to receive pointers to the current and representative work. Stevan Harnad (harnad at mind.princeton.edu)