What are the historical roots of speech recognition? Why is spoken language difficult for a machine to decode, from phonetics to grammar and syntax? Why do we still call tech support to speak to humans, and will computers ever be able to replace them?