In the beginning: .

We review Berwick and Chomsky's a book premised on language as an instrument primarily of thought, only secondarily of communication. The authors conclude that a Universal Grammar can be reduced to three biologically isolated components, whose computational system for syntax was the result of a single mutation that occurred about 80,000 years ago. We question that argument because it ignores the origin of words, even though Berwick and Chomsky acknowledge that words evolved before grammar. It also fails to explain what evolutionary problem language uniquely solved (Wallace's question). To answer that question, we review recent discoveries about the ontogeny and phylogeny of words. Ontogenetically, two modes of nonverbal relation between infant and mother begin at or within 6 months of birth that are crucial antecedents of the infant's first words: intersubjectivity and joint attention. Intersubjectivity refers to rhythmic shared affect between infant and caretaker(s) that develop during the first 6 months. When the infant begins to crawl, they begin to attend jointly to environmental objects. Phylogenetically, Hrdy and Bickerton describe aspects of ' ecology and cognition that facilitated the evolution of words. Hrdy shows how cooperative breeding established trust between infant and caretakers, laying the groundwork for a community of mutual trust among adults. Bickerton shows how 'confrontational scavenging' led to displaced reference, whereby an individual communicated the nature of a dead animal and its location to members of the group that could not see it. Thus, both phylogenetically and ontogenetically, the original function of language was primarily an instrument of communication. Rejecting Berwick and Chomsky's answer to Wallace's question that syntax afforded better planning and inference, we endorse Bickerton's view that language enabled speakers to refer to objects not immediately present. Thus arose context-free mental representations, unique to human language and thought.