From Shared Attention to Shared Language in Human Infants

Homo sapiens possess a unique behavioural system for social action and response, namely, language. Language permits action at a distance by transmitting messages with specifi c meanings from one individual's mind to that of another. It is a peculiar system as compared with other structures in the environment, because the information in language that specifi es meaning is rather abstract and arbitrary. Despite—or perhaps due to—these characteristics, language is the prime medium for 'cultural ratcheting' (Tomasello, 1999) among humans. In cultural ratcheting, behavioural innovations (e.g. tools) spread through a group and are sustained and elaborated upon across generations. For a group to maintain a system of linguistic behaviours, each individual must be able to learn and adapt to the prevailing information structure. Typically, most of the structure of language is learned within a few years of birth, when the human infants are dependent on and in near-constant contact with caregivers. One account for this is genetic determinism: the structure of the human genome makes the acquisition and use of language inevitable. However, there is ample evidence, too complex to summarize here, that nativist views of language development (e.g. the Chomskian 'Language acquisition device,' poverty-of-the-stimulus claims, and mass-media reports of a so-called " language gene ") are either inadequate or blatantly incorrect (see, e.g. Elman et al., 1996; Pullum & Scholz, 2002). There is no doubt that some species-specifi c products and processes of the human genome are necessary for human language learning; however, these products are not suffi cient to explain early language development (MacWhinney, 1999). Most developmental scientists agree that a more complete account of human language must carefully consider infant's social experience. Somehow the structure of social information acquired by