The brain acts as a self-organizing system and does not rely on extensive or
explicit guidance from the environment (Elman et al., 1996; Kelso, 1995).
Self-organized systems usually are in a state of delicate equilibrium deter-
mined in large part by preexisting conditions and to a lesser extent by the dy-
namics of their environment, which provide data through a feedback
mechanism (Smolin, 1997). One result for cognition and language is that even
meager input can have a significant influence (Elman et al., 1996). Although to
casual observers the linguistic input children receive may appear to be limited
and distorted, to a child’s developing brain this input is both rich and meaning-
ful. Adult language is absolutely necessary if children are to develop language,
but infants bring significant innate resources to the endeavor.
The self-organizing characteristics of the brain allow children to catego-
rize similar representations appropriately and cross-reference them in vari-
ous ways. Dogs and cats might exist in a category for pets, but they would be
cross-referenced not only with a category for four-legged animals but also
with words that begin with the letterdand words that begin with the letterc.
Cross-referencing here is not metaphorical: It consists of actual neural con-
nections that link related neurons. The result is a very complex neural net-
work of related items with all their associated features. Exactly how all these
items and features are sorted, stored, and cross-referenced remains a mystery,
but once a mental representation is established in the brain, the child is able to
process it at will. For example, the mental image of a dog eventually becomes
linked to all its associated features, both as a sound and as a graphic represen-
tation of the word—d-o-g.
A similar process seems to be at work with respect to grammar. Children use
their innate ability to organize the world around them to identify the patterns of reg-
ularity—the grammar—that appear in the language they hear during every waking
hour (Williams, 1993). Chomsky (1957, 1965) argued that this process is not pos-
sible because language has an infinite possible number of grammatical utterances
and that the human brain is incapable of remembering them all. He concluded,
therefore, that the brain must have some mechanism for generating the full range of
possible utterances on the basis of a relatively small number of generative rules.
There are at least two errors in this conclusion. First, as de Boysson-Bard-
ies (2001) pointed out, “the human brain contains 10^10 neurons”; ... [each]
“neuron forms about 1,000 connections and ... can receive 10,000 messages
at the same time”; “the number of junctions may be reckoned at 10^15 —more
than the number of stars in the universe” (p. 14). In other words, the human
brain has essentially unlimited storage and processing capacity. The sugges-
tion that the brain is incapable of memorizing innumerable grammatical pat-
terns seems a bit ridiculous in this light. The real cognitive challenge is not
COGNITIVE GRAMMAR 211