guage input (Neville and Mills, 1997). In particular, in a group of Chinese-English bilinguals, delays as long as 16 years in exposure to English had very little effect on the organization of the brain systems important in lexical semantics. That is, the brain system underlying the organization of nouns and verbs was disrupted very little. However, delays of only 4 years had significant effects on aspects of brain organization linked to grammatical processing. Brain organization underlying function words, such as prepositions and conjunctions, was severely disrupted. Similar patterns have been found in studies of congenitally deaf individuals who learned English late and as a second language (American Sign Language was their first language). Deaf individuals displayed ERP responses to nouns and to semantically anomalous sentences that were indistinguishable from those of normally hearing individuals. However, the same deaf individuals displayed aberrant ERP responses to grammatical information. These findings suggest that the systems that mediate the processing of at least some types of grammatical information are much more modifiable by—and therefore vulnerable to—variations in language experience. This is demonstrated again below, in the discussion of interventions with children with specific language disorder.

In general, it seems important that practitioners consider the data generated from studies of the effects—and noneffects—of exceptional circumstances on language learning, for they provide important information on the boundary conditions of language learning. Moreover, these phenomena are the anchor points for theories of language development that take into account the resilience of language learning within more normal ranges of both environmental and organic variation.

As noted earlier, conventional language input is not essential for a young child to develop a language-like system and use it to communicate with others. However, a language model may play a central role in determining how often and when those linguistic properties are used. We noted above, for example, the infrequent use of language to express emotions among children who had been institutionalized. Another example concerns the ability to communicate about objects and events in other than the here and now. Deaf children who are not exposed to usable linguistic input (because their parents do not know American Sign Language, for example) not only use gesture to convey information about the here and now, but they also use it to converse about past, future, and hypothetical events (Morford and Goldin-Meadow, 1997). Linguistic input is thus not essential for a child to communicate about the nonpresent. However, the amount