Many lines of evidence support the concept that the left hemisphere plays an essential and specialized role in language processing in adults. Studies of adults with focal brain injury find that approximately 95% of cases of aphasia are associated with left hemisphere damage (Goodglass, 1993). In the traditional view, damage to the third frontal convolution, Broca's area, is associated with problems in language production, whereas damage to the first temporal convolution, Wernicke's area, is associated with problems in language comprehension. However, the picture is more complex. The ability to predict the location of injury from the aphasic syndrome (or vice versa) is limited, and the right hemisphere remains involved in aspects of language processing, such as interpretation of prosody and metaphor and comprehension of complex syntax (Just et al., 1996). Nonetheless, the left hemisphere contribution to language seems to be necessary.

How, when, and why does the left hemisphere become specialized for language functions? The study of children who sustain focal left hemisphere damage prior to or during language development provides one experimental approach to address these issues. Rarely, children who have not yet learned to speak or who are still developing language skills sustain brain injury to areas of the left hemisphere that typically serve language function in adults. These children afford a naturalistic experimental opportunity to address the theoretical questions about the neural substrate of language learning.

If children with left hemisphere damage prior to language learning subsequently demonstrate serious delays in language development, the implication is that the mechanisms for language development reside within the damaged regions of the left hemisphere, a position called early specialization. Such findings would suggest that the neural architecture for language is determined by innate and probably genetic mechanisms. If, by contrast, children with left hemisphere damage successfully master language skills, the implication is that, at least under extreme circumstances, alternative organizations can be established. Such findings would suggest that the neural architecture of language is an outcome of language learning. In its strongest formulation, this second position asserts equipotentiality, that is, that either hemisphere can serve language functions as long as the neural commitment occurs before language learning forces left hemisphere specialization (Lenneberg, 1967). If children with left hemisphere damage show only minor delays, the implication is that alternative neural organizations are less favorable to language development or processing than the classical language areas (Satz, Strauss, and Whitaker, 1990), an intermediate position called constrained plasticity or ontogenetic specialization. This last position would suggest that some aspects of brain structure may be determined by early and possibly genetic factors, but that the full development of left hemisphere specialization emerges through development.

Challenges in the study of children with early focal brain injury complicate obtaining and interpreting findings. Before modern neural imaging modalities became available, the localization of damage was often uncertain; this review considers empirical studies that used computed tomography or magnetic resonance imaging for lesion location. Because early focal brain injuries are rare, published studies usually consist of small, heterogeneous samples. Age at onset, extent of lesion, time since injury, and associated problems vary within and across these groups, making meta-analyses impossible. The language capabilities of the infants, had they not sustained injury, remain uncertain. Other variables, including presence of seizure disorder or use of anticonvulsant medications, mediate outcomes (Vargha-Khadem et al., 1992). Given all the sources of potential variability, the use of group means to describe brain-injured populations may combine disparate groups and mask important distinctions. A preferable approach is individual profiling in the setting of a contrastive group of age-and developmentally matched children (Bishop, 1983).

An initial question to be addressed in considering the language development of children with focal lesions is their overall intelligence and cognitive profiles. If these children have severe intellectual impairments, then any language deficits should be evaluated in relation to cognitive measures such as intelligence quotient (IQ) tests. Most studies concur that children with focal injury to either hemisphere, even those with total surgical removal of a hemisphere, score at or near the population mean (Bates, Vicari, and Trauner, 1999). In children, unlike in adults, differential hemispheric mediation of verbal and performance or nonverbal functions is not typically found (Vargha-Khadem et al., 1992; Vargha-Khadem, Isaacs, and Muter, 1994; Bates, Vicari, and Trauner, 1999). The near-normal intellectual performance of children with focal damage is a testament to the plasticity of the human brain.