The science of literacy: from the laboratory to the classroom.

C about public education, specifically in the area of teaching literacy, are growing. The Los Angeles Times recently reported that up to 60% of local children may fail their third-grade reading proficiency test. Millions of American students who are not learning disabled but enter school with weak English skills are not learning to read and are ending up in special education classes with lowered expectations and dim prospects. It is not surprising that language learning disabilities are emerging as one of the greatest social issues of our times. A recent briefing on learning disabilities for the Congressional Biomedical Research Caucus* included the following statistics: It costs public schools twice as much to provide special education services to a child. Despite these extra services, twice as many students with learning disabilities drop out of high school. This leads to lower employment rates and higher adjudication rates (85% of juvenile delinquents are learning disabled as are 60% of prison inmates). It is noteworthy that 75–80% of students classified as learning disabled have their basic deficits in oral and written language (1). Finding out ‘‘why Johnny can’t read,’’ in addition to being of increasing social concern, has become a focus of scientific research. Although initially the domain of educational research, more recently a growing interest in the neurobiological basis of higher cortical functions, especially language and reading (and the sensory, perceptual, and cognitive systems that subserve these functions), has captured the attention of neuroscientists. The majority of scientific studies of reading have focused on developmental reading deficits of unknown origin (dyslexia). Early research studies focused mostly on the visual (orthographic) components of reading. However, in this issue of PNAS Talcott et al. (2) point out that learning to read a language depends on acquiring an understanding of both its spoken properties (phonology) and its written form (orthography). In alphabetic languages such as English, printed characters (graphemes) correspond to phonemes, the smallest meaningful unit of sound that amalgamates to constitute a word. A large body of research now has demonstrated that proficiency in phonological analysis of words (i.e., decoding words into their phonetic segments), significantly differentiates dyslexics from controls (3, 4), as well as predicts future literacy skills (5). Although it is widely accepted that dyslexia is characterized by both orthographic and phonological deficits, the precise etiology of these deficits remains the focus of research. Other central research issues focus on whether (i) dyslexia constitutes a distinct disability or rather represents the disadvantageous end of individual differences within a normal distribution, (ii) phonological deficits related to reading failure are speech specific or derive from more basic auditory processing deficits, (iii) dyslexia is a deficit specific to written language or rather the manifestation of a more pervasive delay in language development, and (iv) dyslexia results from genetic, neurological, social, andyor educational causes. A considerable body of research on individuals with oral andyor written language learning impairments (LLI) has shown that, in addition to their weak language and literacy skills, they are less sensitive to dynamic (brief, rapidly changing) sensory stimuli, both in the auditory and visual modality (6, 7). Deficits in detecting rapidly presented or rapidly changing acoustic stimuli have been hypothesized to play a direct role in phonological development and disorders. Specifically, many of the acoustic temporospectral changes that are critical for identifying and discriminating phonemic segments within speech occur within tens of milliseconds, requiring a rate of acoustic processing that has been shown to be impaired in many individuals with LLI (8). However, the extent to which nonverbal acoustic processing rate constraints are related directly to subsequent oral andyor written language development and disorders has been hotly debated in the scientific literature, specifically by scientists who support a modular theory of language specificity in the brain (9, 10). The majority of data supporting language modularity, however, derive from studies with older children or adults, after they already have developed (or failed to develop) language learning skills normally. Thus, they may miss underlying developmental mechanisms that exerted significant effects on the calibration of sensory neural maps, as well as the learning progression for developing phonological categorization and representation. However, recent studies with infants born into families that already have one or more relatives with LLI offer a unique opportunity to observe the processes inf luencing language, prospectively, as it develops. These studies are consistent in showing that, within the first weeks of life, infants with a positive family history for LLI, process temporospectral changes within both nonverbal and verbal acoustic stimuli significantly more slowly than family history-negative infants (11, 12). Furthermore, an infant’s nonverbal auditory temporal processing threshold significantly predicts subsequent receptive and expressive language development, with infants demonstrating the slowest processing rates being most likely to be slower in language development (12, 13). As can be seen in Fig. 1, not only does nonverbal acoustic processing rate predict rate of language development in family historypositive infants, it is equally robust as a predictor of language development in infants with no risk factors.† Taken together, these studies demonstrate populationwide that individual differences along a dimension of dynamic acoustic processing sensitivity, specifically within the time range that is critical for speech perception, affects the course of language learning beginning in early infancy. In this issue of PNAS Talcott et al. (2) extend these findings to the domain of written language. They investigate the hypothesis that sensitivity to dynamic visual