Practices and philosophies about beginning reading instruction vary, but research strongly supports an early emphasis on letter-sound correspondences especially for children at risk for reading failure (Anderson, Hiebert, Scott, & Wilkinson, 1985; Bond & Dykstra, 1967). Reading methods that include explicit, synthetic phonics instruction--isolated letter sounds and blending sounds into words--result in higher first-grade achievement in word recognition and spelling (Adams, 1990); and these effects spread in second grade to comprehension, reading rate, and vocabulary (Chall, 1967). Researchers have investigated individual aspects of phonics instruction-the format, language, and ordering of phonics activities (Carnine, 1976, 1981; Williams & Ackerman, 1971)-- and used these studies as a rationale for a theory of overall program design (Engelmann & Carnine, 1982).
One of the most dramatic demonstrations of the effects of a specific reading program occurred as part of the national evaluation of the federally sponsored Project Follow Through involving 20,000 disadvantaged children across the United States and 22 different models. One model, Direct Instruction (DI), employed Distar Reading (Engelmann & Bruner, 1988a). The Abt Associates report (Stebbins, St. Pierre, Proper, Anderson, & Cerva, 1977), in its analysis of Metropolitan Achievement Test (MAT) reading scores, concluded: "Only the children associated with the Direct Instruction Model appear to perform above the expectation determined by the progress of the non-Follow Through children"(p. 155).
Becket (1977) attributed the success of the DI model in Project Follow Through to the design features of Distar Reading, which "utilize(d) advanced programming strategies which are consistent with current behavior theory, but which go beyond current research on task analysis and stimulus control" (Becker & Carnine, 1980. p. 433). The design of DI programs is rounded on general case teaching, whereby children learn a small set of examples along with strategies for generalizing to a larger set.
One aim in the present study was to examine the contribution of program design to the efficacy of beginning reading programs used as early intervention for young children with learning disabilities. We reasoned that the effects of program design ought to be most apparent in studies employing students who are just beginning the reading process, particularly children who are predicted to fail without careful instruction; specifically, those children who may have documented learning disabilities, or who are among the wider category of children at risk for learning failure. These children have little prior instructional experience to confound the effects of program design.
Both reading programs examined in our research used a synthetic phonics approach, but differed markedly in instructional design (Carnine, Silbert, & Kameenui, 1990). We included one program, DI's Reading Mastery 1, because it is based on an explicit theory of instruction (Engelmann & Camine, 1982) and because its predecessor, Distar, produced remarkably strong achievement effects with economically disadvantaged youngsters. Our second program was Addison Wesley's Superkids. Like Reading Mastery, this program introduces letter sounds in isolation, teaches sound blending, and selects reading vocabulary words that have regular decodable spellings. However, Superkids adopts an entirely different stance on certain other aspects of program design that Adams (1990) has referred to as "unresolved dimensions" of phonics instruction (e.g., the order of letter-sound introduction and use of letter names), and that Gersten and Carnine (1986) have identified as critical elements in effective instruction (e.g., explicit step-by-step strategies, student mastery, specified error corrections, and formative testing coupled with cumulative …