Response to Allington and Woodside-Jiron

Patricia G. Mathes
University of Texas – Houston Health Science Center

Joseph K. Torgesen
Florida State University

(in press, Educational Researcher)

Abstract

The program of research on reading sponsored by the National Institute of Child Health and Human Development (NICHD) is one important source of information for practitioners choosing best practices for early literacy instruction. On the basis of evidence from both normal and disabled readers, this research indicates that multiple factors are critical to early literacy success and that phonological knowledge and skill is one such element. Recently, the utility of this body of research has been called into question by Allington and Woodside-Jiron in an article titled, “The Politics of Literacy Teaching: How ‘Research’ Shaped Educational Policy." This response to that article is written to examine policy making activities that are misrepresented, to correct mistakes made by Allington and Woodside-Jiron about the range and utility of NICHD research, to correct possible misperceptions of the NICHD-supported research effort that might arise in the minds of those who read the Allington and Woodside-Jiron article, and to present new evidence that clarifies the utility of much of this research.

The Response

According to the recent report from the National Research Council titled “Preventing Reading Difficulties in Young Children”(Snow, Burns, & Griffin, 1998), excellent early general education reading instruction is critical to the prevention of the majority of reading difficulties. This report clearly states that unless effective instructional practices are identified and broadly applied in practice, the instructional inequities that commonly divide our children into the elite high performers and rejected low performers is certain to continue. The program of research on reading sponsored by the National Institute of Child Health and Human Development (NICHD) is clearly one important source of information that can assist practitioners in choosing effective practices.

This research and other research funded by other sources should be used to guide practice in much the same way as research supported by other branches of the National Institutes of Health are used to guide fields such as medicine and pharmacology. For example, recent NICHD research findings indicating the importance of infant sleeping positions in reducing the incidence of sudden infant death syndrome (SIDS) has been enthusiastically embraced by the medical community and dissemination efforts are well underway. Unfortunately, the findings from NICHD-supported research that are relevant to the prevention of reading failure have not been similarly embraced by the educational community. Instead, we find serious efforts to discredit and misrepresent these findings (Coles, 1998, Goodman, 1998; Taylor, 1998).

The utility of this research for guiding decisions about instruction was recently called into question by Allington and Woodside-Jiron in the recent article appearing in the Educational Researcher titled, “The Politics of Literacy Teaching: How ‘Research’ Shaped Educational Policy.” Our purpose in writing this response is to examine policy making activities that are misrepresented, correct mistakes made by Allington and Woodside-Jiron about the range and utility of NICHD research, and present new evidence that clarifies the utility of much of this research. For purposes of clarity, and in order to avoid too much repetition of names, we will refer to Allington and Woodside-Jiron throughout the paper as AWJ. Likewise, we refer to NICHD funded reading research as NICHD-RR.

In their article, AWJ state that they “are particularly interested in discerning how advocates advance a specific policy agenda and the role ‘research’ plays in the advocacy” (p. 4).

However, they never actually follow through with their stated purpose. Instead, they focus on one paper by Bonita Grossen titled, “Thirty Years of Research: What We Now Know About How Children Learn to Read” that appeared on the Center for the Future of Teaching’s Webpage. They treat this paper as if it were the primary catalyst of reform within several states (i.e., California and Texas). They then purport to examine the support available for the conclusions about reading instruction made by Grossen. However, this examination takes the form of discounting the utility of the entire 30 years of research on reading, reading development, and reading instruction that has been financially supported by the National Institute of Child Health and Human Development (NICHD).

One of our purposes in writing this response to AWJ is to examine the weight the Grossen paper has had in shaping educational policy and to put into perspective the interpretations and misinterpretations of NICHD-supported research made by Grossen (1997). It is important to understand that the NICHD reading research program and the Grossen article are not one in the same. Without this awareness, it is very easy to accept AWJ’s criticisms of the Grossen article as applicable to the entire 30-year NICHD research base. In our response, we deal separately with issues related to NICHD-sponsored research and Grossen’s interpretation of this research.

A core statement made by AWJ is that the NICHD recommendation that early reading instruction include explicit, systematic teaching of the alphabetic principle, including reading practice using decodable text, is not based on research relevant to general education. At the root of their argument is a simple discounting of the findings of NICHD reading research as unimportant. They accomplish their agenda by:

1. Misrepresenting the scope of NICHD reading research as focusing only on children with reading disabilities.

2.Disowning general education’s responsibility to children who experience reading difficulties.

3. Confusing the issue of credibility and utility of sound research with political advocacy.

4. Discounting and misrepresenting the importance of many outcomes of NICHD intervention research.

5. Over emphasizing Grossen’s (1997) interpretation of the large and complex NICHD research base as simply incorrect and as having a large influence on state reading policies.

We will examine each of these arguments in turn, finding their conclusions largely inaccurate. We hope this additional information and discussion will be helpful to all who are sincerely interested in advancing our understanding of factors and instructional practices that may be critical in helping all children become good readers.

We have attempted to write this response in a tone that will foster rational discussion rather than heated and/or personal argument. We believe that both AWJ and ourselves are sincerely interested in improving the instructional experiences and reading growth of children for whom reading is difficult. We are also equally committed to the methods of science as one means of achieving our aims in this area, and that commitment demands that we be as fair as possible in examining all the evidence relevant to the issues at hand. At present, we have decidedly different points of view on some of these issues; it is our hope that respectful exchange of views will help us all to move to a more accurate and useful synthesis of the knowledge than is presently available.

It should also be noted that this response to AWJ is based on the most recent 1999 variant of the manuscript that appeared in Educational Researcher. We must admit that we find it amusing that AWJ discuss the difficulty they had in writing their critique of the Grossen paper because variants of that paper were found on the CTFL WebPages and in published form in Effective School Practices. We found this complaint amusing because we too experienced difficulty in writing our critique because we too found 3 versions of the AWJ paper that were similar but not identical. Versions of the AWJ article include “Adequacy of a Program of Research and of a ‘Research Synthesis’ in Shaping Educational Policy,” which was made available on the National Research Center on English Learning and Achievement’s WebPage (Allington & Woodside-Jiron, 1997). A second, shorter, but very similar article was published in chapter form (Allington & Woodside-Jiron, 1998), and of course, the most recent reiteration published in Educational Researcher (Allington & Woodside-Jiron, 1999).

Background and Context

Before examining AWJ’s arguments, we first frame the context for this discussion. Unfortunately, we live in an era where the former “great debate” is now characterized as the “reading wars” (Goodman, 1998). Fueling the flames of this “war” is converging empirical evidence on one side (see the recent report from National Research Council edited by Snow et al., 1998) and heated philosophical rhetoric on the other (see Coles, 1998; Goodman, 1998; Taylor, 1998). The unfortunate casualties of this war are the large number of children who experience significant problems becoming competent readers.

In the midst of this debate, there have been repeated calls to reform reading instruction. The result is that typical reading instruction that occurs in today’s primary grade classrooms is qualitatively different than it was even a decade ago (Allington, 1994; Vaughn, Moody, & Schumm, 1998). Today we have an educational community in which many elementary grade and special education teachers have traded in basal reading series for "literature anthologies" that still look much like basal reading series, where ability group instruction has been replaced with whole class undifferentiated instruction, where specific skills are only "taught as needed," and where many teachers believe that "reading should be caught not taught" and that teaching phonics is harmful to young children (Mathes, 1998; Vaughn et al., 1998). Unfortunately, little has changed in terms of academic outcomes over the last 20 years (Snow et al., 1998) and the gaps between children of middle class families and children of poor families have remained fairly constant for the past 16 years (National Center of Educational Statistics, 1996). Presently, over 40% of 4th and 8th graders fail to read at levels considered essential to performing grade level work (National Center of Education Statistics, 1996), and very poor readers consistently make little to no reading gains from year to year (Klingner, Vaughn, Schumm, Hughes, & Elbaum, 1998; Schumm, Vaughn, Elbaum, 1996; Vaughn et al., 1998; Zigmond, et al, 1995). Thus, it appears that we are educating only some, not all, our children to become good readers (Snow et al., 1998).

What Do NICHD Supported Scientists who Study Reading Actually Advocate?

Converging evidence from research at 41 sites nationwide supported by NICHD as well as research supported by other sources such as the U.S. Office of Education has lead reading researchers supported by NICHD to repeatedly conclude that teaching the alphabetic principle is one critical component for preventing or remediating reading failure for many children.

What NICHD supported reading researchers do not advocate is code-only instruction, a focus on skill worksheets, skills taught in a disjointed manner, or skills taught only in isolation. Likewise, NICHD reading researchers do not suggest that beginning readers should be deprived of authentic literature and read only decodable text. What they, as a group, do advocate for is balance (see Fletcher and Lyon, 1998 for a more detailed summary)

Lyon summarizes the NICHD research base in the following manner.

There is evidence suggesting that a constellation of components configured in an instructional reading program that if used properly for a sufficient period of time will contribute to the reading acquisition of many children. At present, these components include instruction in phonemic awareness, the alphabetic principle, the development of speed and automaticity in word and text reading, and the development of reading comprehension strategies within a literature rich environment. The relative importance of using decodable text and predictable text formats for readers who differ on each of these components is not yet well understood and NICHD research is now underway to explicate this issue. It is conceivable that a combination of both types of text formats will be necessary to foster adequate development at different states of reading.

What is inconceivable is that the issue of decodable text stirs such emotion given the tremendous amount of evidence from research on transfer and generalization of skills, in general. The data have shown and continue to show that generalization of basic skills to other contexts and the development of speed and automaticity in applying skills across contexts and formats is critically dependent upon opportunities that provide ample representation of the concept being acquired. This is particularly true of those who have difficulty acquiring a skill. (Lyon, August, 1997, personal communication).

Much Ado About Nothing

We are perplexed why AWJ have chosen to focus attention on the Grossen, 1997 article. From the AWJ portrayal of Grossen’s 30-year paper, one would think that this one paper was largely responsible for shaping the entire early literacy reform efforts within California and Texas. The truth is much different. In our examination of policy efforts, we have determined that in both California and Texas the Grossen article had little or no impact on shaping educational policy.

The reality is that in California, the Grossen 30 years paper postdated passage of key California early reading reform legislation. In Texas, when the Grossen paper first appeared, leaders within the Texas Educational Agency (TEA) recognized that Grossen’s interpretation of NICHD research did not adequately portray the large and complex NICHD research base and the document was not used as a guiding document for the state’s reading initiative. We know these things because we examined the timelines for policy activities within each state, interviewed policy leaders who were involved in reading policy making within each state, and reviewed state produced documents; actions we are surprised that AWJ did not undertake. Thus, it appears that AWJ miscalculated the effect of the Grossen paper by assuming that wide distribution of the paper via the Internet equated to large impact on policy without checking the validity of their assumption. As a result, AWJ’s argument about how “research” was misused in California and Texas is fundamentally flawed. Furthermore, the Grossen paper, to our knowledge, has not been credited by any group, other than AWJ, as being a consensus document or a true synthesis of research. We would rather reserve this designation for scholarly reports that merit it such as the recent NRC report (Snow et al, 1998).

Reading Reform in Texas

In Texas, the roots of advocacy for the inclusion of systematic, explicit instruction in the alphabetic principle as one component of a balanced reading program began well over a decade ago, when the Nation At-Risk report was released (National Commission on Excellence in Education, 1983). At that time, the first author of this paper was actually a teacher of reading in that state, and remembers seeing copies of the report passed around and discussed in a variety of venues, including state conferences, faculty meetings, and even IEP meetings. By 1988, Texas implemented its dyslexia reading legislation for all children who were falling severely behind in reading. Services provided under this legislation are not part of special education or Title 1 funds, but rather are funded by the state. The cornerstone of these services is code-emphasis instruction in reading for children who were falling behind their classmates in reading growth. The state’s attention has only more recently focused on general education classrooms, following recognition that general education reading instruction was failing too many children.While not legislated as in California, the current reform effort began in Texas in 1996 when Governor Bush allocated funds to the Texas Education Agency (TEA) for the state’s current Texas Reading Initiative. As a first order of business, TEA developed the reading initiative’s guiding document, Beginning Reading Instruction: Components and features of a Research-Based Reading Program (Texas Education Agency, 1996). What this report recommends is balanced reading instruction. Specifically, it states that children should have opportunities to: (a) understand and manipulate the building blocks of spoken language, (b) learn about and manipulate the building blocks of written language, (c) learn the relationship between the sounds of spoken language and the letters of written language, in addition to many other elements such as opportunities to write and relate their wiring to spelling and reading and opportunities to read and comprehends a wide assortment of books and other texts. The research and reports cited for this document reveal that (a) Grossen’s work was not utilized, and that (b) the work of many researchers were included – some with funding from NICHD, most with funding from other sources.

The next task of the Texas Reading Initiative was to develop new standards for teaching reading that were consistent with the Beginning Reading Instruction document by replacing the former Texas Essential Elements with the Texas Essential Knowledge and Skills (TEKS). TEKS focuses on providing balanced reading instruction and were developed without referring to the Grossen paper. Since that time, the reading initiative has been focused on staff development efforts.

Reading reform in California

California’s movement toward advocating for a more balanced reading curriculum is more recent than in Texas. In the 1980’s, holistic, literature-based practices were prescribed in the state’s Language Arts Framework, and explicit code-instruction banned. This framework was reaffirmed in 1987 leading publishers to produce the now common “literature based” reading series. By 1992, when NAEP scores were released, California’s reading scores were low. By 1994, NAEP scores were even lower. Over 10 years, the state had had plummeted from near the top to near the bottom in national comparisons (National Center for Educational Statistics, 1996). Furthermore, the drop in reading scores could not be accounted for by the state's changing demographics. Arguably, it was the 1994 NAEP scores more than anything else that lead to California’s reading reform movement.

In 1995, Delaine Easton, a democrat, became California’s State Superintendent of Public Schools. Under Easton’s leadership, the state put together a reading task force. In September of 1995, 2 years before the Grossen paper, the task force released a guiding document for the state’s reading reform. From the California Reading Task Force report, “it was determined that a balanced and comprehensive approach to reading must have: (1) a strong literature, language, and comprehension program that includes a balance of oral and written language; (2) an organized, explicit skills program that includes phonemic awareness (sounds in words), phonics and decoding skills to address the needs of the emergent reader; (3) ongoing diagnosis that informs teaching and assessment that ensures accountability; and (4) a powerful early intervention program that provides tutoring for children at risk for reading failure” (California Department of Education, 1995, p.2). In that same year, the legislature became involved, passing Assembly Bill 1504 (ABC Bill) mandating explicit teaching of phonics, spelling, and math. The legislature then passed Assembly Bill 170 (Baldwin Bill) reiterating the call to teach explicit, systematic phonics as one part of the early reading curriculum. Again, this legislation occurred before the Grossen paper had been written.

In May 1996, again prior to the Grossen paper, the California Department of Education released Teaching Reading: A Balanced, Comprehensive Approach to Teaching Reading in Prekindergartern though Grade Three (California Department of Education, 1996). This program advisory endorsed vigorous early reading instruction including phonemic awareness and explicit systematic phonics as well as instruction and support in vocabulary, reading comprehension, and regular broad, thoughtful reading. Once released, it was quickly adopted by the State Board of Education and the Commission of Teacher Credentialing.

In July of 1996, Assembly Bill 3482 provided money for staff development for teachers and administrators in effective early reading, which included, among other things, staff development on teaching systematic, explicit instruction in the alphabetic principle. In all, the California legislature passed eleven pieces of legislation during 1995-1996, all of which were aimed at the same content (the incorporation of phonemic awareness and alphabetic knowledge instruction into the curriculum). All were passed unanimously by both republicans and democrats and all occurred prior to or simultaneously with the release of the Grossen paper.

Influence of NICHD-RR

Beyond the 30 years paper, NICHD supported Reading Research (NICHD-RR) is portrayed by AWJ as a “policy lever” (p. 5) that is used by those pushing a “code-emphasis” agenda. Reid Lyon and NICHD supported researchers are presented as political advocates of this agenda because they provided expert testimony to various policy-making bodies in California, Texas, and elsewhere. In our view, AWJ greatly overemphasize the role of NICHD supported researchers in stimulating the advocacy for reading reform in California and Texas. While it is accurate to say NICHD supported research influenced decision-makers, as it should, it is inaccurate to present this research program as the primary influence. For instance, in California, NICHD staff and researchers were brought into the state rather late in the reform process. Likewise, numerous individuals provided testimony to that’s state’s legislative body. In fact, Allington himself was one of the first reading experts brought in by the state to help shape policy. In reality, the majority of individuals who provided expert assistance to California’s policy making bodies were in no way affiliated with NICHD and included Marilyn Adams, Richard Allington, Isabel Beck, Robert Calfee, David Chard, Jane Fell Greene, Connie Juel, Adria Klein, Gay Su Pinnell, Nancy Roser, Linda Seigel, Dale Willows, John Shefelbine, and Robert Slavin among others. Given that these individuals have been funded by sources such as the National Science Foundation, The Office of Educational Research and Improvement within the U.S. Office of Education, and private foundation monies, it is simply wrong to characterize NICHD funded research and researchers as having the greatest influence in the state. Additionally, in both California and Texas, movement away from a holistic orientation started many years ago at the grass roots level, typically by upset parents appalled by the lack of reading ability they were observing in their own children.

In sum, for AWJ to imply that NICHD-supported researchers have been pushing their agenda on states is a serious misrepresentation. The researchers they attack have conducted scholarly research in which various approaches have been contrasted. The findings are reported according to the rules of scholarship, which are designed to guard against bias. The fact that NICHD-supported researchers, as a group, support the inclusion of explicit-systematic code instruction has come about because this is the approach supported by the preponderance of data.

One role of any good researcher is to disseminate his or her findings, to put them in the public domain, so that that they can be examined and challenged. However, good science dictates that challenge come in the form of further research -- experimentation, not rhetoric. Likewise, when there is a convergence of findings, such as has recently occurred with regard to the importance of explicitly teaching the alphabetic principle, with multiple researchers, in multiple contexts, and with a variety of different kinds of subject and research designs, then responsible scientific practices dictate that the findings be actively disseminated. This is not advocacy, this is science.

Arguments Undermining the Usefulness of NICHD-Supported Research in Reading

The Scope of NICHD Reading Research

AWJ misrepresent the scope of the NICHD program of reading research as narrowly focused only on children with reading disabilities and as having an a priori focus on phonological core deficits as a primary etiology of reading difficulties. The NICHD research program has been much more extensive than either the Grossen or AWJ articles indicate. NICHD has funded research on a variety of learning processes at all levels including molecular, cellular, behavioral, neurological and educational, using a variety of animal and human models. For more than 30 years, the program has initiated and supported studies that examine early brain development, relationships of language and reading in non-impaired children and adults, social and biological factors in literacy, and other areas that impact reading, but don’t involve disability. The more recent intervention and prevention studies, initiated in 1993 for children with reading disabilities and selectively cited by both Grossen and AWJ, are built on this larger body of research first initiated in 1965.

About 50% of NICHD supported reading research has been devoted to studies on how language, reading, and reading-related processes develop in proficient readers; the other half has examined factors that impede normal acquisition of these skills. Thus, even if the NICHD-supported reading disability intervention studies had never been conducted, and certainly if the Grossen paper had never been written, findings from NICHD-supported research on reading would still be an important source of information with implications for instruction.

Importantly, the two types of NICHD supported reading research (i.e. normal development and reading disabilities) converged independently to show the critical role of alphabetic, or phonetic, reading skills both as a support for normal emergent reading and as an explanation for reading failure in many children (for an excellent discussion of this convergence, see Share & Stanovich, 1995). There is also a very powerful convergence of evidence about the nature of word reading processes in skilled reading (Adams, 1990; Raynor & Pollatsek, 1989) with studies of the word reading difficulties of poor readers (Morrison, 1987; Torgesen, in press). This convergence has helped to provide a focus for many of the NICHD-supported intervention studies on methods for preventing and remediating word-level reading problems.

It is of further importance to understand that the convergence of findings in NICHD-supported research on reading (as well as its convergence with much research not supported by NICHD) was not guided by any systematic, centralized plan among administrators and scientists at NICHD. Typically, research ideas are submitted for support to NICHD where scientists from around the country review them for scientific adequacy. NICHD administrators do not play a major role in deciding which studies receive funding and which do not. This decision is made based primarily on scientific merit. Recently, under the direction of Dr. Reid Lyon, NICHD has escalated its investment in research on reading and reading disabilities. However, the major role that Dr. Lyon has played in guiding the research is to request proposals in the general areas of intervention research as well as research that involves co-ordinated, multi-disciplinary efforts to understand the nature, development, and etiology of reading disabilities. It is important to understand that the questions addressed, measures selected, and design of the studies themselves is all controlled by individual investigators who are not employees of NICHD.

Responsibility for All Children

AWJ argue that NICHD-RR should not be used to guide reform of early literacy curriculum, which, as they point out, is the curriculum of “virtually all children” (p. 2). They ignore the fact that half of NICHD-RR has been conducted on typically developing readers, and assert that the NICHD-RR has not been conducted with general education’s children in general education classrooms. This argument seems to suggest that research conducted on general education’s lowest performing 20% has no bearing on how reading should be taught; implying that children who experience reading difficulties are not the responsibility of general education.

So who are these children for whom AWJ deny responsibility? They are children sitting in general education classrooms being taught by general education teachers who typically have not been taught how to teach reading to children who do not learn to read easily (Moats, 1995). Typically, after being allowed to fail, these children receive labels such as learning disabled, mildly retarded, emotionally disturbed, and attention-deficit-disorder. Some of these children are blamed for their failure to learn because they are limited in their English proficiency, come from poor families, or are simply assumed to be "slow-learners." Most importantly, they are often children who can learn to read at roughly normal levels if provided the right kind of instruction (Foorman, Francis, Fletcher, Schatschneider, & Mehta, 1998). Interestingly, these are the children about whom Allington, in his other writings, appears to be most concerned (see Allington 1994, 1997).

The majority of students who have reading problems are really what can be termed ‘curriculum casualties.’ In large part, these students are poor and of minority status. In total, children experiencing reading problems represent as much as 40% of the total school population (Snow et al., 1998). For these children, we argue that the NICHD-RR provides their teachers with guidance about how to promote their early and continued reading success (Ehri, 1998). To ignore the needs of such a sizable portion of children serves only to promote an underclass trapped in a cycle of illiteracy, prevented from expressing their true potential. For these children, the NICHD-RR has much to say to teachers and program developers about how to provide equitable reading education.

Equitable reading instruction demands the incorporation of research-informed critical components such as explicit, systematic instruction in the alphabetic principle for those children who require it. Equitable reading instruction may well mean providing differentiated instruction for different groups of children within the same classroom, or even providing one-on-one tutoring (Torgesen, 1998). Equitable reading instruction absolutely requires that teachers be knowledgeable in principles and techniques for how to effectively teach critical content. Thus, equitable reading instruction necessitates reforming general education reading instruction to include those elements that are critical to insure adequate reading growth in children who do not learn to read easily (i.e., systematic, explicit instruction applying the alphabetic principle: Snow et al., 1998).

Measures Used in NICHD Reading Studies

One cornerstone of AWJ’s arguments is a disregard for the findings of recent intervention research based on a philosophical belief that the measures used in much of this research aren’t important to “real reading” and therefore do not count. We acknowledge that the measures used in the NICHD-supported intervention research conducted to date are quite different than the “authentic literacy assessment” that is currently advocated by many reading professionals (Paris et al., 1992). “Authentic assessment” attempts to judge children’s ability to apply broad literacy skills to authentic tasks like gathering information for a report, use of literacy as a medium for social interactions, or ability to read a selection and then write a response to it. It also attempts to assess children’s enjoyment, ownership, and involvement in literacy activities both at school and at home. We would certainly not argue against this type of assessment as part of an overall evaluation of reading outcomes for children. After all, if a child can read, but does not apply important literacy skills effectively to every day tasks or reads simply for the joy of reading, then some important goals of literacy instruction have not been attained. However, there must be recognition that for many children, the problem is not one of failure in applying reading skill, but rather the lack of reading skill itself. In contrast to what are often referred to as “authentic measures” of literacy, the recent intervention studies supported by NICHD typically use more objective, quantitative measures of specific reading skills. These measures are chosen because they have demonstrated reliability and validity for predicting academic performance. In contrast, “authentic measures” typically have no known reliability or validity and are often subjective.

A focus of NICHD-RR has clearly been to evaluate carefully the growth of word-level reading skills. The rationale for this focus is the vast preponderance of evidence indicating the serious impact that failure to acquire critical word reading abilities has on overall reading growth for children who end up as poor readers at the end of elementary school (Francis et al., 1996; Juel, 1988; Torgesen & Burgess, 1998). These data suggest that, whatever else a particular intervention accomplishes for at-risk children, if it does not have a significant impact on their ability to read words fluently and accurately in text, then it has not addressed a primary problem they experience in becoming good readers. In other words, attention to reading individual words matters if we want our children to successfully apply broad literacy skills to authentic tasks like gathering information for a report, use of literacy as a medium for social interactions, or ability to read a selection and then write a response to it.

AWJ poke fun at one of the commonly used measures in NICHD-supported research by referring to it as “psuedo-word pronunciation.” The clear implication of their comments is that these measures do not assess real reading skills. On these types of tasks, children are asked to read a non-word such as “surt.” The purpose is to assess the child’s ability to decode using the alphabetic principle. Real words and connected text are not used so that the child must rely on alphabetic knowledge rather than memory or context cues. There is now voluminous evidence that difficulties applying phonetic decoding skills on non-word reading tasks such as the ones used in NICHD-RR is one of the most important defining features of the reading performance of children who experience serious problems learning to read (Bruck, 1990; Rack, Snowling, & Olson, 1992; Siegel, 1989; Siegel & Faux, 1989; Stanovich & Siegel, 1994).

AWJ also misrepresent results about reading comprehension. They state that "the NICHD studies that have implemented direct instruction have found it easier to enhance phonemic awareness and non-word pronunciation performances than to influence word reading, reading rate or fluency, or comprehension achievement" (p.6). What they don't say is that word reading, reading rate, and comprehension, while harder to influence, are influenced. Likewise, the studies they choose to discuss are all kindergarten studies--studies in which children were not yet independently reading text. They discount studies where comprehension gains have been impressive. For example, Foorman et al. (1998) found an effect size of .76 on reading comprehension for gains with children in Title 1 schools receiving direct, balanced code instruction in the general education classrooms.

Another recent study (Torgesen et al., 1999) provided preventive interventions to children identified in kindergarten as the 12% most at-risk for reading failure. This study showed that the most phonemically explicit of three instructional interventions was the only one to produce significant growth in word level skills beyond that attained by children in a no treatment control group. When contrasted against the other two intervention groups, the most explicit condition had an effect size for phonemic decoding of .95 and for real word identification of .50. Although individual contrasts for passage comprehension between the most explicit and less explicit conditions were not statistically reliable, the effect size for comprehension was .43, which is educationally meaningful. In this study, the condition that provided the most intensive, explicit, and systematic instruction in word-level reading skills produced stronger growth in reading comprehension than the other two conditions, in spite of the fact that these latter conditions contained more opportunities to read and discuss text for meaning. Although the authors concluded that, had there been more time for instruction, children in the group receiving the most explicit word-level instruction would have profitted from more opportunities to engage in meaningful experiences with print, the study is nevertheless instructive concerning the importance of powerful word-level instruction for children at-risk for reading failure.

It is important to notice that AWJ present no alternative measures that could provide objective evidence for the efficacy of specific intervention methods. If AWJ know of measures that have been demonstrated to be more reliable and valid than those used in NICHD-RR, it would be very interesting to compare them directly to the measures that are currently being used. Allington did communicate to Grossen that he would need to see enhanced achievement on reading comprehension, composition, reading response, or voluntary reading and writing before he would be inclined to recommend a specific method (Allington, November 1997, personal communication). We agree that all those outcomes are desirable for an overall literacy program, but we do not feel that this minimizes the importance and meaningfulness of more fundamental measures of early reading progress. Is it meaningful that children who are nonreaders at the beginning of a study learn how to read words they have never encountered and that they read connected text accurately? Is being able to read the individual words printed on a page of text likely to increase or decrease comprehension of that text? Would a child who can independently read words never before encountered be more or less likely to read voluntarily? Would a child who understands how the sounds of language map onto printed words using the alphabetic principle be more or less likely to write voluntarily?

Grossen’s Interpretation of NICHD supported reading research

In general, we feel Grossen’s paper has been greatly misrepresented. We concur with AWJ and Lyon and Fletcher (1998) in stating that the specific recommendations go beyond the actual NICHD research base. However, the recommendations are not completely out of line with the research base. The issue is really one of specificity of the recommendations. NICHD-supported research does not specifically tell practitioners how to teach early reading; instead it provides indications about the critical elements of effective instruction. In other words, the research reports describe systematic, explicit instruction as a critical component of instruction for children experiencing difficulties. Grossen went beyond the database in defining the specific procedures that should be used in implementing “systematic and explicit” instruction. NICHD researchers have used a variety of treatment packages and, as a group, they have not recommended exactly how to provide explicit, systematic instruction in the alphabetic principle. For example, NICHD-supported research has not determined an exact order in which grapheme-phoneme relationships should be taught, but does indicate that being systematic about the order of presentation results in better outcomes. Similarly, the optimal balance and timing for use of decodable text in early reading instruction has not been specifically determined, but use of decodable text has been incorporated as one feature of interventions that have assisted children in making the greatest achievement gains. It is important to keep in mind that Grossen’s article was not intended as an all inclusive, comprehensive synthesis of the research. Rather, it was designed to communicate to practitioners basic tenets of good instruction as gleaned from the NICHD reading research program. To put this paper in context, it must be remembered that the majority of teachers and school administrators have little knowledge about how to effectively teach alphabetic knowledge and skills (Moats, 1995; Lyons, 1995). Likewise, terms like “systematic, explicit phonics” are not widely understood.

In our own research, which has involved working with and observing in hundreds of classrooms, we have been struck repeatedly by the numbers of dedicated teachers who do not know even the basics of teaching the alphabetic principle such as the importance of pronouncing sounds the way they are typically pronounced in real words, or the need to show children how to sound out words. Our observations are consistent with recent estimates that only about 1 in 10 teachers has the necessary expertise to teach reading to children experiencing reading difficulties (Lyon, 1995). Thus, we believe that, while not particularly scholarly, the Grossen article represents an often overlooked necessary link in the chain of dissemination. It provides a practical look at the types of activities used by NICHD researchers that have resulted in greater reading gains for children experiencing reading difficulties. The 7 implications of NICHD research the Grossen article identifies are:

1. Begin teaching phonemic awareness directly at an early age.
2. Teach each sound-spelling correspondence explicitly.
3. Teach frequent, highly regular sound-spelling relationships systematically.
4. Show children exactly how to sound out words.
5. Use connected, decodable text for children to practice sound spelling relationships they learn.
6. Use interesting stories to develop language comprehension.
7. Balance [decodable text and authentic stories], don’t mix.

In this discussion, we comment only on those implications that AWJ discussed (i.e. implications 2,3,5 and 7).

Implication 2: Teach each sound-spelling correspondence explicitly

The essential question here is whether or not it is beneficial for children to be explicitly and systematically taught the connections between sounds and letters in words while they are learning to read. When the question is stated at this level, it is apparent that AWJ agree with most NICHD-supported researchers about the benefits of instruction in the alphabetic principle. Where AWJ most directly challenge the support for Grossen’s statements about explicit instruction in sound-spelling correspondences concerns specific methods that can be used to provide this instruction. Whereas Grossen recommends methods that introduce sound-spelling correspondences systematically outside the context of meaningful reading or writing activities, AWJ support less systematic explication of these relationships within the context of activities like writing where they are taught in the service of a goal involving meaningful communication. AWJ simply do not believe that the results of the cited NICHD studies are meaningful.

However, the NICHD-supported studies cited by Grossen are clear that interventions that provide more systematic explicit instruction in sound-spelling correspondences produce better reading growth than those that do not. Perhaps the most convincing of these studies to date was recently reported by Foorman et al. (1998). In this study, the investigators contrasted instructional methods involving explicit instruction in alphabetic reading skills with methods that contained less explicit and systematic instruction in the code. Within methods that contained explicit instruction in sound-letter correspondences, they contrasted two interventions that varied in whether instruction was provided at the single phoneme level or the onset-rime level. The research was conducted in Title 1 schools in intact first- and second-grade general education classrooms. The children who received the most complete and explicit instruction in sound-letter correspondences at the single phoneme level outperformed children in the other instructional conditions on multiple measures including phonetic decoding, word identification, and passage comprehension. In AWJ’s portrayal of this study, the reader is left believing that no meaningful effects were detected because the groups did not perform differently from one another on one of the two measures of reading comprehension. As the authors explained, and as the data suggested, the comprehension measure on which the groups did not differ (i.e., Formal Reading Inventory - Comprehension) was simply too difficult for children in first and second grade; this measure was not sensitive to growth for any of the children in any of the groups. The explicit code group did demonstrate greater comprehension on a measure at the appropriate level of difficulty (ES = .76).

We are also concerned that the second author’s work (Torgesen, Wagner, & Rashotte, 1997a) is quoted out of context as indicating that direct instruction in phonics does not generalize to orthographic reading and reading comprehension. In the study that this quote pertains to, all the intervention programs involved some degree of direct instruction in phonics and they also all produced growth in sight word and passage comprehension skills. The method that involved the most explicit instruction produced the greatest gains in these areas, although the differences for passage comprehension across instructional methods were not statistically significant. Perhaps the most important finding of this study involved the overall success of children identified in kindergarten as the most at-risk for reading failure, and thus grade retention. The children who had been randomly assigned to the most systematic and explicit instructional condition had only a 9% retention rate between kindergarten and second grade, while the children in the other two experimental conditions had a 20% to 25% retention rate and the control children showed a 40% retention rate. The point of the Torgesen et al. (1997a) study is that with early intervention that contains explicit and systematic instruction in alphabetic reading skills, phonologically-based reading problems can, in most instances, be prevented. The major evidence for this statement comes from the finding that the word-level reading skills of these children (which is a critical area of difficulty for these children) were solidly in the average range for both fluency and accuracy at the end of the study.

Scanlon and Vellutino’s (1997) research is also misrepresented as demonstrating that “NICHD-supported researchers indicate letter-sound instruction in early school environments was not the critical instructional difference in fostering early reading achievement in the at-risk children studied” (p. 7). In contrast, what Scanlon and Vellutino (1997) concluded is that kindergarten teachers typically are oriented toward more holistic practices and that “children who enter kindergarten at-risk for reading difficulties, but who are good readers by the end of first-grade...come from kindergarten classrooms in which a greater portion of time is spent in analyzing the structural…aspects of the spoken and written word” (p. 209). What we find most disconcerting about AWJ’s presentation of this study is they choose to ignore the discussion pertaining to the observed inadequacy of exclusively holistic, meaning-based activities for kindergarten children.

We were particularly concerned with AWJ’s treatment of the study by Lovett, Borden, DeLuca, Lacarenza, Benson, and Brackstone (1994). What this study indicated was that teaching sound-symbol relationships explicitly was superior to implicit approaches with older, severely deficient readers. The point was not that they were still behind, but that they made greater progress with one technique over another. Given that these students were already so far behind, finding the tools to help them rapidly achieve more skills is very important, even though they did not achieve 4-years growth during the treatment period, which was actually quite short in this study. Researchers are trained to present their results in a conservative manner, not overstating their implications. The conservative nature of good researchers should not be used as tool to discredit their findings. Achieving positive results with such challenging students should not be written off as insignificant.

A second problem that AWJ have with implication 2 is that Grossen cites 14 references, half of which are commentaries or reviews, rather than original studies. Given that Grossen's purpose was to communicate to practitioners what NICHD researchers recommend based on their research, we see no problem with citing the summaries and commentaries of those researchers, in addition to specific studies. Likewise, if practitioners are interested in gathering the original studies, the original citations are located in the reviews. Since practitioners are less likely to possess the skills necessary to comprehend the original research, reviews may well be more appropriate.

Implication 3: Teach frequent, highly regular sound-spelling relationships systematically

The problem that AWJ have with this recommendation appears to be that the published studies, which are restricted by page limitation, do not provide detail about which sound-symbols were taught and if the sound-symbol correspondence taught to children represented the most frequent sound represented by the letter. It is true that the NICHD researchers have not explained their treatments at this level of detail in their published manuscripts. However, this does not mean that the implication is incorrect. What it does mean is that the researchers assume that those who are interested in this type of specificity will themselves examine the materials used within their research. The fact is that those treatments that have achieved the best results have taught first the sounds that appear in words more frequently and have typically presented the most regular sound spelling relationships as described in Grossen’s article. To determine this, one must examine the actual curricula that achieved the best results. Also, Grossen’s list of 48 sounds is intended to be informational, not specific to any one curriculum. Our experience in working with teachers is that the information presented in the table is not common knowledge. Grossen is very clear that these letter-sound correspondences presented are not from any one program or curriculum, and are only presented to provide the readers with adequate background knowledge. Thus, we do not find this table problematic because it does not have any bearing on any of the research, but rather is helpful for understanding the research.

Implication 5: Use connected, decodable text for children to practice sound spelling relationships they learn

AWJ state that, “None of the NICHD-supported studies isolated or manipulated the decodable text variable” (p. 9). They are correct, but this is not the most important point. The most effective treatments in all of the NICHD-supported intervention studies involved the use of decodable text at some point in the instruction. It is from this fact that Grossen rightly indicates that NICHD research supports the idea that decodable text is an important component of successful reading programs.

The reason the researchers have not isolated decodable text as a distinct component of instruction is that it has been considered a logical component of any intervention that seeks to strengthen the kind of word reading skills that do not involve guessing from context. The words in text cannot be directly identified (either through phonetic analysis or orthographic recognition) until the elements comprising the text have been taught. Thus, decodable text must be linked to a total program of teaching letter-sounds, sounding-out skills, and sight words before it becomes decodable. Second, in light of what is known about transfer and generalization, it makes no sense to teach letter-sound correspondences, sounding-out skills, and sight words strictly in isolation. In fact, to deliberately deprive children of reading text designed to facilitate the integration of these skills would be, in our opinion, unethical. Quite frankly, we are not sure why AWJ take issue with having children read decodable text to promote transfer, generalization, and meaningful integration of alphabetic skills when they, themselves, acknowledge that it is important for children to acquire these skills. To teach skills in isolation, without the text, is reminiscent of the isolated skill and drill phonics of the past-- practices Allington advocates against (Allington, 1994, 1997). Likewise, to ask children to read text that they cannot decode using the alphabetic elements and skills they have been taught, is to communicate to them that the alphabetic knowledge and skill they have spent effort learning is not really relevant to reading, and that they must rely heavily on guessing the identity of words from context.

Implication 7: Balance, don’t mix

Here, AWJ are responding to Grossen’s (1997) statement that, “A common misconception regarding the balance that is called for by the research is that the teacher should teach sound-spelling relationships in the context of real stories” (p. 13). Grossen is making a very similar point about instruction and practice here as she did in implication 5. The essential question is how to reinforce the use of phonetic and visually based word-reading skills in young children. Should they be given opportunities to practice reading in texts that allow them to identify most of the words on the basis of phonetic and visual information in the text, or should they practice reading texts that require them to guess a high proportion of the novel words from context? AWJ are correct in stating that NICHD-RR has not directly examined the proper balance between decodable text and “real stories” in the reading selections given to young children. However, Grossen's recommendation that young children who are beginning to learn to read should be given both kinds of texts for different purposes (decodable text to practice independent word identification skills, “real stories” to experience in supported-reading or teacher-reading contexts) is consistent not only with the recent synthesis of research provided by the National Research Council (Snow et al., 1998) but with a large body of both NICHD-RR and research supported by other sources (Gough and Walsh, 1991; Raynor & Pollatsek, 1989; Share & Stanovich, 1995).

We do want to take direct issue with one of the comments made in the AWJ paper relevant to the issue of “balance.” In attempting to show that NICHD-RR research is inconsistent with Grossen’s recommendations, they cite the testimony of Reid Lyon (1997), who is the director of the branch that supports much of the research, as stating, “NICHD researchers have found that classroom instruction that explicitly addresses the connections between letters and sounds within a literature-rich classroom environment can make a difference between reading failure and reading success.”(p.4). What Dr. Lyon was attempting to communicate here is that many children will learn to read much better if they are given systematic and explicit instruction and practice using phonetic and visually-based word reading strategies in a context that also provides rich opportunities to interact with meaningful text. He was arguing that neither of these important components should be left out. He did not exclude the use of decodable text as a device for practicing word-reading skills, and yet that is the clear implication made by AWJ in their discussion of this citation.

Long Term Impact

Since AWJ wrote their critique, new important research has been released that speaks to the long-term impact of at least some of the intervention research that we feel bears on the utility and generalizability of the NICHD-funded research. Thus we include a brief description here. Presently other longitudinal research examining long-term impact is underway. In a recent study of remedial instruction with children in late elementary school, Torgesen and his colleagues (Torgesen, Alexander, Wagner, Rashotte,Voeller, & Conway, 1999) found that explicit instruction in phonemic decoding skills, coupled with opportunities to apply those skills effectively in reading text, produced dramatic and lasting gains in the reading ability of children who began the study with severe reading disabilities. For example, the children began the study with an average standard score (Mean = 100, S.D. = 15) for phonemic decoding ability of 69. After 67.5 hours of one-to-one tutorial instruction, their score on this test was 93, and a two-year follow-up (during which they had received no additional intervention) showed their average score to be 91. For comprehension of text, the children began with a standard score of 81, with an immediate posttest score of 90, and a two-year follow-up score of 96. Since this group had a measured verbal IQ of 93, their final performance on reading comprehension was slightly higher than one would predict on the basis of general verbal ability.

Conclusions

In truth, we are not clear why the NICHD-RR should be so vigorously attacked when what is actually espoused by this research and these researchers is so reasonable, data-based, and non-exclusionary. The community of scholars whose work in reading has been supported by NICHD has not made any specific, one-size-fits all recommendations. All that has been posited is that, based on all available evidence from both normal and disabled readers, it appears that multiple factors are critical to early literacy success, of which phonological knowledge and skill is one such element. The research suggests that, in order for reading instruction to be maximally effective for all children, it should contain explicit instruction in phonetically and visually-based word-reading strategies--teaching strategies that have been explicitly rejected in the holistic approaches to reading instruction that have been widely utilized in recent years. Allington himself appears to actually support many of the same positions that NICHD supported researchers advocate (see Allington, 1994).

It seems to us an unnecessary exercise in futility to attack the utility of sound research and present a bewildering and misleading picture to practitioners. We are very worried that such activities confuse practitioners who are left believing that they can still pick and choose reading approaches as if choosing food at a cafeteria. Likewise, the constant arguing between "experts" can only leave practitioners distrustful. We propose that what "experts" should be engaged in is acknowledging commonalties and continuing to examine, scientifically, the impact of particular practices for which questions remain.

We would challenge AWJ, and others who are critical of the instructional recommendations derived from NICHD-RR, to conduct equally well-controlled investigations of instructional methods. Although there is clearly room for discussion about what constitutes evidence of instructional effectiveness, we would assert that, at a minimum, the measures used should have demonstrated reliability and validity. In order to make valid systematic comparisons among methods that can be replicated by others, at least a core of objective measures, that provide quantitative data to be used in systematic analyses, should be used. Although case studies and descriptive research have much to contribute in terms of explaining the context and conditions of observed effects, it must also be recognized that these research tools cannot distinguish the efficacy of various methods or approaches and should not be given the same weight as research that experimentally manipulates instructional variables under relatively well-controlled conditions.

As for the Grossen (1997) article, we reject the position that this one paper has had wide spread impact on policy in California and Texas. Instead, we take it for what it is, a practitioner-oriented article designed to communicate some basic principles of instruction that are consistent with what NICHD-RR suggests are important for early literacy instruction designed to prevent reading failure. The recommendations and examples included were not intended to be definitive. We believe that Grossen was well justified to present examples because, in our own experiences working with teachers and administrators, it has become clear to us that these practitioners are sometimes confused by the language researchers use and find specific examples very helpful.

We agree with Allington (1994) when he states that “children are more likely to learn what they are taught than what they are not” (p. 1). We agree with Allington that “it is simply not necessary that some children fail to learn to read well” (p.2).

Without a strong instructional component children are left to their own devices to discover the strategies and processes that skillful readers and writers use. Many children attempt to puzzle through the activities but never discover the thinking patterns that skillful readers employ. We now label these children and schedule them for special instructional programs. It is time, instead, to teach them what they need to know. (Allington, 1994, p. 10.) This statement is completely consistent with evidence gathered by NICHD supported researchers over the past 30 years. However, before we can teach children what they need to know, we must know what it is they need to know. Thus, we must look to data, not dogma, for guidance. If we are ever to achieve equity in reading instruction for all children, we must, as a field, use findings of high quality research as they become available to help us design and implement the best quality literacy instruction to “teach them what they need to know.”

References

Adams, M.J. (l990). Beginning to read: Thinking and learning about print. Cambridge, MA: MIT Press.

Adams, M.J. (1998). The three-cueing system. In F. Lehr and J. Osborn (Eds.). Literacy for All: Issues in Teaching and Learning (pp. 73-99). New York: Guilford Press.

Allington, R.L. (November, 1997). Personal communication. [Electronic mail from R. Allington to B. Grossen, forwarded by B. Grossen to the P. Mathes]