What Science Offers Teachers of Reading

Not long ago, one of us heard a disturbing but revealing story about a school district participating in a research project on early intervention with at-risk readers. Like other successful interventions in reading, this program combined explicit instruction in word analysis (i.e., using knowledge about the alphabetic principle to read unfamiliar words), popularly known as phonics instruction, with ample practice in reading connected text and developing comprehension. However, the school district in question decided to drop the program — despite convincing data of its effectiveness with their own students — because, said district administrators, it was against their “philosophy” to teach phonics.

In contrast, phonics instruction is sometimes presented as some sort of magic elixir for reading difficulties. For instance, on a recent television news show, a prominent teacher educator demonstrated his highly structured phonics program with a group of at-risk youngsters, claiming not only that all children learned to read easily with his program, but that they could do so in kindergarten.

There is a very substantial research literature in the field of reading, but, from these and many other incidents, one wouldn’t know it. Instead, as in the first story, educational decisions that might be made by reference to a base of scientific knowledge frequently are treated as though they are entirely a matter of personal beliefs. And, without knowledge of the scientific evidence about reading, it is difficult for anyone, including educators, to evaluate sweeping claims about the effectiveness of specific programs, like those in the second story.

In this article, we examine what science offers teachers of reading, both general and special educators. Although we think that science has something to offer teachers in many domains, we concentrate on reading for several reasons. First, there is probably more scientific knowledge about how children learn to read, and about why some children experience reading difficulties, than about any other field of education. Second, despite this large body of evidence, the debate about how, and even whether, scientific evidence should inform educational practice has been particularly fierce in the domain of reading. And finally, reading difficulties are a common reason for failure in school, as well as the most frequent reason for identifying children as needing learning disabilities services.

There is a “research-to-practice gap” in many areas of education, not only in reading. Fuchs and Fuchs point out that misconceptions about the relationship between research and practice — such as the idea that researchers do not respect teachers or that empirical research in education is a failure — are common in the professional literature. However, many investigators have emphasized a research-to-practice gap specifically in the domain of reading. For example, Torgesen, Wagner, and Rashotte note that, although we still have much to learn about the best ways to help children with the most severe reading problems, we already know a great deal that often is not applied in educational practice.

Why does this disconnect between educators and scientific knowledge about reading exist? One important reason may be that many teachers simply are not exposed to scientific research on reading in their pre-service or in-service preparation. Prospective elementary and special-education teachers must be prepared to teach many subjects, and they must also have knowledge in areas such as child development, health, and — particularly in the case of special educators — special education law. Thus, especially at the pre-service level, teachers usually receive only a cursory introduction to reading instruction (Moats, 1995,1999; Nolen, McCutchen, & Berninger, 1990). Both general and special educators have reported that their real education occurs after they have entered a classroom.

Specific concerns have been raised about teachers’ preparation for teaching the word-analysis aspect of reading. Even experienced general and special educators may lack formal knowledge about English word structure that is necessary for teaching word analysis to beginning or struggling readers, in which special-education teachers were observed delivering reading instruction to elementary students with learning disabilities over the course of a year, found that there was little direct instruction addressing word analysis, perhaps reflecting teachers’ lack of preparation in this area.

Of course, based upon practical experience, educators may have excellent intuitions about what is effective in reading instruction, and many teachers without knowledge of scientific research still teach well. Neither is scientific knowledge about reading sufficient to make an individual a good teacher. Nevertheless, it makes little sense for educators to be cut off from what science has to offer. Most important, knowledgeable teachers are essential for ensuring that scientific research on reading actually benefits children.

Scientists in a wide variety of fields have emphasized that certain characteristics and values are common to all areas of science, whether one is studying subatomic particles, distant galaxies, or reading. These characteristics and values include the following: gathering evidence through systematic observation and testing; stating claims in ways that are potentially falsifiable; submitting evidence to the scrutiny of the wider scientific community through the process of publication and peer review; attempting to be as objective as possible; considering alternative explanations of evidence; and considering a cumulative body of evidence.

In recent years, scholarly debates about literacy instruction have focused in part on how research itself is conducted — on the research paradigm. For instance, some researchers have chosen more quantitative approaches to research (i.e., studies in which numerical data are gathered and used in statistical analyses) and controlled experimental designs. Controlled experiments involve the use of a control group (one that does not receive the treatment being studied) and holding certain variables constant so that the influence of other variables may be determined.

In contrast to quantitative and controlled experimental designs, other researchers have been oriented toward qualitative and ethnographic types of studies. These studies emphasize conducting research in educational, community, and other social settings, and verbal data (e.g., statements made by students coded into various categories) more than numerical data. Qualitative and ethnographic studies typically focus on the nature of the classroom or home context, as well as the quality of the interactions among teachers, students, and parents. Often they attempt to consider language and literacy learning from the students’ (as opposed to the researchers’ or teachers’) viewpoints — see, for example, Heath, and MacLeod.

Of course, many investigators have combined both qualitative and quantitative types of studies. Nevertheless, divisions over these different approaches to research have led to considerable debate over what constitutes “real science,” with some observers suggesting that a rapprochement between different camps may be impossible.

Because of this debate, we must emphasize that we do not use the term “science” to refer only to one particular method of conducting research. Indeed, many scientific studies of reading simply involve observing and documenting systematically how ordinary children learn to read in actual classrooms. It is important, however, for educators and other consumers of research to be aware of the respective advantages and limitations of different approaches to research. For example, qualitative and ethnographic studies, which emphasize context and the participants’ viewpoints, are extremely useful for identifying important contextual variables in literacy acquisition, such as attitudes toward school learning, linguistic differences, and home experiences with literacy. However, these studies typically lack the controls needed for ruling out alternative explanations of a phenomenon or for determining causality. Thus, questions such as whether one type of reading program is more effective than another, or which cognitive difficulties are causally central to reading disabilities, are the purview of controlled quantitative research.

The point is not that only one method of conducting educational research should be used or that only one method is “scientific.” Instead, different methods are useful for different aims. Furthermore, when implemented appropriately, both quantitative and qualitative methods of research should adhere to the basic principles of science described in the opening of this section.

Examples of well-established scientific findings

The unusually contentious nature of the reading field, combined with the tendency of the media to report primarily controversial or unusual claims, may lead to the impression that well-established scientific findings about reading do not exist. To the contrary, literature reviews of research on reading acquisition, reading instruction, and reading disabilities have converged on some basic conclusions with which few scholars of reading would disagree.

Among these conclusions are the following:

1. Children who are learning to read an alphabetic language such as English appear to pass through a series of common developmental stages.
2. In the earliest stages of learning to read, children’s sensitivity to the sounds in spoken words (i.e., phonological awareness) greatly facilitates their developing word-analysis skills; children who lack this awareness have trouble learning to read.
3. Accurate and fluent word analysis is essential to good reading comprehension, especially as children grow older and encounter increasingly difficult texts.
4. Most beginning readers benefit from explicit instruction in word analysis (i.e., phonics instruction), and for youngsters who are vulnerable to reading difficulties, this instruction appears to be particularly important.
5. In order to become skilled readers, children need much more than just word-analysis instruction — among other things, they need good comprehension instruction and extensive experience reading a variety of texts, such as books, magazines, and newspapers.
6. Instructional approaches that attempt to “match” initial reading instruction to certain characteristics of the reader, notably modality preference, are generally ineffective.
7. Children who experience serious difficulties in learning to read, including those classified as having reading disabilities or dyslexia, typically have difficulties in the areas of phonological processing and word analysis.

Unfortunately, some popular educational practices conflict with this evidence. For instance, for more than a decade, regular-classroom reading instruction has been dominated by the idea that children learn to read the same way that they learn to talk, by being immersed in language and focusing on meaningful communication. Thus, learning to read should be as natural as learning to speak — if only educators duplicate, for reading acquisition, the conditions that typically are in place for oral language acquisition (e.g., emphasize meaning rather than isolated-skill learning and provide extensive experiences with books and print).

However, there are some crucial differences between oral language acquisition and reading acquisition that do make learning to read significantly more difficult than learning to speak for many youngsters. For example, learning to read an alphabetic language requires a more explicit awareness of the phonemes within spoken words than does learning to speak. Furthermore, compared to other alphabetic languages, English has a complex orthography; for instance, many English words, such light and nation, cannot be decoded in a simple letter-by-letter fashion but require the reader to attend to patterns of letters within words (such as igh and tion). These features of written English may help to explain the consistent finding that beginning readers in general, and children with reading disabilities in particular, benefit from explicit, systematic instruction in word analysis — including some instruction involving isolated letters and words. Another educational practice, especially common with poor readers, involves matching the method of reading instruction to a child’s learning style or processing strengths. This approach often focuses upon modality preference, that is, whether the child is a “visual,” “auditory,” or “kinesthetic” learner. Children with a presumed visual or kinesthetic style are then said to require holistic or tactile methods. as opposed to phonics instruction.

However, matching instruction to modality preference consistently has been found ineffective and is not compatible with other kinds of evidence on reading acquisition and reading disabilities. For instance, because poor readers usually have trouble with phonological (i.e., “auditory”) types of tasks, they may indeed appear to be comparatively “visual” or “kinesthetic” in style, and they may also experience considerable difficulty learning word-analysis skills. However, an instructional approach that shuns direct teaching of word analysis is not the answer to these problems, because the evidence suggests that it is not possible to become a good reader in English without acquiring the ability to decode unfamiliar words accurately and rapidly. Rather, poor readers require intensive, highly skilled instruction in word analysis, as well as in many other important reading abilities, such as vocabulary and the use of comprehension strategies.

Certain methods of identifying reading disabilities provide a third example of educational practices that are problematic from a scientific standpoint. The Individuals with Disabilities Education Act (IDEA) includes reading disabilities under the umbrella category of learning disabilities (LD) and requires an ability-achievement discrepancy as one criterion for identification. In many state regulations, the ability-achievement discrepancy is operationalized as a discrepancy between IQ and achievement. That is, in order for a child to qualify for educational services in the LD category, low achievement alone is not sufficient; rather, a child’s achievement must be low relative to his or her IQ.

Many scientific investigators have compared poor readers who meet IQ-achievement discrepancy criteria with poor readers who do not. The latter group of non-discrepant poor readers, who fail to meet discrepancy criteria, generally are children with IQs in the low-average range — children whose IQs are not high enough to yield a significant discrepancy but well above the range of intellectual disability. The usual finding in these studies is that both groups of poor readers have very similar types of reading-related cognitive difficulties, revolving around weaknesses in phonological processing and word analysis. Moreover, both groups of poor readers appear to benefit from similar remedial procedures (Hurford et al., 1994). These kinds of findings have led many scientific investigators to caution against excluding non-discrepant poor readers from educational services — a caution that is in direct conflict with most current educational policies for identifying learning disabilities.

Although the scientific knowledge base about reading encompasses much more than cognitive research, knowledge about the cognitive processes that are important in learning to read (and in acquiring other areas of literacy as well, such as spelling) has proliferated especially rapidly in the past fifteen years or so. The cognitive processes involved in reading include phonological processes, such as phonological awareness; orthographic processes, such as knowledge about letter strings that are valid or invalid in English; and the use of strategies to aid comprehension, such as connecting prior knowledge to new information.

Much of the research on reading-related cognitive processes has important educational implications. Consider the extensively researched area of phonological awareness, which involves sensitivity to the sounds in spoken words, and which is often assessed through oral blending, segmentation, rhyming, or alliteration tasks. Not only can phonological-awareness tasks be used to screen young children for potential reading difficulties, but training phonological awareness — especially in conjunction with a formal program of reading instruction that makes explicit the connections between the phonemes in spoken words and those in written words — may improve reading achievement in these children.

The relative importance of different cognitive processes tends to shift with development. For example, processes such as phonological awareness are most important at the beginning of reading acquisition, whereas other processes, such as the ability to use strategies to aid comprehension, tend to be especially important somewhat later in reading acquisition. Many investigators have outlined models of typical reading development that are framed in terms of the cognitive processes important to reading. Knowledge about typical reading development is extremely useful for designing instruction and for effectively accommodating the wide range of individual differences that exist in any classroom.

Consider, for instance, one second-grade classroom in a middle-class suburban school district. The average readers in this class are at the “fluency building” stage of reading; they are able to read most common words accurately, but not without some effort. These youngsters would benefit from, among other things, an emphasis on instructional activities that promote fluency (e.g., rereading familiar texts). The highest-achieving readers in the class, who can read text written at a fourth or fifth-grade level, already are highly fluent readers. These children do not need to focus on building fluency, but rather would benefit from an emphasis on the vocabulary and comprehension skills involved in understanding the more advanced texts that they are capable of reading. All of these average and above-average readers have good phonological awareness. However, the lowest-achieving readers in the class are functioning at a first-grade level and need help with basic word-analysis skills; among these youngsters are several who lack even rudimentary phonological awareness and who would very likely benefit from phonological-awareness activities.

It is not, of course, that only the high-achieving readers should be taught comprehension skills, or that the low-achieving readers should focus on word analysis and phonological awareness to the exclusion of everything else. Rather, all the children need a broad, comprehensive program of reading instruction, but within the context of such a program, different children need different instructional emphases. These emphases in turn depend at least in part upon children’s phase of development in reading.

However, knowledge about the cognitive processes involved in literacy is at the level of psychological description and does not permit conclusions about the ultimate cause of a given youngster’s reading difficulties. For example, the fact that a particular youngster has a deficit in phonological awareness does not provide proof that the child’s reading difficulty is caused by an innate biological disorder, as opposed to, say, lack of literacy-related experience or inadequate instruction. But poor phonological awareness does suggest that the youngster will need help in acquiring the phonological skills important in learning to read.

Although the kinds of scientific findings described above are extremely useful educationally, we think that science offers educators more than an information base for screening at-risk readers and planning instruction. For one thing, because the scientific knowledge base about reading is continually increasing and evolving, educators need to stay reasonably current with scientific research in their field on a long-term basis. Thus, scientific knowledge is an important resource for ongoing professional development.

In addition, the values and methods of science — such as gathering evidence through systematic observation and testing, and considering alternative explanations of evidence — provide powerful tools for settling disputes and for educational decision-making. As Stanovich (1993-1994) points out, reliance on a political rather than a scientific model for making decisions has created endless problems for education. Some of the extreme pendulum swings that have characterized reading education in particular might be avoided with a scientific model of decision-making. For instance, in our opening anecdotes, we illustrated two extreme positions on phonics instruction, one suggesting that phonics instruction is unnecessary, and the other that it is a panacea for reading difficulties. However, the scientific evidence favors an intermediate view: Good word- analysis instruction clearly is a key component of successful early intervention programs, but even with this instruction, some children still experience serious problems in learning to read.

This is not to say that science can solve every problem, or that scientific knowledge is all that matters. Many other kinds of knowledge contribute to professional growth and are important in making educational decisions — such as, for example, professional experience and judgment. However, experience alone does not necessarily lead people to correct conclusions. For example, despite extensive experience with objects in the physical world, most people do not generate the laws of physics on their own.

Furthermore, many experiences are open to multiple interpretations. For instance, poor readers’ tendency to confuse the letters b and p traditionally has been interpreted as a visual perceptual problem, a reasonable interpretation given the similar visual appearance of the letters. However, current evidence suggests that confusions between b and p, and many other pairs of letters, more likely reflect the fact that the sounds represented by the letters are articulated similarly: b and p are produced in the same way with the lips and tongue, but the former is voiced and the latter is voiceless. In other words, rather than reflecting a visual-perceptual deficit, the b-p confusion typically is language based.

Of course, science and professional experience are not mutually exclusive, but rather complementary and interactive sources of knowledge. As the b-p example suggests, scientific research about reading is an important framework for interpreting observations of specific children, as well as a source of ideas for guiding instruction. At the same time, classroom observations of children are extremely useful in generating ideas for future research, and teachers’ judgment and experience are essential for translating research findings into educational practice. Thus, longstanding collaborations between researchers and educators may yield considerable benefits for both research and practice.

Here are some suggestions for educators interested in becoming more familiar with research on reading and more astute consumers of research:

1. Subscribe to and read several peer-reviewed journals that publish scientific investigations of reading. Examples of such journals include not only Learning Disabilities Research & Practice, but also Reading Research Quarterly, Exceptional Children, Journal of Educational Psychology, Journal of Learning Disabilities, Scientific Studies of Reading, and The Reading Teacher. If you are unsure which journals are most appropriate for you, you can peruse a variety of journals at any university library.
2. Avoid reading only what is familiar, comfortable, and consistent with your personal beliefs. Specifically, seek out opposing points of view on a given issue, such as the best ways to teach word analysis or the utility of strategy instruction. Likewise, read studies that represent a variety of approaches to research, such as both ethnographies and controlled experimental designs, and keep in mind the inherent limitations of different kinds of studies.
3. Don’t be put off by statistics. If you don’t have any formal statistical background (and sometimes even if you do!), the “results” sections of many quantitative studies can be daunting. Of course, one possible solution is to take a basic statistics course if you haven’t already had one, preferably a course oriented toward reading or education. However, you can also learn quite a bit about statistical and probabilistic reasoning by reading some books that are intended for the general public, such as Paulos and Gilovich (1991). In addition, keep in mind that the statistics of articles in peer-reviewed journals are scrutinized by an editor and several reviewers before publication. It is not necessary to be an expert in statistics in order to understand and evaluate the gist of most articles.
4. Judge claims and arguments based on a total body of evidence. The scientific knowledge base about literacy is very substantial and always increasing. Reliable claims about reading do not hinge on a narrow set of studies; rather, they are consistent with a broad range of evidence. It is important to read widely enough among credible scientific investigations of reading to encounter alternative viewpoints and to get the “big picture.”
5. Apply well-established scientific findings in educational practice. Although many aspects of educational policy and decision-making are not in the control of teachers (such as the eligibility regulations on learning disabilities mentioned previously), scientific findings still can be usefully applied by individual educators in everyday practice. For instance, in order to meet the needs of a wider range of children, teachers may choose to supplement a district reading curriculum that is weak in teaching word-analysis skills with their own activities for teaching word analysis and phonological awareness.
6. Beware of false dichotomies. Some examples of false dichotomies in reading education include phonics vs. literature-based approaches to teaching reading, homogeneous vs. heterogeneous grouping, and isolated vs. contextualized skill instruction. In each of these instances, rather than choosing one approach or type of instruction over another, teachers can (and frequently do) combine choices. The scientific question in these situations usually involves finding the right balance (e.g., exactly how much phonics instruction is beneficial, and how does the amount needed vary depending on the characteristics of the children?) — not making a forced choice between two extremes.
7. Be suspicious of simplistic claims. A headline like Many Factors Involved in Reading Achievement might not sell a lot of newspapers, but it is closer to the truth than All Children Learn to Read Easily with Phonics or New Program the Solution to Reading Failure. As we have indicated, there are clear conclusions about reading that can be drawn from scientific research. However, like problems in many other human domains, problems in reading tend to be complex, involve multiple causes, and require more than one solution.
8. Use Internet resources, and evaluate them with criteria similar to those for conventional print materials. For example, although most internet articles are not peer reviewed, you can look for web sites that are continually updated, that represent more than one viewpoint, and that feature work by authors with standing in the scientific community. Some useful web sites for teachers of reading include those of LD Online (opens in a new window), the International Reading Association (opens in a new window), the Center for the Improvement of Early Reading Achievement (opens in a new window), the International Dyslexia Association (opens in a new window), and the U.S. Department of Education (opens in a new window).
9. Attend professional conferences. Conferences are a particularly good way to keep up with current thinking and research in any scientific domain, including reading. Examples of professional groups with annual or regional conferences relevant to teachers of reading include the Council for Exceptional Children (CEC), the International Reading Association (IRA), the International Dyslexia Association (IDA), and the Society for the Scientific Study of Reading (SSSR).
10. Visit model programs and seek mentoring from more experienced educators who are viewed as effective teachers of reading. Seeing scientific knowledge about reading translated into everyday educational practice can be extremely helpful — as can feedback from an especially able colleague.

To sum up, scientific knowledge about reading is not the only kind of knowledge that teachers need, nor does science have all the answers. But given the accumulated knowledge base that is now available about reading, science does have some answers to important educational questions, such as how to identify young children with potential reading difficulties and how to teach them effectively. Furthermore, science also offers educators a powerful tool for solving problems, for making decisions, and for ongoing professional growth.