Early Identification: Predicting Reading Disabilities and Dyslexia

  • 15 February, 2019

Blast from the Past: First posted on February 15, 2019 and reposted on February 10, 2023. This entry seems timely given recent legislative efforts to impose early literacy assessments aimed at identifying dyslexia. Legislatures across the country have been passing laws requiring screening and monitoring assessments. However, when you look at the states that have such legislation we're not seeing improvements in reading achievement. That reminds me that there is a very small research literature on the positive impact of early assessment on learning. Despite potential benefits of early screening, it only really helps if it leads to instructional efforts that successfully address the deficiencies that are identified. Many states are supporting the increased assessment, the relatively inexpensive identification of reading problems. But there has been relatively less concern for providing the additional support needed for more extensive professional development for teachers, and extra teachers and instructional materials for targeted instruction. I suspect that we don't have more research showing learning benefits from early assessment -- not because it is so difficult to develop sound tests but because the instructional follow up is so often inadequate or lackadaisical.

Teacher question:

Prevention of dyslexia and other reading problems should be everyone’s number one priority. Why isn’t there more emphasis on the early identification of reading problems, before they have a chance to ruin children’s lives?

Shanahan response:

In 2018. I was asked to edit an issue of Perspectives of Language and Literacy devoted to this issue. Below is the introduction to that issue and at the end I have included a link so you can follow up on any of the other articles in this issue by an impressive array of scholars who know a lot about the early identification of reading problems.

When I was a young teacher, I taught children with reading problems. Teachers would refer some of their students for evaluation, I would give them a test and decide who I could work with. One youngster that I added to my rolls was a first-grader.

I soon found myself chastised by the district school board for this particular decision.

“Why would you give special reading teaching to a 6-year-old?” I was asked.

In 2018, my decision seems more like “business as usual” than the board’s questions might suggest. These days I would have little to explain for providing extra reading tuition to a first-grader. But why was that so unusual 50 years ago?

The ideology around reading in those days held that students who struggled with beginning reading would eventually outgrow the problem. Low maturity was seen not as something that prevented learning—it simply delayed it. Intervening too early would not help, since the student would still be immature (what 6-year-old isn’t?), and my extra instruction might do harm and was certainly a waste of resources.

The idea of preventing—as opposed to remediating—reading difficulties has been around since the 1930s. However, researchers made little headway with the problem for about 30 years.

The earliest study of the issue that I’m aware of is Chester Bennett’s (1938) An Inquiry into the Genesis of Poor Reading. Bennett’s idea of early identification was to look at second- and third-graders to try to figure out their differentiating characteristics. Given that goal, the study was an abject failure. The author looked at a wide range of characteristics… birth order, speech defects, persistence, physicality, attitude toward school, incidents of crying, fear, headaches, and so on. With the exception of the ordinal birth position and, perhaps, speech defects, the whole list of features was as likely to be the results of reading problems as their cause. The author’s conclusion: researchers should go back to an earlier time in the child’s life. Indeed.

Unfortunately, it was a good long time before researchers took him up on the challenge. Oh, there were small investigations here and there showing that speech problems implicated in reading disability could be detected earlier (Hildreth, 1946), or that using more effective instructional procedures in grade 1 could “prevent” reading problems (Dunklin, 1940; Yoakam, 1943). But there was no concerted effort to develop schemes for predicting who was likely to have difficulty in learning to read—or to develop interventions aimed at disrupting these predictions (rendering the sure failures successful). 

That would change with the landmark contributions of Jansky and deHirsch. Katrina deHirsch was the director of the Pediatric Language Disorder Clinic at Columbia-Presbyterian Medical Center from 1941 to 1972 and her colleague Jeannette Jansky was a learning disabilities specialist. Their book, Preventing Reading Failure: Prediction, Diagnosis, Intervention (1972) provided a longitudinal analysis of more than 400 kindergarten children, in an effort to try to identify—prior to the onset of formal teaching—who would likely fail at reading. (An earlier, less ambitious version of the book had been published in 1966.) Their data led them to conclude that the best approach to early identification was a quick screener to pinpoint which children would struggle, and then a more extensive battery of diagnostic tests (covering a wide range of physical, cognitive, and perceptual variables) to explore the patterns of competencies that would guide instruction.

That effort was far from the last word on the subject and today, I think it is fair to say, much of their scheme has been superseded. However, at least partly due to that work, there is now a clear mandate to figure out which children are likely to struggle—and to do so prior to the onset of that struggle. Unless reading problems can be prevented, or addressed successfully very early, there are likely to be damaging secondary problems (the students’ reactions and responses to their failures) that can only complicate eventual remediation.

These days we have many more variables available to us— variables that go well beyond anything Jansky and deHirsch could have hoped for, including genetic screenings and various kinds of brain scans. Nevertheless, we are still confronted by many of the same problems that their work uncovered more than 50 years ago: the multivariate nature of reading difficulty, the complication of poor or inadequate teaching, false-negatives in prediction, and so on.

This issue of Perspectives on Language and Literacy provides a decidedly contemporary perspective on the early identification of reading difficulties. Mads Poulsen, a psycholinguist based in Copenhagen, Denmark, provides a thoughtful analysis of the need for accuracy in any early prediction model. Any scheme sensitive enough to reveal all students who will eventually struggle inevitably will result in “false-positives”— the misclassification of students with no need of extra learning support. And, schemes that minimize such misidentification will necessarily miss some of those in need. Professor Poulsen explains why that is and what is required to optimize early identification efforts so that they will have practical value.

In the 1930s “early identification” meant revealing those who had failed to learn to read after only a year or so of instruction. These days by early we tend to mean kindergarten. But what if it were possible to figure out who was going to suffer from reading disability years earlier than this? Recent advances in brain science suggest that this possibility may be more than a science fiction dream. Ola Ozernov-Palchik and John Gabrieli are neuroscientists who use brain imaging to identify the neural structures and functions that underlie reading development. Their work is pertinent to the issue of prediction of dyslexia because they explore neuroanatomy at a variety of ages, including infancy. Most studies of the neurological correlates of reading are conducted with already-struggling readers. From such studies it is impossible to discern which differences predate the failure to learn. Since learning to read changes the brain, there is a real need for pre-instruction neural exploration.

Then we explore a couple of practical pedagogical issues in the early identification of reading difficulty. David Kilpatrick, a clinical psychologist and author of the influential Essentials of Assessing, Preventing, and Overcoming Reading Difficulties, explores the role of causation in prediction and assessment schemes. His conclusion: once a reading problem emerges it doesn’t matter much what its etiology—since ultimately etiology cannot determine what assessments to use or which instructional interventions will work.

His discussion of the causes of reading problems made me think about the biggest gap in the prediction literature: No matter how incisively we measure those child factors that suggest future failure…such efforts cannot tell us anything about the instructional environment the student will have to learn within and respond to. Linda Siegel is the former Dorothy C. Lam Chair in Special Education and is editor-in-chief of Perspectives. In this issue, she elaborates on this conundrum, providing a case study of early identification and intervention and how it actually can work within the practicalities of a real school.

Finally, Hugh Catts and Yaacov Petscher, experts in the field of learning disabilities (the former a specialist in Speech, Language and Hearing and the latter a psychologist focused on reading), point us toward the future of early identification. They hypothesize that since reading development is undermined by multiple causal deficits, successful early identification schemes will need to be multifactorial in design and they argue for including computer assisted technology, gamification, and longitudinal models in the development of 21st century early identification efforts. This approach may seem to contradict David Kilpatrick’s claims about the current usefulness of causal explanations in the diagnosis and correction of reading difficulties; but remember, Kilpatrick is explaining the current state of the art in the field, while Catts and Petscher are imagining a future when we will surely know more. If they are correct, then it seems likely that early identification in 2030 will be as different from our 2018 concept as our current efforts are from those of the Jansky and deHirsch era.

Buckle your seat belts; it could be a bumpy—but fascinating and rewarding—ride.

Meanwhile, if I were a kindergarten teacher I’d screen my students early in the year to see what they knew about reading…particularly examining their knowledge of letter names and sounds, their phonological awareness, and awareness of print features (the kinds of skills that Kilpatrick describes). My focus would be on knowledge of literacy rather than on underlying causes or correlates. Although Ozernov-Palchik and Gabrieli and Catts and Petscher’s insights are exciting and hopeful, they are not yet user-ready. I’d implement daily lessons aimed at teaching these early literacy skills, monitoring student progress over the first semester. The screening information, although helpful, is not likely to be sufficiently predictive on its own (Poulsen), both because of the imperfections of testing and the variability evident in classroom environments (Siegel). Predictions based on children’s learning success during those early months improve prediction and are sufficiently accurate to allow for the implementation of intensive early interventions aimed at getting such children on track for success. I hope someday that the future research advances heralded in this issue will render my approach hopelessly outdated, but for now it is likely the best we can do.


Bennett, C. (1938). An inquiry into the genesis of poor reading. Bureau of Publications, Teachers College, Columbia University.

Dunklin, H. T. (1940). The prevention of failure in first grade reading. Teachers College Contributions to Education, 802, 1–111.

Hildreth, G. (1946). Speech defects and reading disability. The Elementary School Journal, 46, 326–332.

Jansky, J., & deHirsch, K. (1972). Preventing reading failure: Prediction, diagnosis, intervention. New York, NY: Harper & Row.

Yoakam, G. A. (1943). An ounce of prevention in reading difficulties. Journal of Consulting Psychology, 9, 125–131.



See what others have to say about this topic.

Ann C Feb 11, 2023 04:25 PM

Proponents of SOR often suggest that the instructional strategy needed by all struggling readers is systematic phonics instruction. In addition to that, I suggest methods which address the wide category of language comprehension as defined by Scarborough’s rope and the transfer of knowledge between reading and writing. I have found many struggling kindergarteners can approach literacy through composing and encoding, then transferring those skills to reading. While I do not dispute the need for phonics instruction, I often see children who are unable to transfer this information to problem solve in reading or writing without carefully planned instruction on transfer. They often can recite all the OG sound cards but cannot blend or segment phonemes. Despite this, the intervention is another round with memorizing letter names and sounds in a different room with a smaller group. Without language comprehension and transfer, I do not often observe significant progress with encoding and decoding.

Andrew Biemiller Feb 11, 2023 07:03 PM

I have a sneaking suspicion that vocabulary problems are often responsible for low reading "comprehension" problems. Addressing that problem is completely different than addressing reading/decoding problems.
Cheer, Andy Biemiller

Timothy Shanahan Feb 11, 2023 07:22 PM


The National Early Literacy Panel found that language measures in preschool and kindergarten provided a strong prediction of later reading ability. Vocabulary measures were predictive, though not especially so... the predictions improved greatly (and their results were strongly differentiated from decoding measures) when batteries of language measures were incorporated (vocabulary but vocabulary along with listening comprehension and/or grammar, etc.). There are a few possible explanations for this: (1) typical early vocabulary measures are simply not very good; (2) vocabulary is a rather imperfect/incomplete measure of language development and it is overall language development that is the real issue here; (3) multiple language measures did better than vocabulary alone because multiple measures tend to be more reliable and reliability is important in prediction equations. My personal bet? Vocabulary is important, but so is language development overall and we should be just as protective of what we do to support growth in those areas as we should be with decoding.



Jill Feb 11, 2023 08:18 PM

Would speech articulation deficits or hearing loss be considered a known predictor of reading difficulties that require early reading intervention or are these not 100% accurate either?

Lauren Feb 11, 2023 11:12 PM

I read that phonemic awareness is a developmental skill which develops between the ages of four and eight. If true, what are the implications of this for students who develop the skill later? Aren't they at an unfair disadvantage?

Janyce Collins Feb 12, 2023 12:53 AM

I am sorry if I seem petty. I can’t stop wondering how a teacher can have three spelling errors in such a short post. WHY do TEACHERS fail to self-edit or use spellcheck? How are our students going to learn if we haven’t?

Timothy Shanahan Feb 12, 2023 02:51 AM

Lauren — yes, most (but not all) children manage to develop PA in that period. It is disadvantageous to try to read without it (though at least some kids learn PA from learning to read). In any event, that’s why it is so worthwhile to teach PA in preschool, kindergarten and early first grade. It is a correctable disadvantage.


Sebastian Wren Feb 12, 2023 06:24 PM

In our "high-dosage tutoring" program for children in K-2 (Literacy First), we see that about 70% of the struggling students we bring into our program do not struggle for long. They respond to daily, targeted instruction, and very quickly learn the skills they need to catch up with their peers. With about 30% of the students, however, their response to intervention is not as clear. Some have persistent problems, even after weeks or even months of targeted daily instruction (1-on-1). These are the students I would describe as dyslexic. The screener does not tell us much -- only that the student has not yet learned essential early reading skills. Their response to high-quality instruction is much more meaningful and informative. In Texas, we are over-burdening our dyslexia specialists with students who really don't need that level of support. We should be using more than just screeners to determine which students need sustained dyslexia services.

Timothy Shanahan Feb 12, 2023 08:38 PM


You really are underlining two important problems with screening. First, there are the false positives --kids who are really doing fine, but whom the tests indicate a problem with. Second, there are what I'll call true positives... kids who really are low in one or another aspect of literacy, but whose deficit is due mainly to lack of exposure. For instance, the kid who comes to school knowing no letters, but who picks that information up quickly and easily in the classroom (it wasn't a gap in learning ability but in exposure, and once exposure was there, the problem was gone). Neither of these groups of kids require remedial instruction -- they just need reasonable quality classroom teaching of the regular daily curriculum. These problems, as you point out, can overwhelm the system, becoming one of the reasons why early assessment isn't paying off (there aren't enough resources to sufficiently address these students' problems).


Mark Shinn Feb 13, 2023 03:09 PM

When the base rate is 0 on so many measures/dimensions/constructs simply based on lack of experience it seem futile to try to differentiate among young children when that energy could be spent in meaningful—and fun—“instruction” in oral language, print and print experience, early reading, stories, rhymes, sounds, letters, number, etc. Really good, well-designed, sequenced, instruction with carefully selected examples so kids can be successful and have fun and learn. Let good curriculum and instruction with appropriately intensive early intervention serve as the screener for young kids. And regular universal growth and development progress monitoring. People forget that that was what Benchmarking was intended to be, not multiple times screening, year after year after year after year. Well checks, like Dr. Visits so educators and parents could help when kids were not developing at the expected rate. A growth model. Some things CAN be detected very early. Some things take time. We all want to help as soon as we can. Sometimes, the best help is a really good teacher with really good instructional tools at the door! IMO

Dr. Bill Conrad Feb 13, 2023 04:07 PM

A well informed and researched article.

To what extent does a failed Balanced Reading curriculum, pedagogy, and assessment system obfuscate the identification of students with reading difficulties?

Could it be that the inability of the adults to teach reading properly in the core curriculum creates “student reading difficulties?”

Timothy Shanahan Feb 13, 2023 04:54 PM


There are many things that can interfere with and make difficult learning to read, including things like cognitive and sensory deficits/ disabilities, genetic proclivities, poverty, racism, neglect, extended illness, and other conditions that lead to lowered opportunities, and, yes indeed, poor teaching, inadequate curriculum, and other problems with the educational environment. None of those alone is usually sufficient to hold a student down, but as those problems pile up kids tend to do worse and worse. A truly balanced reading curriculum would probably not be disabling -- but since no one agrees on what needs to be balanced in such a curriculum, many students were receiving little or no explicit teaching in many of the skills and abilities needed for reading. That, of course, could be terribly disruptive, especially for kids whose households were unlikely to pick up the slack.


Timothy Shanahan Feb 13, 2023 05:40 PM


Speech articulation deficits and hearing loss are often implicated in reading difficulties. However, that a student had either or both of these challenges would not necessarily indicate a need for additional or specialized reading instruction. Research for example finds that speech interventions aimed at improving articulation tend to be quite effective (at improving articulation) but rarely have any positive impact on reading performance. Kindergarten and first-grade teachers should be on the lookout for such children and should refer them to speech and language services as appropriate. Additionally, because these problems can interfere with reading, it would be especially wise to evaluate these students' reading skills and to monitor their progress, with the purpose of not allowing them to fall behind. But if they have these problems and are meeting reasonable criteria on sound reading assessments, nothing more is needed from the reading teacher than the usual quality reading instruction they are expected to provide to everyone.


Jo Anne Gross Feb 17, 2023 04:49 PM

Mark Shinn makes a good point.
Regular Screening and Progress monitoring are two different things.

Miriam Trehearne Feb 19, 2023 10:18 PM

Miriam P. Trehearne February 19, 2023

Dr. Shanahan,

Thank you for your recent blog posting regarding early literacy and dyslexia.
Yes, educators have lots of early literacy screening instruments. But I agree with you, the key to effective early literacy teaching and learning is the successful linking of assessments to research-based and proven instructional practices. That is really the only reason for doing the assessments. Yes, we need to focus on the students’ knowledge of literacy not on causal explanations. In fact, school districts need to focus on prevention of reading problems, starting in kindergarten, more than intervention or remediation. I highly recommend Marie Clay’s seminal work, Learning to be Learning Disabled, where she describes many struggling readers as curriculum casualties, the products of instructional programs which are not responsive to students’ individual instructional needs. To quote Linda Darling-Hammond, teachers can’t teach what they don’t know.

When I was working as a Program Specialist with Special Needs students, I remember a family demanding that their daughter in second grade be labelled dyslexic. They had been told (not by the teacher) that although she was struggling with literacy, they should not worry, because given time she would catch-up. We now know that generally the “gift of time” is not a gift at all. Kindergarten teachers can predict at the end of the kindergarten year where most of the children will be in literacy learning by the end of Grade 1 (Allington, 2011). Connie Juel’s research indicates that there is an 88 percent probability that a student who is a poor reader at the end of grade one will remain a poor reader at the end of grade four. In addition, the research of John Pikulski and others indicates “there is very little evidence that programs designed to correct reading problems beyond second grade are successful.

I agree with you and Dr. Biemiller, (whose work in this area has also had a huge positive effect on teacher knowledge) that vocabulary is important, but so is language development overall and we should be just as protective of what we do to support growth in those areas as we should be with decoding. Unfortunately, this is not the case. Oral language (word and world knowledge) is a curriculum gap area.

Linda Siegel Mar 03, 2023 11:25 PM

Although prediction of reading difficulties is an interesting exercise, attempts at prediction without intervention seem to be merely an academic exercise. I think that we want to combine prediction (really an assessment of critical skills) with intervention. All children who are struggling with reading should receive an appropriate intervention (e.g., structured literacy). Actually we found that systematic phonological awareness and phonics training improved the reading skills of all children, not just struggling readers.

We need good classroom (Tier 1) instruction for all children and progress monitoring for all children.

What Are your thoughts?

Leave me a comment and I would like to have a discussion with you!

Comment *

Early Identification: Predicting Reading Disabilities and Dyslexia


One of the world’s premier literacy educators.

He studies reading and writing across all ages and abilities. Feel free to contact him.