Richard AllingtonKey Idea
State University of New York-Albany, Albany, New York, USA
Board Member, International Reading Association

Transcript from Critical Balances: Early Instruction for Lifelong Reading

DR. ALLINGTON: In one sense, part of what's going on across the country is a debate, the debate about, in one sense, a philosophy of education and a philosophy of teaching and learning and so on.

And if you are interested, I'll just recommend an article in this month's Teachers College Record by Dave Berliner in which he sort of tries to -- to analyze the different notions of what it means to learn and what it means to read and what it means to teach and how different factions literally talk right past each other. It's a powerful statement.

But I'm going to talk more just briefly today about a little framework that I -- that I have been trying to develop, and it has to do with the notion of research-based claims. And I was pushed to do this in part by this conference and also by Jay Goldman, who is the editor of the School Administrator, had asked me to do a piece for the fall issue on research-based claims in reading. And he gave me initially 750 words to do it in, only he let me take 1,500, actually 1,481. That's without editing, however.

He was concerned about it because he said suddenly, at least in the last few years, there's been a ground swell in their advertising copy that's coming into the School Administrator and into the booths at the AASA (American Association of School Administrators) conference and so on, as well as in some of the presentations about research claims: "This is research based," "This is based in scientific research," and "This is based in empirical research," and so on. And he said, you know, it looks like almost everybody has invented the best way to teach reading. Is there a way that you could perhaps give a plain language summary for how we might sort through some of this? I came up with four criteria that folks might use. (More on FAQ6)

The first one's called the convergence of evidence criteria. (More on FAQ6) A series of studies by one investigator or one research team does not a generalization make. But, instead, when we look at research that's been done in multiple sites by multiple investigators and we start noticing that in fact people are coming up with very similar findings, particularly when they're using multiple measures, I think we can argue there is a convergence of the evidence. And I'm going to talk a little bit today about some of the areas of beginning reading where I think there is a convergence and some where I think the people would like to tell us there is a convergence but there's not.

The second criteria I used was what I call the quality of evidence criteria. (More on FAQ6) The quality -- the primary criterion I proposed was the peer and publication in one of the five journals, five journals that routinely produce or publish rigorous, empirical, as well as I suppose nonempirical -- no -- rigorous investigations of -- of reading research, literacy acquisition, and so on. And those are Reading Research Quarterly, Journal of Literacy Research, American Educational Research Journal --

DR. HOFFMAN: Journal of Ed Psych (Educational Psychology).

DR. ALLINGTON: -- Journal of Ed Psych, thank you, and the Journal of Educational Research. And I said I don't think that those are the only five, but I can tell you right now as someone who has been working at a university and working in two different federal research centers for 25 years, everybody is dying to try to get their work published there. If it doesn't get published there, then we usually go to another journal, which doesn't mean that I -- my point simply is that all journals are not equal.

I think what I find most amazing is the number of products for which there are research claims in which there is no published research evidence in any journal, but instead series of what are called site reports, evaluation studies, unpublished dissertations, unpublished monographs. There's a reason they're unpublished.

The third criteria, the comparability criteria, refers to comparability among individuals and school districts. (More on FAQ6) What does the evidence suggest? What kind of implementation did we have? What kind of demographics were in place? Perhaps even what historical era was the study done in? Is a study from 1965 relevant in 1997? It might be, but let's look at the comparability, what kind of kids, what kind of teachers, what kind of classrooms, what kind of communities, what kind of support and so on.

The final one is the one I call compellingness. (More on FAQ6) And the compellingness criteria in one sense, I suppose, is sort of a seat of the pants. It's the most qualitative of all of the judgments, but I -- I put it in because in my -- very honestly, as I read, I start asking myself, is this compelling? Even when I have met the previous three criteria, is this compelling? Am I convinced?

Let me talk now just a little bit about the research in reading and beginning reading in particular. And some of the research that I think has been ignored is what I am going to focus on.

We have three other folks behind me, so there is no reason to try to cover the whole research field. I would pick the stuff that interests me most.

I would suggest that one of the areas in which we have converging evidence on an international scale has already been mentioned by Ms. Allen, and that's that not only are the kids in Texas doing pretty well, but so are the kids across the U.S. (More on FAQ7) In fact, you know, the International Education Association's latest study in 1992 ranked the U.S. fourth graders second in the world. I mean, it's literally the only comparison, international comparison, in which I think we can honestly say U.S. kids met world-class standards.

This recent issue of Educational Researcher had an article by Stedman and another response by Gerald Bracey discussing the fact that that particular study is -- is largely unknown. That meant Bracey says that he's asked audiences over the last two or three years to raise their hand how many people have heard of this study. He gets one hand at about every other meeting.

I have to say that I have been using an excerpt this year out of the local Albany paper that came out last June '96 that was a three-inch column on page 6 that -- that reports the results of that study. It also reports in one paragraph, the second paragraph, is a quote from (Education) Secretary Riley that says this, of course, doesn't mean we're where we want to be and then goes on to point out that parents need to do more, which is another whole day we could spend. Since we have fewer parents with fewer hours these days, when are they supposed to do more and what are they supposed to do it with?

But if you look at the work that the IEA reports that are available through the publications office of the International Reading Association, the report by Warwick Elley and the other by Neil Postelwait, you'll see that there not only is achievement high in the U.S., particularly in the elementary grades, but it points to a couple of other areas that I think there is converging evidence also. One is the importance of -- of sheer opportunity to read and write. (More on FAQ2)

I'll simply argue that the data from the IEA, the data from the National Assessment of Educational Progress, I mean, enormously large-scale databases, as well as the data from some sort of classic studies like Linehart, Sigman, and Cooley back in the 1980s, sort of, again, the heyday of the old skills-based diagnostics prescriptive model, that point to the incredible power or the production function of engaging in reading and writing for improved reading and writing achievement.

I'll suggest also that there is an enormous convergence of evidence. We look at the work of Mike Knapp more recently on teaching for meaning in high poverty schools, his Teachers College Press work. That was a large federal study that they completed about two years ago out of Far West Labs. (Return to PI)

A distinguishing characteristic, I mean literally the first one that jumped out of the data, was the enormous amount of time that kids in these high-achieving, high-poverty schools spent reading and writing, talking, and thinking, as opposed to the kids in the low-achieving schools who spent lots of time, as they did in the Linehart, Sigman, and Cooley study, being drilled, doing dittos, and playing, you know, go-fish phonics games or go-fish sight word games or hidden word search puzzles or vocabulary volcano activities and so on.

I mean, we can go back even further, you know, to the studies that Bert Harris and Al Sewer and others did, and the same findings come out: Kids need to read a lot, they need to write a lot. I think there is no arguing that the evidence has converged to suggest that kids need enormous amounts of time. (More on FAQ2)

I think that the evidence suggests that our middle school and high school kids aren't doing as well internationally or even nationally as the elementary school kids. The first thing I'd point to would be to ask how much they read, because it looks from the national assessment data that the high school kids are reading half as much as the fourth graders and writing half as much, which is an oddity in the sense that as the kids get more and more skilled at things, we seem to expect less and less. It seems to me the higher the grade level the more likely we are to find a teacher who thinks that they don't have the time to waste by having kids read or write.

One of the things that I will argue, there also is, I think, a fair amount of converging evidence that teachers do teach decoding skills, they do teach phonics skills, and so on. (More on FAQ7) We have this evidence in a variety of both recent surveys, large scale, you know, the work that Mike Pressley and Joan Rankin and Linda Yokoi reported in The Elementary School Journal last year, the -- and more recently in the new book that McIntyre and Pressley have out from Christopher Gordon (Return to PI) on strategy instruction, the work that -- that Jim Baumann and Hoffman and the crew from the Reading Research Center have done on their large-scale survey of -- of how teachers teach, sort of a replication of the first-grade studies from the 1960s which found that teachers were teaching phonics in the 1960s. (Return to PI)

I mean, in the Pressley study, you're talking about 95 percent of the primary teachers reporting, "Yes, we teach phonics." If you look at the recent, most recent Hoffman and McCarthy studies of schools here in Texas, of the study from how teachers taught in the last year of using the old basals and the first year of using the new basals, teachers didn't change much. You know, teachers taught phonics, lots of them. Lots of them taught phonics over and above the phonics that was in either basal series, the older skills-based and the newer language- or literature-based series.

In other words, we have, I think, a variety of evidence. And going back, we could pull out the first-grade studies from the 1960s that showed that teachers were teaching phonics.

Now, I mean the question, I suppose, that there seem to be lots of assumptions that nobody teaches phonics anymore. One of the points that was raised at the American Educational Research on California was that, the working assumption being presented by the speakers seemed to be that in 1987 California adopted a new framework and adopted a new set of basals that were literature based, and all teachers changed the way they taught and started immediately not doing anything like they had done before.

A speaker from Great Britain got up and asked the question to Dr. Hoenig, "You know, does this -- it sounds like you're suggesting that these teachers changed dramatically and that runs contrary to all of the evidence on implementation in teacher change. Do you have any evidence?"

And Dr. Hoenig sort of said, well, ya-da-da-da-da-da and then Dr. Treadway said, well, ya-da-da-da-da-da, and then finally the question came, so the short answer is "no." And they said basically, yeah, we don't have any evidence on how teachers taught. No one studied how teachers taught in California after the new adoption.

In many respects, one of the points that Hoffman and McCarthy make in their recent report is almost nobody has studied how reading is taught in Texas. Almost no one has studied how basals or curriculum frameworks influence the way teachers teach.

There are lots of assertions out there on both sides that basals and curriculum materials are a form of mind control that enslave teachers, but the evidence doesn't support it. The evidence doesn't also -- also could be used on the other end to suggest that picking a new basal, picking a new framework, is going to have modest effects, if any, on how teachers teach in Texas or anywhere else.

There are a number of assumptions, and I would point to the work of my mentor, Jerry Duffy, and particularly his most recent paper in the -- in a book by Steve Stahl and crew from Georgia, David Hays, called Instructional Models in Reading that's published by Earlbaum. Duffy's paper is called "Powerful Models are Powerful Teachers." (Return to PI)

Jerry has now just retired from university life, and in this paper he sort of reflects on what the last 25 years at the Institutes of Research and Teaching and his work on teachers and strategy instruction and how to help teachers learn to teach reading better and so on, and he comes to basically the same conclusion -- that it's teachers that matter. (More on FAQ5)

I mean, that was the conclusion of the first-grade studies. Curriculum didn't matter much, teachers matter; that almost any kind of curriculum that was put in place worked for some people and didn't work for others; (Back to FAQ5) (Return to PI) that there were implementations districtwide where the same -- Program A did wonderfully in this district and terribly in this district; that the difference between the best and the worst implementations of -- of the same program were always miles wider than the average differences between programs implemented across a wide variety of sites.

I think there is converging evidence that kids need to develop strong independent decoding skills. I don't think there is a convergence of evidence on how to best do that. (More on FAQ1) And, in fact, one of the things that sort of troubles me is that we're getting what I will argue is a false dichotomy sort of waved in front of us. And the false dichotomy that's being waved in front of us sort of contrasts or pits direct, systematic phonics versus incidental, opportunistic phonics.

And I'll suggest that the work that Mike Pressley has been doing, the work that Duffy has done, the work that Hoffman and McCarthy have been doing, and a variety of other folks, including the study that Mike Pressley and I are now running with sites here in Texas, sites in California, sites in Wisconsin, New York, and New Jersey, observing teachers, first-grade teachers in classrooms suggests that the very best teachers, the ones that have been nominated as exemplary, almost always teach in a way that I would call planful, direct, and opportunistic. (Back to FAQ2)

If we look at the Pressley survey, you know, less than a quarter of the teachers report using a commercial decoding curriculum, of these exemplary teachers. Almost all of them report teaching, you know, phonics and decoding strategies every day, but they teach it in the course of reading and writing activity.

We see this as an argument between decontextualized and contextualized, you know, teaching of decoding. And often the direct and systematic is sort of argued as decontextualized.

I think the difference also that I would argue is that phonics as a strategy versus phonics as knowledge. I mean, a fundamental difference in some of the data that's being put out is "Does this kid use the phonics knowledge that he has effectively in an integrated strategy, you know, to produce meaning from text?" There's a difference between effective, strategic use of phonics knowledge and pronouncing nonsense syllables. (More on FAQ6) (Return to PI)

I'll finally argue that perhaps what we really see is the difference between novice and expert teachers. In other words, we see teachers who are very expert about kids, about the first grade, about teaching reading and writing, and they invariably teach quite differently than teachers that aren't very expert. And teachers that aren't very expert may in fact find themselves enslaved by materials. They may find themselves doing decontextualized, unsystematic kinds of instruction.

But it would be an argument, I think, from the weight of the evidence, the converging evidence, that we need to invest more in teacher expertise, not necessarily more in curriculum materials or frameworks. (Back to FAQ8) We start thinking about the notion of -- of what it means to -- to be an effective first-grade teacher, effective at the beginning.

I can say I think there is convergence that suggests, the data that suggests, phonemic awareness and developing it in kids is important. I don't think there is a shred of evidence that we have any convergence of evidence how that might best be done. (More on FAQ1) We've got data that suggests that you can do it at four and five by playing a lot of word games with -- with small groups of kids. We've got some data that suggests you could do whole-class stuff. You've got some data that suggests one-to-one is pretty powerful, some that suggests that doing it with drill and Elkonin boxes, and others suggest doing it with invented spelling and writing activities.

In other words, I think that the conclusion one can draw from the research is that there is no adequate research base, no converging evidence that suggests that -- that we know how best to teach phonemic awareness.

I'd suggest that the convergence of evidence also says about 80 percent of the kids develop it (phonemic awareness) by the middle of first grade right now in the classrooms we've got, taught by teachers who have never heard the word. (Back to FAQ7)

Is the best way to solve the problem of the other 20 percent or 15 percent, depending on -- on which study you look at going back into the '70s, is the best way to do that to implement and change kindergarten and first-grade instruction systemwide statewide, or is it to try to be a little bit more, you know, notion of a surgical strip and target the kids who seem to have difficulty in the classroom context?

I'll simply suggest that we don't have evidence to suggest which way is the most powerful, most effective. We start looking at how to intervene to teach kids effective decoding skills, we don't have a lot of convergence. We start looking at the -- the evidence from a variety of folks.

I'll just leave you with a couple of quotes here. The first one is from Jeanne Chall:

Imposing a set systematic phonics program on a teacher who is knowledgeable about reading and keenly attuned to the strengths and weaknesses of her pupils may very well destroy the beauty of what she has already achieved. A beginning code emphasis program will not cure all reading ills, it cannot guarantee all children will learn to read easily, nor have the results of meaning emphasis programs been so disastrous that all academic and emotional failure can be blamed on them, as some proponents and publishers of new code emphasis programs proclaim.
This is 1968.

I'll suggest that even with the lack of agreement on how to go about teaching phonics that what we do know is that traditional phonics programs don't represent an advance. (Back to FAQ3) From Marilyn Adam's book, she argues simply, "The basic phonics curriculum is inherently intractable, slow, inefficient, and worse, except for students who essentially know how to read before it's begun, it's likely to be ineffective."

In other words, she's arguing here that filling in short vowel dittos doesn't teach kids to read, that the only kids who can do it are the ones who can already read, and they don't need it -- and the ones who can't do it can't learn from it.

Pat Cunningham, in her review, her review for the National Reading Conference research session on phonics, which is reprinted in the back of her book Phonics They Use, concludes at the very end, in the last paragraph, "The kinds of phonics instruction we need and which we should advocate is not the old phonics. It's not rules and jargon in a worksheet." And she makes a strong case using her work and the work of Linnae Ehri and Irene Gaskins and others for a decoding- by-analogy approach not a decoding-by-sounding in the sense of a letter by letter left to right sounding activity.

What I am going to suggest is that the converging evidence suggests that it's important that kids do in fact learn to decode. It's important if they're going to learn to decode that they learn about phonemic awareness, but what we don't have, we don't have data that -- that provides us with much in the way of saying this is exactly how we should do it. (Back to FAQ1)

The question we were asked to talk to is: "What should policymakers pay attention to?" And I'm going to close this off with one last overhead here. And I decided that, you know, policymakers ask questions like this: "Will smaller class size produce higher achievement?" And the answer they want is "yes, no."

Researchers, as Ernie House has pointed out, talk this way. Researchers say, well, "If smaller class sizes are met in a rationale manner, and the teachers act upon the reduced class size to deliver more personalized instruction, then you could expect higher achievement, but. . ."

And in one sense that's part of the problem we're faced with, that people are trying to take research in which "if," "and," "then," "but's" are the only way we can talk about it and turn it into "yes/no's." We can't turn a lot of the research into "yes/no's."

I'll simply argue that the NICHD studies, which have gotten a lot of press recently, simply have exaggerated claims for what they found, exaggerated claims for the kinds of effects and for what should be done in terms of American reading instruction. These exaggerations are just overwhelmingly obvious when you sit down and read the studies. They have not independently manipulated important variables like decodable texts. They have not independently manipulated variables like direct and systematic phonics instruction and so on.

I've got a new pin and I'm going to start marketing them. I don't have them today, but you can order them.

It's called "Show Me the Research." And what I am going to suggest is that everybody in this room may be better off to actually start reading the research, to actually start reading it and asking, what does it tell us?

Connie Juel and Diane Roper Schneider's study from the 1985 Reading Research Quarterly has often been given to me as a reference for the decodable text. Quoting that article, "The interpretation of the results of this study do not constitute advocacy of any one specific approach to beginning reading instruction. In particular, the results should not be interpreted as supporting an explicit phonics method since induction of learning of letter sounds," and so on.

It seems when you go back and read the studies people are much more careful when they write than sometimes when they talk. Sometimes people are more careful, and the original author is more careful, than people who are using the original author's work to advance a particular form of evidence as evidence or a particular approach or recommendation. (Back to FAQ6)

We've got three other people that are going to come up here and talk a little bit. I will simply suggest that we can start looking at converging evidence, but there are radical claims being made today that can't be supported by the research.

And we can't -- I mean, if we just sit back and say, oh, well, the research says, and we don't go out and do our homework, then -- then we're the ones to blame.

Thank you.

Return to Main Speakers


Reading Online, www.readingonline.org
Posted October 1997
© 1997-2000 International Reading Association, Inc. ISSN 1096-1232