Turning Decoders Into Readers

by Robert Pondiscio
June 9th, 2011

I’m a fan of PBS’s John Merrow.  He is the rare television journalist—perhaps the only one—who has the interest, background and sufficient airtime to give thorny education topics the nuanced treatment they deserve.  The other night he devoted nearly ten minutes of PBS’s Newshour to an intriguing question:  Can a good school have bad test scores?  To answer it, he and his producer Cat McGrath visited P.S. 1, a South Bronx school that appears to be filled with bright, eager learners and devoted teachers, yet is “failing” as judged by its terrible reading scores.  What’s going on?

“We discovered that the FIRST graders at that school were reading confidently and competently, Merrow writes on his blog, “but the fourth graders weren’t according to the results of the state test. Is this a paradox, or a full-blown contradiction?”  Merrow attempted to figure out where things leave the rails between first and fourth grades–an earnest, but ultimately frustrating piece that correctly diagnoses the problem, but fails to uncover or sufficiently examine its root causes.

Merrow starts by correctly pointing out that there is a big difference between “reading” in the first grade and “reading” in fourth grade.  Indeed, they’re hardly the same activity. Observing a phonics lesson in a first grade classroom, he points out that “Ms. Hunt’s students seem to be getting it. What they are doing is called decoding, but decoding is only half the battle. Understanding what the words mean is a much harder skill called comprehension. It’s where many children fall flat.”

For starters comprehension is not a “skill” at all.  Your ability to read with comprehension depends on many things.  You must be able to decode.  You must know all (or nearly all) of the words.  And you must know at least a little about the subject matter of the text to construct a mental model that allows you to make meaning correctly.   “My dog is sleeping on the couch” is easy to understand; “ My Havanese is snoozing on the divan” means more or less the same thing as long as you know about dogs and furniture, and understand that “snooze” is a synonym for sleep. 

Only 18% of P.S 1′s 4th graders are reading at or above grade level.  The good decoders have failed to become strong readers. What happened?  One 4th grade teacher says the children’s home lives start to take a toll.

“They’re not as innocent anymore. They’re realizing the things that are affecting their schoolwork. You know, I mean, I have homeless students in my room. I have students with fathers in jail. There’s drugs. So, that obviously comes into play at a certain point as well.”

Another 4th grade teacher suggests the grind of test prep and test anxiety is the issue.  “The system takes the fun out of reading,” observes Brenda Cartagena.  “I want them to read for enjoyment. I want them to grab that book because it’s fun. I tell them, reading, you travel, you meet new friends, you learn how to do new things. But it’s very difficult, you know? They take the joy out. And it’s hard to infuse it back.”

Full disclosure: I spent a significant amount of time talking to producer McGrath as she and Merrow prepared the piece.  I stressed the importance of vocabulary and background knowledge and how reading comprehension, unlike decoding, is not a transferable “skill” at all.  How the tests children take are de facto tests of general knowledge.  To what degree, I wondered, does the instruction these South Bronx kids receive reflect that?  Having taught at a nearby school in the same district a few blocks away, I suspected the answer is “not at all.”   To their credit Merrow and McGrath looked at the tests.  Merrow writes on his blog:

“We looked over past tests, and, sure enough, the passages were about subjects that poor kids in the south Bronx may not be familiar with (cicadas or dragonflies were two of the subjects, for example). Answering the questions did require inferential leaps, just as we had been told.

“So we asked to talk with a couple of fourth graders who were reading below grade level, and here’s where it got complicated.  As you will see in the NewsHour piece, both children, one age 9 and the other 11, handled the passages and answered all the questions. Maybe the personal attention helped, but they read easily and drew inferences correctly. We only ‘tested’ a couple of kids, but both were below grade-level, their teacher assured us.”

Again, did they “read easily?”  Or did they decode easily?  And I’m not as confident as Merrow that they “drew inferences correctly.”  Here’s what viewers saw Monday night on the Newshour:

JOHN MERROW: I wondered how the fourth-grade class might perform on the state test this year, and asked Ms. Cartagena to send me two of her students who were reading below great level.

Jeannette, who is 9, came first.

STUDENT: So far, I have hoped to find many new species.

JOHN MERROW: I asked her to read a passage about dragonflies from last year’s state test.

STUDENT: About 5,500 dragonfly species buzz around the world. Who doesn’t like — like looking at these amazing insects?

JOHN MERROW: What are species?


JOHN MERROW: Kinds. It’s kinds of species. Right. Exactly. Yes.

Exactly right?  It is impossible to know, based on this exchange, if the child understands “species” as well as Merrow assumes or if she has a sufficient grasp of what a dragonfly is to apply the concept.  As a teacher, I’d want to probe more for understanding, “if you’re looking at two dragonflies, how can you tell if they are different species?” you might ask.  If she said they might be different colors or have different shaped wings, I’d feel reasonably confident that she understands the basic idea.  If she says “one’s male and one’s female” or can’t explain the difference at all, then the concept is still shaky, or she might not know enough about dragonflies to apply it. Either way it would impact her ability to draw inferences and make meaning from the passage.

Given that the achievement gap long predates test-driven accountability, you could sensibly argue that that testing makes the problem worse, but it cannot be the root cause.  Similarly, the idea that “real life catches up with kids” by 4th grade is unsatisfying.  If reading comprehension is a skill like riding a bike or throwing a ball through a hoop (it’s not), it is not an ability you would suddenly lose if your father was sent to prison or you were evicted from your home.

What Merrow either didn’t probe or did not air is what – what exactly – the instruction given to these children in 2nd, 3rd and 4th grade looks like.  Are they being steeped in a content-rich curriculum that would make it less likely that concepts like cicadas, dragonflies and species would be unfamiliar at test time?  Or is the school operating, as most do, on the incorrect  assumption that reading comprehension is a transferable skill?  That decoding + engagement + content-free reading strategies is enough to guarantee success? When this formula fails, as it inevitably must, it is normal to look to outside causes like poverty, fractured families and test anxiety as root causes.  These things certainly work against student engagement and achievement, but they are clearly not the cause of failure. 

Merrow is due a lot of credit for  taking a nuanced view of reading and and asking the right question: why doesn’t early decoding success automatically turn in to comprehension success?  But ultimately the piece doesn’t provide the answer.

There’s No Such Thing as a Reading Test

by Robert Pondiscio
June 16th, 2010

“Children who do not learn to read proficiently by the end of third grade are unlikely ever to read at grade level,” writes Sara Mead in the July/August issue of The American Prospect.  The issue features a special section titled “Reading By Grade Three” that examines the crisis in early childhood literacy.  In addition to Sara’s piece, which lays out the case for national action on early childhood literacy, Cornelia Grumman,  executive director of the First Five Years Fund, looks at the need to get kids off to a good start even before formal schooling even begins; the New America Foundation’s Lisa Guernsey cautions against letting the clear need for improved early literacy translate into classrooms that are all skills and no play; Gordon MacInnes of The Century Foundation describes how providing low-income kids with “stable, high-quality preschool and kindergarten” has made a difference in New Jersey.  Lots more good reads; the whole package can be found here.   E.D. Hirsch and I contributed a piece as well, a version of which is below.  It looks at how a fundamental misconception about the nature of reading leads to mischief in how we teach and test it.   

There’s No Such Thing as a Reading Test
E.D. Hirsch Jr. and Robert Pondiscio

It is among the most common nightmares.  You dream of taking a test for which you are completely unprepared having never studied or even attended the course.  For millions of American schoolchildren, however, it is a nightmare from which they cannot wake, a Kafkaesque trial visited upon them each year when they are required by law to take a reading tests with little preparation.  Eyebrows are already being raised.  Not prepared!?  Why, preparing for reading tests has become more than just an annual ritual for schools.  It is practically their raison d’être!

Schools and teachers may indeed be making a Herculean effort to raise reading scores, but paradoxically these efforts do little to improve reading achievement and to prepare children for college, career and a lifetime of productive, engaged citizenship.  This wasted effort is not because, as many would have it, our teachers are lazy or of low quality.   Rather, too many of our schools labor under fundamental misconceptions about reading comprehension, how it works, how to improve it, and how to test it.

Reading is not a Skill

Reading, like riding a bike, is an ability we acquire as children and generally never lose.  Some of us are more confident on two wheels than others and some of us, we believe, are better readers than others.  We view reading ability as a broad, generalized skill that is easily measured and assessed.  We judge our schools and increasingly individual teachers based on their ability to improve the reading ability of our children.  When you think about your ability to read—if you think about it at all—the chances are good that you perceive it as not just a skill, but a readily transferable skill.  Once you learn how to read you can competently read a novel, a newspaper article, or the latest memo from corporate headquarters.  Reading is reading is reading.  Either you can do it, or you cannot. 

This view of reading is only partially correct. The ability to translate written symbols into sounds, commonly called “decoding,” is indeed a skill that can be taught and mastered.  This explains why you are able to “read” nonsense words such as “rigfap” or “churbit.”  Once a child masters “letter-sound correspondence,” or phonics, we might say she can “read” since she can reproduce the sounds represented by written language.  But clearly there’s more to reading making sounds.  To be fully literate is to have the communicative power of language at your command—to read, write, listen and speak with understanding.  As nearly any elementary school teacher can attest, it is possible to decode skillfully yet struggle with comprehension.  And reading comprehension, the ability to extract meaning from text, is not transferable. 

Cognitive scientists describe comprehension as “domain specific.”  If a baseball fan reads “A-Rod hit into a 6-4-3 double play to end the game” he needs not another word to understand that the New York Yankees lost when Alex Rodriguez came up with a man on first base and one out; he hit a groundball to the shortstop, who threw to the second baseman, who relayed to first in time to catch Rodriguez for the final out.  If you’ve never heard of A-Rod or a 6-4-3 double play and cannot reconstruct the game situation, you are not a poor reader.  You merely lack the domain specific knowledge of baseball to fill in the gaps.  

Even simple texts, like the ones our children read on their all-important reading tests, are filled with gaps—presumed domain knowledge—that the writer assumes the reader knows.  Research also tells us that familiarity with domain knowledge increases fluency, broadens vocabulary (you can pick up words in context), and enables deeper reading and listening comprehension.

A simple model, then, would be to think of reading as a two-lock box, requiring two keys to open. The first key is decoding skills. The second key is oral language, vocabulary and domain-specific or background knowledge sufficient to understand what is being decoded.   Even this simple understanding of reading enables us to see that the very idea of an abstract skill called “reading comprehension” is ill-informed.  Yet most U.S. schools teach reading as if both decoding and comprehension are transferable skills (more on that in a moment). Worse, we test our children’s reading ability without regard to whether or not we have given them the requisite background knowledge they need to be successful.    

Who is a “good reader?”

Researchers have consistently demonstrated that in order to understand what you’re reading, you need to know something about the subject matter.   Students who are identified as “poor readers” often comprehend with relative ease when asked to read passages on familiar subjects, outperforming even “good readers” who lack relevant background knowledge.  One well-known study looked at junior high school students judged to be either good or poor readers in terms of their ability to decode or read aloud fluently. Some knew a lot of about baseball, while others knew little. The children read a passage written at an early 5th-grade reading level, describing the action in a game. As they read, they were asked to move models of ballplayers around a replica baseball diamond to illustrate the action in the passage. If reading comprehension was a transferable skill that could be taught, practiced and mastered then the students who were “good” readers should have had no trouble outperforming the “poor” readers.  In fact (and perhaps intuitively) just the opposite happened.  Poor readers with high content knowledge outperformed good readers with low content knowledge.  Such findings should challenge our very idea of who is or is not good reader: if reading is the means by which we receive ideas and information, then the good reader is one who best understands the author’s words.

You have probably felt the uncomfortable sensation of feeling like a poor reader when struggling to understand a new product warranty, directions for installing a computer operating system, or some other piece of writing where your lack of background knowledge left you feeling out of your depth.  Your rate of reading slows.  You find yourself repeating sentences over and over to make sure you understand.  If this happens only rarely to you, it is because you possess a broad range of background knowledge—the more you know, the more you are able to communicate and comprehend. The implications of this insight for teaching children to read should be obvious: The more domain knowledge our children receive the more capable readers they will become.

The message has not reached American classrooms, however.  A stubborn belief in reading comprehension as a transferable skill combined with the immense pressures of testing and accountability has created something like a perfect storm—ever more time is being wasted (and wasted is not too strong a word) on scattered, trivial and incoherent reading.   A study sponsored by the National Institute of Child Health and Human Development found that only four percent of 1st grade class time in American elementary schools is spent on science, and two percent on social studies.  In third grade, about five percent of class time goes to each of these subjects.  Meanwhile a whopping 62% in 1st grade and 47% in 3rd is spent on language arts. 

Most young American children spend anywhere from 90 minutes to two-and-a-half hours a day in something educators call the “literacy block,” an extended period which might include  reading aloud, small group “guided reading,” independent writing, and other activities aimed at increasing children’s verbal skills.  Reading instruction largely focuses on teaching and practicing all-purpose “reading comprehension strategies”—helping students to find the main idea of passage, make inferences or identify the author’s purpose.  The general idea is to arm young readers with a suite of all-purpose tricks and tips for thinking about reading that can be applied to any text the child may encounter.   Careful readers may be thinking, “if the ability to understand what you read is a function of your domain-specific background knowledge, then how is it possible to teach all-purpose reading strategies?”  It’s a question well worth asking.  And one that seldom is.  

Reading strategies figured prominently in the 2000 report of the National Reading Panel, and reading strategies work, to a point. Reading comprehension scores tend to go up after instruction in strategies, but it’s a one-time boost.  The major contribution of such instruction is to help beginning readers know that text, like speech, is supposed to make sense.  If someone says something you don’t understand, you can always ask that person to repeat, explain, or give an example.  Reading strategies offer similar workarounds for print.  They’re not useless, but repeated practice seems to have little or no effect.

“The mistaken idea that reading is a skill—learn to crack the code, practice comprehension strategies and you can read anything—may be the single biggest factor holding back reading achievement in the country,” wrote Daniel T. Willingham, Professor of Psychology at the University of Virginia, recently in the Washington Post.  “Students will not meet standards that way. The knowledge base problem must be solved.”

Tests Worth Teaching To

Once the connection between content and comprehension becomes clear, two conclusions come almost unbidden.  Our present system of testing reading ability is inherently unfair.  Since reading comprehension is not a transferable skill, unless we ensure that all children have access to the same body of knowledge, the student with the greater store of background knowledge will always have a strong advantage.  The second conclusion is even worse: the content-neutral way that we teach reading, as a discrete set of skills and strategies, is counterproductive and even irresponsible.

If our schools understood and acted upon the clear evidence that domain-specific content knowledge is foundational to literacy, reading instruction might look very different in our children’s classrooms.  Rather than idle away precious hours on trivial stories or randomly chosen nonfiction, reading, writing and listening instruction would be built into study of ancient civilizations in first grade, for example, Greek mythology in second , or the human body in third.  Recently, the Core Knowledge Foundation has been piloting precisely such a language arts program in a small number of schools in New York City and elsewhere.  Initial results are promising, however it is crucial to remember that building domain knowledge is a long-term proposition.  All reading tests are cumulative.  The measurable benefit of broad background knowledge can take years to reveal itself.

At present, teachers are tacitly discouraged from taking the long view.  Indeed, what incentive would a 2nd grade teacher have to emphasize content that might not show up on a test until 6th grade, if even then?  There is more upside for teachers in doing exactly what they chiefly do now – test prep, skills and strategies – unless we actively incentivize a domain-specific approach to language arts.

Let us propose a reasonable, simple, even elegant alternative to replace the vicious circle of narrowed curriculum and comprehension skills of limited efficacy, which over time depress reading achievement.  By tying the content of reading tests to specific curricular content, the circle becomes virtuous.   Here’s how it would work:  let’s say a state’s 4th grade science standards includes the circulatory system, atoms and molecules, electricity, Earth’s geologic layers and weather; Social Studies standards include world geography, Europe in the Middle Ages, the American Revolution and the U.S. Constitution, among other domains.  The state’s reading tests should include not just fiction and poetry, but nonfiction readings on those topics and others culled from those specific curriculum standards.  Teachers would still teach to the test, emphasizing domain specific knowledge (because it might be on the test), but no one would object, since it would help students not only pass the current year’s test, but to build the broad background knowledge that enables them to become stronger readers in general. 

The benefits of such “curriculum-based reading tests” would be many:  tests would be fairer, and a better reflection (teacher quality advocates take note) of how well a student had learned the particular year’s curriculum.  The tests would also exhibit “consequential validity” meaning they would actually improve education.  Instead of wasted hours of mind-numbing test prep and reading strategy lessons of limited value, the best test-taking strategy would be to spend time learning the material in the curriculum standards—a true virtuous circle.

By contrast let’s imagine what it is like to be a 4th grade boy in a struggling South Bronx elementary school, sitting for a high-stakes reading test.  If you do not pass, you are facing summer school or repeating the grade.  Because the school has large numbers of students below grade level, they have drastically cut back on science, social studies, art, music—even gym and recess—to focus on reading and math.  You have spent the year learning and practicing reading strategies.  Your teacher, worried about her performance, has relentlessly hammered test-taking strategies for months.

The test begins and the very first passage concerns the customs of the Dutch colony of New Amsterdam.  You do not know what a custom is; neither do you know who the Dutch were, or even what a colony is.  You have never heard of Amsterdam, old or new.  Certainly it’s never come up in class.  Without background knowledge you struggle with most of the passages on the test.  You never had a chance.   Meanwhile, across town, more affluent students take and pass the test with ease. They are no more bright or capable than you are, but because they have wider general knowledge—as students who come from advantaged backgrounds so often do—the test is not much of a challenge.  Those who think reading is a transferable skill and take their background knowledge for granted may well wonder what all the fuss is about.  Those kids and teachers in the Bronx struggle all year and fail to get ready for this? Why, all the answers are right there on the page!

It ends, as it inevitably must, in the finger pointing that plagues American education.  Do not blame the tests.  Taxpayers are entitled to know if the schools they support are any good, and reading tests, all things considered, are quite reliable.  Do not blame the test writers.  They have no idea what topics are being taught in school and their job is done when tests show certain technical characteristics. It is unfair to blame teachers, since they are mainly operating to the best of their ability using the methods in which they were trained.  And let’s certainly not blame the parents of our struggling young man in the South Bronx.  Is it unreasonable to assume that a child who dutifully goes to school every day will gain access to the same rich, enabling domains of knowledge that more affluent children take for granted?

It’s not unreasonable at all.  That’s what schools are supposed to be for.  The only thing unreasonable is our refusal to see reading for what it really is, and to teach and test reading accordingly.

Poor Speller? Blame Your G-E-N-E-S

by Robert Pondiscio
October 28th, 2008

Some people have a way with words.  Othurs not weigh haf.  According to the Times of London, it could just be your genes. 

In the past, poor spelling was attributed to all manner of things, from bad schooling to a lack of moral fibre. But science is offering a new explanation. A difficulty with spelling could be rooted in your genes and in the way that your brain is wired. These findings stem from research into the language disorder dyslexia, but they are proving important for the wider population. Biology, it seems, not only influences those with dyslexia but also people without the syndrome. If you are a bad speller you can blame your grey matter.

Simply deciphering the written word is the most complex task your brain will face says John Stein, Professor of Neuroscience at Oxford University Medical School, who notes that written language is a relatively recent invention.  “It was invented only 5,000 years ago, notes Stein.  “It is piggybacked on to our linguistic ability, which was invented 30,000-40,000 years ago.  The consequence is that many people fail to read or spell.”

Reading Strategies: A Little Goes a Long Way

by Dan Willingham
August 28th, 2008

Yesterday I argued that the knowledge readers bring to a text is essential to reading comprehension. But does even a knowledgeable reader comprehend automatically? Mustn’t the reader apply comprehension strategies to extract meaning from the text? The short answer is that teaching students comprehension strategies does help, but too much time is currently devoted to them.

Reading comprehension strategies include things like question generation (students are taught to generate questions about a text and then answer them) comprehension monitoring (students are taught to become aware of when they do not understand), and summarization (students are taught techniques to summarize meaning). Often, multiple strategies are taught.

The National Reading Panel  reviewed 205 studies examining the effectiveness of teaching students reading strategies, and there is little doubt that they help, and that the effect is sizable.

There are two aspects of the data which deserve special attention because they hold implications for classroom application. First, the effects of teaching students reading strategies are weak or absent before the third grade.  This finding is readily understandable—students are still learning to decode, and simply can’t juggle in mind the tasks of decoding, comprehending, and trying to implement a strategy. It’s only when decoding has become fluid so that the reader doesn’t need to think about it much that enough mental space is free to accommodate a strategy.

Second, when it comes to teaching students to use reading strategies, shorter programs seem just as effective as longer programs. This finding is crucial, because it ought to make us think differently about what reading strategies actually do. It’s natural to think that strategies improve the reader’s skill in extracting meaning from a text. But if that were true then more practice ought to make you better at it. Instead, comprehension strategies feel less like a skill and more like a trick—something like “check your work” in mathematics. It’s a very smart thing to do, and students should be explicitly taught to do it, but it doesn’t require extensive practice.

What might the trick of comprehension strategies be? A good guess is that they encourage students to think differently about reading. There is so much emphasis on decoding in early reading instruction (as there must be) that it is understandable that a student might think “If I’ve decoded, then I’ve read it.” But an adult knows that if you get to the bottom of a page and don’t know what you’ve read, you haven’t really read it, even if you’ve decoded everything. That conception of reading—that the point is communication—must click for students, and comprehension strategies may have most of their impact in getting students to think about reading as something they do to understand. Once they understand that, most of what comprehension strategies advise is something that students will do naturally: try to find the main idea, check your own comprehension, and so on.

The bottom line is that teaching comprehension strategies is a good idea, but it appears to be a one-time boost. There is no evidence that more practice yields more benefit. More information on this subject can be found here: http://www.aft.org/pubs-reports/american_educator/issues/winter06-07/CogSci.pdf