Philip GoughKey Idea
University of Texas, Austin, Texas, USA

Transcript from Critical Balances: Early Instruction for Lifelong Reading

DR. GOUGH: Good morning.

AUDIENCE: Good morning.

DR. GOUGH: I must begin by correcting Kathleen Jongsma, who inflated my reputation by referring to me as the Barbara Pierce Bush Regents Chair of Literacy.

This happened to me once before when, on the one occasion I met Governor Bush, he introduced me to a group as "the chair."

And I said, "I must correct you. I'm not a chair, I'm a professor."

And he said, "What's the difference?"

And I said, "$900,000."

We're still looking for the $900,000, by the way.

Let me begin by saying something about research. I hope you'll forgive me for reading my opening remarks, because I wanted to make sure that I had them exactly the way that I wanted them. And I'm afraid that on my feet I'm not as skilled as Dick Allington. I can't be assured that I will say the right words.

Now, my background is in experimental psychology. I'm what Morton Botel would have called an outsider. I was trained there in the scientific method, and that method is under attack from all sides these days.

Deconstructionists argue that the world is a text, and that since texts are anchored only in other texts, they may be assigned any meaning you choose. Social constructivists argue that all beliefs are socially constructed and that which belief is taken to be true is a matter of which group you belong to and what power it has. Cultural relativists argue that western science is simply our myth, no better than that of any other tribe. Feminists hold that science is nothing more than a tool to preserve male's hegemony and that they have their own ways of knowing.

Nowhere is this clash of paradigms more evident than in education. Hardly an issue of the monthly Educational Researcher appears without an article challenging the scientific method. In the most recent issue, for example, we are even asked, "Are our research epistemologies racially biased?"

I believe science's critics could not be more wrong. By any criterion you can name, the scientific method is the best way of investigating nature that our species has devised in 2,000 years. (More on FAQ6)

Perhaps some other alternative methodology will yield some new knowledge about reading, but I'll believe it when I see it. Much of what is called research in education these days would not pass muster in the eyes of scientists.

The fundamental idea of the scientific method is that if you wish to know whether X is related to Y, there is only one way to determine this. One must hold all else constant, either physically or statistically, vary X, and see if and how Y changes. (More on FAQ6)

The scientific method is not difficult; indeed, it might be called stringent common sense. Nor is it the property of any privileged class. Anyone can use it. To be sure, it is easier to apply within a laboratory, where I typically work, than in the classroom, but that does not mean we should turn away from it. On the contrary, since the empirical questions which beset us in education are so important, we should use every effort to do the best science we can to answer them. We need more, not less, rigor.

For example, Dick Allington criticized the research done here at the University of Houston by Foorman, Fletcher, and Francis. I agree with many of his criticisms, but what would the scientific response to this be?

I'm reminded of the fact that when George Pavlidis offered evidence that dyslexics cannot control their eye movements, every laboratory in America tried to replicate his results and failed. That's what I have been trying to persuade my friends on TAWL, Teachers Applying Whole Language, to do in the past year and a half, and it's what I would ask Dick to do. But I fear that the TAWLers believe that since they already know the truth, there is no point in such research. I believe they're wrong.

Now, what has the scientific method taught us about reading? Here I must turn to my overheads. What I would like you to do first is to consider what it has taught us about us about the skilled reader. Let me begin by showing you a quote from Marilyn Adams, "The single immutable and nonoptional fact about skillful reading is that it involves relatively complete processing of the individual letters of print."(More on FAQ1)

Ken Goodman does not believe this. In his 1993 Phonics Phacts, he replies, "Rather than simply stating how wrong I believe this nonoptional fact to be, I'm going to let you demonstrate that to yourself through the following simple experiment."

Well, let me give you the experiment. [Figure 1] A student of Ken's named Victor Frederick Golasch asked a group of college students to read this passage; I will give you a second to read it. If I now conceal it and ask you how many errors did you detect in the passage, what would you say?

There are actually six errors in the passage, and here they are. Let me show you the passage again and let you look. Notice there are two "the's," "boat" is misspelled, "though" instead of "through," so she had to take the boat "apart" should be one word, and she should "have" planned ahead rather than "of."

Well, we replicated his experiment and we found much as he did; namely, that most college students don't detect most of the errors in the message. But does that mean that the errors had no effect on the reading of the text?

What we then did was to ask another group of 66 readers to read a clean version of the text; namely, the one you see on the lower half of the overhead. We then examined their reading time on each sentence, and here are the results. That is, regardless of whether they noticed the error in the text or not, those who read the error-filled text took longer on the line than those who read the clear text. Remember: These data are from those who failed to notice the error.

Ken concludes his argument by saying, "The real world of reading is making sense of print, not recognizing words." I would argue that the real world of reading is making sense of print by recognizing words. (More on FAQ1)

Now, this leads me to the question of: How accurate are skilled readers? Reading about miscue research would persuade you that everybody makes miscues. We wondered about that and thought we should do some of our own, so we've asked my class of students in the psychology of reading to read aloud a text. [Figure 2]

Here are the total number of errors they made in reading Keith Stanovich's How to Think Straight about Psychology. Here are the total number of errors they made in reading James Joyce's The Dubliners. And what I would have you note is that the modal student, that is, the typical college student, makes no miscues at all. So what are we learning by asking people to read material which is too difficult for them, which is the standard practice in miscue analysis?

My contention is that skilled readers simply don't make errors. (More on FAQ1) I have told Dick this morning that my last miscue was in 1967, but . . ..

How do they manage this? Well, I maintain that it's because skilled readers are simply great decoders; that is to say that what they do is simply look at print and effortlessly and automatically recognize the word on the page. That is so much easier to do than any other option, like using context to figure out what the word means, that no one should resort to using the text to try to figure out what the meaning is. (Return to PI)

How do I know this? Well, in the first place, studies of eye movements tell us that the skilled reader fixates on or near every word in the passage. That is to say, skilled readers do not skip words; they look at every word. (More on FAQ1)

In fact, the presence of a single error in a line. To go even further than the Golasch study, we had people read a passage from Ken Goodman. And what we did was to place a single X in a line of text.

Now what I am going to show you is just one representative line, but the same is true of all the lines in the text. [Figure 3] That is, here is the experimental passage and it contains "virtually all human babies learn to speak their home languxge" and replaced the A in language with an X. And we had the other half of our subjects read a clean version of it.

The yellow bar represents the reading time on that line, and the presence of a single X in a single word in an entire line of text costs the reader a quarter of a second. Decoding is important, but I don't want anyone to leave this room thinking that Gough or anyone else believes that decoding is all there is to reading. In fact, what I would argue is that reading consists of getting the words off the page and then understanding those words; that is, that reading involves decoding times comprehension. (More on FAQ1)

And we reflect this in what we call the "Simple View of Reading," namely that "R equals D x C." (More on FAQ1) Given that R, D, and C vary from 0, where you can't do it, to 1.0 where you can do it like Tiger Woods, what this conveys is the fact that if you are poor at decoding or poor at comprehension, you must be a poor reader.

Now, educators don't like to break down reading into skills, as we all know, but I think it's important to do so because I think these two skills are extremely different and separate. We know they're correlated; that is to say, the skilled decoder tends to be a skilled comprehender, the poor decoder tends to be a poor comprehender, but they are dissociated. (More on FAQ1)

If we put them in a 2 x 2 table and we say, okay, here are kids who can decode and kids who cannot, here are kids who can comprehend and here are kids who cannot, what you see is there are four cells. Most readers fall in the northeast diagonal; that is, they are either literate or they are garden-variety reading disabled. But we can find children who can decode skillfully and not understand. We call them hyperlexic.

We can also find too many children who can understand perfectly if you read the story to them or you talk to them, but cannot decode; we call them dyslexic.

We have abundant evidence that this simple view of reading provides a very precise description of reading skill. For example, in a large study done at the Southwest Educational Development Laboratories by my colleague and friend Wes Hoover, the Southwest Educational Development Laboratories went into four small towns in Texas and measured the reading, the decoding, and the comprehension skills (in this instance, the listening comprehension skills) of a couple of hundred Mexican American children in these towns.

When we took their scores on reading -- and their reading, by the way, was not a word list of anything of the sort; it was reading a series of passages and answering questions about the reading and retelling, in effect. The listening comprehension was assessed in the same way, save that the experimenter read the story to the child and then asked her to retell it. The decoding was measured by asking the children whether they could read a short list of pseudo-words, nonwords -- decontextualized decoding if you will. When we took their scores on decoding and multiplied them times their scores on comprehension, we predicted their reading ability all but perfectly; that is, there is nothing left. If you know how well a child can read three pseudo-words and you know how well that child can listen and retell a story, you know how well that child reads. (Back to FAQ6)

Well, the simple view also gives us a take on why it is the relations between these parts change over the years. Back to our formula, R equals D times C. The stories we ask children to read are very simple, so in terms of the formula C would equal 1. Well, if C equals 1, then R equals D. And so we would expect to see that among children, the ability to decode, the ability to read isolated words or to decode pseudo-words will tell you how well they read.

Here is a study carried out in Connecticut with some 700 children which makes the point very effectively; that is, in the beginning, when you look at first graders, the correlation between their ability to read isolated words and their ability to read and retell a text is nearly perfect. That correlation, as the materials we asked them to read get more difficult, and as decoding becomes more difficult, what we see is that that correlation decreases.

By contrast, think of the college student. Most college students, the vast majority, are skilled decoders. That is, for them, D equals 1 or something close to it. Well, if D equals 1, what does R equal? C. So what we would expect to see is that among college students, reading and listening ought to be highly correlated, and that correlation should grow as a child gets older. And sure enough, it does.

This is a representation of a dozen different studies showing what happens to the correlation between reading and comprehension across grades, and what it clearly does is to grow from about .4, up to near perfection by the time the child is in college.

All right. What does this suggest? It suggests to me that a very essential part of reading is knowing what I call the cipher. I use that term because I much prefer it to what I think is a hideous term, the grapho-phonics system. I would maintain that what a child needs to know in order to read a language like English is the system of letter/ phoneme correspondences that the language contains. Now, that system is enormously complex and far more complex than most people realize. (Back to FAQ1) Had we time, I would try to demonstrate that fact, but for now, let me just demonstrate why I believe it is that the child needs to know the system.

There are two major reasons. The first is the novel word. You and I are old enough and have read enough so that it is rare for us to run into a novel word. I can't even remember the last time I saw a word in print that I had not seen before. But for the beginning reader, this is a constant experience.

That is to say, the beginning reader comes to school with a vocabulary containing something like 5,000 words, and every one of those words is going to be encountered for a first time. So for a very long time, perhaps the first three grades, children are primarily encountering words in texts which are already part of their spoken vocabulary. How do they recognize them?

Well, Ken and his colleagues in whole language would say that you must use context. And I would agree, use the context all you want. But let me remind you that context is a false friend, in the following sense; that is, if you ask people to guess the content words in a text, the modal content word, the typical content word, cannot be guessed by any of a hundred people. Had we time, I'd demonstrate that for you.

So context will let you down with respect to the content words, the nouns and verbs and adjectives that carry the message that the text embodies. On the other hand, context will help you quite a bit with respect to the function words, the articles, the pronouns, the prepositions. But you don't need help with them. They're all familiar. You know them all. So context is in that sense a false friend. The child instead needs the cipher, needs to be able to look at "mist" or "desk" or "dress" and simply realize that that's "dress," that's "mist," that's "desk."

The second problem is the memory problem. And here what we're facing is the fact that in order for something to become a sight word, in order for something to be stored as something you can recognize automatically, you need to see that word very often. (More on FAQ2)

Back in the '30s, Gray did a study in which he argued that a child needs at least six exposures to a word in order to make it a sight word. What few of us realize, I think, is how rarely words occur. Here, for example, is a count of the words that are used in a study done by NAEP on children's oral reading back in 1992, and what you'll see is that here is the word "the" out here. It occurs very often. In fact, the word "the" accounts for 7 percent of all the words in running text. If you taught the child the word "the" and 11 other words, they could read 25 percent of the New York Times. (More on FAQ2)

The important point, though, is back at this end of the distribution. What you will see is that the typical word in a children's book or in any text, for that matter, occurs but once; only a single time. How is one to get those exposures?

Well, here I would concur with Dick in saying that a crucial factor in early reading is reading itself, doing a great deal of reading and a great variety of reading, because that's the only way you're going to encounter most words. (More on FAQ2) Most words occur very rarely. (Return to PI)

Does the child who knows the cipher read differently than the child who doesn't? Yes. I'll go briefly over this. The child who knows the cipher, that is, the child who can read pseudo-words, reads faster and more accurately than a child who doesn't. They show a smaller frequency effect. They show a larger regularity effect. They don't make substitutions when they're oral reading, whereas the child who does -- who does not know the cipher does.

The child with the cipher will create neologisms, will mispronounce words. Finally they will create substitutions, but their substitutions are a result of their application of the cipher.

Children who don't know the cipher will substitute old words. They're reading a text and all they'll do, they come to a word they don't know and they'll use context and the picture as best they can, but the chances are what they'll do is they'll throw in a word that they've encountered before in their reading texts.

The same thing is true with respect to their spelling. That is, I am very fond of invented spelling, for example, because one of the things that happens in invented spelling is that spelling becomes phonetic, and this indicates the presence of phonemic awareness, this indicates some knowledge of the cipher. (More on FAQ2) But children who can read pseudo-words aloud spell more accurately, more phonetically, insert fewer letters, and never make nonalphabetic intrusion.

The child who lacks phonemic awareness, the child who lacks the cipher, will substitute things like digits, ampersands, and exclamation points in their spelling. And the presence of those things in a child's spelling test is a sure sign that the child does not know the cipher. Well, what do we know about teaching the cipher? Here I would agree with Dick once again. That is to say, I don't know what the best way is to teach the cipher, but I think I disagree with him with respect to phonemic awareness.

That is to say, I think there is a mountain of evidence which says that the child will not learn to read unless she has awareness that the spoken word can be decomposed into segments. (More on FAQ2)

Let me try to make clear that phonemic awareness is not phonics. That is, questions which tap phonemic awareness are ones like, what's the first sound in "fish," what's the last sound in "drum"? I can break a word into pieces like this, "dog" would be (sounding out) "d-o-g." Can you break sun into pieces? Notice that there is no reference to letters in any of this, and phonemic awareness has nothing to do with letters.

Does it matter? You bet it does. Tunmer and Nesdale found that children who have phonemic awareness may or may not be able to read pseudo-words. But children who do not have phonemic awareness cannot read pseudo-words. But if you don't have phonemic awareness, you damn well will not learn the cipher. It is necessary but not sufficient. And I believe our early instruction ought to emphasize phonemic awareness. (More on FAQ2) (Return to PI)

Do children have phonemic awareness? Well, here is how much they have on entering the first grade, according to some data we collected with a thousand Austinites. That is to say, roughly half of the children enter the first grade able to tell you what the first phoneme in a word is. Fewer can tell you what the last phoneme is and still fewer can segment "dog" into "d-aw-g."

Hallie Yopp in a recent article in The Reading Teacher showed that the ability to do this (phonemic awareness tasks) in kindergarten predicts reading success in the fourth grade better than anything that we've yet discovered. (More on FAQ2) And all of this leads us to the point that what we should be doing is teaching phonemic awareness in kindergarten.

Dick says 80 percent of the children will get it. I guarantee you that the 20 percent of the children who don't get it will not learn to read.

So I would argue that what research tells us is that early reading instruction should crucially involve phonemic awareness. (Back to FAQ2)

Thank you.

References

Bradley, L., & Bryan, P.E. (983). Categorizing sounds and learning to read - A causal connection. Nature, 301, 419-421.

Gollasch, F.V. (1980). Readers' perception in detecting and processing embedded errors in meaningful text, University of Arizona, Tucson, AZ.

Gough P.B. (1983). Context, form, and interaction. In K. Rayner (Ed.), Eye movements in reading: Perceptual and language processes. New York: Academic Press.

Gough, P.B., & Griffith, P.L. (1992). Reading spelling, and the orthographic cipher. In P.B. Gough, L.C. Ehri, & R. Treiman (Eds.), Reading acquisition. Hillsdale, NJ: Erlbaum.

Gough, P.B., Hoover, W.A., & Peterson, C. (1996). Some observations on the simple view of reading. In C. Cornoldi & Oakhill (Eds.), Reading comprehension difficulties. Hillsdale, NJ: Erlbaum.

Gough, P.B., & Tunmer, W.E. (1986). Decoding, reading, and reading disability. Remedial and Special Education, 6-10.

Hoover, W.A., & Gough, P.B. 9990). The simple view of reading. Reading and Writing, 2, 127-160.

Just, M.A., & Carpenter, P. A. (1980). A theory of reading: From eye fixations to comprehension. Psychological Review, 87, 329-354.

Lundberg, I., Frost, J., & Petersen, O. (1988). Effects of an extensive program for stimulating phonological awareness in preschool children. Reading Research Quarterly, 23, 263-284.

Matthews, M.M. (1966). Teaching to read. Chicago, IL: University of Chicago Press.

Rack J.P., Snowling, M.J., & Olson, R.K. (1992). The nonword reading deficit in developmental dyslexia: A review. Reading Research Quarterly, 27(1), 28-53.

Rayner, K., & Pollatsek, A. (1989). The psychology of reading. Englewood Cliffs, NJ: Prentice Hall.

Share, D.L., Jorm, A.F., Maclean, R., & Matthews, R. (1984). Sources of individual differences in reading acquisition. Journal of Educational Psychology, 76, 1309-1324.

Return to Main Speakers


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