Do Readers Perceive Every Word the Same Way Every Time?
The Science of Reading conference, a small group of academics and journalists interested in phonics instruction, insists that beginning readers must be taught as total beginners who know nothing about reading. Children have been sheltered from literacy in communities and homes until they get to school. They have to be carefully, systematically taught from the ground up.
To prove their case, they cite evidence that many children can’t recognize all of the letters of the alphabet, don’t have highly developed phonemic awareness, and know little of orthography when they enter kindergarten. They take a simple view and argue that children already know plenty about language—vocabulary, making comments, asking questions, listening. Reading teachers must teach them what they don’t already know about letters and sounds. Forget about the language they already know how to use. Just use it. Read to them. Talk with them. Don’t let language interfere with phonics instruction.
These Scientists insist that beginners must be taught to recognize and pronounce every word in every phrase, clause, and sentence perfectly as single word units—to a mastery criterion—before they can be expected to read real texts. “A fat cat sat on a mat” is an example of a criterion of mastery. Indeed, for Scientists these phonetically regular words and many others, according to the fat cat criterion, ought to be pronounced accurately when presented in isolation in a list before a child should begin to read for-real text.
The sentence might just as well read “A fat mat sat on a cat” as far as the beginning reader is concerned. Try as we might, it’s really hard to strip meaning and structure from reading, but the Scientists insist upon it. Dalle-3, the bot I used to create these images, repeatedly froze on the “fat mat sitting on a cat. (“Sorry! I’m unable to create that image”). Finally, I got this one:
Say hello to Mat. Note that the AI insisted on making sense of the fat mat sentence. As a voice created from mathematical probabilities, the AI could find no possibility that a fat mat could sit on a cat. But Fat Matt, a young man, could sit with a cat. Given even a shred of conventional syntax, Dalle-3 had little trouble with this one: “Sat a cat on a mat fat.” Both the cat and the mat are fat.
If a child can put together twelve phonemes into a string of distinct morphemes (f-a-t, c-a-t, s-a-t, m-a-t) blended with consonants surrounding four short a’s with 100% accuracy, this reader matriculates to instruction in reading as we know it in the ordinary sense of the word “reading.” It matters not whether the child reads for sense or for nonsense. Reading words aloud with speed and fluency is the goal.
It’s nonnegotiable, it’s scientific, it’s the law in most states.
*****
Relying on a single strategy to teach beginning reading, the phonics instructor hears children read the sentence “the horse was eating hay in the barn” as “the (stop) h-h-h h-or-se horse was (stop) e-e-ting eating (stop) hay (stop) in (stop) the b-ba-barf-b-ar-bar-n-barn.” The child focuses all of their effort on using the graphophonic system to reproduce the sentence in speech. Do nothing to draw learner attention away from the internal spellings of words.
Assuming the phonics instructor teaches words like “the” as sight words (high-frequency words, Dolch words, whatever you like) memorized as wholes, words like “horse” are not high-frequency words necessarily memorized but must be sounded out as a series of phonemes before the verb “was eating” is decoded. Whether the reader comprehends the scene is immaterial. The instructor cares about the pronunciation of a horse, its meaning nor its function.
If a reader reads aloud “The house was eating hay in the barn,” ordinary readers would find several concerns. First, houses don’t usually eat, and if they did, why hay? Now, hay could be in a barn, horses are kept in barns, horses eat hay, house looks and even sounds a bit like horse, houses eating in barns doesn’t make sense—the word must be [horse]. As before, it was all but impossible for the bot to conjure a house eating hay in a barn. Have a look at what I needed to prompt the bot to generate this image:
A political reading scientist who insists on the print-to-sound 100% accuracy requirement diagnoses the reading problem exhibited here as “phonics-based.” “House” as a graphophonic cue doesn’t track “horse.” They look and sound different. Pay no mind to differences in meaning.
Practicing the r-controlled vowels (-or) ought to be the next instructional step, says this scientist. Significantly, this style of reading scientist has convinced almost every state to bar every teacher from letting a child read to the end of the sentence and leave the miscue uncorrected to return for a second pass. Phonics instruction is immediate, uncontaminated by meaning. Reading by definition is one word at a time.
Why might a teacher want to let a child see more context instead of stopping to sound out the word? Why not stop the child and focus attention on sounding out the word “horse?” Lots of reasons. First, I would want to know if the child even attempts to self-correct—and when. The expected response “horse” is followed by “was eating.” The observed response “house” is semantically impossible. The issue for me is not whether “I need to drill this child on r-controlled vowels,” though I might very well take that up as a hypothesis and do a quick assessment—after I attend to the biggest issue.
Is this child reading for meaning? If so, I will observe to see whether a spontaneous attempt at self-correction ensues. A child who regresses (moves their eyes to the left in the visual array of letters) following the word “eating” and tries to sound out the word “horse” shows evidence of “metacognition” or “comprehension monitoring.” This sort of self-monitoring for comprehension is a mindset, a disposition, a habitual way of thinking, that can be developed using an instructional approach called Reading Recovery.
Reading is comprehension; phonics is sounding out words to serve comprehension. In order for phonics to be useful, readers must be aware of their miscues (something’s wrong) inside a sentence and take self-corrective action, which may include phonics analysis, which can be supported by spreading activation of semantics and glimpses of syntactic relations. Test yourself. Analyze this miscue to ascertain whether the reader is attending, or not attending, to phonics cues, semantic cues, and/or syntactic cues: “The well-dressed man looked at his partner with love and infection in his eyes.” Of course, the expected response is “affection.” What produced this miscue?
Take away semantics and syntax. The only feature that changes horse into house is one letter—r vs u. Which squares with your ordinary understanding of reading? Checking oneself continuously to see whether the text makes sense or checking to make sure you are pronouncing the words correctly? I realize it’s not fair to ask lay people to answer a question requiring considerable expertise, but this is what the Science of Reading conference is doing.
Marie Clay, the inventor of Reading Recovery, decribed the self-correction mindset as the engine of self-development and improvement. Scaffolding beginners who can’t keep up with the class to learn to orchestrate and integrate three ever-present, fundamental cueing systems as resources for self-correction and self-learning activities is powerful education. Technical training in spelling patterns and sounds, while important, is sideline work.
So as a reading teacher, I recommend civil disobedience. Break the law if you have tenure. Let the reader read on about houses eating hay until the sentence ends. I’d wait for a split second, hoping to see a spontaneous regression with corrective activity. If not, I’d say “check it,” and I would keep saying “check it” until the child is habituated to the idea that reading must make sense and that fix-ups are needed routinely.
If the child struggles to identify the problem, I might ask “Which word isn’t making sense?” or I might say “Take another look at this word.” As a last resort, I might ask “Do houses eat hay?” and comment on the behavior I saw that I want to see repeated. “That was really smart to go back and figure out what didn’t make sense.”
*****
Advocates for Reading Recovery, often called ‘three cueing’ by the Science conference, know about the advantages of learning to analyze words phonetically as a route to automatic word recognition. These teachers agree with phonics teachers that young readers need to develop automaticity and fluency—instant word recognition followed by oral reading that sounds natural. Some children profit from direct, explicit instruction in phonics—not as a blanket prescription day in and day out for all children. The issue isn’t just that ‘house’ isn’t ‘horse’ and simply can’t be read as ‘horse.’ The issue is what do we do to help readers grow a mindset to foster self-development—to recognize that ‘house’ can’t be ‘horse’ in their independent reading when no one is watching.
Since the 1990s when I had the good fortune of teaching at a university serving as a regional training ground for Reading Recovery teachers, I’ve read and taught a fair amount of research about miscue analysis and various approaches early reading teachers take toward teaching students to orchestrate cueing systems and become more and more expert until they experience reading as mature readers.
Two principles fairly well saturated with evidence in mainstream reading science convince me that Marie Clay uncovered a very good way to help early readers who struggle with the mechanics of text. First, prior knowledge is crucial for comprehension, including knowledge of the semantic and the syntactic dimensions of language. Second, linking automatic in-the-moment decoding and comprehending to metacognitive monitoring work from the very beginning is crucial. Teaching children to sound out words is important side work. Making it the central focus of instruction wouldn’t be my recommendation.