When Instruction Becomes Doctrine: What the Science of Reading Is Still Getting Wrong
Mark Seidenberg’s recent response to The New Yorker article “Alphabet Soup” raises concerns that many of us have been voicing quietly, and sometimes not so quietly. He’s right about several things: Structured Literacy has been stretched far beyond its original purpose, explicit instruction has become overextended, opportunity costs are real, and sentence-level language, writing, and knowledge-building have been pushed to the margins of elementary classrooms.
Where I part ways with Seidenberg isn’t in the concern. It’s in the explanation. Specifically, I disagree with how explicit instruction is being defined, what Structured Literacy was actually designed to address, and why, despite good intentions, it continues to fail many learners, including many students diagnosed with dyslexia.
This isn’t a debate about whether instruction should be explicit.
It’s a debate about what should be made explicit, for whom, and in what order.
Explicit instruction does not require phonemic awareness tasks
Yes, children need explicit instruction when they begin learning to read and write. That point has been well established for decades. But somewhere along the way, that idea hardened into a much narrower assumption: that explicit instruction must mean phonemic awareness drills, isolated sound–symbol instruction, discrete trial training, and memorization of so-called “sight” or “heart” words.
That leap does not follow from the evidence.
Young children can be introduced to literacy as a system that connects meaning, structure, spelling, and sound from the beginning, all without stripping language down to its most abstract and unstable parts. Being explicit does not require delaying meaning. It does not require atomizing language.
Explicit does not mean linear. Explicit does not mean sound-first. Explicit does not mean starting with the weakest part of the system and hoping everything else will eventually attach.
Structured Literacy was built for a narrow theory of dyslexia and that theory fails many learners
Structured Literacy is often described as the approach for dyslexic learners. In practice, it was built around a phonological-deficit model of dyslexia.
The first problem is that many dyslexic learners don’t actually fit that profile. Dyslexia often co-occurs with broader language, lexical, and working-memory challenges. A sound-only explanation doesn’t capture that reality.
But the deeper problem is this: even learners who do have phonological weaknesses are not helped by sound-first instruction.
Under Structured Literacy frameworks, these learners are asked to rely most heavily on the very processes that are least stable for them such as phonemic awareness tasks, repeated segmentation, and sound-based generalization. These tasks carry the highest cognitive and memory load exactly where the learner is weakest.
What these learners need is not more attention to sound.
They need instruction grounded in:
- meaning
- stable word forms
- consistent spelling of bases
- morphological relationships
- words used in sentences and real contexts
Sound becomes learnable after it is anchored to meaning and structure—not before.
A brief clinical example
Consider a nine-year-old student who had received three years of high-fidelity Structured Literacy intervention. He could segment and blend accurately, identify phonemes on demand, and decode controlled text at grade level. On paper, the intervention had “worked.”
But he couldn’t reliably spell common words. He couldn’t explain the meaning of unfamiliar vocabulary he read aloud. He couldn’t write a coherent sentence without extensive support. When asked to spell words ending in /əs/, he guessed <ous>, <us>, and <is>. This is because the sound alone offered no reliable guidance.
When instruction shifted away from phonemic awareness drills and toward studying word families, base elements, suffixes, and meaning relationships, everything changed. His spelling stabilized quickly. His vocabulary growth followed, and his sentence writing improved. Only then did phonological patterns begin to consolidate; not as isolated skills, but as part of a coherent system.
This is not an anomaly. It aligns with decades of research showing that spelling, vocabulary, and morphology are central to building stable word representations.
Implicit learning works—for some learners
Seidenberg is right that implicit (statistical) learning plays a powerful role for many typically developing readers. Some children pick up spelling conventions, vocabulary, and syntax simply through exposure.
But not all children do.
Some learners do not acquire these patterns incidentally no matter how much they read or how rich their language environment is. These students are often described as needing “more phonics,” when what they actually need is explicit instruction in structure and meaning, not increased attention to sound.
Spelling makes this especially clear. English spelling cannot be learned through sound alone. Homophones, suffixes, base-element consistency, and morphological conventions all require explicit structural knowledge. Learners who struggle with spelling are not failing to memorize sound. They are missing the organizing system.
Over teaching phonics has real costs
Here, I strongly agree with Seidenberg. Over teaching carries consequences.
In many schools, literacy instruction now dominates the day, often at the expense of science, history, writing, and sustained oral language work. That trade-off is not neutral. Background knowledge, sentence comprehension, and vocabulary are not enrichment. They are necessary prerequisites for skilled reading.
In trying to raise reading scores through ever-more intensive phonics instruction, we have cut away the very experiences that make reading meaningful and generative.
The deeper problem: a linear, gated model of literacy
At its core, the prevailing framework treats literacy as a sequence:
decode → then comprehend
phonics → then language
accuracy → then thinking
This model delays meaning, gate-keeps language, and disproportionately harms learners who cannot move efficiently through sound-based stages. It also underestimates learners who are ready to engage with complex language much earlier.
Integrated approaches, such as Structured Word Inquiry (SWI), do not reject explicit instruction. They reject the assumption that literacy must be built from sound upward, one isolated component at a time. They treat morphology, etymology, phonology, and meaning as interdependent from the start.
When instruction becomes doctrine
Taken together, these issues point to something deeper than imbalance or overreach. We have confused explicit instruction with sound-first instruction, remediation with universal pedagogy, and scientific inquiry with policy mandate.
I’m not arguing against method. I’m arguing for instruction grounded in the structure of English and in how learners actually come to understand written language.
When a single framework is mandated as universally appropriate, it becomes insulated from evidence. When outcomes fail to improve, something we’re already seeing in national assessments, the explanation shifts to “lack of fidelity,” and the solution becomes more of the same: earlier, longer, stricter.
That logic protects systems and methods while children continue to struggle.
If literacy instruction is going to improve, it has to remain accountable to learner response and open to revision, not just frozen into doctrine by policy.
