About the New York Times article claiming the reading wars have been won. Questions and Ponderings by Dr. Sam Bommarito
Clarification about the article. It is from a TIME100 Next article. See screen capture above.
Let’s begin with the key questions we must ask about the article and its topic. Is the LETRS program really the silver bullet that the article implies it is? The short answer is no. Here is why I say that:
- The LETRS program has been shown to improve students’ decoding scores and teachers’ knowledge of phonics. On its own, it has not been shown to raise comprehension scores. https://riveteducation.org/a-timely-reminder-there-are-no-silver-bullets-even-with-professional-learning/. Direct, explicit instruction in synthetic phonics raises students’ decoding scores. It also improves teachers’ phonics knowledge. But as analysts like Johnson and Thomas have pointed out in both their books and blogs on the subject, LETRS training improves students’ decoding scores, but does not significantly improve their comprehension scores. In addition, some supporters of LTRS and other programs focused on teaching synthetic phonics have made some misleading claims. Before readers accept claims about the “Mississippi Miracle” at face value, they should examine them carefully. At the end of the day, the “miracle ” applies mainly to gains in decoding scores, not comprehension scores. The programs in the “Miracle states” did much more than add synthetic phonics. The net result is that researchers like Rachel Gabriel report districts are having buyer’s remorse about the choices they made LINK. Rachel cites the work of Rachel Canter LINK, which indicates that in her home state of Mississippi, “When adjusting for demographic factors such as poverty, we’re in first place.” This shatters the belief that many seem to have that all you need to do to improve students’ reading test performance is to add synthetic phonics to the program. Other demographic factors do matter. The so-called Mississippi Miracle involved much more than simply adding LETRS.
- Within the SOR community, there is now an interest in a different form of analytic phonics. This form of phonics gets students to the “takeoff point” of teaching themselves phonics in months, not years. It also uses up less classroom instructional time, allowing more time for comprehension-related activities. For details, see my blog interview with Nora Chabazi from EBLI, the most cited source for this form of phonics: LINK.
- When LTRS was created, research on the brain as a problem-solving machine had not yet emerged. See my blog post outlining the results of that newest brain research LINK, and its implications for the “Great Debate”. The newest studies fundamentally challenge some assertions SOR advocates make about the brain’s role in the reading process LINK.
- The results of mandating analytic phonics in England gives us a peek at what we can expect in the United States. See Dr. Michael Rosen’s analysis of the effects of mandating phonics in England. You can find it on his Oct 2n Facebook post this year. The upshot: mandating phonics didn’t work, even after millions of dollars were spent.
- My overall position on the “Great Debate” is that treating this issue as a debate is counterproductive. My position is that we need to explore and use evidence from ALL sides and give all students access to both analytic and synthetic phonics. LINK.
================================================================================
I’ll leave the reader with three additional things to ponder. One is what the NRP actually said about reading instruction. Here is the excerpt:
Add to that what Dr. Neil deGrasse Tyson said about how scientists handle questions. They DON’T debate. They argue, exploring all sides of the issue. They treat the “truth” as being written in Xerox, not in stone. SCIENTISTS CHANGE THEIR MINDS WHEN NEW EVIDENCE COMES ALONG. The best we ever have is the “current consensus”.
Duke and Cartwright, 2021: Active View of Reading model and added effect sizes for selected parts of the model.
============================================================================
ESPECIALLY NOTICE THE EFFECT SIZE FOR BRIDGING PROCESSES AS THEY APPEARED FROM THE SLIDE ABOVE
The effect size for the bridging processes is considerably larger than the effect size for word recognition. That means there could be a real advantage to adding programs like Rasinski’s Fluency Development Lesson Plan, which is based on decades of Rasinski’s research on the best ways to teach fluency.
============================================================================
Overall, I must respectfully disagree with Dr. Louisa Moats about whether the issues of how best to teach phonics are settled. I also must question her implication from the Times Article that the problems inherent in teaching LETRS would be solved with better implementation over time. In many districts, more than enough time has passed, and the promised results haven’t materialized.
Please remember, the NRP outlines five ways to teach phonics, not just one. Yet folks promoting Moat’s brand of SOR have outlawed all but one. Many of the approaches that involve analytic phonics (think problem-solving phonics) have a proven track record of improving BOTH decoding and comprehension. My disagreement with her position, and with the position of some of her supporters, in no way diminishes the importance of what she did by pointing out that many schools of education were not teaching synthetic phonics. Too many of them were totally focused on analytic phonics, and because of that, many children suffered. However, the solution to that should not be to do to the kids who thrive on analytic phonics what was done to the kids who needed synthetic phonics. Analytic phonics is NOT GUESSING; IT IS PROBLEM SOLVING. What she and her supporters are doing hurts the kids who thrive on analytic phonics. I still believe that if we consider ideas from ALL SIDES, we can build ways to teach most kids most of the time. It’s time to explore the common ground already emerging from the current discussions LINK.
Dare to Dream!
Dr. Sam Bommarito, the guy in the middle, taking flak from all sides
If you are interested in seeing future posts, please subscribe to the blog (top right corner of this post)
Let:
P = explicit phonics instruction
R = successful reading comprehension
A central claim at issue is that P is necessary for R:
R → P
Bommarito repeatedly answers as though the claim were that P is sufficient for R:
P → R
He then points to evidence consistent with:
¬(P → R)
and treats this as evidence against:
R → P
But:
¬(P → R) ⇏ ¬(R → P)
Failure of sufficiency is not failure of necessity.
That substitution recurs throughout the piece. “Not sufficient” becomes “not effective.” “Not solely causal” becomes “not causally important.”
Neither inference follows.
The LETRS argument has the same defect.
Let:
E(X,Y) = X has a measurable effect on Y
LETRS ≠ synthetic phonics ≠ curriculum ≠ statewide reform
Yet evidence is moved among these objects as though they were interchangeable.
From:
¬E(LETRS, comprehension)
it does not follow that:
¬E(synthetic phonics, reading)
Still less does it follow that:
¬E(Mississippi reforms, reading)
Changing the object under discussion does not strengthen an argument. It changes the argument.
The Mississippi section fares no better.
Suppose:
M = P + X₁ + X₂ + …
where M is the Mississippi reform package, P is phonics, and O is the observed gain in reading outcomes.
If:
M → O
then O cannot be attributed to P alone.
But it does not follow that P contributed nothing to O, that P was unimportant, or that P was unnecessary.
Those conclusions are absent from the premises.
The effect-size argument is weaker still.
Bommarito observes:
ES(B) > ES(W)
where B denotes bridging processes and W denotes word recognition.
He then infers an instructional advantage for a specific fluency program.
But:
ES(B) > ES(W) ⇏ Eff(I_B) > Eff(I_W)
where I_B and I_W are interventions targeting B and W.
An effect size associated with a construct is not an estimate of the causal effect of an intervention intended to change that construct.
And even if:
Eff(I_B) > Eff(I_W)
were established, it would still not follow that:
Eff(Rasinski) > Eff(alternatives)
That requires another empirical premise.
The conclusion is two inferential steps beyond the evidence.
The “new brain research” argument has a simpler problem.
Its form is:
Y is newer than X.
Therefore:
Y undermines X.
No.
To undermine X, one must show:
Y → ¬X
or:
(Y ∧ X) → ⊥
“Newer” establishes chronology, not contradiction.
Then there is the claim about children who supposedly “thrive on analytic phonics.”
Let:
A = analytic phonics
S = synthetic phonics
The claim requires:
∃G : Eff(A | G) > Eff(S | G)
where G is an independently identifiable subgroup of students.
Where is the evidence for G?
If G is defined merely as:
G = {x : A works better than S for x}
then the proposition that A works better for G is true by definition.
That is a tautology, not evidence that such a group can be independently identified or prospectively assigned to analytic phonics.
The NRP argument has the same structural weakness.
From the fact that the NRP identifies several components:
{C₁, C₂, C₃, C₄, C₅}
it does not follow that:
C₁ ≈ C₂ ≈ C₃ ≈ C₄ ≈ C₅
in efficacy, evidentiary support, or instructional priority.
A taxonomy is not an endorsement.
A list is not equivalence.
A category is not a causal estimate.
The repeated appeal to “all sides” is not a scientific principle either.
If:
Evidence(H₁) > Evidence(H₂)
then rational evidentiary weighting is:
Weight(H₁) > Weight(H₂)
not:
Weight(H₁) = Weight(H₂)
The existence of competing positions does not create evidentiary parity among them.
Finally, the article under discussion appeared in TIME, not The New York Times.
That is not a logical fallacy.
It is simply an avoidable factual error in the title of a post devoted to correcting other people’s interpretation of evidence.
The larger problem is more serious.
Again and again, the piece relies on invalid substitutions:
¬sufficient ⇒ ¬necessary
¬sole cause ⇒ ¬important cause
construct association ⇒ intervention effect
newer ⇒ truer
classification ⇒ equivalence
multiple causes ⇒ any particular cause is unimportant
None of those implications is valid without additional premises.
Strip away the substitutions and missing premises, and much of the argument disappears.
You said “Please READ the Original Papers Before Calling This a “Mic Drop”
Dr. Bommarito, I am not asking you to retract your blog post.
I am not asking you to endorse the Science of Reading, abandon balanced literacy, or accept my interpretation of the evidence.
I am asking you to do something much simpler:
**Please read the papers yourself, carefully and in full.**
I strongly recommend it.
Because there is a serious problem with your post, and it is not fundamentally a disagreement between competing camps in reading instruction.
**The instructional conclusions you draw are not conclusions the study establishes.**
Your entry is an example of the public relations of tactic discredit and discount. Attack the credentials and conclusions of the writer. The implication that I don’t read the studies is absurd.
The point is that I read ALL the studies, from many different sources and draw very different conclusions. I’m not the only one drawing very different conclusions.
I am going to with this topic in my post tomorrow.
Best,
Dr. Sam
Dude, it’s not the New York Times.
Here is the link: https://time.com/article/2026/09/23/louisa-moats-letrs-training-reading-interview/
It is listed on Time.com
It is a TIME 100 article.
All that information is readily available on my post.
Not sure what your concern is.
To err on the side of caution I added this to the start of the post “Clarification about the article. It is from a TIME100 Next article. See screen capture above.”