Two Out of Three Reading Comprehension Lessons Stay on the Surface, Even With a Great Curriculum

A few months ago on the show we talked about the Mississippi miracle and why it disappears by eighth grade. Phonics instruction got a whole state's fourth graders from the bottom of the country to above the national average. Then, by middle school, the gains stalled. The question we were left with was what happens in the years after decoding, when kids are supposed to be reading to learn.

This brief from SRI Education is one of the more concrete answers we have seen. It has nothing to do with dyslexia specifically. The researchers wanted to know what actually happens during reading comprehension lessons in ordinary elementary classrooms, in districts that have done everything right on paper, and the answer turned out to be that most of the time not much happens.

It is a fifteen page learning brief rather than a journal article, and it is free to read. We will summarize it here and then get into why we think it matters for families like ours.

What they did

SRI was hired in 2024 to study four urban districts: Aldine in Texas, Baltimore City, Guilford County in North Carolina, and Richmond in Virginia. All four get support from the Charles and Lynn Schusterman Family Philanthropies for their literacy work, and Schusterman paid for the study. Worth knowing, though the findings are not flattering enough to look like a sales pitch.

What makes these districts interesting is that each had adopted a highly rated, knowledge-rich reading curriculum at least five years before the study. The programs were Core Knowledge Language Arts (CKLA), Wit & Wisdom, and EL Education. If you have heard the phrase "high-quality instructional materials," or HQIM, these are the programs people mean. Five years in, the researchers were not looking at a rollout. They were looking at what mature implementation looks like.

During the 2024-25 school year they:

• Observed comprehension lessons in 111 classrooms, grades 1 through 5, across 24 schools, using a trained observation protocol from the University of Virginia.
• Surveyed 539 teachers about how often they used the curriculum.
• Sat in on 63 teacher planning meetings.
• Interviewed teachers, coaches and principals at 12 schools.

The authors describe it as one of the strongest observational studies of comprehension instruction in fifty years, and having read it, we do not think that is much of a stretch. Most classroom research of this kind involves a handful of teachers, and this one involved hundreds.

Surface-level versus robust comprehension

The whole brief hangs on one distinction, so it is worth getting clear.

Surface-level comprehension is when students treat reading as a set of tasks to complete. They find the details, label the parts, and answer literal questions. Their understanding stays at the level of what the text says.

Robust comprehension is when students build a full mental model of what the text means. They still find the details, but they connect those details to the point of the whole thing, including what is implied rather than stated.

The example the researchers use is a fourth grade CKLA lesson about the Revolutionary War. A surface-level lesson has kids explain the training methods Baron von Steuben used at Valley Forge. That meets a standard, and it is a fine thing to know. A robust version of the same lesson has kids explain those methods and then connect them to why they mattered: without French and German help during that winter, the Continental Army might not have survived to beat the British.

It is the same text and the same lesson plan either way. The only difference is whether the teacher steers toward the bigger meaning or stops at the task.

The good news first

Teachers were using the materials. Somewhere between 72% and 89% of teachers, depending on the district, said they used the curriculum daily or almost daily, and the classroom observations backed that up. In most of the observed lessons, the large majority of time was spent with students actually working with a text for comprehension. Teachers were not drifting off into creative writing or unrelated discussion.

Kids were participating, too. Nearly every lesson had at least some students actively involved, and more than half regularly had everyone involved. Nobody was lecturing for thirty minutes and nobody was doing extended silent seatwork.

The researchers call this a "high and consistent floor." After five years of work, these districts have gotten the basics of implementation nailed down, which is harder than it sounds and deserves some credit.

The main finding: two thirds of lessons stayed on the surface

Here is the problem hiding in plain sight, as the authors put it.

In 64% of lessons, the stated purpose of the lesson only supported surface-level comprehension. In 67% of lessons, the actual work teachers and students did only got them to surface-level understanding. Only 27 of the 111 lessons, about one in four, involved work that built toward robust understanding of the text.

That pattern held in every district, not just in the average.

Because the observers were in each district for about a week, they often saw the same unit, and sometimes the exact same lesson, taught in different classrooms. That turned out to be the most revealing part. One third grade Wit & Wisdom teacher had students find and label literal and nonliteral language in a book about the moon landing, and stopped there. Another teacher, same lesson, did the labeling and then asked what Neil Armstrong actually meant by "one giant leap for mankind." One fourth grade CKLA class found details that described the characters in The House on Mango Street. Another found the same details and then talked about how they revealed themes of discrimination.

These were the same books, on the same days, in the same districts, and the depth of the lesson came down to the teacher.

The curriculum itself did not predict which way a classroom went, either. Two of the districts both used CKLA. One had the lowest rate of robust lessons of the four, at 7%. The other had the second highest, at 30%.

What the deeper lessons had in common

The researchers went back through the 27 robust lessons to see what set them apart. Six practices showed up more often, listed here in order of how strongly they were associated with deeper work:

1. Text-specific analysis. Students closely analyze a shared text, with both literal and inferential questions, and have to back up claims with evidence.
2. Activating prior knowledge. Connecting the text to what students already know, including from earlier texts in the unit.
3. Explaining and modeling. The teacher thinks out loud about how they are making meaning, such as pulling several details together into a main idea.
4. Instructive feedback. Not just "right" or "wrong," but probing, rephrasing and extending what students say.
5. Text-based reasoning. Students do most of the intellectual work, orally or in writing, rather than the teacher doing it for them.
6. Peer learning. Students make sense of the text together, building on each other's thinking.

There is a catch the authors are careful about. None of these practices works on its own. A teacher can run a beautiful text analysis routine and still stay on the surface if every question is "where does the character live?" Even inferential questions about a character's motives lead nowhere if they never get connected back to what the whole story is doing. The practices only help when they are pointed at a big idea.

Why this keeps happening

The brief does not blame teachers, and neither would we. It points at the system around them.

If a district's goals are written in terms of narrow standards like text structure or vocabulary, teachers will teach text structure and vocabulary. If professional development is about pacing guides and handouts, teachers will get good at pacing guides and handouts. And if teachers do not deeply know the content of the texts themselves, they will fall back on the safe surface-level version of the lesson. That last one is easy to underestimate. These curricula ask elementary teachers to guide kids through Don Quixote, Civil War medicine, the circulatory system and the ratification of the 19th Amendment. A teacher who last studied those topics twenty years ago is going to be wrestling with the text right alongside the students, and the observers saw exactly that.

The recommendations follow from the diagnosis: districts should define what deep comprehension means to them and say so out loud, refocus their existing training and coaching on the how rather than the what of the curriculum, and fix their observation tools. On that last point, the authors note that the quick walkthrough checklists principals use tend to reward things that are easy to see from the doorway: posted objectives, kids participating, the curriculum visibly in use. Every one of those was already happening in these districts, and none of them says anything about depth.

One more detail that we think parents should know. The brief points out that state reading tests and the common benchmark assessments cannot reliably measure separate comprehension standards. A kid's score on "identifying text structure" and their score on "finding the main idea" are too intertwined to pull apart. So when a school pulls up the data and decides your child needs a targeted drill on text structure, it is treating a number as more precise than it is, and the drill itself is likely to be another surface-level lesson.

What to keep in mind

We want to be careful not to overclaim here, and the authors are careful too.

• Four districts, all urban, all with outside foundation support. That is a specific slice of American schools. Suburban and rural districts, or districts without this kind of coaching infrastructure, could look better or worse.
• One week per district. The observations are a snapshot. A lesson that looked surface-level on Tuesday might have been building toward a synthesis on Friday, though the authors say they saw enough related lessons to feel confident about the pattern.
• Depth was judged by trained observers using a rubric. That is about as good as this kind of research gets, but it is still a judgment call about a thirty-minute window.
• No student outcome data yet. The brief says an analysis of achievement data is coming. Until then, the claim that robust lessons produce better readers rests on prior research rather than on these classrooms.
• It is a learning brief, not a peer-reviewed paper. SRI is a serious research organization and the methods are described in detail, but this has not been through journal review.

The bottom line

The science of reading conversation has, for good reason, been mostly about phonics. This brief is a reminder that there is a lot of reading instruction after phonics, and we know much less about how well it is going. Districts that have done the hard work of adopting a strong curriculum and getting teachers to use it every day still had two thirds of their comprehension lessons stop at finding and labeling things.

For a dyslexic kid, that hurts twice over, because the surface tasks lean on the skill they lack while the deeper work where many of them shine mostly does not happen.

The hopeful reading is the one the authors land on. A quarter of lessons did get there, with the same materials and the same kids, because the teacher kept steering toward what the text meant. Teachers can learn to do that with good coaching, and you can do it at home tonight with whatever book is on the nightstand.

If you have seen this in your own kid's classroom, or if your district is doing something about it, we would love to hear about it.