Beyond Reading Comprehension Strategies

We need to be clear on what we mean by “comprehension strategies”—and reevaluate the usefulness of the concept.

---

For decades, schools have tried to teach reading comprehension by focusing on a defined set of “skills and strategies,” including making inferences, finding the main idea of a text, and summarizing. Partly because standardized reading tests appear to measure students’ mastery of such strategies, this kind of instruction has taken up many hours of the school day every week, year after year, edging out other subjects like social studies and science, especially at the elementary level.

Cognitive psychologist Daniel Willingham has long argued, based on the results of numerous studies, that students get all the benefits of strategy instruction after only about 10 hours. In other words, skills like “finding the main idea” aren’t the same as skills like swimming or playing tennis: they don’t continue to get better with practice. (Some have argued there’s a difference between comprehension skills and comprehension strategies, but many educators use the terms interchangeably. I’m not convinced there’s a significant difference. Willingham seems to use “strategy” to mean the instruction and practice, and “skill” to mean what students acquire from it.)

Willingham has suggested that teaching comprehension skills is more like reminding students to check their work in math: It may remind them that they need to find the meaning of a text, but it won’t necessarily enable them to do it.

Most recently, Willingham has noted that after reviewing 12 major meta-analyses, he found it didn’t matter whether students got four hours or several hundred hours per year of strategy instruction. “There was no evidence that increasing instructional time for comprehension strategies—even by 400 percent!—brought any benefit,” he observed.

Two Kinds of Reading Instruction

At the same time, Willingham has introduced a new wrinkle. It has been pointed out to him, he writes, that his description of comprehension strategy instruction could mean that “schooling should include phonics instruction, some work to support [reading] fluency, and then perhaps two weeks of comprehension strategy instruction. What’s the point of anything else if comprehension can’t be taught?”

Clearly, that can’t be right, he says—and then goes on to differentiate “types of comprehension.” One type, like making inferences, leads students to a “fairly basic” understanding of a text. These are the strategies that don’t continue to improve with practice. That’s probably because students acquire them through oral language processes and are “already quite good at them when they start school.”

“If a typical five-year-old heard, ‘Michael dropped Jessica’s remote-controlled car. Jessica punched him,’ she would readily draw a causal connection between these ideas,” Willingham writes. Still, he continues, children may not draw inferences to connect ideas when they read because they don’t realize it’s important to do that. So strategy instruction can help nudge them to that recognition.

On the other hand, he argues, there are comprehension processes that “offer more sophisticated analysis, and these need to be explicitly taught.” The examples Willingham offers are things like evaluating arguments or assessing the quality of writing, or discipline-specific tools of analysis like understanding why a novelist uses foreshadowing, how to interpret historical documents by considering the source, and distinguishing between correlation and causation.

One ardent defender of comprehension strategy instruction has cited Willingham’s more recent work to me, in an email, as a “retraction of his suggestion that ten lessons max out the benefits of strategies.”

I don’t see it that way at all. Not only is he clearly standing by his previous position, he’s even going further to suggest that four hours of strategy instruction is the “max out” point.

You could argue, as my email correspondent seems to be doing, that Willingham is also now saying that a different kind of strategy instruction—not the kind measured by the tests that have been used in the studies—doesn’t max out in that way. In other words, there are benefits to teaching those tools of more sophisticated analysis for more than a few hours. But if you read Willingham’s writing carefully, it’s clear he’s not applying the term “strategy instruction” to teaching students to do things like understand the difference between correlation and causation.

Willingham seems to be arguing, essentially, that we don’t need to teach things like “making inferences” for more than a few hours, but we do need to keep teaching students how to do those other more sophisticated things—across the curriculum—indefinitely.

The Picture Is Even More Complicated

While I have the utmost respect for Willingham’s work, I don’t completely agree with that neat dichotomy. I do agree that we can’t really teach students how to make inferences in the abstract, because that’s something kids learn naturally. If a toddler touches a hot stove and burns herself, she will likely infer that the same thing will happen if she touches it again. But that doesn’t mean that four hours of instruction in making inferences—or, more accurately, reminding students that they need to apply their ability to make inferences to text—will be enough.

I’m not questioning Willingham’s analysis of the data. Rather, I’m saying that the data focuses on a particular kind of strategy instruction—the kind that puts the skill or strategy in the foreground rather than the content. Researchers do that because they’re trying to assess whether a particular intervention—like teaching inferencing—actually works.

Researchers may provide their subjects with some “background knowledge,” but only to the extent necessary to evaluate the effectiveness of the strategy. For example, if a reading passage is about coal mining, and students don’t know anything about that, researchers may be provide them with the information and vocabulary they need to understand the passage at least at a superficial level.

It’s understandable that researchers generally rely on reading passages on topics kids aren’t already learning about. It may be easier for them than using passages from, say, a social studies textbook that’s part of the curriculum (although at least a few researchers have done that, with good results). More fundamentally, researchers may be trying to separate the effects of students’ content knowledge from the effects of instruction in the comprehension strategy—which is, in my view and that of some others, essentially impossible.

In any event, that doesn’t mean classroom instruction—where the objective is enabling students to learn rather than assessing the effects of a particular strategy—should follow the same model the research does. And yet that’s become the standard classroom approach: give students texts on random topics they may not know anything about, maybe explain a few unfamiliar vocabulary words, and have them practice whatever the skill of the week is.

What the studies tell us is that continued practice in that kind of skills-focused comprehension instruction doesn’t provide benefits. But that doesn’t mean that after a few hours of instruction and practice in making inferences, students will never again need help making an inference.

As teachers have undoubtedly noticed, students might be able to make an inference about a simple story but struggle to make one about a more complex text, especially if it assumes a lot of background knowledge they don’t have. One study done by ACT found that what distinguished college-ready readers from others wasn’t their general ability to do things like find the main idea. It was their ability to do that with complex text.

I’m also dubious that a strategy like making inferences leads to nothing more than a basic understanding of a text, as Willingham posits. Depending on the context, it seems to me that it could easily resemble the comprehension processes Willingham categorizes as being more sophisticated. When students learn to evaluate a historical document by considering its source, for example, aren’t they making an inference about the motives or interests of the person who wrote it?

Teachers will likely need to support students in making inferences, and engaging in a host of other cognitive processes, throughout their school careers, across a range of subjects—not just for a few hours. But instead of thinking of what they’re doing as comprehension strategy instruction, teachers should see it as helping students make sense of whatever they’re trying to read—or trying to learn.

In other words, they should put the content in the foreground rather than the strategy. Maybe being guided to make an inference would help students with one text, but “finding the main idea” would be better suited to help them with another.

Some Strategies Can and Should Be Taught Explicitly

That’s not to say that the concept of strategies is entirely useless. But the label has been used to lump a bunch of things together that seem to me to be inherently quite different.

Some comprehension strategies, like making inferences, can’t be taught in the abstract and aren’t truly transferable. Whether you can apply your ability to make an inference is going to depend on whether you have the necessary background knowledge—which could be knowledge of the topic but could also include more general knowledge of academic vocabulary and the kind of complex syntax used in written language.

Other strategies, like finding the main idea of a text and summarizing (which are actually pretty much the same thing), involve a combination of knowledge and transferable skills. These aren’t things most kids learn to do naturally. And while their ability to do them will depend in part on having enough relevant background knowledge, it is possible to teach kids how to engage in the process of generalization that is a prerequisite skill. Once students have had some practice summarizing a text (or finding the main idea), it’s likely to be easier for them to do it in other contexts, assuming they have the necessary background knowledge.

But a skill like summarizing is actually far more complex and difficult than we’ve acknowledged—and learning how to do it requires far more than a few hours of vague instruction and practice. Usually, students are just told that summarizing means they should “include the important information,” but many will need explicit guidance in order to do that. They may need repeated instruction, practice, and feedback in asking and answering questions about a text, like who, what, when, where, why, and how, and then combining the answers into a sentence or paragraph.

Ideally this would be done not just through reading instruction but also writing instruction that is explicit, logically sequenced, and embedded in the content of the curriculum—in other words, a method like The Writing Revolution. Rather than thinking of things like summarizing and finding the main idea just as reading comprehension strategies, we need to start seeing them as writing strategies, and indeed, learning strategies.

Connecting Literacy and Learning

That would involve recognizing that reading and writing aren’t entirely separate processes, and that neither of them is separate from acquiring knowledge in “content areas” like history and science. If you’ve learned how to construct a coherent paragraph with a good topic sentence, you’ve learned how to “find the main idea and details” in a powerful way—and you’ve deepened your understanding of whatever content the paragraph is about.

And we still need to put content in the foreground. It's alluring to think there’s a particular strategy, or set of strategies, that holds the key to all comprehension. Unfortunately, that’s not the case. Even summarizing, which has a lot of evidence behind it, has its limits. If I were having students analyze the imagery in a poem or the foreshadowing in a work of fiction, that’s probably not the strategy I would choose.

One benefit of viewing literacy as a subset of learning, rather than an entirely separate field, is that it takes a heavy burden off of teachers of English or “reading.” Especially at higher grade levels, where instruction is departmentalized, teachers of social studies and science—or really, any subject—need to also see themselves as literacy teachers. By the same token, English and reading teachers also need to see themselves as content teachers.

We have evidence that when that happens, students can benefit greatly. An elementary curriculum called MORE, which combines social studies and science topics with techniques like teaching about prefixes and suffixes, has been found to improve both decoding and comprehension through at least fourth grade.

“There are very few literacy interventions that improve reading comprehension in third grade and beyond,” the lead creator of MORE, Professor James Kim, has been quoted as saying. “And there are even fewer that improve reading comprehension through fourth and fifth grade.”

Some high schools that serve the many older students who struggle with reading are experimenting with teaching literacy skills across the curriculum. Some students need one-one-one remediation, but they and their classmates also benefit from in-class activities like taking apart the word “intermolecular” in chemistry or connecting words with mathematical symbols in math class.

Teachers of any grade level or subject may need some training in foundational reading skills in order to help struggling students. It would be enormously helpful for them all to get training in how to teach writing effectively, which would include how to embed writing activities in whatever content they’re teaching. They would also benefit from effective curriculum materials that incorporate the kinds of questions most likely to help students understand and analyze that content—and, crucially, good training in how to use those materials.

What I would not recommend is training in comprehension strategy instruction of the sort that many teachers have received. The assumption that we can teach comprehension in the abstract has done far more harm than good.