Natalie Wexler on Memory and the Writing Effect | Literacy and the Science of Learning

Writing is hard—and teaching writing is even harder. But science tells us it’s well worth the effort, because writing flexes the mental muscles that nurture literacy and learning.

Host Natalie Wexler connects cognitive science to specific writing practices that transfer information from working to long-term memory and require students to retrieve and elaborate on that information. She’s joined by psychologists John Sweller and Jeffrey Karpicke, whose research has identified effective instructional and academic strategies for teaching, learning, and lightening students’ cognitive loads.

“Writing isn't just a product—it’s part of the process of learning. In fact, evidence shows that having students write about what they’re learning can result in dramatic cognitive benefits,” Wexler says.

Learning and putting new information to use is a two-way process: students must first transfer new information from working to long-term memory. Then they must be able to remember that information by retrieving it from their memory stores. Writing supports both. 

Karpicke describes an experiment in which college students read science texts in different conditions. Compared to students who read the text once, twice or created a concept map, students who read the text once and then wrote down everything they remembered, recalled significantly more about the topic a week later. 

Many studies have found the same result: writing boosts memory. But not all writing has the same impact. Writing prompts that require elaboration, such as “how” or “why” questions, help expand and strengthen understanding by drawing new connections to the material. And writing is not equally effective for all students. Inexperienced writers can be so cognitively overwhelmed by the task of writing that it actually impedes learning.

Wexler explains how teachers can ease the cognitive burden on students who are learning to write. First, they can ask students to write about content they've already learned about, so they don’t have to juggle new information in working memory along with the cognitive demands of writing. That approach also helps deepen students’ knowledge of curriculum content.

Sweller describes how teachers also can provide opportunities for “deliberate practice,” which can make foundational literacy skills automatic. For example, students who have mastered spelling rules don’t have to think about spelling when they write. Higher-order writing skills never become completely automatic, but practice helps. For example, students who practice distinguishing between complete sentences and fragments, with feedback from a teacher, eventually “develop a gut sense of what makes a sentence a sentence,” Wexler notes.

These processes work together to enhance student writing—which accelerates literacy and knowledge—not as an end-product, but an active part of the learning process.

This podcast is produced by the Knowledge Matters Campaign and StandardsWork. Follow the Knowledge Matters Campaign on Twitter, Instagram and Facebook. Search #knowledgematters to join the conversation.

Production by Tressa Versteeg. Original music and sound engineering by Aidan Shea.

Transcript

Natalie Wexler

Students are constantly bombarded by new information. That’s what school is all about. Science tells us that if we want them to be able to engage in higher-order thinking, they need to retain at least some of that new information in long-term memory and be able to draw on it when they need it. So the question is, what can educators do to ensure their students, across grade levels and subjects, are truly absorbing the knowledge that can determine their future success?

This is Episode 5 of the Knowledge Matters Podcast, Season 3: Literacy and the Science of Learning. I’m Natalie Wexler, your host for the last two episodes. In Episode 1, my co-host Dylan Wiliam covered the basic architecture of cognitive load theory. In this episode, I’m applying that theory to an area where it should be applied, but rarely is: writing. 

First, though, we need to address an elephant in the room—and that’s AI. You’re probably aware that students can now feed a prompt into an AI tool like ChatGPT and, in less than a minute, get back an essay that’s better than what most high school or even college students can produce. 

Some commentators have argued that means we don’t really need to teach kids to write anymore. Students can just edit or tweak a document produced by AI. That would be nice, because writing is hard—and teaching it can be hard too.

But what that argument overlooks is this: writing isn't just a product—it’s part of the process of learning. In fact, evidence shows that having students write about what they’re learning can result in dramatic cognitive benefits. If students rely on ChatGPT instead of doing the work of writing themselves, they're missing out on those benefits—and on learning.

In this episode, we’ll explore the potential benefits of writing. We’ll also talk about why many students don’t get those benefits, and how we can ensure they do. 

So what are the potential benefits? For one thing, writing can help students retain information in long-term memory and access it when they need it. That’s important because the capacity of working memory,  the aspect of our consciousness that tries to make sense of new information, is so limited. 

John Sweller is an emeritus professor of psychology at the University of New South Wales in Australia,  who is considered the father of cognitive load theory. You heard from him in episode one. I also spoke to him about the limitations of working memory and how long-term memory can help us work around them. 

John Sweller

We can only hold two, three, maybe four—and process—units, elements of information at a time. And we can hold them in working memory for about 20 seconds. After that they're gone, unless we keep rehearsing them to keep them in working memory. 

So how do we function? Well, we can transfer information from working memory into long-term memory. And we're constantly doing that. And our long-term memory holds unlimited amounts of information—or there must be limits, we don't know where they are—for unlimited amounts of time. It's enormous.

We are transformed once we have information in long-term memory, which we can transfer back into working memory. We become different people. It's a truism to say education is transformational. That's why it's transformational.

Natalie Wexler

So, the more information we have stored in long-term memory, the easier it is to acquire new information. 

But how do we get information into long-term memory in the first place? What really helps is to be able to attach meaning to the new information—and one good way to do that is to explain it to someone else. You might just tell someone about it. Or you could…write about it.

The next hurdle is getting the information out of long-term memory when we need it. In other words, retrieving it—as you may remember from Episode 2. Hundreds of studies have shown that the more you practice retrieving an item of information that you may have slightly forgotten, the more likely you are to be able to retrieve it in the future. One way to engage in retrieval practice is through a quiz or a test. In fact, retrieval practice is sometimes called “the testing effect.”

But an iconic retrieval practice experiment didn’t involve testing or quizzing. It involved writing. Let’s hear from one of the researchers who worked on that study: Jeffrey Karpicke, a professor at Purdue University.

Jeff Karpicke

We had college students read very brief texts about different science topics. And in a retrieval condition, we had them read the material and then set the material aside and take five, ten minutes or so to just write down as much as they could remember in any order. And that's what we call a free recall task.

Natalie Wexler

Other groups of students in the experiment were asked to read the text once, read it twice, or create a concept map of the text while looking at it. Then, all the groups were tested a week later. The result? Students who had put the text aside and written down everything they could remember were able to recall far more of the concepts than students in the other groups. So the “testing effect” could also be called a “writing effect.”

Education researchers have also done experiments with students in 1st through 12th grade, known as “write-to-learn” studies. These researchers have looked at what happens when students write about what they’re learning in math, science, and social studies. A 2020 review of 56 of these studies found that on average, when students write about what they’re learning—in any subject and at any grade level—they retain more information and understand it better. 

We’ve talked about why writing might help you retain information and retrieve it when you need it. But why would it help you understand the information better? That has to do with another concept from cognitive science called elaboration. It’s been found that when students ask and answer certain kinds of questions—why, how, can you give an example—their comprehension becomes deeper. Here’s Jeff Karpicke again.

Jeff Karpicke

So elaboration, I think really broadly, just refers to any kind of activity where you are, or a student is going beyond the material that they are given. Elaboration can mean taking content and organizing it in a way. It can mean taking content and focusing on stuff—stuff is maybe not the technical term, but I think you understand what I mean—information in there that is unique and distinctive. It can refer to instances where a learner is relating that material to themselves or bringing their own, like, additional personal knowledge to the material. 

Elaboration can mean a lot of things. And elaboration is—it’s a goal, I think, that we go for when we're trying to enhance student learning. We want students to engage in deep, meaningful, thoughtful processing. And that is what we really mean when we're talking about elaboration.

Natalie Wexler

You can engage in elaboration orally. But again, it’s also something we do when we write, if we’re writing well.

Jeff Karpicke

I mean, a lot of writing tasks are probably naturally elaborative. They're not just asking a learner to just kind of repeat something that they had seen before, but they're asking them to maybe make some new connections or put something together in a new way. 

Natalie Wexler

Now we’re up to three possible benefits from writing: transferring new information to long-term memory, retrieval practice, and elaboration. But wait, there’s more. 

Learning to write well can also boost students’ comprehension of written language. That’s because of a potential barrier to reading comprehension that doesn’t get enough attention: syntax, or sentence structure. The sentence structure of written language is almost always more complex than the sentence structure we use in conversation.

For example, written language is more likely to use the passive voice, or constructions like an appositive—that means a phrase describing a noun. In the sentence John Sweller, a prominent cognitive psychologist, is considered the father of cognitive load theory, “a prominent cognitive psychologist” is the appositive. That’s not the way most of us talk, but you see it a lot in written language.

There are also certain words, like moreover or despite, that appear frequently in writing but not in conversation. If students aren’t familiar with that kind of syntax and vocabulary, it can seriously interfere with their comprehension of text. And they’re unlikely to become familiar with it just from carrying on conversations.

Here’s where writing provides an additional benefit, beyond the ones supported by cognitive science research. If students are taught how to use complex syntax and vocabulary in their own writing, they’re in a much better position to understand it when they encounter it in their reading.

It’s clear that writing can enhance students’ learning. But it doesn’t always have that effect. Remember those “write-to-learn” studies that showed students learn more when they write about what they’re learning? That was true on average. But in almost one-fifth of those studies, there was a negative effect. In other words, when students wrote about what they were learning, they learned less.

So, why would that be? Unfortunately, there wasn’t enough information in these studies to provide a clear answer. But another experiment may be able to shed light on the question. Jeffrey Karpicke, the Purdue researcher who conducted the “free recall” experiment with college students, also did some similar experiments with 4th graders. 

In one of those experiments, Jeff and his colleagues had all the kids read a text and asked one group to then put it aside and write down everything they could remember. That was the retrieval practice group. But instead of remembering more than the other groups, as the college students did, those kids performed no better on later tests than the 4th graders who hadn’t written anything down.

Jeff Karpicke

Those first experiments with young children, um there were no differences between the retrieval group and a group that just reread the material.

Natalie Wexler

So Jeff and his colleagues adjusted the experiment. Rather than just asking kids to write down everything they could remember, they provided more support, like asking them specific questions about what they had read. With these changes, those fourth graders actually were able to recall more than the other groups. 

Jeff Karpicke

There are two ingredients, I guess, to making retrieval practice a successful learning activity. One is that when you give people something to learn, and then you're going to ask them to engage in some kind of retrieval, they have to be successful. 

There's a flip side to it, which is that you want people to be successful when they're retrieving, but it can't be too easy. The retrieval needs to be a little challenging. And that's, that’s, that’s difficult. That's the challenge here is kind of—for educators, for students, for anyone who's going to engage in retrieval is—finding that sweet spot where I'm going to be successful at practicing retrieval, but it's not trivially easy.

Natalie Wexler

Retrieval practice can work for young students as well as older ones, but these studies show that you have to be careful. Jeff thinks one problem in their initial experiment was that the text the fourth graders were asked to read was too long. But another factor might be that writing imposes a tremendous cognitive load on working memory. That’s especially true if you’re an inexperienced writer. Remember, working memory can hold only four or five things for just a matter of seconds. When you're writing, you have to juggle way more than that—or try to.

If you’re young, or you come from a different alphabetic system, you might have to figure out how to form letters. Then there’s spelling, choosing what words to use, how to structure your sentences, how to organize your thoughts. And, of course, understanding the content you’re trying to write about. It’s pretty easy to become cognitively overwhelmed. Here’s John Sweller again.

John Sweller 

If you're still learning how to write letters and if you're still learning how to spell words and if that limited working memory that we were talking about earlier is consumed by that process, nothing else much happens, because it can't happen.

Natalie Wexler

College students, presumably, are pretty experienced writers, so the task of writing itself doesn’t impose that heavy a burden on their working memory. But the 4th graders in Jeff Karpicke’s later experiment? Maybe they were so overwhelmed by the task of writing that they didn’t have the cognitive capacity to understand or retain the information they were trying to write about. In other words, they experienced too much cognitive load.

Much of John Sweller’s work, and much of the research on cognitive load, has focused on math and science. A lot of the evidence comes from studies of what is known as the “worked example effect.” In those studies, some students are shown examples of math problems that have already been solved, or “worked,” while others are left to figure out how to solve the math problems on their own.

John Sweller

One of the major findings of cognitive load theory is the worked example effect, where if you, if you give people worked examples and have them study those worked examples, they learn more and become better problem solvers than if they engage in problem solving. We've got countless experiments demonstrating that, most of them in mathematics type areas, in science. But the same applies in humanities.

Natalie Wexler

In the context of writing, you might call a worked example a “mentor text”—a piece of exemplary writing that students can use as a model. Many teachers already use mentor texts to teach writing—and they have their value. But it’s possible to appreciate a great piece of writing without being able to produce something that resembles it. You might know, for example, that you should vary your sentence structure, or use subordinating conjunctions. But actually doing it can be challenging.

What all this adds up to is that if we want students to get those cognitive benefits from writing, it may not be enough just to ask them to write. We also need to modulate the heavy cognitive load that writing imposes on working memory. 

One way to do that is to have them write about topics they already know something about. That might sound obvious, but students are often asked to write about topics that they haven’t learned much about. Many schools use a separate writing curriculum that has its own topics and provides students with only limited information about any of them.

Even if their reading curriculum has writing prompts related to its content, that curriculum still might not provide much information. Many reading curricula jump from topic to topic, treating each of them superficially, because the focus is primarily on teaching comprehension skills. 

John Sweller 

It's not just American schools. I suspect it's worldwide. It's certainly the case here in Australia. You've got a student who's trying to work out how to write, how to spell, how to put words together. And then they have a topic that they know very little about. You can't do all of that at once.

Natalie Wexler

Students are—or should be—learning about topics in some depth in social studies and science. But writing is considered to be the province of English or literacy teachers, so students are rarely asked to write in those other subject areas, at least at the elementary level.

Of course, you could have students write about their own lives or experiences. They certainly know that content. But that would waste a golden opportunity to use writing to reinforce and deepen their knowledge of curriculum content. And it’s unlikely to equip them to do the kind of expository or analytical writing they’ll be expected to do later on.

So one way of managing the cognitive load of writing is to have students write about topics in the curriculum they’ve already acquired some knowledge about, either through reading or class discussion—ideally both. But often even that won't be enough, because writing is that hard.

What many students need is something cognitive scientists call “deliberate practice.” It’s a concept that can be applied to teach any complex skill. A teacher first breaks down the process into manageable chunks and gives students practice with the chunk they need. The teacher also gives prompt, targeted feedback. When students have more or less mastered one chunk, the teacher gives them another chunk, maintaining the kind of effort that leads to learning–also known as “desirable difficulty.”

Once students have acquired skills in long-term memory through deliberate practice, they have more capacity in working memory for higher-order thinking. For example, if a student who is learning to play the violin masters the fingering, she then has more capacity for things like expression and intonation.

So, how does this apply to writing? With foundational writing skills, the idea is to practice them to automaticity. Here’s John Sweller again.

John Sweller 

When something becomes automated, we can do it without using working memory anymore. It bypasses working memory. It goes directly from long-term memory, goes into working memory— but working memory can handle it without any difficulty. In other words, if when you're writing, if you have to carefully think about: how do I spell this word? Your writing skill is going to decrease immediately.

And of course, young kids are continuously in that situation. We as writers are in that situation periodically, but not often. Normally we write easily and fluently. We don't have to even think about spelling. We know it. It's automated.

Natalie Wexler

Higher-order writing skills, on the other hand, will never become completely automatic, but practice can help a lot. Let’s say a student doesn’t understand the difference between a complete sentence and a sentence fragment—which is the case for many students, including many in high school and even college. The traditional approach is to provide a definition like: a sentence needs a subject and a predicate and must express a complete thought. That’s too abstract for many kids. They might memorize the definition and still write in fragments rather than complete sentences. 

A better approach is to give students a series of phrases with no punctuation or capitalization, some of which are complete sentences and some of which are fragments. They need to determine which phrases fall into which category. If they practice that enough, with prompt feedback, they develop a gut-level sense of what makes a sentence a sentence.

Even a writing activity this basic can reinforce and deepen knowledge of content. Once they’ve identified the sentences and the fragments, the students then need to add information to turn the fragments into sentences. 

To do that, they need to retrieve information from long-term memory and put it into their own words. If the activity is embedded in curriculum content, it can serve as retrieval practice, of both the concept of a sentence and of the content students are writing about—and provide the opportunity for elaboration as well.

What I’m describing is taken from a method called The Writing Revolution. I didn’t create the method, but I’m the co-author of a book explaining it. The method begins at the sentence level, but it goes all the way through argumentative essays. Crucially, it teaches students how to create clear, linear outlines for paragraphs and essays. 

Even if you’re good at writing sentences, writing at length imposes its own heavy cognitive load. Your working memory may be trying to juggle questions like: What was I going to say next? or Did I already say that? If you can offload that burden to an outline, you have more working memory capacity to devote to expressing yourself in writing.

After thinking about all of this for a while, I came up with a hypothesis: If you have a curriculum that is rich in content beginning in the early grades, and you combine that with a method like The Writing Revolution, students will experience the potential cognitive benefits that come from writing. They’ll also become familiar with complex sentence structure, boosting their reading comprehension. They might learn to write well too. And that can happen even with the most challenging student populations.

So I went to a high-poverty school district that was combining The Writing Revolution method with a curriculum that was rich in content but—like many curricula—didn’t actually teach students how to write. 

Tune in to the next, and final, episode of Season 3 to find out what I saw when I went there.

The Knowledge Matters Podcast is co-hosted by me, Natalie Wexler, along with Dylan Wiliam and Doug Lemov. To learn more about my work, visit: www.nataliewexler.com. You can sign up there for my newsletter, Minding the Gap, and order my latest book, Beyond the Science of Reading.

This podcast is produced by the Knowledge Matters Campaign. Learn more about this episode and their work at: knowledgematterscampaign.org. There, you can find curriculum resources, blogs, and sign up for their periodic newsletter. 

To catch all of the Knowledge Matters Podcast, Season 3: Literacy and the Science of Learning, make sure you subscribe to the Knowledge Matters Podcast on Apple Podcasts, Spotify, or wherever you listen to podcasts. Thanks for listening.