S7 E3 - What Does Research Really Say About Decodable Texts?

Subscribe to the Podcast

Well, hello, hello. Welcome to the Structured Literacy Podcast recorded here in Pataway, Burnie, Tasmania, on the lands of the Palawa people. I'm Jocelyn, and today we're kicking off a new two-part series, and we're going to be talking about a topic that in some corners is being debated, and that is decodable text.

An article was recently published in a teacher magazine by a prominent researcher and academic about how decodable texts can hinder reading progress. And while I don't disagree with the substance of this person's argument, that keeping students on fully decodable text for too long does them a disservice, I wish that this person had presented a more nuanced discussion of what current research does actually say, as well as helping teachers with some practical guidance so that they can make critical decisions in the classroom.

It's Not Black and White

Now, if you've been listening to this podcast for a while, you know that I'm not one for black and white thinking. I'm not interested in this versus that debates that create false dichotomies and leave teachers feeling like they have to pick a side. What I am interested in is understanding what the research actually tells us, knowing what it doesn't tell us, and thinking about how we can use that information to make sound decisions for the children we teach while being aware that there's more work to be done in this area.

Over this two-part series, we're going to explore decodable text from multiple angles. Today in episode one, we're going to dig into the research. What do we actually know?

What don't we know yet?

And importantly, what do we need to be cautious about when we're trying to be research informed?

In the next episode, we're going to get really practical and I'll walk you through how to think about text selection and progression in your classroom.

The Question of the Research

So the central question of the research that I reviewed was basically, what kinds of books should beginners actually read while they are learning phonics and early reading? We know we have decodable texts, which are those books engineered so that most of the words can be sounded out with the letter sound patterns that children have already been taught, plus a small set of high-frequency, irregular words that have been broken down and taught well. The orthography of these texts is tightly controlled. Pretty much it's everything is as it seems. And then we have levelled, or sometimes called predictable, texts. These are books organised by general difficulty, like sentence length, vocabulary, pictures, and the topic. They often contain many words that students can't decode with their current code knowledge. They rely on repetition, pictures, and context for the students to lift words from the page.

Now, what difference does it make if children practice reading decodables versus levelled texts, especially when both groups are getting phonics teachers? That's what we want to know. And I'm really sorry, but the answers are mixed and fairly nuanced. So before we dive in, I want to let you know I'm going to be using terms like "seem to indicate" or "may say," because there are some pretty significant limitations to the research currently available to us. What we don't want to do is be dogmatic about what the research says without understanding the nuance, because that approach doesn't serve us well. So we don't want to completely dismiss everything, and we don't want to take everything at face value. And I think that as professionals, we are able to make this distinction and have a professional conversation about what this might mean for us.

The Limitations

So in looking at the studies in this area around decodable text, there's lots of limitations. We see small sample sizes, we see teacher-selected participants, quasi-experimental designs where whole classes or schools are assigned to conditions, meaning that teacher enthusiasm and skill level differs across the classrooms. We also see program developers involved in evaluations, which can lead to conflicts of interest. Meta-analyses report that most studies included in the meta-analyses show moderate to serious risk of bias with non-randomisation, so it's not randomised and being selective about which students they're choosing, and often incomplete reporting. Some had difficulty even identifying and coding the different types of texts used in the studies because the authors didn't describe them in enough detail.

So the evidence we have here is informative. It could be a bit directional, but it's not rock solid. So instead of taking any single study as the final word and the evidence and leaning in to confirmation bias, "See, I told you so because that one study says," what I'm going to ask you to do is come with me on this little journey where we look at patterns across multiple studies and see where the studies agree with each other.

Also, most studies comparing decodable use to levelled text use run for a single year or less. So often just a term or a semester. So we don't have strong long-term experimental evidence that follows children from the start of schooling through to upper primary, where phonics instruction is consistent, but that different students read different texts to see what's happened. This means that we cannot say that a decodables-heavy, early diet causes problems later on, as has been suggested in the last little while. We also can't say that starting with levelled texts leads to more robust reading down the track. These questions are open. We don't have settled facts here. So with that context in mind, let's look at what research does and doesn't tell us.

Murphy Odo, 2024

So what does research seem to indicate? Well, firstly, decodables help with decoding. A 2024 meta-analysis by Murphy Odo focused specifically on decodable texts with children without reading disabilities. In that meta-analysis, they found a small average advantage for conditions using decodables in word reading. So reading decodable texts help improve word reading, and a slightly bigger effect for pseudo-word reading, so nonsense words. Individual studies align with this.

Jenkins et al., 2004

So Jenkins and colleagues in 2004 looked at first-grade tutoring using decodable texts versus less controlled texts within a systematic synthetic phonics approach. The decodable conditions where the students read decodable texts showed modest benefits in decoding, especially for pseudo-words. So we have a correlation across what we're seeing here.

Mesmer, 2005 & Mesmer, 2009

Mesmer in 2005 and 2009 found that first graders reading more decodable texts showed higher accuracy and more phonics-based attempts at unfamiliar words.

Lee, 2020

And even Lee and Lee in 2020, and I say even because these were Korean sixth grade students learning English, found that decodable English texts led to greater gains in word identification than high-frequency non-decodable texts. And in reading this, for me, it's a little bit of a "Duh, Fred!" moment, because the whole point of decodable texts is to help students get better at decoding words, at sounding them out, at practicing using the patterns they've just been learning.

What it Means

So what we're seeing in the research is indicating that when texts closely match the phonics patterns children have been taught, decodable texts support more accurate word reading and better phonological decoding, sounding out with the sounds, than less controlled texts, particularly in the early stages. So that's one thing we can say.

What we also see from looking at a range of studies is that different texts seem to prompt different reading strategies. And here's what's a bit interesting, it's that children do different things with different texts. And again, I think for those of us with experience in using a range of texts with students of different ages, we're like, well, yep, we knew that already. In decodable texts, children more often try to sound out words using letter sound knowledge, and the errors they make are more phonological. They're related to the sounds that they can see represented on the page. In level or less-decodable texts, the mistakes that children make are often linked to context and figuring out what makes sense based on what the picture is on the page. So they might say house when the word says home, but both words look sort of similar to a beginning reader and it matches what the picture says.

The important thing here is that decodable texts anchor children in the alphabetic principle. Less decodable texts invite them to guess. And that's the message that we've been hearing over the last few years that we can be confident about.

Now there is some examination of whether different sorts of texts help with comprehension, but this is less useful to us because... actually, I'm going to tell you why in just a moment. So stick with me while we look at set for variability and where this sits in the research.

Set for Variability

So there's a theoretical idea in reading research called set for variability, where a reader generates an approximate pronunciation from the letters on the page. Then when what they've said doesn't match what makes sense, they notice, they go back to the word, they look at the code on the page, they also think about context, they think about morphology and all these ideas, and they go, hang on, let me tidy that up. So I'll give you an example. So a student might be reading, and the sentence might actually say, "We read the book today." And they say, "We read the book today." And then they go, "Well, hang on, that doesn't really make sense. We read the book today. We re- Oh, we read the book today." That is set for variability in action. That student isn't guessing, they're using code knowledge to have a go at getting the word off the page, recognising that something doesn't sound right and then fixing it up.

Now, why this is important is that there are some studies, one in particular, that looked at the impact of reading both decodable and levelled texts within an intervention program for slightly older students. And what they found was that students who had both types of texts did better than students who only had one type of text. So only had decodables or only had the levelled text. And here's my hypothesis, and I'm flagging this as a hypothesis with a complete big flashing sign that says OPINION ALERT, because I don't want to present this suggestion as established fact. Because that result is pretty interesting to me, and I'm trying to think about how that's come about and what might be getting us there.

So here's my thought. What if this worked because the decodable text grounded the student in effective decoding using phonics? But then the levelled text that presented words that were not immediately phonically decodable based on the student's knowledge provided an opportunity to develop set for variability. We know we don't want to keep students in only decodables, fully 100% decodables forever, because that does get in the way, they can never read anything else. But what could have been happening in this study where you got a better result?

So my hypothesis is that the two different texts for the student who had the foundations of decoding and phonics already established was that it worked on different things. The decodable text helped decoding, and the levelled text helped with building set for variability, which could have been like a pseudo-opportunity for set for variability training. The fact is we don't know because we don't know what happened when the students in that intervention study were reading that levelled text and came across a word that they hadn't seen before. We don't know whether they used multi-cueing or whether they actually were practicing set for variability. But that's my guess here.

But I want to be really clear about my thinking and observations based on what I've seen with students. Because remember, evidence-informed practice is not "we only do what is explicitly outlined in the research." It's "we take what we know from research, what research indicates." We don't think we know better from truly robust, replicated, longitudinal research. But we can also use our judgement. What do we know if we're experienced and we know how to get results? What do we know from what we see? And both of those things are only in alignment when the results of the students are also in alignment. So when we've got the data to show that what we're saying actually leads to positive outcomes and we're in alignment with what makes sense to us as a professional, cross-referencing this with the knowledge as we have it now from research, then we can be reasonably confident that we're not going to break the kids.

So I'm not saying there's only one way. I'm not saying run out now and give every Year 2 student two different types of texts. And we're going to get into why that's a bad idea, but it's an interesting discussion.

The problem is that these tests used in the research don't actually measure for set for variability or distinguish the student's accurate reading from pure guessing from the context. So what we've got here is a plausible explanation, but not a proven mechanism. So that's what I'm saying, don't run out and change everything you're doing, but just have a think about it.

Early Benefits Diminish

The other thing that research seems to indicate is that these early benefits of decodable texts diminish over time. So decodable texts
get the student to a strong foundation of decoding, but keeping the students in decodable text beyond the point they're needed doesn't get you where you want to go. So you need a different type of text. The two different texts have two different purposes. They're not on a continuum, but they do different jobs.

Switch Emphasis

So Hebert argues that after initial phonics is established, emphasis should switch to how texts are sequenced and curated based on difficulty like vocab and sentence length. So that idea that level texts were grounded in isn't necessarily diabolically wrong. It's just that it was introduced too early. Decodables have a clear impact early on. And then we can use other texts that are available to us to stretch students, to learn to flex words and fix up pronunciation. That also doesn't mean, though, that you need to do a running record and choose a level of a text based on what that test tells you. So I'm not suggesting we go back to running records and benchmark assessment.

Being Cautious

In interpreting all of this, there's some things we really need to be cautious about. Some of the studies suggested comprehension gains using levelled texts over decodable texts. But this all gets a little bit murky because we know how central background knowledge and vocabulary is to comprehension. So we don't know whether these gains were truly because reading the levelled text made the child a better decoder and therefore better able to understand, or whether there were more contextual clues or whatever was going on. So I think we want to be pretty cautious in thinking about that.

Now, over the past few years, the messaging in our profession has been decodables, decodables, decodables in early years classrooms. But I think what we're seeing in this recent work is pushing us away from this either/or thinking. We want to be cautious about black and white policies that say, thou shalt only ever do this in this grade and only ever do that in this other grade. Because different children develop these skills at different times. You could have a Foundation student who enters the classroom not actually needing fully decodable text all of the time. So you might have that student read the fully decodable texts to practice new code they're learning, but they can also read levelled texts because they're not guessing. They can use set for variability very early on.

You might also have a Year 3 or 4 student who's only just developing a solid foundation of the alphabetic principle and really needs purely decodable texts for independent reading for a little while longer. So it's not about grade here, and it's not about there being the one answer for every student. We have to understand reading development.

Another caution in this research is that many big phonics trials use strong phonics instruction, but don't use decodable text as connected text practice. Conversely, some balanced programs do use decodables in parts. So the research isn't necessarily comparing apples with apples. So even though there might be phonics positive results in the research, we can't automatically assume that that means anything about decodable texts. So we have to be clear about what part each practice plays in the instruction.

So here's what we can actually take away. And in the next episode of the podcast, I'm going to give you even more nuance about recommendations to use decodables wisely and effectively. Decodable text aligned with the phonics lessons you are teaching support better word reading and pseudo-word decoding in the early years. Just having a decodable text in the mix isn't going to do anything for you. So these texts push children to use phonics rather than guessing, and that's where we want them to be in these very early stages.

And just for context, I'm going to say for most students, Foundation and the first half of Year 1, roughly, once decoding is more secure, when students have a really good handle on the complex code, and I don't mean 257 graphemes, I mean 70-ish graphemes or so, and we can see that they are secure in the decoding, then adding in less decodable texts, I just generally call them mostly decodable, and using them with the right prompts and responses when there's an error, likely help students learn to flex and deal with more complex words as they develop set for variability, where they can tidy things up when they misread a word.

Research suggests using both types of texts may be more beneficial than just one, but there are specific times and points of development when each has their place. Very broadly, early on use decodables to get that decoding engine running. Then we can carefully layer in at appropriate times and for specific purposes richer text. And there'll be more details about that in the next episode.

Conclusion

So take heart from what the research is telling us that decodables do the job we think they're doing, which is to help decoding be stronger as long as we're carefully aligning the text with what we're actually teaching. But be cautious about over-interpreting any kind of short term results as being a result forever.

And above all, just avoid that decodables versus real books argument. Don't even go there. Books are just books, they have words in them, and we as teachers can be selective about how we assign reading material for different children to help maximise their journey.

So there you have it, a little bit of a look at what the research about decodable text currently tells us. And I'm looking forward to the day when we say, "Wow, isn't it great, we have all this robust, super well conducted research that we can feel confident in." And until that time, remember those three elements of evidence-informed practice, what the research says, carefully interpreted, with our experience, when we have it, when we know that we know how to get runs on the board, we can rely on our judgement. Always, always, always looking at the data, looking at the progress of students to make sure that we're actually sound in our decision making.

In the next episode, we're going to get super practical and I'm going to walk you through how to make some confident decisions in the classroom and in our schools. Thanks so much for joining me today. And remember that if you are being evidence informed, always with an eye on the data to make sure we're getting it done, you are not going to break the kids.

Until next time, happy teaching, everyone. Bye.

References:

Murphy Odo, D. (2024) The use of decodable texts in the teaching of reading in children without reading disabilities: a meta-analysis. Literacy, 58: 267–277. https://doi.org/10.1111/lit.12368.

Jenkins, Joseph & Peyton, Julia & Sanders, Elizabeth & Vadasy, Patricia. (2004). Effects of Reading Decodable Texts in Supplemental First-Grade Tutoring. Scientific Studies of Reading - SCI STUD READ. 8. 53-85. 10.1207/s1532799xssr0801_4.

Mesmer, H. A. E. (2005). Text Decodability and the First-grade Reader. Reading & Writing Quarterly, 21(1), 61–86. https://doi.org/10.1080/10573560590523667

Lee, J. S. (2020). An emerging path to English in Korea: Informal digital learning of English. In M. Dressman & R. W. Sadler (Eds.), The handbook of informal language learning (pp. 289–301). Wiley Blackwell.