Should We Teach with Decodable Text?

25 August, 2018
48 Comments

Teacher question:

Please share your thinking as well as research referencing the occasional use of decodable texts for small group reading instruction in grades K-2.

Shanahan response:

This is not a highly researched topic. There have been only a handful of studies into the effectiveness of decodable texts since the term was first used back in the 1980s. And, truth be told, they are kind of mess; with little evident agreement about what decodable text is, what it should be compared with, and what outcomes we should expect to derive from it.

Research has less solved the problem—is it helpful to use decodable texts with beginning readers—than demonstrated how complicated even simple ideas can be. (I’ve longed argued that in the social sciences this is likely the most profound impact of research. It forces us to operationalize constructs which often reveals how squishy soft our thinking is).

First issue… what is decodable text? Originally, the term was used relatively, to suggest a continuum (Juel & Roper-Schneider, 1985). Basically, phonics-oriented decodable texts used a preponderance of “words where all letters followed their major sound patterns.” The contrast was the basal readers of the time that employed a lower proportion of such words. Words like pet, big, nap, and dot would fit the decodable definition, while words like cow, pear, and come would not (cow follows a common spelling pattern, but not the major sound pattern for o, and pear and come are irregular).

How regular did the words have to be to merit designation as decodable? And, doesn’t decodability change with learning? As children know more and more spelling/pronunciation patterns, then those less frequent patterns become decodable, too, right?

Some researchers tried setting percentages of decodability and others worked with shifting definitions based upon kids’ learning. And, still others simply analyzed texts purported to be decodable and found a preponderance of simple, regular spelling patterns with short vowel sounds (Wolf, 2018).

You see the problem… if we all define decodable text in different ways, no matter what my research findings, someone can reject them simply by saying that wasn’t “what I meant by decodable.”

In that first study (Juel & Roper-Schneider, 1985), they found that the kids who had worked with relatively decodable texts were more able to decode at the end of grade 1. However, those kids also presumably got more phonics instruction during that time. One can’t tell from that study whether the difference was due to the phonics or the phonics plus decodable text.

Later studies, though with more exact definitions of decodability, found that decodable text either led to advantages or disadvantages. For example, Mesmer (2005) found that kids were more likely to try to decode decodable text (duh), but leveled texts (less decodable) led to greater fluency (Mesmer, 2010). Some studies (Cheatham & Allor, 2012; Compton, 2005) concur with the first Mesmer study, but that’s okay because others support the second (Priec-Mohr & Price, 2017). And, then there are those with mixed results (Chu & Chen, 2014).

The problem with all of these studies is that it is unclear what the right outcome would be. The ones that found the less decodable texts to be superior found that kids were able to read the leveled books more fluently in grade 1. However, phonics early on does bring with a bit of disfluency—because instead of just remembering a simple pattern or reading already known words, the phonics users have to figure out the unknown words (Barr, 1975). Many leveled readers can be read “fluently” even by non-readers because they don’t necessarily depend much on reading. And, the differing degrees to which students in the various conditions are taught phonics is a huge confound in the other direction.

Perhaps the best study of the problem was conducted by Jenkins and colleagues (2004). They compared the effectiveness of text with 85% decodability with text that had only 11% decodability (and these percentages were based on the texts’ match with the patterns that students had been taught). And they found? That degree of decodability made no difference. It made no difference in decoding, word reading, passage reading, or reading comprehension. (The two groups that received instruction with texts at these two levels of decodability both outperformed a group that received no additional instruction, so the teaching was effective in both cases).

Of course, there are a number of studies and evaluations of phonics instruction in which the program under investigation included instruction in decodable text (e.g., Foorman, Francis, Fletcher, Schatschneider, et al., 1998; or look at the phonics studies synthesized in the National Reading Panel Report or the various reports of What Works Clearinghouse). These phonics programs were successful, but there is no evidence that they would have been any less successful without the decodable texts. To be perfectly honest, evaluations of phonics programs that lack such a component are usually effective, as.

This lack of positive support may be surprising. If phonics instruction were useful—and it is—one would think that opportunities for concentrated practice of those phonics skills would be beneficial. And, perhaps it is. But proving that is going to require a lot more study—and with much more care in considering the nature of the decodable and less decodable texts (and with more attention to issues like repetition—Mesmer, Cunningham, & Hiebert, 2010).

Two final thoughts:

First, given that there have been no evident negative effects associated with the use of decodable text within some of the successful phonics programs, I think it’s safe to say that it is okay to use such materials as a very small part of instruction. Perhaps there is a small practice effect, and perhaps there isn’t—but such teaching isn’t likely to hurt either, at least when kept minimal.

Second, English is complex and the sounds associated with particular letters and letter combinations depend upon the letter’s position in syllables, morphology, and etymology. That’s why so much is made these days of “statistical learning.”

Readers start to discern that b is more often associated with the /b/ sound in words like big and bad much more often than it will serve as a silent letter (such as in words like bomb or climb), and readers learn to respond accordingly. If unsure of the pronunciation of a b word, go with /b/; you’re more likely to get it right. Presenting students with lots of decodable text, text that’s much more regular that normal text, might mess up some of these cognitive calculations. Dick Venezky and Dale Johnson (1973) long ago showed that adults attribute sounds to letters in proportions more reflective of their appearance in children’s primers than of the actual proportions in which they appear in English more generally. (This is a problem for both leveled readers and decodable texts.)

I think it’s okay to use decodable texts as part of phonics instruction, but such practice should be severely limited, and even beginning readers should be reading more than decodable texts.

Comments

See what others have to say about this topic.

Meg Galloway Aug 25, 2018 08:18 PM

Can you comment on the use of decidable texts (even ones like Mac and Tab, etc.) for young students who have dyslexia and have already failed to learn to read with classroom levered texts? In my experience, most dyslexia remediation programs include decidable text, though some like the Slingerland Multisensory Language Approach uses “meaningful phrases” (verb or noun phrases) pretaught before reading non-controlled cannneced text to allow words above the student’s level to be read -or at least recognized-with accuracy and fluency. Another reason to get out of decidable texts quickly, perhaps, is the low vocabulary level. Can you comment on that? Thank you.

Meg Galloway Aug 25, 2018 08:19 PM

Sorry for typos! So much for respondents by with a phone.

Jeanette Aug 25, 2018 10:47 PM

Tim,

Thank you for the interesting post on decodable text. I am curious as to what other kinds of text you would advocate to teach reading to beginners if decodable text should be "severely limited?" I've always learned that decodable text are important so that our students can practice what they are learning in explicit, systematic phonics instruction. (And certainly some students need more of this kind of practice than others). Of course reading aloud to students is important as well, so we are exposing them to more complex text and vocabulary, and helping them improve their listening comprehension. However, if we are talking about getting beginning students to practice reading on their own, what other kind of books besides decodables would you recommend for them to practice their newly learned skills? Also, do you recommend something different for a struggling reader as opposed to a typically developing reader?

Thank you,

Jeanette

Tim Shanahan Aug 26, 2018 12:57 AM

Meg and Jeanette— no matter what you’ve been told or which group of kids you are concerned about, there is no research evidence supporting the idea that decidable text works.

Tim

Harriett Janetos Aug 26, 2018 02:03 AM

Thanks for such an interesting blog. This is a timely topic because many K-1 teachers at my school are concerned about the use of predictable text in beginning reading. Here's an excerpt I recently sent to Fountas and Pinnell, whose Benchmark Assessments my district uses.

"I've just read your article in this month's The Reading Teacher, and it's provided me a good opportunity to communicate something I've thought about ever since receiving training in your Benchmark Assessments. You state:

'Most of us have taught in times of shifting literacy mandates and polarizing views that can leave us feeling frustrated, overwhelmed, confused, and even cynical . . . We can maintain our equilibrium in the face of these challenges when we are rooted in a set of worthy values, a solid foundation of research and data, evidence from observable reading and writing behaviors of our students over time, and good common sense.'

These sentiments exactly describe my experiences and feelings over the past three years since my district has started using your program. My frustration is rooted in your A-D predictable texts, which I believe do not have a "solid foundation of research and data", and the co-writers of the Common Core Foundational Skills--Louisa Moats and Marilyn Adams--agree.

Louisa Moats, in Learning to Read, says:

'These books are called “predictable” because once the repetitious theme or pattern is learned, the pictures are reviewed, and the child is rehearsed, he or she can memorize and “pretend read” with enjoyment. When the child is excessively dependent on context to read, he or she often cannot read the words of the story if they are removed from the book itself. Predictable books are not suitable for beginning instruction in sounding out and blending words because there is not enough control of the phonic patterns to allow the child to practice what he or she has learned . . . Predictable texts do not support children’s growing understanding of the alphabetic principles of English.'

Marilyn Adams, in an email to me, says:

'The problem with books like these is that there is little -- or in the case of the last word, negative -- incentive for the kids to pay attention to the composition of the words--and first-grade reading is critically about getting one’s arms around same.'

My experiences as a reading specialist working with struggling first and second graders support these views. And now that our K teachers are required to use your assessments, I find that I have to give my own decodable assessments so that I can see whether students can decode connected text and not simply read sight words and pictures."

The Reading League Aug 28, 2018 06:05 PM

Dear Dr. Shanahan,
We were glad to read your blog entry on the question related to teaching with decodable text. The topic is of great interest to us because, as you point out, research designed to tease out the influence of different text types on beginning reading is sparse, the research that has been done hasn’t isolated the influence of text types very well, and for this and other reasons, study findings are murky when we try to make use of them. We’ve recently received questions from others who read your blog entry as they try to make meaning of your recommendations, so we wanted to respond to them within the context of your blog so our questions, discussions, and recommendations remain linked.
To lay out some initial logic, if one accepts that teaching phonics to beginning readers is well-supported by research, then the value of decodable text can be tied to this research, as well as tied more broadly to learning theory research (and as a bonus, to logic!).
To first draw on research directly related to beginning reading before we go broad, if beginning readers learn that letter “a” says /a/, “m” says /m/, “n” says /n/, etc., and we initially give our readers texts mostly constrained to these learned letter-sound correspondences, this will help them apply what they have learned to sound out words, where accuracy should be pretty high. As children learn new letter-sound correspondences, these can be distributed among the previously learned letter-sound correspondences, thus broadening the word selection used in the texts while still promoting high word reading accuracy. Although there is a temporary phonetic restriction in the words featured in these decodable texts, the payoff is a “high signal, low noise” instructional condition where guessing at words isn’t required, and cues outside of the words themselves are generally not needed. This initial groundwork for learning to read is important, but of course, it is only a short step on the continuous trail in the development of literacy. This is the logic part of how children are taught to decode words.
The research supporting the use of decodable text then relates to a broader foundation of research related to learning theories that include the value of controlling the complexity and organization of knowledge introduced to beginning learners, the facilitation of deliberate practice in assimilating component skills, etc. So back to the point made above. If there is a strong research base to teaching beginning readers letter-sound correspondences – then deliberately choosing texts to facilitate readers in applying this knowledge only makes sense. The alternate plan is to deliberately avoid this sort of application, which seems neither logical, nor supported by learning theory research, even if the use of decodable texts isn’t yet unequivocally backed up by study findings. The question isn't which text type is better per se, but which text has value related to accomplishing a given learning objective. If the objective is to teach children how to blend letter-sound correspondences to read words in connected text, then decodable text seems the way to go - at least at the beginning, until the code is learned. Of course, students should have access to richer literature (read aloud, shared reading, etc.) at other times of the day in order build vocabulary, background knowledge, and the value of literature.
In the absence of studies we'd want to see that would unequivocally support the claimed benefits of using decodable text, teachers need to make decisions based on what research there is and the learning objectives they have. We think the recommendation from your blog “to severely limit” the use of decodable texts is what is kicking up the dust. Out of context, it could imply that the use of any decodable text may be harmful, but your recommendation is embedded within a complex sentence where you make a few points simultaneously, so it's worth untangling the sentence. It’s one of those times when a close read is warranted, along with a comprehension check from the author, provided he’s still available for comment.
The first idea to be parsed from the aforementioned dust-kicking sentence is that using decodable texts as part of phonics instruction is okay. You do make this point, so teachers who teach phonics where the use of decodable text is considered an active ingredient can collectively heave a sigh of relief. The second point - that teachers should severely limit the use of decodable texts – means (we hope) that you want teachers to know that decodable text should comprise a small portion of the instructional day. In an attempt to define this point in a way that is useful to instructional decision-making, we’ll speculate that this limit could be 10-20 minutes or so of reading decodable text per day linked to phonics instruction, which is in line with many well-designed reading intervention studies yielding positive effects.
A final point - We agree that a lack of an operational definition for decodable text makes the research-to-practice road even bumpier, but no text type is really well-operationalized in research, so in the absence of this, we'll still need to rely on researchers to describe what they mean by decodable texts, leveled texts, etc. and then make inferences based on those pesky details related to the operationalization of constructs, research design, and all those other abstract things that unfortunately do matter to our understanding of how to teach beginning reading.
Here is a link to one of our free teacher education events about the topic of Lesson-to-Text Match: https://www.youtube.com/watch?v=uRhKsFHMEuw&t=2s
Sincerely,
The Reading League

Tim Shanahan Aug 29, 2018 02:43 AM

Reading League— Your logic is impeccable, but your empiricism is weak.

No matter how carefully you lay out the logic, there must be a reason why studies aren’t finding that, as you propose, if explicit phonics works then decodable text must work, too. You need to explain why phonics programs that dont include decodable text still work (and as far as I can tell work as well as phonics programs without such text), why direct tests of the proposition haven’t upheld it, or why we should ignore research evidence concerning the potential damage it does.

I’ve learned in my life that many things are logical: the earth stands still because I don’t go spinning off of it, the sun goes around the earth because I can watch it come up in the East in the morning and set in West in the evening, and if ships go too close to the edge of our obviously flat earth they’ll be destroyed.

We use systematic study to allow us to get beyond those beliefs in what we find to be logical, but that actually are incorrect. So far, research has not been particularly kind to the idea of only exposing kids to decodable text, though I understanfpd that is upsetting (and while I diasagree don’t think I’m hoping you’ll venture too close to the edge of the earth).

If you decode my blog entry carefully, you’ll see that I defended some brief practice with such artificial texts in pursuit of your logic, but cautioned against the kind of sole use of such text that you espouse here because of the existing empirical evidence. I’m always perturbed by phonics proponents who champion research when it supports their belief and reject it when it doesn’t. That seems more like politics or religion than any science of pedagogy. Who knows? Someday the preponderance of evidence may go the other way and I’ll need to change my mind to be sure I’m serving kids best... it must be more comfortable to know that no matter what the evidence may show you will always be able to support the one true pedagogy and that you don’t need to be concerned about what works best or what does long term harm. I’d love to have that level of self satisfaction... instead I’ll keep trying to get it right, and I’ll trust that science will continue to tease out such answers.

Good luck.

Elizabeth Nonweiler Aug 30, 2018 11:06 AM

http://www.shanahanonliteracy.com/blog/should-we-teach-with-decodable-text#sthash.HETlHn8f.eFWDOl0W.dpbsskip to main content

Should We Teach with Decodable Text?

It seems to be impossible to copy the content.
My response:

First, my definition of a decodable text is a text the reader can decode independently without guessing what the words are. To be realistic, I call a book decodable if the reader can decode almost every word. So, if you can decode almost every word in a text on advanced physics, that text is decodable for you. However, for a five year old and just beginning to learn to read, very few books are decodable.

I know of a small study by Marlynne Grant that showed improvement when decodable books were introduced to a phonics programme that was already successful:
http://www.nonweiler.demon.co.uk/page26a.html
“In 1999 we used structured Snappy Lesson® teaching as before but included the Phonics First Books as children’s first experience of reading books for themselves. The results showed an additional end of year gain of 5 months reading age attributable to the Phonics First Books. Children’s first experience of reading books with their phonics and blending was successful. So that phonics and blending were
being reinforced and established as the primary strategy for reading, without children having to guess or memorise. This avoided the children having to use mixed reading strategies to read their first books, thus developing wrong neural circuits.”

I have anecdotal evidence from my own experience. Here is one example: It is about a little girl I met twice. When she was five years old and had been at school for more than seven months in England, since schools were expected to teach synthetic phonics, I showed a book that was easy to decode at an early stage. She looked at the pictures and made up a story with no relation to the text. I asked her to read some single words out of context and she read them accurately. I found she had been sent home with four books that were not decodable for her, with instructions to parents to talk about the pictures, ask children to think what the text might be and, lastly, to try sounding out simple words, i.e., phonics as a last resort. So, when presented with a story book, she thought she was meant to make up a story. Months later, I met her again and asked her to read another book with words I was sure she could decode. This time she read the first word accurately and then guessed the second wrongly, based on the first letter and the context. Added to the way she had been taught to read texts, she had problems with concentration and was later diagnosed with ADHD. When I asked her to try again, she threw the book on the floor and ran away. I understood. Why would anyone with poor concentration want to make the effort to read a word from beginning to end, if she has been encouraged to guess, found guessing easier than phonics and, as a result, had had little practice using the phonics she had been taught?

So, until reliable evidence proves me wrong, I will continue to promote the use of decodable books for beginning readers.

Catheryn Sale Sep 06, 2018 07:49 AM

Hi Tim
I have shared your blog with Reading Recovery teachers I work with. Our discussions lead us to believe that texts for emergent and early reading (5-8 years) have control of some sort (words, letters, sounds, punctuation, language structures and so on); that writers and publishers produce materials that follow their assumptions about reading acquisition; and the exposure our children have to texts influences how they come to build their reading processing systems. Hopefully our children become literate and can read many different texts. The teachers I work with when asked: what are the characteristics of books do you look for when choosing books for your children to read? in a quick 5 minute paired sharing and 5 minutes bringing ideas together - gave responses in these categories: child experience, language and print. A) Child experience (the child needs to be successful, get enjoyment, make connections with the story, link to a personal interest, some children will needs more natural language structures, the child's experience and subject matter is important, teachers may like to follow a series of characters that interest children (Baby Bear books), some children like humourous books, and sometimes book choice links to classroom topics). B) Language (there needs to be familiar vocabulary, repetitive language structures to encourage fluency, variety of sentence and language structures, variety of text types (genre), repeated vocabulary for early readers). C) Print (There needs to be clear print layout, a variety of fonts - but not overlarge, volume of print, variety of print, use of punctuation - you may choose texts with more direct speech to concentrate on punctuation usage).
Our understanding is that the purpose (intention) of reading is to make meaning and the success (criteria) of reading is to comprehend.
Decodable texts seem to challenge engagement, motivation and the purpose of reading for us.
Of course phonics is taught (even systematically - determined by the teacher's analysis of the child's learning need) probably more synthetically in writing leading from phonology to orthography, and analytically in reading, for example, linking known to unknown. We conclude that we want variety of texts to be used in emergent literacy.
Question 1: What is your position on variety of texts to encourage flexibility in problem-solving?
The teacher, of course, helps the child to co-construct literacy acquisition. We don't have a 'program' we design and teach what we do, under a theory of acquisition selecting from many different procedures.
Question 2: Do you think teachers are capable of being effective with this approach or do we need direct prescriptive instructional programs to follow?

Mandy Cuneo Sep 01, 2019 01:12 PM

Learning to decode is one essential part of reading that research shows if taught systematically and explicitly allows nearly all students to learn to read. The joy of reading is in reading books not just words or short phrases. Decodable texts are stories - there should be characters and a problem even at the simplest level, that children will be able to decode and read because they have the knowledge of the letter sound relationships needed to read that book and can understand the situation of the character.
The problem with predictable texts is that a significant group of children use the pictures and the repeated sentence structures to appear to learn to read. These children often have poorer phonemic awareness, difficulty blending individual sounds to make words and looking at pictures and memorising sentence patterns is easier. The familiar contexts of the texts means the meaning is easier to glean. This stops them from developing skills in decoding and masks reading problems. Typically, children with this type of problem will become apparent in Year 3 or 4 when the content is on unfamiliar topics and the vocabulary and sentence structures more complex and reading is expected to be more fluent and independent.
Children who intuitively decode and encode words, they notice letter patterns, syllables, prefixes and suffixes, are more likely to learn to read seamlessly but may not comprehend so well. However, some children will be able to read, retell and comprehend through the process of reading increasingly complex predictable texts until they are able to independently read.
This however is not the only part of literacy and English education, even in the early years. Being read to is an important part of learning to read. These texts would not be decodable texts but rather quality literature. Parents and siblings read to children. They also should not be decodable texts.
Children who intuitively decode will move much more quickly though the decodable texts and will likely simultaneously, begin to read the quality literature available to them in the classroom, library and their home.