Blogs About Reading
Shanahan on Literacy
Literacy expert Timothy Shanahan shares best practices for teaching reading and writing. Dr. Shanahan is an internationally recognized professor of urban education and reading researcher who has extensive experience with children in inner-city schools and children with special needs. All posts are reprinted with permission from Shanahan on Literacy.
Which Texts for Teaching Reading: Decodable, Predictable, or Controlled Vocabulary?
I'm looking for help with information or resources about text types for early readers. We have decodable text, text with high-frequency words, and predictive text. It seems like a reasonable strategy to provide our fragile readers with more opportunities to read these low-complexity texts while we shore up issues with phonological awareness. Many teachers over the years have complained to me, an instructional coach, about a lack of available texts to meet the need of students as they proceed through the year and the text complexity increases. Even with popular curriculum programs, teachers usually have very limited options with beginning reader texts, and it isn't clear how the different types are meant to be used or the benefits of each. Do you have any advice for novice teachers about using different text types with our vulnerable readers?
The role of text in reading instruction has always been a big instructional question for parents and teachers — but it has not drawn the same kind of research interest as many other issues.
Nevertheless, the research does provide clues and it suggests that kids are likely to be best off in classrooms that provide them with a mix of these text types rather than a steady diet of any one of them — nor do I see the progression through these as developmental, with kids graduating from one kind of simplified text to another.
Let’s start with the basic premise that when someone is beginning to learn to read (or to learn almost anything else), the teacher is going to need to ease the way a bit; simplifying the process so the learner can actually engage. Every beginning text scheme that has been tried (e.g., controlled vocabulary readers, predictable texts, decodables, language experience stories, words in color, initial teaching alphabet) is not exactly like the texts that we read because it is a simplification made to allow youngsters to get started.
A second premise is that every scheme to simplify a process and to support beginners is somewhat misleading because the simplification is sure to make some important change to the process they are trying to learn. If a youngster is trying to ride a two-wheeler, training wheels might be a great place to start, but those extra wheels mislead these novices with regard to how to balance when riding.
There is nothing particularly unique about the potential negative impacts of simplifications and supports. Doctors who prescribe crutches are always concerned about potential nerve damage from the crutches and from the muscle atrophy that they might promote. Likewise, social policymakers worry over the role welfare plays in discouraging work. Neither group eschews these supports — they are needed — but they make serious efforts to try to avoid the downsides.
Sadly, advocates of various beginning reading schemes usually appear oblivious to the problems their favorite support systems present to beginning readers. They love the fact that the texts they champion allow kids to read early on. But they ignore the fact that their beginning reading texts — like everyone else’s — differ from the actual universe of texts that we read, and the more these texts diverge the greater the danger that they will be misleading to at least some kids.
Controlled vocabulary readers limit texts to a handful of words that are used repeatedly. New words are added gradually. The major approach to learning these words is memorization. Initially, because they start with so few words, these texts sound very stilted, but as kids memorize more and more words, controlled vocabulary texts sound more and more like language.
For instance, consider this sequence of pages from the old Dick and Jane Readers:
pg. 1: Dick
pg. 2: Jane
pg. 3: Dick and Jane
pg. 4: Dick and Jane run.
pg. 5: Jane and Dick run.
Pg. 6: Dick runs.
pg. 7: Jane runs.
pg. 8: Run, run, run.
Decodable texts on the other hand try to minimize the numbers of words that students won’t be able to decode. Initially, these texts too sound very artificial since the words they include are limited to very few letters and the same letters over and over. They, too, eventually become more like real language as they proceed.
Here is an example of what is meant by decodable based on the old Linguistic Readers. The idea would be to only introduce such a text once the students had some command of the following phonemes /k/, /m/, /f/. /l/, and the phonogram or word base at.
pg. 1 The cat is fat.
pg. 2: The mat is flat.
pg. 3: The fat cat sat on the mat.
Predictable texts start with more natural sounding language right out of the gate, but instead of requiring the novice readers to rely on memorized words or mastered letter sounds, the readers must depend upon repetition, context, and pictures to guess at words.
Here is the beginning of perhaps the most famous of all predictable books, Bill Martin’s Brown Bear, Brown Bear:
pg. 1: Brown Bear, Brown Bear what do you see?
pg. 2: I see a yellow duck looking at me.
pg. 3: Yellow duck, yellow duck what do you see?
pg. 4: I see a red bird looking at me.
All of these texts “work” in terms of getting kids started with reading.
However, each has problems. For instance, controlled vocabulary readers tend to steer kids towards guessing at unknown words based upon the words in their memory. Thus, the child who has memorized when, upon confronting unknown words like which or where, will tend to “read” these, too, as when (Barr, 1975; Biemiller, 1970) — not a very efficient strategy with an alphabetic language.
Decodable texts, too, can be problematic as they tend to steer kids away from meaning, and at times even away from real words. Kids who are used to strong phonics support and decodable texts tend to try to sound words out more than do other kids (Cheatham & Allor, 2012). But when this doesn’t work (and it doesn’t always work), these kids end up producing nonsense words (mispronunciations based on the sounds they know) or they balk and don’t even read words that they can’t decode easily (Barr, 1975; Biemiller, 1978).
And, predictable texts lead kids to read the pictures instead of the words — not a reading approach at all. In fact, studies show that, since the print isn’t really needed to make sense of many predictable books, the kids learn to ignore the words (Ehri, 1992; Whri & Sweet, 1991; Juel, 1991) and to rely mainly on the context — though such use of context is alien to proficient reading.
Which of these texts should you use?
A basic finding in educational psychology is that simplification or making diverse forms consistent for the purposes of teaching speed acquisition. But they also reduce the learners’ abilities to generalize or transfer these skills to the greater complexity of the actual forms that one needs to learn.
For example, it has been found that providing readers with consistent and simple sound-symbol relations speeds their learning — but when you then ask them to read a more diverse orthography such as the one we use in English, then they are less able to make the needed adjustments (Levin, Baum, & Bostwick, 1963; Levin & Watson, 1963).
If the goal is better beginning reading, this relying heavily on any one of these approaches is pretty smart. If, however, the goal is to teach reading — you know, the kind of reading you and I do — then heavy dependence on any one of these schemes is shortsighted.
Personally — based on my own experiences as a primary grade teacher — I would use all of these kinds of text. My thinking then, and my thinking now, is that the way to prevent someone from being hurt by over dependence on a crutch is to employ a variety of crutches; deriving the benefits of each, while trying to minimize potential damages.
It is very reasonable to employ decodable texts. It gives kids a chance to practice their phonics in a favorable text environment — an environment in which there aren’t likely to be many words that can’t be figured out easily.
But those “experts” who claim that kids should only read such texts for some length of time are just making that stuff up. Research is not particularly supportive of such an artificial text regime (Adams, 2009; Jenkins, et al., 2004; Levin, Baum & Bostwick, 1963; Levin & Watson, 1963; Price-Mohr & Price, 2018). “Teaching children to expect one-to-one consistent mapping of letters to sounds is not an effective way to promote transfer to decoding at later stages in learning to read” (Gibson & Levin, 1975, p. 7).
Please don’t misunderstand where that quote comes from; Eleanor Gibson and Harry Levin were big explicit phonics proponents in their day, but they also believed in following the research.
Those who pushback against any who would dare to present anything other than decodables text to kids often complain that anything else is too hard or discouraging to kids. But that’s where those other text simplifications — that have their own problems — come in.
Having young students reading both simple decodable texts along with controlled vocabulary readers keeps them from being overwhelmed by difficulty — but also prevents them from trying to depend upon memory or simple decoding so much that these approaches do damage.
The consistent use of any simplification regime runs the risk of teaching kids that scheme rather than how to read. The research in this area supports the idea of easing the way for beginning readers, but not so much that kids fail to develop good reading habits (analyzing the words) or that leads them to perseverate on disruptive views (such as the idea that words will be cued by pictures or that they should expect the spelling system to depend upon one-to-one mapping of letters and sounds).
I’m not a big fan of predictable text in this equation, because it discourages kids from looking at the words. However, even these texts are okay for very brief times. In my classrooms, kids worked with these kinds of texts once a week or less — along with the basal readers, linguistic readers, and language experiences stories that made up the lion’s share of their reading. Predictable texts are fun, they allow a level of early success unmatched by the other texts and they do encourage kids to try to keep reading meaningful and fluent; nothing wrong with any of that.
For a long time, I’ve advocated for substantial amounts of instructional time devoted to decoding, fluency, comprehension, and writing. Decodable texts can be an important part of the decoding instruction, but I’d make controlled vocabulary readers the base of my reading comprehension instruction. Predictable texts can be a lot of fun, too, once in a while.