When Learning Only Has To Be “Good Enough”
The purpose of these posts is to bring more of what is known from basic research on reading, language, and learning into decisions about how to teach children to read. In the Gretkzy [sic] post I used some humorous spelling errors to draw attention to the fact that the way that words are read changes as skill develops. Skilled reading involves parallel processing of multiple letters, using one’s accumulated knowledge of words. It is rapid and automatic, like a reflex, and sometimes it yields errors, such as reading Gretkzy as Gretzky. This process is almost wholly different from beginning reading. We teach children to read words letter by letter, left to right (let’s call this the serial decoding procedure) because it’s the on-ramp to reading, not because it’s what skilled reading involves (the parallel procedure).1 The goal isn’t for the child to become exceptionally good at using the serial procedure, which is slow and effortful and only works for a limited set of simple words. Rather, it is for the child to learn enough from using this procedure to begin reading words like skilled readers, in bigger chunks processed in parallel. The immature, serial decoding procedure enables the mature, parallel one.
Consider once again the hoary metaphor of learning to ride a bike: At the outset the learner needs explicit instruction and extensive practice using training wheels (or other supports). The goal is not to become highly skilled at riding this way; it is to learn enough to be able to transition to riding without them. The analogy to learning to read should be obvious.
Teachers’ backgrounds vary enormously and I am never sure who knows what. Was the observation that what beginning readers are taught to do is not the same as what skilled readers do a surprise to you? Let me know! It’s important. Thinking that the serial decoding procedure is also used in skilled reading would encourage overteaching because what could be the harm in it? Because serial decoding is actually different from skilled reading, however, overemphasizing it is a problem: it comes at the expense of activities that enable the child to make the transition. Returning to the bicycle analogy, staying with training wheels too long takes away from activities needed to make the transition to skilled riding, such as learning to maintain balance while braking or learning to dismount. Note also that there is such a thing as overlearning, in which one kind of behavior (such as riding with training wheels or reading letter by letter) is learned so well it interferes with learning a new one.2
Responses to the Gretkzy post suggest that many teachers are not aware of these differences between beginning and skilled reading. Because my observations are inconsistent with their understanding of the science of reading (SoR), some people will take this as another case of scientists not being able to agree on anything. The same thing happened with my posts about other questionable elements of the SoR such as phonemic awareness instruction. But no, disagreements among scientists aren’t the issue here. Researchers disagree about many things (it’s intrinsic to how science progresses) but there is little controversy about the basic facts I’ve described. The dissonance that some educators experience in reading these posts arises from a different source: discrepancies between the simplistic version of the “science of reading” they’ve been taught and aspects of the actual science. This gap is harmful for teachers and their students and needs fixing. That’s why I do these posts.
Why Serial Decoding Is So Prominent in the SoR
Here are some reasons why a person, or curriculum, or school district might overemphasize serial decoding. Are any of them relevant to you?
1. Serial decoding of words is the appropriate on-ramp for reading words, it was neglected in the WL approach, and teachers want to get it right. That is all good, but the additional question is: what is it the on-ramp to? Where is the reader going?
2. Serial decoding doesn’t exist in a vacuum; it’s tied to overemphasis on phonemic awareness. The SoR in this country (but not the UK) usually starts with teaching children to pronounce phonemes, as spoken units analogous to graphemes. This is based on a gross misunderstanding of research on phonemes and phonemic awareness, as discussed here. The SoR recommends extensive PA training on the mistaken view that it is necessary for learning grapheme-phoneme correspondences (phonics) and using them to decode words. It’s a mistake because treating spoken words as if they consist of phonemes depends on exposure to letters.
Instruction about phonemes, phonics, and serial decoding go together. In each case, I’ve argued (phonemes, phonics, here, there) that less is more: The reader needs to acquire knowledge that is good enough to segue into mature reading rather than achieving mastery of a training wheels procedure. Rather than 44 phonemes, a simple sound associated with each letter is good enough to start decoding. Rather than dozens of phonics rules, a much smaller number is sufficient to enable the child to crack the code. Rather than limiting the child to decoding words letter by letter, left to right, instruction can and should encourage using larger chunks.
3. The emphasis on serial decoding is an overcorrection of the most egregious Whole Language fallacy. The major tenets of the “science of reading” approach can be understood as reactions to specific problems with Whole Language. WL didn’t do an adequate job with teaching basic skills; the SoR focuses on extensive teaching of basic skills. WL assumed that phonics was only necessary for poor readers who didn’t respond to standard instruction; the SoR recognizes that learning connections between print and sound is essential for all readers. Whole Language assumed that learning to read is like learning a first, spoken language; the SoR recognizes that it’s not, insofar as reading requires explicit instruction at the outset, and so on.
The emphasis on serial decoding in the SoR is also a response to WL, specifically psychologist Ken Goodman’s “guessing game” idea. Goodman thought that readers only need to sample the letters in a word because the rest could be filled in using knowledge of the language and the topic of the text. (Marie Clay and Frank Smith made similar claims.) This guessing process was then incorporated in WL curricula, via the “3-cueing” method and other activities. The process Goodman described is a terrible on-ramp to reading. I won’t repeat the reasons here; the topic is covered elsewhere (e.g., my book). The most charitable thing one can say about guessing is that a skilled reader with extensive knowledge of the language and the topic of a text can do something roughly like that some of the time (e.g., skimming a text for a specific bit of information). The grievous error was taking this as the starting point for beginning readers.
What is the opposite of sampling the letters in words? Reading every letter from left to right, no skipping allowed. And so we arrive at serial decoding, which is indeed the right place for readers to start. But, with so much focus in the SoR on getting this right, the fact that it is only an interim step on the path to skilled reading has been overlooked.
4. Structured Literacy, LETRS, and other SoR resources focus on extensive instruction related to serial decoding, but say little about the transition to skilled reading. The Reading Rope and Ehri’s orthographic mapping account discuss what develops as children acquire reading skill, but not how that is achieved. They didn’t consider how explicit instruction enables implicit (unconscious) learning, how to coordinate explicit and implicit learning activities, or how the relations between them change with the development of reading skill.
In short, there are plenty of reasons why educators might get carried away with serial decoding. The extent to which this is happening isn’t known. Extensive instruction about phonemes, phonic rules, and serial decoding is apparently standard in many “science of reading” classrooms and incorporated in SoR curricula. Science of reading zealots on social media attempt to police fidelity to such curricula. Some teachers say they have little room to deviate from the prescribed progression, which extends over multiple years. We need hard evidence about what is happening in SoR classrooms and how it is affecting literacy outcomes. In the meantime the possibility of overteaching these basic skills is a realistic concern that needs to be taken seriously.
My next post will examine how instruction might promote the transition from serial decoding to the parallel process used in skilled reading. I don’t have a complete answer but there are some very promising directions to consider.
______________
¹ The distinction between letter and grapheme isn’t important for this discussion.
² There is an extensive research literature on what are called entrenchment or overlearning effects in which learning one thing extremely well makes it difficult to learn something new. For example, learning and using the phonology of one language makes it difficult to speak a second language without an accent.
