Why Synthetic (Bottom-Up) Reading Instruction?
Updated: Apr 4
All words consist of individual sounds, called “phonemes,” which seamlessly blend together when the word is spoken. The word SHEEP, for example, is a blend of three such phonemes: the sound of SH (symbolically /sh/), the sound of EE (symbolically /E/ or “long E”), and the sound of P (symbolically /p/).
So… SHEEP = /sh/ + /E/ + /p/
If you articulate these 3 phonemes, in order, blending them one after the other, you can’t avoid pronouncing the word SHEEP. If, in addition, the word SHEEP is in your spoken vocabulary, you’ve just comprehended, and therefore read, a written (encoded) word. (There are 41- 44 phonemes in spoken English, depending on where you’re from. For a list featuring US pronunciation, see here.)
Anyone faced with teaching a child to read must make a fundamental choice. Shall I teach reading as a top-down skill, starting with the final product (whole words) and then have the child analyze those words to discover all the letter/sound relationships they reveal? Or shall I teach reading as a bottom-up skill, starting, not with whole words, but with individual letter/sound relationships, and then have the child blend those individual sounds to create (synthesize) whole words?
Advocates of top-down instruction are faced with a dilemma from the very beginning: if the child can’t yet read any words, how can instruction start with whole words? It’s at this point that the “sight word” becomes a necessity. It’s a word that must be deliberately rote-memorized by the beginner, based on the word’s visual characteristics rather than on the sound value of its individual letters – something akin to memorizing a phone number or a meaningless password like Sj4h#k.
Concretely, how are children expected to memorize sight words? Well, the words are placed on “word walls” where children see them throughout the day. Children are also given repetitive (“patterned,” “predictable”) books to “read.” These books are all structured the same way. For example, if the goal is to teach the sight words I, TO, EAT, and LIKE, then each page of the “story” will have the text “I like to eat X” where X progressively changes, as the child turns the pages. So page 1 might be "I like to eat apples," page 2, "I like to eat pizza," page 3, "I like to eat bananas," and so on for 8-12 pages. By seeing these four words repeatedly on each page, the hope is the child will manage to memorize them.
But how are children expected to read what’s being eaten (the apples, pizza, bananas, and so on)? These are complex words for a beginner, and they appear only a single time in the story. The solution is this: each page has a picture of that exact food. It’s at this point that “guessing” becomes a method for reading. Children are actively encouraged to guess the meaning of an unknown word based on:
an accompanying picture
what might “sound right”
the word’s first letter
Whole Language (1975 – 2000) and Balanced Literacy (2000 – present) have dominated reading instruction in schools, public and private, for nearly half a century. (Read the history here.) Both are top-down methods that start with sight words, guessing, and “invented” spelling. The main difference between them is that top-down Balanced Literacy includes some top-down phonics.
By “top-down phonics” I mean any type of phonics (e.g. Analytic Phonics, Analogy Phonics, Onset-and-Rime Phonics) that requires the child to master a large number of sight words before such phonics can commence. Analytic Phonics requires something to “analyze.” Analogy and Onset-Rime Phonics require something from which to “analogize.” That “something” is sight words. (For a discussion of these types of phonics, see here.)
In contrast to all these top-down strategies, there is one, and only one, bottom-up method: Synthetic Phonics. Because it starts with letters and the sounds those letters symbolize – and then blends those sounds to create whole words – there is no need for either sight words or guessing. (For a more complete description of Synthetic Phonics and its contrast with Balanced Literacy, see here.)
Given today’s choice between top-down Balanced Literacy and bottom-up Synthetic Phonics, what should a reading teacher or parent choose? The following will help you decide.
The Four Advantages of Bottom-Up Instruction
Advantage #1: Today, almost everyone agrees that a child needs Phonemic Awareness (PA) if he or she is ever to become a skilled reader. Illiterate people, both children and adults, can’t hear individual phonemes in spoken words without specific training. So, what’s the easiest, quickest, and most direct way to help a child become aware of the phonemes that make up individual spoken words? It’s this: tell the child directly what 4 or 5 of them are, and have the child repeat them, in isolation, until she’s saying each one accurately. At the same time, show her a letter that symbolizes each one. With these 4 or 5 letter-sound correspondences in place, start showing her how to blend them into whole words.
Example: Teach the 5 phonemes /m/, /n/, /s/, /a/, and /e/ and the letters that represent these sounds: M, N, S, A, and E. (Note: the 2 vowel sounds are the “short” sounds you can hear at the beginning of the words APPLE and END.) Now show her how to blend these sounds into the words MAN, MAM, MEN, SAM, ANN, NAN, MESS, SASS, MASS, AM, AN, ASS, SEN, SAN, NES, NAS, EM, EN, ES, SEM, MEM, NAM, SESS, NEN, NEM.
Maybe you’re thinking: “Hold on Parker, those last 13 aren’t even words.”
True enough. But not too far down the line the child will be faced with reading such words as SENt, SANd, NESt, NASty, EMber, ENvy, ESsay, SEMi, MEMber, dyNAMic, obSESS, emiNENt, and NEMesis. So, it’s hardly a bad thing to have her learn to “read” these 13 word parts now.
The primary goal here, in the first few weeks of kindergarten, is learning some phonemes, the letters that represent them, and, especially, blending. That some of the blends are only parts of words is ok.
As soon as another 3-4 phonemes are added to the above five, there will be more than enough actual words to deal with.
Once the above is accomplished, in the first month of instruction, the child will be reading – not memorizing sight words, not guessing – but genuinely reading. He’ll already be starting to realize how the whole thing works: letters in words stand for sounds and those sounds can be blended to figure out what the word actually says. This is called the alphabetic principle.
If the teacher then reverses the above and asks the child to spell a word he has just recently built by blending, the child will likely be able to do so. Spelling is harder than reading. It involves segmentation: hearing the word’s phonemes, and then attaching the correct letter to each phoneme heard. This task becomes much easier if the child just recently built the word, phoneme by phoneme, by blending.
Contrast the above with how Phonemic Awareness must be approached using any top-down method. Since top-down methods start with whole words, children must become aware of phonemes by hearing them, not individually, but in spoken words. However, in spoken words, those phonemes are in their coarticulated form, that is, they bleed into one another – a fact that makes hearing individual phonemes notoriously difficult for beginners.
Let this sink in for a moment. The bottom-up method, Synthetic Phonics, makes PA as easy as possible by teaching the isolated phonemes directly and then blending the phonemes into whole words. Top-down methods, all of them, make PA as hard as possible by requiring beginners to hear and distinguish phonemes in their coarticulated spoken form.
The result is that PA has become something of an obsession in top-down Balanced Literacy. Since so many children are not becoming aware of the phonemes in whole words, elaborate phonemic exercises have been proposed to remedy the situation. These remedies range from pre-phonics phonological exercises such as clapping out syllables, rhyming, alliteration, and onset-rime separation, to “advanced” oral-only exercises involving phonemic deletion, phonemic substitution, and even phonemic reversal.
None of this is necessary for a typically developing child in a bottom-up Synthetic Phonics classroom. Such a child will become aware of phonemes because each phoneme is explicitly taught and practiced. Such a child will become aware of the alphabetic principle because blending all through a word is explicitly taught and practiced. Already back in the late 90s, the US National Reading Panel cautioned against what it viewed as the “rush” to teach PA apart from blending with letters. It also debunked the notion that PA was a “new” concept:
“In the rush to teach phonemic awareness, it is important not to overlook the need to teach letters as well. The NRP analysis showed that PA instruction was more effective when it was taught with letters. Using letters to manipulate phonemes helps children make the transfer to reading and writing.” (2-33)
“It is important to note that when PA is taught with letters, it qualifies as phonics instruction. When PA training involves teaching students to pronounce the sounds associated with letters and to blend the sounds to form words, it qualifies as synthetic phonics. When PA training involves teaching students to segment words into phonemes and to select letters for those phonemes, it is the equivalent of teaching students to spell words phonemically, which is another form of phonics instruction. These methods of teaching phonics existed long before they became identified as forms of phonemic awareness training. Although teaching children to manipulate sounds in spoken words may be new, phonemic awareness training that involves blending and segmenting with letters is not. Only the label is new. (2-34)
Advantage #2: For any child, reading comprehension, the ultimate goal of all reading instruction, will proceed faster if that child is taught using bottom-up Synthetic Phonics.
[Note: If this seems like a controversial statement to you, it’s likely that you’re unaware of the Simple View of Reading (SVR). Please take the time now (it’s worth it!) to read about the Simple View here. It won’t take you long, and my SVR blog fully explains why Synthetic Phonics is the best choice for jump-starting reading comprehension. For the sake of efficiency and space, the arguments I make in the SVR blog are not restated here.]
Top-down proponents like to stress the importance of “reading for meaning” right from the start of instruction. They fear an initial focus on blending individual phonemes will produce students who can transform print into sound, yet not understand the sound they’ve just produced. (“Word-calling” or “barking at print” are the pejoratives some of them use.) The fear is unfounded. Unless the teacher has chosen completely inappropriate texts, the beginning reader will be reading material that he would understand if that material were spoken to him.
Here’s what renowned Harvard researcher Jeanne Chall found after her comprehensive study of top-down methods versus “code emphasis,” her term for Synthetic Phonics:
“The results are better, not only in terms of the mechanical aspect of literacy alone, but also in terms of the ultimate goals of reading instruction – comprehension and possibly even speed of reading. The long-existing fear that an initial code emphasis produces readers who do not read for meaning, or with enjoyment, is unfounded. On the contrary, the evidence indicates that better results in terms of reading for meaning are achieved with the programs that emphasize code right at the start…” [Jeanne Chall, Learning to Read: The Great Debate, p 307].
[Note: You may be thinking: well, if both phonemic awareness AND reading comprehension are better served using a bottom-up approach to instruction, then the case has been made. But do keep reading. The next two advantages are even more compelling.]
Advantage #3: Orthographic Mapping and Self-Teaching, both essential abilities for skilled reading, will start developing for most students within the first few months of competent bottom-up instruction. In top-down instruction, such development takes years – or they never develop at all.
[Note: Orthographic Mapping and the Self Teaching Hypothesis are discussed in detail in this blog, but I’ll recap these concepts here.]
Orthographic Mapping is what enables a child to go from word identification (via blending or “sounding out”) to instant word recognition. If a word is “mapped,” it no longer needs to be slowly sounded out. Instead, it’s immediately recognized as a sight word. But be careful here. Don’t confuse this type of sight word creation, which is automatic and unconscious, with the sight words that all top-down methods require to get started. The latter require conscious, deliberate effort on the part of the child precisely because the child doesn’t know how the word’s letters symbolize sounds. (These 2 utterly different types of sight words are discussed here.)
So, what skills must be in place for orthographic mapping to commence? Linnea Ehri, the researcher who coined the name “orthographic mapping,” says the child must be able to form “complete connections between the letters seen in the written forms of words and phonemes detected in their pronunciations.” To form these “complete connections,” the child needs 2 skills: knowledge of “how most graphemes (single letters or letter pairs) symbolize phonemes in our conventional spelling system” and “the ability to decode words never read before by blending letters into a pronunciation.” (Word Recognition in Beginning Literacy, p21)
Once a child decodes a word successfully a few times – estimates range from one to six times – that word automatically becomes a sight word for her.
David Share’s Self Teaching Hypothesis complements and corroborates Ehri’s Orthographic Mapping. A skilled, educated reader has about 60,000 sight words in memory. Recognizing that no amount of deliberate memorization and/or guessing could account for such a phenomenal number, Share claims that self-teaching must be involved:
“Each successful decoding encounter with an unfamiliar word provides an opportunity to acquire the word-specific orthographic (spelling) information that is the foundation of skilled word recognition. A relatively small number of successful exposures [1-4] appear to be sufficient for acquiring spelling representations, both for adult skilled readers and young children. In this way, decoding [blending] acts as a self-teaching mechanism or built-in teacher enabling a child to independently develop both word-specific and general spelling knowledge. Although it may not be crucial in skilled word recognition, decoding may be the principal means by which the learner attains skilled word recognition.” [Cognition 55, (1995), p 155]
Share claims there are only two “co-requisites” needed in order for this self-teaching capacity to begin developing in a new reader. Notably, these co-requisites are the same two cited by Ehri for orthographic mapping: knowledge of grapheme-phoneme correspondences and the specific phonemic awareness skill of blending. Share calls these two co-requisites the sine qua non of reading acquisition (the indispensable ingredient without which skilled reading is impossible). (p 156, 173) Here’s what he says:
“There is strong evidence for a causal role of phoneme synthesis (blending) as a twin co-requisite (alongside grapheme-phoneme knowledge) for successful reading acquisition. This conclusion is supported by both laboratory and field studies. Phonics programs which explicitly teach blending produce superior results compared to ‘analytic’ programs which generally do not include a blending component” (p 194)
“There is strong support for Ehri’s view that spellings can only be memorized when linked to phonemes detected in pronunciations. The process of letter-by-letter decoding and blending (amalgamating) into an integrated spoken unit, or in short, bottom-up decoding, may be ideally adapted for orthographic mapping. Spelling, of course, is another such process which obliges the explicit processing of letter order and letter identity.” [Journal of Experimental Child Psychology 72, (1999), p 123]
The example I used at the start of this essay (involving the letters S, M, N, A, and E), and, indeed, any bottom-up reading program, begin instruction precisely with the co-requisites cited by both Ehri and Share for orthographic mapping and self-teaching to commence: knowledge of letter/sound correspondences, and blending.
By contrast, in top-down reading programs, children spend the bulk of first-year instruction (kindergarten in the US) wasting precious time, chiefly by memorizing sight words, guessing their way through “predictable” books, and expressing themselves using invented spellings. Judging by the schedule laid out in the popular Balanced Literacy book, Words Their Way (by Donald Bear), children won’t “discover” the needed letter/sound correspondences for all-through-the-word blending until 4-5 years have elapsed. Thus, both orthographic mapping and self-teaching get delayed 4-5 years.
Worse, many children in BL classrooms never get instruction in all-through-the-word blending. These are the children who will go through life as crippled readers, unable to self-teach, unable to do the “mapping” required for automatic sight words to accrue. Some of these kids will end up in the “school-to-prison pipeline” where fully 85% of juveniles who interact with the court system are functionally illiterate. (Sixty percent of the adults in the US prison system are illiterate.)
[Note: So far, I’ve argued that bottom-up Synthetic Phonics provides an enormous dividend for beginning readers in three crucial areas: phonemic awareness, reading comprehension, and orthographic mapping. There is, however, a fourth reason for adopting a bottom-up approach to reading – and I think it’s the most important reason of all.]
Advantage #4: Bottom-up instruction makes the tasks of learning to read and spell logical for children. This is no small matter. From the time children first begin to speak, the question they ask most often is why? Children need to be given reasons if they’re to make sense of their world. In her book, Why Our Children Can’t Read, Diane McGuinness expresses it this way:
“Children, like adults, have problems paying attention to something they can’t do or don’t understand. They have a limited capacity to hold information in mind, and this capacity shrinks to zero when the information makes ‘no sense’.” (p169)
A child should agree that the letters M-A-N say “man” not because some authority figure says so, but because a knowledgeable teacher carefully leads the child to understand why this fact is logical, rational, and even inevitable. The only way to do this is to start teaching the full code, with blending, right at the start of reading instruction. Then the word MAN is simply the sum of its blended parts:
MAN = /m/ + /a/ + /n/
Taught to read in this manner, most children catch on quickly, and they delight in the way the code works. More importantly, they become convinced that reading is rational, worthwhile, easy, and fun. Their mindset becomes “I can do this” rather than “Oh no. Not more sight words!”
There’s no avoiding the fact that human brains work most efficiently when they can make logical connections. That’s precisely what’s behind Ehri’s prerequisite (above) for orthographic mapping to succeed: once a child can make “complete connections between the letters seen in the written forms of words and phonemes detected in their pronunciations,” sight words learning becomes automatic.
There’s nothing surprising in any of this. We can fill the first 2 years of reading instruction with difficult-to-memorize sight words, with guessing strategies, and with invented spellings. But if we do so, what will happen, inevitably, is that many students, among them some of our brightest and most inquisitive boys and girls, will decide that reading is arbitrary and therefore not worth their effort. Since they can’t “figure it out” they’ll lose interest and give up. Some will, understandably, “act out” their frustration. Then, the top-down system responsible for their failure will categorize them as “reading disabled” or “dyslexic.” This is a scandal.
It’s not too hard to place yourself in the shoes of millions of struggling, frustrated children. Just do the following short exercise.
An Exercise in Top-Down Learning
Memorize these 7 sight words any way you can. (I used a simple substitution algorithm to take you back to the time when you couldn’t read):
bpm means “the” qv means “in”
jwg means “boy” bew means “two”
zqdmz means “river” bpmg means “they”
aie means “saw”
Once you’ve memorized these 7 sight words, “read” this 8-word sentence:
Bpmg aie bew jwga nqapqvo qv bpm zqdmz.
Let’s suppose you’ve persevered and you’ve been successful:
They saw two boys xxxxxxx in the river.
Now you must guess the meaning of that unknown word, nqapqvo. So ask yourself: what could “make sense” in this context? Swimming? Fighting? Wading? Rafting? Peeing? Drowning? You may be getting a tad frustrated at this point. You may be wondering: who in the world would teach beginners to read via memorization and guessing?
Here’s where Balanced Literacy books conveniently provide a picture of 2 boys, standing by a river, holding fishing rods. “Look at the picture” your teacher tells you, or “Look at the first letter.”
Now, do you feel as though you’ve done any genuine reading? Based on this exercise, would you conclude reading is a logical skill worth your time and effort? Would you like to keep at this for a year or more?
Advocates of top-down reading instruction have been highly influential in Teacher Colleges, professional organizations (e.g. ILA, NCTE), and school administrations for nearly a century. This education establishment is supported by a multi-billion-dollar publishing industry and its well-paid, enormously popular, yet misguided authors.
If you’re a Balanced Literacy teacher, it seems to me that you can remain so in good conscience only if one or the other of the following is true:
1) You claim – and you can make the case – that all four of the above advantages of bottom-up Synthetic Phonics are false. I say “all four” because even if only one remains, it still presents a strong case for Synthetic Phonics.
2) You claim there is an advantage of top-down instruction so compelling that it overrides these 4 advantages of bottom-up instruction. (I would certainly like to know what that might be.)
It won’t suffice to say that some children do indeed learn to read via Balanced Literacy. Though the statement is true, it’s also true that some children will learn to read no matter how inadequate their reading instruction. Snow and Juel’s dictum is relevant here: synthetic phonics is “helpful for all children, harmful for none, and crucial for some.” (I would change that last word to “most.”)
It’s time – way past time – to end the needless suffering caused by a century of top-down instruction. Phonemic awareness, reading comprehension, orthographic mapping, and the ability to make the entire enterprise logical for our children, are all substantially enhanced in a Synthetic Phonics classroom.
In addition – and this is no small matter – reading science seems to concur.
© Copyright: Stephen Parker, 2019
Reading Teachers and Parents: To see what bottom-up instruction looks like, you can easily download any of my FREE Synthetic Phonics books right here.
Special thanks to Pamela Snow, Dylan Wiliam, and Jennifer Chew for their helpful suggestions in the drafting phase of this blog.