“Readers read familiar words by accessing them in memory, called sight word reading. With practice, all words come to be read automatically by sight, which is the most efficient, unobtrusive way to read words in text.” Ehri (2005)
Do not confuse ‘sight-recognition’ words with ‘sight-words’ (the irregular words that must be memorized, such as ‘plaid’), or with memorized words. A student that memorizes ‘b-e-d’ spells ‘bed’ will find reading error-prone, effortful, and slow. Comprehension will be very low. By contract, a reader using sight-recognition words is not aware of any effort in reading.
The effects of dyslexia are serious and obvious, and only a few readers become dyslexic, which is strongly suggestive of a medical or genetic problem. Yet many dyslexics become strong readers with corrective intervention.
This blog post presents a line of evidence suggesting that many cases of reading disability are simply the result of a child missing or skipping a step on the normal path to learning to read. These students, hold onto inappropriate reading strategies that slow their acquisition of sight-recognition words. Without sight-recognition words, they remain struggling readers with poor comprehension. We review theory, research, and corrective approaches.
The epigraph at the top of this posting is from Linnea Ehri’s famous paper describing the phases that “normal” readers follow to acquire their inventory of sight-recognition words. She proclaims the central role of sight-recognition words, and describes four progressive strategies to recognize words that are not yet in the sight-recognition inventory:
- Pre-alphabetic: reading words by memorizing their visual features or guessing words from their context. For example ‘look’ has two eyes.
- Partial-alphabetic: recognizing some letters of the alphabet and guessing prompted by context. For example “The cat in the h__”.
- Full-alphabetic: using simple spelling-to-sound mapping to blend unfamiliar words. For example “bat”.
- Consolidated-alphabetic: using advanced spelling-to-sound mapping, larger chunks, and analogies to blend unfamiliar words. For example “blast”
In each case, if a word is successfully identified, then the reader may add it to his inventory of sight-words regardless of which strategy was used. Ehri’s processes are hard work, but reading a sight-word is fast, effortless, obligatory, and automatic. So there is a huge gain in acquiring the sight-word.
Ehri calls her steps ‘phases’, but they overlap. An emerging reader might recognize some words, guess others if the context allows, blend if he is able, and of course skip as the last resort. Every reader follows a unique path on this model. Even skilled readers must fall back on the simpler strategies for complex spellings; I have no idea how to blend ‘chthonic’ or ‘hemorrhoid’, but they are sight-words to me now and I surely acquired them long after I could decode.
This posting focuses on how sight-words become available to be automated. Just as important is whether a successfully identified word will actually be added to inventory, but that must wait for another blog post. Logan (1997) provides an excellent review for the curious.
Over time, a child reduces the effort of reading by building his inventory of sight-recognition words, and also shifts towards more-advanced strategies as his reading skills improve. Read that sentence again. That is the standard model of how non-dyslexic readers develop.
Ehri’s model is not complete, and the smaller details are disputed. As I learn more about English spelling, I become convinced that morphology, the units of meaning, becomes more important than phonics once a child has learned to blend. But that is also for another posting.
Ehri is only describing early reading, other factors like vocabulary, grammar, and world knowledge are involved in advanced reading; the “Matthew Effect” describes the subsequent feedback loop driving a strong reader forward. Yet, Ehri’s excellent model describes the first steps and provides the key idea behind this blog posting.
Ehri’s model does not address reading failure. Students who do not progress through her phases will obviously fall behind their peers, but that alone isn’t reading disability. In Ehri’s model, a student who has stalled at a certain phase may quickly catch up if provided with extra tutoring or practice.
But a struggling reader is not frozen in amber, and that isn’t what happens. Reading is a developmental process, and a beginning reader is not the same as an intermediate one. Educators focus on decoding skills in the early grades, and comprehension skills in the later ones. Books become longer and more complex. A reader who falls behind soon finds himself in a different environment.
In their book Off Track: When poor readers become “learning disabled”. Spear-Sterling and Sternberg expand Ehri’s model to include the child’s continuing development. They suggest that a child who strays from the normal path of reading may get lost in the woods, but still continues to develop reading skills. Unfortunately, those skills may be inappropriate or even sabotaging, and this child may then follow a faulty path. It is the failure to bring an “Off-Track” poor reader back to the main path that will turn him into a disabled reader.
Off-Track describes common patterns of reading disability, and explains how to assess them and correct them. The common pattern in older students that we see repeatedly at the Community Reading Project is the ‘Compensatory Reader’, who gets stuck at the Partial-alphabetic stage of recognizing words using a few letters and prompted by context.
Compensatory Readers are experts at guessing, and may have memorized a surprising number of words. They hone their guessing skills and resist moving into decoding. Reading tutors compound the damage by helping them improve at guessing, teaching poor readers to ask whether a word sounds right, looks right, and makes sense.
But guessing is error-prone and requires constant monitoring and back-tracking. These kids may be the best readers in grade 1, but by grade 4 the wheels have fallen off. We are often amazed at how well a student can read with memorized words, often fooling their parents and teachers into thinking they can read, but their vocabulary and comprehension tends to remain very low. (Hint: ask your child to read the credits from a movie poster, because names cannot be guessed and cannot be sight-words since your child has not seen them before).
Another pattern from “Off-Track” that we see are the “Non-Automatic” readers who decode accurately but do not build an inventory of sight-recognition words in return for their efforts. Their reading is accurate but slow and effortful, focused on the individual words at the expense of attending to the larger text. Again, they fool their parents and teachers since they can ‘read’ almost every word, but they remain slow readers with poor comprehension.
To understand why leaving the main path of reading can be a spiral into disability, consider a two-finger typist who has practiced until he can peck away with surprising speed. But he cannot progress to become a skilled typist because he is off-track. To speed up, he must purposefully spread all 10 fingers, unlearn and relearn typing from scratch, and suffer much slower typing until he recovers.
Typing isn’t an exact analogy, an off-track reader does not have the physical feedback of spreading his fingers on a keyboard and must use his executive function to monitor reading methods. He may not realize that he depends on guessing, or even that he shouldn’t be. Older readers start to engage with harder texts, and the pressures to get through the reading become intense. Eventually negative behavior patterns begin to kick in. An off-track typist may choose to improve, but an off-track reader will not easily find his way back without intensive intervention.
“Off-Track” is not an intuitive idea. It suggests that “more of the same” isn’t enough, we have to do something extra or different to help a reader who strays off the path. It is commonly said that dyslexics learn differently or have to be taught differently although there is no evidence for learning styles. Well, here is a learning theory that describes exactly how such a ‘difference’ requirement might arise.
Read Like a German Dyslexic
Is there any experimental evidence behind these theories of reading? Yes there is. Dr Ehri and her graduate students at CUNY have delivered bookshelves full of empirical support for her model. Google Scholar lists dozens of her papers, and they are all worth reading.
But if you are like me, you worry about confirmation bias. Of course Dr Ehri’s research findings will support her theories. So I like to find unrelated research, preferably in another field and asking unrelated questions.
The most compelling research I found emerges from Heinz Wimmer’s work with German-speaking dyslexic students. Wimmer (1996) was interested in how German dyslexic students perform on non-word reading. Non-words are carefully constructed nonsense words that shine a light on the component skills of reading, a student who can ‘read’ them is demonstrating decoding skills – deciphering from the alphabetic spelling of a word to its intended pronunciation. Non-words cannot be sight-read since the student has never seen them before, and they cannot be guessed from context.
Like Torgesen (see our earlier blog post), Wimmer went around to schools and found the worst readers, but a problem arose immediately in the screening tests; those readers were making almost no reading mistakes – they were hitting ceiling on accuracy. They were just slow readers, slowness turns out to be the main symptom of dyslexia in Austria. Wimmer had to be content with measuring speed because his ‘dyslexic’ students could read with perfect accuracy.
Wimmer’s study incorporated both age-matched and reading-level-matched control groups. With them, Wimmer was able to rule out many common explanations of dyslexia such as poor memory representation, visual deficits, impaired articulation speed, knowledge of phoneme-grapheme correspondences, or low skill automatization. What was left was the phonological deficit explanation of dyslexia – a possibility that these students might suffer from slow access to phonological memory representations. We will return to this idea in a later blog post.
Obviously, it is hard to support a medical or genetic justification for dyslexia if German dyslexics do not exhibit the same problems as English ones – surely they are not physically or genetically different. If the ‘cure’ for dyslexia is to move to Germany, then there is something odd in our understanding of reading deficits. But in what way are Germans and English readers different?
In a follow-up study the following year, Wimmer teamed with Oxford University to put this issue into the spotlight. Wimmer & Goswami (1997) compared normal German and English students between the ages of 7 and 9, the period when children learn to read, to see how the component skills of reading develop with age. They found that the youngest Germans were more accurate at non-words than the oldest English readers, and even the slowest young Germans were faster than older English readers.
By German standards of accuracy and speed, almost all the normal English readers were dyslexic. Interestingly, the Germans were uniformly good readers in all age groups; the English included readers who did well, but also others who did very badly.
The design of this study allowed the researchers to infer how the readers were internally processing words at different ages. Wimmer & Goswami suggested that Germans moved into reading by starting with strong decoding skills and then using these skills to develop sight-recognition vocabularies. They seemed to be starting Ehri’s model at the third step – the Full-alphabetic phase of blending sounds into words, without an initial inventory of sight-words.
By contrast, the English started at Ehri’s first and second steps by memorizing and using guessing strategies to build their inventory of sight-recognition words, then developed third-step decoding skills to increase that inventory. The large variations in English non-word reading abilities was explained by situating the child in those transitions – the ones who had not yet learned to decode did poorly, the decoders did well. Every German reader is a decoder, and Wimmer expressed surprise at the number of the older English students were still guessing instead of decoding.
English vs German Readers
How can it be that Germans learn to read differently and have a different experience of dyslexia? There are two major differences between English and German beginning readers.
First, German has a ‘transparent orthography’, there is a strong one-to-one relationship between how a word is spelled and how it is pronounced. Diacritics are used to mark vowels for different pronunciation, and hardly any letters represent more than one sound (although, as in English, there are often several ways to spell a sound). If I taught you how to pronounce the letters and set you to read a German newspaper aloud, you could ‘decode’ it accurately. Not knowing German, you would not understand a word, but a German-speaking listener could comprehend what you were saying.
English on the other hand has an ‘opaque orthography’, the same spelling may have different sounds (for example the ‘ie’ spelling in ‘thief’, ‘tied’, and ‘friend’), and there are many ways to spell sounds, for example, the /i/ sound might be ‘e’ as in ‘be’, ‘ee’ as in ‘meet’, ‘ie’ as in ‘chief’, ‘ei’ as in ‘receive’, ‘i’ as in ‘ski’ , ‘ey’ as in ‘key’, ‘y’ as in ‘hurry’, and several others.
It is our opaque writing system that forces young readers to start by memorizing words and guessing, then ease their way into decoding and chunking.
What about dyslexia in other languages? Consider Italians: Paulesu et al. (2001) conducted a brain-scan study of English, French, and Italian dyslexics; we will consider his brain-scan findings in a later posting. What makes his study interesting here is the background. In order to match for age and IQ, Paulesu had to find university students with dyslexia – easily done in France and England, but dyslexia isn’t common in Italy. Italian has one of the most transparent orthographies, and only uses 32 phonemes compared to 44 in English and roughly 60 in German. Paulesu had to screen 1200 students for impaired phonological skills in order to identify 18 ‘dyslexic’ students – and this group’s reading and non-word decoding skills were markedly superior to the matched English and French dyslexics.
Or Spanish: Serrano & Defior (2008) found the same results – Spanish has a ‘transparent’ orthography and Spanish dyslexics are slow but accurate readers. It seems that if your language (or more precisely, the spelling system used in your language) allows you to skip the guessing and start at the third step of Ehri’s model, then you aren’t at risk of becoming a ‘Compensatory Reader’. But you STILL have to develop phonological processing skiils, and without them you will still become a dyslexic ‘Non-Automatic Reader’.
Wydell & Kondo (2003) reported on a 16-year-old English-Japanese bilingual student who was a superior reader in Japanese and a dyslexic in English. Like Paulesu’s competent Italian dyslexics, he had impaired phonological skills, but seemingly only English was affected. The authors noted that less than 0.1% of children have a reading disability in Japan, which they attribute to the transparent nature of Japanese orthography (logographic for Kanji, syllabic and highly consistent for Kana), and also to school instruction that drills skills to automaticity. It’s a perfect example, in Japanese he doesn’t have to guess (but he is likely slow), and in English he is only reading by guessing words from context and first-letter.
The second difference is that German students are systematically taught synthetic (sound-to-spelling) phonics before they are introduced to books or reading (Johnston & Watson 2003). Sound-spelling pairs are slowly introduced and then immediately used for word recognition, and children are taught rhyming words that illustrate spelling patterns. English students are often introduced to books before they can read, and learn to read with a combination of sight-recognition words (‘look-and-say’ instruction) and training on common sounds made by each letter (regular phonics). Rhyming words are taught for vocabulary, usually with disregard for spelling variations (bed, dead, said). Of course, it is much easier to teach synthetic phonics in a transparent orthography.
Finally, consider again the ‘dyslexic’ German readers from the first study. They could decode accurately, which is almost trivial with German orthography, but they were slow. They decoded as you would with the German newspaper, but seemingly without developing an inventory of sight-recognition words. This is, of course, the “Non-Automatic” readers described in “Off-Track”. Germans start at the third phase of Ehri’s model, and that is the first step they might get stuck.
If it is slowness that prevents the readers from forming sight-words, then German dyslexia can be ‘fixed’ with an intervention that strengthens the sound-spelling skills of decoding and and develops a sight-recognition inventory. A spelling intervention fits the bill, presenting the morphological structure and etymology of bases and affixes and allowing affixing rules to be drilled and mastered explicitly, and supported by ‘repeated reading’ to develop fluency.
And that is exactly what happened in Ise & Schulte-Körne (2010), where German dyslexics in grades 5 and 6 were given an intensive spelling-based remediation resulting in strong gains in both spelling and reading speed. We will return to Ise & Schulte-Körne’s study when we discuss how to design an intervention, for now we simply say that their results are predicted by McGuinness (2004) whose analysis of the NRP studies directs us to the teaching of reading and spelling as a single discipline.
Getting Back On-Track
This blog post has presented a line of evidence that reading disability is the result of missing a step on the normal path to reading, and not a genetic or medical disability. That’s probably not enough to convince a skeptical reader, but I will explore other lines in future posts.
The significance of the “Off-Track” model of reading disability is that it directs us to address specific obstacles with tailored remediation techniques.
The free BLENDING program on this website is designed specifically for older “Off-Track” Compensatory Readers. It drills simple phonic recognition tasks, removes context from the guessers, forces the first-letter readers to look at all the letters, practices spelling segmentation tasks, and builds sight-recognition inventories. It typically takes 2-3 weeks of 20 minutes a day to work through it. It’s free – give it a try.
For “Off-Track” Automatic Readers (accurate but slow readers), we recommend the method of Repeated Reading (Dowhower, 1989), and the free FLUENCY program on this website is a useful implementation. We will write about this technique in the future.
We started this blog with Torgesen’s intervention for seriously dyslexic readers withering away in LD classrooms; how some of them bounced back after a short intervention. This result is beginning to look less mysterious.