What went wrong with reading

What it looks like from where you are sitting

Your child reads a book you have read together twenty times, and reads it beautifully. You open a new one and it falls apart. They look at the picture before they look at the word. They say pony where the page says horse, and dad where it says father. They reach a word they have not met, stop, and look up at you.

That is not a child who is not trying, and most of the time it is not a child with a difficulty either. It is what reading looks like when somebody has been taught to work out a word from everything except the letters in it. For about thirty years, in a great many English-speaking classrooms, that is exactly what children were taught to do.

They were taught to guess

The method has a name. It is called three-cueing, and the three cues are meaning, structure and visual: what would make sense here, what would sound right here, and what the letters look like. The letters are the third of the three, and in practice they were often the last resort rather than the first move. A child stuck on a word was told to look at the picture, think about the story, get their mouth ready for the first sound, and have a go.

For the first two or three years it works, or it looks as though it does. Books for the youngest readers are short, the pictures carry most of the meaning, and the sentences are predictable enough that a good guesser can get through one without reading it. The trouble arrives at about eight, when the pictures thin out and the words get longer and less guessable. That is the point at which a lot of parents are told, for the first time, that their child is behind.

The idea travelled under several names. Whole language, then balanced literacy, which promised a bit of everything and in practice often kept cueing at the centre of it. Reading Recovery, a one-to-one programme for six-year-olds who were behind, was built on the same idea, and it was used with the children who were already finding reading hardest.

It is worth saying plainly that this is not a story about teachers. Teachers were trained in this, by universities and by the programmes their schools had bought, and many were told that the alternative had been discredited.

What the research actually found

The research had in fact settled. Congress asked for a review of it, and in 2000 the National Reading Panel reported back. It said that instruction has to teach children to hear the separate sounds inside a spoken word, has to teach the letters that spell those sounds explicitly and in an order, and has to build fluency, vocabulary and comprehension alongside them. The first two are the ones cueing skipped.

There is a second finding that matters more than it sounds. When researchers track the eyes of a skilled adult reader, the eyes do not skip along sampling a word here and there. They land on very nearly every word, and take in very nearly every letter. Fluent reading is not guessing that got good. It is letter by letter work that became so fast and so automatic that it stopped feeling like work at all.

Which means guessing is not a stage on the way to reading. A child who is guessing well is not nearly there. They are practising a different skill from the one they will need.

How it got into the classroom

Research does not walk into a classroom on its own. Somebody has to publish it, sell it, train people in it and put it in a catalogue, and for decades the people doing that were selling the other thing. Programmes built on cueing were bought by thousands of school districts, and the training that came with them taught the method to the teachers who used it.

The reporting that took all of this apart is Sold a Story, an investigation by Emily Hanford for APM Reports, first published in 2022. We are not going to retell it here. It is better than any summary of it, it is free, and you can listen to it yourself.

Two things belong next to it. The publishers and authors at the centre of that reporting have since revised their materials to put more weight on phonics, and leaving that out would not be fair to them. And we have nothing to do with any of it: TaleTykes is not affiliated with APM Reports or with anybody else named on this page, has been reviewed by none of them, and is not endorsed by anybody.

Where it stands now

These figures are about reading in the United States, not about TaleTykes. The first is the size of the problem. The other two are the reason to be hopeful about it.

Most of the country has now changed its reading law or its policy, and a good deal of that happened after the reporting landed. A statute is not a classroom and it does not teach anybody, so this will take years to arrive where your child is sitting. But the direction is no longer seriously in doubt.

What you can actually do

Two conversations, and neither of them asks you to become an expert in any of this.

With the school

• Ask which reading programme they use, by name, and when it was last changed.
• Ask whether phonics is taught to every child, explicitly, in a set order, rather than only to the ones who get stuck.
• Ask what a child is told to do when they meet a word they do not know. If the answer involves the picture, or what would make sense in the sentence, that is the thing this page is about.
• Ask whether the children are screened early, and what happens for a child the screen picks up.
• Ask kindly. Almost nobody in that room chose the programme.

At home

• When your child stops at a word, resist the sentence on the tip of your tongue, which is what would make sense there. Say the sounds with them instead, left to right, and let them put the word together and finish it themselves.
• Cover the picture, if the picture is doing the reading.
• Keep books they can nearly manage within reach, and read the harder ones to them. Listening to a story well above their own reading level is where a great deal of the vocabulary comes from.
• Let it be short and daily rather than long and occasional.

And read with them. The strongest thing in the reading research is a grown-up in the room talking to a child about a story, and that has never been a piece of software.

What we are, and what we are not

We built this against the background above, so it would be strange not to say where it sits. Here is what the software does today.

• Books built from the sounds of their stage. For a child of about five to seven reading in English, every word in a story is one they can sound out from the letter sounds that stage has taught, apart from a small set learned by sight and the names of the characters. The order we follow is published in full on the phonics page, and it is the same list the writing is held to rather than a description of it.
• A word they can tap. Tapping a word says it aloud and shows the letter groups it is built from, so ship comes apart as sh, i and p rather than as four separate letters.
• Reading along. For the two youngest age groups the words are highlighted one at a time as they are read.
• A level we describe honestly. We estimate how hard the text is on a scale of our own, and every screen that shows it says so. It is not a Lexile measure, not a school reading level, and not a measure of your child.

And here is what it is not.

• Not a reading programme. Decodable text is a scaffold. It is applied to one age band, it is taken away as a child moves on, and it is not a substitute for being taught to read by somebody whose job that is.
• Not evidence about children. We have measured no reading outcome in any child, and there is no study. What we have is a reason for building it this way, which is not the same thing and should not be read as though it were.
• No reading age, no grade and no percentile. We will not hand you a number that ranks your child against other children, because we are not in a position to and because it would not help you.

There is more about how the levels are worked out on the reading page, and you can read a whole book without an account on the sample shelf.