The science of reading: what it says, what it doesn't, and how to teach by it

What it is

"The science of reading" isn't a program or a single method. It's a name for decades of research, from psychology, linguistics, education and neuroscience, on how children learn to read and why some struggle. In schools, the phrase has come to mean the teaching that research points to, especially for the first years: explicit, systematic phonics, built on phonemic awareness, with plenty of reading of real text.

A useful frame is the simple view of reading, proposed by Philip Gough and William Tunmer in 1986: reading comprehension is the product of two things, decoding (turning print into words) and language comprehension (understanding those words once you have them). If either is zero, reading comprehension is zero. A child who decodes fluently but has a small vocabulary will struggle, and so will a child with a rich vocabulary who can't work out the words on the page.

From that frame come the pieces teachers will recognize:

• Phonemic awareness: hearing and working with the individual sounds in spoken words (the three sounds in ship).
• Phonics: how letters and letter groups represent those sounds, taught in a planned order, with practice blending sounds into words and segmenting words for spelling.
• Fluency: reading accurately, at a reasonable pace and with expression, so attention is free for meaning.
• Vocabulary and knowledge: the words and background a reader needs to understand what they decode.
• Comprehension: making sense of text, including the strategies good readers use.

The National Reading Panel's findings, published in 2000, cover all five. Anne Castles, Kathleen Rastle and Kate Nation's review, written to help end the long-running "reading wars", explains why phonics is central in an alphabetic writing system like English, and also what else children need to become expert readers.

Why it works

Written English is a code for spoken English. Children already know thousands of spoken words when they start school; what they don't know is how the marks on the page map onto those words. Teaching the code directly, in a planned order, gives them a way to work out words they've never seen in print instead of guessing or memorizing shapes.

The strongest evidence comes from comparisons between systematic phonics and other approaches. Linnea Ehri and colleagues' meta-analysis for the National Reading Panel pooled 66 comparisons from 38 experiments. Systematic phonics beat unsystematic or no phonics, including whole language, with a moderate effect that lasted after the teaching ended. The effect was larger when phonics began in kindergarten or first grade than later. It helped decoding, word reading, comprehension and spelling, and it helped children at risk of reading difficulties and older struggling readers. Small-group and whole-class phonics did about as well as one-to-one tutoring.

The panel's findings were challenged. A group of researchers re-analyzed the data and argued the effect was smaller than reported. Karla Stuebing and colleagues answered by re-analyzing the same studies with the critics' own coding, and found small-to-moderate effects favoring systematic phonics, consistent with the panel; the largest effects came when phonics sat inside fuller, more intensive reading instruction. The disagreement is real, but it's about how big the benefit is, not about which way it points.

The What Works Clearinghouse's 2016 practice guide for kindergarten through third grade, by a panel chaired by Barbara Foorman, rates two recommendations Strong: develop awareness of the sounds in speech and how they link to letters, and teach students to decode words, analyze word parts, and write and recognize words. It rates daily reading of connected text as Moderate, and teaching academic language skills as Minimal. Minimal means the rigorous evidence is thin or inconsistent, not that the practice doesn't work.

How to start, by grade

Examples

• A fifteen-minute phonics lessonKindergarten–1st grade

  Two minutes of oral blending and segmenting. Two minutes reviewing last week's letter-sound patterns with cards. Five minutes introducing today's pattern (say sh makes one sound, show it, blend sh-i-p, sh-o-p). Three minutes spelling words with the pattern on whiteboards. Three minutes reading decodable sentences that use it.

• Word chainsGrades K–2

  Write cat on the board. Change one letter at a time and have children read each new word: cat, cap, map, mop, top, tip, ship. Then reverse it: you say a word, they change one letter on their whiteboard to spell it. Every change makes them attend to each sound.

• Reading a hard word, not guessing itGrades 1–3

  When a child is stuck, say "Look at the letters. What sounds do they make?" rather than "What would make sense?" or "Look at the picture." Point to each part as the child blends. If the word has an irregular part, show that part and say it.

• Fluency with repeated readingGrades 2–5

  Partners read a 150-word passage to each other three times across the week. The listener follows along and marks skipped or misread words. On Friday each reader does a one-minute read with you. The repetition builds accuracy and speed on the same text; new passages each week move it to new words.

• Knowledge-rich read-aloudGrades K–3

  During a unit on oceans, read aloud a nonfiction book well above the class's decoding level, stopping to talk about four or five new words and how they connect. Children who can only decode simple text are still building the vocabulary and knowledge they'll need when their decoding catches up.

Common mistakes

Teaching children to guess words from pictures and context.

Cues like "look at the picture" or "what would make sense" let children skip the letters, which is exactly the skill they need to practice. Prompt them back to the print, and save context for checking a word once it's decoded.

Phonics without a sequence.

Teaching letter-sound patterns as they come up in books isn't systematic. Ehri's meta-analysis compared systematic phonics with unsystematic phonics and found systematic was better. Use a written order and stick to it, with review built in.

Too much time on oral phonemic awareness alone.

Phonemic awareness matters, but the practice guide's Strong rating is for sound awareness linked to letters. Move from oral-only games to letters quickly, usually within the same lesson.

Leveled books before children can decode.

Books chosen only by level often contain patterns children haven't been taught, which pushes them back to guessing. In the early stages, give them decodable text that matches what's been taught, alongside read-alouds for vocabulary.

Thinking phonics is all of reading.

The same research says children also need fluency, vocabulary, knowledge and comprehension. A classroom that only does phonics produces decoders who don't understand what they read. Keep read-alouds, talk and knowledge-building units going from day one.

Stopping phonics too early for students who still struggle.

Older students who can't decode won't catch up by reading more of what they can't read. Give them systematic intervention in small groups; the meta-analysis found it helped older struggling readers too.

Time and cost

Time: a daily phonics and phonemic awareness lesson in kindergarten through second grade, as one part of a longer literacy block that also includes read-aloud, writing and comprehension. The lesson is short and brisk; the daily habit matters more than the length.

Money: a scope and sequence and decodable texts are the essentials. Commercial programs cost money; free decodable text and lesson routines exist (ours are below).

Training: the biggest cost. Teaching phonics explicitly takes a working knowledge of the sound structure of English and of the order patterns are taught in. Plan time for teachers to learn it, and expect the first months to feel slow.

Debates and open questions

How big is the phonics effect? The panel's meta-analysis reported a moderate effect. Critics re-analyzed the data and argued it was smaller. Stuebing and colleagues replied that the critics' own analyses don't contradict the panel and that effects of that size are practically significant, and the critics replied again. The argument is about the size of the advantage and what to compare it with.

Synthetic or analytic phonics? Ehri's meta-analysis found that synthetic phonics (sounding out and blending letter by letter) and larger-unit approaches (working with onsets, rimes and chunks) produced a similar advantage. The evidence supports being systematic more than it supports one style of phonics over another.

What about comprehension and knowledge? The National Reading Panel found that vocabulary instruction leads to gains in comprehension and that teaching a combination of comprehension strategies works best. Castles, Rastle and Nation argue that moving "beyond phonics" is as important as getting phonics right. The research on how best to build knowledge across the grades is less settled than the research on early decoding.

Is the research old? The National Reading Panel and Ehri's meta-analysis cover studies up to 2000. The 2016 What Works Clearinghouse guide and the 2018 review by Castles and colleagues reach the same conclusions about early decoding with later research, which is why the rating here is Strong rather than resting on one report.

Tools that do it

Free, on this site, nothing to sign up for. Class lists stay in your browser.

• Decodable sentences by phonics pattern: short decodable text for each pattern, for daily practice after you teach it.
• Sight word lists and flash cards: high-frequency word lists; teach the regular parts by sound and point out the irregular letters.
• Reading fluency timer and WCPM calculator: time one-minute reads and work out words correct per minute for repeated reading.
• Read-aloud books for the projector: free books to read aloud for vocabulary and knowledge while decoding catches up.
• Reading passages with questions: leveled passages for fluency and comprehension practice in grades 2 and up.
• How to teach phonics step by step: a longer guide to the daily phonics lesson, the order of patterns and decodable practice.
• 25 phonemic awareness activities: five-minute oral activities to link to letters.

Questions teachers ask

Is the science of reading just phonics?

No. Phonics is the part with the strongest evidence for beginning readers and the part many classrooms were missing, so it gets the attention. The same research covers phonemic awareness, fluency, vocabulary, knowledge and comprehension, and says children need all of them.

What grade should phonics start?

Kindergarten. Ehri and colleagues found larger effects when systematic phonics began early than when it began after first grade. A common plan teaches letter-sound patterns from kindergarten through second grade, with multisyllable decoding continuing into grades three to five.

Do older struggling readers need phonics?

Often, yes. The meta-analysis found that systematic phonics helped older students with reading difficulties as well as beginners. A quick oral reading check shows whether an older student's problem is decoding, comprehension or both.

What's wrong with three-cueing?

Three-cueing teaches children to identify words using meaning, sentence structure and the letters, often in that order. The trouble is that it lets children guess from context and pictures instead of decoding, which is the skill the phonics research supports teaching directly. Use context to check a word after it's decoded.

Are decodable books necessary?

The practice guide rates daily reading of connected text as Moderate. Decodable text is a practical way to make that daily reading match what's been taught, so children practice decoding instead of guessing. Children move on from them as their knowledge of the code grows.

Does the science of reading mean no read-alouds or real books?

No. Read-alouds of rich books are how young children build vocabulary and knowledge they can't yet get from their own reading. The science of reading changes what children read on their own while they learn the code, not whether they hear good books.