Decodable readers and leveled readers, free at your child's level
Decodable vs leveled readers, what the reading research says about each, and how Peabody's free first books are checked against the sounds a child knows.
A decodable reader uses only the letter-sound patterns a child has been taught, plus a few common words learned by heart, so every word can be sounded out. A leveled reader is sorted by overall difficulty and often lets a child guess from pictures and patterns. Beginners need decodables; leveled books come once decoding is secure.
Peabody (peabody.kids) is a free AI-driven adaptive learning platform for kids, free through the end of 2026, with a free library and stories written at each child's level. Its first books are checked against the sounds a reader has met. AI writes the stories; code decides the rest, including whether a story is decodable enough to show.
Decodable and leveled, side by side
| Decodable readers | Leveled readers | |
|---|---|---|
| Built from | The phonics patterns taught so far, plus a short list of heart words | Overall features: sentence length, vocabulary, layout, picture support |
| What a child does with a new word | Sounds it out | Often guesses from the picture, the first letter or the pattern |
| Labelled by | Phonics step (short a, digraphs, silent e…) | A letter (A–Z) or number |
| Best for | Kindergarten and first grade, while decoding is being learned | Once a child decodes most words, for fluency, language and stories |
| Main weakness | Can read stilted; limited vocabulary | Rewards guessing habits beginners need to outgrow |
What the research says
Phonics comes first, and early. The National Reading Panel's analysis (Ehri, Nunes, Stahl and Willows 2001, Review of Educational Research) pooled 66 comparisons: systematic phonics had an effect of d = 0.41 overall, and 0.55 when it began in kindergarten or first grade, against 0.27 later (SAGE). A reanalysis put the size lower, around 0.24, but not the direction.
Decodable text helps beginners sound out, and is one criterion among several. Cheatham and Allor (2012, Reading and Writing) reviewed the studies and found decodable text raises beginners' use of sounding-out and their accuracy, while arguing it should be one way to choose text, not the only one (doi). The University of North Carolina's literacy institute calls the evidence inconclusive and recommends decodables as a temporary scaffold (UNC TRI). Both say the same thing in practice: use them while decoding is new, then move on.
Guessing is a stage to grow out of. In Ehri's phases of word reading, guessing from a picture or a first letter is the partial-alphabetic stage, before a child uses all the letters (UFLI summary). Leveled and predictable books, built to be readable with picture support, invite that strategy.
Each successful sound-out teaches the word. Share's self-teaching hypothesis (1995, Cognition) holds that every time a child decodes a new word on their own, they learn its spelling a little (PubMed). Studies cited in the research report behind this post put the number of exposures to learn a new word form at about four to six, and more for weaker readers. That is the case for text a child can actually decode: it turns reading into practice.
Decoding is not the whole of reading. Castles, Rastle and Nation (2018) put it plainly: phonics is necessary for decoding and not sufficient for comprehension, which depends on language, vocabulary and knowledge (Ending the reading wars). Reading aloud to a child and talking about the story still matter.
Why AI-written stories need a check
Language models are poor at writing to a reading level on request. Day and colleagues (2025) asked several models to simplify passages to a fourth-grade level and none of the model and prompt combinations hit the target (Readability Consortium preprint). The research report found nothing published on controlling model text at the decodable level, which sits below where readability formulas work at all. So a generated first book has to be checked word by word, not trusted.
How Peabody's first books are made decodable
- Every level 1 story written for a child is checked against that reader's own patterns. The code counts the distinct words a child can't yet sound out with the phonics patterns they have met, their sight words, the story's names and its word lists. More than six and the story is marked too hard, then repaired or rewritten. The cap is never zero, on purpose, so stories stay warm rather than stilted.
- Patterns come in one at a time. A new reader starts on word families one short vowel at a time (at, ig, op, ug, et), then blends, two-letter sounds, silent e, endings, two-part words and vowel teams. A new pattern comes in once the newest is sticking (right three days in a row) or after three stories with it.
- Pattern books. Eleven library books repeat one sentence frame with only the last word changing, a short sound-it-out word where one fits, with a little surprise near the end. The story writer makes the same kind for a level 1 child on the first sight-word lists.
- The picture confirms; it doesn't replace. Every page of a level 0 or level 1 book has a big picture of the thing it names, and the word wears the picture's colour. The changing word is kept decodable so the picture checks what the child sounded out rather than giving it away.
- Then the books level up. From level 2 the decodability check retires and the levels are measured by sentence length, words per screen and word familiarity, as in the reading levels cheat sheet. The practice side follows Peabody's sixteen-step phonics scope and sequence.
The free library holds 201 hand-written stories from the level 0 floor to about eighth grade, and all but two of the scarier ones are open without an account: open the free library and let your child pick a book.
What this doesn't show
- Peabody cannot hear a child read. It never records a child, so it can't measure accuracy or fluency aloud.
- Peabody doesn't measure Lexile, and doesn't give guided reading letters.
- Peabody has no efficacy study of its own; the research numbers belong to their studies.
Learn more elsewhere
- Reading Universe: a free map of reading skills with classroom videos and phonics checks.
- IES guide: foundational skills for K-3 reading: four recommendations rated for evidence, with take-home tip sheets for parents.
- Shanahan on Literacy: a careful blog by a former member of the National Reading Panel.
For about seventy more, checked and sorted by who they're for, see where to learn about AI.
Questions parents ask
Are decodable readers better than leveled readers?
For a beginner learning to decode, yes: they let a child practise sounding out instead of guessing. Once a child decodes most words, leveled books and real stories matter more for fluency, vocabulary and enjoyment. Most children need both, in that order.
When should my child stop reading decodable books?
When they decode most new words at their level without stopping, usually somewhere in first grade. The UNC literacy institute describes decodables as a temporary scaffold. Peabody's own check retires at level 2 for the same reason.
Where can I find free decodable books?
Peabody's library is free and needs no account, and its level 0 and level 1 books are built around short sound-it-out words with a picture on every page. A child's own stories at level 1 are checked against the sounds they have met.
What are heart words?
Common words with a part that can't be sounded out yet, like said or the. Decodable books include a few, taught by heart right after the phonics step they fit.
Checked against how Peabody works on 18 October 2026. Written with help from AI. Peabody is a reading app, not a reading assessment: if you're worried about your child's reading, talk to their teacher.