What is a graded reader?
A graded reader is a book written or adapted for language learners, with its vocabulary, grammar and length limited to a defined level. Some are original stories written within those limits. Others retell novels, biographies or true stories in simpler language.
The point is to let you read whole books in a language you’re still learning. Unsimplified novels and newspapers are written for fluent readers, so a learner below an advanced level meets unknown words in almost every paragraph and ends up decoding instead of reading. A graded reader removes most of those obstacles: you can follow the plot, read at a natural pace and finish what you start.
Graded readers aren’t the same as children’s books. A picture book for native speakers assumes years of listening to the language and freely uses words that adult learners seldom need. Nor are they parallel texts, which print a translation on the facing page. A graded reader is written entirely in the language you’re learning, usually for teenage and adult learners, at a level you can manage.
How graded readers control vocabulary and grammar
Graded readers control difficulty with a word list for each level, counted in headwords. A headword is a base word together with its regular forms, so “walk” also covers walks, walked and walking. A writer at a given level uses only the headwords on that level’s list, plus a few extra words explained in a glossary or shown in a picture.
Series usually begin at a few hundred headwords and climb in steps to a few thousand. Oxford Bookworms, for example, runs from 250 headwords at its Starter level to 2,500 at Stage 6. Most major publishers also map their levels to the CEFR, which helps you compare series, although the match is approximate.
Grammar is graded too. A low level might allow only the present and simple past, short sentences and a single plot line. The passive, conditionals and reported speech arrive at later stages, along with longer sentences, more characters and more left for the reader to infer.
Here is one moment written at three levels:
- Unsimplified: “Having missed the last train, Elena resigned herself to a long, cold walk home along the deserted coast road.”
- Intermediate: “Elena missed the last train, so she had to walk home. It was a long, cold walk on the empty road by the sea.”
- Beginner: “Elena is late, and the last train is gone. She walks home on the road by the sea. The road is empty, and it is cold.”
The story survives each step down. What changes is the vocabulary (resigned, deserted), the grammar (a participle clause becomes plain sentences) and how much the reader has to work out alone.
Why graded readers work
Graded readers work because they make extensive reading possible: reading large amounts of easy, interesting text for its meaning rather than to study it. Richard Day and Julian Bamford set out the approach in Extensive Reading in the Second Language Classroom (1998): easy material, plenty of it, chosen by the reader, read fairly fast for pleasure or information, without a dictionary at every turn.
Paul Nation and Rob Waring, who have written widely on graded readers, explain why vocabulary control is the key ingredient. A word has to be met many times before it sticks, and course books alone can’t provide enough meetings. Unsimplified text can’t either for most learners, because it holds too many unknown words to read in quantity. Graded readers solve both problems: a controlled set of words recurs across thousands of pages, at a density you can actually read.
Two other ideas point the same way. Stephen Krashen’s input hypothesis, developed in the late 1970s and 1980s, holds that we acquire language by understanding messages slightly beyond our current level, and a well-chosen graded reader delivers that kind of input page after page. Paul Nation’s four strands framework treats fluency development as a strand of its own, practiced with material that feels easy. Reading an easy graded reader quickly is fluency practice of exactly that kind.
How to choose the right level
The right level is the one where you know nearly every word: a few unknown words per page, not a few per line. Laufer (1989) suggested about 95% of words known for reasonable comprehension, and Hu and Nation (2000) found that about 98% was needed to read fiction without help. On a page of 200 words, those percentages translate like this:
| Unknown words on a 200-word page | Words known | What that level is good for |
|---|---|---|
| 0–1 | 99.5–100% | Fluency: read fast for speed and enjoyment |
| 2–5 | 97.5–99% | Your main reading: new words learned in context |
| 6–10 | 95–97% | Slow, careful study of a short text, not volume |
| More than 10 | Below 95% | Too hard for now: drop a level |
To test a book, read the first two pages without a dictionary and count. If you can follow what’s happening and meet no more than a handful of new words per page, you’ve found your level. If you’re guessing in every paragraph, drop down. Reading slightly too easy costs you little; reading too hard usually ends with an unfinished book.
A few habits make every book count:
- Keep moving. Skip a word the first time you meet it, and look it up only if it keeps coming back.
- Finish books. Volume matters more than any single title, so choose stories you want to know the end of.
- Mix levels. Do most of your reading at your main level and some a level below, for speed.
- Listen too. If the book has audio, listen to a chapter after you’ve read it.
- Keep a few words. Save the handful that kept recurring and review them with spaced repetition.
Where graded readers come from
Graded readers come mainly from educational publishers, and how many you can find depends heavily on the language you’re learning. For English the choice is enormous: series such as Penguin Readers, Oxford Bookworms and Cambridge English Readers offer original stories and adapted classics at many levels, often with audio and activities.
For widely taught languages such as Spanish, French, German and Italian, several publishers produce graded series, and there are dedicated graded readers for learners of Chinese and Japanese. Beyond that, the shelf thins quickly. Learners of Hungarian, Persian or Turkish may find only a few titles at each level, or nothing at the level they need.
Because you need many books rather than one, libraries, school resource centers and secondhand shops are worth using. And since levels aren’t standardized across publishers, a “Level 3” in one series can be harder than a “Level 4” in another. Judge by the page, not the label.
Where graded readers fall short
Graded readers are an excellent bridge from course books to unsimplified text, but they have limits worth planning around:
- Scarcity. Outside English and a few major languages, there may be little to read at your level.
- Narrow topics. Series lean toward fiction and adapted classics. Graded books on football tactics, chemistry or local politics are rare.
- Flattened style. Simplifying a novel strips out much of its voice, and the lowest levels can feel bland.
- A gap at the top. The highest levels stop at a few thousand headwords, while Nation (2006) estimated that reading unsimplified written English with 98% coverage takes 8,000–9,000 word families. The step from the top level to a real novel is still steep; see how many words you need to read.
- Someone else’s word list. A graded reader recycles the publisher’s list, not the words you’re trying to learn this month.
None of this makes them less useful. It means they work best as one source of reading among several.
What AI-generated graded text adds, and what it doesn’t
AI-generated graded text fills several of these gaps, but it trades editorial quality for flexibility. A language model can write a story or rewrite an article at a chosen level on demand, in any language it handles well, on any topic, and built around the exact words you’re studying, so those words recur while you’re reviewing them.
| Aspect | Published graded readers | AI-generated graded text |
|---|---|---|
| Quality control | Written to a word list and edited | No editor; level control is approximate |
| Length | Whole books, good for sustained reading | Usually short texts |
| Topics | Mostly fiction and adapted classics | Anything you choose |
| Languages | Plentiful for English, scarce for many others | Any language the model handles well |
| Your vocabulary | Recycles the series word list | Can be built around the words you’re learning |
The trade-offs deserve attention. AI text can include unnatural phrasing or plain errors, and quality varies from language to language. Level control is approximate, because no fixed word list governs what appears, and a rewrite can blur or drop details. Treat AI text as reading practice, not as a reference: check any grammar point you plan to reuse, and verify names, numbers and facts against the original before relying on them.
Using graded stories and rewrites in LingunAI
LingunAI produces graded text on demand, built around your own vocabulary. In StorySplice, you choose a language, a CEFR level from A1 to C2, a length of up to 500 words and the words you want to practice, picked from your flashcards or typed in. The AI writes a complete story at that level that uses those words, with a full translation below it and 8–15 vocabulary words explained.
The article reader does the same for real articles: paste the text, choose a level, and LingunAI rewrites it in the language you’re learning, keeping the facts, though the lowest levels may drop minor details. In both, you can select any word that stops you and save it as a flashcard.
It isn’t a library of edited books, and it shouldn’t replace one where good graded readers exist. A sensible split is published readers for long, sustained reading, and LingunAI for short texts aimed at your words, for topics no series covers and for languages with few graded books. The full method is in how to learn a language by reading.