The Antimoon method: a critical review

Antimoon is a website started in 2001 by two Polish learners of English, Tomasz P. Szynalski and Michał Ryszard Wójcik, to explain how they reached a very high level without living abroad. Their advice was written for English, long before Anki was fashionable, and it still shapes how many self-taught learners think about input and flashcards.

Most of it has aged well. One part has not: the instruction to avoid mistakes, which sounds reasonable and quietly stops people from ever speaking. This review separates the two.

What the Antimoon method actually recommends

The method is often summarised as "input, no mistakes, SRS". The site's own overview lists six steps:

1. Motivation. Become someone who likes learning the language, because nothing else survives years of study.
2. Dictionary. Get a good one and use it constantly, including for pronunciation and example sentences.
3. No mistakes. Try to use correct language from the beginning.
4. Pronunciation. Learn the sounds of the language early and pay attention to how each word is pronounced.
5. Input. Read and listen to a lot of correct sentences. Antimoon calls this the most important part.
6. SRS. Put words and phrases into a spaced repetition program and review them regularly.

Two details are usually lost in retellings. First, Antimoon does not rank flashcards above input: input comes first, and the SRS is there to keep what input delivers. Second, the content is supposed to be enjoyable. The site is explicit that funny, thrilling material beats worthy material, and that grammar rules should be consulted only when needed.

The authors are also sceptical of classes, which in their view produce slow progress and learners who still make basic errors after years.

What Antimoon gets right

The input-first core matches what most acquisition research and most successful self-taught learners report: you cannot produce sentences you have never met. The method's model, in which correct sentences accumulate in your head until you can build similar ones, is a plain-language cousin of comprehensible input.

It was also early on three points that are now common advice:

• Pronunciation is trained at the start, not repaired later.
• A learner's dictionary with example sentences is a study tool, not a last resort.
• Flashcards work better with sentences and phrases than with bare word lists, an idea that later grew into sentence mining.

And it places responsibility with the learner. No teacher, app or method does the reading for you. Kungfucius has looked for a loophole here and can report that there is none.

Criticism one: the no-mistakes rule

Antimoon's page on mistakes states that you can damage your English by writing and speaking. The reasoning is that every wrong sentence you produce makes the same error more likely next time, until it becomes a habit. The recommended behaviour is to be careful and slow, use simple language, look things up, and, if you still make many mistakes, stop speaking and writing and concentrate on input.

There is a fair point inside this. Errors can settle in, and a learner who has said the same wrong thing for five years will struggle to change it. Linguists call the end state fossilisation.

The problems are with the dose.

• It treats production as a risk, not a skill. Speaking is partly a motor and retrieval skill. It improves by doing it under mild pressure, and input alone does not train it.
• It has no exit condition. "Stop speaking until you make fewer mistakes" does not say when to start again. Careful people can wait for years.
• The mechanism is the authors' own reasoning. That each uttered error reinforces itself is plausible, but Antimoon presents it as an argument, not as a research finding, and the site itself concedes that learning pronunciation involves many mistakes.
• It discards feedback. A mistake that gets noticed, by you or by a tutor, is one of the fastest ways to find out what your input has not yet taught you.

A fairer version of the rule: do not practise errors on repeat, and do not avoid speaking to prevent them. Kungfucius's longer argument is on the page about mistakes.

Criticism two: the SRS workload

The Antimoon authors describe reviewing in SuperMemo for about 15 minutes a day over eight years, and on many days spending another hour or two adding new items. The reviewing is modest. The adding is not.

An hour or two of card-making is an hour or two not spent reading or listening, which the method itself calls the most important part. For most learners with jobs, the SRS should be the smallest slice of the day, fed by words that input has already made familiar. Settings and daily limits are covered in the guide to spaced repetition.

There is also a temperament issue. The method suits people who enjoy building and maintaining a system. If you are not one of them, the collection becomes a second job, and the language becomes the thing you do after the job.

Criticism three: it was built for English

Szynalski and Wójcik learned a language with an alphabet they already knew, a vast supply of entertainment and no tones. The site notes that the method can be adapted to other languages, but it does not do the adapting. For Chinese, Japanese and Korean, three things change.

Antimoon assumes	What changes in Asian languages
You can start reading real texts early	Chinese and Japanese need several hundred characters before even graded readers are comfortable, so an explicit foundation phase comes first
Pronunciation is learned from dictionary transcriptions and listening	Mandarin, Cantonese, Thai and Vietnamese tones need early feedback from a person or a recording of yourself, which means speaking and getting it wrong
Correct output follows from enough input	Politeness levels in Japanese and Korean depend on who you are talking to, and are hard to absorb without real conversations

The tone point is where the no-mistakes rule fails most clearly. Nobody learns the Mandarin third tone by reading about it. You produce it badly, someone winces, and you adjust.

Does the Antimoon method work

It worked for its authors, and the input and SRS parts are well supported by experience elsewhere. Followed literally, it tends to produce learners with excellent reading and listening, a large passive vocabulary and a reluctance to open their mouths.

Whether that counts as working depends on your goal. For reading novels or passing the reading and listening sections of an exam, it is close to ideal. For living in Osaka or negotiating with a landlord in Taipei, it leaves a gap that only speaking closes.

The learner who waits until he can speak without mistakes will have a flawless silence.

What to take from Antimoon and what to change

Keep the spine, adjust the dose:

• Keep large amounts of enjoyable input as the main activity.
• Keep early attention to pronunciation, and for tonal languages add a tutor or recordings of yourself from the first weeks.
• Keep the SRS, capped at a small daily slice and filled from your own reading and listening.
• Change the no-mistakes rule to "notice and fix": speak early in low-stakes settings, and write down the corrections you receive.
• Add a foundation phase for characters or script before expecting input to carry you.

If you want this arranged as a daily plan, the Kungfucius method is the same idea with the speaking put back in.