a field guide, built from the papers

Every language you've ever spoken, you learned by being soaked in it. Nobody has ever learned one from a worksheet.

Immersion learning is the attempt to reproduce that soaking on purpose. It has fifty years of research behind it, some of it genuinely good, and a lot of folklore piled on top that the research doesn't support. This site separates the two, then shows you how to build the thing in a house in the suburbs, around a job.

i+1
Krashen's shorthand: input a half-step past your level is what moves you forward
669,498
people in the biggest language-learning dataset ever assembled (Hartshorne et al., 2018)
317
experiments in the meta-analysis behind spaced practice (Cepeda et al., 2006)
Grade 6
when Canadian immersion kids reach near-native French comprehension. Their speaking never catches up.
the idea, stripped down

Three things the research actually agrees on

01

You acquire language by understanding messages

Not by studying how it works. Krashen's input hypothesis says comprehension comes first and grammar tags along as a side effect, which is why listening and reading do most of the work and speaking shows up late, on its own schedule. The catch people miss: the input has to be understood. A Chinese news broadcast at week one is not input, it's weather.

02

Volume and variety of exposure is the mechanism

Nick Ellis's usage-based work frames acquisition as your brain running statistics over everything it hears. More samples, from more speakers and contexts, produce a better estimate of the language. That's the whole argument for hours-per-day over hours-per-week, and for mixing your sources instead of living inside one podcast.

03

Input alone leaves you unable to speak properly

The Canadian immersion programs proved this the expensive way: kids with years of all-French schooling comprehended like natives and wrote and spoke with stubborn grammatical holes. Swain's output hypothesis explains why. Producing language forces processing that comprehension lets you skip. So immersion needs a speaking component eventually, or it builds a very fluent listener.


two versions of the same idea

Moving to the country, or faking it convincingly

Full immersion

The original: go live where the language is spoken. It works, and the Canadian data from 1965 onward is the proof, but it also comes with the catch that data found. Even total immersion produced students who couldn't speak accurately, because their classrooms were teacher-centred and nobody ever pushed them to produce anything. The country helps. It isn't magic.

Home immersion

The AJATT and Refold communities spent two decades working out how to reproduce the mechanisms without a plane ticket: native media for input, tutors over video call for the negotiation of meaning, phone in Mandarin for ambient exposure, Anki for spacing. The internet made the access problem trivial. What's left is the attention problem, which is harder and much less discussed.


the rest of the site

Where to go from here

"The best methods might also be the most pleasant, and, strange as it seems, language acquisition occurs when language is used for what it was designed for, communication." Stephen Krashen, Principles and Practice in Second Language Acquisition, 1982, p.7