Refold / Mass Immersion Approach

Refold↗ (the rebranded Mass Immersion Approach, formerly AJATT-inspired MIA) is a modern, staged roadmap for acquiring a language through massive comprehensible input first, with output and sentence mining layered in once your brain has enough raw material. It's the gym program for immersion: huge volume of reps now, gains you can speak later.

What it is

Refold is a structured, phase-based immersion method created by Matt vs. Japan (Matt Bonham) and a team of polyglots, evolving out of the Mass Immersion Approach (MIA) and Khatzumoto's AJATT ("All Japanese All The Time"). It takes the loose, hardcore "just immerse forever" ethos of AJATT and turns it into an actual roadmap with stages, milestones, and tooling — something a normal human with a job can follow.

The core bet is simple and unapologetically input-first: you cannot output what you have not first absorbed. So you front-load enormous amounts of listening and reading in your target language, train your ears and your intuition for months, and only deliberately practice speaking and writing once your comprehension is already strong. Speaking isn't banned — it just isn't forced before your brain is ready. It emerges on a foundation of thousands of hours of input.

Refold organizes the journey into roughly four stages:

• Stage 1 — Prep: learn the sound system, install your tools (Anki↗, a media player with subtitles, a dictionary), and pick your immersion content. Build a few hundred words of base vocabulary so input isn't pure noise.
• Stage 2 — Comprehension: the heart of the method. Pile up listening and reading hours. Start Sentence Mining — pulling real sentences you almost understand out of native content and feeding them to an Spaced Repetition (SRS) deck. The goal is to grow from "I catch a few words" to "I understand most of this without subtitles."
• Stage 3 — Output: once comprehension is high, deliberately activate the language. Speak with tutors and partners, write, get corrected, and let your mined vocabulary go from passive recognition to active production.
• Stage 4 — Polish: refine accent, fix fossilized errors, and push toward near-native range.

If you've read about Refold's intellectual cousins, you'll recognize the DNA: it's Comprehensible Input (Krashen) operationalized into a checklist, fused with the data-nerd tooling of the Anki community.

The evidence

Refold isn't a peer-reviewed study; it's a synthesis. But its load-bearing claims rest on real, well-cited research.

Input is the engine. Stephen Krashen's Input Hypothesis (i+1) argues that we acquire language by understanding messages slightly above our current level — not by studying rules. Decades of research on extensive reading and listening (work by Paul Nation, Stephen Krashen, and reading researchers like Beniko Mason) consistently show that high-volume comprehensible input grows vocabulary and grammatical intuition. Nation's work on vocabulary coverage is especially relevant: you need to know roughly 95–98% of the words in a text to read it comfortably, which is exactly why Refold tells you to mine the unknown 2–5% rather than drown in content you can't parse.

The Silent Period is real and useful. Krashen and Tracy Terrell, in The Natural Approach, described a Silent Period where learners build competence before producing speech. Refold institutionalizes this: don't grind output in Stage 2; let it incubate. This is also a cousin of Automatic Language Growth (ALG), which takes the "don't force output early" idea even further.

Spaced repetition fights forgetting. Sentence mining leans on Hermann Ebbinghaus's Forgetting Curve and modern work on Retrieval Practice & Interleaving: retrieving a word right before you'd forget it strengthens memory far more than re-reading. Anki's scheduling algorithm is a direct application of this.

Motivation matters more than method. Here's the honest part. Zoltán Dörnyei's research on the Science of Motivation shows that sustained effort — driven by engaging content you actually want to consume — predicts success better than any single technique. Refold's strength is that it points you at native media you enjoy. Its risk is the opposite: the rigid Anki grind can crush the Affective Filter and burn people out. Many Refold dropouts don't fail at immersion — they fail at the spreadsheet. Be honest with yourself about that.

What Refold gets right: input volume is non-negotiable, output is built on it, and tooling can systematize the boring parts. What it sometimes oversells: the optimal mining intensity, and the implication that more cards always means more gains. They don't. Comprehension hours do.

How to actually use it

No "fluent in 30 days" promises here. This is a multi-year program where the reps compound. Here's the no-bullshit Languy training plan.

1. Prep your gym (1–2 weeks). Learn the alphabet/sound system so you can hear words, not mush. Install Anki, grab a subtitle tool like Language Reactor & Immersion Tools, and a pop-up dictionary. Knock out a 500–1000 word frequency deck so your first immersion isn't total static.

1. Pick content you'd watch anyway. A show you love beats a textbook you tolerate. This protects motivation — your real bottleneck. Need ideas? See Finding Comprehensible Input and Comprehensible Input Platforms.

1. Immerse daily — log the reps. Aim for a sustainable daily dose (start at 30–60 min, build up). Do active immersion (full focus, looking things up) plus passive immersion (audio in your ears while you do chores). Volume is the gain. Track hours like sets at the gym — boring, but it's the truth.

1. Mine sentences, not word lists. When you hit a sentence you understand except for one word, turn it into an Anki card. One unknown per card ("1T" / one-target). This is the engine of Vocabulary Acquisition — context glues words to memory far better than isolated lists. (See Sentence Mining for the full technique.)

1. Do your Anki reps every single day. Skipping days lets the Forgetting Curve win and reviews pile into a hated mountain. Small daily sets. Keep the Affective Filter low — if Anki becomes torture, cut card count before you quit entirely.

1. Wait to speak — then go hard. Don't force output in Stage 2. When you can comfortably follow native content, then start Speaking: How Output Emerges with tutors, exchange partners, or your AI Language Coach. Output activates what input already installed.

1. Polish last. Once you're conversational, target Pronunciation & Accent and persistent grammar errors deliberately.

Languy's verdict: Refold is a brilliant skeleton. Don't let the SRS tail wag the immersion dog. More hours understanding stuff you love, fewer hours fighting your card backlog.

Resources

• Refold.la — the official free roadmap, guides, and language-specific decks. Search "Refold roadmap."
• Matt vs. Japan (YouTube) — the founder's videos explaining the philosophy and sentence-mining workflow.
• AJATT (alljapaneseallthetime.com) — Khatzumoto's original blog; the spiritual ancestor. Raw, fanatical, foundational.
• Anki — the free SRS that powers the whole method (paid on iOS as AnkiMobile). See Anki: The Complete Guide.
• Language Reactor↗ — browser extension for mining sentences from Netflix and YouTube. See Language Reactor & Immersion Tools.
• Stephen Krashen, The Input Hypothesis (1985) — the academic backbone of input-first learning.
• Paul Nation, Learning Vocabulary in Another Language — the research on vocabulary coverage and frequency.
• MIGAKU↗ — paid all-in-one mining toolkit built by people from the MIA scene.

Related

• Comprehensible Input (Krashen)
• The Input Hypothesis (i+1)
• Sentence Mining
• Spaced Repetition (SRS)
• Automatic Language Growth (ALG)
• The Silent Period
• Anki: The Complete Guide
• Language Reactor & Immersion Tools