How Apilingual works

There is a particular disappointment that only intermediate learners know. You finished the course. You can order coffee, describe your weekend, conjugate the present tense in your sleep. Then you open a real novel in the language — or stand next to two native speakers talking at ordinary speed — and understand almost none of it. The gap between what your app taught you and what the language actually is turns out to be enormous, and nothing on your phone was built to cross it. This page is about why that gap exists, what the research says about closing it, and exactly what Apilingual does and doesn't do about it.

The intermediate plateau

Beginner apps are extraordinarily good at the first thousand words and the core grammar, because that material is finite, orderable, and the same for everyone. But a language is not a thousand words. A typical novel draws on a working vocabulary many times larger than any beginner course covers, and its sentences are shaped by native intuition rather than by a syllabus. So the moment you step outside the curated content, you meet a wall: the words on the page are distributed by real-world frequency, not by lesson order, and most of them are ones you have simply never been shown.

This is the intermediate plateau, and it is where most learners quit. Not because they are bad at languages, but because the tooling abandons them precisely when the work gets hard. Streak-based beginner apps optimise for the stretch they are good at and then leave you at the edge of it. What you need next is not more of the same graded sentences. What you need is a way to read and hear the real thing without drowning in it — and something that keeps track of the ground you gain, because at this stage progress stops being a straight line and starts being thousands of individual words, each in its own state of half-learned.

Comprehensible input

The most influential answer to "how do people actually acquire a second language" is the input hypothesis, associated with the linguist Stephen Krashen. Its claim, in plain terms: you acquire language by understanding messages — by reading and listening to material you can mostly follow — rather than by consciously studying rules. Krashen's shorthand for the ideal difficulty is i+1: input just one step beyond your current level, comprehensible enough that meaning carries you, but with enough that is new to pull you forward. Comprehension does the teaching; the new bits get absorbed almost as a by-product of understanding the whole.

It is worth being honest about the state of this research, because overselling it is exactly what the streak-and-badge apps do. The input hypothesis is influential but contested. Critics have long argued that its central claims are framed so loosely they are hard to test or falsify — the philosopher-linguist Kevin Gregg's well-known critique makes exactly this case — and that "i+1" is more a useful metaphor than a measurable quantity. Few researchers today believe understanding input is the only thing that matters. What does hold up well is the weaker, more practical version: reading a lot of material at the right level helps, and helps meaningfully. Meta-analyses of extensive reading find a real, positive effect on comprehension and vocabulary — though the studies vary in quality and the gains take volume and time, not a fortnight. We build on the durable part of the evidence, and we would rather tell you it takes months than promise you fluency by summer.

Why input alone isn't enough past B1

Even taking comprehensible input seriously, reading by itself has a well-documented ceiling. You can meet a word a dozen times in context and still not quite own it — you recognise it when you see it, and lose it the moment you need to produce it. Vocabulary researchers such as Paul Nation have long described knowing a word as a matter of degree and of reaching the coverage a text demands: incidental exposure lays the groundwork, but deliberate, spaced retrieval is what moves a word from vaguely familiar to reliably usable. Input gets words in front of you; it does not, on its own, guarantee they stay or that you can summon them.

This is why a serious system has to do two things reading alone won't. First, it has to track vocabulary — to know which words you have met, which you are still shaky on, and which you truly own — because you cannot deliberately practise what you cannot see. Second, it has to treat mastery as more than a single number. Recognising a word when you read it, recalling its meaning when prompted, and producing it yourself are three distinct skills, learned in roughly that order and each earned a different way. Grinding the easy end of that ladder — reviewing words you already recognise — feels productive and teaches you almost nothing. The point of tracking is to spend your effort where it actually moves you: on the words at your edge, at the stage they are actually stuck.

How Apilingual operationalizes this

Apilingual is built to make that theory concrete in a tool you would actually use. It starts with the reader. You bring real content — an article, an ebook, a podcast, an audiobook, a video — and it becomes a lesson in which every word is one tap from its meaning. Reading the real thing stops being a war of attrition against the dictionary, so you can get the volume the research asks for.

Underneath the reading, every word carries a status — new, learning, known — and that status is yours, per word, per language. This is the part most apps hide inside an algorithm and we deliberately surface: because the app knows your words, it also knows what a given text will cost you, which is what lets it point you at material that is mostly known with just enough new. Words you are still learning feed a spaced-repetition schedule that brings each one back around the moment you are about to forget it, rather than on a fixed calendar — the hard ones sooner, the ones you have made yours fading into the background. That practice moves a word deliberately through recognition, recall, and production, giving each stage the kind of retrieval it actually needs. For learners starting closer to zero, guided stories provide a structured on-ramp — whole levels built to introduce the right words in the right order — before you graduate to content of your own.

One honest boundary: today, the live practice engine is a vocabulary system — tracked words, adaptive scheduling, and the recognition-recall-production progression. Grammar, writing, and translation practice are in development and are not switched on yet, so this page does not claim them. The reading-and-vocabulary loop above is what the app does now, and it is the part the evidence supports most directly.

What we deliberately don't do

A method is defined as much by what it refuses as by what it includes. Apilingual has no hearts, no lives, no punishment for putting the phone down for a week — the plateau is not a motivation problem to be solved with dread. It has no artificial lesson gates: your own library is open, and you read what you want to read in the order you want to read it, because the whole premise is that the content you actually care about is the content that keeps you going. And mastery cannot be farmed by grinding easy reviews. A word does not count as known because you tapped "easy" enough times; it counts when you have shown you can recognise, recall, and produce it. The number is meant to tell you the truth about where you stand, not to flatter you into a streak.

None of this is a shortcut. Comprehensible input works, but it works on volume and time, and we would rather say so than sell you a countdown to fluency. What we can honestly offer is the missing tooling: a way to read the real language without drowning, and a vocabulary you can see, own, and actually build on.