There's a specific kind of language learner who can conjugate anything, has finished three courses, knows what the subjunctive is for — and cannot follow two Spanish speakers talking about their weekend. It's an extremely common profile, and it isn't caused by lack of effort. It's caused by a diet almost entirely made of exercises and almost never of actual language.
The counterweight to that is an idea from the linguist Stephen Krashen, developed in the late 1970s and 80s and argued about ever since: we acquire language when we understand messages slightly beyond our current level. He labeled it i+1 — your current competence (i) plus a small step past it.
Learning versus acquiring
Krashen's most useful distinction is between two things people lump together. Learning is conscious: rules, tables, explanations, the stuff you can recite. Acquisition is unconscious: the sense that "en la mesa" sounds right and "en el mesa" sounds wrong, with no rule retrieved in between.
Fluent speech runs on acquisition. There isn't time to consult a rule mid-sentence. Conscious learning is useful — it speeds things up, it lets you self-correct, it stops you inventing patterns that don't exist — but it's a support system, not the engine. The engine is a large volume of understood input.
Nobody learned their first language from a grammar table. They spent years drowning in language they mostly understood, at a level that kept nudging upward.
The strong form of Krashen's hypothesis — that input is all you need and conscious study contributes nothing — is not where the field landed. Most researchers now think explicit instruction and deliberate output help, especially for adults, who don't have a decade of daily immersion to spare. But the core claim has held up well: massive comprehensible input is necessary, and it's the part most learners are missing.
What "comprehensible" actually means
This is where good intentions usually die. A learner decides to embrace input, puts on a Spanish Netflix series, understands roughly one word in six, and quits after twenty minutes with a headache. That wasn't input. That was noise.
Reading research gives fairly concrete numbers here. Work by Paul Nation and others on lexical coverage — the share of words in a text you already know — suggests that for comfortable independent reading you want to know around 98% of the running words. That's roughly one unknown word in fifty, or about two per page. Around 95% you can still get by with effort and context. Below that, comprehension falls apart, and so does your ability to guess new words from context, because the context itself is full of holes.
98% sounds impossibly high until you realize how skewed word frequency is. The most common couple of thousand word families do most of the heavy lifting in ordinary speech and everyday text. This is the whole argument for frequency-ordered vocabulary: you're not memorizing words, you're buying access to content. Each chunk of high-frequency vocabulary unlocks a whole tier of material you can now actually read.
Can you follow the story without stopping?
The practical version of the 98% rule: if you can read a page and follow what's happening, pausing once or twice at most, it's at your level. If you're stopping every sentence to look something up, it's too hard — not because you're not ready to work, but because you've stopped reading and started decoding. Decoding builds very little.
Why "slightly over your head" is the sweet spot
Input that's fully comprehensible teaches nothing new. Input that's incomprehensible teaches nothing at all. The band in between — where you follow the meaning, but a handful of words and structures are new — is where acquisition happens, for a reason that's almost mechanical.
When you understand the situation but not one word in it, your brain has everything it needs to assign that word a meaning: context, expectation, the shape of the sentence. Meet the word again in a different comprehensible context and the meaning sharpens. Meet it a dozen times and you own it — not as a translation pair, but with its register, its typical collocations, the prepositions that follow it. That texture never comes from a dictionary entry.
This is also how grammar goes from "known" to "automatic." You can be told that Spanish uses the imperfect for background and the preterite for events. That's ten seconds of learning. What makes it automatic is meeting the contrast a few hundred times in stories where it obviously mattered.
The affective filter, or: stress makes you stupid
Krashen's other contribution worth keeping is the affective filter: anxiety, boredom, and self-consciousness act like a valve on how much input gets through. Two learners can receive identical material and acquire wildly different amounts depending on whether they're relaxed and interested or tense and bored.
The practical implication is unromantic but real: material you enjoy beats material that's objectively better for you. A trashy thriller you're actually curious about will teach you more than a literary novel you're slogging through, because you'll do far more of it. Volume beats prestige, every time.
Where learners get stuck
The intermediate wall
Beginner material is everywhere. Native material assumes ten thousand words. The gap between them is where most people quit — there's simply not enough content at the level where you're ready to read fluently but not ready for a newspaper. Graded readers exist for exactly this reason, and the learners who break through are almost always the ones who found a supply of level-appropriate stories and read a lot of them.
Waiting to feel ready
There is no point at which you'll feel prepared to start reading. You start reading badly, with something easy, and it gets less bad. The first easy book is the hardest one you'll ever read.
Confusing looking things up with reading
Dictionary lookups are useful in small doses and corrosive in large ones. If you look up every unknown word, you never build the tolerance for ambiguity that real listening demands — and real conversation offers no pause button. Look up words that block the plot. Let the rest go; you'll meet them again.
Input without any consolidation
The mirror-image failure. Pure extensive reading works, but slowly, because a word met once in chapter three and not again for six weeks usually evaporates. Pairing input with a bit of spaced repetition — pulling the words you actually met into a review queue — dramatically speeds up the climb. Input creates the memory; spaced review keeps it alive until natural exposure takes over.
How to actually do this
- Front-load high-frequency vocabulary. The first couple of thousand common words are what make content comprehensible in the first place. This is the one phase where deliberate study pays enormous dividends.
- Read at 95–98% known words. If you're stopping constantly, drop a level. There's no prize for suffering.
- Choose interesting over impressive. Enjoyment determines volume, and volume determines results.
- Read a lot, look up a little. Two or three lookups a page, not twenty.
- Recycle what you met. Send the words that keep tripping you into a review queue rather than hoping to bump into them again.
- Add listening once reading is comfortable. Same principle, harder mode — no pausing, no rereading.
Built to close the intermediate gap
The intermediate wall is the reason Lernlang exists. The app front-loads the core Spanish vocabulary that covers the overwhelming majority of what you'll meet in real shows, books, and conversations, teaches grammar as patterns inside real examples rather than tables, and then hands you stories written to match what you already know — so you're reading in that 95–98% band from early on instead of waiting years for native material to become bearable.
When you do want to go outside the app, you can photograph a page of a book or paste in song lyrics; Lernlang translates it and drops the unfamiliar words into your review deck. Your own content becomes your input, and the words you meet in it stop disappearing.
The short version
Exercises tell you about a language. Input gives you the language. The goal of every course, deck, and grammar explanation should be to get you to the point where you can consume real content — and then to get out of the way, because from there the content does the teaching.
If you want one change that reliably moves people forward, it's this: cut your exercise time in half and spend it reading something you can mostly understand and genuinely want to finish.