The recognizable problem
You're three years into learning a language. You pick up a novel intended for native speakers and find you recognize most of the words on the page — eighty percent, maybe more. You feel real progress. Then you sit down for a conversation with a native speaker, and the same words don't come. You reach for one and what arrives is a blank, or its near-synonym from your weaker repertoire, or — worse — nothing at all and a long pause while your face heats up.
The standard explanation is that your passive vocabulary is bigger than your active vocabulary. You can recognize words you can't produce. The fix, the standard advice says, is to convert one into the other: drill flashcards in production mode, force yourself to use the words in writing and speaking, repeat until the recognition deepens into production capacity.
This framing is everywhere. It's in textbooks, in YouTube channels, in app marketing, in the Reddit threads where intermediate learners try to make sense of the intermediate plateau. It feels right because it describes the symptom accurately: there's a clear gap between what you recognize and what you can produce, and closing that gap feels like it should be the work.
But following the standard advice produces the same plateau most intermediate learners hit. You drill more. You force more production. The gap doesn't close, or it closes for a few words while the next layer of "passive vocabulary" rises up behind. After a while you start to suspect the work itself is doing the wrong thing.
It is. The framing is wrong — or wrong enough that working from it doesn't unlock anything.
Why the active/passive dichotomy misleads
The dichotomy implies two stores of vocabulary in your head. A passive store, where the words you recognize but can't produce live. An active store, where the words you can produce on demand live. And some conversion mechanism that moves words from one store to the other, presumably through retrieval practice.
This is a useful-sounding model. It's also, as far as anyone can tell from the way vocabulary actually behaves in real learners, not how it works.
There aren't two stores. There's one store with words at varying degrees of depth.
A word at shallow depth: you've met it once, in one context. You can recognize it again in something close to that context. You can't produce it freely because you don't have the felt sense of where it lives — what register it carries, what situations call for it, which near-synonyms it isn't, which collocations it travels with. You know the word in the same way you know a name you heard at a party last week: the connection is thin and the surrounding pattern is mostly missing.
A word at mid-depth: you've met it across three or four different contexts. You can produce it in those contexts but not in others. The pattern around it has started to fill in, but only along the edges you've actually traveled.
A word at deep depth: you've met it across many varied contexts, in different registers, attached to different topics, used by different speakers. You can produce it in conversations you haven't had yet, because what's in your head isn't a single example to retrieve — it's a felt range. You know where the word lives, even when you encounter it in a situation you've never seen before.
The "active vs passive" frame collapses this gradient — shallow, mid, deep, varying word by word, varying domain by domain — into a binary. Words are either in the active store or the passive store. Conversion is the work.
The binary is wrong. The gradient is right. And the difference matters, because it changes what the work actually is.
The real bottleneck — depth, not store-conversion
What feels like "passive vocabulary that needs activating" is mostly shallow depth on the word. The word is in your head, but only one slice of it is. You recognize the slice when you see it; you can't reach for it freely because there's no surrounding pattern to navigate to.
What feels like "active vocabulary" is words you've met across enough varied contexts that the pattern has filled in. You can deploy them across new situations because you have a sense of when they fit. The retrieval feels active because, from the inside, it doesn't feel like retrieval — it feels like reaching for the word that obviously belongs.
The gap between "I recognize this word" and "I can use this word" isn't a wall to break through. It's a depth gradient to fill in. The mechanism isn't conversion from a passive store to an active store. The mechanism is adding contexts to a word until you can interpolate to new ones.
This changes the diagnosis. The intermediate learner who recognizes thousands of words but can't speak isn't sitting on a vast passive vocabulary waiting to be activated. They're sitting on a vast vocabulary at shallow depth — words they've met once, in one context, with no surrounding pattern. The work isn't to "activate" what's already there. The work is to deepen what's already there by meeting it again across different contexts.
That distinction sounds small. It isn't. It changes what counts as useful practice.
Conversion-style work, applied to shallow-depth vocabulary, tends to make the shallow connection slightly less shallow — and only for that one context. You can drill the flashcard for ascertain a hundred times in production mode and you'll get faster at producing ascertain in the artificial flashcard context. You will not, by that practice alone, become better at producing it in a real conversation that calls for it in a register you haven't met before.
The plateau most intermediate learners hit is the plateau of having a lot of shallow-depth vocabulary and treating the problem as activation when the problem is depth.
Why "drill the words more" doesn't fix it
The standard advice for closing the active/passive gap is to drill more, and drill in production mode. Make the flashcards harder. Write sentences with the target words. Force production. Pile up the reps.
The trouble is that drilling the same word in the same context — even with the difficulty notched up — doesn't deepen its range. You become better at producing that word in that one context. You become better at producing it in the flashcard context. You don't become better at producing the word in situations you haven't seen.
This is the visible failure mode of the heroic Anki user with ten thousand cards in their target language. They can pass the flashcards. They've trained the retrieval connection to a degree most language learners would envy. And they still freeze in conversation, because what they've practiced is retrieval intensity within one context — the flashcard — while what conversation actually requires is range across contexts they've never seen.
Retrieval intensity within one context is not the same as range across contexts. The first is what flashcards train. The second is what conversation rewards. Drilling more of the first does not produce the second.
This is the part of the standard advice that lands wrong on intermediate learners. They drill harder, in production mode, with all the discipline their schedule allows. They do not get appreciably better at speaking. They blame themselves for not drilling enough, or for not drilling the right way, when the real problem is that the activity itself wasn't pointed at the right mechanism.
There's a second failure mode in the standard advice that's less talked about. The "active vs passive" frame implicitly treats vocabulary as words-in-the-abstract. A word is either active or passive — globally, as a property of the word. But vocabulary is domain-bound. You can be deeply ranged on words that come up in your work domain and shallow on the same surface forms in a casual register. You can be conversationally deep in one domain and barely functional in another. The "active vs passive" dichotomy can't see this, because it treats the word as the unit. The real unit is the word in a domain, and depth is per-pair.
This is also why the standard advice produces the same disappointment over and over. It points you at the word, when the actual work is at the level of the word-in-context. Drilling the word doesn't deepen the context. Only varied contexts deepen the context.
What actually deepens a word — varied contexts
In my own English-learning journey, the moments that built actual usable vocabulary weren't moments of structured study. They were years of reading software documentation in Russia — an accidental immersion across thousands of varied technical contexts, where I'd meet the same word in three different specifications in a single afternoon — and, much later, months of high-volume low-stakes real conversation while driving Uber Eats, where strangers in passenger seats kept handing me new phrasings of the same idea in voices my structured courses had never sampled. The mechanism that built usable English in my head was range across contexts, not retrieval intensity within them.
The longer version of that story belongs in a different piece. The mechanism is what matters here.
Practical implications
If the real problem is depth and the mechanism is range, the practical advice changes.
Stop treating active/passive as a binary. Treat each word as having a depth profile, varying across domains. The work isn't to "convert" any specific word from passive to active. The work is to deepen the words you care about — across enough varied contexts that the felt sense fills in.
For most intermediate learners, that means meeting the same word across multiple different sentences, in different domains, in different registers — five varied sentences for one word will do more for usable vocabulary than fifty repetitions of one pair. This is true even when the fifty repetitions are with production-mode flashcards. The fifty repetitions deepen one context. The five varied sentences deepen the word's range.
The practical sources of varied contexts are the ones you'd expect, and they're worth naming because they tend to get dismissed in favor of more drilling. Reading native material at your level forces you to meet familiar words in unfamiliar combinations. Listening to native speech — podcasts, interviews, films, anything that has people talking instead of presenting — gives you the same words across speakers, registers, and topics. Real conversation with native speakers, even short and clumsy, puts the words into your hands in contexts that aren't prefabricated. Comprehensible input in the Krashen sense — material at your level plus a little — operates on the same principle: meaning emerges from encountering language in varied real contexts, not from analytical drilling. The post-Duolingo habit guide walks through specific ways to build these inputs into a daily routine.
And, for the words you specifically want to deepen, tools that explicitly give you multiple varied sentences per word can do the same work as wide reading, more efficiently, for the words you've actually chosen. Anki shared decks, with their context-collapsed translation pairs, don't do this. Tools built around context-rich examples do.
What this looks like over time: the "active vs passive" dichotomy quietly stops being a useful frame. You start noticing depth instead. Words you recognize but can't quite use shift into words you can use after a few more encounters in varied contexts. Words you couldn't even recognize a month ago start showing up in new sentences, and each encounter deepens them further. The gap doesn't get closed by conversion. It gets filled in by depth.
Where LAFwords fits
LAFwords is built around exactly this principle. Each word you study arrives across ten varied real-world sentences spanning different senses and registers, returning at the moment you're about to lose it. The ten sentences aren't there to teach you to use the word in those exact ten situations — they're there to give you a felt sense of the word's range, so you can deploy it in the hundred situations you haven't seen yet. Spaced repetition (FSRS) cements the contexts; the leverage is the spread.
LAFwords isn't a substitute for reading and listening to native material. Wide native input is irreplaceable; nothing about LAFwords' design tries to replace it. What LAFwords does is give you depth on specific words, on demand, when you've decided you want to deepen them — the way you might once have used a flashcard, but with the unit of practice changed from the pair to the range.
In practice this means: you encounter a word in a novel or a podcast or a conversation, type it into LAFwords, and within a minute or two it comes back as a card with ten varied sentences across senses and registers, with native pronunciation and an image. You meet the word again across those ten contexts, then again when spaced repetition brings it back. The word deepens — not because it's been drilled more, but because it's been met in more places.
For intermediate learners who've been stuck on the active/passive plateau, this is the change in mechanism that actually moves things. It's not "active vs passive." It's depth across varied contexts, on the specific words you want to deepen.
7-day free trial, no credit card. The principle holds with or without the specific tool.
Closing
The intermediate plateau isn't a wall between you and "active" vocabulary. It's a gradient of depth across the vocabulary you already have. The work is to deepen the words you have — across varied contexts — not to acquire some other kind of vocabulary that's stored differently.
The standard advice is wrong in a small way that produces a large effect. It points you at the word and asks you to drill harder. The mechanism that actually deepens vocabulary points you at the contexts and asks you to vary them.
If you take one thing from this piece: stop counting words and start counting contexts per word. The number that predicts whether you can use a word isn't how many times you've seen it. It's how many varied situations you've met it in. Ten varied contexts will give you the word's range and neighborhood — and through that, the ability to use it in the hundred contexts you haven't seen yet.