Is vocabulary more important than grammar?

For most learners at most stages, yes. Broken grammar usually still communicates, as anyone who has understood 'yesterday I go store' can confirm, but a missing word communicates nothing at all. Vocabulary knowledge is also one of the strongest predictors of reading comprehension in a second language, ahead of syntactic knowledge in most studies. Grammar still matters for precision, and past a certain point imprecise grammar changes your meaning. The point is about sequencing rather than importance: vocabulary is usually the constraint that runs out first.

How many English words do I need to know?

It depends on what you want to do. Around 3,000 word families covers most everyday conversation. Reading novels and newspapers without help takes roughly 8,000 to 9,000 word families, which is the level at which about 98 percent of the running words on a page are already familiar. An educated adult native speaker sits somewhere around 15,000 to 20,000. Word families rather than words, so govern, governs, governed and government count once between them.

Does the language you speak limit how you think?

Not in the strong sense the famous quote implies. The idea that you cannot think what you have no word for, usually attached to the Sapir-Whorf hypothesis, has not held up well. People have thoughts they struggle to articulate, coin words for new concepts, and grasp distinctions their language does not mark. What does hold is a weaker version: the words you have influence what is easy to notice, hold onto, and say quickly. For a learner that weak version is the useful one.

Why Vocabulary Matters More Than Grammar

Vocabulary matters more than grammar because a missing word doesn’t produce a wrong sentence. It produces silence. Broken grammar still communicates; an empty slot doesn’t. Research consistently finds vocabulary size to be the stronger predictor of how much you actually understand, and it is the constraint most learners never notice hitting.

What Wittgenstein actually said

The line everyone quotes:

The limits of my language mean the limits of my world.

Ludwig Wittgenstein, Tractatus Logico-Philosophicus, proposition 5.6.

You have probably seen a different version. “The limits of my language are the limits of my universe” circulates widely on posters and study accounts, and it is not what he wrote. The German runs “Die Grenzen meiner Sprache bedeuten die Grenzen meiner Welt”: bedeuten is “mean,” Welt is “world.” The differences look cosmetic and aren’t. “Mean” describes a relationship where “are” asserts an identity, and “universe” reaches for something grander than the word he chose.

The larger misreading is what the sentence was for. The Tractatus is a book about logic and the boundary of what can meaningfully be said. Proposition 5.6 sits inside an argument about the structure of thought, not inside advice about how many words to learn this month.

So borrowing it for language learning is a repurposing. It happens to be a good one. But it is worth knowing that you are repurposing it, because the literal reading is exactly where the idea breaks, and that comes up later.

The limit you can’t see

Here is the strange thing about a vocabulary gap. It does not feel like a gap.

When your grammar fails, you notice. You hear yourself say “I have went” and something snags. Feedback arrives, from a teacher, a correction, or your own ear catching up half a second late.

A missing word sends no such signal. You don’t reach for it and find it broken. You don’t reach for it at all. The sentence quietly routes around the hole: you say “the money you get back when you return something” instead of refund, or you drop the point and say something simpler that you can actually say. From the inside that feels like having spoken. It almost never feels like having lost something.

This is the part of Wittgenstein’s line that earns its place in a learning context. Not that language builds reality, but something far more mundane and more useful. You cannot miss what you cannot name. The edge of your vocabulary is invisible from where you are standing, which is precisely why it is the constraint people underestimate.

Vocabulary or grammar: which one is stopping you?

Run the comparison directly.

“Yesterday I go store and buy three apple.” Tense, plural and article are all wrong, and you understood every bit of it.

Now reverse it. “The tribunal remanded the case pending disclosure.” The grammar is faultless, and if those content words are not in your vocabulary the sentence tells you nothing whatsoever.

That asymmetry shows up in the research as well. Vocabulary knowledge ranks among the strongest predictors of second-language reading comprehension, and studies that weigh it against syntactic knowledge generally find vocabulary carrying more of the load. It makes sense mechanically: grammar encodes the relationships between ideas, vocabulary supplies the ideas themselves. A relationship can often be inferred from context. Content you don’t have cannot be inferred from anything.

None of this makes grammar optional. Past a certain level imprecision costs you real meaning, and “I’ve been working here for three years” and “I worked here three years” are different facts about your life. The claim is narrower than the headline suggests, and it is about sequencing rather than value. For most learners, at most stages, vocabulary is the binding constraint. It is what runs out first.

How much vocabulary buys how much understanding

Vocabulary research usually works in terms of coverage: the share of running words in a text that you already know. The thresholds turn out to be sharper than you would expect.

Roughly 95 percent coverage is treated as the floor for reasonable comprehension with some guessing from context. Around 98 percent is where unassisted reading becomes comfortable, because at that density only about one word in fifty is unfamiliar, which is few enough to absorb without stopping.

That sounds reachable until you see the price in word families.

Vocabulary (word families)	Roughly what it covers
1,000	Most of everyday casual speech
2,000	About 80 percent of running words in ordinary written text
3,000	Most everyday conversation; the base for graded readers
5,000	Around 95 percent of written text, enough to read with support
6,000–7,000	About 98 percent of spoken discourse
8,000–9,000	About 98 percent of novels and newspapers, unassisted
15,000–20,000	The range of an educated adult native speaker

Word families rather than individual words, so govern, governs, governed, government and governing count once between them.

The shape of that table is the whole story. The first 2,000 word families do an enormous amount of work, which is why beginners feel fast. Then the curve flattens hard. Moving from 80 percent to 98 percent coverage costs roughly four times the vocabulary, because each word you add appears less often than the one before it. You pay more per word for less visible return, and it happens at exactly the stage where progress stops feeling like progress.

Why intermediate feels like a plateau

That curve explains the most common complaint in language learning. Everything improved steadily for a year, and then it seemed to stop.

Nothing stopped. The high-frequency words were arriving almost incidentally, because they show up often enough that ordinary reading and listening teach them for free. The mid-frequency band does not behave that way. A word sitting in the 5,000 to 9,000 range might appear once across several months of reading, which is nowhere near often enough for incidental acquisition to do the job.

This is the honest case for deliberate vocabulary study, and it is a narrow one. You do not need to drill the first 2,000 words, because exposure will hand them to you anyway. You need a system for the mid-frequency band, because exposure alone will not get you there in any reasonable span of time. We worked through the numbers in how many words you need to be fluent.

Knowing a word and having a word

There is a second gap hiding inside the first one.

Every count above measures recognition. You see the word, you know what it means. That is not the same as producing it at the moment you need it, and the distance between the two is wide. Most learners recognise several times more words than they can actually deploy.

For Wittgenstein’s line, this is the part that bites. The ceiling on what you can express is not your recognition vocabulary. It is the much smaller set you can retrieve inside the half-second a real conversation allows. A word you would recognise instantly on a page but cannot summon in speech is, in any practical sense, not available to you.

Which makes “learn more words” an incomplete instruction. Moving a word from recognition into production is a different job with different mechanics, and it is the one that changes how you sound. We covered how that works in active versus passive vocabulary.

Where the quote overstates it

Now for the part the posters leave off.

Read literally, “the limits of my language are the limits of my world” is strong linguistic determinism: the claim that you cannot think what you have no word for. That idea, usually filed under the Sapir-Whorf hypothesis, has not held up. People have thoughts they struggle to put into words, invent vocabulary for concepts that had none, and understand distinctions their own language never marks. If language strictly determined thought, none of that could happen.

What survives is a weaker and better-supported claim. Language influences what is easy to notice, easy to hold in mind, and easy to say quickly. Not what is thinkable, but what is reachable.

For a learner the weak version is more than enough, and it has the advantage of being true. You are not a smaller person in English than in your first language. You are a slower one, a blunter one, someone constantly routing around holes. The thought is already there. What is missing is the direct path to it.

That reframes the goal in a way worth keeping. You are not building a larger mind. You are deleting detours, which turns out to be the same project as thinking in English.

What to actually do about it

If vocabulary is the binding constraint, the response is narrower than “read more.”

Work in frequency order. Rare words feel more impressive and pay less. The mid-frequency band is where the return is, because that is where the gap between your vocabulary and the text sits.
Learn words inside sentences. A word paired with a single translation trains a lookup. A word met in a real sentence, with its typical partners, trains the thing you need in conversation.
Space the reviews. Mid-frequency words appear too rarely for natural repetition to maintain them, so the repetition has to be scheduled instead.
Push words into production. Recognition is cheap and it plateaus. Produce the word from a prompt rather than reading it and nodding.
Keep the daily load finishable. A number you hit every day beats a bigger number you abandon in week three.

How Vocaby fits

Vocaby is built for the band this article is about. The library runs past 29,000 English words, which reaches well into the mid-frequency range that ordinary exposure delivers too slowly to be useful.

Every entry carries IPA, real audio, a plain-English definition and example sentences, so a word arrives as a usable thing rather than a translation pair. FSRS spaced repetition then decides which words return on a given day, which is what keeps mid-frequency vocabulary from quietly decaying between the rare occasions you meet it in the wild. You can browse the range in the word library.

Wittgenstein was writing about logic, not about English study. Borrow the sentence anyway. Just borrow the modest reading of it: the words you can reach are the thoughts you can deliver, and that set is a good deal smaller than the set you recognise.

Download Vocaby on the App Store to build the vocabulary that actually reaches your mouth, with audio, IPA, examples and spaced repetition on every word.