Here is an experiment you can run on yourself in ten seconds. Conjugate hablar in the preterite. Most people who've had a year of Spanish can do it, haltingly, by walking down the list: hablé, hablaste, habló…
Now, without pausing, say: "I spoke to her on Tuesday but she'd already left."
The first task takes a second and feels like knowing Spanish. The second one is the one you actually wanted, and it tends to fall apart — not because the rule is missing, but because retrieving a rule and running down a list is far too slow to happen inside a sentence you're also busy constructing.
Two kinds of knowing
Researchers draw a line between explicit knowledge — facts about the language you can state — and implicit or automatized knowledge, the fast, unconscious system that produces the right form while your attention is on what you mean.
A conjugation table builds the first one very efficiently. That's what it's for, and an hour with a good table genuinely teaches you something you didn't know. The trouble is that the two systems don't convert into each other for free. Explicit knowledge is available when you have time to consult it — writing an email, editing, doing an exercise. Conversation does not give you that time. Under load, you use whatever is automatic, and a table is never automatic.
You can correct the mistake you just made
Everyone who's studied grammar formally knows this feeling: you say something wrong, hear it, and fix it a beat later. Your explicit knowledge is fine — it arrived, just too late to be useful. That lag is the entire gap between having learned a rule and having automatized it, and closing it takes a different kind of practice, not more tables.
What the research actually says
It's worth being careful here, because this argument gets overstated in both directions.
The largest synthesis of the question — Norris and Ortega's 2000 meta-analysis of instructed second language acquisition — found that focused grammar instruction produces large gains, and that explicit instruction outperformed implicit instruction. If you've been told that research proves grammar teaching doesn't work, that's not what it found. Teaching rules helps.
What it also found is more interesting: instruction that draws attention to a form while learners are dealing with meaning ("focus on form") produced effects roughly equivalent to instruction that isolates the form and drills it ("focus on forms"). Same size effect, and only one of the two also builds the ability to process the language in real time. If the outcomes tie, the version that happens inside real sentences is the better buy.
And then there's readiness
The other finding worth knowing is Manfred Pienemann's Teachability Hypothesis. Across a lot of studies, learners acquire certain structures in a fixed developmental order, and instruction can't make them skip a stage. Teaching a structure several steps ahead of where a learner currently is produces very little; teaching the one they're nearly ready for works well.
The practical read: a syllabus can't impose an order on your brain. When a rule "won't stick" despite being clearly explained, the most likely explanation isn't that you're bad at grammar. It's that you met it too early — and that meeting it later, after a few hundred encounters with the surrounding language, will feel almost easy.
Why the sentence is the right unit
A conjugation table teaches a form stripped of everything that would let you use it. A sentence teaches five things at once:
- The form — llegó.
- What it means here, which a table can't show: why preterite and not imperfect in this particular sentence.
- What it attaches to — the preposition it takes, the word order around it.
- How it sounds, including where the stress lands, which is precisely what distinguishes hablo from habló.
- A retrieval cue you'll actually encounter again, because sentences recur and paradigms don't.
That last point does most of the work. The reason grammar eventually becomes automatic is repeated exposure to the pattern in varied contexts, with meaning attached. A table gives you one context — the table — and no meaning at all.
What to do instead
None of this is an argument for never looking at a rule. It's an argument about sequence and proportion.
- Read the rule once, briefly. Two minutes and one clear example. Explicit knowledge is cheap to acquire and genuinely helps you notice the pattern in the wild. Just don't mistake having read it for having learned it.
- Then meet it forty times in sentences. This is the part people skip, and it's the part that does the work. Sentences you mostly understand, with the target form doing a job you can see.
- Practise the way you'll perform. If you want to speak, some of the practice has to be production under mild time pressure — saying it, not selecting it from four tiles. Skill transfers best to conditions that resemble the training.
- Let errors sit. A form you're not developmentally ready for won't be fixed by being corrected harder. Keep reading and listening; come back to it in a month and it will often have quietly resolved itself.
- Use tables as reference, not as study. They're a good place to check something. They're a bad place to spend an hour.
Could you do it without understanding the sentence?
Fill-in-the-blank drills often can be completed by pattern-matching the blank to the subject pronoun, with the meaning of the sentence entirely ignored. If an exercise can be passed without comprehension, it is training something other than the skill you want. The useful version forces you to understand the sentence in order to choose correctly.
Rules explained once, then practised in sentences
Lernlang teaches grammar as patterns inside sentences you can already mostly read, with the explanation kept short and the practice kept contextual. A placement check finds what you already handle so you're not re-drilled on it, and the exercises are built so the sentence has to be understood before the answer can be chosen — no guessing from the pronoun.
The summary
Grammar instruction works. Conjugation tables are a fine way to receive it, and a poor way to practise it. The rule takes two minutes to learn and several hundred encounters to automate, and almost all of the value is in the second part — which is why the learner who read one explanation and then read forty thousand words of Spanish ends up speaking better than the one who did the whole workbook.
If you take one thing from this: the question isn't whether you know the rule, it's whether the rule arrives in time. Only sentences teach that.