“I understand English, but I can’t speak it.”
This is one of the most common frustrations in language learning.
You watch a video and understand most of it.
You read an article without translating every sentence.
Someone speaks to you in English and you know what they mean.
Then they ask:
What do you think?
And suddenly the language seems to disappear.
You know the words.
You know the grammar.
You understood the question.
But producing an answer feels completely different.
Students often interpret this as a contradiction:
If I understand English, why can’t I speak English?
There is no contradiction.
Understanding a language and producing it require overlapping but different operations.
And once we understand that distinction, the problem becomes much easier to diagnose.
Recognition is not production
Imagine seeing the word:
although
You understand it immediately.
Now close the page.
Can you retrieve the same word at exactly the moment when your own sentence requires that relationship?
That is a different task.
Recognition begins with something already present.
The word appears.
The sentence appears.
The speaker provides the language.
Your brain has to interpret it.
Speaking begins before the language exists.
You have an intention.
A situation.
A relationship between ideas.
And now you must construct language for it.
That difference is fundamental.
When you listen, someone else has already made thousands of decisions
Consider a simple sentence:
I would have gone if I’d known.
When you hear it, the speaker has already done enormous linguistic work.
They have selected:
- the relevant events;
- their relationship;
- the counterfactual perspective;
- the tense structure;
- the auxiliary verbs;
- the contraction;
- the word order;
- the vocabulary;
- the pronunciation;
- the rhythm.
You receive the finished product.
Your task is to reconstruct its meaning.
When you speak, you must make those decisions yourself.
That is why comprehension can develop much further than spontaneous production.
Vocabulary can exist in several different states
Students often say:
I know this word.
But what does know mean?
A word may be:
recognizable — you understand it when you see it;
retrievable with effort — you can remember it after several seconds;
available in controlled practice — you can use it when the exercise clearly requires it;
available spontaneously — it appears when you need it during real communication;
flexible — you can use it across different contexts, registers, and combinations.
These are not identical levels of knowledge.
A learner may recognize 10,000 words while having far fewer readily available for spontaneous speech.
That is normal.
The solution is not necessarily to learn another thousand words.
Sometimes the existing vocabulary needs to become more accessible.
Grammar can have the same problem
Consider a learner who understands the difference between:
I went.
and
I have gone.
They can explain the rule.
They complete grammar exercises correctly.
But during conversation, the distinction collapses.
The immediate reaction is often:
I need to study the grammar again.
Perhaps.
But perhaps the grammar is already understood.
The real problem may be that spontaneous speech gives the learner no chapter title saying:
USE PRESENT PERFECT NOW.
The learner has to interpret the situation first.
What am I trying to communicate?
Which relationship matters?
What perspective am I taking?
Only then can the grammatical form be selected.
Knowing a grammar rule is therefore not the same as being able to make the grammatical decision in real time.
Speaking begins before words
This is why “just speak more” is incomplete advice.
Before a sentence appears, something else happens.
You decide:
What do I want to say?
What matters most?
What does the listener already know?
Am I describing an event, result, process, possibility, obligation, opinion, contrast, cause, or condition?
How direct should I be?
What register fits this situation?
Language gives form to those decisions.
Speaking difficulty can therefore occur at several levels.
The learner may lack a word.
Or grammar.
But the learner may also be slow at converting an intention into a linguistic structure.
That is a different bottleneck.
Translation can help — and it can also become an extra step
Another common explanation is:
You can’t speak because you translate in your head.
This is too simplistic.
Translation is not inherently bad.
Comparing languages can be extremely useful.
It can reveal differences in structure, meaning, and conceptual organization.
The problem arises when every spontaneous sentence requires a long chain:
idea → complete sentence in first language → translation → grammatical checking → pronunciation → speech
That is a lot of processing.
A more efficient system gradually develops shorter routes between communicative intention and the target language.
But that does not happen because someone orders the learner:
Stop translating.
It develops through repeated meaningful use and increasingly direct associations.
The problem may be retrieval speed
Suppose you know the word you need.
You can remember it in ten seconds.
In a vocabulary exercise, ten seconds is acceptable.
In conversation, ten seconds can feel enormous.
The other person is waiting.
You become aware of the silence.
You start searching harder.
Another word disappears.
Then grammar becomes uncertain.
Finally you say:
Sorry, my English is terrible.
But perhaps your English is not terrible.
Perhaps retrieval is too slow for the communicative conditions.
That suggests a very different intervention from learning more grammar.
The problem may be too much monitoring
Advanced learners often face the opposite problem.
They know a great deal.
And precisely because they know so much, they can detect many possible mistakes before speaking.
A sentence begins internally.
Then:
Is that the right tense?
Should it be in or on?
Is that word formal enough?
Maybe the article is wrong.
Would a native speaker say it like that?
Perhaps there is a better expression.
The learner keeps editing a sentence that has not yet been spoken.
Knowledge has become a monitoring system.
The result can be paradoxical:
more knowledge, less fluency.
The goal is not to remove accuracy.
It is to learn when accuracy monitoring should intervene and when communication must continue.
The problem may be dependence on prompts
Classroom exercises often contain hidden support.
Complete the sentences using the Present Perfect.
Choose one of these five words.
Answer using the vocabulary from Unit 7.
Transform the sentence.
Match the expressions.
The student performs well.
Then someone asks an unexpected question at dinner.
All the support disappears.
This reveals something important.
The learner may have developed conditional access to the language.
When the task identifies the linguistic tool, the learner can use it.
When reality provides only a communicative problem, the learner must select the tool independently.
That selection itself needs training.
Conversation practice does not automatically solve everything
At this point the obvious solution seems to be:
I just need more conversation.
Often, more conversation is useful.
But conversation can also reproduce the same limitations indefinitely.
Imagine a learner who speaks every week using:
- the same safe vocabulary;
- the same grammatical structures;
- the same avoidance strategies;
- the same approximations.
Communication succeeds.
Nothing forces the system to change.
After two years, the learner has accumulated hundreds of hours of conversation but still feels trapped inside the same language.
Practice becomes developmental when it contains enough challenge, feedback, noticing, and reconstruction to modify the system.
So the question is not simply:
How much do you speak?
It is:
What changes because you speak?
Mistakes are not necessarily the main problem
Some learners become obsessed with eliminating mistakes before speaking freely.
But communication cannot wait until the language system becomes perfect.
Children do not wait.
Native speakers do not speak perfectly in every situation either.
Adults reformulate, hesitate, abandon sentences, choose imprecise words, and repair misunderstandings constantly.
The useful ability is not:
never produce an imperfect sentence.
It is:
continue operating when language is imperfect.
That includes paraphrasing.
Repairing.
Clarifying.
Asking.
Restarting.
Using a simpler structure.
These are not signs of failed communication.
They are communication skills.
Paraphrasing is one of the missing bridges
Suppose you want to say:
screwdriver
but you cannot remember the word.
A learner dependent on exact vocabulary stops.
A more flexible speaker might say:
The tool you use to turn a screw.
The exact word is missing.
Communication continues.
This ability changes the entire experience of speaking.
You no longer need complete linguistic availability before opening your mouth.
You need enough language to construct another route.
Paraphrasing therefore deserves deliberate practice.
It converts partial knowledge into usable communication.
Understanding can create a dangerous illusion of readiness
Comprehension feels convincing.
You watch an English-language video.
Everything makes sense.
You think:
Yes. I know English.
Then speaking exposes a different reality.
This can feel humiliating.
It should not.
The comprehension was real.
It simply measured another capability.
This distinction exists far beyond languages.
A mathematics student can understand a teacher’s solution and still be unable to solve a new problem.
A programming student can understand code and still be unable to design a solution from scratch.
A chemistry student can recognize a reaction and still fail to predict what happens under changed conditions.
Following someone else’s completed reasoning is easier than generating your own.
Language is another instance of the same phenomenon.
Why repeating after videos is not enough
Shadowing and repetition can be useful.
They train pronunciation, rhythm, processing speed, and familiarity with structures.
But repeating:
I would have done it differently.
after a speaker is not the same task as independently deciding to express a counterfactual judgment and constructing the sentence yourself.
One trains reproduction.
The other trains generation.
Both can belong in learning.
They should not be confused.
A useful progression moves from support toward independence
Instead of jumping directly from explanation to completely free conversation, learning can move through changing conditions.
For example:
Stage 1 — Recognition
Can you understand the structure?
Stage 2 — Reconstruction
Can you rebuild it when part of it is removed?
Stage 3 — Controlled retrieval
Can you produce it when the communicative situation clearly requires it?
Stage 4 — Choice
Can you decide whether the structure is appropriate among alternatives?
Stage 5 — Spontaneous production
Can you use it while attention is primarily on meaning?
Stage 6 — Transfer
Can you use the underlying distinction in a new context that was never practised?
This is not a rigid universal method.
It is a way of seeing why:
I understand it.
does not automatically imply:
I can use it.
Your teacher should diagnose where the chain breaks
This is also why individual instruction should not simply provide more material.
If comprehension is already strong, more explanations may not address the bottleneck.
The useful question becomes:
Where does the transition fail?
recognition → reconstruction?
reconstruction → retrieval?
retrieval → decision?
decision → production?
production → transfer?
Two students who both say:
I understand English but I can’t speak it.
may therefore need completely different work.
One needs vocabulary retrieval.
Another needs less monitoring.
Another needs grammar decisions under time pressure.
Another needs pronunciation confidence.
Another needs experience maintaining communication with incomplete knowledge.
The symptom is shared.
The mechanism is not.
How often should you practise speaking?
Frequency matters, but there is no universal number.
We explored the broader problem of lesson frequency here:
For speaking specifically, frequent contact can be valuable because retrieval needs repeated activation.
But ten minutes of genuinely independent production may sometimes reveal more than an hour of heavily supported conversation.
Again, quantity is not the entire variable.
The conditions matter.
Do you need a native speaker?
Not necessarily.
Native-speaker interaction can provide extremely valuable exposure to natural language, pragmatics, rhythm, vocabulary, and real communicative variation.
But the ability to diagnose why a learner understands but cannot produce language is a pedagogical problem, not simply a native-speaker-status problem.
We examine that distinction here:
The useful teacher is the one who can identify what is preventing the learner from moving from comprehension toward independent production.
Private lessons are valuable only if they target the bottleneck
If you already understand English reasonably well, paying someone to explain material you already understand may create very little additional value.
The lesson should respond to the actual problem.
That is why we argue that the value of private teaching lies not simply in access to an hour of teacher time but in the quality of instructional decisions made during that hour:
Individual instruction becomes meaningful when the learner’s response changes what happens next.
What can you do if you understand English but cannot speak it?
Do not begin by assuming that you need more of everything.
Instead, identify where speech fails.
Ask yourself:
Can I retrieve words without seeing them first?
Can I explain an idea when the exact word is missing?
Can I choose grammar without being told which grammar is being tested?
Do I construct sentences directly, or translate complete sentences before speaking?
Do I stop because I lack language, or because I am afraid the language is imperfect?
Can I continue after making a mistake?
Can I speak about unfamiliar topics, or only rehearsed ones?
Does my language survive when I have less time to think?
The answers point toward different interventions.
Test availability, not familiarity
Here is a simple principle.
Do not ask only:
Does this look familiar?
Ask:
Can I produce it without seeing it?
Then:
Can I choose it without being told to use it?
Then:
Can I still use it while thinking about something else?
Then:
Can I use it in a situation I have never practised?
Each step removes support.
What remains is closer to available ability.
Speaking is not the final chapter of learning
Many courses implicitly create this sequence:
grammar first,
vocabulary second,
speaking later.
But speaking is not merely where previously learned knowledge is displayed.
Speaking itself changes knowledge.
It reveals what is retrievable.
It exposes missing distinctions.
It forces decisions.
It creates failures that explanation alone cannot predict.
It makes the learner reorganize language under pressure.
So speaking should not be treated only as a test after learning.
It is part of learning.
The goal is not to know more English than you can use
Accumulating knowledge can feel productive.
Another grammar topic.
Another vocabulary list.
Another video.
Another application.
Another textbook.
Sometimes that is necessary.
But if the central bottleneck is production, eventually the learner needs a different question:
How much of what I already know is actually available to me?
That question changes the learning strategy.
Instead of endlessly expanding the warehouse, you begin improving access to what is already inside.
Understanding is real progress
If you understand English but struggle to speak it, your comprehension is not fake.
It is evidence that substantial parts of the language system already exist.
The task is not to start again.
The task is to change the conditions under which that knowledge becomes accessible.
From:
I recognize it.
to:
I can retrieve it.
From:
I can retrieve it when prompted.
to:
I can choose it.
From:
I can choose it slowly.
to:
I can use it while communicating.
And eventually:
I can use it in situations I did not prepare for.
That is not one jump.
It is a sequence of transformations.
And once we see those transformations, “I understand English but I can’t speak it” stops being a mysterious failure.
It becomes a diagnosable learning problem.
Language Learning at Levitin Language School
Levitin Language School works internationally with adults, children, teenagers, students, professionals, and migrants.
We do not assume that a learner who struggles to speak simply needs more vocabulary, more grammar, or more conversation.
The first task is to identify where usable language breaks down.
Our broader educational architecture connects:
Languages
Academic Subjects
Language + Subject
because the same distinction between recognition and independent production appears throughout learning.
Explore Levitin Language School:
Explore Language Learnings — our U.S. educational direction:
Explore the educational work of Tymur Levitin:
https://timurlevitin.blogspot.com
Contact Levitin Language School
WhatsApp / Viber: +380932913429
Telegram: @START_SCHOOL_TYMUR_LEVITIN
Email: notification@levitintymur.com
Tymur Levitin
Founder & Director, Levitin Language School
Languages • Academic Subjects • Language + Subject
© Tymur Levitin. All rights reserved.
