Somebody asks what your project is about and you explain it in four minutes. Clearly. With an example. They understand it, they ask a follow up question, you answer that too.
An hour later you open a blank document to write the same thing down and produce two sentences, both of which you delete.
This is common enough that most people assume it says something about them. Usually something unkind: that they are lazy, or disorganised, or that they never really understood the material and the fluent explanation was a bluff.
None of those follow. The explanation is more boring and more useful.
Before going further: this article is about a practical workaround, not about diagnosis or treatment. If difficulty with writing is significantly affecting your work or study, that is worth raising with a qualified professional, and nothing here substitutes for that.
Speaking and writing are not the same task wearing different clothes
They feel related because both produce language. Underneath they make very different demands.
Speaking to somebody is a shared task. You have a listener supplying structure in real time through their face, their questions and their confusion. You do not plan the whole explanation before starting. You produce a bit, read the response, adjust, continue. The other person is doing half the organisational work and neither of you notices.
Writing gives you none of that. There is no listener, no feedback, no signal that you are being understood, and no external structure at all. There is a blank page, which supplies nothing.
The classic account of this in writing research comes from Linda Flower and John Hayes, whose 1981 paper in College Composition and Communication described writing as a set of processes running at once rather than a neat sequence: planning what to say, translating it into sentences, and reviewing what you have produced, with all three interrupting each other constantly.
That is the part that matters here. When you write, you are generating and evaluating in the same moment. You produce a clause and simultaneously judge it, check the spelling, wonder whether the reader will follow, reconsider the whole structure, and decide the opening is wrong.
When you speak, almost none of that judging happens, because there is no time and because spoken language is allowed to be imperfect.
Why the load lands so unevenly
Everyone experiences some version of this. It is much heavier for some people, and it is worth naming who without turning it into a diagnosis.
People who find sustained attention and task initiation effortful, which includes but is not limited to people with ADHD, are carrying more of that simultaneous load with less capacity to spare.
People with dyslexia or related differences are spending working memory on transcription itself, on spelling and letter forms, which leaves less available for the actual thinking.
People writing in a second or third language are running an extra translation process on top of everything else, and often monitoring for errors far more anxiously than a first language writer would.
And people who are anxious about a specific reader are evaluating every sentence against an imagined critic before it is even finished, which is the fastest way to stop a first draft that exists.
What these have in common is not a lack of ideas. It is that the writing task is asking for several things at once and one of the channels is congested.
The workaround is one idea: separate generating from judging
Almost every practical technique that helps with this is a version of the same move. Do the producing first, without evaluating, and evaluate afterwards as a separate task.
Freewriting is that. Talking to a colleague before you write is that. Outlining is a partial version of it.
Speaking is the most complete version, because it is the mode in which you already generate fluently and in which self editing is almost impossible. You cannot delete a spoken sentence. That constraint, which is a nuisance in conversation, is exactly what makes it useful here.
So: explain it out loud, record it, and turn the recording into text. Then edit the text, which is a different task and usually a much less blocked one.
How to actually do it
Set a phone recording and talk for five minutes.
Talk to somebody, not into a void. This part sounds silly and it is the part that determines whether it works. Imagine a specific person who does not know the subject. A friend outside your field, a new colleague, a version of yourself from two years ago. Speaking to a real audience in your head restores the structure that the blank page removed.
Do not plan first. Planning is the thing you are trying to skip past.
Use a prompt to start rather than starting cold. “The thing I am trying to explain is.” “The problem with the current approach is.” “If somebody only remembered one thing from this it would be.” Any of these will get you moving, and the first thirty seconds being rubbish does not matter because you are going to cut them.
If you dry up, say so out loud and keep going. “I do not know how to say the next bit, but roughly it is that…” That sentence, transcribed, is often the clearest statement in the whole recording.
Then convert recordings into text. Upload the file, or record straight in, and you get punctuated text back along with a summary. The steps take a couple of minutes. Around 50 languages are supported, which matters if you think most fluently in a language other than the one you eventually have to write in. Accuracy sits around 95% on clean audio, so a quiet room is worth more than an expensive microphone.
Export it as TXT or DOCX and open it next to your empty document. Vomo also gives you a summary alongside the full text, which is worth reading first on the days when even opening the transcript feels like too much.
The transcript is not the draft, and pretending otherwise will put you off the whole method
This has to be said clearly, because the disappointment is predictable.
Spoken language transcribed looks bad on the page. It repeats itself, doubles back, has half finished clauses, and contains the same point made three times in slightly different ways. Reading your own transcript for the first time is a mildly humbling experience for everyone.
That is fine, because what you have is not a draft. It is raw material, and specifically it is raw material for a task that is not blocked.
Editing is a different cognitive job from generating. You are no longer producing from nothing, you are reacting to something that exists, and reacting is much cheaper. Most people who cannot start a document can comfortably cut one.
Three passes and you are usually there.
Delete about half. Anything repeated, any false start, any sentence that exists only to get to the next one.
Find the sentence where you said it best. There is nearly always one place in the recording where the explanation landed cleanly, usually not at the beginning. That sentence is frequently your opening paragraph.
Then reorder. Speech arrives in the order you thought of things. Writing needs the order the reader needs. This is the only pass that requires real work and it is a structural task rather than a generative one.
What this is not
Three boundaries, and they matter more here than the technique does.
It is not a treatment, and it is not evidence of anything about you. It is a work method, in the same category as writing an outline or using a timer.
It is not a replacement for formal support. If you are a student or an employee and writing difficulty is a persistent barrier, the routes that actually change your situation are assessment and formal accommodations, through your institution or employer. Those exist, they are a separate track from anything in this article, and a workaround that helps you get through Tuesday does not remove the case for them.
And it is not a fix for not knowing what you think. If you cannot write the section because the underlying argument is not resolved, you will not be able to say it either, and the recording will make that obvious within ninety seconds. That is genuinely useful information, delivered faster than a morning of staring at a document would have delivered it.
Where it will not help
Some people find speaking without a real listener as difficult as writing without a reader. If talking to an empty room stalls you the same way, try talking to an actual person and recording that conversation instead, with their agreement.
Highly formal registers do not come out of speech easily. Legal drafting, academic conventions, anything with a rigid house style. You can generate the content this way and you will spend longer on the form pass.
And it does not remove the need to write. It moves the difficult part from the start to the middle, which for a lot of people is the difference between a document that exists and one that does not.
The privacy bit, briefly
You will end up recording yourself thinking out loud, which tends to be more candid than anything you would type. Half formed opinions about a project, frustrations, things about colleagues.
Worth checking that recordings are encrypted in transit and at rest, that they are not shared with third parties, and that you can delete an individual recording rather than only the whole account. Vomo does those. And it is a reasonable habit to delete the audio once the text exists and the piece is written, because the transcript is the thing you needed and the recording of you muttering is not.
Try it on the thing you have been avoiding
Not on something new. On the document that has been open in a tab for nine days.
Five minutes of talking, two minutes of converting, twenty minutes of cutting. The output will not be finished and it will be a great deal further along than the two sentences you keep deleting.
On cost, since this is often something people trial on their own rather than through work: the free tier covers 30 minutes of transcription a week, which is roughly six of these sessions and plenty to find out whether it suits you. Unlimited is $1.92 a week if it turns out to be how you write from now on.
You could always explain it. The explanation was never the problem.