Somebody told me I use too many em dashes.
Reader, I do.
I use them in emails. I use them in the novel. I used four of them last week in a text message to my sister about a dog.
And instead of apologising, which felt undignified, I went away and worked out why. Which turned into a rabbit hole. Which turned into this, because it turns out the answer is not "bad habit" and it is not "lazy grammar."
It's that some of us are still using a punctuation system that the rest of the language quietly abandoned about four hundred years ago.
Nobody has ever said a semicolon out loud
Here is the thing I have only recently been able to articulate.
I don't see sentences. I hear them.
A sentence arrives as a sound — a shape, a rhythm, a specific number of beats — and my entire job, at that point, is to get it onto the page before it evaporates. I am not constructing it. I am transcribing it.
And the page is not built for sound.
Speech has no semicolons. Nobody in the history of human conversation has ever deployed one out loud. What speech has is pauses of different lengths, emphasis, changes of direction, and an enormous amount of self-interruption.
The em dash is the only mark on the page that transcribes any of that.
A semicolon is a grammatical claim. It asserts that two independent clauses are related in a balanced, structurally equivalent way, and that the writer has assessed this and stands by it.
A dash says: breath, swerve, carry on.
One of those is a legal document. The other is how people actually talk.
And this is the bit that genuinely surprised me
I assumed, going in, that the dash was the modern, informal, slightly degraded option. The one you reach for when you can't be bothered to work out what you meant.
It's the opposite. It's the original.
Punctuation did not begin as grammar. It began as breathing.
Around 200 BCE, Aristophanes of Byzantium, working at the library at Alexandria, proposed a system of small dots placed at different heights on the line to mark pauses of different lengths — explicitly to help people say the words out loud. Greek playwrights had already been dropping vertical dots between phrases for the same purpose. Orators marked up their own speeches beforehand with signs for pauses, for breath, for rhetorical effect.
Jerome, some six hundred years later, laid out scripture per cola et commata — by phrases — so that a reader would know exactly where to take a new breath.
None of these people were marking grammar. They were scoring speech. Punctuation was, at the beginning, a set of stage directions.
And then, somewhere between the fifteenth and the seventeenth centuries, as silent reading took over and printing standardised everything, the whole system changed jobs. Punctuation stopped indicating pauses for the voice and started indicating the grammatical construction of the sentence for the eye.
It went from being about the ear to being about the architecture.
The linguist Wallace Chafe wrote about this in the eighties — the idea that both writers and readers carry a "covert prosody," an auditory imagination of intonation and hesitation running underneath written language, and that punctuation is the small, partial, leaky way we make some of it visible.
So when I say I hear sentences rather than see them, I'm not describing a quirk.
I'm describing the older use of the equipment.
And the em dash is the last prosodic mark left standing in a system that otherwise went over to grammar. It is the only one that still means pause here, change direction rather than these two clauses stand in the following relation.
Of course we cling to it.
The dyslexia part
Here's where it gets interesting, and slightly annoying.
The research on dyslexia and punctuation is fairly consistent, and it does not say what you might hope. A 2023 study comparing students with dyslexia to matched controls found that they use the same inventory of punctuation marks — they know what all of them are — but they use fewer of them, and make more errors with them. Work in higher education has found much the same, with a medium effect size for punctuation and capitalisation errors even after you strip out spelling.
The researchers describe it as insecurity. Not incapacity. Insecurity.
And when I read that, I thought: well, obviously.
Because if every formal punctuation mark is a place where you might be visibly, correctably wrong — and if you have spent your entire education being visibly, correctably wrong on the page in front of people — then you are going to gravitate, hard, towards the one mark that cannot be got wrong.
You cannot misuse a dash. That's the whole point of it. It makes no grammatical claim, so there is no claim to fail. It is the only piece of punctuation in the English language that is entirely safe.
That isn't laziness. That's a completely rational response to thirty years of red biro.
The ADHD part
The other half is about where sentences go to die.
Every connector other than the dash requires you to commit before you write it. A comma, a colon, a semicolon — each one demands that you already know the relationship between the thing you've just written and the thing that's coming. You cannot deploy them innocently. You have to stop and adjudicate.
And that stop happens at precisely the moment you are holding the rest of the sentence in working memory.
Which is the problem, because working memory is the thing that ADHD reliably compromises. The research on adults is not subtle about it — worse performance on tasks requiring rapid processing and active manipulation of what you're holding, more omissions, more errors. And the written-expression literature notes that it shows up as text that's shorter, more error-prone and structurally messier, with the underlying mechanism barely investigated.
I can tell you what the mechanism feels like from inside.
The rest of the sentence is not written down anywhere. It is not in a plan. It exists in a small, badly lit, temporary room at the back of my skull, and it has a shelf life of about four seconds.
Ask me to leave that room to go and adjudicate a semicolon, and when I come back the sentence has gone. Not gone fuzzy. Gone. I can usually retrieve the idea. Almost never the phrasing — which was the good bit, and the entire reason I was writing it down.
The dash costs nothing. It requires no ruling. You don't have to know what the relationship is yet; you can work that out later, in daylight, with tea, in the edit, where such decisions belong.
The momentum survives. And because the momentum survives, so does the sentence.
I genuinely think this is a large part of how I wrote forty-three thousand words in a week. Not talent. Not discipline. Certainly not planning.
Deferred punctuation.
Brackets whisper, commas bury
There's a third thing, and it's the one I'd defend hardest.
The dash lets you interrupt yourself without demoting the interruption.
My brain does not produce clean single-track sentences. It produces a main idea with an urgent parenthetical attached, arriving simultaneously, both convinced they have priority.
Brackets whisper — anything inside them is instantly downgraded to a footnote the reader is invited to skip. Commas bury — comma-clauses sink into the sentence and lose their edges.
A dash lets the aside arrive at full volume and keep its own weight.
Which matters when, as is frequently the case in my writing, the aside is the best bit.
An honest caveat
I want to be clear about what is established and what is me.
Established: punctuation began as prosody and later became grammatical. Dyslexic writers use fewer punctuation marks and make more errors, and researchers describe this as insecurity. ADHD compromises working memory, and written expression suffers.
Not established: that dyslexic and ADHD writers use more em dashes than anybody else.
Nobody, as far as I can find, has counted. It is a hypothesis built out of three true things and one writer's experience of her own head. If somebody wants to go and count, I would read that paper immediately.
But it fits. And it explains something that "you overuse dashes" doesn't.
The problem, which I do accept exists
If everything is a dash, everything sits at the same pitch.
You get prose that is all swerve — every clause equally urgent, equally interrupted, equally breathless. Exhausting to read, and worse, nothing lands, because landing requires that some things be quieter than others.
I know, because I've written that page. Usually at one in the morning.
What I do about it now — and I only noticed I was doing it when somebody pointed it out — is use the one-line paragraph instead.
Like that.
Same swerve. Same breath. Same refusal to specify the grammatical relationship. But it gives me something the dash can't, which is hierarchy. A line on its own is loud. A line inside a paragraph is quiet. I get to choose.
The dash is a breath inside a sentence. The line break is a breath between them. Once you have both, you can build something with actual dynamics instead of one long urgent monotone.
Which, now I write it down, is a decent description of learning to write at all.
A handrail, not a habit
So that's the defence, and it isn't aesthetic.
The dash is an accommodation. It's the thing that lets a fast, leaky, easily-derailed, red-biro-scarred brain get a whole thought down intact — with the aside attached, at full volume, before the small badly lit room empties itself.
It's a handrail.
I would rather have the sentence, slightly over-punctuated, than the grammatically impeccable version of a sentence I can no longer remember.
And if you were taught that this was sloppiness: it isn't. It's the older system. We're just the ones who never stopped hearing it.
Writers — what's yours? What's the small, faintly indefensible habit that turns out to be load-bearing?
— Mallory
Sources and rabbit holes
Punctuation and the Prosody of Written Language — Wallace Chafe, Written Communication, 1988
A history of punctuation — Aeon
Punctuation: history and development — Britannica
French Students with Dyslexia Facing the Punctuation System: Insecurity, Inventory Use, and Error Studies — Brain Sciences, 2023
Beyond spelling: the writing skills of students with dyslexia in higher education
Written expression in boys with ADHD: the mediating roles of working memory and oral expression — PubMed
Metacognitive interventions in text production and working memory in students with ADHD