Hand a manuscript back to a writer with a single note in the margin — “you’ve been reading a good deal of Cormac McCarthy lately, haven’t you?” — and watch the reaction.
Usually, it is a startled yes. The tell is all through the draft: the dropped commas, the biblical run-ons, the nouns made to do the work verbs should be doing. Writers take on the cadence of whatever they are reading the way a room takes on the temperature of an open window.
After enough years at the editing desk, you stop being surprised by it. What a person has been reading leaks into their sentences before they have decided to imitate anyone, and usually before they are aware they are doing it at all.
That leakage is the unglamorous mechanism behind a piece of advice every teacher repeats and almost none of us explains well. I spent a working life in Australian classrooms before this company, and I must have told a great many students to read more. The instruction is true but lazy.
Reading more of the wrong things, in the wrong way, does very little. What separates writers whose prose keeps improving from those who stall is not volume. It is a particular kind of attention — and that attention can be learned, watched, and described precisely.
The reading that feeds writing is slow on purpose
The cognitive scientist Maryanne Wolf, Professor-in-Residence at UCLA and director of its Center for Dyslexia, Diverse Learners, and Social Justice, has spent her career mapping what she calls the brain’s “deep reading circuit” — the processes that engage when we read attentively rather than skim. In Reader, Come Home (2018), she argues that this circuit is not innate the way speech is. It is built, slowly, through practice, and it can erode. The processes she lists — inference, analogy, perspective-taking, critical analysis — are precisely the operations a writer performs in reverse when constructing a sentence that has to carry weight.
Wolf’s worry, set out in a widely read 2018 essay for The Guardian, is that screen reading trains a different reflex: the eye darting for keywords, taking the gist, leaving before the hard part. She is careful not to make a tidy panic of it — she grants that the digital brain gains real things, like quicker task-switching — but the trade is real. Skimming and deep reading are different skills, and the one that builds writers is the slow one. It is why a writer who takes in a great deal of text online and almost no books often produces prose that is dense with information and flat in tone. They have been practising extraction, not absorption.
Serious readers read like saboteurs
There is a difference between reading a novel and reading a novel the way a writer does. The novelist Francine Prose calls it close reading, and her book Reading Like a Writer (2006) is the most useful account of it I know. Prose describes slowing to the level of the single word — asking why this verb and not its synonym, why the sentence breaks where it does, why a paragraph ends on the image it ends on. She is teaching readers to reverse-engineer choices, to treat finished prose as a record of decisions rather than a smooth surface.
That is the habit I see most reliably in the writers whose work needs little from me. They read with a low-grade suspicion. When a paragraph lands, they stop and ask what did that. When a chapter drags, they do not merely feel bored — they find the sentence where their attention slipped and work out why. The point is not to imitate. It is that the reverse-engineering builds a private store of solved problems. A reader who has noticed how a good author manages a jump in time, or buries exposition inside an argument between two characters, holds a tool the reader who merely enjoyed the book does not.
The vocabulary argument is real, but smaller than people think
The most concrete and best-evidenced benefit of heavy reading is the least interesting one. In their 1998 paper “What Reading Does for the Mind,” Anne Cunningham and Keith Stanovich showed that print exposure predicts vocabulary, general knowledge, and verbal skill even after you control for raw intelligence and education. Their striking finding was that the language of books is simply richer than the language we speak — children’s books carry more rare words than the conversation of college graduates. Readers do not gain a larger vocabulary because they are cleverer; they gain one because books keep handing them words that speech never would.
For writers this matters, but not in the way the self-help version implies. A larger vocabulary does not by itself make better prose — the most overwritten manuscripts I see come from people showing off words they have only just acquired. What heavy reading actually supplies is range: the ability to feel that begin, start, and commence are not interchangeable, that each carries its own register and weight. That discrimination, not the word count, is what lets a writer choose the right one without stopping to think. It is built by exposure and very hard to fake.
Fiction does something stranger
Reading also trains the part of writing that has nothing to do with words — the modelling of other minds. In a 2012 essay for the New York Times, Annie Murphy Paul gathered the neuroscience suggesting that reading narrative activates brain regions as though the reader were living the events: descriptions of texture light up sensory cortex, descriptions of movement light up motor areas. The claim that matters for writers is the social one. Following characters through a plot is, in effect, sustained practice at inferring what other people want, fear, and conceal — what psychologists call theory of mind.
The research here is contested and should not be oversold; some of the headline studies on fiction and empathy have struggled to replicate, and I would not lean my weight on them. But the underlying logic survives the caveats. Writing that works almost always rests on an accurate model of the reader’s mind — anticipating where they will be confused, what they will assume, when they will get ahead of you. I see the failure of this constantly: technically clean prose written by someone who has plainly never pictured the person reading it, explaining what is obvious and skipping what is not. Heavy readers of narrative tend to have that model running by default. They have spent thousands of hours rehearsing it.
What attention to reading actually changes
Set these together and the mechanism is no mystery. Deep reading builds the circuit that lets a writer hold a long argument in mind. Close reading stocks the store of solved problems. Print exposure supplies the range to choose words by feel. Narrative reading keeps the model of other minds warm. None of it comes from reading fast, and none of it comes from reading only the kind of text that rewards skimming.
It is also why the advice to “read more” so often fails the people who take it most to heart. They read more, but they read the way the screen has trained them to — fast, extractive, gist first — and then wonder why their writing has not moved. The writers who improve are doing something quieter and harder to see: they are reading slowly enough to be changed by it.
Which returns me to the margin note. The McCarthy tell is usually offered as a small embarrassment, evidence that the writer was being derivative. It is the opposite. It is proof the deep circuit is working — that the prose was getting in, sentence by sentence, doing exactly what reading is meant to do to a writer. The remedy is never to read less. It is to read widely enough that no single voice runs the table, and closely enough to know whose voice you are borrowing, and why.
