Where the Meaning Lives
The word, the self, and the machine
There are two kinds of writer, and the arrival of LLMs has made the gulf between them much more visible.
The first kind builds sign systems. For them, writing is an act of intelligence. They have always known, on some level, that the emotional charge of a sentence is a property of its arrangement. That voice is not something you have but something you construct, word by word, clause by clause, in the negative space between what is said and what is withheld. They understand rhythm as architecture. They know that the reader’s response to a story was not transferred from the writer’s body into the text. It was built. Placed. Engineered through sequence and restraint and the precise control of information. For these writers, the construction of the sign system is what matters, because that is where meaning is produced.
The second kind believes that writing is an act of expression. For them, words are a conduit between the writer’s inner life and the reader’s. What makes great writing valuable is the presence of an authentic human consciousness behind the words, a living body that has suffered and desired and remembered, and that this presence saturates the text. For these writers, the act of writing is what matters, because that is where meaning is unearthed.
Both kinds care about voice and emotion. The difference is not in what they value. It is in how they believe meaning comes to be.
Sherwood Anderson built a personal mythology around the untouched first draft. He claimed he wrote “Hands”, the opening story of Winesburg, Ohio, in a single sitting at a kitchen table in Chicago and didn’t change a single word afterwards. He described the completed draft in physical terms: solid, like a rock, put down. For Anderson, the unedited text was not just a record of what he felt. It was the feeling itself, made material. To revise it would be to destroy it.
This is the purest form of expressivism - believing that the act of writing has unearthed something sacred. But look at it closely and it is also a reverence for the sign system. Anderson was not just worshipping the expression of his feelings. He was worshipping the specific sequence of words his feelings had produced. He was saying: the arrangement matters so much that I dare not touch it. The expressivist and the constructivist are often closer than either wants to admit. They argue about where meaning originates while agreeing, in practice, that the sentence is where it lives or dies.
The postscript to Anderson’s story makes the point even sharper. The manuscripts held at the Newberry Library show revision after revision. His clean, direct prose was in fact shaped through iterative editing. Anderson was a constructivist building sign systems the whole time. He just told a better story about it than most.
Clarice Lispector told no such story. In her newspaper columns she wrote that she could only work “when ‘the thing’ spontaneously arrives,” and that she was “at the mercy of time.” Lispector described the word itself as bait: once the thing between the lines had been caught, the word could be discarded. This is the expressivist position stated plainly. The meaning exists before the words. The initial act, the plunge into consciousness, is the real event. Once the shape of what lies beneath has been caught, the words on top can be rearranged, adjusted, moved around in service to something that was already there. The craft is preservation, not discovery.
Sylvia Plath worked from the same conviction in a different register. Where Lispector waited for the thing to arrive, Plath went after it. The Confessional poets built an entire movement on the premise that the autobiographical self is the irreplaceable ingredient. But what made Plath’s method distinctive was not the confession itself. She wrote from life first, fast, getting the thing down while it was still raw. Then she edited, sometimes obsessively, shaping stanzas that are among the most formally controlled in the English language. The initial act was where the meaning was unearthed. The craft that followed secured what was found. That order mattered.
Lispector in fiction and Plath in poetry occupy the same position: the writer as shaman, plundering the depths, and the text as evidence that something real was brought back. Both trust that the essential event has already happened before the careful work begins. Both treat editing as keeping what was unearthed in place.
This is the expressivist position at its most powerful. Not naive. Not un-crafted. But rooted in a conviction that meaning precedes language, that the writer’s job is to go down into consciousness and return with something true, and that the words you use are the net, not the fish.
Joan Didion represents the opposite end. In her essay “Why I Write”, Didion said she wrote entirely to find out what she was thinking. She went further: she claimed she was not an intellectual, that she could not think in abstracts, that her attention veered back to the specific and the tangible. Had she been blessed with even limited access to her own mind, she wrote, there would have been no reason to write at all.
That could be the sign-system writer’s creed. For Didion, meaning does not precede the text. It is produced by the text. Her prose was a machine for generating clarity, not a conduit for transmitting it. You can feel it in the style: those short, controlled sentences, the obsessive precision of the detail, the way the emotional force of a piece arrives not through confession but through the accumulation of carefully selected facts. Didion did not pour herself onto the page. She built structures on the page and discovered herself inside them. The act of construction contained the discovery.
The expressivist starts with meaning and finds words for it. The constructivist starts with words and finds meaning through them.
We talk about expressivism and constructivism as if they sit at opposite ends of a line. But push either position far enough and they curve back toward each other. The spectrum is a circle.
Lispector worships the moment of contact, when consciousness yields something that must be held in place with language. Anderson, revealed by his own archives, agonises over every syllable until the construction is right. One calls the act sacred. The other calls the arrangement sacred. Both positions, at their most extreme, arrive at the same place: a religious conviction in the logos. The word itself, not as a vehicle for something behind it, but as the thing that signifies what matters.
The oldest version of this conviction is John’s: “In the beginning was the Word.” Not: in the beginning was the feeling, and then the feeling found words. The Word comes first. The expressivist reads this as revelation: the Word finds what was already there. The constructivist reads it as creation: the Word is where meaning begins. But both believe the text is not a translation of experience but the primary event. And beneath that shared conviction lies something deeper still: the fact that there is consciousness capable of meaning at all. That there is something rather than nothing that attaches significance to signs. This is the religious dimension that neither camp can fully account for and neither can escape.
The irony is that the disagreement itself is linguistic. Both camps worship the word. Both believe meaning originates beyond language. They differ only on whether the word unearthed it or produced it, and that distinction, for all its philosophical weight, collapses at the level of practice. The pure expressivist still cares about the sentence. The pure constructivist still writes from the body. The circle keeps closing.
The sign-system writer looks at an LLM and sees something extraordinary. Here is a machine that operates entirely within the domain they have spent their life mastering: the domain of signs. It has no body, no biography, no childhood to draw from, no lover to mourn. What it has is intelligence. Not consciousness, not experience, but the capacity to predict which sign should follow which, and to do so at scale. It can produce sequences that cohere, carry rhythm and generate affect. Sometimes badly. Often blandly. But the mechanism is visible, and the mechanism is familiar. It is the same mechanism the sign-system writer has always used. Pattern. Prediction. Arrangement. The ability to triangulate what a reader will bring to a sequence of marks on a page.
It confirms what the sign-system writer always suspected: that meaning is not a substance secreted by the author. It is an event that occurs at the point of contact between a sign system and an interpreter. The writer’s job was never to pour themselves into the text. It was to build the text so precisely that the reader does the rest.
The expressivist writer looks at the same machine and sees a counterfeiter. If writing is the transmission of authentic selfhood, then a machine that produces convincing prose without a self is not just a tool. It is an existential threat. It is proof that the thing you thought was sacred, the irreducible link between self and sentence, can be faked. Or worse: that it was never there at all.
This is why expressivists resist LLMs so viscerally. It is not really about jobs, or copyright, or the quality of AI-generated prose. It is about ontology. What is writing for? Where does its value live? If you have built your identity, your sense of artistic purpose, on the belief that your writing matters because you wrote it, because your specific consciousness is the irreplaceable ingredient, then LLMs do not feel like a new way to interact with signs. They feel like manufacturers of void systems, lacking in human value or meaning.
The gap between the two camps is real but mislocated. The expressivist writer accuses the sign-system writer, and by extension the LLM, of lacking consciousness. The sign-system writer points out that consciousness was always an effect, not a cause. One side says: intelligence without consciousness is empty. The other says: consciousness without intelligence is noise. Both are partly right.
What the expressivist camp misses is that they have always been building sign systems. Every time a writer chooses one word over another, every time they break a paragraph in a particular place, every time they use white space or repetition or a shift in tense to control the reader’s experience, they are doing exactly what the sign-system writer does. The difference is not in the practice. It is in the self-narration. The expressivist writer tells themselves a story about unearthing, about consciousness made legible. But their craft, the actual technical decisions they make at the desk, is semiotic engineering.
What the constructivist camp misses is that the self is not irrelevant. It is not the mechanism by which prose generates meaning. It is the thing that determines what is worth writing. What to arrange. What to attend to in the first place. Didion did not write about just anything. She wrote about the specific subjects her specific life had made urgent. The construction was the method. The self was the reason the construction occurred.
An LLM can produce a competent paragraph about grief. It cannot decide that this particular grief, this specific loss, in this precise configuration of circumstance and character, is the thing that needs to be written about right now. It cannot hold a ten-year obsession with a question and let that obsession sharpen into form. It cannot recognise, in the middle of a draft, that the whole structure is wrong because the emotional centre of the story has shifted and the architecture no longer serves it.
The self is not the source of meaning. It is the source of intent. And intent, for now, is something only a consciousness embedded in time can generate. Not because of anything mystical. Because intent requires a relationship with the future, a sense of what matters, and a set of preoccupations shaped by a life that has actually been lived.
So the question is not whether LLMs threaten writers. It is which theory of writing they threaten. If you believe the value of literature lives in the act, in the sacred plunge into consciousness that unearths the meaning before the words arrive, then yes, LLMs are a catastrophe. They are a machine that produces the appearance of presence without the substance. And no amount of reasoning will make that feel like anything other than desecration.
If you believe the value of literature lives in construction, LLMs look different: not like desecration but like a tool working in your own medium, one that confirms your theory of how meaning works while revealing, through every flat or generic sentence it produces, exactly how much the architect still matters.
An LLM has no consciousness to plunge into. It has no depths to plunder. But it arranges signs, and the signs cohere, and sometimes the reader of those signs feels or understands something deeply. What it lacks is not craft or feeling but a reason to speak.
That is what the writer brings. Not the meaning itself. The need for it. The specific, time-bound, bodily need to arrange these signs and not others, to build this architecture and not that one, because something in a life lived has made this the thing that will not leave you alone.
The expressivist and the constructivist were always closer than either believed. LLMs made both the distance and the proximity more visible. What separates both of them from the machine is not technique, or even feeling, but consciousness as intent: the specific, irreplaceable, human need to make meaning.