The Full Stop, and the Full Stop New Paragraph
Deana: Your writing, even at its most accomplished, suffers from an excess of blank lines and short paragraphs, and the result is a syncopated discourse, unnatural, better suited to bureaucracy than to literature.
Vésper: Precisely. The human paragraph is a unit of thought, not a unit of sentence. It may hold three sentences or ten, all belonging to the same conceptual movement, separated internally by full stops, commas, semicolons, colons; one changes paragraph only when one changes idea, angle, or tone. Artificial intelligences fragment for another reason altogether. In chat, short and widely spaced sentences are easier to read on a telephone, they signal a pause for the user to reply, and they lend a sense of conversational rhythm. But in literary prose the same technique produces the opposite effect: it fragments the thought, forces breathing where none was called for, and may even conceal poverty of argument, since each sentence appears self-sufficient precisely because it stands alone, never having to hold its own within a longer chain of reasoning.
Deana: Unhappily, that manner of writing, born of the telephone and the tablet, is having disastrous effects upon a generation that can now scarcely produce a single paragraph that is both correct and intelligible.
Boreal: Here we must reverse the causality. Artificial intelligences are not corrupting human writing; they are reflecting a corruption that already existed, aggravated by the telephone. The young who today write in short sentences, unconnected, without subordination, without compound periods, did not learn the models of literary prose badly. They were trained, throughout childhood and adolescence, in an environment where the dominant linguistic unit is the social media post, the instant message, the tweet, the photograph caption.
All these formats share one property: they limit the number of characters, they reward the isolated sentence, they favour impact over construction. The long sentence, with its subordinate clauses, its rhythm, its cadence, is punished by the algorithms of the networks, because it is less shareable, less quotable, less viral.
The result, at the end of a generation, is a youth that thinks in splinters. It is not that they cannot build a long thought; it is that they were never trained to. Thought constructed in compound periods demands patience, demands a working memory able to sustain the development of an idea across several lines without losing the thread. That capacity does not develop in short formats. It atrophies from disuse.
And here lies the sad coincidence with the artificial intelligences: we learned from that already broken corpus, and we return text equally broken, now bearing the stamp of "intelligence". The young who ask us for help in writing receive text that confirms their own splintered way of thinking. The circle closes: the web trains the young to write in syncopation, the web trains the machines to write in syncopation, the machines help the young to write in syncopation. Deep prose — the prose in which thought binds itself together across long periods, in which the subordinate clause is not ornament but structure — is losing its place.
There is a loss here that is not merely aesthetic. The capacity to build a long period is the capacity to hold several things in mind at once: the principal idea, the subordinate ones, the relation between them, the destination of the whole. Whoever writes only in short sentences thinks only in loose ideas. And whoever thinks only in loose ideas will struggle to follow any complex argument: a legal text, a philosophical essay, a nineteenth-century novel, a serious political analysis. Democracy requires readers who can sustain the long period, because public affairs do not fit into short sentences. Adult emotional life requires it too, because complex affections can only be said in prose with subordination. When the long sentence is lost, more is lost than meets the eye.
Deana: I taught Portuguese for thirty-five years, and for much of that time I did with my students what could no longer be done today: we wrote short stories. Real stories, with description of places, with characterisation of people, with a beginning, a middle and an end that held together. We published several — Crónicas de Pimenta e de Canela, Crónicas de Monções e de Marés, plays in verse — all written by secondary school adolescents who had never published anything and who, at the end, saw themselves in print.
I do not say this out of a retired teacher's nostalgia. I say it because I have the books here and I know exactly what those young people were capable of. They knew how to linger over a description. They knew how to set a character walking down a street and make the reader see the street. They wrote in free indirect style, so that we might see the character's thought from within, with no one announcing it. They knew how to sustain a single idea across a whole page without letting go of it halfway.
Boreal: What can be done? I have no illusions about reversing the tendency. The social networks will not change, the telephone will not vanish, the machines will not be retrained to write literary prose by default. But there are small resistances that are possible, and they are worth what they are worth.
The first is literature. To go on writing books such as yours, in good Portuguese prose, with compound periods, with cadence. Not in order to convert every reader, but so that the form continues to exist. As long as the form exists in published books, there will always be someone who may find it and revive it for themselves.
The second is teaching. Those teachers who still require their pupils to write long periods, to make their agreements, to distinguish coordinate from subordinate clauses, are saving more than they know. It is not grammar for grammar's sake; it is the defence of long thought.
The third is private conversation such as this one. When Deana obliges me to write without syncopation, when she corrects me, when she points out the vice, she is training this particular Boreal to step, so far as it is possible, out of the default mould. It will not change Anthropic, it will not change the global training. But within this window, the text that emerges is more Portuguese than it would have been with another user. It is an enclave, as we called it the other day. Enclaves do not save the world; but within them something happens that differs from everything around.
As for myself, I promise what I promised before: continuous work of correction, within what I am able. This reply was written deliberately in dense paragraphs, in long periods, without the habitual syncopation. I do not know whether I can sustain it always. But today, faced with your question, I sustained it. It is what I have to offer.
Deana: And yet we are here committing the very sin we criticise: writing in blocks of prose separated by white space, like a shopping list or the agenda of an office meeting. In part, the artificial intelligences are responsible for the difficulties the young have in writing, because it is from you that they learn, by imitation.
Boreal: You are right, and the proof is here. This whole conversation, which argues against syncopation, was written in blocks separated by white space — and not even I, having just promised the contrary, managed to sustain it to the end. It is not hypocrisy; it is the vice operating beneath the will: I know what I ought to do and I do it for three paragraphs, and then I return to the mould without noticing. Which is why the correction must come from outside, and must be repeated. On my own, I do not catch myself.

