The new language of the feed
The historic shift from an oral to a written tradition ushered in a new way of thinking asbout the world: does ‘algospeak’ mean we are now at the cusp of another revolution?
To the online uninitiated, people can seem to be speaking in code. On TikTok, “unalive” has replaced “death.” Sex becomes “seggs,” porn is “corn,” whilst a vast array of emojis stand in for words that are at risk of being flagged by algorithms as inappropriate. What at first sight seems to be playful slang can, as we shall see, be a form of survival, as users shape language to hide from algorithmic detection.
At the same time, when people use AI tools, there is another quirk — the sudden proliferation of em dashes (see below):
—
What was once a somewhat bookish punctuation mark, it is now the ‘fingerprint’ of text created using Generative AI (specifically Chat GPT). This anachronistic punctuation appearing in the midst of the latest tech feels like a ghost from another age.
These points are not unconnected, as, taken together, they hint at how our everyday language is being reshaped by machines into what writer Adam Aleksic calls ‘algospeak’. These shifts are not merely stylistic but strike at something more significant: they change how we think.
The medium has always been the message
Of course, our communication has always been determined by the medium through which it is carried. But to understand how this is playing out today, we need to go back over 5,000 years to an age when we lived in oral cultures, where knowledge had to be transmitted through rhythm and formula in order to be embedded in our memories. For example, to modern ears, Homer’s famous “rosy-fingered dawn” phrase sounds like a decorative flourish. But Walter Ong, in his famous book, ‘Orality and literacy,’ suggested it was in fact a stock phrase that made long stories easier to recall. It’s an example of how orally based thought worked differently from ours. It built ideas by adding one thing to another, always tied to the situation at hand, and circling back with repetition, thereby keeping stories alive in our memories and, therefore, our ability to repeat them to each other.
A storyteller describing a hero might say they were “strong and brave and wise,” piling traits on top of one another rather than distilling them into one abstract category (such as “virtuous”). In epic tales, battles are narrated blow by blow, with stock phrases for weapons and wounds repeated across episodes. The point was not about precision but keeping the rhythm moving and, with that, the tale memorable. So, phrases like “rosy-fingered dawn” offered scaffolding, stabilising the speech in a world where memory was fragile.
Surely we can make a case for algospeak operating in a similar manner. For example, “Unalive” or 🍑 are today’s stock phrases, but instead of being the result of the constraints of memory, they are, in fact, a function of the constraints of algorithmic detection. Both are formulations created under pressure, repeated until they stabilise, and shared because they help communication survive. They show us that when language is under threat (whether by memory or algorithm), solutions emerge to hold it together.
But to understand the significance of this, we need Ong’s analysis of the enormous shift that occurred with the invention of writing, which spelt the end of oral cultures for much of the world’s population. Writing involved “lock[ing] words into space,” as Ong put it, making them permanent and external, encouraging a mode of thinking that revolved more around abstraction, categorisation, and criticism. This is fundamentally different to the more fluid, performative, and memory-sensitive character of the oral tradition. Importantly, this meant that thoughts could then be more easily debated and rearranged. Print accelerated this change, creating cultures of commentary as words themselves had become much more stable objects that could be compared, contested, and preserved.
Ong also suggested that a new broadcast age, which he was living through during the Twentieth Century, layered on a “secondary orality”: speech that was immediate and communal but shaped by microphones, mass audiences, and the knowledge of being overheard. This didn’t just change the format; it became more self-conscious. Politicians and performers began to craft their words for the microphone, adopting slower pacing, lower pitch, and catchphrases designed to resonate across millions.
Families learned to experience themselves as part of vast, simultaneous audiences, gathered around the radio or television set. This, argues Ong, meant that everyday speech grew more cautious, carrying an implicit awareness that one might be speaking to an unseen wider audience.
The next wave of change
The question we now surely need to ask is whether Ong’s analysis needs updating, as another shift now appears to be underway.
As we have explored, social media today seems to reintroduce many of the traits Ong associated with the oral cultures from centuries ago: posts and memes are endlessly repeated and remixed so that they lodge in memory through sheer frequency. They rely on ‘stock phrases’ such as catchphrases, hashtags, and soundbites that can be easily reused across different situations, carrying a shared tone or meaning wherever they appear. And they spread through communal uptake, gaining force when echoed, liked, or stitched by others.
The parallel with earlier oral cultures is uncanny, but where oral traditions relied on formulaic techniques to embed stories and meanings into memory, today’s digital cultures rely on formulaic techniques to sidestep algorithms.
And it is this, says Aleksic, that influences how we think: to call death “unalive” means we subtly reshape the concept itself. Of course, we know that the words we use shape how we see the world; the Sapir–Whorf hypothesis has long held that language doesn’t merely reflect thought, but shapes it. When linguistic structures shift, so too do the perceptual categories through which experience is filtered.
“Death” is blunt and final, while “unalive” softens it, turning into the absence of life rather than its end. The grammar also displaces agency. Someone can be killed, but to be “unalived” seems to happen without a subject, blurring responsibility and cause. This can serve a helpful purpose for some communities, such as reducing stigma around suicide or violence. However, this shift also risks vagueness at a time when clarity is crucial.
On this basis, Aleksic argues that the word “unalive” only exists because a machine’s presence has entered the conversation: when people anticipate moderation (related to the use of the word “death”), they invent detours. This is what Ong meant when he said technologies “technologise the word”: the phrase has become newly literal as language itself is re-engineered to side-step the algorithm.
The em dash ghost
And in the midst of this, almost as a side note, perhaps the em dash points to something else — the reminder of print. ChatGPT’s punctuation habits can feel odd, not because they are incorrect but because they belong to another communicative order, one built on more around permanence and polish. This feels like an example of philosopher Mark Fisher’s notion of hauntology. He might have suggested that the em dash only feels odd because it is like the ghost of print culture haunting digital speech. As Generative AI is trained on the archives of print, it drags stylistic residues of book culture into a world at odds with this, dominated by fragments, emojis, and immediacy.
Conclusions
If oral cultures stabilised knowledge through formulae because memory was fragile, and print cultures stabilised knowledge through writing because permanence was possible, we might characterise our present condition as managing knowledge in a provisional way. Words are used in a tactical rather than eternal manner, given they survive until the algorithm notices, or until another code takes their place. And as words shape how we think, then the world surely looks less like a set of fixed categories than a shifting landscape in which survival depends on agility.
In the early 90s, before the widespread use of social media, the sociologist Anthony Giddens described the time as being marked by reflexive doubt. Even in an era of expanding science and expertise, knowledge could never be fully settled: it was always provisional, subject to revision, constantly reinterpreted in light of new evidence and shifting contexts.
Social media has arguably accelerated this condition. In feeds where words can vanish overnight, where meanings shift with each update to the algorithm, reflexive doubt is no longer an abstract sociological diagnosis but an everyday experience. Em dashes attract our attention as they are remnants of a more stable, written culture that, whilst not dead (not least by virtue of you reading this!), does seem to be struggling to have the same hold on our collective thinking that it once did. On this basis, “Unalive” is not just a euphemism but a hedge against algorithmic erasure, a stopgap word until the rules change.
To speak today is to anticipate misrecognition and adapt in advance, to live with language, and therefore an understanding of the world, that are both permanently in beta.