Skip to main content
The Vernacular After Geography
The Standard with Ears · Arc 1 · TAM_EAR_1-02

The Vernacular After Geography

Dante went hunting for a language that belonged to no place, and could not find one

In a hurry? Read the executive summary.

TAM-EAR.1-02 · The Standard with Ears · The Approximate Mind

Dante was banished from Florence in 1302 and never went back. Somewhere in the first few years after that, moving between the households of other men, he started a treatise about language. He planned four books. He finished one and part of another, and then the manuscript stops in the middle of a sentence in the fourteenth chapter of book two, and nobody has ever worked out why. It was lost for two centuries. It was written in Latin.

The treatise is an argument for the vernacular. The case that the language you learn from the woman who nurses you is nobler than the language you learn from a schoolmaster is made, at length and with considerable heat, in the language of the schoolmasters. Dante seems to have found this unremarkable. There was nowhere else to make an argument.

The distinction he draws in book one is the useful thing. There are two channels by which a person comes to have a language. The first is the one every child gets: you acquire it from the people around you, without instruction, without rules, by imitation and immersion, before you know it is happening. The second requires a teacher, a text, a discipline, and years. Dante calls the second one grammatica, and what he means is Latin, though he does not use the word. It does not change, because it was designed not to, and it was designed not to because a language nobody can alter is a language that can carry an empire’s paperwork across four centuries.

Then he goes hunting. He works through the Italian peninsula, region by region, city by city, and he counts at least fourteen distinct vernaculars, and he rejects every one of them. Bologna gets a hearing and is dismissed. Rome is treated unkindly. Even his own Tuscan does not survive the audit. What he is looking for is a vernacular that is illustrious, that hinges the others, that would be at home in a court and at a tribunal, and that belongs to no single city because it is somehow present in all of them.

He does not find it. At the opening of the sixteenth chapter he says he has hunted across the whole country without catching the animal, and he compares what he is after to the panther of the bestiaries, whose scent hangs in every part of the wood and which is never once seen.

What makes a vernacular a vernacular
#

Dante is not sorting languages by how they sound, or how many people speak them, or how much literature they have, or how far they are from Latin. He is sorting them by acquisition. A vernacular is whatever you got by being somewhere, from the people who were also there, without anyone setting out to teach you. That is the whole criterion. It is relational and it is procedural, and it says nothing at all about content.

There is a demonstration of the difference available to anyone who has met a trained phonetician. A good one can produce almost any accent on demand, accurately enough to fool the people who own it. Ask what happens when the performance stops and the answer is that they revert, immediately, without deciding to. The trained register is a costume. The acquired one is the body underneath it, and the body is what you dream in, count in, and argue in when the argument gets bad enough.

Which also explains why the second channel never displaces the first. Instruction always arrives after acquisition has already happened, onto a surface that is already formed.

Institutions can only ever overlay a vernacular, because nothing official has ever gotten there first.

Which means the vernacular was only ever geographic by accident.

It was geographic because for the whole of recorded history the people around you were the people near you. Immersion required proximity. A child in Bologna acquired Bolognese because Bologna was what was within earshot, and the fourteen vernaculars Dante counted were fourteen answers to the question of who happens to be standing close enough to be copied. Distance did the sorting. Nobody designed that arrangement either.

The vernacular was never a property of places. It was a property of whoever was in earshot, and for four thousand years those were the same thing.

They are not the same thing now, and the split happened without an announcement.

A fifteen year old in a mid-sized American city and a fifteen year old outside Manila acquire, over the same eighteen months, the same cadences, the same comic timing, the same three-beat structure for a self-deprecating joke, the same set of words for degrees of social embarrassment, the same rising terminal on a declarative sentence. They have never met. Neither has been taught any of it. Neither could name a single rule. Both would fail to explain to an adult why one phrasing is correct and the other is not, while being completely certain about which is which.

That is Dante’s first channel operating exactly as he described it. Immersion, imitation, no instruction, below the level of awareness. The only thing that has changed is what constitutes being around.

Meanwhile the fifteen year old in the American city cannot fully parse how her sixty year old neighbor two doors down constructs a sentence, and the neighbor cannot parse hers, and both of them were born within nine miles of where they are standing.

The feed is not a supplement to the kitchen and the street. It is a rival to them, running during the same few years when acquisition is possible at all, and on most days for more hours.

The feed is the nurse now. It meets Dante’s criterion cleanly, on every clause, and it has no location.

The institution that will not admit to being one
#

Every previous machine for making a standard language announced itself.

The Académie française was founded, chartered, staffed with forty people, and published a dictionary. The schoolroom had a curriculum and a cane. The King James translators were appointed and their names are known. The BBC had a pronunciation unit with a policy, and if you disliked the policy you could write to them about it and somebody would write back. Whatever else these were, they were legible. You could see who was doing the standardizing, what rules they had adopted, and roughly why. They could be argued with, resisted, lobbied, satirized, and in several cases overthrown.

The ranking that decides which cadence a hundred million teenagers see this quarter has no charter, no address for correspondence, and no theory of language at all. It has an objective, and the objective is attention. Nobody inside the building set out to shape a language, which is why nobody inside the building thinks the comparison applies to them.

It applies. What is being exerted is a standardizing pressure on the world’s adolescents at a rate no ministry of education ever managed, and its being a side effect changes nothing about its size.

An institution that does not know it is one cannot be reformed, because there is nobody inside it whose job it is to receive the complaint.

The Académie could be mocked into retreat. It was, repeatedly. There is no equivalent move available against a selection function, and not because the function is powerful. The function is not claiming anything. You cannot win an argument with a party that has not made a claim.

There is a second consequence, and it may be the larger one. What arrives announced gets met with something. A child in a Latin lesson knows a lesson is happening, and part of her is holding the material at arm’s length even while learning it, because that reflex is what being taught feels like. The feed triggers nothing. It presents as your friends, your interests, your own attention finding its own way, and none of that summons the resistance a curriculum summons.

An institution that goes unrecognized may encode more deeply than one that does not, for exactly this reason: it never meets the immune response that official instruction has always had to overcome. Dante’s Latin announced itself as the learned language and was resented accordingly. This one arrives as company.

The standard made out of us
#

Latin did not come from the villages. Sanskrit did not come from the villages. Court French, received pronunciation, literary Arabic, the King’s English: in every previous arrangement, the standard was materially different stuff from the vernacular beneath it. It had a separate ancestry, a separate training, a separate set of custodians. That difference is what gave the vernacular something to be. A vernacular is defined against something, and what it was defined against was always a language that had come from somewhere else.

The current standard is assembled out of the vernacular itself.

The ranking is not the only machine in the house, and the second one works differently. Everything the models produce, they produce from what people already said. The register that comes back is a composite of the registers that went in, weighted, smoothed, and returned. There is no separate ancestry. There are no custodians with a different training. When the machine writes a sentence, it writes a sentence made of everyone’s sentences, and it hands it back to the people it took it from.

Everyone’s is too generous, and the word doing the work is weighted. What went in was not the world’s speech in proportion. It was mostly written, mostly English, mostly by people with the leisure to post at length, and the composite that comes back carries that slice’s habits as though they were the habits of language itself. Promotion to a standard has always selected one vulgar tongue out of many. Nothing about this arrangement changed that, and the question of whose vernacular got promoted is large enough to need its own accounting.

So the pole is gone. There is nothing above the vernacular that is not made of vernacular, and therefore nothing to lean away from. A teenager in 1955 could tell you exactly what she was not talking like, and could hear the difference in a single vowel. A teenager now is talking against a composite of everything teenagers have said, including her own contribution to it, returned to her at high volume within the year.

Every standard before this one was made of different material from the speech beneath it, and that difference was the thing the speech beneath it had to push against.

Sanskrit could never have become Prakrit. That was the guarantee. That is what has stopped being true.

What the feed cannot carry
#

A vernacular is not a word list. Words are the top layer of a stack, and they are the layer that has always traveled fastest, which is why the word list is what people mean when they say a dialect is dying. Below the words there is syntax, which travels more slowly. Below the syntax there is prosody, the melody and stress of a sentence, which is harder to pick up from a screen than anyone assumes. And below the prosody there is everything the body does.

How close you stand. What you do with your hands while you wait. The pause length that reads as respectful in one place and as hostile in another. When to look away. How you hold a knife against an onion, and the particular economy of movement that says a person learned it from someone rather than from a video. The rhythm of pouring. What your face does before you disagree. Whether you touch the person you are reassuring.

None of that is written down anywhere, in any form, and none of it is transmitted by the feed, because the feed carries only what a camera and a microphone can resolve and what a model can compress. The embodied layer is acquired the way it has always been acquired, which is by standing near somebody for years and copying them without noticing.

So the stack comes apart. The lexicon delaminates and goes global, at the speed of the network. The syntax follows more slowly. The body stays exactly where it was, in the kitchen, in the doorway, at the funeral, transmitted at the old speed by the old method, and it stays local for a reason that has nothing to do with anyone protecting it: it cannot be encoded, so there is nothing to harvest.

Which produces a strange creature and a real consolation. A person can now sound like the whole world and move like a single street.

The vernacular after geography is a vernacular of the hands.

We wonder whether the body stays local because it truly cannot be encoded, or only because nobody has yet had a sufficient reason to try, and whether we would be able to tell the difference from inside the years before someone did.

For now the older thing holds. Two people can share every word they use and still be unable to sit in a room together comfortably, and the gap between them will be made of timing, and neither will be able to say what it is.

Dante would have understood the animal better than we do. He spent a book hunting a vernacular that had no city, that was somehow in all of them and resident in none, and he came up empty because in 1305 there was no institution capable of manufacturing such a thing. The scent was everywhere. There was nothing in the wood.

There is now, and it is not a panther. It is a selection function with an objective that has nothing to do with language, and it did not have to be hunted, and it does not have to be caught, because it comes to the house.


This essay is the second of the mechanism arc in The Standard with Ears, and reads as an instance of the accumulation-period argument rather than a fresh claim: the latency that produced fourteen Italian vernaculars was an artifact of who stood within earshot, and the earshot has changed. Its closing consolation, that the embodied layer stays local because it resists encoding, is the series’ own claim at this point in the argument, and The Body in the Data withdraws it. The question of whose vernacular was promoted, raised here and left open, belongs to The Harvested Slice. It stands in conversation with The Monoculture and The Curation Economy in the main series.

How this essay connects to others across The Approximate Mind.

Convergence of method is treated there as a systemic exposure. Here the convergence sits upstream of method, inside acquisition itself, where two teenagers who have never met arrive at the same cadence during the few years when a first register can still be formed at all.
The composite that comes back was assembled from a slice that was mostly written, mostly English, and mostly produced by people with the leisure to post at length. Whose vernacular got promoted is the question of the default, asked one layer up from where this essay leaves it.
Every standard before this one was made of different material from the speech beneath it, and that difference was what the speech had to push against. The question of whose words a machine returns has a specific answer here: they are ours, weighted, which removes the pole.
Be Youreframes
The loss of a pole to lean away from stops looking like an absence. An averaged register exerts steady pressure toward a center that belongs to nobody, so a speaker talking against a composite of everything speakers have said is not unopposed. She is being pulled.
The Amber Question takes up the registers that entered despite this essay's channels, and asks what persistence in weights amounts to: a composite with no speaker inside it, preservation or taxidermy left unsettled.
The accounting this essay named in one paragraph and handed forward arrives: The Harvested Slice asks what selected the slice, and answers with Bembo, the Prakrit hierarchy, and the composition of the crawl.
Dante and the Theory of the Vernacular
  1. Alighieri, Dante. De vulgari eloquentia. Edited and translated by Steven Botterill, Cambridge University Press, 1996.
  2. Ascoli, Albert Russell. Dante and the Making of a Modern Author. Cambridge University Press, 2008.
  3. Shapiro, Marianne. De vulgari eloquentia: Dante’s Book of Exile. University of Nebraska Press, 1990.
Standardization and the Institutions That Do It
  1. Anderson, Benedict. Imagined Communities: Reflections on the Origin and Spread of Nationalism. Revised ed., Verso, 2006.
  2. Lodge, R. Anthony. French: From Dialect to Standard. Routledge, 1993.
  3. Milroy, James, and Lesley Milroy. Authority in Language: Investigating Standard English. 4th ed., Routledge, 2012.
Acquisition, Embodiment, and What Transmission Requires
  1. Eckert, Penelope. Linguistic Variation as Social Practice: The Linguistic Construction of Identity in Belten High. Blackwell, 2000.
  2. Labov, William. Principles of Linguistic Change, Volume 2: Social Factors. Blackwell, 2001.
  3. Mauss, Marcel. “Techniques of the Body.” Economy and Society, vol. 2, no. 1, 1973, pp. 70-88.
  4. Polanyi, Michael. The Tacit Dimension. University of Chicago Press, 1966.