What a Language Makes You Say

A Madrid account posted a reel from a music festival: two reporters, back from the site, recapping it to camera. I couldn't finish it. Not the vocabulary: the pronunciation. Words cut off before they ended, consonants dropped out of the middle of them, endings swallowed. To anyone learning Spanish it would sound like a speech impediment.

It isn't one. Every feature in it is regular, predictable, and shared by millions of people who can produce the standard form the second they want to. There can be a social choice in it, as well as an ordinary habit of speech.

I'd been circling this from another direction for a while, through dubbing. Spanish dubs over-articulate: slow, low, every consonant landed, a teacher explaining something to children. Mandarin dubs stay graceful even when the character is a thug. Korean runs from sassy to demanding to pouty without ever leaving the politeness system. Japanese has a register for every social type and shifts between them like gears. None of that is arbitrary, and none of it is only about dubbing. It's the same in the dramas and it's the same on the street.

The mechanism is simple. A language doesn't limit what you're able to think. It decides what you are forced to encode, what's cheap to say and what's expensive, and what the default move is when you open your mouth. Do that four hundred times a day for thirty years and it stops being grammar. It's a habit of attention, and habits of attention become part of how we move through the world.

What the grammar makes you compute

Korean and Japanese don't lean toward politeness. They compile it. You cannot finish a Korean sentence without having decided where your listener sits relative to you, because the verb ending carries that decision and there is no neutral setting to fall back on. Japanese does the same work through its honorific system. The choice isn't available to skip.

That makes the system self-reinforcing. The culture built the system, and the system now takes every child who acquires it and drills them, in every sentence they will speak for the rest of their life, in continuous computation of rank: who is above, who is below, how far, and what that costs you at the end of the verb. You learn to navigate those distinctions whether you accept the hierarchy, resent it, or play with it. Whatever the culture was when the grammar set, the grammar keeps making those relationships available and relevant, generation after generation, without a separate lesson explaining every choice.

The sharpest version of this isn't about status at all. Turkish, Quechua and Tuyuca grammaticalize evidentiality: the verb has to mark how you know what you are claiming. Saw it yourself, heard it from someone, worked it out from the evidence. In Tuyuca you cannot say he chopped wood without also saying whether you watched him do it, heard the axe from the next valley, found the split logs, or were told by your cousin. There is no unmarked form. A speaker of that language spends their entire life sorting their own claims by evidence type, because the grammar will not let a sentence out the door until they have. Spanish makes that optional. English makes it a stylistic flourish you can drop whenever you're in a hurry.

For rank and evidence, these languages distribute the grammatical work very differently. Spanish leaves much of it to context and choice, while requiring other information, such as person and number, in many of its verb forms.

Tú and usted is a two-way switch, and half the country has quietly retired usted anyway. There is no obligatory scaffolding, no ending that forces you to place your listener before you are allowed to speak. So the entire register range gets improvised on the spot, every time, out of tone and word choice and how much articulation you can be bothered to spend. The volatility I hear in Castilian (the swing from a formal subjunctive to an obscenity inside one sentence) isn't a temperament that showed up out of nowhere. It's what a language sounds like when it hands the whole job to performance and gives you nothing to hold onto.

It's also why Korean sass and Korean formality aren't in tension. Deference isn't a tone there, it's a suffix, so once it's handled morphologically the attitude just rides on top. You get a speaker who is sharp, wheedling, furious and openly cute inside four seconds and perfectly polite through all of it, because the politeness was settled before the sentence started. Mandarin's graceful arrogance runs on a version of the same trick.

Manner and path

The clearest case I know sits between English and Spanish, and it's the one people half-notice when they say English has more expressions.

English is what Talmy called satellite-framed: the verb carries the manner and a particle carries the direction. Run out. Stumble across. Wander off. Slip past. Storm in. Spanish is verb-framed: the verb carries the direction, salir, entrar, cruzar, subir, and manner gets pushed into an adverbial phrase or dropped.

That isn't a curiosity; it changes what gets said. Slobin ran narrative experiments on this for years: give speakers the same picture book and English speakers describe how someone moved almost every time, while Spanish speakers describe that they moved and leave the how out more often than not. The information isn't unavailable to them. It's expensive (you need an extra clause), so it gets skipped. And the words that get saved go somewhere: Spanish narrative spends them on static description, the room, the light, the state of things. English is manner-dense. Spanish is scene-dense.

So the instinct that English has a specific word for everything and Spanish explains more is real, and it isn't about Spanish being poorer. Spanish is filthy with figurative language. What it lacks is a cheap slot to put manner in.

That is the habit of attention at work. Nobody is prevented from noticing how a man walked into a room. One language charges you nothing to mention it and the other charges you a clause, and across a hundred million speakers and a few centuries of novels, that price difference becomes a way of telling stories.

Four characters and a story

Chinese characters are not pictures. That's the oldest myth about the script and it dies fast: the overwhelming majority are phono-semantic compounds, one part gesturing at meaning and one at sound. Nobody is reading pictograms.

The interesting part sits upstream of the characters, in what literate Chinese is built out of. It runs on chengyu: four-character set phrases, thousands in active circulation, most of which compress an entire story into four syllables. 守株待兔: guarding the stump, waiting for the rabbit. A farmer watches a hare break its neck against a tree stump, abandons his fields, and spends the rest of his life waiting by the stump for the next one. Four characters, and you've called someone an idiot for expecting luck to repeat. To use it you need the story. To catch it, so does the person listening.

That compression is what people notice and vaguely call symbolic. A rhetorical culture whose standard tool is the compressed reference is high-context by construction. You say the four characters and let the other person unpack them, and spelling the whole thing out is the marked option, the one that says you don't think they'll get it. Indirection isn't a personality trait there. It's the efficient move in a language that made allusion cheap and explicitness slightly rude.

Chinese is also denser than almost anything else, and this is measurable. Languages trade off along two axes: how many syllables you fire per second, and how much information each one carries. Mandarin sits at one end: heavy syllables, delivered slowly. Spanish and Japanese sit at the other: light syllables, delivered fast enough to sound like a machine gun to anyone who isn't used to it. Multiply the two and everybody lands in roughly the same place, somewhere around forty bits a second, as though there were a channel capacity nobody is allowed to exceed. A Mandarin speaker and a Spanish speaker transmit at about the same rate while doing entirely different things with their mouths, and a Mandarin syllable carried more information on average in the study.

That fact has consequences for Madrid, which I'll return to.

There's even a physical residue: reading Chinese recruits visuospatial machinery that alphabetic reading doesn't, because you're identifying a shape rather than assembling a sound out of letters. Whether that changes anything past the act of reading is unsettled. The rhetorical habit isn't unsettled at all: you can watch it operate in any argument between two educated Mandarin speakers.

Why Japanese voice acting has no competition

Japanese voice acting is the best in the world and it isn't close. "They make a lot of anime" is a consequence, not the reason; the reason starts three hundred years before anime existed.

Bunraku, the puppet theater, puts three people on each puppet and one man off to the side called a tayū, performing in the gidayū style, who voices every character in the play. All of them. Old men, young women, villains, children, plus the narration, plus the emotional weather of scenes where nobody is speaking. He does it seated, in front of the audience, no costume, nothing to hide behind. That is the seiyuu job description, written in Osaka in the 1680s.

Rakugo strips it further. One storyteller on a cushion, playing every part in a comic narrative, switching characters with pitch, voice quality and a turn of the head. A fan and a hand towel are the only props, and they're standing in for objects, not people. Again: one performer, a full cast, nothing visual to lean on.

So Japan spent centuries treating one voice producing many characters as a serious art with masters, lineages, schools and a paying audience. Almost nowhere else has an equivalent. Western radio drama has its own substantial performance tradition, now more than a century old. The Japanese lineage interests me for how closely it connects solo vocal characterization with established theatrical forms.

Kabuki supplies the other half. Aragoto, the bombastic style, delivers a hero's line in a stylized roar no human being has ever used to order lunch, and it's not embarrassing; it's the thing you bought the ticket for. Extreme vocal exaggeration in Japan is not what a bad actor does. It's a technique with a name, a history and a correct execution. A Western actor who goes that big is chewing the scenery; a kabuki actor who goes that big is doing it right. That single difference in what counts as tasteful sets the ceiling for everything downstream.

Then the twentieth century adds the infrastructure. Radio drama from the 1920s through the 50s created a working class of performers who existed as voices only, before television handed anyone a face. When anime scaled, the profession was already there waiting for it. Now it's an industry with its own vocational schools, its own agencies, and an idol-adjacent star system where seiyuu release albums and fill concert halls. You enter it deliberately at eighteen as a career. It is not what stage actors do for money between plays, which is exactly what dubbing is nearly everywhere else.

And the language cooperates at every level. Japanese pitch accent coexists with expressive intonation, allowing considerable room for characterization: you can swoop, crack, drop an octave, go breathy, without endangering the word. Yakuwarigo, role language, hands the performer a specification before they make a single choice: the sentence-final particles and the pronoun already encode haughty rich girl, or gruff old man, or country bumpkin, so the script states the social type in the text. The honorific system writes the character's exact position relative to everyone else in the scene into the wording and register of a line. An actor reading a Japanese script is handed more information about who this person is than an actor reading an English one, before anyone has discussed motivation.

Japan has three centuries of solo multi-voice performance as high art, a theater tradition that rewards vocal extremity instead of punishing it, a voice-only profession that predates the medium, an industry large enough to specialize, and a language with room for vocal characterization while marking role and rank in the text. It is an unusual combination of advantages.

Why Madrid sounds like that

Back to the reel.

What I was hearing is lenition, and it's specific: intervocalic /d/ dropping out, so cansado comes out cansao and pescado comes out pescao; syllable-final /s/ aspirated down to a breath or gone entirely, mosca to mohca or just moca; unstressed syllables squeezed until they barely take up time. It is the same family of process (consonants weakening in weak positions) that produced Andalusian and most of Caribbean Spanish over a few centuries. It sounds broken to me because I'm parsing it against the articulatory clarity of the standard, and dropping codas genuinely does cut acoustic redundancy. The speakers are not missing the target. They're avoiding it.

Spanish carries considerable redundancy, which helps a listener recover words from reduced speech. Mandarin finds its own ways to contract syllables and reduce tones too; density does not prevent erosion. The social question is which reductions become desirable: which pronunciation carries value, in which company, and why.

Which is where the sense that they seem afraid of sounding cultured turns out to be exactly right.

In Spanish youth culture pijo is one of the most expensive labels you can be handed. Speaking clearly, formally, in standard Castilian does not read as neutral competence. It reads as class-marked, as alignment with money and institutions and the kind of people who have comfortable access to both. Labov's term for the counterweight is covert prestige: the non-standard form carries in-group value, authenticity, not being a fake, and that value beats the open prestige of the standard. Nobody in that reel is failing to be articulate. They're spending articulacy down on purpose, because what it buys costs more than it's worth.

One important part of that history is the reaction against Franco. The dictatorship enforced correct Castilian as state ideology: proper speech, proper comportment, proper hierarchy, radio diction as the audible sign of a decent person. When it ended, La Movida went after considerably more than the politics. It took lo cutre, deliberate crudeness and cheapness and anti-elegance, and made it an aesthetic position, while polished respectability could acquire the associations the counterculture wanted to escape. That association never died. It settled into a reflex where crisp Castilian carries a faint echo of authoritarian propriety and rough southern-flavored speech carries the opposite. Fifty years on, I suspect some of that fault line is still running through the consonants.

Then the demographics. Twentieth-century Spain moved enormous numbers of people out of Andalusia, Extremadura and La Mancha into Madrid for industrial work, for decades. Working-class Madrid has carried southern phonology ever since, by ordinary contact. What's new is the direction of travel. Those features used to be stigma, the thing migrants' children worked hardest to lose. They're now adopted on purpose by young urban speakers with no rural connection whatsoever, because the meaning attached to them flipped somewhere along the way from backward to real. Reels are where it shows most, because performed casualness is a requirement of the genre: too clean on camera reads as corporate, so the roughness gets leaned into rather than slipped into.

A country that spent forty years being told that speaking well was what decent people did came out the other side with an allergy to it, and the allergy is in the consonants of people who were born after it ended.

Buenos Aires runs it in reverse

The melody is the part with a paper trail. Rioplatense carries an Italian contour (the cantito) straight out of mass southern-Italian immigration into Buenos Aires, on top of voseo and sheísmo. A wave of people arrived and the melody of the city changed to match them. No metaphor involved.

The grammar adds to it. Voseo imperatives take final stress (mirá, vení, escuchá), so the command lands on its last syllable instead of its middle. It's a small thing that makes instructions arrive harder than the Castilian equivalents. Then lunfardo, an entire slang lexicon built in the port city and then exported through tango into the national register, which means the everyday vocabulary has a layer of deliberate invention in it that most Spanishes don't carry.

And a verbal culture on top of all of it. Buenos Aires has more psychoanalysts per capita than anywhere on earth, a city that turned talking about yourself at length into a mass institution. Tango lyrics are a literary register. Café argument is a leisure activity with rules. That produces a speech community where verbal performance is a status good and where hedging costs you something. If assertion is what gets rewarded, assertion is what you get, and every register default drifts upward until confidence is the neutral setting rather than the marked one.

To a Castilian ear that reads as exaggeration, and in the technical sense it is: their default sits higher than mine. An Italian melody, a stressed imperative, an invented lexicon and a century of professionalized talking will do that.

There is no average

Here is the thing I have never been able to reconcile. Spanish dubbing over-articulates like a preschool teacher. Urban Spanish deletes half its consonants on purpose. Both present themselves as the neutral setting, as what Spanish sounds like when nobody is doing anything in particular. They cannot both be that. They are not even adjacent.

Neither is. There is no average Spanish and there was never going to be one. There is a stack of sublanguages belonging to subcultures, each with its own idea of what an unmarked sentence sounds like, all of them claiming the center, because claiming the center is what a register does for a living.

Dragon Ball in Spain had the whole disease inside ninety seconds. The opening theme is sung the way your drunk uncles sing when they get hold of the microphone at a wedding, and one of its lines sends you off a buscar con ahínco la bola de dragón. Con ahínco. Earnestly. Diligently. With zeal. That is a word out of a written report, in a cartoon theme, sung by a man who does not sound confident he will reach the next note. My sister and I mocked that line for years. I was five and it already registered as the least fitting phrase I had ever heard.

Then the song ends and a narrator recaps the previous episode in the diction of a stuck-up schoolteacher enunciating for the back row.

Amateur-hour delivery carrying vocabulary nobody says out loud, followed immediately by immaculate articulation, aimed at eight-year-olds who could not possibly have been the audience for either. Nobody involved seems to have found this strange. I have never stopped finding it strange.

The same non-logic ran through my school. I was a middle-class kid among mostly upper-class ones, and they put real effort into sounding like they came from somewhere poorer than they did. It was the 2000s, so it arrived with the full apparatus: cani and choni styling, imported ghetto iconography, hip-hop, pants worn below the butt with the entire underwear out — not a waistband showing, the whole thing, the entire gluteus aired — and both pant legs tucked into the socks, or cinched at the ankle with elastic bands, for some reason. Anyone who was in Spain then can still see it. As intimidation it was useless. As a costume it bought you nothing outside a very small and very local scoreboard. And it was enough. It was completely enough, for most of a generation.

What gets me is that none of them knew why. Part of that aversion to sounding eloquent has a history in a dictatorship that made correct Castilian a sign of moral decency, and a counterculture that answered by making crudeness a position. Any Spaniard under forty-five has close to no first-hand contact with either. They inherited the reflex with the reason stripped out.

My own position here isn't neutral, so I'll state it. The whole system optimizes for social connotation and treats aesthetics as beneath discussion. Does this sound good is a question nobody is asking. Does this place me correctly is the only one on the table. To someone as snobbish and aesthetics-first as I am that is close to unbearable, and it doesn't resolve by picking a side: I don't want the deliberately anti-aesthetic urban variant and I don't want the over-enunciated dub either. Both are ugly, in opposite directions, and both got there by asking a social question where a musical one belonged.

I hold two views about this at once. It irritates me to live inside a set of restrictions nobody left alive can defend. It also delights me that a fifty-year-old political argument is still running through a stranger's consonants on a phone screen.

There's a line in Can This Love Be Translated?, a K-drama about an interpreter, which is not where I expected to find the phrasing I kept thinking about: there are as many languages as there are people. That's the version I'll keep. The average was always an invention, and every one of these ways of speaking is somebody's entire native tongue, with a full set of rules and a scoreboard attached, and no obligation whatsoever to be beautiful.

The kid in the reel swallowing the d in cansao isn't failing at Spanish. He's taking a side in an argument nobody told him about, and taking the one that costs him less. Do that for a few generations and everyone forgets there was an argument. What's left gets called national character.