The Nine Patterns Across Seven Languages

The comparison runs over four kinds of committed artifact: the page data itself, the ratified name list each language works from, the harvest document recording how people in that language write about the nine behaviours, and the rulings each build wrote down while it was making them. All four sit in one repository and all four are dated.

Two standards of proof are in play and this page keeps them apart. A ruling is a fact about this archive: it was made, it was recorded, and anybody can read it. A phrasing is a claim about how a language is actually used, and it counts as evidence only if somebody opened the page it sits on and read the string in place.

That is not a hole in this record so much as a description of what kind of record it is. A finding about which register a file is entered through survives a refused connection, because it is a fact about where a cluster of searches lives rather than a string somebody typed. A quotation does not survive it. Everything below is the first kind.

File 001 is the file about distance. The reader watches themselves move away from the people who get close, and cannot say why. In English it is The Disappearing Pattern, and the English name describes something the reader does to themselves.

In every other language the evidence pointed at a verb, and at a verb the reader applies to other people. Spanish reaches for alejarse. German for wegstoßen. French for s'éloigner. Brazilian Portuguese for afastar-se. Italian for allontanarsi. Japanese for 遠ざける, a transitive verb chosen over the nearest alternative because the alternative is what relationship writing in Japanese recommends, and a file named after the recommended move reads as advice rather than as a pattern.

A result found six times by six passes that could not see each other is the strongest single item in this record. It is also the narrowest. It says the point of entry into file 001 has the same shape everywhere it has been measured. It does not say the file carries the same name anywhere, and it does not.

The obvious rendering of file 001 in a Romance language is the noun for disappearance. It was the first candidate in Spanish, in Brazilian Portuguese, in French and in Italian, and it was refused in all four.

Each refusal was made under the same test. Every candidate name in every language is scored on five axes before it is ratified, and one of those axes asks whether the word collides with a named atrocity, a political event or a legal instrument anywhere the language is spoken. That axis is why the four refusals happened and it is why they happened separately.

The Spanish ruling records the word as a live national category across three regions rather than as a metaphor. The Brazilian ruling records the same collision, notes that it is the third language to reach it, and calls its own case the heaviest of the three, because there the word belongs to families rather than to law. The French candidate failed on a standing criticism from the language's own authority, plus a second reading as a missing persons case. The Italian candidate failed on two named legal instruments at once. German banned the nearest compound from copy outright.

None of those rulings cites another. Four language builds, four different bodies of national history, one axis, one answer.

Four files were harvested in enough languages to compare properly. Every one of them split.

File 002 is the apology file. Spanish found no first person route through the apology at all, and had to enter the file through guilt. Brazilian Portuguese found the same thing independently and made the same move, with the strongest sub-register being guilt for taking up space. German found the front door open: people write about apologising constantly, in the first person, and the German file is entered at the apology itself. Italian found the same open door, on the richest platform genre in its set, and its harvest says plainly that the Portuguese ruling is not inherited. Two languages could not name the behaviour and two could, and no amount of reasoning from either pair would have produced the other.

File 003 is the testing file. Spanish, German, Brazilian Portuguese and Italian all found the same thing: people do not write about the manoeuvre, they write about jealousy and distrust, and the file has to meet them there or go unfound. German and Portuguese each add the same complication from opposite sides of the world. In this subject area, in both languages, the ordinary word for a test means an online quiz, so a file named with it would have read as the quiz pattern to the only audience it was written for. Japanese needed none of that. Japanese has an established term for exactly this behaviour, used in the first person about oneself, so the Japanese file is entered through its own name and the workaround the other four needed would have been an error imported from a document.

File 004 is the file about being drawn to what harms. Spanish enters it at attraction. Brazilian Portuguese does not, and this is the one comparison in the set that inverts rather than varies: the verb for attraction is fully occupied by therapy marketing, and what Brazilians write about themselves is the verb for choosing. That is a different sentence with a different reader inside it. One is a misfortune and the other is an admission, and a page framed on attraction addresses a passive reader the search data does not contain.

File 005 is the compliment file, and it produced the second convergence in the set. Spanish, German and Brazilian Portuguese all enter it through deserving, each in its own words, each found separately, and the German harvest records its own line as the German form of the Spanish one. Japanese enters it through neither a topic nor a feeling but through a fixed phrase: the sentence the person hears themselves say when the compliment lands. In Japanese the recognition surface for this file is a sentence rather than a subject.

You have a word for this. It is almost certainly not the word an article would sell you.

A term that returns no first person searches is article inventory written by marketers, and the archive does not build toward it however large the volume looks. That rule carries into every language. The list of terms it applies to does not.

Brazilian Portuguese is the proof. Two of the Spanish saturation findings replicated in Brazil exactly and one inverted: the noun for putting things off is dead in Spanish first person search and alive in Brazilian, with its own condition category on a clinical platform and users conjugating the verb about themselves. A builder who had copied the Spanish table would have been right twice and wrong once, with nothing on hand to tell which was which.

The same pass produced the shape underneath all of it. Every dead term in the Brazilian table is one the Brazilian content market has produced heavily: coaching vocabulary, attachment vocabulary, and money vocabulary written by people who sell financial products. People who are actually suffering write the behaviour instead. The archive answers the behaviour, so a saturated term is favourable rather than awkward. It pushes a real sufferer into describing what happens, which is the thing these pages are for.

One distinction inside this had to be written down twice before it held. Saturation governs what the archive builds toward: no title, no address, no entry point. It says nothing about whether a word may appear in ordinary prose. Conflating the two turns a targeting ruling into a ban, and the strongest saturation finding in the Portuguese harvest is deliberately absent from that language's banned list for exactly that reason.

The archive speaks to one person. Which grammatical form it uses to do that is settled once per language, before any page is written, and the six decisions have almost nothing in common except the test they were made under: which form does not make a person who has just done the thing again, at one in the morning, feel handled.

Spanish took the familiar second person, because it reads as neutral written Spanish everywhere, including to readers whose spoken form differs. German took the familiar form for a reason no other language reproduces: the German file vocabulary is already administrative, so the formal pronoun stacked on top of it would have turned the Archivist into the office that keeps your file. French found no neutral option at all. Every available form places the voice in a region, and the French ruling names that cost instead of hiding it. Brazilian Portuguese took the standard Brazilian address. Italian settled it on grammar rather than tone: the Italian formal register takes third person agreement, so a recognition line written in it is the archive talking about the reader while claiming to look them in the eye, and there was no second option to weigh.

Japanese is not a pronoun ruling at all. It is a politeness level ruling, set at the plain polite register with nothing above it and nothing beneath it, which means the directness the voice depends on has to be carried by the rest of the sentence rather than by the form of address. Gendered sentence-final particles are banned in the Archivist's voice for the same reason the marks are banned everywhere else.

One clause holds in all six. The ruling binds inside the safety blocks too, which is where the pull toward formality is strongest and where yielding to it costs the most. A person in an acute moment addressed formally has been handed a form rather than a hand.

Every language routes a reader in difficulty to real help, and no two of them route to the same instruments. Each language currently carries six to eight verified routes, composed from a single file per language, never retyped onto a page. What counts as a national number, who operates it, whether it answers around the clock and whether it costs the caller anything are different questions in each territory, and every answer had to be established on its own.

Spanish arrived at the same place from the other direction, and later. One word meaning free was removed from every Spanish crisis route after those routes were already public, and ten Spanish pages that had retyped their safety block instead of composing it never received the edit. Nothing about those ten is dangerous to a reader. They are the standing proof that safety copy kept in a second place drifts away from its source in silence, which is the whole reason the composition rule exists.

Every route in every language currently ships under a standard weaker than opening the source page, and each file says so in those words, with the date. The number has to appear identically across at least three independent results, at least one of them the operator's own domain or a government domain, and one source disagreeing disqualifies the route outright. Italy has an organisation held out of the archive on exactly that: a disagreement about cost. An absent resource is lawful. An unverified one is not.

No number appears on this page. A findings record is not a crisis surface, and printing a route here would be the eleventh copy.

Some rules are the same rule in every language. Some are the same law expressed in a different unit. And some are a rule in one language and a defect in another. Telling the three apart is most of the work in a language build, and getting it wrong fails in the direction that stays quiet.

The opening ceremony is the clean case. Every page in the archive opens on a first sentence of exactly seven words, and that count is a rhythm rather than an information budget, so it does not flex per language. Japanese puts no spaces between words, so the counter that enforces the law returns one for an entire page and the law passes on any input of any length. The law did not change. The unit did, to a character band derived by measuring every seven-word opening already live in two other languages.

The shape rule for names is the case that fails. The archive tells a name from ordinary vocabulary by its capital letter, mechanically. German capitalises every noun, so the same rule would have banned ordinary German words from prose across the whole archive. Japanese has no case at all, so there the rule does not merely fail, it does not exist, and it had to be replaced with an explicit class field on every canon entry before a single Japanese page could be written.

Then there is the class that fails without saying so, and Italian produced the worst instance found anywhere in this project. Four consecutive Latin script languages had treated accent folding and word anchoring as one operation, because in those four they happened to coincide. In Italian the accents sit on function words: the accented and unaccented forms of the same two letters are two different everyday words. Folding a rule's needle on any of them turns it into a needle on the commonest word in the language. And the anchor most rules are built from is defined over the ASCII alphabet, so an anchor placed against an accented letter matches nothing at all. In Italian the word it kills is the first word of nearly every question the archive asks, so a rule written that way passes every page while checking none, and a rule that checks nothing looks exactly like a page set with nothing wrong in it.

Two more, and both are about ownership rather than grammar. French found one coined framework already holding the territory these pages describe, and had to refuse the single word French searchers reach for most. Italian found three, belonging to three different authors, and two of the three are also what the searcher types, which puts the evidence rule and the ownership rule against each other head on. Italian paid a price French did not: two of the seven callout labels could not take their literal rendering, because the literal rendering would have placed another author's framework word inside the archive's own label set.

And one thing carried better than it does in English. The hope refrain is built on a root pair, a word in its written form and the same word in its rewritten form, and the door it belongs to locks to that pair. French carries the pair intact and more cleanly than Spanish does. Where a language will not give a clean pair, the argument is made again by other means and the loss is recorded rather than forced.

The archive never marks the reader's gender. Where a sentence would have to mark it, the sentence is written again into one that does not. No slash, no symbol, no inserted vowel, in any language. It is a craft rule rather than a political one: the surface of a page stays open so the reader supplies themselves, and a mark on the page is the archive announcing a position it does not take.

What was not law until recently is that the rule has to be worked out against each language's actual grammar before the first page is written. Four of the first seven archives were completed without it. The remediation cost four separate passes.

The ruling cannot be carried from one language to another, and those five rows are the proof. Spanish and Portuguese build the compound past on an auxiliary that does not vary, so the whole exposure is one adjective slot. French builds it on an auxiliary that agrees with the subject, and that reaches into the archive's own vocabulary: the French name for file 002 is built on a verb of exactly the kind that agrees, so a ruling that did not answer that case would have left the hardest page in the language unwritable. German has neither exposure, which is precisely why the one construction German does have went unseen for so long. Japanese has no grammatical slot for it at all, and its exposure is lexical instead.

The last row is worth more than the other four. A zero that explains itself is a finding. A zero that does not is an unasked question.

No automatic rule was proposed for this in any language, and the refusal is deliberate. Catching it mechanically would require knowing which noun a participle attaches to. The marks are enforced by rule; the agreement is not, and a document implying otherwise would be claiming a protection nobody is providing.

Stating the gaps is part of the record rather than a caveat attached to it.

No searcher is quoted here, in any language. The English pages carry eighty-eight blocks of quoted phrasing. The other six languages carry none, and that pair of numbers is checked against the live page data every time this page is rebuilt rather than asserted in prose.

Several sectors in several languages have no harvest at all, and the pages in them are written from the mechanism rather than from evidence about how that language phrases the behaviour. German named four of its nine files and every arena as unharvested. Italian named its arena, observer, reach and bridge sectors, and its Swiss region. Brazilian Portuguese found one line in the whole money arena and nothing at all in the creative one. Those are quality limits on named pages, recorded at the page.

The file by file comparisons above are drawn only from the files each harvest actually reached. Where a language did not harvest a file, this record says nothing about that language and that file, rather than reasoning across from a sibling language. Reasoning across is the specific error the saturation finding exists to name.

And the entire quotation layer is one environment change away from being restorable. Every harvest document keeps its source list intact. When the network refusal lifts, each list is walked again and a quoted block returns one quote at a time, as each is confirmed, never wholesale.