00 Foundations

A First-Principles Framework for Thirteen Languages — Volume 00: Foundations§

English · Spanish · Brazilian Portuguese · French · German · Russian · Turkish · Hebrew · Arabic · Hindi · Mandarin · Japanese · Korean

0Why this exists§

The French volumeThe French volume made one move: it separated primitives from derived constructs, so grammar could be reasoned about instead of memorized. That move survives contact with thirteen languages. But between two languages you can get away with a comparison; between thirteen you need a theory — because the differences stop looking like a list of quirks and start looking like what they actually are: a small number of engineering decisions, made differently.

This volume is that theory. It makes four claims:

1. The meaning space is universal and finite. Every language draws from the same inventory of things-that-can-be-grammatically-marked. No language marks anything outside it.
2. Grammar is obligation. What makes languages differ is not what they can express — all can express everything — but what they force you to express in every sentence.
3. Every language is a profile: a subset of dimensions wired to "mandatory," plus a packaging strategy, plus a word-order scheme. Thirteen languages = thirteen profiles over one board.
4. Learning a new grammar = diffing profiles. Once you hold the board, language number fourteen is not a new subject. It is a configuration file.

The proof that claim 1 is even coherent comes from an unexpected place: engineered languages, above all John Quijada's Ithkuil, which spent five decades enumerating the board exhaustively. We will use it the way physicists use limit cases — not to learn it, but to see what natural languages are choosing not to do. That is Section 12.

1The Theorem§

"Languages differ essentially in what they must convey and not in what they may convey."

This single sentence is the load-bearing wall of the whole framework. Read it twice, because it inverts how grammar is usually taught.

Any language can express any meaning. Mandarin has no tense marking, and Mandarin speakers are not confused about time — they say "yesterday" when it matters. Russian has no articles, and Russians are not confused about which cup you mean. English has no evidential marking, and you can always add "apparently" when you didn't see it yourself.

So the grammar of a language is not its expressive range. The grammar is its disclosure requirements — the meanings you are forced to commit to whether or not you care, in every single sentence, on pain of ungrammaticality:

• In English you cannot mention a cat without deciding the cat or a cat. Definiteness is mandatory.
• In Russian you cannot report an action without deciding completed or ongoing. Aspect is mandatory; there is no neutral verb.
• In Turkish you cannot describe a past event without deciding did I witness this or learn of it secondhand. Evidence is mandatory.
• In Japanese you cannot finish a verb without deciding what my social relationship to the listener is. Register is mandatory; there is no unmarked form.
• In Hindi you cannot say "I eat" without disclosing your own gender — khātā hūṁ if you're a man, khātī hūṁ if you're a woman.
• In French you cannot get through the sentence without gender-tagging every noun — and in Arabic and Hebrew you cannot even say "you" without first deciding the gender of the person you're talking to.

2One event, thirteen forms§

Here is a single event — a cat ate a fish, both known to speaker and listener — filed in all thirteen jurisdictions. This table is the whole framework in miniature; we will return to it constantly.

Thirteen sentences. One event. Every difference above is one of exactly three kinds of decision: which dials are wired to mandatory (Sections 4–5), how the marks are physically packaged (Section 6), and in what order the pieces are laid out (Section 7). There is no fourth kind.

3The stack§

Every grammar in this book — every grammar, period — compiles through the same five layers:

The French volumeThe French volume lived almost entirely in L1 plus fragments of L2 and L5. The upgrade is seeing all five — because what makes Japanese feel alien to a Spanish speaker is not L1 (identical), but different settings at L2 (politeness in, person out), L3 (agglutinative), L4 (head-final), and L5 (no agreement at all).

4Layer 1: The five primitives — and the one that wobbles§

The five primitives survive unchanged from the French volume. Every language on this list — every human language examined by typology to date — builds propositions from:

ENTITY and ACTION are rock-solid: every one of the thirteen has a recognizable noun class and verb class, and typologists have found no clear counterexample anywhere. RELATION and REFERENCE always exist as functions, though the machinery varies wildly.

QUALITY is the wobbler — and the wobble is one of the deepest facts in this book.

"Adjective" is not a stable universal category. It is disputed territory between noun and verb, and each language files QUALITY under a different flag:

5Layer 2: The Dial Board§

This is the heart of the framework. Below are the fourteen dimensions that recur across the thirteen languages — the dials. For each language, each dial is ● wired-mandatory, ◐ available-optional (or partial), or ○ absent from the grammar (expressible only lexically).

The nominal dials — what you must disclose about an ENTITY§

* (ungrammatical) "Marked objects" = differential object marking (DOM), explained below.

The verbal dials — what you must disclose about an ACTION§

Read the columns, not the rows, and patterns jump out that no bilateral comparison can show:

Gender is a European-and-Semitic obsession. Seven languages wire it mandatory; the entire eastern half of the table (Turkish, Mandarin, Japanese, Korean) has none, anywhere, not even in pronouns — Turkish o means he, she, and it. A billion-plus people live entirely without grammatical gender and lose nothing. It is a tagging system, not a feature of reality.

Person and Social are complementary — almost a conservation law. Spanish packs maximal who into the verb (hablo = I-speak, six distinct forms) and zero social relation. Japanese packs zero who (tabeta doesn't know its own subject) and maximal social relation. Both drop pronouns constantly — Spanish because the morphology licenses it, Japanese because context does. Two opposite dial settings produce the same surface behavior for opposite reasons.

Evidence is rarer than it should be. Only Turkish wires it fully: every past-tense verb chooses -di (I witnessed) or -mIş (I infer/was told). You cannot gossip in Turkish without flagging that it's gossip. The sleeper: German has a covert evidential — Konjunktiv I (Er sei krank "he is-said-to-be sick") is mandatory in news reporting to mark secondhand claims. And Ithkuil, of course, wires a nine-way evidence dial into every verb (§12).

Aspect is wired somewhere in almost everyone — but at different depths. English puts it in auxiliaries (was eating / has eaten). Spanish fuses it into endings (comió/comía). Mandarin makes it the only verbal inflection-like thing it has (了/过/着). Russian goes deepest: aspect is wired into the lexicon — verbs come in perfective/imperfective pairs (писать/написать) and there is no aspect-neutral way to say anything. You don't conjugate your way into Russian aspect; you choose which verb exists.

Differential Object Marking — the hidden sibling pattern. Four languages on this list mark some direct objects and not others, by the same logic with different hardware: Spanish (personal a: Vi a María vs Vi la película), Turkish (accusative -ı only on definite objects: balık yedim "I ate fish" vs balığı yedim "I ate the fish"), Hebrew (et only before definite objects), Hindi (ko on definite/animate objects). Russian half-joins via animacy (animate objects borrow the genitive form). Five grammars, one rule — the more definite or alive the object, the more it gets flagged — and no traditional textbook connects them, because each is taught as a local quirk. This is the framework earning its keep: a quirk in one language is usually a dial you'll meet again.

Classifiers are individuation machinery. Mandarin, Japanese, and Korean treat bare nouns as unbounded stuff — closer to "catness" than "a cat" — so counting requires a unitizer: 一只猫 yì zhī māo "one animal-unit of cat," 三条鱼 sān tiáo yú "three long-flexible-units of fish." English does this too, but only for actual stuff: two sheets of paper, three head of cattle. The dial exists everywhere; three languages wired it to everything.

6Layer 3: Packaging — the four codecs§

Same meanings, four physical strategies for attaching them. Every language on the list uses one dominant codec:

Watch one meaning — roughly "I will not be able to come" — pass through three codecs:

• Mandarin: 我不能来 wǒ bù néng lái — four separate words: I / not / able / come. Zero compression.
• Spanish: no podré venir — podré fuses able+future+1st-person into one opaque ending. Lossy compression: you cannot cut -ré into its component meanings.
• Turkish: gelemeyeceğim — one word: gel-eme-yecek-im = come + unable + future + I. Lossless compression: every meaning is a visible, separable slot.

The fourth codec deserves its own exhibit, because nothing in European experience prepares you for it. In Arabic and Hebrew, the dictionary entry is not a word — it is a consonantal skeleton, and grammar is the vowels you thread through it. The root k-t-b ("writing") in both languages:

Same root, same machine, two languages — which is why Hebrew and Arabic form one learning cluster despite sharing little day-to-day vocabulary. The templates are themselves the grammar: Hebrew's seven binyanim and Arabic's ten forms are valence dials (§10) implemented as vowel patterns. In Semitic, vocabulary is grammar. There is no boundary.

One more packaging fact, easy to miss: French runs two codecs at once. Written French is a fusional language with rich agreement (ils parlent); spoken French has eroded most endings into silence (parle/parles/parlent are homophones), collapsing toward the isolating pole — which is why spoken French, like English, requires subject pronouns while Spanish doesn't. The orthography preserves a grammar the mouth abandoned. English took the same erosion path three centuries earlier — which is why the most Mandarin-like language in Europe is English.

7Layer 4: Linearization — one master switch§

Word order looks like thirteen arbitrary conventions. It is mostly one parameter with a cascade of consequences: does the head of a phrase come first or last?

Head-initial (verb before object, preposition before noun): English, Romance, Semitic, Mandarin-ish. Head-final (object before verb, postposition after noun): Turkish, Japanese, Korean, Hindi.

The Greenberg correlation bundle — properties that travel together:

Three things to take from this table.

First, the four SOV languages are typological siblings with zero shared ancestry. Turkish, Hindi, Japanese, Korean — four families, one architecture: verb last, postpositions, everything modifying a noun stacked in front of it. This is why I said earlier in this series that Hindi is "genetically European, typologically Japanese." Profiles, not family trees, are what you actually feel when learning.

Second, the bundle is a tendency, not a law — and Mandarin is the exhibit. Mandarin is SVO like English but puts relative clauses before the noun like Japanese — a genuinely rare combination among the world's languages. Greenberg's correlations are statistics, not physics. The framework must be honest about this: the board constrains; it does not dictate.

Third, German is a disguise. German main clauses look V2 — verb rigidly second: Heute esse ich Fisch ("today eat I fish"). But subordinate clauses unmask an SOV chassis: …weil ich Fisch *esse ("…because I fish eat"), and compound verbs bracket the whole clause: Die Katze *hat den Fisch gefressen. German is a head-final language wearing a head-initial mask in main clauses — learn it that way and half its "weird" orders become one rule.

And the deepest linearization insight of all, in programming terms, since this framework speaks that dialect:

There is a third system besides position and label, and it belongs to East Asia: topic-prominence. Japanese は wa, Korean 은/는 eun/neun, and Mandarin's topic-comment structure (那条鱼，猫吃了 "that fish — the cat ate it") organize the sentence around what we're talking about rather than who did what. Japanese grammaticalizes the difference outright: wa marks topic, ga marks subject, they can co-occur in one sentence (象は鼻が長い zō wa hana ga nagai — "as-for elephants, the nose is long"), and no European language has anything equivalent. When Japanese feels slippery to a European-language speaker, this dial — not the script, not the politeness — is usually the real culprit.

8Layer 5: Propagation — how far a tag spreads§

The French volumeThe French volume's stylesheet metaphor generalizes into a real theory: gender/number/case tags are a type system, and agreement is type-checking. A language assigns each ENTITY a type; the question is how many other words are forced to echo it.

Three propagation facts are worth the price of the book:

Russian's past tense agrees in gender, not person — он читал / она читала (he read / she read-FEM) — because the Russian past is historically a participle: an adjective. The "weird rule" is a fossil of a costume change (§10): a verb that dressed as a QUALITY a thousand years ago and never fully changed back. First-principles reasoning turns an exception into an etymology.

Hindi's agreement switches targets. In imperfective clauses the verb agrees with the subject. In perfective transitive clauses the subject gets locked behind the ergative ne — and the verb, needing something to agree with, agrees with the object: Billī ne machhlī khāī — "ate-FEM," agreeing with the fish. This is the single most disorienting mechanism on the thirteen-language list, and it falls out of two dials interacting: aspect (perfective) × case (ergative). Not a quirk — a collision.

Arabic breaks its own type system on purpose: non-human plurals take feminine singular agreement (deflected agreement). Books — plural, masculine — agree as if "she." Every type system has undefined behavior somewhere.

9The verb as struct§

The French volumeThe French volume called a conjugated verb a compressed data packet. Here is the full comparative payload — what each language's verb actually carries:

Stare at the two extremes. The Spanish packet knows everything about who and nothing about the room. The Japanese packet knows everything about the room and nothing about who. And Mandarin ships an empty struct and lets context do all disclosure. All three work perfectly. Grammar is not information the mind needs — it is information the language insists on. That is Jakobson's theorem made concrete in a struct diagram.

10Costume changes: the four great conversions§

The French volumeThe French volume's sharpest idea: complicated grammar is usually one primitive wearing another's costume. Cross-linguistically there are four great conversions, and every language on the list owns machinery for each.

V → E (action as thing)§

V → Q (action as property) — and the reorientation that matters most§

Participles exist everywhere (a broken window, une fenêtre cassée). But the head-final languages take this conversion to its logical extreme, and it is the single biggest mental rewiring on the list for a European-language speaker:

Turkish, Japanese, and Korean have no relative pronouns. No who, no que, no der. Instead, an entire clause converts to a QUALITY and stacks in front of the noun like a giant adjective:

• Turkish: balığı yiyen kedi — "the fish-eating cat" = the cat that ate the fish
• Japanese: 魚を食べた 猫 sakana o tabeta neko — "[ate-the-fish] cat"
• Korean: 생선을 먹은 고양이 saengseon-eul meogeun goyangi — same architecture

You built this in English without noticing: "a fish-eating cat" is the same conversion; head-final languages just let it scale to any clause of any size. Meanwhile Hindi opts out with a third strategy, the correlative — a paired-pointer construction with no equivalent in Western Europe: jo billī ne machhlī khāī, *vo bhāg gaī* — "which cat ate the fish, that-one** ran away." Two linked clauses instead of one embedded one.

E → Q (thing as property)§

English stacks nouns bare (chocolate cake); French exposes the RELATION with a preposition (gâteau au chocolat); German fuses into compounds (Schokoladenkuchen); Japanese links with の, Mandarin with 的, Turkish with the possessive-compound (okul kapısı "school its-door"). And Semitic has the construct state — a dedicated grammatical form for smashing two ENTITIES together: Hebrew beit ha-sefer "house-of the-book" (= the school), Arabic baytu l-mudarrisi "house-of the-teacher" — where, elegantly, definiteness is disclosed only once, on the final member, and spreads backward through the chain.

Q → E (property as thing)§

Spanish lo bueno ("the good [of it]"), German das Gute, English the poor. The neuter article is a nominalization machine. Minor, but it completes the square: all sixteen ordered pairs of primitive-to-primitive conversion occur somewhere; these four are the industrial-scale ones.

11Three unifications§

Three case studies where the framework collapses "unrelated facts" across thirteen languages into one mechanism each.

11aPossession is location + existence§

Seven of the thirteen languages have no verb "to have." The European transitive-possession verb is the anomaly, not the norm:

English, Romance, German, and Mandarin (有 yǒu) own a "have"; the rest express possession as an ENTITY existing in a RELATION to you. Which is what possession is — "have" is a derived construct that seven languages never bothered to derive. Mandarin's 有 even moonlights as the existential ("there is"), sitting exactly on the seam.

11bThe copula is a patch§

Section 4 set this up; here is the closure. "To be" exists to let ENTITY and QUALITY predicates do a verb's job in languages where they can't do it alone:

• QUALITY filed under ACTION → no copula needed for adjectives: Japanese takai, Mandarin tā hěn gāo, Korean descriptive verbs.
• QUALITY filed elsewhere, patch omitted in present tense → zero copula: Russian, Arabic, Hebrew — all three simply juxtapose (she doctor) and patch only in past/future.
• Patch installed and then split → Spanish and Portuguese run two copulas, dividing the work by a dial most languages leave to context: ser for essence, estar for state. Es aburrido (he's boring) vs está aburrido (he's bored). Portuguese adds ficar for becoming — a three-way split.

One dial — does QUALITY predicate alone, and if not, how is the patch configured — explains the Russian zero, the Japanese conjugating adjective, and the Spanish ser/estar wall, three of the most notorious stumbling blocks in three different classrooms.

11cNegation has four addresses§

Every language negates; the machinery lives at one of four addresses:

Plus two refinements the table hides: Mandarin and Arabic select the negator by tense/aspect (不 bù general vs 没 méi for non-completion; Arabic lā present, lam past, lan future) — the negator carries temporal information so the verb doesn't have to. And English is the strangest of the thirteen here: negation (and questions) require summoning an auxiliary from nowhere — do-support: I eat → I do not eat. No other language on the list conjures a dummy verb to host its negation. When English feels "easy," remember it hides this piece of genuine typological exotica.

12The limit cases: what engineered languages prove§

The claim in Section 0 — that the meaning space is finite and enumerable — is not a metaphor. People have enumerated it. Constructed languages are experiments at the edges of the dial board, and four of them mark the corners of the possibility space that our thirteen natural languages live inside.

Ithkuil: every dial wired to mandatory§

John Quijada spent ~45 years (begun 1978; published 2004; fully redesigned as New Ithkuil, 2023) building the maximal grammar: a language where nearly every dimension a human language could obligatorily mark, is obligatorily marked, with the explicit goal of expressing "deeper levels of human cognition briefly yet overtly and clearly" while minimizing ambiguity.

Every Ithkuil word (a formative) runs through ten fixed slots — a Turkish verb perfected — and must disclose, among other things:

The famous party trick (from the 2011 version): Tram-mļöi hhâsmařpţuktôx — two words — "On the contrary, I think it may turn out that this rugged mountain range trails off at some point." Nobody speaks Ithkuil fluently, not even Quijada; the disclosure load exceeds real-time human processing. That is the finding. Ithkuil is not a usable language; it is a proof by construction that the dial board is finite, enumerable, and vastly larger than any natural language's selection from it.

Now the payoff for us: every "exotic" obligatory category in our thirteen is one Ithkuil dial, wired on. Turkish evidentiality = Validation, nine values collapsed to two. Spanish ser/estar = a sliver of Essence/state distinctions. Mandarin classifiers = Configuration/Perspective, paid at the numeral. Japanese politeness ≈ Bias-adjacent social disclosure. Russian aspect = the Aspect dial welded into the dictionary. Nothing in any natural language's grammar is sui generis. It is all selections from one board — which is precisely why the profile method (§13) works.

The other three corners§

• Lojban (Logical Language Group, 1987–): grammar as predicate logic. No noun/verb distinction at all — every content word is a predicate with numbered argument places (klama = "x1 goes to x2 from x3 via x4 by means x5"). It dissolves our L1 primitives into pure RELATION — proving the five primitives are a human packaging convention, not a logical necessity. Its "attitudinals" are a full Bias system as optional particles.
• Toki Pona (Sonja Lang, 2001): the opposite pole — ~120–137 words, virtually no obligatory anything. The minimal viable language. Proves the floor is astonishingly low if you let context carry everything. Mandarin sits closest to this corner among natural languages; Toki Pona shows how much further the slider goes.
• Natural Semantic Metalanguage (Wierzbicka & Goddard, 1972–): not a conlang but a research program with the same shape — the empirical hunt for the semantic floor: 65 semantic primes (I, YOU, WANT, KNOW, GOOD, BAD, BECAUSE, IF…) claimed to be lexicalized in every human language tested, out of which all other meanings can be paraphrased. The five primitives of the French volume are, in effect, the grammatical cousins of NSM's lexical floor. (Suzette Haden Elgin's Láadan, 1982, deserves its footnote here: a feminist conlang that made evidence and speech-act intent mandatory sentence particles — Turkish's dial, deliberately rediscovered.)

Natural languages are points in the space these four corners frame: more obligatory than Toki Pona, less than Ithkuil, packaged human-style rather than Lojban-style, floored on something like NSM's primes.

13Why the board is finite: grammar is eroded vocabulary§

One question remains before the method: why these dials? Why do unrelated languages keep grammaticalizing the same dozen-odd meanings — tense, aspect, definiteness, evidence — and never, say, an obligatory marker for "expensive"?

Because grammar is not designed. It condenses — out of vocabulary, along a small set of one-way erosion paths (the field is grammaticalization; Heine & Kuteva catalogued the pathways). Words with the right kind of general meaning get used constantly, wear down phonetically, glue onto neighbors, and fossilize into obligatory machinery:

• Futures condense from want/go/have. English will was willan "to want"; gonna is going to mid-condensation, right now, in your mouth. Spanish cantaré is medieval cantar he — "I-have to-sing" — the auxiliary fused onto the infinitive's tail. Mandarin 要 yào "want" is condensing toward "will" as we speak.
• Articles condense from demonstratives and "one." The < þæt "that"; un/una < Latin unus; French le < ille "that one." Languages without articles (Russian, Hindi, Japanese) simply haven't run this program — and Colloquial Arabic and Hebrew ran it long ago (al-, ha-).
• Aspect condenses from "finish." Mandarin 了 le < 了 liǎo "to finish." Spanish acabar de. English have-perfects.
• Evidentials condense from perfects. Turkish -mIş was a perfect ("has come") that drifted into "evidently came" — the standard pathway to hearsay marking.
• Negation cycles. French pas was the noun "step" (je ne marche pas — "I don't walk a step"), recruited for emphasis, then mandatory, and now it is eating ne itself: the Jespersen cycle, mid-stride.

14The method: profiling a new grammar in an afternoon§

Everything above compresses into a working procedure. To acquire the grammar-map of any new language — including the ten on your list you haven't started — extract its profile with twenty questions. Answers to these place the language on every layer of the stack:

Packaging & order (5):

1. Codec: isolating / agglutinative / fusional / templatic?
2. Basic order — and is it position, label (case/particles), or topic that assigns roles?
3. Head-initial or head-final? (One answer predicts adpositions, genitive order, and relative-clause direction — check for Mandarin-style exceptions.)
4. How do relative clauses work: pronoun-after, participle-before, or correlative?
5. Verb-second or verb-bracket phenomena? (German-type masks.)

Nominal dials (5):

1. Gender/class: how many, and does it reach verbs?
2. Number: obligatory? A dual? Classifiers for counting?
3. Definiteness: articles, case-tricks (Turkish), topic-tricks (Japanese), or nothing?
4. Case: how many, on nouns or on articles? Any ergative split?
5. Differential object marking — are definite/animate objects flagged?

Verbal dials (6):

1. What does the verb pack? (Person? Gender? Nothing?)
2. Tense-based or aspect-based system? Are aspect pairs lexical (Russian-type)?
3. Mood: is there a live subjunctive/irrealis?
4. Evidence: obligatory (Turkish), covert (German Konjunktiv I), or absent?
5. Social: T-V, multi-level (Hindi), wired-into-every-verb (Japanese/Korean), or none (Hebrew)?
6. Negation's address: before, sandwich, in-packet, or predicate-swap? Tense-sensitive negators?

Derived machinery (4):

1. How does V→E work (infinitive, gerund, nominalizer, template)?
2. Copula: full, zero-in-present, or split? (Predicts how QUALITY is filed.)
3. Possession: "have" or exist-at-me?
4. Questions: particle (and does it move?), inversion, or intonation?

Twenty answers ≈ one page ≈ the skeleton a textbook spreads across four hundred. The textbook is still needed — for the ten thousand facts of vocabulary, irregulars, and idiom — but you will file every one of those facts into a slot you already have, instead of building the shelving while the boxes arrive. That is the entire practical promise of this framework: not that grammar becomes unnecessary, but that it becomes addressable.

And one more use: the profiles predict transfer costs. Group the thirteen by profile distance rather than family, and the learning clusters reorganize: English sits nearer Mandarin (analytic, positional, aspect-flavored) than nearer its cousin Russian. Hindi's cheapest neighbors are Japanese and Korean, not its cousin French. Spanish→Portuguese is a dialect-scale hop; Hebrew→Arabic is one codec shared. You already knew the family tree. The profile map is the one that prices your effort.

15The thirteen profiles§

The twenty questions, pre-answered. One paragraph of spec per language — these are the seeds of Volumes 01–13.

ENGLISH — IE·Germanic. Fusional eroded nearly to isolating. SVO rigid, roles by position; head-initial. Dials: definiteness ●, number ●, tense ●; gender/case relics in pronouns only; no social, no evidence, no classifiers. Packet outsourced to auxiliaries; do-support for negation/questions is its private exotica. Costumes: gerund -ing, bare noun-stacking. Trap for its speakers: assuming the and tense are universal necessities. They are minority tastes.

SPANISH — IE·Romance. Fusional. SVO flexible, pro-drop (person morphology licenses silence). Dials: gender ●2 with full NP concord, definiteness ●, preterite/imperfect aspect ●, live subjunctive ●, T-V ● (vos in the Río de la Plata), DOM ● (personal a). Split copula ser/estar = grammaticalized essence-vs-state. Clitic pronouns stack in rigid order (se me lo). Traps: ser/estar, subjunctive triggers, aspect choice — the three live dials English lacks.

PORTUGUESE (BR) — Spanish's profile with three twists: the inflected infinitive (person on the abstract form — unique here); ficar making the copula a three-way split; and spoken BR reorganizing person (você/a gente + 3sg) while clitics retreat — a system mid-erosion, French-style, worth watching in real time.

FRENCH — IE·Romance. The two-codec language: written = full fusional concord; spoken = eroded toward analytic, hence mandatory subject pronouns (English's solution, independently rediscovered). Negation sandwich with ne dying. Liaison surfaces agreement the spelling preserves. Gender ●2, subjunctive ● (narrower than Spanish). Trap: the writing documents a different language than the sound.

GERMAN — IE·Germanic. Fusional. SOV chassis, V2 mask; verb brackets (hat … gefressen). Dials: gender ●3 (arbitrary — learn noun+article as one word), case ●4 carried on articles, T-V ●, covert evidential ◐ (Konjunktiv I for reported speech). Separable prefixes scatter the packet across the clause. Compounding = industrial E→Q. Traps: article-borne case; the two word orders are one grammar, not two moods of chaos.

RUSSIAN — IE·Slavic. Dense fusional. Order free — case does roles, order does information flow. Dials: case ●6, gender ●3, aspect ●● welded into the lexicon (verb pairs; no neutral form), zero copula in present, no articles, possession as u menya "at-me." Past tense agrees in gender (fossil participle). Motion verbs split uni/multidirectional — another lexicalized dial. Traps: aspect-pair choice and case chains; everything else is bookkeeping.

TURKISH — Turkic. Pure agglutinative slot-machine; vowel harmony welds the seams. SOV, postpositions, participle-before relatives, genitive-before. Dials: case ●6, evidence ● (-di/-mIş — witnessed vs inferred, unskippable), possessive suffixes on nouns, DOM ● (accusative = definite), zero gender anywhere. Question particle mi floats to whatever it questions. No "have": var/yok exist/not-exist. The natural language nearest Ithkuil's architecture. Trap: none of your European reflexes transfer — which is exactly why the profile method shines here.

HEBREW — Afro-Asiatic·Semitic. Templatic: root × binyan (7 templates = voice/valence dial). SVO (modern). Dials: gender ● on nouns, verbs, and "you" (ata/at), definiteness ● (ha-), DOM ● (et before definite objects only), no T-V at all — the list's one egalitarian pronoun system. Zero copula in present; possession yesh li; construct state (smichut) for E→E fusion. Trap: reading the root through the vowels — vocabulary and grammar are the same skill.

ARABIC (MSA) — Semitic. Templatic, ten verb Forms = valence dial. VSO (dialects SVO). Dials: gender ● incl. verbs and "you," dual ● (kitābān "exactly-two-books" — a number value everyone else dropped), case ●3 (eroding in speech), definiteness ● (al-). Broken plurals re-template the noun (kitāb→kutub); non-human plurals agree feminine-singular. The real trap is external: diglossia — MSA and the dialects are two profiles, and you must choose which one you're learning.

HINDI — IE·Indo-Aryan. Genetically European, typologically Japanese: SOV, postpositions, head-final. Dials: gender ●2 reaching the verb (and the speaker: khātā/khātī hūṁ), split ergativity ● (perfective transitive → ne on subject, verb agrees with object), politeness ●● (tū/tum/āp, three levels, verb-marked), DOM ● (ko), no articles. Correlative relatives (jo … vo); compound verbs add aspect flavor (khā liyā "ate-up"). Trap: the ergative flip — drill it as a mechanism, not an exception.

MANDARIN — Sino-Tibetan. The isolating pole: empty verb struct, zero agreement, no tense/gender/case/articles/plural. What is wired: classifiers ● (nouns are stuff; counting needs a unit — 只/条/本), aspect particles (了/过/着), topic-comment architecture, 把/被 reorganizing objects, 的 as the universal modifier-glue (and prenominal relatives despite SVO — the bundle-breaker). Tones are lexical, not grammatical. Trap: nothing is hard except believing that nothing is mandatory; the discipline is in the discourse, not the morphology.

JAPANESE — Japonic. Agglutinative, SOV, head-final; particles as keyword arguments (が/を/に/で/の); topic は vs subject が grammaticalized. Dials: politeness ●● — no neutral verb exists (taberu/tabemasu + honorific/humble axis meshiagaru/itadaku); tense ●; person ○, number ○, gender ○ — the anti-Spanish. Adjectives are verbs; negation is an adjective (tabenai → tabenakatta); adversative passive (雨に降られた "was adversely rained upon"); classifiers ●. Giant-adjective relatives. Trap: not the script — the は/が and register dials, which have no European analog.

KOREAN — Koreanic. Japanese's typological twin with zero genetic kinship — the strongest single proof on the list that profile, not family, is what you feel. Agglutinative SOV, particles (이/가, 을/를, 은/는), prenominal relatives, classifiers ●, person/gender ○. Speech levels ●● (three live: -다/-어요/-습니다) plus subject-honorific -si-; verb-final endings stack tense+honor+level+mood. Hangul is featural and learnable in a weekend — the lowest-cost script on the list attached to one of the higher-cost register systems.

16The series, and how to use it§

Each language volume runs the same fixed structure, so the volumes are diffable — you can lay Spanish beside Japanese and see, dial by dial, exactly what changes. That is the point of a framework: it makes comparison mechanical so insight can go where mechanics can't.

And the honest caveat, stated once: this framework is a map, not a vehicle. Knowing that Spanish wires preterite/imperfect while English doesn't will not make comió/comía come out of your mouth under pressure — only drilled sentences do that (that system lives elsewhere in this project, and it is the engine; this book is the atlas). What the map buys is different: no rule ever again arrives as an arbitrary humiliation. Every strange thing a language does to you, you will recognize as a dial you have seen before, wired on. Grammar stops being thirteen mysteries. It becomes one machine, configured thirteen ways.

Sources§

• John Quijada, New Ithkuil: Grammar Design (v1.3.2, Feb 2023) — ithkuil.net design doc; A Grammar of the Ithkuil Language; New Ithkuil overview (Conlang wiki); Ithkuil (Wikipedia)
• Roman Jakobson, "On Linguistic Aspects of Translation" (1959) — the obligatoriness theorem
• Anna Wierzbicka & Cliff Goddard, Natural Semantic Metalanguage — 65 semantic primes (Wikipedia); primes guide
• Joseph Greenberg, "Some Universals of Grammar" (1963) — the word-order correlation bundles
• Bernard Comrie, Aspect (1976); Heine & Kuteva, World Lexicon of Grammaticalization (2002)
• The World Atlas of Language Structures (wals.info) — the features-×-languages matrix this framework's tables miniaturize
• Lojban (lojban.org); Toki Pona (Sonja Lang, 2001); Suzette Haden Elgin, Láadan (1982)