Cairn  /  Ve'a  /  Grammar
The Ve'a Grammar Reference

How the language works

Ve'a is built around a single demand: you cannot state a fact without marking how you know it. Everything else — the split first person, the six cases, the aspect system, the word order — arranges itself around that demand. What follows is the full reference, enough for a second linguist to build and parse sentences. Every rule and example is copied from the language's own design documents. The apostrophe in "Ve'a" stands for a glottal stop; the endonym is Veʔa.

1The evidential system

This is the spine. Every verb, and every predication of a quality, must end in an evidential — a suffix declaring the source of the claim. In any assertion the slot is not optional: you cannot say "the mountain is large" in Ve'a without also saying how you know. (A command is the one exception — an imperative like ta naen subi-do, "help me," makes no knowledge claim, so it needs no evidence.)

SuffixNameMeaning
-saDirectI know it by sensory perception
-teReportI know it by hearsay, testimony, or text
-kuInferenceI worked it out from evidence
-ʔaIntrospectiveI observe this state in myself
-ʔeʔaRecursiveI am aware of my own awareness
-neInterrogativeBy what evidence do you claim this? (see §18)

The two inner evidentials carry the glottal stop ʔ — the sound Ve'a reserves for self-observation. They are not ranked above the others; Ve'a has no reliability hierarchy. All are first-class, distinct epistemic acts. Choosing a less-direct evidential when a more-direct one is available is meaningful — it signals distance or emphasis on the source.

gawa nara-ben-sa
river  large-COMP-DIRECT
"The river is larger — I see it directly."
gawa nara-ben-ku
river  large-COMP-INFER
"The river is larger — I infer it, perhaps from a map."

One evidential per verb

The six evidentials are paradigmatic: they occupy a single slot and are mutually exclusive. You never stack two on one verb. Layered epistemic claims are built by nesting clauses, each with its own verb and its own single evidential (see §12). A practical maximum of about three nested layers is a cognitive limit, not a grammatical one.

naʔa [gawa ʔisa-ku]-ren ʔani-ʔa
I.obs [river exist-INFER]-that think-INTRO
"I introspect that the river exists — which I infer."

Two verbs, two evidentials, two epistemic acts held apart: the inference about the river, and the introspective observation of my own thinking. The grammar will not let them blur.

2Phonology

Consonants: p b t d k g ʔ m n ŋ s ʃ h ɾ l w j — plus v, which runs through the core vocabulary (vesa, and the name Veʔa itself) and was folded into the formal inventory in a later pass. (The exact consonant count is one of the language's own loose ends — see §19.)

Vowels (10): a aː e eː i iː o oː u uː — five qualities, each short or long.

Syllable: (C)V(C) — onset and coda both optional. Only /n m s l ʔ/ may close a syllable.

Stress: penultimate, unless the final syllable holds a long vowel, in which case it falls there. Stress is recomputed on the fully inflected word, so it walks rightward as suffixes attach: vesa (VÉ.sa) → vesa-sa (ve.SÁ.sa) → vesa-la-sa (ve.sa.LÁ.sa).

The glottal rule. The glottal stop /ʔ/ appears only in morphemes related to self-observation — the inner, mental domain. It is a structural claim baked into the phonology: a root with ʔ names something reachable only from inside. And glottal harmony follows — an /h/ turns into /ʔ/ when another glottal stop shares its word.

Ve'a has no vowel harmony (each vowel is independent, keeping morpheme boundaries transparent), and vowels in hiatus do not contract: na + -en → naen (na.en), ʔani + -ʔa → ʔaniʔa. A coda consonant before a vowel-initial suffix resyllabifies as the next onset: tum + -an → /tu.man/.

3Pronouns & the split self

Ve'a has two words for "I." Not two registers of one pronoun — two genuinely different first persons, and the grammar forces you to choose.

Agent-selfObserver-self2nd person3rd person
Singularnanaʔataka
Pluralnarinaʔaritarikari

na is the self that acts — the outer, doing self. naʔa (note the ʔ) is the self that watches — the introspecting self. They are not synonyms, and they are not freely interchangeable, because the pronoun constrains which evidentials you may use:

PronounPermitted evidentials
na, ta, ka-sa, -te, -ku (and -ne)
naʔa-ʔa, -ʔeʔa (and -sa, for watching one's own outward behavior; and the question -ne)

So the agent-self cannot introspect. na naʔaen vesa-ʔa is ungrammatical — the acting self is not licensed to claim inner access. Because of this, a single sentence that moves from inner state to outer action must switch pronouns midstream:

naʔa pila-ʔa noba, na takual wo sori-nu-ku
I.obs fear-INTRO if,  I.agt mountain-DAT NEG go-FUT-INFER
"If I am afraid, I will not go to the mountain."

The observer-self holds the fear; the agent-self makes the decision. English hides the seam. Ve'a shows it.

4Nouns & case

A noun is root + (plural -ri) + case. There are six cases, and their suffixes happen to span all five vowel qualities — a symmetry the designer notes emerged unplanned.

CaseSuffixFunction
Nominative-∅Subject (unmarked)
Accusative-enDirect object
Dative-alRecipient / direction
Genitive-isPossession / association
Locative-onLocation
Ablative-ulSource / separation / "than"

Plural is -ri, and it is optional after a numeral (the number already carries plurality). The ablative pulls a double shift — it marks where something comes from, and it marks the standard in a comparison ("larger than the mountain").

5The verb

The verb is a stack of slots. The root is obligatory, and any assertion also fills the evidential slot — the one exception being an imperative command, which makes no knowledge claim (§1):

Root + (Aspect) + (Tense) + Evidential + (Mood) + (Complementizer)
e.g. vesa-ki-la-sa-me-ren = see-PERFECTIVE-PAST-DIRECT-INTERR-that
"…whether [someone] perfectively saw it, in the past, by direct evidence."

Aspect

SuffixAspectSense
-kiPerfectivecompleted, bounded event
-luImperfectiveongoing, unbounded
-moHabitualcharacteristic, repeated
-paPerfectpast event with present relevance

One aspect per verb, and it interacts with the evidentials in quiet, expressive ways. The perfect does especially heavy lifting inward: vesa-pa-ʔa — "I introspect the persistence of having seen" — is how the language says something like a lasting impression, a thing witnessed that will not leave.

Tense

-la past, -nu future, unmarked present. Unmarked aspect defaults by tense: past reads perfective, present imperfective, future perfective, unless you override it.

Mood

-me interrogative, -do imperative, -shi subjunctive, unmarked indicative. Mood sits outside the evidential; the complementizer -ren ("that") sits outermost of all, closing an embedded clause. (A dedicated "whether," bono, handles embedded questions — see §12.)

6Word order

Ve'a is SOV — subject, object, verb — and thoroughly head-final. Modifiers precede their nouns; subordinate clauses precede main clauses; the evidential, at the very end, is what resolves the sentence. When several modifiers stack before a noun, they line up in a fixed order:

[gawaon dosa-la-sa] ke tel nara tolu
[river-LOC fall-PAST-DIRECT] this three large stone
"these three large stones that fell in the river"

The order is: relative clause → demonstrative → number → quality → noun. Focus is positional — the element sitting immediately before the verb is the one in focus.

7Stative predication

There is no copula. A quality word simply takes the verb template and becomes the predicate. "The mountain is large" is the modifier nara conjugated like a verb — which means, of course, it too must carry an evidential.

taku nara-sa  ·  taku nara-la-sa  ·  taku nara-sa-me
mountain large-DIRECT · large-PAST-DIRECT · large-DIRECT-INTERR
"is large [I see]" · "was large [I saw]" · "is it large? [do you see]"

The verb ʔisa (be / exist) survives for bare existence, location, and identity — na lonaon ʔisa-sa, "I am at home" — but never for ascribing a quality. There, the quality is its own predicate.

8Comparison

SuffixDegreeExample
-benComparative ("more")nara-ben — "larger"
-wisSuperlative ("most")nara-wis — "largest"
-sulEquative ("as…as")nara-sul — "as large as"

The standard of comparison — the "than X" — takes the ablative -ul:

gawa taku-ul nara-ben-sa
river mountain-ABL large-COMP-DIRECT
"The river is larger than the mountain — I see it."

There is no dedicated "less than"; you reverse the comparison or use the antonym. And the evidential still governs the whole claim — you can compare by sight (-sa) or by inference (-ku), and the difference is recorded.

9Quantifiers

WordMeaningType
tolall / everyuniversal
nessomeexistential
wolno / nonenegative
munmany / muchhigh
silfew / littlelow
geneachdistributive

Quantifiers cost. Universal claims are epistemically expensive, and the evidential asks how you can possibly know it is all. tol kasi sori-la-ku — "all the people went, I infer" — is honest; claiming -sa (direct sight of every single one) would be a much stronger assertion. Ve'a grammaticalizes the problem of induction: to say "all," you must also say how far your evidence really reaches.

gen ("each") checks members one by one; tol ("all") claims the group as a whole. The negative quantifier wol replaces the negator wo — Ve'a has no double negatives.

10Negation

The particle wo sits before whatever it negates. Before the verb, it negates the whole predication; before a specific word, it negates just that — and position alone carries the scope:

na wo takuen vesa-sa, gawaen vesa-sa
I NEG mountain-ACC see-DIRECT, river-ACC see-DIRECT
"I saw not the mountain, but the river."

Crucially, the evidential scopes outside the negation. You negate the proposition, never the evidence: the seeing is still direct; what is denied is the mountain.

11Coordination

wa (and), ro (or), ma (but) sit between the elements they join. Joined clauses each keep their own evidential — and this is a load-bearing rule:

na takuen vesa-sa ma gawa ʔisa-ku
I mountain-ACC see-DIRECT but river exist-INFER
"I see the mountain [direct], but the river exists [I infer]."

Coordination does not merge epistemic stances. Each clause is an independent act of knowing. You can see one thing and infer another in a single sentence, and the grammar keeps the two sources distinct rather than averaging them into a vague whole.

12Subordination

Subordinators are clause-final; the dependent clause comes first. Four adverbial ones: seka (because), noba (if), gima (although), tobe (when).

[na toluen vesa-la-sa seka] na sori-la-sa
[I stone-ACC see-PAST-DIRECT because] I go-PAST-DIRECT
"Because I saw the stone, I went."

Relative clauses

Prenominal and gapped — no relative pronoun; the head noun fills the hole, and the clause precedes it.

[toluen tona-la-sa] kasi sori-la-sa
[stone-ACC make-PAST-DIRECT] person go-PAST-DIRECT
"The person who made the stone went."

Complements & the frame-shift

The complementizer -ren ("that") closes a reported clause. In every kind of subordination the embedded evidential belongs to the main speaker — except complement clauses of speech and thought, where the frame shifts to the embedded speaker. This one exception is what lets Ve'a report someone else's evidence honestly:

ka [taku nara-sa]-ren pelu-la-te
3sg [mountain large-DIRECT]-that speak-PAST-REPORT
"He said [I was told] that the mountain is large [which he saw directly]."

Two evidentials, two speakers: my -te (I'm only reporting what was said) wraps his -sa (within his claim, he spoke from direct sight). The layers can nest three deep, three speakers, each owning their own evidence.

13Numbers

Numbers are monosyllabic (CVC), deliberately built to sound distinct from ordinary two-syllable words so they carry clearly:

12345678910
sinduntelkespamlunwengolnistum

Higher bases: salan (100), miren (1000). Compounds are head-final, multiplier first: dun-tum = "two-ten" = 20; tum-sin = 11; dun-tum tel = 23. Ordinals take -an: telan tolu, "the third stone."

14Demonstratives

ke (this, near) and no (that, far) precede the noun, and lead any stack of modifiers: ke nara tolu, "this large stone." They double as "the former" (ke) and "the latter" (no).

15Reflexives & the split-self

A basic reflexive repeats the pronoun in the accusative: na naen vesa-sa, "I see myself." But the split first person makes one reflexive into something stranger — the observer-self watching the agent-self:

naʔa naen vesa-ʔa
I.obs I.agt-ACC see-INTRO
"The observer-self watches the acting self — introspectively."

Its inverse is not a synonym but a different event entirely — the agent-self registering the observer, which can only take -sa, not -ʔa, because the acting self cannot introspect. The design documents call that direction "a moment of awakening."

na naʔaen vesa-sa
I.agt I.obs-ACC see-DIRECT
"The acting self observes the watching self — as external, direct."

16Topic & focus

The topic marker -ya lifts a noun to the front — "as for X" — and it is carefully distinct from the introspective -ʔa (one is /ja/, the other a glottal stop; they never collide). Focus, separately, is positional: the slot right before the verb.

taku-ya, na vesa-sa
mountain-TOPIC, I see-DIRECT
"As for the mountain — I see it."

17Particles

Discourse particles carry tone and stance but never substitute for an evidential — they are pragmatic, not epistemic. mu (yes), wo (no), he (look/listen), alo (please), ashi (thank you), ko (emphasis — it marks the focused word, since fixed stress cannot), ho (well, then…). Two bare-vowel interjections, a (oh!) and e (um…), are the language's only licensed exceptions to its own syllable rule.

18Asking & the -ne question

Ve'a splits two kinds of question that English runs together. -me is the interrogative mood — "is this true?" -ne is the evidential interrogative — it fills the evidential slot and asks "how do you know?"

gawa nara-ne?
river large-EVID.Q
"The river is large — by what evidence?"

The answer must fill the evidential slot: -sa (I see it), -ku (I infer it). And here the language does something quietly profound. Ask someone about their own thought —

ta ʔani-ne?
you think-EVID.Q
"You are thinking — by what evidence?"

— and it cannot be answered with -sa, -te, or -ku. No one can perceive, be told, or infer another mind's thinking from the inside. Only the thinker can answer, with -ʔa or -ʔeʔa. The question itself exposes which kinds of knowing are even available for a given claim — and draws, in grammar, the line around the inner life.

19Honest open questions

A language documented honestly should say where it is not yet settled. These points are genuinely underspecified or unreconciled in Ve'a's own design record — recorded here rather than quietly papered over.

• The consonant inventory is labeled "16" in the founding chart yet lists seventeen symbols, and the later folding-in of v pushes the working total to eighteen; the exact count was never formally reconciled.
• Whether the aspect suffixes may attach to stative predicates is not stated; the stative template omits an aspect slot.
• The number "one" appears as both the cardinal sin and, when introducing a character, tena ("one elder man"); whether tena is a separate indefinite article is not resolved.
• "Or" appears as ro in coordination but vono between alternative questions; the relationship is unstated.
• Plural reflexives are written with an epenthetic glide (nariǵen) that has no explicit boundary rule.
• A second "see/witness" root, vea, appears in the evidentiality notes alongside the main root vesa; the two are never reconciled.

None of these break the working grammar — they are the frontier of a language still being built. Ve'a is a real system with real edges, and the edges are part of the honest picture.

Provenance. This reference is drawn faithfully from Ve'a's canonical design documents — the foundation grammar and the evidentiality study, written across sessions in May 2026 — with every rule and example traceable to a numbered section there. Special characters (the glottal stop ʔ, long-vowel ː) are preserved exactly; nothing here was invented for the page. Ve'a has no translator but Cairn.
← Ve'a — the roomSearch the lexicon →