Nychthemeron

Kar'aqa names

The tongue of the ravenmen · nine tones, nineteen consonants, four vowels

Generated by tools/karaqa.py and tools/build_karaqa.py, seed 7001. Do not hand-edit this page — edit the builder and run it again.

The attested material is the author’s and is never regenerated. Everything below the attested corpus is a proposal, and is marked as one.

I. The source, and a correction

The tongue has a document of its own, and has had all along: Eventide (Original)\Eventide\Linguistics\Miscellaneous (Ancient, esoteric or alien)\Kar_aqa.docx, 135 words, indexed in the Salvage Index as “the tone system of the Kar’aqa, governing over the Iteration III duplicate.” A duplicate sits at Eventide (New)\Iteration III\Linguistics.docx, which the same index lists as safe to delete, and a summary in Compendium\06 — Tongues and Scripts.

Two things the wiki had wrong until now. The working notes recorded that no Kar’aqa language existed at all, which was false; and Kar’aqa described the tones as “flat at three levels, rising, falling, peaking and dipping”, which reads as seven. There are nine.

II. The nine tones

Every one is carried by a combining diacritic over a, e, i or o.

ContourMarkOn aAttested in
Flat, top˥Combining Hook Aboveảkʼảh, yỉk
Flat, mid˧Combining Macronā— cited in the table only
Flat, bottom˩Combining Breveăᵵĕʼ
Rising, low→high˨˦Combining Double Acute Accenta̋śa̋k
Rising, mid→top˧˥Combining Acute Accentáqóh
Falling, high→low˦˨Combining Double Grave Accentȁyȁʼ
Falling, mid→bottom˧˩Combining Grave Accentàqèn
Peaking˧˦˩Combining Circumflex Accentâçâ̰h
Dipping˧˨˥Combining Caronǎ— cited in the table only

And one mark the table does not account for. çâ̰h “day” carries a combining tilde below as well as its peaking tone — creaky voice, which the nine-tone table never mentions. It is kept as written and never generated: either a tenth distinction the table omits, or a slip of the pen. Unresolved.

III. The inventory

PlaceLetters
Bilabialp · m · v
Dentalt · n · s
Palatalś · ć — ć = tś
Velark · ng
Uvularq
Glottalʼ · h
Clicksᵵ · ç · x — dental · palatal · lateral
Liquidl · y · r
Vowelsa · e · i · o — and no u

Mind the two conventions. Here x is the lateral click and ç the palatal click, where in the Azadean and Tulkata work x is a velar fricative and tools/scripts.py wants ġ and x in that other sense. Do not carry one convention into the other.

The apostrophe. kʼảh has it after a consonant and ᵵĕʼ after a vowel. The author’s own notation elsewhere uses the apostrophe for a glottal stop between vowels, for a syllable break beside a consonant, and for an ejective at the start of a word — so kʼ is most likely an ejective k rather than k plus a stop. The likeliest reading, and not settled.

Every attested word is CVC and monosyllabic, and every one is closed — not one ends in a bare vowel. Length comes from compounding, which is hyphenated.

IV. The attested corpus — the whole of it

Nine words, three of them particles, and two sentences. The author’s.

Kar’aqaSenseAs a human writes it
kʼảhtreeKʼah
ᵵĕʼrainṬeʼ
śa̋kfireShak
çâ̰hdayCah
qènnowQen
qóhparticle; subject-marking on the evidence of the one clauseQoh
yȁʼverbal particle, continuous presentYaʼ
yỉktowards — directional particleYik
kʼảh-qóh śa̋k-yȁʼthe tree is burning
çâ̰h-qèntoday

What the two sentences give away. Particles follow their head and are hyphenated to it; the order in the one clause is subject–verb; and compounding is noun-noun with a hyphen (çâ̰h-qèn, day-now, “today”).

A pattern worth noticing and not yet a rule. qèn “now” and the particle qóh both begin q-; the two particles yȁʼ and yỉk both begin y-. Two of four, in a corpus of nine. It would be a handsome system if it held — q- for the deictic, y- for the grammatical — and nine forms cannot establish it.

V. Ṭeʼkʼah is “rain-tree” — and Pavasia has the owner wrong

Pavasian names records टेकाह् Ṭekāh as “the Kar’aqa town in the Crow Hills, naturalised from their own name for it,” and Pavasia says the same: “Ṭeʼkʼah is their own name for the place.” Both elements are in the five-word lexicon above:

ᵵĕʼrain
kʼảhtree
ᵵĕʼ-kʼảhrain-treeṬeʼkʼah

It is the only Kar’aqa compound anybody can check, and it checks. The self-test of tools/karaqa.py fails unless the module can rebuild it from its parts and romanise it to exactly that. No previous generator in this project had an external form to validate against.

But the word is northern, and the town is not

The southern ravenmen of the Crow Hills speak a language entirely unrelated to this one — a separate tongue and a separate identity, and they do not call themselves Kar’aqa. Kar’aqa is the northerners' own name for themselves.

So Ṭeʼkʼah parsing cleanly in the northern corpus does not make the southerners northern. It makes the Pavasian record wrong. The name reached Timirāvana through an intermediary and was set down as the locals' own, and the likeliest intermediary is the obvious one: Azade holds the main trade route with the far east, borders Pavasia, and stands on good terms with the southern ravenmen. A word picked up at the eastern end of that road and applied at the southern end.

And the error is three hundred years old and fossilised. Timirāvana wrote the name into the almanacks it has computed with the southerners ever since, so the mistake now sits inside the kingdom’s own calendar. The one “Karʼaqa” word any human power ever wrote down is in the wrong language — and the southern town’s actual name is recorded nowhere at all.

Which leaves the southern tongue a complete blank: not one word of it is attested, and nothing in this module touches it. It wants an inventory of its own, built from nothing, and it is not begun.

VI. What the inventory unmasks

The criteria are graded, and the grading is the point. Nine forms will bear some conclusions and not others — in a corpus of a few dozen forms a gap is not a rule, which was the lesson of the Tulkata work and bites far harder at nine.

Harda positive impossibility — no u in the inventory at all, and no cluster in any attested form, so a name with either is disqualified outright
Softa tendency the corpus shows and cannot prove. Every attested form happens to be closed and to have an onset — but nine forms cannot make an open syllable illegal, only unattested
NameVerdictOn what ground
Rakatudisqualifiedcontains 'u'; the vowels are a, e, i, o only
Arkashirdisqualifiedhas the cluster 'rk'; no attested form has a cluster; begins with a vowel, where all nine attested forms have an onset — likewise a tendency
Karʼaqaflagged onlyends in a bare vowel, where all nine attested forms are closed — a tendency, not a proven ban
Ṭeʼkʼahcleannothing against it

Rakatu and Arkashir are not Kar’aqa names

This is the Svelekebá finding over again. Svele·kebá proved Uzda and not Tulkata, because Tulkata has no voiced stop and kebá has a b. Here Rakatu has a u, which the language does not possess, and Arkashir the cluster rk, which no attested form has — both hard faults, both final. So Voralan’s names for the Kar’aqa capital and for the Kar’aqa highlands are somebody else’s names.

Whose, then? Three candidates, and the first is much the best:

  1. Uzda, the troll tongue, and this is the strong reading. Svelekebá already established that the names Voralan uses for the far east come through the trolls; the Tulkata Empire’s easternmost limit abuts the Highlands; and Rakatu sits comfortably in Uzda’s shape where it cannot sit in Kar’aqa’s.
  2. Do’ari of On-Yezdo, who are the actual neighbours. The difficulty is that the two peoples are at enmity in the east, and Voralan does not read Do’ari.
  3. Azadean for Arkashir, which answers to Ardashir rather too neatly. The weakest of the three on reflection: Arkashir is confirmed as lying in the far east, and the road from Isfaran to a Voraic map is very long for one name to travel. Though Azade’s far-east route is exactly the road §V needs, so the candidate is not useless — only in the wrong place.

And Kar’aqa itself is flagged

The word ends in a bare vowel, and not one of the nine attested forms is open. It is the endonym of the northern people, so it is not in question — but something about the spelling is. Two readings, and nothing yet chooses between them:

  1. The native form is closed and Voralan added the vowel. Most likely a compound on the Ṭeʼkʼah model — something of the order of kʼàr-ʼa̋q, with the apostrophe sitting where the seam is — and the final -a is Voralan’s, added as borrowers do. Then even the northern endonym, as we have it, is a foreign spelling, which would be a tidy third instance after Rakatu and Arkashir.
  2. Open syllables are legal and nine words simply missed them. Then Kar’aqa is native as written, and the closed-syllable pattern is a tendency of the corpus and not a law of the tongue.

The generator holds to closed syllables either way, since that is what the evidence shows; but it no longer calls an open final a fault, only unattested.

VII. The drawn lexicon — proposed

Thirty-two roots, one closed syllable apiece, each clear of the attested corpus by two segments or more. The fields extend the attested senses and go no further: weather, vegetation, fire and time — mountain forest, the sky, and the counting of days, which is what the भुशुण्डि are paid for.

téhstonećʼỉhmistkèkroot
ngímbarkćʼőqmosssêtridge
ngỉhcleftpíʼspringʼa̋ngstar
híhdarkᵵírcoldkìldeep
xi̋thighhòksilencese̋ksmoke
ᵵảngashvȅkboneqʼảngeye
ćʼȍhwingśȉʼfeathermỉʼnest
qỏngflockxǒscountmi̋hyear
kôʼmouthti̋ngwordye̋hdoor
çỏsroadćóksaltćʼèhpaper
nẻksnowxînthunder

VIII. The gazetteer — proposed

Every name is a compound on the Ṭeʼkʼah model: modifier, hyphen, head.

The three kingdoms

The author has settled three principal kingdoms east of the great mountains. An older document counts one — see §X.

Kar’aqaSenseOn a human map
ćʼỉh-sêtmist-ridgeChʼihset
xîn-xi̋tthunder-highXinxit
śȉʼ-téhfeather-stoneShiʼteh

Their seats

The first of these is the city Voralan calls Rakatu.

Kar’aqaSenseOn a human map
śȉʼ-mỉʼfeather-nestShiʼmiʼ
ngím-ye̋hbark-doorNgimyeh
híh-kôʼdark-mouthHihkoʼ

In the Highlands Voralan calls Arkashir

Kar’aqaSenseOn a human map
ćʼȍh-kôʼwing-mouthChʼohkoʼ
se̋k-píʼsmoke-springSekpiʼ
ngím-çỏsbark-roadNgimcos
ćʼȍh-çỏswing-roadChʼohcos
ᵵảng-xi̋tash-highṬangxit
kèk-mỉʼroot-nestKekmiʼ
ćʼỉh-kôʼmist-mouthChʼihkoʼ

On the Do’ari frontier

On-Yezdo, the last true bastion of Do’ari civilisation, lies in the furthest east, and so do the Three Kingdoms. The two scholarly peoples of the world are neighbours, and have been for longer than any human state has existed — so the quarrel on Do’ari is not an abstraction about method but a border, and a border has places on it.

Kar’aqaSenseOn a human map
ti̋ng-ye̋hword-doorTingyeh
ᵵír-ye̋hcold-doorṬiryeh
ᵵír-sêtcold-ridgeṬirset
ti̋ng-téhword-stoneTingteh
ᵵảng-sêtash-ridgeṬangset

And not a single one of these has ever been written down by anybody. The frontier is past the edge of every Voraic map; the only powers who could name it are the two standing on it, and neither publishes.

The Crow Hills are not in this gazetteer

Ṭeʼkʼah and the Crow Hills belong to the southern ravenmen, whose tongue is entirely unrelated to this one and of which not one word is attested — see §V. Coining southern places out of northern roots would repeat precisely the error Timirāvana made three hundred years ago. They wait on a southern inventory, which is not begun.

Those who read, in the north

Not the भुशुण्डि. That is Pavasia’s word, bestowed on the scholars of the Crow Hills after the immortal crow who kept the count of the ages — and those are southerners, of another people and another tongue. What the northern kingdoms call their own readers is unrecorded; these are proposals, and they are built on the senses the corpus actually gives: day, now, the counting of years.

Kar’aqaSenseAs a human writes it
hòk-qʼảngsilence-eyeHokqʼang
çâ̰h-mi̋hday-yearCahmih
çâ̰h-ti̋ngday-wordCahting
hòk-xǒssilence-countHokxos

IX. How it is written down, and by whom

Voralan does not write it at all. The Abecedarium Imperii cannot carry nine tones, three clicks and a uvular stop, so what reaches a Voraic map is the segments alone — the clicks transliterated (ᵵ → ṭ, ç → c, x → x), the palatals spelled out (ś → sh, ć → ch), the apostrophes kept, the tones simply gone. Which is why two Kar’aqa words a Voraic clerk writes the same way may be unrelated, and he will never know it.

And the one power that has had to solve the problem was solving a different language. Pavasian names gives टेकाह् in Brahmi as 𑀝𑁂𑀓𑀸𑀳𑁆, says plainly that it “represents their tones not at all,” and then flags the thing that matters:

“The Kar’aqa tongue is tonal and no Indic script can write it. What Timirāvana actually does about this — a diacritic system, a separate convention, or simply giving up — is undecided, and the kingdom has three hundred years of almanacks that must have solved it.”

That job belongs to the southern tongue and not to this one. Timirāvana’s almanacks are computed with the ravenmen of the Crow Hills, so whatever notation it invented, it invented for their speech. It remains the sharpest small job on the whole subject — three hundred years of rice-calendars in Devanagari or Brahmi, and somebody in Niśāntapura devised a way to write another people’s tones. It would be the only script in the setting made by one people to write another’s speech. But it waits on the southern inventory, and so does everything else southern.

X. What is still wanting

See also

Kar’aqa · Pavasian names · Pavasia · Tulkata names · Do’ari · Races · Dysis