Hermeneia is a voking language for querying what a particular can become under a use.
It is the query language for Vocarium, a trope database of vokeable particulars. Hermeneia does not merely select rows or retrieve records. It calls stored particulars into relation with a purpose, context, use-model, warrant, or challenge.
Hermeneia asks:
Can this particular become an answer here?
What does it call forth?
What happens when it crosses context?
What breaks when it is challenged?
What previous licence must now be withdrawn?A normal query language asks:
SELECT records WHERE condition;Hermeneia asks:
invoke this particular
under this use-model
show what is preserved, lost, licensed, or insufficientHermeneia is therefore not only a retrieval language. It is an imminence language: a language for asking what a stored particular is about to count as, become, or disclose under a declared use.
Hermeneia is organised around voking operations.
To voke is to call a stored particular into a use-relation.
| Operation | Meaning |
| ----------- | -------------------------------------------------------- |
| invoke | call a particular into a declared use |
| evoke | call forth neighbouring or latent particulars |
| convoke | call multiple particulars together |
| transvoke | call a particular across context, bearer, record, or use |
| provoke | call a particular into challenge or stress-test |
| intervoke | call between trope paths or candidates |
| revoke | call back or withdraw a previous licence |
Implemented end-to-end and tested: invoke. The other six are recognised by keyword and refused as unimplemented (their bodies are not parsed) — status table in docs/SPEC.adoc.
Hermeneia does not select rows. Hermeneia vokes tropes.
invoke "authentic language"
under use_model "critical-paraphrase"
show verdict, loss, warrantPossible result:
verdict: p-insufficient
loss:
quality Attenuated(6)
bearer Present
context Present
record Present
reason:
The paraphrase "real language" preserves a loose ordinary association
but loses the theoretical force carried by "authentic".evoke around "authentic language"
within context "de Man / Romantic irony"
limit quality_loss <= Attenuated(3)This asks what nearby tropes, claims, quotations, paraphrases, or warrants are resonant with the original particular without exceeding the chosen loss bound.
convoke [
"authentic language",
"poetry",
"myth",
"madness"
]
under use_model "critical-reading"
show constellationThis brings several particulars into a shared interpretive scene.
transvoke "authentic language"
from context "source quotation"
to context "student paraphrase"
show path, attenuation, failed_floorsThis traces what happens when a trope crosses from one use-context into another.
provoke "real language"
against "authentic language"
under use_model "technical-paraphrase"
show failure_pointsThis stress-tests a candidate substitution or interpretation.
revoke verdict "v_2041"
because warrant "source_check_failed"This withdraws a previously licensed answer when its warrant, evidence, or use-model no longer holds.
Vocabulary track: trope-particularity-workbench → haec → trope-checker (verdict). Store/query track: haec → vocarium → hermeneia → trope-checker (verdict). trope-checker depends on nothing (HC-1); the workbench depends on nothing; vocarium and hermeneia meet at the M2 StoreProvider adapter (vocarium ADR-0002: wire, do not link).
Hermeneia queries Vocarium, the database of vokeable particulars.
Vocarium holds:
tropes
transformation paths
grades
warrants
use-models
verdicts
Hermeneia asks:
what can these particulars become under use?Haec records transformations. Hermeneia queries them.
Haec:
attenuate[2] u
project[{quality, record}] v
fuse[source_quote] x y
Hermeneia:
invoke v under use_model "critical-paraphrase"
transvoke x from source to essay
provoke y against submitted_claimHaec is the transformation language. Hermeneia is the voking language. Vocarium is the store.
Hermeneia separates three things that ordinary query languages often collapse:
retrieved — the database found it
interpretable — it can be meaningfully read under this context
licensed — it is p-sufficient under this use-modelA result may be retrieved but not licensed. It may be interpretable but insufficient. It may be p-sufficient under one use-model and fail under another.
That distinction is the point.
Hermeneia is an experimental query language for trope-aware, loss-aware, warrant-aware databases. It is intended for work on particularity-preserving computation, interpretive databases, AI-assisted scholarship, evidential reasoning, paraphrase discipline, and transformation-aware knowledge systems.