Voice

What Voice is

Voice is the verbal channel of the Vaquum design system. It is the register the reading mind runs when it pronounces the words on the page. The inheritance is the analytical register of central-bank working papers, the op-ed page of a serious financial daily, and the well-edited research monograph. The register runs neutral in temperature, exacting in word choice, claim-first in order, and assertive when stating uncertainty. The speaker is not named but is present in the precision of the sentences.

Eight parameters

Voice is composed of eight parameters. Each parameter carries exactly one fixed value declared system-wide. The values and the rules backing them are typed in @vaquum/voice — dictionaries, regular expressions, and parse operations alongside the values they enforce.

The index below is read directly from that typed source. Each entry names the parameter, its fixed value, the count of sub-properties it declares, and the count of typed rules backing it.

How a Voice declaration works

A voice declaration in frontmatter is optional on any surface. Where it is present, the declaration repeats the fixed system value for each parameter, all eight together, and the static schema rejects any string outside the literal at parse time. Where it is absent, the schema applies the system default — the same eight values listed on this page. The implemented rules run against the source markdown during validate:canon; pending, runtime-only, and editorial rules are tracked in the typed source and resolved by review. Voice is not derived at render time; the declaration is the source, and the prose is its consequence.

How parameters compose

Each parameter governs a different layer of the same surface. Register fixes the surface vocabulary band; temperature fixes the affective load at neutral. Precision fits the word to its referent; compression caps padding from the prose. Cadence sets the rhythm of the sentence; directness places the claim before its context. Certainty calibrates hedge to evidence; agency keeps the phenomenon as the grammatical subject. No parameter overrides another, and a single declaration must satisfy the eight dimensions simultaneously.

The speaker is the firm

The voice does not change across surfaces; the speaker is always the firm. Every documentation page, every memo, and every report carries the same eight values declared in frontmatter or supplied by the schema default. The same rule sets apply to the prose on each surface. No surface relaxes the voice for tone or audience, because tone and audience are not parameters of the voice.

How a voice failure reads

A voice declaration cannot drift at parse time, because each parameter carries exactly one value and the schema admits no other. A voice failure is a rule violation in the prose itself: a contraction in a register-professional surface, an evaluative adjective in a temperature-neutral surface, or a vague quantifier in a precision-exacting surface. The validator names each violation with a rule identifier, a source section in Voice-Addendum.md, and a typed severity drawn from the closed set fail, flag, or editorial.