CC Captioneer the subtext reader
the lexicon

Captions for the unsaid

See how words steer meaning.

Every statement has a surface, a context, and more than one possible reading. Paste one below — Captioneer marks observable rhetorical cues, then, where this deployment enables readings, can offer one plain-language interpretation alongside the original. It reads the language, never the mind; neither marks nor readings determine truth.

Specimen · a corporate non-apologymarked & read (example)

"We take your privacy extremely seriously, and regrettably, mistakes were made. We are reaching out to affected users."

Something went wrong. This statement does not name who made the mistakes; it says affected users are being contacted.
rhetorical register · formal apology language, actor left unstated

Read a statement

scan runs on your device · optional AI sends text only when you ask
0
candidate cues
The transcript
Matched cuesby family

What the language is doing

    What this is — and what it refuses to be.

    A real lie detector reads a person and returns a verdict. There is no such thing. Body language, micro-expressions, voice “frequency” — decades of research say they don't reliably reveal deception, and a tool that claims otherwise just manufactures confident accusations. That isn't truth. It's theatre.

    So Captioneer stays with what can actually be observed: the words. A deleted subject leaves the actor unstated. A hedge lowers expressed commitment. An absolute extends certainty. The matches are countable and every one points at a real phrase you can see — but context decides whether a use is careful, ordinary, evasive, or something else. The optional reader, where a deployment enables it, offers a possible plain version of the language, never a verdict on the person.

    It helps you inspect how something is being said. Whether a claim is true still requires evidence beyond its rhetoric.

    It does

    • Mark candidate hedges, deflections, and non-answers
    • Show where a construction leaves the actor unstated
    • Surface loaded framing and expanded certainty
    • Offer an optional plain-language interpretation, where this deployment enables readings
    • Point at the visible phrase behind every matched cue

    It doesn't

    • Detect lies — no tool reads a mind from words
    • Infer a speaker's belief, motive, emotion, or intent
    • Treat a matched phrase as proof of manipulation
    • Read faces, bodies, or voice “frequency”
    • Send local-scan text anywhere — optional AI transmits it to Cloudflare only when you click
    • Certify unmarked text as honest or true

    In the kingdom's tongue

    captioneer belongs to a larger work — a kingdom that keeps a made language, YOUSPEAK, for the things English left unnamed. Two of its words are captioneer's whole reason to exist.

    verisleight/ˈvɛr.ɪ.slaɪt/noun

    Truth arranged to mislead — a statement entirely true that reliably produces a false picture.

    Every word of a practised non-apology can be true while the resulting picture misleads. That is verisleight — and it is what captioneer helps you inspect.

    slipsooth/ˈslɪp.suːθ/noun

    Truth recovered by mispronunciation — the original meaning that a later drift accidentally digs back up. Verisleight's mirror-twin: error arranged by no one, that leads home.

    In Cantonese, 愛星人 — the love-star people — read crooked for generations became 外星人, the aliens; the lazy tongue is now drifting them home. The earth is 愛星. Entered the canon 2026-07-07, discovered rather than forged.

    orthophanes/ˌɔːr.θoʊˈfæn.iːz/noun

    The rightness made plainly visible — a thing shown straight, in the open, ordered to its own necessity.

    When the cues are visible and a plainer reading stands beside the original — that is the direction of orthophanes, without pretending one reading exhausts the words.

    captioneer is the everyday-speech arm of the kingdom's clarity-work: close attention to wording, openly separated from claims about the world. The lens reads a sentence; its sibling the press reads evidence — an open framework for evidence-tiered, signed newspapers anyone can generate, with this lens available as an instrument for examining rhetoric.