content-locale-humanize-th · git:20260705.2fd4f06 · 2026-07-05 · sha256 f2c52dd71de09985
content-locale-humanize-th git:20260705.2fd4f06A
Immutable. This exact content is served forever at /api/v1/blob/f2c52dd71de09985.
--- name: content-locale-humanize-th title: "Thai Native-Fluency Calibration" description: "Per-language calibration anchors for detecting AI-slop and translationese in Thai (th) target text — auto-loads alongside content-locale-humanize when the target language is Thai." license: Apache-2.0 compatibility: "Extends content-locale-humanize's AI-Slop & Translationese dimension. Load both together — this file has no rubric of its own." domains: content rules: - content(thai) - match(\bth-TH\b) - match(\bThai\b) - match(ภาษาไทย) - semantic(check if this Thai translation sounds native) - semantic(eliminate translationese from this Thai text) --- ## Overview Sourced calibration anchors for Thai, feeding `content-locale-humanize`'s AI-Slop & Translationese dimension. This is a calibration aid, not the checklist — reason natively beyond it (see the core skill's "why structure, not word lists" section). ## Instructions Sourcing confidence: weakest of any language in this set. A targeted search turned up Thai-specific AI-detection tooling but no independently-documented word-level or phrase-level AI tells — don't invent a plausible-sounding list to fill the gap. Lean entirely on the structural checks in `content-locale-humanize` for Thai: source-syntax interference (Thai has no articles, no verb tense inflection, and a different clause-topicalization pattern than English — literal English sentence structure carried into Thai is unusually visible), sentence-rhythm clustering, connector-pileup, and paragraph-opener templating. Thai's lack of spaces between words and different clause-boundary conventions also make burstiness/rhythm judgments harder to make by eye — read for natural clause-break rhythm rather than word-count. If a genuine Thai-specific AI-tell word list becomes available, add it here — this file exists as the attachment point. ## References No solid per-language sourcing found as of this writing; see Overview.