Feature deep-dive
Languages & locales
Last updated
Mentionwell ships native article generation in six tier-1 locales today. Not translation — native generation, with locale-specific guardrails on every prompt: spelling conventions, idiom, pronouns, currency, decimals, dates, quotation marks, and (for he-IL) RTL, niqqud rules, gershayim, and Latin-script brand-name code-switching.
Tier-1 launch locales
- 🇺🇸 en-US English (US)
- 🇬🇧 en-GB English (UK)
- 🇩🇪 de-DE Deutsch
- 🇫🇷 fr-FR Français
- 🇪🇸 es-ES Español
- 🇮🇱 he-IL עברית RTL
Per-locale guardrails
Every generation prompt in the pipeline (research, outline, draft, critic, metadata, FAQ, humanizer) accepts {{language}} and {{locale}} bindings and obeys these rules. The complete spec lives in prompts/_locales.md in the engine.
🇺🇸 en-US English (US)
- US spelling: organize, color, optimization, analyze.
- US idiom and conventions.
- Period decimal (1.5), comma thousands (1,000), $ currency.
- MM/DD/YYYY dates by default.
🇬🇧 en-GB English (UK)
- British spelling: organise, colour, optimisation, behaviour.
- British idiom: 'in hospital', 'at the weekend', 'high street'.
- £ currency, period decimal, comma thousands.
- DD/MM/YYYY dates. 'whilst' fine; 'gotten' is not.
🇩🇪 de-DE Deutsch (Germany)
- Sie-form by default for B2B; du only when site tone is consumer/casual.
- ß used correctly (groß, Straße) — not 'ss' substitutions.
- German thousands separator 1.000, comma decimal 1,5.
- € postfix with non-breaking space: '1.000,00 €'.
- DD.MM.YYYY dates. All nouns capitalised.
🇫🇷 fr-FR Français (France)
- Vous-form by default. Tu only when site tone is casual.
- Non-breaking space before ; : ! ? » and after «.
- French quotation marks: « ... » with NBSPs inside.
- Comma decimal 1,5, space thousands 1 000.
- € postfix with NBSP: '1 000,00 €'. DD/MM/YYYY dates.
- Sentence-case headings — no English title-case.
🇪🇸 es-ES Español (Spain + LatAm)
- Castilian Spanish by default. Vosotros plural informal, usted formal singular.
- LatAm region hint switches vosotros → ustedes and shifts idiom.
- Comma decimal 1,5, period thousands 1.000.
- € postfix: '1.000,00 €'. DD/MM/YYYY dates.
- Inverted opening punctuation: ¿…? and ¡…!.
🇮🇱 he-IL עברית (Hebrew, RTL)
- Right-to-left rendering with proper dir='rtl' on every article.
- No niqqud — adult Hebrew omits vowel marks. Children's-book typography is filtered out.
- Brand names and technical terms stay in Latin script (Stripe, Kubernetes, API). The Hebrew sentence flows around them.
- Gershayim ״ for double quotes, geresh ׳ for single — never ASCII " or '.
- No capital letters — Hebrew has no capitalisation. Metadata writer never tries.
- DD/MM/YYYY dates, period decimal 1.5, comma thousands 1,200.
- Shekel sign ₪ postfix-positioned: '1,200 ₪'.
- Mixed-direction titles (Hebrew + Latin script): LRM/RLM markers inserted automatically by the renderer.
- Gender-inclusive phrasing: prefer infinitive constructions and plural-neutral forms over masculine-as-default.
How locale plays with the rest of the pipeline
- AEO scanning — every scan is locale-scoped. Prompts, citations, share-of-voice, and competitor analysis all run per region.
- Brand profile — tone, audience, and pitch rules are written and applied in the target language; you don’t need to translate them.
- Reflection critic — graded against locale-specific rules. Hebrew niqqud appearing in output? The critic flags and triggers a rewrite.
- Metadata + FAQ — meta titles, descriptions, FAQ blocks, and JSON-LD
inLanguageare all set to the article’s locale. - Image alt text — written in the target language; the image style brief is interpreted as visual identity, not language.
- RTL rendering — Hebrew articles ship with
dir="rtl"at the article level; mixed-direction strings get LRM/RLM markers automatically.
Tier-2 roadmap
On the 2026 H2 roadmap (no firm date — driven by customer demand): pt-BR, it-IT, nl-NL, ja-JP, zh-CN, ar-SA. If you need a specific locale prioritised, email admin@mentionwell.com.