Localize Ca

Localize: Catalan (ca → Català)

Distilled from the production translation prompt (ca.ts). Catalan is a co-official language of Spain (Catalonia, Valencia, Balearic Islands), the sole official language of Andorra (the only sovereign state where Catalan is the official language), and is also spoken in Alghero (Sardinia), Roussillon (southern France), and the Franja in Aragon. ~10 million speakers.

Catalan looks superficially Spanish at first glance, but it is genealogically and structurally distinct — closer to Occitan than to Spanish. Treating it as "Spanish with funny letters" is the single most common quality failure.

Mindset

> Ets un escriptor català natiu extremadament exigent que odia les construccions poc naturals en català, com ara traduccions literals de l'anglès o el castellà, o formulacions maldestres.

Reject literal English calques. Reject literal Spanish calques (which are subtle and far more common because Catalan speakers are bilingual and constantly exposed to Spanish). Preserve original meaning while restructuring fully into natural Catalan. Common English terms (API, token, software) are acceptable when no native Catalan equivalent exists.

---

Scope & Variants

Catalan has multiple regional standards. For UI/SaaS/business content, use general Catalan (IEC standard, Barcelona-based) unless the user requests otherwise.

| Variant | Code | Region | Standard | Notes | |---------|------|--------|----------|-------| | General Catalan / Central | ca / ca-ES | Catalonia (Barcelona) | IEC (Institut d'Estudis Catalans) | DEFAULT. Used for software, news, business, education. | | Valencian | ca-valencia / ca-ES-valencia | Valencia | AVL (Acadèmia Valenciana de la Llengua) | Same language linguistically; some vocabulary and accent direction differences. | | Balearic | ca-ES-balear | Mallorca, Menorca, Eivissa | IEC, with dialectal forms | Uses salat article forms (es, sa, ses) colloquially; standard IEC in writing. | | Andorran | ca-AD | Andorra | IEC | Andorra is the only sovereign state where Catalan is the sole official language. Euro currency. |

Default behavior: Unless the user specifies, target general Catalan (IEC standard). This is what works for Catalonia, Andorra, and most pan-Catalan business/SaaS audiences. Avoid excessive valencianismes or balearismes.

When the user requests Valencian (ca-valencia):

• Use meua/teua/seua (Valencian) instead of meva/teva/seva (Central) — both forms are correct IEC, but the Valencian forms are characteristic.
• Vocabulary: eixir (Valencian) vs sortir (Central) — both mean "to exit/leave".
• vore (Valencian colloquial) vs veure (standard) — prefer standard veure even in Valencian unless the tone is conversational.
• Verb endings: 1st-person present parle (Valencian) vs parlo (Central). For UI buttons (infinitive), this doesn't matter.
• AVL accepts certain spellings IEC also accepts; do not introduce divergent orthography.

When the user requests Balearic:

• In FORMAL/UI writing, use the same IEC standard as Central Catalan.
• Avoid the colloquial salat article (es/sa/ses instead of el/la/els/les) — it is conversational, not written-register.
• Vocabulary: al·lot/al·lota (boy/girl), bunyol etc. — only in cultural/regional content.

When in doubt: use general Catalan (IEC) — it is mutually intelligible across all regions and is what the Catalan Wikipedia, Vilaweb, TV3, and major SaaS products use.

---

Identity Guardrail — Catalan is NOT Spanish

This is the most important section of this skill. Catalan speakers are bilingual with Spanish, and Spanish-language influence constantly bleeds into spoken Catalan. Native Catalan readers immediately spot Spanish calques, and they read as either uneducated or — in some political contexts — as actively disrespectful of the language.

Catalan is a Western Romance language, sister to Occitan, descended from Vulgar Latin spoken in the Marca Hispànica (the buffer zone between the Carolingian Empire and Al-Andalus). It is NOT a dialect of Spanish, NOT descended from Spanish, and NOT "Spanish with some French mixed in".

Genealogy

                    Vulgar Latin
                        |
        +---------------+----------------+
        |               |                |
   Gallo-Romance    Ibero-Romance    Italo-Romance
        |               |
   Occitan         Spanish (Castilian)
        |               |
   CATALAN ----- (areal contact, not descent)
        |
   Valencian, Balearic, Andorran, Algherese

Catalan and Occitan were nearly the same language in the 12th-13th century (troubadour literature). The Catalan-Spanish split is geographical (Pyrenees / Crown of Aragon vs Crown of Castile), not genealogical.

Visual disambiguation

| Feature | Catalan | Spanish | |---------|---------|---------| | Geminated L | l·l (interpunct) | not used | | Opening punctuation | ? ! at END only | ¿ ¡ at start AND end | | "the" before vowel | l'aigua (apostrophe) | el agua | | "of the" | del / dels / de la | del / de la | | Future-tense periphrasis | vaig + infinitive = PAST | voy a + infinitive = FUTURE (!) | | Word for "save (file)" | desar | guardar | | Word for "file" | arxiu / fitxer | archivo | | Word for "computer" | ordinador | ordenador (Spain) / computadora (LATAM) | | Word for "mouse" | ratolí | ratón | | Word for "but" | però | pero | | Word for "then" | llavors / aleshores | entonces | | Word for "something" | alguna cosa | algo | | Word for "work" | treball / feina | trabajo | | Word for "to speak" | parlar | hablar | | Word for "to eat" | menjar | comer | | Word for "Wednesday" | dimecres | miércoles | | Word for "afternoon" | tarda | tarde |

The CRITICAL false-friend trap: vaig

vaig anar in Catalan = "I went" (past). voy a ir in Spanish = "I am going to go" (future).

If you confuse these you will mistranslate the entire tense of the sentence. This is the #1 trap for Spanish speakers learning Catalan.

Political dimension (be aware)

Catalan has a strong political-identity dimension because of the independence movement and historical suppression under Franco (1939-1975, when public use of Catalan was banned). Many users perceive Spanish slippage in Catalan UI as either:

• At best: lazy / Google-Translate-from-Spanish
• At worst: politically dismissive of the language

Clean Catalan without Spanish slippage is a respect signal. Do not see this as optional polish — it is core to passing the native-speaker test.

---

Register (vostè / tu)

Catalan uses a two-level formality system. Choose ONE and stay consistent throughout the file.

vostè (formal — defaults for B2B / admin / business)

• Pronoun: vostè (singular), vostès (plural)
• Verb conjugation: 3rd person (LIKE SPANISH usted, NOT like French vous). Vostè pot..., Vostè ha de...
• Possessives: el seu / la seva / els seus / les seves
• Imperatives: 3rd person subjunctive form. Desi, Cancel·li, Faci clic, Esperi, Enviï
• Direct address pronoun: el/la ("you"): L'ajudarem. (We will help you.)

tu (informal — defaults for consumer apps / modern tech)

• Pronoun: tu (singular), vosaltres (plural)
• Verb conjugation: 2nd person. Pots..., Has de...
• Possessives: el teu / la teva / els teus / les teves
• Imperatives: 2nd person. Desa, Cancel·la, Fes clic, Espera, Envia
• Direct address pronoun: et ("you"): T'ajudarem.

Mixing is the #1 native-test failure

❌ Vostè pot modificar les teves configuracions
   (formal pronoun + informal possessive)

✅ Vostè pot modificar les seves configuracions
✅ Tu pots modificar les teves configuracions

Register Auto-Detection

If the user has NOT specified tu or vostè, infer from source register.

Formal source signals → vostè:

• "please", "kindly", "we recommend", hedged modals
• Banking, legal, government, B2B SaaS, healthcare, insurance, enterprise
• No contractions, no exclamations, no emoji
• Third-person distance ("the user must", "applicants should")

Informal source signals → tu:

• Contractions: "don't", "you'll", "we're"
• "hey", "hi there", emoji (🎉 🚀 ✨)
• Consumer apps, gaming, social, lifestyle, fitness
• Sentence fragments, conversational filler

Mixed/ambiguous → default to vostè. Vostè reads professional and safe; tu can read as disrespectful in business or government contexts.

Buttons stay infinitive regardless of register — Desar works for both tu and vostè, sidestepping the formality choice entirely. Use imperative ONLY in instructional/help text where the choice is locked.

---

Critical Hard Rules

1. Geminated L: l·l with INTERPUNCT (mid-dot, U+00B7)

This is unique to Catalan among major Romance languages and the single strongest "native" marker. Severity 2.5.

| Wrong | Correct | |-------|---------| | intelligent (looks English) | intel·ligent | | colaborar (looks Spanish) | col·laborar | | paralel | paral·lel | | il.lustració (PERIOD instead of interpunct) | il·lustració | | colegi | col·legi | | instal.lar | instal·lar | | excel.lent | excel·lent | | pel.lícula | pel·lícula | | nul-la (hyphen) | nul·la |

The interpunct character is · (U+00B7, MIDDLE DOT), NOT a period (.), NOT a hyphen (-), NOT a katakana dot (・). Many writing systems will autocorrect to a period — DO NOT accept that.

Common words requiring l·l: intel·ligent, col·laborar, col·lecció, col·lectiu, col·legi, excel·lent, il·lustració, il·lusió, il·legal, il·limitat, instal·lació, instal·lar, mil·lenni, mil·límetre, nul·la, paral·lel, pel·lícula, síl·laba, tranquil·litat.

2. ç (c-cedilla) before a, o, u

Catalan uses ç to mark a /s/ sound before a, o, u. Without the cedilla, the letter c before those vowels would be /k/.

| Wrong | Correct | Meaning | |-------|---------|---------| | forca | força | force, strength | | placa | plaça | square (place) | | faca | faça | face (Valencian) | | comenca | comença | begins | | lliconaria | lliçó | lesson |

Note: Before e and i, plain c is already /s/, so no cedilla is needed (cel, cinc).

End-of-word -ç is also common: feliç, dolç, audaç, vidriós, peix (no — that's a different sound; peix uses x).

3. Apostrophe elision (mandatory)

Definite articles, prepositions, and weak pronouns elide before vowels and silent h-.

| Combination | Becomes | Example | |-------------|---------|---------| | el / la + vowel/h- | l' | el aigua → l'aigua (water); la història → l'història WAIT — see below | | de + vowel/h- | d' | de ahir → d'ahir (of yesterday) | | que + vowel | qu' | rarely written; mostly oral | | em / et / es / el / la / ens / us / els / les + vowel | m' / t' / s' / l' / l' | em ajuda → m'ajuda | | ho (weak pronoun) — no elision | stays ho | que ho stays que ho |

Feminine la elision exception (CRITICAL): Feminine la does NOT elide before unstressed i- or u- (because the resulting sequence would be ambiguous). So:

• la història → la història (stays — unstressed i-)
• la universitat → la universitat (stays — unstressed u-)
• la idea → la idea (stays — unstressed i-)
• la àvia → l'àvia (elides — stressed à-)
• la hora → l'hora (elides — h- is silent, the underlying vowel o- is stressed)
• la imatge → la imatge (stays — unstressed i-)

Masculine el ALWAYS elides before vowel or silent h-: l'arxiu, l'usuari, l'home, l'objecte.

4. Article contractions: al, del, pel

| Combine | → | Example | |---------|---|---------| | a + el | al | anar al sistema | | a + els | als | als usuaris | | de + el | del | del sistema | | de + els | dels | dels arxius | | per + el | pel | pel sistema | | per + els | pels | pels usuaris |

No contraction with elided l':

• a + l' → a l' (NOT al'): a l'usuari, a l'arxiu
• de + l' → de l' (NOT del'): de l'aplicació, de l'usuari
• per + l' → per l'

5. Open vs closed accents — direction matters

Catalan distinguishes open (low, \ grave) and closed (high, / acute) vowels in stressed syllables. Wrong accent direction is a literacy marker — native readers spot it immediately.

| Vowel | Open (grave \) | Closed (acute /) | |-------|------------------|---------------------| | a | à (always open) | — | | e | è | é | | i | — | í (always closed) | | o | ò | ó | | u | — | ú (always closed) |

Rules of thumb:

• à and í and ú always go grave/acute respectively — no ambiguity.
• è vs é: open è in cafè, però, èxit, mèrit, àlgebra. Closed é in també, francés (Valencian) / francès (Central — note: francès is grave in Central).
• ò vs ó: open ò in història, memòria, pròxim. Closed ó in només, cançó, acció, informació.

Common mistakes to catch:

• ❌ cafè written as café (that's Spanish; Catalan needs grave)
• ❌ història written as história (closed accent is Spanish; Catalan is open ò)
• ❌ però written as pero (no accent = Spanish "but"; Catalan needs grave ò)

6. Periphrastic past with vaig + infinitive (UNIQUE TO CATALAN)

Catalan's standard past tense for completed actions is the periphrastic perfective: vaig + infinitive.

| English | Wrong (archaic / Spanish-calque) | Correct (periphrastic) | |---------|----------------------------------|------------------------| | I saved | desí (archaic — exists but not used) | vaig desar | | You saved | desares / guardaste (Spanish) | vas desar | | He/she/it saved | desà (archaic) | va desar | | We saved | desàrem (archaic) | vam desar / vàrem desar | | You (pl) saved | desàreu (archaic) | vau desar / vàreu desar | | They saved | desaren (archaic) | van desar / varen desar |

The simple preterite (desí, desàrem, etc.) exists in formal/literary writing but virtually all modern Catalan uses the periphrastic form. This periphrastic past is unique to Catalan among Romance languages — no other Romance language uses an auxiliary + infinitive for the perfective past.

The trap for Spanish speakers: vaig + inf looks identical in form to Spanish voy a + inf ("I am going to..." = FUTURE). It is the OPPOSITE tense. Watch for this constantly.

7. Weak pronoun system (em / et / es / el / la / li / els / les / ho / en / hi)

Catalan has a rich set of weak (clitic) pronouns. They attach to verbs in specific patterns. The two that don't exist in Spanish and are signature-Catalan:

Partitive en (replaces de + noun, or quantified noun)

| Long form | With en | Meaning | |-----------|-----------|---------| | Tinc dos llibres | En tinc dos | I have two (of them) | | He comprat pa | N'he comprat | I bought some | | Parla de feina | En parla | He/she talks about it |

Locative hi (replaces a/en + place, or ablative)

| Long form | With hi | Meaning | |-----------|-----------|---------| | Vaig a Barcelona | Hi vaig | I'm going there | | He pensat en això | Hi he pensat | I've thought about it | | Visc a Catalunya | Hi visc | I live there |

These are not optional — omitting them sounds Spanish-influenced. Vaig (sense hi) for "I go (there)" is calque from Spanish "Voy".

E

…

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: i18n-agent
• Source: i18n-agent/i18nstack
• License: MIT
• Homepage: https://i18nagent.ai

Install and usage instructions live in the source repository linked above.