Languages in Exponential #

At a glance, on 3 November 2026. A default installation (the shipped share/locale, share/translations and settings/i18n.ini, no extension) has:

A country is green when a locale ships for it whose language the administration shows translated: with a translation of its own or the first translation of its fallback chain (i18n.ini [TranslationSettings], FallbackMode=first, see section 13). English counts as translated: it is the language of the source texts. The map as a vector image: doc/eng-us/images/languages/coverage-map.svg; outlines: Natural Earth, public domain.

The uninhabited territories without a locale by design are green too: Antarctica (AQ), Bouvet Island (BV) and Heard Island and McDonald Islands (HM). Exponential serves these territories in every language; no local conventions exist.

Made by ./console exp:i18n coverage from the shipped files; ./console exp:i18n coverage --check and the test expI18nCoverageMapTest fail when the figures or the map are out of date.

The one place that says which languages Exponential supports, at which level, and how to add more. The figures were taken on 15 October 2026, those of the administration translations on 26 October 2026 (Exponential 6.0.15).

In short: 390 locale files (383 languages and countries, the test language Lorem Ipsum and 6 variations, 43 of them written right to left) can be chosen as content languages and siteaccess locales; 156 administration translations exist, all of them complete: Afrikaans (afr-ZA), Akan (aka-GH), Albanian (sqi-AL), Amharic (amh-ET), Arabic (Saudi Arabia) (ara-SA), Armenian (hye-AM), Assamese (asm-IN), Asturian (ast-ES), Azerbaijani (aze-AZ), Bambara (bam-ML), Basque (eus-ES), Belarusian (bel-BY), Bengali (Bangladesh) (ben-BD), Bhojpuri (bho-IN), Bodo (brx-IN), Bosnian (bos-BA), Breton (bre-FR), Bulgarian (bul-BG), Burmese (mya-MM), Catalan (cat-ES), Cebuano (ceb-PH), Central Kurdish (ckb-IQ), Chechen (che-RU), Croatian (cro-HR), Czech (cze-CZ), Danish (dan-DK), Dari (Afghanistan) (fas-AF), Dhivehi (div-MV), Dogri (doi-IN), Dutch (Netherlands) (dut-NL), Dzongkha (dzo-BT), English (Australia) (eng-AU), English (United States) (eng-US), Esperanto (epo-EO), Estonian (est-EE), Ewe (ewe-GH), Faroese (fao-FO), Filipino (fil-PH), Finnish (fin-FI), French (France) (fre-FR), Fula (ful-SN), Galician (glg-ES), Ganda (lug-UG), Georgian (kat-GE), German (Germany) (ger-DE), Greek (Hellenic) (ell-GR), Greenlandic (kal-GL), Guarani (grn-PY), Gujarati (guj-IN), Hausa (hau-NG), Hawaiian (haw-US), Hebrew (heb-IL), Hindi (hin-IN), Hulaulá (huy-IL), Hungarian (hun-HU), Icelandic (isl-IS), Igbo (ibo-NG), Imperial Aramaic (arc-IQ), Indonesian (ind-ID), Irish (gle-IE), Italian (Italy) (ita-IT), Japanese (jpn-JP), Javanese (jav-ID), Kabyle (kab-DZ), Kannada (kan-IN), Kashmiri (kas-IN), Kazakh (kaz-KZ), Khmer (khm-KH), Kikuyu (kik-KE), Kinyarwanda (kin-RW), Kirundi (run-BI), Klingon (tlh-QO), Konkani (kok-IN), Korean (kor-KR), Korean (North Korea) (kor-KP), Kurdish (kur-TR), Kyrgyz (kir-KG), Lao (lao-LA), Latin (lat-VA), Latvian (lav-LV), Lingala (lin-CD), Lithuanian (lit-LT), Lower Sorbian (dsb-DE), Luxembourgish (ltz-LU), Macedonian (mkd-MK), Maithili (mai-IN), Malagasy (mlg-MG), Malay (Malaysia) (msa-MY), Malayalam (mal-IN), Maltese (mlt-MT), Manipuri (mni-IN), Māori (mri-NZ), Marathi (mar-IN), Mongolian (mon-MN), Nepali (nep-NP), Nigerian Pidgin (pcm-NG), Northern Sami (sme-NO), Norwegian Bokmål (nor-NO), N’Ko (nqo-GN), Occitan (oci-FR), Odia (ori-IN), Oromo (orm-ET), Pashto (pus-AF), Persian (fas-IR), Polish (pol-PL), Portuguese (Brazil) (por-BR), Portuguese (Portugal) (por-PT), Punjabi (pan-IN), Quechua (que-PE), Romanian (ron-RO), Romansh (roh-CH), Russian (rus-RU), Sanskrit (san-IN), Santali (sat-IN), Scottish Gaelic (gla-GB), Serbian (Srpski) (ser-SR), Serbian (Српски) (srp-RS), Shona (sna-ZW), Simplified Chinese (chi-CN), Sindhi (snd-PK), Sinhala (sin-LK), Slovak (slk-SK), Slovenian (slv-SI), Somali (som-SO), Spanish (Mexico) (esl-MX), Spanish (Spain) (esl-ES), Standard Moroccan Tamazight (zgh-MA), Sundanese (sun-ID), Swahili (Kenya) (swa-KE), Swedish (swe-SE), Syriac (Iraq) (syr-IQ), Tajik (tgk-TJ), Tamil (India) (tam-IN), Tatar (tat-RU), Telugu (tel-IN), Thai (tha-TH), Tibetan (bod-CN), Tigrinya (tir-ET), Tongan (ton-TO), Traditional Chinese (Taiwan) (chi-TW), Turkish (tur-TR), Turkmen (tuk-TM), Ukrainian (ukr-UA), Upper Sorbian (hsb-DE), Urdu (Pakistan) (urd-PK), Uyghur (uig-CN), Uzbek (uzb-UZ), Vietnamese (vie-VN), Welsh (cym-GB), Western Frisian (fry-NL), Western Punjabi (Shahmukhi) (pnb-PK), Wolof (wol-SN), Xhosa (xho-ZA), Yiddish (yid-US), Yoruba (yor-NG), Zulu (zul-ZA). The names are the English ones of the locale files.

Every locale is plain INI: nothing here needs a PHP extension at run time.

Contents #

1. Locales, translations, content languages and siteaccesses
2. Support levels
3. Every locale
4. Every administration translation
5. The world's major languages
6. Common languages still missing
7. Adding a locale
8. Adding or completing an administration translation
9. Right-to-left languages
10. No PHP extension at run time
11. The limit of 62 content languages
12. Test languages: Lorem Ipsum
13. Translation fallbacks

The languages not translated yet #

Exponential knows 171 languages: that many different language codes (the three letters before the dash) are found in the shipped locales share/locale/*.ini, variants and the test language left out. The administration shows 151 of them translated: at least one of their locales has a translation of its own, falls back to the first translation of its chain (FallbackMode=first) or is English, the language of the source texts. For the other 20 it shows the source texts in English; each is a candidate for a new translation (see section 8).

1. Locales, translations, content languages and siteaccesses #

Four things are easily mixed up:

A locale file is all a content language needs. Adding a locale file only makes the language selectable: it adds no content language, changes no siteaccess and uses none of the at most 62 content language slots (30 on 32-bit PHP; see section 11).

The details are elsewhere and not repeated here:

• Content languages: adding, showing, translating, moving and removing the languages content is written in.
• Translations and languages: the state of the administration translations, making your own strings translatable, the HTML language tags.
• The interface language choice: the list box of the right toolbar, its groups, its A–Z list and the coverage it shows.
• The fallback locale of a content object.

2. Support levels #

A country without a translation of its own speaks the one i18n.ini [TranslationSettings] FallbackLanguages[] names, at the same level: every German-, French- and Spanish-speaking country falls back to ger-DE, fre-FR and esl-ES (Latin America too: esl-ES is the complete one, esl-MX lacks some kernel strings and four extensions; esl-MX itself takes esl-ES for those extensions), the Arabic countries to ara-SA, and the other languages likewise (por-MZ to por-PT, ita-CH to ita-IT). The translator looks the fallback up for the kernel and for each extension on its own. expUISwitchLanguageProviderTest fails when a country of a translated language has neither. The "Translation coverage" column below counts only a locale's own translation; the interface language choice shows the fallback's. Chains, the automatic same-language fallback, rules of extensions, ./console exp:i18n and Setup > Interface languages: section 13.

3. Every locale #

share/locale holds 268 files on 18 October 2026: 49 that existed before 6.0.15, 170 added in 6.0.15, the test language qlo-ZZ (section 12) and 48 added on 18 October 2026 (the remaining countries of Spanish, French, Arabic and German, generated the same way from CLDR 48). The add form of Setup > Languages offers the 280 without a variation (a variation such as ger-DE@euro only overrides a few keys of its base file and is chosen in site.ini), sorted by English name, each with its native name, its country (unless the name already gives it) and its code, and a search field that matches any of them: bengali, বাংলা, ben, bangladesh all find ben-BD. The interface language choice offers the same locales in its A–Z list, a native list box with type-ahead (typing "Kis" jumps to Kiswahili).

The columns:

• Direction: RTL for a language written right to left (section 9).
• Script: the ISO 15924 script the file's names are written in: ICU's likely script of the language, checked against the letters of the file's own native name and month names (the file wins: ser-SR is Serbian in Latin script).
• Source: CLDR 44 for a file generated from Unicode CLDR 44 in 6.0.15; older, regenerated from CLDR 44 for an older file that was broken and generated again; hand-made names, CLDR 44 conventions where CLDR has no data for the language and the names were written by hand; hand-made for a file written entirely by hand; older, hand-made, fixed in 6.0.15 for an older file with single values corrected by hand; older, hand-made for the files that came with earlier releases, unchanged.
• Translation coverage: the share of today's interface strings the locale's own administration translation translates, as the interface language choice shows it (expUISwitchLanguageCatalogue::coveragePercent(), rounded down; English is the source language and always 100%); — without a translation.

The table is generated from the locale files; do not edit it by hand.

How the generated files were made, and where they differ from CLDR, is in section 7. Notes on single locales:

• Latin (lat-VA) is a classical language: no country speaks it as a mother tongue, and the Holy See (Vatican City), where it is an official language, stands as its country. CLDR 44 has no Latin data, so its numbers, currency (euro) and date patterns follow the Holy See's conventions (CLDR it_VA) and its names are written by hand: Latina, nominative month names (Ianuarius), the ecclesiastical day names the Holy See uses (Dominica, Feria secunda ... Sabbatum); the long date shows the short month (Feria tertia, 14 Nov. 2023). Its country names are CLDR's English ones.
• Hebrew (heb-IL) existed before; its file named windows-1255 and held double-encoded month and day names. It was regenerated from CLDR he_IL in 6.0.15 (UTF-8, he-IL, ₪, CLDR's Hebrew country names). hin-IN and kor-KR were regenerated as well: hin-IN had English names ("Hindi (Devnagari)", English months and days) and country names with characters lost to ?; kor-KR had an invalid long date (the year marker 년 before the year) and short day names equal to the long ones.
• Fixed by hand in 6.0.15, the rest of the older file kept: chi-HK (its long date printed 月 twice, and it was the one file with a byte order mark), chi-CN (short Sunday 周日), ell-GR (native name Ελληνικά instead of "Greek (Hellenic)"), ukr-UA (Українська instead of "Ukrainian"; month names in the form used inside a date, 15 жовтня instead of 15 Жовтень, with the stand-alone form beside them). kat-GE lost a capital its generator had given it (Georgian has none in running text: ქართული).
• Still to improve in older files: ser-SR and srp-RS mix English and native ("Serbian (Српски)", and srp-RS capitalizes its day names), chi-CN and chi-TW keep their old date patterns (2025年 十月 15日, readable but not CLDR's 2025年10月15日). The short month names of rus-RU (Янв, Май) are neither of CLDR's forms, and som-SO names the months in the traditional way (Bisha Tobnaad); these files have no stand-alone group for them yet.
• Dhivehi (div-MV): ICU 74 has no Dhivehi data. Numbers, the rufiyaa (symbol Rf) and date patterns follow CLDR en_MV; the language name (ދިވެހި, English "Dhivehi", the Maldivian spelling), the currency name and the month and day names in Thaana are written by hand and need a native speaker's review. CLDR starts the week on Friday in the Maldives; eZLocale knows only Monday and Sunday, so the week starts on Sunday.
• Yiddish (yid-US): CLDR 44 has Yiddish only as yi_UA (until CLDR 43 yi_001, "world"), and a locale code needs a two-letter country. The largest Yiddish-speaking community lives in the United States, so the file has CLDR's Yiddish names with the conventions and currency of the United States.
• Imperial Aramaic (arc-IQ) and Hulaulá (huy-IL, Jewish Neo-Aramaic) have no CLDR data and are written in Hebrew square script. arc-IQ takes the conventions of CLDR syr_IQ (Iraqi dinar) and its Aramaic month and country names, transliterated from Syriac; the day names are the Aramaic ones (חד בשבא ... שבתא). huy-IL takes the conventions, month and country names of CLDR he_IL (new shekel) and the Jewish Aramaic day names (חד בשבא ... מעלי שבתא, שבתא). The language and currency names are written by hand; each file names the source of every value and needs a review.
• Dari (fas-AF) is the Persian of Afghanistan, as in CLDR (fa_AF, display name "Dari", دری); fas-IR stays "Persian".
• Central Kurdish (ckb-IQ, Sorani, Arabic script, right to left) has no ISO 639-2 code (639-2 has only kur for all of Kurdish, which kur-TR, Kurmanji in Latin script, already uses), so it takes its ISO 639-3 code.
• Klingon (tlh-QO) is hand-made, for fun and as a test of a locale without CLDR data. tlh is the ISO 639-2 code of Klingon; QO lies in the ISO 3166 user-assigned range (QM to QZ), so the country is None (private-use region QO) and the HTTP tag is the bare tlh, which claims no real place (like eo for epo-EO). It has no currency (ISO 4217 XXX, sign ¤), numbers as eng-US, the 24-hour clock, Marc Okrand's day names (DaSjaj ... lojmItjaj) and the months as "month" plus a number (jar wa' ... jar wa'maH cha'), in the Latin transcription. It is offered in the A–Z list of the interface language choice only, in no group. The Klingon script pIqaD is not in Unicode proper (the ConScript Unicode Registry places it in the Private Use Area, U+F8D0 to U+F8FF) and is never used in interface texts or locale files; a site that wants it for decoration loads a ConScript-compatible font itself, while the text stays in the transcription.
• Interlingua (ia_001) is not offered: CLDR has it only for the region "world" (001), and a locale code needs a two-letter country.
• Test languages, marked [LanguageSettings] TestLanguage=true (Lorem Ipsum, qlo-ZZ), are listed last in the table, as "test language only", and counted nowhere in this document.

Scripts, fonts and line breaking #

A locale file holds text only; whether a script shows depends on the fonts of the reader's system. The administration designs (admin4, admin4l and the classic admin share components.css) use system fonts only and download nothing. Their font token --a4-font keeps its Latin families first (so Latin text is unchanged) and then names, through --a4-font-intl, the common system fonts of the other scripts: Chinese, Japanese and Korean (--a4-font-cjk: PingFang, Hiragino, Microsoft YaHei and JhengHei, Yu Gothic, Meiryo, Malgun Gothic, Apple SD Gothic Neo, Noto Sans CJK), Thai (Leelawadee UI, Thonburi, Noto Sans Thai), Devanagari (Nirmala UI, Kohinoor Devanagari, Noto Sans Devanagari), Arabic and Hebrew (Segoe UI, Noto Sans Arabic, Geeza Pro, Noto Sans Hebrew, Arial Hebrew). Han characters are shared by Chinese, Japanese and Korean but drawn differently in each, so the order of --a4-font-cjk follows the interface language (:lang(ja), :lang(ko), :lang(zh-TW), :lang(zh-HK): that language's fonts first; Simplified Chinese first otherwise). The classic admin and the Admin UI add --a4-font-intl after their own Latin families. A font that also carries Latin letters (Droid Sans Fallback) is left out on purpose: on a system without the Latin families it would take over Latin text as well. Any script not covered falls back to what the system offers.

Line breaking follows the language of the page or of an element with lang:

html:lang(ja), [lang]:lang(ja) { line-break: strict; }
html:lang(ko), [lang]:lang(ko) { word-break: keep-all; overflow-wrap: break-word; }
html:lang(th|lo|km|my), [lang]:lang(th|lo|km|my) { overflow-wrap: break-word; }

Japanese avoids a break before small kana and closing punctuation; Korean breaks at the spaces between words, as it is set; Thai, Lao, Khmer and Burmese, written without spaces, break at the word boundaries of the browser's dictionary, with break-word as a safety net for a run it does not know. Chinese needs nothing: browsers break between ideographs. The Klingon script pIqaD is never used (see Klingon above).

Indigenous languages of the Americas #

Since 6.0.15 the locales of 22 Indigenous languages of the Americas ship with Exponential, from the Arctic to the Andes. A locale file makes the language a content language: items, URL aliases and searches in the language, its own name in the language lists, and the dates, numbers and currency of its country. The administration shows these languages in English, or in Spanish where a siteaccess lists esl-ES as its interface language; see the help page for the setup.

• Month and day names: where Unicode CLDR has data (Cherokee, Lakota) the file is generated from it. Navajo, Cherokee, Inuktitut, Cree, Ojibwe, Guarani and Mapudungun carry month and day names written by hand; the Alaska languages and Mohawk keep the English names, Aymara, Nahuatl, Yucatec Maya and K'iche' the Spanish ones, as texts in those languages commonly use them. Each file says so in its header.
• Ojibwe has two locales, oji-CA and oji-US: the language is spoken on both sides of the border, and the two differ only in country, currency and date order.
• Administration translations of Indigenous languages of the Americas: Quechua (que-PE), Guarani (grn-PY), Greenlandic (kal-GL) and, from the Pacific, Hawaiian (haw-US) are complete.
• Scripts: Cherokee is written in its own syllabary (U+13A0 to U+13FF, U+AB70 to U+ABBF) and Inuktitut in Canadian syllabics (U+1400 to U+167F); every other language here uses Latin letters, some with special ones (ł, ŋ, x̱, ẽ, the glottal stop ʼ). Windows, macOS, iOS and Android ship fonts for both syllabaries; a public design that sets its own web font names a fallback such as Noto Sans Cherokee or Noto Sans Canadian Aboriginal.

4. Every administration translation #

Measured on 16 October 2026 (the ten translations added on 18 October 2026 against the 17,763 strings of that day, the three added on 19 October 2026 against the 17,834 strings of that day, Swahili, added on 23 October 2026, against the 18,404 strings of that day, the eighteen added or completed on 24 October 2026 against the 18,411 strings of that day), in two ways (both rounded down). On 26 October 2026 every translation was measured again against the 18,811 current strings: all 73 translate at least 99% of them, 38 every one (Yiddish, Vietnamese and Indonesian, added the same day, against 18,812); the other 35 lack only the 22 strings of the data table module (kernel/datatable/module), and kal-GL also leaves 25 longer sentences unfinished:

• Own file finished: finished messages of the file divided by all its messages, obsolete ones left out, each context, source and comment counted once. php bin/php/ezchecktranslation.php <code> prints a similar "Percent finished".
• Current strings translated: how many of the 18,839 strings in share/translations/untranslated/translation.ts (every string the kernel and its designs ask for today) the file translates, same context, source and comment, finished and not empty. This is what an administrator sees, and what the interface language choice shows as coverage: the rest shows in English.

Most translations were complete for the strings of older releases; 6.0 added about 10,000 strings (admin4, the new Setup pages, audit, packages, jobs and more) that only eng-US, ger-DE and the six machine-assisted translations translate so far. English is the source language: the two strings eng-US lacks show their English source text, and the interface language choice counts English as 100%.

Notes:

• Sanskrit (san-IN, संस्कृत भाषा) was added on 28 October 2026, machine-assisted, in modern standard Sanskrit in Devanagari as used in Sanskrit software localisation, with action nouns on buttons (संरक्षणम्, रद्दीकरणम्, निरसनम्, सम्पादनम्), the polite imperative with कृपया in instructions, the danda as sentence end and sandhi applied so that placeholders and Latin words stay free, Hindi (hin-IN) serving only to check modern terms, with a glossary (share/translations/san-IN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Santali (sat-IN, ᱥᱟᱱᱛᱟᱲᱤ) was added on 28 October 2026, machine-assisted, in standard written Santali in the Ol Chiki script with ᱾ as sentence end and Latin digits, Santali grammar, the bare verb on buttons (ᱨᱟᱠᱷᱟᱣ, ᱵᱟᱛᱟᱞ, ᱢᱮᱴᱟᱣ, ᱥᱟᱯᱲᱟᱣ) and placeholders that stand free, Bengali (ben-BD) and Hindi (hin-IN) serving only to check terminology, with a glossary (share/translations/sat-IN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Tongan (ton-TO, lea fakatonga) was added on 28 October 2026, machine-assisted, in the official Tongan alphabet with the fakauʻa ʻ and the macron vowels ā, ē, ī, ō and ū, the polite common register, the address koe, the bare verb on buttons (Tauhi, Tamateʻi, Fakatonutonu) and computing loan words from English in Tongan spelling (faile, sēva, kēsi), with a glossary (share/translations/ton-TO/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Bodo (brx-IN, बर’) was added on 28 October 2026, machine-assisted, in standard written Bodo in Devanagari with Bodo grammar rather than Assamese or Hindi, the bare verb stem on buttons (रैखा, बन्द, मेटा, सोलोंथाय), the polite request with नाजा and the full stop as sentence end, Assamese (asm-IN) and Hindi (hin-IN) serving only to check terminology, with a glossary (share/translations/brx-IN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Esperanto (epo-EO) was added on 28 October 2026, machine-assisted, following the Fundamento with the established computing vocabulary of the Esperanto localisations, the address vi, the infinitive on buttons (Konservi, Nuligi, Forigi, Redakti), the letters ĉ ĝ ĥ ĵ ŝ ŭ and placeholders that never take an ending, with a glossary (share/translations/epo-EO/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Guarani (grn-PY, Avañe'ẽ) was added on 3 November 2026, machine-assisted, in Paraguayan Guarani and the official alphabet of the Academia de la Lengua Guaraní (nasal vowels with the tilde, the puso as a straight apostrophe), the address nde, the imperative on buttons (Eñongatu, Embogue, Embosako'i), the Academia's terms where it has one and the established Paraguayan Spanish loan where Guarani has none (marked in the glossary), and suffixes joined to placeholders with a hyphen, with a glossary (share/translations/grn-PY/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete.
• Quechua (que-PE, Runasimi) was added on 28 October 2026, machine-assisted, in Southern Quechua and the official three-vowel alphabet of Peru, the address qam with the possessive -yki, the imperative on buttons (Waqaychay, Chinkachiy, Hukchay), Spanish loan words in Quechua spelling where Quechua has none, and suffixes joined to placeholders with a hyphen, with a glossary (share/translations/que-PE/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Northern Sami (sme-NO, davvisámegiella) was added on 28 October 2026, machine-assisted, in the common written standard with the computing terms of the Sami localisations, the address don, the infinitive on buttons (Vurket, Gaskkalduhttit, Sihkkut, Rievdadit), Sami rather than Norwegian or Finnish words and placeholders that never take a case ending, with a glossary (share/translations/sme-NO/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Kashmiri (kas-IN, کٲشُر) was added on 28 October 2026, machine-assisted, in standard written Kashmiri in the Perso-Arabic script, right to left, with the Kashmiri vowel letters of the standard orthography, Kashmiri grammar rather than Urdu, the polite imperative on buttons (محفوظ کٔرِو, منسوٗخ کٔرِو, مٹٲوِو, تَرمیم کٔرِو) and placeholders that stand free, Urdu (urd-PK) serving only to check terminology, with a glossary (share/translations/kas-IN/GUIDE.md); "Login" in the user, session and audit lists is the login name. The kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• N’Ko (nqo-GN, ߒߞߏ) was added on 28 October 2026, machine-assisted: the common Manding language in the N’Ko script, right to left, without tone marks, with the terms of the Bambara glossary written in N’Ko letters, the address ߌ and the bare verb on buttons (ߡߊߙߊ, ߘߊߓߌߟߊ, ߖߐߛߌ), with a guide (share/translations/nqo-GN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Faroese (fao-FO, føroyskt) was added on 28 October 2026, machine-assisted, in standard written Faroese with the computing terms of the Faroese term work, the address tú, the infinitive on buttons (Goyma, Ógilda, Strika, Rætta), Faroese rather than Danish or Icelandic forms and placeholders that never take a case ending, with a glossary (share/translations/fao-FO/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Dogri (doi-IN, डोगरी) was added on 28 October 2026, machine-assisted, in standard written Dogri in Devanagari with Dogri grammar and postpositions rather than Hindi ones, the polite address tus, the polite imperative on buttons (सहेजो, रद्द करो, मेटो, संपादन करो) and the danda as sentence end, Hindi (hin-IN) and Western Punjabi (pnb-PK) serving only to check terminology, with a glossary (share/translations/doi-IN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Western Frisian (fry-NL, Frysk) was added on 28 October 2026, machine-assisted, in the official spelling of 2015 with the computing terms of the Frisian localisations, the polite address jo, the infinitive on buttons (Bewarje, Annulearje, Fuortsmite, Bewurkje) and Frisian words rather than Dutch ones, with a glossary (share/translations/fry-NL/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Manipuri (mni-IN, মৈতৈলোন্) was added on 28 October 2026, machine-assisted, in standard written Meiteilon in the Bengali script as the locale has it, with Meitei grammar and suffixes, the polite request on instructions, the verbal noun on buttons (থম্বা, কেন্সেল তৌবা, ফাহনবা, এডিট তৌবা) and placeholders that never take a suffix, Bengali (ben-BD) and Assamese (asm-IN) serving only to check terminology, with a glossary (share/translations/mni-IN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Upper Sorbian (hsb-DE, hornjoserbšćina) was added on 28 October 2026, machine-assisted, in standard Upper Sorbian with the vocabulary of the Upper Sorbian dictionaries and localisations, the courtesy form Wy with its capital, the infinitive on buttons (Składować, Přetorhnyć, Zhašeć, Wobdźěłać), Upper Sorbian rather than Lower Sorbian forms and placeholders that are never declined, with a glossary (share/translations/hsb-DE/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Standard Moroccan Tamazight (zgh-MA, ⵜⴰⵎⴰⵣⵉⵖⵜ) was added on 28 October 2026, machine-assisted, in the Tifinagh alphabet of IRCAM and its standard vocabulary, the address in the second person singular, the imperative on buttons (ⵃⴹⵓ, ⵙⵔ, ⵙⵙⴼⴹ, ⵙⵏⴼⵍ) and product names, codes and acronyms in Latin letters, with a glossary (share/translations/zgh-MA/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Konkani (kok-IN, कोंकणी) was added on 28 October 2026, machine-assisted, in standard written Konkani in Devanagari, the official script of Goa, with the polite address tumi, the polite imperative on buttons (जतन करात, रद्द करात, मेटयात, संपादन करात) and Konkani forms rather than Marathi ones, Marathi (mar-IN) serving only to check terminology, with a glossary (share/translations/kok-IN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Dhivehi (div-MV, ދިވެހި) was added on 28 October 2026, machine-assisted, in standard written Dhivehi in the Thaana script, right to left, in the polite register of Maldivian government texts, with buttons in the polite imperative (ރައްކާކުރޭ, ކެންސަލްކުރޭ, ފޮހެލާ, އެޑިޓްކުރޭ), Arabic loans where Dhivehi uses them and placeholders that stand free, with a glossary (share/translations/div-MV/GUIDE.md); "Login" in the user, session and audit lists is the login name. The kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Kabyle (kab-DZ, Taqbaylit) was added on 28 October 2026, machine-assisted, in the Berber Latin alphabet of Kabyle with the letters č, ḍ, ǧ, ḥ, ɣ, ṛ, ṣ, ṭ, ẓ and ɛ, the address in the second person singular, the imperative on buttons (Sekles, Sefsex, Kkes, Ẓreg) and the computing terms of the Kabyle localisations of Mozilla and LibreOffice (afaylu, aqeddac, tazarkatut), with a glossary (share/translations/kab-DZ/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Romansh (roh-CH, rumantsch) was added on 28 October 2026, machine-assisted, in the common written standard Rumantsch Grischun with the terms of the Pledari Grond and the Romansh localisations, the polite address Vus, the infinitive on buttons (Memorisar, Interrumper, Stizzar, Modifitgar) and Romansh words rather than German or Italian ones, with a glossary (share/translations/roh-CH/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Occitan (oci-FR, occitan) was added on 28 October 2026, machine-assisted, in its Languedocien form and the classical spelling of the Conselh de la Lenga Occitana, with the computing terms of the Occitan localisations, the polite address vos, the infinitive on buttons (Enregistrar, Anullar, Suprimir, Modificar) and Occitan words rather than French or Catalan ones, with a glossary (share/translations/oci-FR/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Ewe (ewe-GH, Eʋegbe) was added on 28 October 2026, machine-assisted, in the Ewe alphabet with the letters ɖ, ɛ, ƒ, ɣ, ŋ, ɔ and ʋ, the address wò, the bare verb on buttons (Dzra ɖo, Tutu ɖa, Trɔ asi le eŋu) and computing loan words from English in Ewe spelling (faɛl, sɛva, kase), with a glossary (share/translations/ewe-GH/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Breton (bre-FR, brezhoneg) was added on 27 October 2026, machine-assisted, in the unified spelling (peurunvan) with the computing terms of the Ofis Publik ar Brezhoneg and of the Breton localisations, the polite address c'hwi, the verb-noun on buttons (Enrollañ, Nullañ, Dilemel, Kemmañ), the mutations of the grammar and placeholders that never mutate, with a glossary (share/translations/bre-FR/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Dzongkha (dzo-BT, རྫོང་ཁ) was added on 27 October 2026, machine-assisted, in standard written Dzongkha in the Tibetan script with tsheg and shad, the polite neutral register of Bhutanese government texts, the bare verb on buttons (སྲུང་བཞག, ཆ་མེད་གཏང་།, བཏོན་གཏང་།, ཞུན་དག) and Dzongkha computing terms, Tibetan (bod-CN) serving only to check terminology, with a glossary (share/translations/dzo-BT/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Bambara (bam-ML, Bamanankan) was added on 28 October 2026, machine-assisted, in the official Latin alphabet of Mali with the letters ɛ, ɔ, ɲ and ŋ, the address i, the bare verb on buttons (Mara, Dabila, Jɔsi, Yɛlɛma) and computing loan words from French in Bambara spelling (fisiye, dosiye, sɛriwɛri), with a glossary (share/translations/bam-ML/GUIDE.md) that the N'Ko translation shares; the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Dari (fas-AF, دری) has had its own catalogue since 27 October 2026, machine-assisted, adapted row by row from the complete Iranian Persian catalogue to Afghan usage: معلومات for information, بلی and نخیر, the Afghan spelling of loanwords (ویب، انترنت، کمپیوتر، تیلفون) and the words Afghan writing uses where Iranian Persian has its own (فولدر، لینک، آدرس، رمز عبور، فرمایش for a shop order, تخلص, ولایت), in Arabic script, right to left, with a glossary (share/translations/fas-AF/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete, and the locale no longer falls back to fas-IR. A review by native speakers is welcome.
• Asturian (ast-ES, asturianu) was added on 27 October 2026, machine-assisted, in the standard of the Academia de la Llingua Asturiana with the computing terms of the Asturian localisations, the address tu, the infinitive on buttons (Guardar, Encaboxar, Desaniciar, Editar) and Asturian words rather than Spanish ones, with a glossary (share/translations/ast-ES/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Korean (North) (kor-KP, 조선말) was added on 27 October 2026, machine-assisted, adapted row by row from the complete South Korean catalogue to the cultured language of the DPRK (문화어): the initial ㄹ and ㄴ of Sino-Korean words kept (리용, 련결, 렬), ‑였‑ after ㅣ and ㅐ (되였습니다), no sai-siot (수자), dependent nouns and auxiliary verbs attached (할수 있습니다, 확인해주십시오), 《》 quotation marks and the North Korean terms where they differ (화일, 클라스, 그루빠, 페지, 자료기지, 차림표, 봉사기, 오유, 찰칵), with a glossary (share/translations/kor-KP/GUIDE.md) that lists the computing terms kept from kor-KR where North Korean usage is uncertain; the kernel, ezoe and ezjscore catalogues are complete, and the locale is generated from CLDR (ko_KP). kor-KP does not fall back to kor-KR. A review by native speakers is welcome.
• Wolof (wol-SN) was added on 27 October 2026, machine-assisted, in the official Latin orthography of Senegal with the letters à, é, ë, ó, ñ and ŋ, the address yow, the short imperative on buttons (Denc, Neenal, Far, Soppi) and computing loan words from French in Wolof spelling (fisie, dosie, serwëer), with a glossary (share/translations/wol-SN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Luxembourgish (ltz-LU, Lëtzebuergesch) was added on 27 October 2026, machine-assisted, in the official orthography of 2019 with the n-rule, capitalised nouns, the polite address Dir, the infinitive on buttons (Späicheren, Ofbriechen, Läschen, Änneren) and Luxembourgish words, many of them from French, rather than the German ones of ger-DE, with a glossary (share/translations/ltz-LU/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Ganda (lug-UG, Luganda) was added on 27 October 2026, machine-assisted, in the standard Luganda orthography with double vowels and consonants and the letter ŋ, the address ggwe, the short imperative on buttons (Tereka, Sazaamu, Sangula, Longoosa) and computing loan words in Luganda spelling (seeva, fayiro, kiraasi), with a glossary (share/translations/lug-UG/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Macedonian (mkd-MK, македонски) was added on 27 October 2026, machine-assisted, in the standard literary language in the Macedonian Cyrillic alphabet (ѓ, ќ, ѕ, џ, љ, њ, ј), the polite address Вие, the imperative on buttons (Зачувај, Откажи, Избриши, Уреди) and Macedonian words rather than the Bulgarian or Serbian ones of bul-BG and srp-RS, with a glossary (share/translations/mkd-MK/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Turkmen (tuk-TM, Türkmen dili) was added on 27 October 2026, machine-assisted, in literary Turkmen in the Latin alphabet of Turkmenistan with the letters ç, ä, ž, ň, ö, ş, ü and ý, the polite address Siz, the short imperative on buttons (Ýatda sakla, Ýatyr, Poz, Redaktirle) and the polite imperative in -yň/-iň in instructions, with a glossary (share/translations/tuk-TM/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Welsh (cym-GB, Cymraeg) was added on 27 October 2026, machine-assisted, in standard written Welsh with the computing terms of the Welsh Government's term bank, the polite address chi, verb-nouns on buttons (Cadw, Canslo, Dileu, Golygu), the mutations of the grammar and placeholders that never mutate, with a glossary (share/translations/cym-GB/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Tatar (tat-RU, Татар теле) was added on 27 October 2026, machine-assisted, in literary Tatar in the Cyrillic script with the letters ә, ө, ү, җ, ң and һ, the polite address Сез, the verbal noun on buttons (Саклау, Баш тарту, Бетерү, Төзәтү) and the polite imperative in -гыз/-гез in instructions, with a glossary (share/translations/tat-RU/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Basque (eus-ES, euskara) was added on 27 October 2026, machine-assisted, in the unified standard (euskara batua) with the computing terms of the Basque localisations, the address zu, the dictionary form of the verb on buttons (Gorde, Utzi, Ezabatu, Editatu), placeholders kept free of case suffixes and Basque words rather than the Spanish ones of esl-ES, with a glossary (share/translations/eus-ES/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Kyrgyz (kir-KG, Кыргызча) was added on 27 October 2026, machine-assisted, in literary Kyrgyz in the Cyrillic script with the letters ң, ө and ү, the polite address Сиз, the verbal noun on buttons (Сактоо, Жокко чыгаруу, Жок кылуу, Түзөтүү) and the polite imperative in -ңыз in instructions, with a glossary (share/translations/kir-KG/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Galician (glg-ES, galego) was added on 27 October 2026, machine-assisted, in the norm of the Real Academia Galega with the computing terms of the Galician free-software localisations (ficheiro, contrasinal, cartafol, ligazón), the address vostede, the infinitive on buttons (Gardar, Cancelar, Eliminar, Editar) and Galician words rather than the Spanish or Portuguese ones of esl-ES and por-PT, with a glossary (share/translations/glg-ES/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Mongolian (mon-MN, Монгол хэл) was added on 27 October 2026, machine-assisted, in Khalkha Mongolian in the Cyrillic script with the letters ө and ү, the polite address Та, the verb in -х on buttons (Хадгалах, Цуцлах, Устгах, Засах) and the request in -на уу / -нэ үү, with a glossary (share/translations/mon-MN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Belarusian (bel-BY, беларуская) was added on 27 October 2026, machine-assisted, in the official orthography of 2008 with і, ў and the apostrophe, the polite address вы, the infinitive on buttons (Захаваць, Скасаваць, Выдаліць, Рэдагаваць) and Belarusian words rather than the Russian ones of rus-RU, with a glossary (share/translations/bel-BY/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Tibetan (bod-CN, བོད་སྐད་) was added on 27 October 2026, machine-assisted, in modern standard written Tibetan close to the Lhasa standard, in the Tibetan script with the tsheg between syllables and the shad as the sentence end, the established Tibetan computing terms and bare verbs on buttons, with a glossary (share/translations/bod-CN/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Kikuyu (kik-KE, Gĩkũyũ) was added on 27 October 2026, machine-assisted, in the standard orthography with the letters ĩ and ũ, the address wee, the short imperative on buttons (Iga, Tiga, Thiria, Cenjia) and loan words in Kikuyu spelling, mostly through Swahili (swa-KE), with a glossary (share/translations/kik-KE/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Hawaiian (haw-US) was added on 26 October 2026, machine-assisted, in the modern orthography with the ʻokina and the kahakō, with the computing words of the Hawaiian Lexicon Committee where they exist (terms without one were coined from Hawaiian roots and are marked "coined" in the glossary), the address ʻoe, bare verbs on buttons and a glossary (share/translations/haw-US/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome, above all of the coined terms.
• Uzbek (uzb-UZ) was added on 26 October 2026, machine-assisted, in standard literary Uzbek in the Latin alphabet of 1995 with the polite address Siz, verbal nouns in -sh on buttons and Uzbek words rather than the Turkish ones of its relative tur-TR, with a glossary (share/translations/uzb-UZ/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Sindhi (snd-PK) was added on 26 October 2026, machine-assisted, in standard Sindhi in the Perso-Arabic script (right to left) with the polite address توهان, adapted from the Urdu catalogue (urd-PK) only for shared terms and written in Sindhi grammar, with a glossary (share/translations/snd-PK/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Western Punjabi (pnb-PK) was added on 26 October 2026, machine-assisted, in Punjabi as written in Pakistan in the Shahmukhi script (right to left): the Gurmukhi catalogue (pan-IN) transliterated and then adapted to Pakistani vocabulary, with the Urdu catalogue (urd-PK) as the reference for Perso-Arabic terms, and a glossary (share/translations/pnb-PK/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Burmese (mya-MM) was added on 26 October 2026, machine-assisted, in the formal written style (sentence endings သည် and ပါသည်, not the spoken တယ်), the address သင်, verbs in -ရန် on buttons and a glossary (share/translations/mya-MM/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Fula (ful-SN, Pulaar of Senegal), Lingala (lin-CD) and Oromo (orm-ET) were added on 26 October 2026, machine-assisted, each with a glossary (share/translations/ful-SN/GUIDE.md, lin-CD/GUIDE.md, orm-ET/GUIDE.md): Pulaar in the official Latin orthography of Senegal, with French loan words in French spelling for technical terms that have no Pulaar word; Lingala of Kinshasa, with the same rule for French terms; Oromo in the Qubee Latin alphabet. The kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is especially welcome, above all for Pulaar, which has little software vocabulary in common use.
• Bhojpuri (bho-IN), Maithili (mai-IN) and Nepali (nep-NP) were added on 26 October 2026, machine-assisted, in Devanagari script, each with a glossary (share/translations/bho-IN/GUIDE.md, mai-IN/GUIDE.md, nep-NP/GUIDE.md). Bhojpuri and Maithili were adapted from the Hindi catalogue (hin-IN) for shared technical terms and are written in their own grammar; Nepali takes Nepali forms, not Hindi ones. A review by native speakers is welcome.
• Nigerian Pidgin (pcm-NG) was added on 26 October 2026, machine-assisted, in the English-based spelling of the BBC News Pidgin service with Pidgin grammar (dey, don, go, na) rather than English with a few words changed, with a glossary (share/translations/pcm-NG/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Javanese (jav-ID) and Sundanese (sun-ID) were added on 26 October 2026, machine-assisted and adapted from the Indonesian catalogue (ind-ID) for shared technical terms, each with a glossary (share/translations/jav-ID/GUIDE.md, sun-ID/GUIDE.md): Javanese in ngoko alus (the everyday forms with respectful words for the user), Sundanese in the loma register with polite forms for the user's own actions. The kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Armenian (hye-AM) was added on 26 October 2026, machine-assisted, in Eastern Armenian with the formal address (Դուք), infinitives on buttons, the Armenian full stop ։ and question mark ՞, guillemets and placeholders standing free, with a glossary (share/translations/hye-AM/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by a native speaker is welcome.
• Kazakh (kaz-KZ) was added on 26 October 2026, machine-assisted, in Cyrillic script (the script of the locale and of Kazakh software today; the move to the Latin alphabet decided in 2017 is planned up to about 2031) with the polite address (Сіз), verbal nouns on buttons and placeholders standing free, with a glossary (share/translations/kaz-KZ/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by a native speaker is welcome.
• Shona (sna-ZW) was added on 26 October 2026, machine-assisted, in standard Shona with the respectful plural address (imi), verb stems on buttons, Shona loanwords for technical terms and placeholders standing free, with a glossary (share/translations/sna-ZW/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by a native speaker is especially welcome.
• Georgian (kat-GE) was added on 26 October 2026, machine-assisted, in Mkhedruli script with the formal address (თქვენ), verbal nouns on buttons and placeholders standing free, with a glossary (share/translations/kat-GE/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by a native speaker is welcome.
• Malay (msa-MY) was added on 26 October 2026, machine-assisted, in standard Malaysian Malay (Rumi script) with the polite address anda, the computing terms of Dewan Bahasa dan Pustaka and a glossary that marks where Malaysian usage differs from Indonesian (share/translations/msa-MY/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. msa-SG and msa-BN fall back to it. A review by a native speaker is welcome.
• Thai (tha-TH) and Lao (lao-LA) were added on 26 October 2026, machine-assisted, in the polite, neutral register of current software without sentence-final particles, words written without spaces and placeholders standing free, each with a glossary (share/translations/tha-TH/GUIDE.md, share/translations/lao-LA/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by native speakers is welcome.
• Aramaic was added on 26 October 2026, machine-assisted. syr-IQ is Modern Aramaic in Syriac script (Assyrian/Syriac), unvocalized and right to left, with the polite plural address and a glossary (share/translations/syr-IQ/GUIDE.md); syr-SY falls back to it. arc-IQ (Imperial Aramaic) and huy-IL (Hulaulá, Jewish Neo-Aramaic) are written in Hebrew square script and are no translations of their own: their catalogues are the syr-IQ text transliterated letter for letter (their GUIDE.md gives the mapping). The kernel, ezoe and ezjscore catalogues of all three are complete. A review by native speakers is welcome.
• Yiddish (yid-US) was added on 26 October 2026, machine-assisted, in the YIVO orthography and right to left, with the formal address and a glossary (share/translations/yid-US/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by a native speaker is welcome.
• Vietnamese (vie-VN) and Indonesian (ind-ID) were added on 26 October 2026, machine-assisted, each with a style guide and glossary (share/translations/vie-VN/GUIDE.md, share/translations/ind-ID/GUIDE.md): Vietnamese in a neutral register with "bạn" where a pronoun is needed, Indonesian in standard spelling (EYD) with the formal "Anda". The kernel, ezoe and ezjscore catalogues are complete; ind-ID no longer falls back to the source texts. A review by native speakers is welcome.
• Scottish Gaelic (gla-GB) was added on 26 October 2026, machine-assisted, in the Gaelic Orthographic Conventions (grave accents only) with the formal "sibh" and a glossary in the Akerbeltz terminology (share/translations/gla-GB/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by a native speaker is welcome.
• Māori (mri-NZ) was added on 26 October 2026, machine-assisted, in te reo Māori with macrons in the conventions of Te Taura Whiri i te Reo Māori and the register of New Zealand government software, with a glossary (share/translations/mri-NZ/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by a native speaker is welcome.
• Filipino (fil-PH) was added on 26 October 2026, machine-assisted, in the KWF spelling with the address "mo" and the English loanwords Filipino software uses, with a glossary (share/translations/fil-PH/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by a native speaker is welcome.
• Chechen (che-RU) was added on 26 October 2026, machine-assisted, in Cyrillic with the palochka (Ӏ, U+04C0) and the formal "шу", with a glossary (share/translations/che-RU/GUIDE.md); the kernel, ezoe and ezjscore catalogues are complete. A review by a native speaker is welcome.
• Chinese (chi-CN, chi-TW) is complete since 24 October 2026, machine-assisted with a style guide and glossary per language (mainland: 帐户, 单击; Taiwan: 帳戶, 按一下, 垃圾桶). chi-HK has no catalogue of its own any more and falls back to chi-TW, chi-SG to chi-CN. A review by native speakers is welcome.
• Bengali, Urdu, Marathi, Telugu, Tamil, Gujarati, Kannada, Malayalam, Punjabi (Gurmukhi), Odia, Hausa, Amharic, Yoruba, Igbo, isiZulu, Somali, isiXhosa were added on 26 October 2026, machine-assisted with a style guide and glossary per language. Ambiguous words have no per-context translations yet, and a few strings without a reliable term stay in English until a native speaker review (the allowlist says so). A review by native speakers is especially welcome for the African languages.
• Russian, Catalan, Serbian, Bulgarian, Lithuanian, Portuguese (Portugal), Afrikaans were completed or added on 25 October 2026, machine-assisted with a style guide and glossary per language (share/translations/<code>/GUIDE.md) and checked by the same tests as every other language. ser-SR is not translated on its own: it is srp-RS transliterated letter for letter into Latin script. eng-AU is generated from the source texts with a reviewed word list (Australian spelling, GST instead of VAT). nno-NO (Nynorsk) and eng-NZ have no catalogue of their own any more and fall back to nor-NO and eng-AU. A review by native speakers is welcome.
• Spanish (esl-ES, esl-MX) was complete for an older release; the strings added since were translated for 6.0.15, machine-assisted, in each country's own usage. A review by a native speaker is welcome.
• Arabic (ara-SA) is new: Modern Standard Arabic for every Arab country, machine-assisted and checked against a glossary in a second pass; product names, file names and settings stay in Latin letters. A review by a native speaker is welcome.
• Hebrew (heb-IL), Latin (lat-VA) and Klingon (tlh-QO) are complete since 6.0.15. They are machine-assisted and should get a review before a site relies on them: Hebrew by a native speaker, Latin by a Latinist (new terms follow the conventions of modern Latin software translations), Klingon in the style of the Klingon Language Institute. Klingon is written in the Latin transcription of Marc Okrand's The Klingon Dictionary (nej Search, mej Logout, pegh mu' Password, SeHlaw Setup, veQ Trash), with descriptive compounds for newer concepts.
• French (fre-FR) is complete since 17 October 2026, measured against the 17,295 strings of the untranslated file of that day. Its older hand-made translations stay; the strings added since, and about ninety older ones that had lost a placeholder, a tag or a line break or had stayed English, were translated machine-assisted. A review by a native speaker is welcome. The catalogues of ezoe and ezjscore are complete too.
• Ukrainian (ukr-UA) is new and complete since 17 October 2026, measured against the 17,295 strings of the untranslated file of that day, machine-assisted throughout; the catalogues of ezoe and ezjscore are complete too. A review by a native speaker is welcome. Ukrainian has three plural forms and the source texts only two (%count item(s)), so a few counts read awkwardly for 2 to 4; most such texts are phrased as "label: %count".
• Turkish (tur-TR, new), Polish (pol-PL), Italian (ita-IT), Portuguese (Brazil) (por-BR), Dutch (dut-NL), Swedish (swe-SE), Norwegian Bokmål (nor-NO), Danish (dan-DK), Czech (cze-CZ) and Croatian (cro-HR) are complete since 18 October 2026 for the kernel's 17,287 strings, machine-assisted with a glossary per language; the catalogues of ezoe and ezjscore are complete too. The 478 messages added to the untranslated file while the work was under way were translated the same way when it was merged, so each covers all 17,763 strings of 18 October 2026. Forms of address: Polish imperative on buttons and actions ("Zapisz", "Otwórz"; questions keep the infinitive) and impersonal sentences, Turkish siz, Italian tu, Brazilian Portuguese você, Dutch je, Czech and Croatian formal vy / vi, Swedish, Norwegian and Danish du. The older translations of these languages that had broken placeholders, tags, line breaks, English leftovers, wrong terms or mixed forms of address were redone; obsolete entries stay as they were. A review by a native speaker is welcome.
• Persian (fas-IR, new), Pashto (pus-AF, new) and Kurdish (kur-TR, new, Kurmanji in Latin script) are complete since 19 October 2026 for all 17,834 strings of that day, machine-assisted with a glossary per language, its terms settled after the first pass, and reviewed in a second pass; the catalogues of ezoe and ezjscore are complete too, those of the other bundled extensions come with their own releases. All three address the user formally (Persian shomā, Pashto tāso, Kurmanji hûn); buttons and menu items are nouns or verbal nouns. Persian is standard written Persian with the Persian letters ی and ک and the zero-width non-joiner; Dari (fas-AF) has had its own catalogue since 27 October 2026, see above. Kurmanji has no catalogue for other countries yet. A review by a native speaker is welcome.
• Swahili (swa-KE, new) is complete since 23 October 2026 for all 18,404 strings of that day, made the same way; the catalogues of ezoe and ezjscore are complete too, those of the other bundled extensions come with their own releases. It is standard Swahili (Kiswahili sanifu) with the terms used in Kenya; Tanzania (swa-TZ) falls back to it. Swahili has no formal pronoun of address: sentences speak to the user in the singular and ask with tafadhali (please), buttons and menu items are short imperatives (Hifadhi, Futa), as in other Swahili software. A review by a native speaker is welcome.
• Hindi (hin-IN), Korean (kor-KR), Hungarian (hun-HU), Romanian (ron-RO), Slovak (slk-SK), Slovenian (slv-SI), Bosnian (bos-BA), Albanian (sqi-AL), Estonian (est-EE), Latvian (lav-LV), Icelandic (isl-IS), Irish (gle-IE), Maltese (mlt-MT), Lower Sorbian (dsb-DE) and Greenlandic (kal-GL), all new, and Japanese (jpn-JP), Greek (ell-GR) and Finnish (fin-FI), whose older translations covered about a third, are complete since 24 October 2026 for all 18,411 strings of that day, made the same way; the catalogues of ezoe and ezjscore are complete too, those of the other bundled extensions come with their own releases. The older Japanese, Greek and Finnish translations were kept where they were right and fit the glossary, and redone otherwise. Strings that need a different word in different places ("Store" as the shop tab or as Save, "Order" as a shop order or a sort order, "State", "Login", "View") are translated per context. Forms of address: formal in Hindi (āp), Korean (hamnida style), Japanese (desu/masu), Greek, Hungarian (Ön), Romanian (dumneavoastră), Slovak, Slovenian, Bosnian, Albanian, Latvian and Lower Sorbian (Wy); the singular, as usual in their software, in Estonian, Finnish, Icelandic, Irish, Maltese and Greenlandic. Buttons are infinitives in Slovak and Latvian (as in Czech), nouns in Korean, Japanese, Greek and Hungarian. Each language's GUIDE.md has the details. A review by native speakers is welcome, above all for Greenlandic, Lower Sorbian, Maltese and Irish.
• Each extension releases its own catalogues. Of the ten languages above, ezoe, ezjscore and explayouts_ui (the Layouts tab) carry complete ones; the other extensions show their texts in English in these languages until they get one.
• Two directories have no locale file of the same meaning: ser-SR and srp-RS are both Serbian (ser-SR in Latin script, srp-RS in Cyrillic; both locale files exist).
• Translations of extensions live in each extension's translations/ directory and are not counted here.