Locale

Origin & evolution

The modern locale tag is standardised by the IETF as BCP 47, a stable name for a series of RFCs. The current syntax lives in RFC 5646, "Tags for Identifying Languages", which obsoleted the earlier RFC 3066 and the 1995 RFC 1766. A tag is hyphen-separated subtags, each adding precision: a language subtag (en), an optional script (Hant for Traditional Han), and an optional region (GB). The individual subtags are catalogued in the IANA Language Subtag Registry, maintained independently so the registry can grow without reopening the RFC.

The format spread because everything adopted it. Windows, Java, C#, JavaScript's Intl, the HTML lang attribute, and the ICU libraries all speak BCP 47. That convergence is why a locale string set in a design file flows untranslated through the whole stack.

How it works in practice

A team shipping to Switzerland cannot get away with a single "German" build. They need de-CH, fr-CH, and it-CH. The language subtags pull three translation sets; the shared CH region carries the Swiss franc, the day-month-year date order, and the apostrophe digit grouping (1'000'000). When the same product launches in Germany, de-DE reuses the German strings but switches the currency to euro and the grouping to a full stop. Language and region vary on separate axes, and the locale tag is what lets them vary independently.

Locale vs. its neighbours

• Language is one axis of a locale: the words. A locale adds region, and region changes formatting and currency even when the words are identical. en-US and en-GB share a language and disagree on dates, spelling, and the dollar-versus-pound symbol.
• Region is the other axis: the place, which governs currency, number and date formats, and legal defaults. A region without a language is incomplete; CH alone cannot tell the product whether to render German, French, or Italian.
• Internationalisation (i18n) is the engineering that makes a product able to take any locale: externalised strings, Unicode, flexible layouts. Localisation (l10n) is the work done per locale. The numeronyms were coined at Digital Equipment Corporation in the 1980s, counting the letters between the first and last character. A locale is the unit l10n operates on.

In the graph

In the Unified Product Graph, a locale sits in the localisation region and acts as the join point for going global. A product reaches it through product_localised_in_localeProductlocalised inLocalehierarchy. From there a locale draws its translated strings through locale_translated_via_translation_bundleLocaletranslated viaTranslation Bundlehierarchy and its deeper market-fit work through locale_adapted_via_cultural_adaptationLocaleadapted viaCultural Adaptationhierarchy. Pricing attaches separately, so a locale can carry its own regional_pricing💴Regional PricingLocalisationLocation-based pricingView reference → without entangling currency with language. Keeping the locale as a first-class node, rather than a string buried in a config file, is what lets the graph answer "what does shipping ja-JP actually require?" by following its edges.