Unicode CLDR: The Local Conventions Behind an Interface

Why localization needs locale data

Translating the label “Total” does not settle how the number next to it should appear. A product also needs decisions about dates, currencies, ordering and the regional conventions relevant to its users.

The practical lesson is to separate textual translation from locale-sensitive presentation. A translator should not need to rewrite a date string every time an application displays a new date.

An illustrative formatting problem

Imagine an invoice page translated into five languages. The amount, currency and date come from structured data. If a developer assembles them with a single hand-written pattern for every locale, translated labels may surround ambiguous or unfamiliar values.

Our suggested test is to define the intended locale for each scenario and render representative amounts and dates through the application's supported formatting system. Compare the result with the product's documented audience expectations. Include edge cases such as zero, large amounts and dates that could be read in two orders.

These are editorial testing suggestions, not a list of required CLDR conformance checks or a claim about exact formatting in a particular runtime.

Version support is an implementation question

A published CLDR release does not mean that every browser, operating system or library already uses it. Our recommendation is to record both the reference data version and the deployed runtime when investigating differences.

For a product team, upgrading a library can be as relevant as editing a translation. A discrepancy may arise from an older data bundle rather than from the target text.

Connection to LISA and AI

CLDR is a Unicode project with its own development history. It belongs in this collection as a technical foundation for localization, not as a direct successor to a LISA standard.

For AI-assisted workflows, our recommendation is to use structured values and appropriate runtime formatting instead of asking a model to invent presentation rules on each request. AI may help identify a suspicious example, but it should not silently redefine a locale's formatting behavior.