What is a language code?

A language code is a short standardized identifier for a language, such as en for English or de for German, most often the two-letter form defined by ISO 639-1. Email platforms use language codes in two places: to label the language versions of a template, and to record on each contact which language that person reads. Sending the right version to the right contact comes down to comparing those two codes.

How language codes work

The common standard is ISO 639-1, which assigns a two-letter code to each major language: en, de, fr, es, cs. The related standards ISO 639-2 and ISO 639-3 define three-letter codes, and a few languages have two of them, which is why German appears as both deu and ger. Two-letter codes are the form most web and email software expects.

A language code identifies the language only. It is a separate thing from character encoding, which decides how the letters of that language are stored as bytes, and a message in any language is normally encoded as UTF-8.

Language codes vs locales

A locale adds a region to the language. The tag de-AT means German as used in Austria, en-GB means British English, and pt-BR means Brazilian Portuguese. The combined form is called a language tag and is defined by BCP 47 (RFC 5646), the standard behind the browser's Accept-Language header and the HTML lang attribute. Software locales often write the same thing with an underscore, as in de_DE or en_US.

The region matters for formatting: dates, decimal separators, currency and some spelling. It matters much less for choosing a language version of an email, because a reader in Vienna and a reader in Berlin both read a German email. Routing by language therefore usually reduces a locale to its base language, and the regional part is kept only where formatting depends on it.

How to store language codes

Contact data collects language values from many sources, and each source writes them differently. One import has de, another has de_DE, a signup form wrote German, and someone typed Deutsch by hand. A single consistent format is easier to segment, export and debug than a mix that happens to work.

The practical rule is to store the same codes the templates use, which for most platforms means two-letter ISO 639-1 codes. Values are best cleaned on the way in: Français and french mapped to fr in the CSV before an import, and forms that offer a fixed list instead of a free-text field. A periodic review of the distinct values in the language property finds what slipped through.

Language code in Lettr

Each language version of a Lettr multilingual template is identified by a two-letter ISO 639-1 code. The code comes from the editor's built-in language list and cannot be typed, so a template cannot hold two regional variants of one language, such as British and American English.

Contact values do not have to be that exact. At send time Lettr compares each recipient's communication language with the template's languages under three rules, strictest first, and the first rule that matches wins:

1. Same tag: case, surrounding spaces and _ versus - are ignored, so DE, De and de with a trailing space all match de.
2. Same base language: the region is dropped in both directions, so de-AT and de_ch match de, and pt-BR and pt-PT both reach the pt version.
3. Same language by name: a case-insensitive alias table maps spellings to codes, so german, Deutsch and ger match de, and Français and french match fr.

The alias table covers 19 languages: Chinese, Czech, Dutch, English, Finnish, French, German, Italian, Japanese, Korean, Polish, Portuguese, Romanian, Russian, Slovak, Spanish, Swedish, Turkish and Vietnamese. It includes less obvious forms such as cz for Czech, jp for Japanese, flemish for Dutch, brazilian for Portuguese and castellano for Spanish, along with native-script names such as 中文, 日本語 and 한국어. The same 19 languages are the ones the automatic unsubscribe footer and the hosted preferences and unsubscribe pages are translated into.

Languages outside the table still route under the first two rules. A template with a hu version reaches contacts stored as hu or hu-HU, and only a name such as hungarian is not recognised. A value that matches nothing, such as it when the template has no Italian version or a typo like germn, gets the primary language. A list such as de,en also matches nothing, because each contact has one language. The campaign's review panel lists the unmatched values with their counts before the send.

The docs still recommend clean codes. A template language de and contacts stored as de match on the strictest rule, which leaves nothing to debug. The Contact Language page has the full table of accepted spellings.