The languages the search stack accepts: ISO 639-1 codes, exactly the set the installed text_stemmer reports — :fr, :en, :de, …

iex> SearchCore.Language.supported?(:fr)
true
iex> SearchCore.Language.supported?(:klingon)
false

There is one spelling for a language and no aliases: search_core does not decide what a language is called, and does not carry a table that could disagree with the stemmer. Which languages exist is asked of Text.Stemmer at call time, never frozen in, so upgrading text_stemmer makes any language it adds usable immediately — no search_core release, no recompile.

Need a human-readable, localized language name for a UI? That is what the ISO code is for: hand it to ex_cldr_languages, which does it properly in every locale. Storing English names here would be a worse copy of a solved problem.

Algorithm variants

Snowball ships more than one algorithm for a few languages, addressed by a qualified code: :en_porter, :en_lovins, :nl_porter. A variant still belongs to its language — :en_porter is English, so it takes the English stopword list — which is what base/1 is for. Use base/1, not the raw code, for anything keyed by language rather than by algorithm.