We have a set of tests for consistency in the data for language, script, and country. The following is a draft description what those consistency checks should aim for.
Default script, language
* default = most used in writing; currently if modern, otherwise historical.
- For each script encoded in Unicode, the default* language is present in the script metadata.
- For each language used in CLDR, there is a default* script
Implications for Language-Country population data (LCPD)
- If a base-language has a CLDR locale, then it is in the LCPD for at least one country.
- If there is a CLDR country locale for a language, then that language+country is in the LCPD.
- For each country, get the language most widely used as a written language in that country. That language+country combination is in the LCPD.
- When a significant proportion of the language use in a country is in a non-default script, that script is marked in the LCPD.
- When a script is not EXCLUDED in UAX#31, then we have at least one language-country pair in the LCPD.
- If a language has a significant* literate population in a country, the pair is in the LCPD. This target is fuzzier, but definitely
- anything >1M, or
- >100K and either official (real, not honorary) or 1/3 of the population.
Implications for Likely Subtags
Likely Subtags are built from the language-country population data, plus the script metadata, plus an exception list.