fontswebfontsgoogle-webfontsglyph

Languages supported by "latin" vs "latin-extended" glyphs in fonts on Google Web Fonts?

Google Web Fonts Select Character Sets

Some fonts on Google Web Fonts support multiple "character sets". The thing is, if the web font I use only serves the "latin" glyphs, users who translate the page to a language whose glyphs aren't supported will clearly notice the messed up text.

I'd like my web fonts to support the most popular languages in the world aside from English, for example, Spanish, German, French, etc.

For this purpose, I'd like to know, which languages exactly, the "latin" and "latin-extended" cater to, individually.

I expect the answer to look like:

Latin Character Set & Supported Languages:

- ..........
- ..........
- ..........

Latin-Extended Character Set & Supported Languages:

- ..........
- ..........
- ..........

I couldn't find this info in Google Web Fonts documentation, or by Googling.

Solution

  • Latin

    aka Unicode Latin1-Supplement (U+0080 to U+00FF) is meant to support primarily Western European languages (as you mentioned French, German, Spanish, also Portuguese, Italian, Irish, Icelandic, languages of Scandinavian countries and unintentionally also other languages mentioned in the list below). English is supported by standard ASCII. ASCII (first 127 chars, 95 of them are graphemes U+0020 to U+007E) was placed as the very first block in Unicode named Basic Latin. This block is considered as a part of "Latin" and is usually supported even in non-latin fonts allowing them to be used as system fonts (most non-localized low-level programs have ASCII hardcoded).

    Latin Extended

    Latin Extended on Google fonts means practically block Latin-Extended-A (U+0100 to U+017F) which should (combined with "Latin") support all European based latin-written texts. Internet emerged in the USA, so ASCII was its native code. Then ISO-8859-1 (Latin1) standard for upper half of 8bit codepages was defined to support Western Europe, which was transformed to Latin1-Supplement Unicode block. Other 8bit ISO-8859 European Latin standards (Latin 2 East, Latin 3 South, Latin 4 North) were merged and moved to Latin-Extended-A block. These Latin standards shared many characters with Latin 1, so almost all European languages (except for Maltese, Latvian, Lithuaian) in "Latin-Extended" range require also Latin1-Supplement. This means that fonts supporting "Latin-Extended" should naturally include "Latin" category, but some fonts may decide to support only chosen language (like Lato font which supports only Polish characters - the author is a Pole). Therefore if you don't aim specifically for some of the mentioned three languages and if you check "Latin Extended", make sure to check also "Latin".

    In Unicode, there is also Latin-Extended-B block which added support mostly for non-European Latin alphabets, Azeri Ə and Romanian Ș, Ț (to fix previous mistake), but these characters are often replaced with Ä, Ş, Ţ from Extended-A (albeit my Romanian friend told me that it is unacceptable substitute). Support also includes Vietnamese Ơ, Ư (but this has its own category on Google fonts) and some African languages, which also require Latin-Extended-Additional block.

    African Latin languages are mostly not supported by Google's Latin Extended category (the list of compatible Google fonts is below). There are even more exotic C, D and E extensions (252 characters total) containing outdated and today mostly useless letters and symbols. This table sums this up (not 100% correct, just to get the idea of the blocks main intention):

    --------------------------------------------------------------------
| Unicode Latin Set         | Latin Support       | Google Name    |
|==================================================================|
| Basic Latin (aka ASCII)   | English             |                |
| Latin1-Supplement         | Western European    | Latin          |
|------------------------------------------------------------------|
| Latin Extended A          | European based      | Latin Extended |
|------------------------------------------------------------------|
| Latin Extended B          | non-European        | Vietnamese     |
|------------------------------------------------------------------|
| Latin Extended Additional | African             |                |
|------------------------------------------------------------------|
| Latin Extended C, D, E    | Historical, Exotic  |                |
--------------------------------------------------------------------

    How Google categorizes fonts

    From observation, if the font contains SOME characters from Latin Extended A block, Google places it into Latin Extended category. There is no information if the font supports ALL characters from the block: the Glyphs tab in font details doesn't display all glyphs in font.

    Officially, GoogleFonts categorizes Latin into 6 "glyph sets" described here (credits to @Andj in the comments) which don't exactly map to Unicode character sets. The link contains full list of language support (hundreds of latin languages), but it seems that the rules are not applied strictly, at least at legacy fonts added before these rules. The current status is rather unclear as the "language" filter on Google fonts confuses language, writing system and unicode block. For time being, to determine the language support try to display local characters from the list below.

    Languages support

    From the list of latin-written alphabets below inspected on Omniglot and other sources, I do not count:

    Please comment if something important is missing or if some minority language is used in electronic communication. Bolds are official major country-wide languages. In this list there are languages spoken by at least hundreds of thousand people.

    ASCII (Basic Latin, often supported even in non-latin fonts)

    Clasical Latin, Aymara (Bolivia) Afrikaans (south Africa), Asturian (Spain), Corsu (France), Dutch, Fijian, English, Greenlandic, Gaelic (Scotland), Gilbertese (Kiribati), Haitian, Hiligaynon (Philippines), Lombard (Italy), Malay, Shona (Zimbabwe), Sicilian, Swahili (central Africa).

    Latin

    Latin Extended

    Latin Extended, African (mostly not supported in Latin-Extended fonts). Full support of Africa alphabet has Ubuntu, Fira Sans, EB Garamond, Tinos, News Cycle, Didact Gothic, M Plus, Sawarabi, Cousine, Caudex, Judson, Andika (and of course Noto, see below). (This section is incomplete and will be revised yet.)

    Combining diacritical marks

    Alternatively, the font may support the Combining Diacritical Marks block: U+0300 to U+036F. For example, Ř can be typed either as U+0158 (aka precomposed character) or as R + U+030C. Program supporting Unicode should both display and treat the same and provide some API to deal with it - like String.normalize() to decompose diacritics - but if the program or font doesn't support repertoire, the combining diacritical mark might end up a bit misplaced (like too low umlaut on Ɛ̈ it seems to get fixed in this font), see this very detailed Unicode Q&A on this topic.

    Non-latin characters in Latin languages

    Many Latin fonts support some characters outside of Latin scope, as they are common in Latin texts, namely:

    If your font doesn't support them, I recommend to try and see how it combines with fallback font like in this sentence (to copy and paste incl. the bullet sign)

    • “We sell ‘cheap’ capacitors in range μF–mF, 2€ per pack”

    Customizing fonts

    You might want to customize some fonts (if their licence allows it) by Font Squirrel service or use them as a backup.

    Fonts with extensive amount of characters:

    If you really like some font that lacks support of some diacritics, it is quite easy to add the support using Font Forge. In that case read the font license carefully: from the legal point of view, font is software.