Google Noto.

This report by Patrick Burgoyne is a year old now, but has only recently come to my attention, and I thought I’d see what you people think (I’m sure those of you who know about typefaces have long been familiar with it):

In what is being billed as one of the largest typeface projects in human history, Google and Monotype have collaborated to create a unifying system that spans more than 100 writing systems, 800 languages, and hundreds of thousands of characters. Some of the languages featured have never been digitised before and have only previously existed as inscriptions on stone or on ancient documents.

The name, Noto, comes from Google’s initial brief to Monotype which was ‘no more tofu’. ‘Tofu’ is the nickname used for the blank boxes that appear when a computer or website is unable to display a particular character because there is no font support for that language. Google’s challenge was to create fonts for all of the 800 languages included in the Unicode Consortium standard for software internationalisation, which includes many little-spoken or so-called ‘dead’ languages, thus eradicating ‘tofu’ from our screens. For each language, Noto includes letters in multiple serif and sans serif styles across up to eight weights, as well as numbers, emoji, symbols and musical notation.

The project involved hundreds of researchers, designers, linguists, cultural experts and project managers around the world. To create the Tibetan characters, Monotype worked with Tibetan monks and Buddhist scholars. Fulani speakers from West Africa have been provided with the first digital alphabet in their language, while there is even a version in Ogham, the Old Irish script used between the first and sixth centuries.

Seems like a good idea, and I like the fact that they handle Fulani and Ogham, but I’m sure there are problems with it, because there are problems with everything.


  1. I love the concept of it, and dream of a world in which you’d be able to write anything in a given (super?-)typeface. On the other hand, as I’ve noted in a couple LH comments in the past, Noto does seem a little underbaked in places. For example, IPA stress and length marks render very poorly in it. Another concern is that most of the non-Latin writing systems are only available in a sans-serifesque style, which rather undoes the sense of a unified “look” if you want to pair them with serifed Latin. (And I’ve also had a problem with them grabbing punctuation marks and numerals from my default system font, rather than from Noto Sans.)

    But my pettiest complaint is that Noto Serif includes the little optional left nub on ß without including it on ſ, on which it’s based. Make up your mind, people! (Even the reverse, including it on ſ but not on ß (as the default font here appears to do?), would make a bit more evolutionary sense.)

  2. IPA rendering, real, good, IPA rendering with the brackets and diacritics etc. is quite hard to find. Even Junicode and Brill botch it. I’ve resorted to using Charis SIL for IPA, even though I don’t like it for the rest of the body text, just because it spaces things properly.

    > the little optional left nub
    I’m not surprised; it seem that, the better the Unicode handling, the worse is the typographical sensibility of a font (with fake small caps, auto-generated sizes etc.). More generally, I feel like we can have good, open, technically satisfying software or good design, but not both. I often fantasize of undergoing design & typography training, specifically in order to combat this lack of design sense in all but the most restricted walled gardens…

  3. I don’t think it’s that simple, alas. Michael Everson’s specialized fonts are tasteful and interesting; Everson Mono (which covers all non-CJK characters in Unicode as of a certain date) is so butt-ugly I can’t bear to use it as my monowidth font (though not as bad as Code 2000). There’s something about setting out to make a “consistent” Unicode font that just tends to produce hopeless results. Perhaps it’s like convertible sofas (the classic type, not the futon type): bad as sofas, bad as beds.

  4. IPA stress and length marks render very poorly in it.

    IPA would be way down on my list of concerns; it’s never going to look pretty, and it’s only used by specialists.

  5. There’s a choice– you can do fonts and typesetting beautifully in a narrowly defined area, or you can do it unbeautifully in a widely defined area. The best example is Donald Knuth’s work in typesetting mathematics– thanks to Knuth’s work, there is a universally accepted ‘right’ way to typeset mathematics. Knuth was motivated by the ugly choices for typesetting the landmark texts he wrote on computer programming– and fixing that took him twenty years.

  6. Marja Erwin says

    Strobe warning for the link in the original post. It flashes while loading and the images flash thereafter.

    How readable will these fonts be? I currently use OpenDyslexic for the Roman alphabet, because the words don’t turn into chickenstrokes. Although they can still turn into gray blobs. Some people use Comic Sans instead, because the letters are more unique, and less likely to switch places.

    … Of course, if it’s coming from Google, they may all be animated fonts, which spin, and zoom if users try to scroll, with an autoplaying page-flip system instead…

  7. See this blog post:
    which discusses Noto and compares it to other fonts.

  8. Apparently Noto doesn’t have small caps. Tsk.

  9. We surely need to talk about Noto Sans CJK and Noto Serif CJK.

    Adobe and Google have collaborated to produce a unified font for Traditional and Simplified Chinese, Japanese, and Korean. Depending on which language is chosen, the right glyph variant of the same Chinese character can be used (see here for examples).

    This massive CJK font was released as Noto Sans CJK and Noto Serif CJK by Google using Latin glyphs from the Noto families, and as Source Han Sans and Source Han Serif by Adobe using Latin glyphs from Adobe’s Source families. There are more details about the project on Adobe’s CJK Type blog.

    The Noto Latin fonts are serviceable (for the screen, not really for print) but nothing to write home about. But the cool thing is that with this project, you now have freely available web fonts that at least meet the minimum standards for serious text in many under-served scripts. For instance, I’ve used them for Khmer, Lao, Tibetan, and Burmese in this blog post. This is really useful since a scandalous proportion of devices come without any Burmese fonts, for example.

    I’ve also been using Brill a lot lately to create graphics with text for my language-related writings due to its excellent coverage for Latin and Greek as well as Cyrillic (it was designed specifically for academic publishing, after all), though I see from the Cipanglossia blog post linked above that it doesn’t support most non-Slavic or archaic Cyrillic letters. I like Brill for its balance of professionalism and personality (hello, serif-less ascenders in the italic), reminiscent of Baskerville but actually taking inspiration from handwritten texts produced by English writing masters in the 18th century. I haven’t found much to complain about its IPA symbols, though perhaps my expectations have been lowered by years of struggling with inadequate fonts.

  10. FYI: DejaVu Sans Mono is a very aesthetically-pleasing fixed-width font with good unicode coverage.

  11. > DejaVu Sans Mono is a very aesthetically-pleasing

    I can’t fathom for what manner of æsthetics would the above sentence fragment be true…

  12. De fontibus non est disputandum…

  13. David Marjanović says

    IPA would be way down on my list of concerns; it’s never going to look pretty, and it’s only used by specialists.

    IPA is designed to look pretty, to such excesses as ɛ not being the actual Greek ε just to make sure it doesn’t stand out (let alone ɣ vs. γ); and it’s not only used by more specialists than anything comparable, it’s also occasionally used by non-specialists, even some schoolbooks.

  14. By the way, their talk about covering Fulani refers to the alphabet called Adlam which was invented in the late 1980s, and was discussed in a previous post on this site.

    This blog post from Monotype has slightly more information about designing Noto Adlam. They were able to work with Abdoulaye and Ibrahima Barry, the Guinean brothers who invented the script, which allowed them to note that “Adlam was the first time in Monotype’s 100+ year history that the type designers worked directly with the script inventors”.

    Of course, as noted in the previous thread, Fulani already had two serviceable writing systems before Adlam was invented. Evidently Adlam has proved quite popular, so it is great that it was included in the project. But the statement in the Creative Review article that “Fulani speakers from West Africa have been provided with the first digital alphabet in their language” needs to be revised.

    You can see Noto Sans Adlam and a couple of other typefaces here. This does also bring into question whether Noto was the very first digital typeface for Adlam, or just one of the first.

  15. Traditional Mongolian is, as usual, a problem case.

    Searching Noto Traditional Mongolian on Google, you immediately find posts at github noting rendering issues.

    According to one Marekjez86, “Unicode is re-working Mongolian. I expect some results toward end of this year. We will NOT make any changes in Mongolian until we hear from the Unicode Consortium.”

    So we can keep our fingers crossed.

Speak Your Mind