19

Recently I've resumed work on a font I'm currently developing, meant to be released as a completely free, open-source, OFL-licensed font designed specifically for use in academic and formal writing.

Besides being free and open source, it aims to have exceptional symbol coverage among many different languages (in particular supporting Latin-based scripts, Greek, and Cyrillic-based scripts), as well as coming with optical sizes, which can be thought of as extra families of the font designed specifically for certain sizes, like for footnotes or book covers.

Currently the regular style of the font looks like this (please ignore the currently defective spacing at certain parts):

enter image description here

I have a few plans to better support linguistics in relation to the design of the font:

  • First, I plan to add extensive support for phonetic notation, including in particular the IPA, but also the extIPA and Americanist phonetic notation.

  • I plan to follow the example of established phonetic fonts while designing everything, aiming for the best design possible, such as discussed here.

  • I plan to support a really wide array of languages, from largely used languages such as Spanish, German, Vietnamese, Russian, Ukrainian, Greek, etc. all the way to low-resource African languages like Hausa, Igbo, and Yoruba, low-resource languages using the Cyrillic scripts such as Abkhaz and Mari, etc.

    Part of the reason for this, at least when it comes to linguistics, is to facilitate the writing of articles in the phonetics/phonology/structure/etc. of those languages, allowing for a unified font to be used in writing, and dispensing with the need to use multiple fonts due to lack of glyph support.

However, not being a working linguist myself, it is evident to me that I significantly lack the actual experience to be able to effectively identify most of the issues that might come up in actual practice. To this end, I would like to ask:

Question. What are difficulties linguists have run into while working with commonly used fonts in their works?

What should I focus on to be able to best design a font that will amply satisfy the needs of working linguists, based on the issues they have encountered in their work and research experience?

6
  • See also the Linguistics Meta discussion for this question, which preceded it: link.
    – Emily
    Feb 29 at 17:00
  • 1
    Not a full answer, but the Unicode Phonetic Extensions and Phonetic Extensions Supplement blocks are of potential interest in a number of linguistics contexts (especially when it is preferable to use a local phonetic alphabet instead of the IPA), and most fonts don’t have full coverage (if they have any coverage at all). Mar 1 at 3:01
  • @AustinHemmelgarn Oh, I didn't know about these; I'll be sure to provide full support for them in the font. Thank you so much for letting me know!
    – Emily
    Mar 1 at 22:41
  • wait, Japanese didn't have native words for "bed" and "dog"???
    – user253751
    Mar 2 at 19:23
  • 1
    @user253751 I think that quoted example is saying that what changed was that ドッグ was pronounced more like ドック. 犬 (ken or inu: dog) is old. Bed is more interesting. ja.wiktionary.org/wiki/寝台 looks to have been used in early 20th century. 寝床 might also have been used. Mar 6 at 18:23

3 Answers 3

15

This sounds like an excellent project. I hope you succeed and look forward to seeing the end product! These are some of the main requirements for my particular field of linguistics, Indo-European studies. For more general/theoretical branches of linguistics, there will be other needs related to formulas, syntax trees, etc.

My current job consists in large part of typesetting and laying out academic books (not only Indo-Europeanist ones, but within the humanities), and there are a few very frequent and very frustrating shortcomings that I face over and over again.

Aside from a solid support of IPA, which it seems you’ve already pretty much got covered, much of what I typeset requires the font to support:

  • characters used to transcribe Indo-Europeanist things (and various other commonly found transliterations), such as velars with inverted breves ⟨k̑ g̑⟩, vowels with inverted underbreves ⟨i̯ u̯⟩, vowels with macron + breve ⟨ā̆ ē̆ ō̆ ū̆ ȳ̆…⟩ macron + accent ⟨ā́ ḗ ī́ ṓ ū́…⟩
    ** [Notice how, at least on macOS, the fallback font in the browser paints the macron and acute over the e differently to the others since ḗ is the only one that exists as a precomposed character]
  • advanced Ancient Greek, including accents and breathing marks, historical letters (qoppa, sampi, etc.), alternative letter forms (like ⟨ϑ⟩ for theta), etc.
  • advanced Cyrillic, including historical variants
  • OpenType caps features (caps, smcp and c2sc) for all of the above, including in bold and italic styles

Just these four requirements, especially the final one, whittle the currently available options down to just two or three fonts, of which one is a modified version of Minion commissioned especially for Indo-Europeanist use and not publicly available; and another was commissioned and published by another publisher of linguistic works who have stipulated that it may not be used by competing publishers.

In addition to these requirements, various others occasionally pop up, such as support for Mycenaean Greek, Linear A/B, various obscure manuscript ligatures, Tocharian (not yet encoded in Unicode), etc. However, these are so obscure that I would not expect a generally linguistics-oriented font to support them; there are specialist fonts for them.


Ideally, a linguistics-oriented font should also support Devanagari, CJK, Arabic and pretty much any other script you can think of, to ensure consistent line height, letter size and stroke width across scripts. In your sample image, for example, the kana look like they’re from a different font that doesn’t fit with the rest of the text very well (no serifs and different stroke width, line height and letter size).

But supporting a world of scripts is a massive task and not doable by a single person – Google’s Noto font suite were made by hundreds of people over several years and are still far from complete.

5
  • Hi Janus, it's really nice to meet you! Thank you so much for such a detailed and extremely useful reply, as well as for your well wishes.
    – Emily
    Feb 29 at 19:53
  • Would you be open to communication with me in the future, as I go on implementing these features on the font? I believe having the input of such a knowledgeable specialist like yourself would greatly improve what I can achieve with the font, and would also allow me to eventually provide comprehensive support for the issues you face in your day to day work.
    – Emily
    Feb 29 at 19:53
  • Regarding the addition of other scripts, this is a point that I really want to improve the font on, but also recognize the enormous effort involved in doing so. Currently I've decided to set for a compromise: first I'll develop all the styles and optical sizes of the font for Latin, Cyrillic and Greek, and after this is finished I plan to consider expanding the font to other scripts, such as Georgian, Hebrew, Thai, Devanagari, Coptic, and even Japanese, among others.
    – Emily
    Feb 29 at 19:53
  • 2
    @Emily Certainly, I’d be happy to help where I can. If at some point you plan to move the project to Github, you can connect to me there (@koskoshneta). I’m sure there will be others who can say more about the needs of theoretical linguists, which will, I’m sure, be quite different to the historical linguist’s. Feb 29 at 22:00
  • Thank you so much, your advice and expertise will surely prove invaluable in designing the font to better cater to linguistics usage; I truly appreciate it! I'm currently working on setting up a repository for the project on GitHub, and will let you know as soon as it is ready. Thank you once again!
    – Emily
    Mar 1 at 22:42
9

First of all, your font looks already really good in the main roman typeface and has good distinctive features.

In the support of West African orthographies the design of italics becomes a bit tricky because of the letters F with hook (uppercase Ƒ, lowercase: ƒ) and V with hook (Upper case Ʋ, minuscule: ʋ). A way to deal with this is carefully choosing the shapes of italics for f, v, and subsequently also w, see Knappen 1993. Fonts for Africa: The fc fonts. TUGboat 14, 104–106. Another problem is the letter Eng or Engma (capital: Ŋ, lowercase: ŋ) which has several different shapes for the capital letter and communities may have strong feelings which of the shapes looks right, but unfortunately they don't agree (see, e.g., this mail message by Lorna Priest from SIL).

There are probably other possible pitfalls, one that comes to my mind is the Esperanto lowercase letter ĥo: It is better to place the hat to the right of the main stem than directly above it.

P.S. The documentation of the TIPA package for LaTeX by Fukui Rei 2002 is also an excellent source of information and inspiration.

5
  • The eng issue should be largely solvable by mapping specific variants to specific languages, in the same way as ş (s with cedilla) is often made to be automatically substituted for ș (s with comma) in Romanian and Moldovan specifically, due to the historically poor support for the latter. Mar 1 at 12:18
  • Yes, but s with cedilla and s with comma below have now different Unicode code points. For the capital Eng there is no such differentiation, alternate forms need to be but somewhere (most easily, into the private use area; or using some open type features for selecting glyph variants) Mar 1 at 14:13
  • Yeah, it’s not a complete parallel anymore, but the substitution logic is still the same: lots of fonts continue to use language-based substitution to select the comma variant for Romanian even if the cedilla variant is used, for backwards compatibility. Whether the variant maps to a Unicode point isn’t necessarily relevant to that since substitution rules work on glyphs, not code points. Mar 1 at 14:30
  • And here is a question about another specific letter (Capital letter Esh): linguistics.stackexchange.com/q/30221/9781 Mar 1 at 21:26
  • @SirCornflakes I'm really happy you like the current version of the font! Thank you so much for your advice and pointers to resources, knowing this will be exceedingly helpful when designing this part of the font!
    – Emily
    Mar 1 at 22:43
4

I am not a working linguist, but have often needed to use linguistic notation in things that I am writng. As Janus Bahs Jacquet mentioned, good coverage of letters with diacritics, including ones that are not precomposed characters, is helpful. I have noticed that some fonts do not position the breve correctly when a vowel with a macron and breve is italicized, e.g. on my computer, ē̆ shows up on one browser with the ˘ slightly too far to the left, and on another with it much too far to the right:

enter image description here enter image description here

Taking care of details like this will surely be appreciated.

1
  • Thank you for letting me know about this issue, I'll be sure to account for it and in general properly support stacked diacritics in the font as best as I can
    – Emily
    Mar 1 at 22:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.