Ask a Typography Expert: Talking fonts with Siva Kalyan

I thought it might be fun to have a chat with local hero @skalyan, who knows an awful lot about fonts and typography. Rather than try to write up a whole article, I thought it might be more fun to just start a topic with a Q&A style.

:point_right: Anyone else, please feel free to pop in a question at any point.

I had the pleasure of meeting Siva a few years back at a conference in Canberra, and we had a grand old time yammering. So, let’s yammer a bit more…

Siva, maybe you could tell us a little about your background and how you became interested in linguistics and typography?


Sorry I didn’t reply to this until now! (Christmas holidays and whatnot, plus I just moved to a new city.)

My interest in typography has two sources, neither of which have much to do with my background in linguistics:

  1. My interest in writing systems. I learned a couple of Indic writing systems growing up (Tamil and Devanagari), and studied Japanese in high school, which eventually led to my taking classes in Japanese calligraphy. Plus I’ve always been interested in invented writing systems: I learned Tengwar long before I actually read Lord of the Rings, and used to spend hours browsing the “Constructed scripts” section of Omniglot as a teenager. I even tried creating a vertically-written Brahmic script (inspired by the Vine alphabet, and indirectly, Mongolian), but was never happy with it.
  2. Learning (La)TeX. Before I got into linguistics, I studied maths, and learned TeX and LaTeX as a matter of necessity. My first introduction was Michael Spivak’s The Joy of TeX, which opened my eyes to things like hyphenation, curly quotes, and the difference between en and em dashes. I was immediately hooked by the almost God-like level of control that is possible with TeX, and I spent a considerable part of my teenage years either conducting typographic experiments in it (such as this one), or daydreaming about such experiments. (Eventually I realised that the only thing I enjoyed about maths was the typesetting—which is when I switched to linguistics, which I liked for its own sake.)

At some point I came across the world of commercial fonts (of the sort you might find at Hoefler & Co., Monotype, or MyFonts), and professional-quality font editors (such as FontForge and Glyphs). This led to my attending the Type Design Intensive summer program at the University of Reading a couple of times, as well as (more recently) conferences organised by the Association Typographique Internationale. I haven’t really created my own complete font yet (at least, for an established writing system!), but I have some ideas, and hope I can do so in the future. (I do have experience editing, extending, and converting fonts, though.)

Well, that’s (some of) my background! (And a taste of some of the yammering that Patrick and I get up to :slight_smile:.) I’m happy to answer any questions you may have about linguistics and typography—or to just point out things that I find cool (trust me, there’s a lot!).


It me.

Can I just pause to say that I enjoy this story.

Alas, it looks like the program is offline?

One think I’d like to ask you about is your opinion on fonts for phonetics. Whatever one thinks of SIL (and I for one have… opinions :sweat_smile:), their fonts are fantastic. One perennial favorite for me is the Gentium typeface. I think the font is handsome but the real killer feature for me is how well diacritics work. Stacking, ties, tone, all that stuff seems to Just Work™.

Gentium has been around for some time, and I don’t keep up with the typography world. Do you have any opinions about this and other typefaces for phonetic notation?

:point_right: I went ahead and changed the title of this topic — I hope others feel free to ask font/typography/Unicode questions in this thread. Come to think of it, maybe we could start a series of Ask Docling topics like this one!


To be fair, I later learned to enjoy maths for its own sake as well—it just took time (and finding the right teachers).

Oh no! Well, at least you can find the latest (pre-pandemic) cohort’s work here. (And previous cohorts’ work is still visible at the Internet Archive.) Unfortunately, they don’t seem to have continued running the program during the pandemic. I do know that the program leader (Gerry Leonidas) feels very strongly that typeface design cannot be taught remotely, so I suspect that even if they do have the resources to move it online, he would prefer not to.

Ah, Gentium! Did you know that it actually grew out of a master’s project at the University of Reading? As Gerry said when he revealed this to me, “All roads lead to Reading!”.

Gentium is handsome indeed; you won’t be surprised to learn that its creator (Victor Gaultney) is a calligrapher. And indeed, its support for diacritics is second to none. However, I find that visually, it only really works in large sizes (footnotes set in Gentium can be pretty near unreadable).

A widely-used non-SIL alternative, with equally good support for IPA and diacritics, but better suited for use at small sizes, is Brill (commissioned by the publishing house of the same name). The designer (John Hudson) is a professional typographer, and is famous for designing typefaces for minority languages (look up his foundry Tiro Typeworks for samples of his other work; I would also point out his work—together with Fiona Ross—for the Murty Classical Library of India).

Another good option is Google’s Noto Sans and Noto Serif, which at this point are basically “reference implementations” of the Latin blocks of Unicode. The design is (deliberately) not very exciting, but everything should work exactly as advertised (Disclaimer! I’ve never actually tried the Noto fonts for IPA, probably because the design is too bland for me). Fun fact: several years ago, IKEA switched from using Verdana as its official font to using a customised version of Noto Sans! (You’ll still see a mix of both fonts in its physical stores, though.)

Now if you really insist on fonts that aren’t backed by either a (para)missionary organisation or a large company—don’t worry! There are several options for you (though none designed by professionals, as far as I’m aware).

First up is Linux Libertine (and its spinoff, Libertinus). This is probably best known as the official typeface of Language Science Press, and thus probably has the best “street cred” of any open-source font for linguists. I was around for the early versions of Linux Libertine (and even remember suggesting that they add more Old Church Slavonic characters!), at a time when their IPA support was a bit dodgy. Thankfully, a side effect of its being taken up by a serious linguistics publisher seems to have been that nearly all of the Unicode support issues seem to have been ironed out by now.

Next comes Junicode, which is aimed more at medievalists, but contains pretty good IPA support. There’s not much to say about this, except that it has a “classic” feel, and contains support for Gothic, runes, and a whole bunch of abbreviations found in medieval manuscripts.

Finally, Cardo, which, like Junicode, is also aimed at medievalists. It’s pretty old, and hasn’t been updated in a while, but looks very good, and has decent IPA support.

Boy, that was a long post—and I didn’t even get around to commercial fonts that support IPA! I think I’ll leave it there for now, and give people a chance to react/ask questions.

1 Like

Oh, and here’s a “fun” font that you would never expect to have IPA support, but actually does: Pecita. Yes, you saw that right: it’s a handwritten font that supports IPA! Wild.

1 Like

Speaking of Brill and Tengwar…today Toshi Omagari released a Unicode font for Tengwar that is specifically designed to harmonise with Brill!

Also, I thought I’d mention a few fonts that have only partial support for IPA, but are so beautiful that it’s worth contributing to the development of their IPA character set if you have the time:

  1. EB Garamond. In theory this is based on the Egenolff-Berner specimen of Claude Garamond’s typefaces—but in practice the letter shapes have been “cleaned up” in a way that reflects the later history of interpretations of Garamond to a considerable extent. There is also a corresponding math font.
  2. Crimson Text. Along with Linux Libertine (mentioned earlier in this thread), this was one of the earliest high-quality OpenType fonts to become available. I actually contributed a number of IPA characters early in its development, so you may well see some of my handiwork! (Also, the creator does some really interesting research on building a computational model of letter spacing based on visual neuroscience.)
  3. Coelacanth. This is a dreamboat! It’s based on the Centaur font that’s probably installed on your system if you have a Windows computer—which in turn is based on Nicolas Jenson’s classic roman (see Adobe Jenson for a commercial interpretation), which looks like something straight out of the Harry Potter movies. Unlike Centaur (but like Adobe Jenson), Coelacanth comes in multiple optical sizes—i.e., there are different versions designed specifically for headings, footnotes, etc. It would take a whole new post (or a whole new thread!) to explain what’s so great about this—but take my word that it’s great! Oh, and don’t even get me started on the character set (which goes way, way beyond IPA or even extended Latin!).

Finally, I thought I’d close out by mentioning a couple of monospaced fonts that have good IPA support. (I think you can answer for yourself the question of why anyone would want a monospaced IPA font!)

  1. Iosevka. This is my all-time favourite monospaced font, and the one I use in (almost) all my text editors, IDEs, and terminals. In addition to IPA, it supports a wide range of Unicode characters, including extended Latin, Greek, Cyrillic, and all sorts of useful dingbats. The letters are all designed to fit inside a 2-by-1 rectangle, which has the advantage that it meshes perfectly with Chinese and Japanese fonts (unlike most other monospaced fonts!). Plus it contains a wide variety of programming ligatures (which I have a soft spot for)!
  2. Fixedsys Excelsior. Anyone remember what MS-DOS used to look like? Yes, it’s that font—but with support for not only IPA, but a whole slew of writing systems, including Gothic (in both senses, i.e. Fraktur as well as Wulfila’s alphabet), Arabic (though without vowel marks), and even Tamil (though not monospaced)! Admittedly it’s not very practical (or beautiful)—but it’s nice to know that it exists!

Thanks for sharing these beautiful fonts!

I wish there were a specimen page for IPA that tested examples of diacritic stacking, ties, tone marking, all the stuff linguists have to deal with. Then it would be possible to compare how well different fonts support those features.

This is the closest thing I’ve found, but the specimen itself seems quite limited:


Oh, wow! This covers quite a few of the fonts that I mentioned (and many that I didn’t)!

I imagine PHOIBLE (together with a browser extension for substituting fonts) would also work in a pinch.

1 Like

Maybe we could talk Unicode a little too?

If a linguist were to ask you “what are the minimum things I need to learn to have a useful understanding of Unicode”, how would you reply?

I guess at this point, since practically all software supports Unicode unproblematically, the most important things to know are (1) what characters are available in Unicode, and (2) where to find fonts that support them.

For (1), two resources come to mind:

  1. DecodeUnicode. This is where you can see the full range of Unicode in all its glory, artistically presented. There’s also a book (which I have), and even a movie (which I used to have as my screensaver!). Unfortunately, there’s not much by way of metadata. For that, you have to go to:
  2. The Unicode Consortium. This is the non-profit organisation that is in charge of deciding what goes into Unicode (and hence, they’re the ones you have to convince if you want an obscure writing system added in!). In recent years the Unicode Consortium has occasionally been in the news for their rôle in deciding which new emojis to add to the standard (they’ve even been mentioned a few times on Stephen Colbert’s show!). More seriously, their charts are the go-to reference for what characters are in Unicode, what they’re officially called, and (in some cases) what they’re used for. Special shout-out to the Unihan database, for Chinese characters (including Korean and Japanese variants!); especially delicious is the fact that the metadata for a large number of characters include not only readings in Mandarin, Cantonese, Japanese (both on and kun), Korean, and Vietnamese, but even Middle Chinese (useful if you’re reading Tang poetry!).

Anyway, where was I?… Oh yes, the Unicode Consortium. You can find links to all the proposals that have ever been made to the Unicode Consortium here. Many of these are super interesting to read, as they essentially provide literature summaries for individual writing systems (or characters within such writing systems); see e.g. the proposal to encode the (historical) Nandināgarī writing system of South India, or the document that introduced the “multiocular O” (ꙮ) character (Fig. 42, p. 46) in Cyrillic Extended-B.

Also, did you know you can support Unicode by adopting a character? Most people seem to do it with emojis or dingbats, but there’s no reason to be so limited!

Anyway, besides the websites mentioned above, I strongly recommend playing around with a Character Map-type application. The best I’ve found so far is gucharmap in Linux, though there is also an application called Ultra Character Map for Mac that has almost the same functionality. Character Map in Windows has very poor functionality in comparison, but unfortunately I don’t know of a good alternative (I haven’t really looked, as it’s been years since I used Windows).

That about covers it for figuring out what riches are embedded in the Unicode specification. As for where to find fonts that support specific Unicode ranges…well, Noto Fonts are your one-stop shop! (Note that they include a large variety of archaic and obscure writing systems, including highly complex ones such as Egyptian hieroglpyhs, Mesopotamian cuneiform, Javanese, and Grantha. All Noto fonts are commissioned from professional typeface designers, and as far as possible, are made in consultation with native users of the scripts.)

I should mention one widely-used piece of software that doesn’t support Unicode by default: (La)TeX. Fortunately, thanks to the folks at SIL (I know…) there are Unicode-compatible versions of these programs, called Xe(La)TeX, which can be used almost as a drop-in replacement. (There’s also Lua(La)TeX, which also supports Unicode, is not SIL-associated, and is allegedly superior to Xe(La)TeX in some ways, but I’ve never used it myself, so I can’t really comment on it.)

I just realised that I completely missed the point of the question, which was “What are the minimum things I need to learn…?”, not “What is everything I need to learn…?”. Oh well—I’ll leave this up, and hope that you can figure out a useful subset!

1 Like