Combining marks and outlier letters.

James Puckett

Is it worthwhile to put combining mark anchors in letters like ð, ß, ŋ? Are there enough mad linguists out there that a commaaccent will inevitable end up under a capital eszett? Does this stuff add so much bloat to web fonts that it might delay a user from loading two dozen tracking scripts by a millisecond? I don’t see an obvious reason to set this stuff up. But the Brill fonts can handle any crazy thing I throw at them (except y+ogonek, that looks gross) so I worry that there really is someone who needs every crazy possibility to work.

Andreas Stötzner

Combining mark anchors is an intriguing concept – in theory. In practice, it hardly ever pays off the trouble of doing it thoroughly in an extended font. To my experience, those users who tend to be fond of such niceties (academic users in particular) also tend to be rather short of money when it comes to fund this sort of cumbersome sophistication.

Mark Simonson

FWIW, it's pretty simple to add combining mark anchors in Glyphs. You put them in anyway for lots of glyphs just to get automatic accent building to work. The [mark] and [mkmk] features can then be generated automatically.

James Puckett

Mark Simonson said:

FWIW, it's pretty simple to add combining mark anchors in Glyphs. 

I think Fontlab VI does the same thing, although I’m not sure if Fontlab does it by default like Glyphs. If Fontlab VI does ship with this as the default behavior then we’ll probably see combining mark support in the majority of new fonts. 

Paul Miller

In my experience (which is admittedly limited) anchors don't add that much bloat to the file.  But they are tiresome and time consuming to add and test out.

If you are adding anchors anyway then adding a few more won't be a problem.

However it is unlikely anyone will ever use them on this sort of character.

Denis Moyogo Jacquerye

Some African languages put diacritics on nasal consonants so ŋ́, ŋ̀, ŋ̂, ŋ̄, ŋ̈ and probably others need to work if they are within the scope of your glyph set.

Christian Munk

In Danish linguistics, ð̞ is an essential sound. But it usually works fine.

One that might cause trouble is ŋ̩ (U+014B, U+0329). I don't know if some phoneticians use ŋ̍ to avoid collision.

Ogonek is used for nasal vowels in some Indigenous American languages, so they can appear on any vowel character, and perhaps also vowel-like characters such as v and w, though I have never seen that.

I don't know about eszett, but it looks like it might take a cedilla: ß̧. (https://en.wikipedia.org/wiki/Cedilla). It is certainly not a phonetic symbol, so I think it may be someones idea of an şş- or sş-ligature. I wouldn't worry about it.

Andreas Stötzner

Christian Munk said:

… about eszett, but it looks like it might take a cedilla: ß̧ …

Andreas Stötzner

Never came across this one. Are there any real-world proofs? What for does Wiki record this?!

Simon Cozens

Christian Munk said:

Ogonek is used for nasal vowels in some Indigenous American languages, so they can appear on any vowel character.

And just for more fun, their position should change: European ogoneks connect to the bottom right of the character, Indigenous American ones are centered.

Adam Jagosz

Simon Cozens said:
European ogoneks connect to the bottom right of the character, Indigenous American ones are centered.

I guess adding a locl feature for the European languages would be easier, since there is less of them? Probably not the safest way to go for a general purpose font, though...

Christian Munk said:
In Danish linguistics, ð̞ is an essential sound. But it usually works fine.

What do you mean? If the designer doesn't create the anchor, it won't work. I assume you mean most fonts you came across supported this?

A different but related question, does it make sense to add anchors to accented characters? Sadly, it's the only way to get automatic compound accents in FontForge. But I guess they (the anchors in accented characters) are even less useful in real life applications.