Unicase accents

Nick Shinn's picture

How far should a unicase typeface go in reducing the number of characters in a font?
Should unicase fonts include accented characters?

Florian Hardwig's picture

Hi Nick,
please define (your) unicase. Really just one form (‘uniformed’) of each character?
What has unicase to do with reducing the character count? In a ‘normal’ font, you’d have both Ä and ä, both À and à. Why would you want to drop the accents when unifying these two cases?

I’ve been told that it once was okay to omit the accents on capitalized French text. But that had technical reasons (the typewriter?) and shouldn’t be done anymore. Maybe a Frenchman/-woman can enlighten us …

Nick Shinn's picture

I'm thinking Alphabet 26 unicase, which seems to be something of a norm. (Caps, with lower case a, e, m, and n--we'll get into Greek and Cyrillic later...)

Unicase reduces the characters in a font.
The Alphabet 26 model also reduces the form of the typeface to a single height, by removing extenders.
Continuing with this paring down, the next step is to remove accents, which upset the clean line.
The Greek all cap setting is generally done without accents.

Perhaps linguistic practice can be organized into categories:

1. Languages without accents--English, Dutch/Afrikaans, Welsh, Uzbek, Filipino, Indonesian, Javanese, Malay, Hausa, Malagasay, Oromiffa, Zulu, etc. (Did I get that right?!)

2. Languages with lowercase and titlecase accents, where it's the norm to do without accents on all cap settings.

3. Languages with lowercase and titlecase accents, where it's acceptable to omit accents on all cap settings in certain circumstances such as display typography--for which unicase is generally intended.

4. Languages where accents are indispensible in all settings.

Florian Hardwig's picture

Okay, I see.
As your aim is to avoid anything that upsets the clean line, I assume German umlauts count here as ‘accents’ too.
Having no umlauts would make such a font unusable in German language (I can only speak for that one), even in display typography only – thus group 4. Why that?
There are a lot of words that – sometimes completely – change their meaning without the dots. Here just a few:

  • abspülen [to do the dishes] — abspulen [to wind (off)]
  • ächten [to proscribe; outlaw] — achten [to respect; esteem]
  • drücken [to press] — drucken [to print]
  • fördern [to promote, further] — fordern [to demand]
  • Händel [quarrel] — Handel [trade]
  • spüren [to feel; sense] — spuren [to obey]
  • stützen [to support] — stutzen [to hesitate; trim]
  • täuschen [to deceive] — tauschen [to change; swap]
  • zählen [to count] — zahlen [to pay]

Right, one can substitute any umlaut by vowel+e – but only in obsolescent e-mail ‘typography’. If you want to broaden your language support, you could try to integrate the dots. Could be an interesting challenge. URW once did ultrabold display versions of a couple of classics, like Gill Sans, Antique Olive and Broadway.

Oh yes, there is room for improvement. ;°)

Thomas Phinney's picture

In general, my philosophy is, would a user reasonably expect these things to work, or at least be happily pleased if they do? If so, it should be in the typeface.

So for my part, when I did unicase alternate small caps for Hypatia Sans, I included every accented variant available with the regular letters. So for example the typeface supports unicase Vietnamese.

The unicase alternates also interact with some of the stylistic sets, so for example the alternate "e" with the slanted crossbar has a unicase version, as does the simpler form of the "a." Both of those need to get all their accented versions, too.

I did however draw the line at polytonic Greek unicase. The Greek unicase is only monotonic. :)

This sort of stuff is a good example of how glyph counts can rapidly spiral out of hand, mind you. Although I think adding these kinds of extras is "good" or makes a "better" typeface, I can totally sympathize with people who choose to draw the line some time earlier.



Nick Shinn's picture

If you make a face with lower case x-height and small caps the same height and same weight, you don't need to create any new glyphs, just make a couple of new OT classes that recombine lc and sc.

An easier way might be to write a unic feature that:
1. Repeats the c2sc lookup
2. Repeats the smcp lookup
(Those two turn any keyed character into a small cap)
3. Substitutes lower case a, e, m and n, along with their accented versions

IMO, small caps should be substantially larger than lower-case x-height for text work, but that isn't a priority for display--which is where unicase is more likely to be used anyway.


Thomas, how is the Hypatia Greek unicase appointed? Please show!

Thomas Phinney's picture

I'm afraid I don't have the temporal bandwidth to do a graphic sample of that right now. Maybe Miguel or somebody else who has a copy of Hypatia Sans will feel like it.

Must go get some sleep now. :/


Syndicate content Syndicate content