Typophile - What would Ross Mills say? - Comments

> I try to maintain

hrant — Mon, 13 Mar 2006 01:29:41 -0800

> I try to maintain stylistic independence between the two scripts

Which is what I’ve tried to do with the pair of masters; although only as much as is needed within the confines of authenticity for each script - which really isn’t huge in the case of Armenian & Latin: mostly slant and serifness (especially the latter) need to be different.

> clients who want to pay for a multiscript typeface don’t want
> to pay for a separate multiscript typeface for each script

While some clients don’t want to pay for typefaces, period. :-/
This is really just another case of education, and you win
some you lose some. But anyway I sell the various components
of N&P individually: one client might want just the two masters,
another might want the Latinocentric half (for some good ol’
chauvinistic setting :-) and I can even envision somebody only
buying the Latin subordinate part of the Armenocentric set!
It’s my personal favorite actually, and I wouldn’t mind seeing
an entire book set in it. It would be unique in style and highly
readable. I promise.

> I try to treat neither script as stylistically dominant

Again that’s something I’ve tried to do (or actually, incorporate into the system) with N&P, although some people (including the CR jury) might see a strong Armenian flavor throughout. I’m not sure that isn’t more hindsight than insight though. If you show Patria to somebody, will he say “oh, it looks Armenian”? No way. But maybe he’ll say “it looks foreign”? That’s not impossible, but I think it’s unlikely, mostly because people are now used to seeing all kinds of strange fonts; Patria is pretty mainstream in comparison.

hhp

Adam: your Ethopic typeface

John Hudson — Sun, 12 Mar 2006 20:43:35 -0800

Adam: your Ethopic typeface can also be called “matched bi-alphabet face”?

Yes, but I thought Nick was asking specifically about matching with Canadian aboriginal syllabics.

In Nyala, the Ethiopic typeface, the Latin exists in a supporting role to the Ethiopic. So it might be seen as one half of what Hrant advocates: a typeface in which one script is clearly dominant in determining the style and the other subservient to it. The Nyala Latin was designed for the specific purpose of setting some Latin words within Ethiopic text, based on examples of such documents I had examined which typically used e.g. Times or similarly ill-matched design for the Latin. If I were intending to set just a few words of an Ethiopic language in the midst of Latin text, I would not use Nyala for the Latin because I don’t think it is an appropriate Latin text face (Gary Munch remarks that it has some merit as a Latin display face independent of the Ethiopic). So it isn’t really a bi-matched system in the same way that Ross’ Euphemia is, but the Canadian syllabics lend themselves very easily to matching a Latin sans serif.

Nyala is something of an unusual approach for me, because the Ethiopic script so obviously has dominance over the Latin. More often, I try to maintain stylistic independence between the two scripts, and address harmony in terms of weight and proportion, aiming for matching typographic colour as far as possible without compromising either script. The reason for this is pragmatic: clients who want to pay for a multiscript typeface don’t want to pay for a separate multiscript typeface for each script, as required by Hrant’s Nour/Patria system. So I try to treat neither script as stylistically dominant (except, sometimes, where this decision has been taken independently of my design, e.g. 300 years ago by Peter the Great).

> you’re on to

hrant — Sun, 12 Mar 2006 20:43:24 -0800

> you’re on to something

Thank you.
But yes, there’s a lot more to do/add/fix.
Not least the serifs on the Latin caps.

> the lightest weight of Patria Armenian is very classic

Actually, it’s classic Latinization. :-/
But that’s OK here because it’s a subordinate.

> it’s noticeably lower than the Latin x-height.

1) Even though the Latin has no strong x-line horizonals?...
2) It’s noticeable when you zoom in (and you’re a type designer), but not when you’re immersed (and you’re a layman).

> The Armenian looks smaller

1) Don’t base it all on a couple of samples - that’s too deterministic; for one thing “dearest” contains fully 5 (out of 7) instances of glyphs that are unusually wide in Patria. But even in that case you might find that if you focus beyond the plane of the text, the apparent sizes come sufficiently* close (not least because Latin’s paucity of extenders*). I couldn’t say how many man-months of squinting and de-focusing I did before arriving at these relative proportions. :-)
2) Again, there’s a perceptual and cognitive difference between deliberation and immersion.

* See below.

> You’ve not met your “apparent size” requirement.

Of course not absolutely.
It’s a balance, a dance of gentle gives and takes.
And the same with the other factors I mentioned.

> It’s a misconception that book faces require small x-height.

I don’t know what you mean by “small”, but it’s
certainly no misconception that the extenders
are more than just pretty appendages.

The more you’re shooting for immersive reading, and the larger
the point size, the smaller the x-height should be (up to a point).

> a type’s apparent size is determined almost completely by it’s x-height!

Nope. Especially in something like Armenian, with its rich extender space (or even Polish, with its rich diacritics) the x-height is certainly the most important, but you can’t simply ignore the rest. Even in English the ascenders with their frequency and the descenders with their heft do make a difference. And this is where the notably subtle x-height changes in N&P come from.

> although perhaps not in Georgian — but that’s the exception

Exception? No such thing. It’s just a pushier case. Armenian is just like Georgian in that way, simply less extreme. So any anti-Regularity decisions that apply to Georgian also apply -to a lesser extent- to Armenian (and everything else). Like I said, more important than where the threshold might be is that there is a threshold at all.

> No loss.

(I think you missed my point.)
If there’s no loss, then there’s no reason to blindly align things.

hhp

Hrant, you’re on to

Nick Shinn — Sun, 12 Mar 2006 20:15:08 -0800

Hrant, you’re on to something with your “doubly bi-lateral” approach, and the component parts are fine (the lightest weight of Patria Armenian is very classic), but you haven’t delivered everything you claim and there are still some family design issues to resolve.

The Armenian “x”-height of Nour/Patria is clearly indicated by serifs (more so than its baseline), and it’s noticeably lower than the Latin x-height. The Armenian looks smaller and more condensed (compare the rhythm of vertical strokes), while the Latin is larger and rounder. You’ve not met your “apparent size” requirement. Rather than dethroning Formal Regularity, what you have is two out-of-sync regularities.

the particular usage of the non-x-height space varies greatly

It’s not about the non-x-height. It’s an axiom of typography that a type’s apparent size is determined almost completely by it’s x-height! In Latin, Greek, Cyrillic, Armenian, etc., (although perhaps not in Georgian — but that’s the exception).

sure, use Quadraat Headliner for a book

It’s a misconception that book faces require small x-height. Dante, for instance, or Cartier Book, have ample x-height, while Bernhard Modern, with a tiny x-height, is more of a display face. Small text fonts (eg ITC Bodoni Six) have exaggerated x-heights.

Even when a multi-script font is used to set single scripts at a time, what’s the loss if it happens to also be suitable for proper multi-script setting?

No loss. When Western foundries publish a typeface with Latin and Cyrillic scripts, they’re making the design available for use in Russia, Ukraine, etc., in situations which will usually not involve bilingualism with Latin script.

Or you could say a deeper

hrant — Sun, 12 Mar 2006 19:17:21 -0800

Or you could say a deeper order. Anyway, I agree.
My contention is that as a rule “formal regularity”
has been limited entirely to things like lining up
x-heights and copy-pasting serifs.

hhp

hpp: “…the particular

enne_son — Sun, 12 Mar 2006 19:02:16 -0800

hpp: “...the particular usage of the non-x-height space varies greatly, affecting apparent size. So for example for the former the Latin will look unusually large (which it does) while in the latter it will look too small (since the Armenia extenders are much stronger). Talk about disrupting immersive reading (not to mention unwittingly conveying to the reader that one script is more “important” than the other)”

This sounds very, very right to me from a perceptual processing point of view. But the argument about bringing down formal regularity is a non-starter because compatability of apparent size across different structural-mode deployments of cartesian space is simply a formal-functional or optical-grammatical criterion of a higher, more inclusive and demanding order.

Actually, there are many

hrant — Sun, 12 Mar 2006 17:53:48 -0800

Actually, there are many more than two, but I agree that those
two cases (which I call “parallel” and “hierarchic” in my article)
are the main ones.

Just FYI though, a third type of bilingual document, which is
actually quite prevalent, is the dictionary, and that tends to be
much more complex than the “main” two. Even a system like
N&P’s might have trouble keeping up, but of course it will
manage much better than the one-dimensional stuff.

> The requirement for the latter is most definitely
> to have a common x-height across scripts

Absolutely not. Why not? Because if you do that the
other-script snippet can look unusually small or large.

The fact that guidelines like the x-height are not in fact
populated by straight horizontal lines means that there’s
actually a lot less strict alignment necessary, and instead
things like apparent size need to be weighted heavily.

The five main factors that need to be balanced against
each other very carefully in a multi-script typeface are:
authenticity, readability, even color, apparent size, and
regularity (like alignment, but also stylistic consistency).

> the Latin glyphs can be tailored to the requirements of the other script.

If you ignore that a font’s vertical proportions affect its
readability (at a given point size), sure, use Quadraat
Headliner for a book... :-/

Precedent of expectation is only a small factor in actual functionality.

> Big x-height for use with Cyrillic, small for use with Armenian.

Which however totally doesn’t work, because the particular
usage of the non-x-height space varies greatly, affecting
apparent size. So for example for the former the Latin will
look unusually large (which it does) while in the latter it
will look too small (since the Armenia extenders are much
stronger). Talk about disrupting immersive reading (not to
mention unwittingly conveying to the reader that one script
is more “important” than the other).

> perhaps bilingual typography is not an important reason for multi-script typefaces

1) Of course it is. Everywhere from the -expanding- European
Union to cultures with “secondary” scripts trying to keep up.
2) Even when a multi-script font is used to set single scripts at a time,
what’s the loss if it happens to also be suitable for proper multi-script
setting? The answer I guess is the Modernist-coziness of the designer.

—

What’s the solution?
Formal Regularity needs to be brought down from its position on the throne.

hhp

>However, in general perhaps

sii — Sun, 12 Mar 2006 17:29:09 -0800

>However, in general perhaps bilingual typography is not an important reason for multi-script typefaces,

This is likely true, but these days its rare for a non-Latin font to be produced which doesn’t include a matched or custom Latin. Users have come to expect at least Code Page 1252 coverage. There can be technical and performance issues in fonts with no Latin set.

Re the x-height problem:

Nick Shinn — Sun, 12 Mar 2006 15:57:32 -0800

Re the x-height problem: there are two kinds of bilingual document.
One is where the same content is repeated in side by side columns, translated.
The other is where a word or two of a foreign script is inserted into the text.

A common x-height is not required for the first, and as Hrant argues, may give undue weight to one language over the other.

The requirement for the latter is most definitely to have a common x-height across scripts, if the typographer is going to the trouble of using one typeface.

How can this situation be resolved?

There is sufficient variety of x-height types in Latin script that the Latin glyphs can be tailored to the requirements of the other script.

So a multi-script typeface should probably have x-height options for Latin script. Big x-height for use with Cyrillic, small for use with Armenian.

However, in general perhaps bilingual typography is not an important reason for multi-script typefaces, which exist to make a typeface available to different alphabet users, and will rarely be used in bilingual documents where different scripts will be thrust up against one another?

Chester, it is precisely

hrant — Sun, 12 Mar 2006 14:33:07 -0800

Chester, it is precisely what I think users need that motivated me to create N&P. Not all users certainly, but actually users that are often ignored. Furthermore, it’s important to realize that the sociocultural reality and needs of users for whose script you’re making a font for can be -and usually are- different depending on the script. An Armenian person who might be interested in N&P is quite likely to have different priorities than let’s say a Brit who’s interested. Different cultures need their fonts to do different things for them. And taking this sort of thing into account is Good Design.

Here’s a good overview of the system:
http://www.themicrofoundry.com/other/nour&patria/N&Psys.gif

> I just wonder whether the users of the system are using it as is intended.

Ah, the ever-elusive Intent. :-) You probably know even better than me that the original intent of a type designer is the first thing to go when the font hits the road. Which is OK, really - I’m a pragmatist too, and I simply can’t angry over things I can’t affect.

That said, I certainly do have an intent for the thing. But it’s also way too early to know if most users will “get it”, since Nour has only really been used once so far (in Emily Artinian’s book that I mentioned in the CR awards thread). One thing that helps though is that for somebody (especially an Armenian person) to go looking for something like this, to find it, and pay good money for it, it’s quite likely that the intent will not only get through, but resonate strongly with the customer. N&P isn’t something most people would buy just for its looks (it doesn’t even look that impressive formally I don’t think - although Patria customers have in fact told me that they like its somewhat rigid but sensitive forms - which is actually the intent). This is fortunately exactly what happened with Emily: she looked for 2-3 months, paid the relatively high price pronto, and her tesimonials are the juiciest things a type designer could ever want! :-)

> should all be designed to hold together æsthetically.

Agreed. But of course, and as you probably agree, there are other
factors besides aethetics that needs to be balanced in as well.

What’s encouraging to me is that you seem to think that N&P does
in fact enjoy this stylistic cohesiveness. I’m very grateful for
that because that half of the coin is difficult for me to gauge,
and the other half (functional authenticity) I already think is
working to satisfaction. So thanks!

> I believe that this was the approach in developing the
> new typefaces for Microsoft’s Longhorn/Vista OS.

And this sort of sensitivity is indeed highly welcome, and a recent phenomenon (thankfully gradually overpowering the heady cultural chauvinism of “70s-style” non-Latin design). Nonetheless, the one-font-per-script approach is a damaging over-simplification.

Nick, the bulk of your questions are in fact addressed fully in my article. Which is not to say you’re likely to appreciate the answers! :-) Just FYI though: Spatium #4 is sold out; but Hyphen there’s plenty of, and your interest in Greek type might provide extra motivation. Basically, if you’re asking these questions because you’re interested in the answers (or at least some answers) then my article should help.

But since the article is not (yet) available publicly,
I will try to briefly address some of your points here.

> what this typeface system does is offer stylistic alternates of the “other” script.

Style is exacly what I’ve tried too look beyond,
not to ignore it certainly, but to balance it in
a way that users need.

> doesn’t offer Armenian typographers as much
> to work with as their Latin counterparts

Certainly there’s room for expansion (there always is). But pointing to the lack of an italic for Nour ignores a basic fact of Armenian typesetting: thanks to our floating emphasis mark, we don’t need italics at all the same way that Latin users do. So sure, it would be nice to have (and I don’t think it’s Latinization) but sort of like smallcaps is to Latin. Before I make italics for Nour, I would make different optical sizes.

Asmik? The inclusion of an italic doesn’t make up for its hopelessly Latinized vertical proportions. I know, because that’s exactly the type of Armenian fonts I used to make. :-/

> Constantia (John Hudson) does a good job of being “doubly bi-lateral”

No, that’s not what it means.

> a puzzling inconsistency between scripts: different x-heights.

That is, of course, exactly a feature, not a bug. Imposing strictly uniform vertical proportions across scripts is nothing short of cultural and functional suicide. The variation of vertical proportions certainly can’t go unchecked; but neither should it be shunned. And it is in fact partly for reasons of immersive readability than vertical proportions usually cannot be imposed across scripts.

Take for example Georgian. Would you make its vertical proportions match Latin, even though only 4 out of its 33 letters are confined to the x-height? And if you don’t impose vertical proportions in a Georgian/Latin system, where’s the threshold? Do you impose it for Armenian/Latin or not? Or what about the big two: Cyrillic (with is extender-starved lc) and Greek? Actually, the determintion of a threshold is moot; the important thing is that a threshold does exist.

In fact I would suggest (and actually already have) that even within the Latin script different languages ideally need different proportions - at least in terms of the Em space: Polish for example, with its heavy dose of diacritics, is probably better served set at a slightly smaller point size (but with the same line-to-line leading) when set parallel to English text.

hhp

there’s only one such

Nick Shinn — Sun, 12 Mar 2006 10:12:09 -0800

there’s only one such system on the planet, for any pair of scripts: Nour&Patria.

It is a unique approach, but it’s kind of “cutting off one’s nose to spite one’s face”.
With the premise that a single Latin style and a matched single Armenian style is not possible (a political statement, if not a defeatist design philosophy), what this typeface system does is offer stylistic alternates of the “other” script.

So, in OpenType terms, Nour = A (Armenian) + L.salt (Latin stylistic alternate)
Patria = L + A.salt (Armenian stylisitic alternate).

It may be politically sensitive, but the result is a complex system, a variation of the “Roman + Italic + Caption” style categories in Fairmont, that doesn’t offer Armenian typographers as much to work with as their Latin counterparts, by not having enough “design-space” to provide a clearly-contrasted Roman + Italic. Armenian is conceived of as the “Caption” element in the triumvirate. The type designer’s message seems to be that the notion of Roman/Italic contrast is un-Armenian, unacceptable Latinization.

Do Armenian typographers not use the layout convention of contrasting “Roman” against “Italic”? A face like Asmik would suggest otherwise.

I would imagine that typographers setting Armenian would be inclined to use the B weight of Patria as “Roman”, and the A weight of Nour as “Italic”.

For a text face, the serifed Cleartype font Constantia (John Hudson) does a good job of being “doubly bi-lateral” in the relationship between Latin and Cyrillic. What makes the difference is, surely, starting both scripts at the same time, with a single, comprehensive design space. Not fragmentation.

As a bilingual system, Nour or Patria has a puzzling inconsistency between scripts: different x-heights. What happens when a word or phrase of one script is inserted into a paragraph of the other? Surely the abrupt change of x-height will disrupt immersive reading?

Hrant, fair enough. I guess

Sun, 12 Mar 2006 09:47:47 -0800

Hrant, fair enough. I guess I’m just not up on your Patria/Nour project, or the ideas behind it. I’m just such a pragmatist - and you have to be as a type publisher, even if being a type designer is more “romantic” - that I wonder how users work with the fonts.

I gather that there are four fonts in the grid: Pure Latin / Pure Armenian / Latin with an Armenian accent / Armenian with a Latin accent. I understand the concept, and I appreciate it; I just wonder whether the users of the system are using it as is intended.

And again, perhaps I’m a dirty modernist, but it seems to me that a type system or multiple scripts — or a single big-ass OT font conatining multiple scripts — should all be designed to hold together æsthetically. Often, additional scripts are just that: additional. They have been designed after the Latin script, and so are beholden to the Latin. When a project has been conceived as an integrated, cross-referenced, cohesive multi-script face, that seems like the ideal approach. And I believe that this was the approach in developing the new typefaces for Microsoft’s Longhorn/Vista OS.

Hrant, from my brief perusal of CR’s type contest issue, it seems that the Patria/Nour system is very handsome and very cohesive, and I compliment you on it. I’d love to see a PDF, if you have one posted on your site somewhere.

So, excuse my earlier dig. I couldn’t help myself...

PS; what were we talking about? Oh yeah: Inuktitut syllabics...

OK, OK, you can tack on

hrant — Sun, 12 Mar 2006 09:28:11 -0800

OK, OK, you can tack on “as far as I know”. But really that’s about it. N&P is the only “known system” shall we say that has a double pair of masters and subordinates that makes it a balanced, versatile system. Without such multilateralism a multiscipt font system is one-dimensional, allowing for authentic, functional setting in all its scripts only in very narrow applications. Not that I’m not trying to spread the word: my article spills all and has been published in two journals (so far); I’ve given a talk about it at two conferences; and I preach one-to-one as well, like when Michail Semoglu showed me his beautiful reverse-Latinized design, and I suggested he do it both ways, not just one.

And perhaps it’s not even a very big claim, for the simple reason that many people think it’s not worth the trouble, or even that it’s a bad idea. Modernism is the dominant ideology and it tells people that a solution like mine is “too complex”; the most Modernism can apparently be happy with is the one-font-per-script concept (and really, it would much rather have everybody just use the Latin script). While the reason I think it’s just complex enough is simply that multiculturalism is itself highly complex!

> Has Matthew Carter done anything that could
> be called a “matched bi-alphabet face”?

As far as I know the most Carter has done is the Cadmus/Skia
pair, which is however a superb example of reverse-Latinization.

http://www.themicrofoundry.com/other/cadmuskia.gif

> Even if Hrant wasn’t the first to suggest this

I would of course really love to hear of any conceptual
predecessor, especially since I value raw ideas so much.

hhp

As I recall the ‘system’

sii — Sun, 12 Mar 2006 09:25:22 -0800

As I recall the ’system’ requires two families, say you want to support Latin and Greek. You’d have two families in the system. Family 1 would be for setting mostly Latin with a few Greek words and would be a Latinized Greek and a second family for setting mostly Greek with a few English words - that would include a Greekized Latin. Even if Hrant wasn’t the first to suggest this I think he was the first to make a ’system’ that works this way.

Adam, would that make it the

Sun, 12 Mar 2006 09:11:02 -0800

Adam, would that make it the second only such system on the planet? Surely there can’t be two only such systems. I think the only way to resolve this is the Man’s Way: arm-wrestling. If anyone else pops up with another only such system, we’ll have to have a round-robin tournament. Has Matthew Carter done anything that could be called a “matched bi-alphabet face”? Because I’d put money on Matthew...

What would Ross Mills say?

sii — Tue, 7 Mar 2006 21:56:59 -0800

Jeremy Tankard’s latest...

http://www.inuitpaper.com/arjowigginsinuitfont.htm

Cheers, Si