What makes a typeface 'sing?'

Ryan Maelhorn's picture

Type design is hard. I laugh at myself about it. How could it possibly be so hard? An A is an A is an A. Plus, I've been making these forms since I was 5 years old. How can this be so hard?

It's amazing how fast you can tell if a typeface is great or not as well. We may disagree between eachother about which ones are best, but it takes no scientific investigation to decide if a typeface is beautiful or not. Your mind becomes made up about this before you can even read the words. It's pretty much instant.

Making a competent typeface is not that hard. At least that's what I have found. You know what I mean by competent, something that is skilfully done, well executed, but yet somehow lacking that 'thing' that makes the best faces sing. The most common adjective for a competent face would be -- boring. It works, it's legible, it holds true to itself, but it isn't interesting, it isn't overtly attractive. It is boring. Bland.

What does one have to do to take a competent face an make it beautiful? why is one face merely competent and another is mind jarringly beautiful? What is this ungraspable thing that makes a typeface sing? I wonder if some of the more experienced designers here would share their thoughts on this.

hrant's picture


But what kind of beauty are you talking about? If you mean the aesthetic kind, to me that's boring.


Bendy's picture

And beautiful faces are not the best text faces.

What I look for in a typeface is to see nuanced design, using tricks I've not seen before, where every tiny decision makes sense and fits with everything else. What else? Originality of thinking/concept and of execution/aesthetics. Accomplished point placement, assured curves. Absolutely regular spacing. Consistency, harmony, tension.

>Making a competent typeface is not that hard.

I wouldn't agree, but maybe we have different ideas of what constitutes a 'competent typeface'.

Ryan Maelhorn's picture

This is an example of what I'm talking about.

It works, it's very legible, blah blah blah blah. There's just something missing from it. Something that would take it from being competent to being truly great. But, what is that something?

hrant's picture

Sorry, that's not remotely "competent" to begin with.


Ryan Maelhorn's picture

No? What makes you feel that way?

Bendy's picture

In that example, there are a number of basic errors. Curves (and straight sections!) are malformed. Strokes and terminals are inconsistent. There are no optical corrections at joins. Proportions between letters are not regular. Heights and widths are all over the place. Spacing is uneven. In short, it hasn't had enough time spent on it, it doesn't show any sensitivity to letterforms. I don't think there's a way to make this design 'great', as it doesn't demonstrate expertise or originality.

Ryan Maelhorn's picture

Bendy, while all you post is quite correct, these critiques are all technical, and can be fixed in a non-artistic, mechanical way. To me personally, even if it was adjusted optically, heights and widths adjusted, etc etc, it still wouldn't really sing. This has to do with something at a much more base level, and that's what I'm chasing after. That's what I'm trying to figure out. That's what so bedeviling, and that's what I'm hoping to learn more about by posting here.

But please don't get it twisted, as they say, I appreciate your comments nonetheless. Thank you.

Bendy's picture

Design critiques are necessarily rational, but attempting to solve design problems through a mechanical approach is likely not to produce better results than what appears above. That's why originality or artfulness come into play, as I intended to mean above.

John Hudson's picture

Look, Ryan, there really isn't a quality of typeface design that can be considered independent of all those things you dismiss as what can be 'mechanically solved'. You can't even begin to approach the qualities of a good typeface, a type that sets and reads well and that attracts the reader, without solving those basic issues of proportion, weight and optical refinement. It is only through those qualities that a type has any chance of 'singing': there is no song independent of them.

cerulean's picture

In short, expertise is competence in greater supply. And everything about letterforms is a product of collective human perception, psychology, and culture, and as such cannot be reduced to a matter of simple math without compromising the effect.

There is no sharp line between craft and creativity. Nobody who asks "where do good ideas come from" has ever gotten a satisfactory answer, which is almost entirely a reflection on the inadequate nature of the question.

Frode Bo Helland's picture

I don’t have the time now, but my guess is an optically corrected, width-revised and evenly spaced, version of the design above could highlight some of the things that makes typefaces “sing”.

enne_son's picture

Paraphrasing "Huit thèses pour (ou contre ?) une semiologie de la peinture" (Hubert Damisch):
[...] le travail [...] sur le signifiant, [...] producteur d'une plus-value.

The "travail sur le signifiant" is done at the optical-grammatical tuning level of co-ordinating proportion, weight, contrast and construction, and the "plus-value" -- achieved through individuation, or adding nuance, on all these dimensions (of proportion, weight, contrast and construction -- is gestural-atmospheric. Where I use gestural-atmospheric and talk of gestural-atmospheric force or singularity, others use personality. The term "atmospheric" can be found in Ovinck, with reference to type, and the term "gestural" in Henk Krijger with refernce to type and typography. For example the assertiveness of Quadraat and the legendary Jan Van Krimpen restraint in type design are gestural-atmospheric qualities.

When choosing a typeface for a job, gestural-atmospheric identity (depending on the subject matter) and optical-grammatical integrity (co-ordinate with the vehicle) are my prime reference or decision axis.

It takes well-resolved optical-grammatical integrity and real gestural atmospheric distinctiveness, singularity and clarity for a type to sing. The "travail" (work) on the signifying components (letters) is at the heart of it.

hrant's picture

Good stuff guys.

Ryan, even if I pretend that there can be a real separation between technique and spirit, and I imagine simply addressing the former, I can't put my finger no what to subsequently add to make it "special". There's no formula, no patch - it has to come from within. Nevermind that there's no universal "special".


dezcom's picture

Whatever may make a typeface "sing" is not an object like vocal chords. Vocal chords make sounds. Some people can make these sounds become song. They do this with a combination of natural talent and hard work. Those with the lesser amount of talent may need to greatly increase the amount of work. If you are looking for an ingredient to identify and quantify and concoct as a magic potion, join those who seek the holy grail in a lifelong search to no avail. If you want to design type, just do the work and then do it again and again and again. If you work hard enough, you may begin to hear a very, very faint melody.

John Hudson's picture

Although I think Chris is extending the singing metaphor a bit far, his comments suggest to me a counter question to Ryan: what constitutes 'song' in typography? Rather than asking what quality or qualities cause a typeface to sing, ask what the song sounds like.

oldnick's picture

What makes a typeface sing?

Well, among all the available voices, it's probably the typeface’s tenor, I suppose.

But, if you’re tone-deaf, it might not sing on key…

quadibloc's picture

This reminds me of the quote from Frederic W. Goudy which I referenced in another thread:

"If there were an individual, readily recognized quality or characteristic which the type designer could incorporate in drawings that would make any one type more beautiful, legible, or distinguished than another, it is obvious that only type of that kind would be designed."

Yes, there are no cheats when it comes to achieving excellence - in a typeface, in music, in TV or movies, in other fields. Even when you do everything right, the result isn't guaranteed to catch the public fancy.

Rob O. Font's picture

Being' on an album cover.

dezcom's picture

So John, how far can I extend a metaphor before it stops singing? ;-)

Nick Shinn's picture

Even when it’s over, the fat lady is still going strong.

Ryan Maelhorn's picture

This has been a great conversation!

I'm moving my design over to the critique section.

Té Rowan's picture

Its resonance with the user.

Nick Shinn's picture

It’s the quality of drawing.
For instance, that’s what distinguishes one revival of old metal faces from another.
Or, in the case of a typeface like Ed Benguiat’s Souvenir, it’s what takes a typeface with a difficult concept that doesn’t sing much, and gives it a strong voice.

dezcom's picture

"Rather than asking what quality or qualities cause a typeface to sing, ask what the song sounds like."

It sounds like one Vandercook clapping ;-)

Charles_borges_de_oliveira's picture

Its the fine details that make a typeface. Anybody can make a circle but add shading and highlights to it makes it stand out. Same with type. By getting your proportions and weights right and to where everything rhythmically flows is one of the keys to making your type less boring. Not sure if any of this made sense.

abattis's picture

Ryan, when you say, "It works, it's very legible, blah blah blah blah" -- well, its the blah blah blah that is just blah blah blah for you that makes this overall, blah. There is 'just something' missing from it, and that thing is "taking care of all the details," the details which you have glossed over :-) Ben named some of the details, surely there are many kinds of details to consider.

What I would suggest is to delete most of the glyphs you have and refine a core set of 'key shape' glyphs to a very high standard of 'fit and finish.' The lowercase "adhesion" is used at the University of Reading, and I suggest you do your own analysis of 'key shapes' that are particular to your design.

Given a limited character set, you can still test a typeface's typesetting performance using a text generator tool - like http://www.adhesiontext.com or http://justanotherfoundry.com/generator. When and only when you think you have something that is singing, should you build out the character set -- because writing is systematic, the tune will carry over.

John Hudson's picture

Good suggestions Dave, although my own approach is a little different, and perhaps better suited to 'singiness'. The test word 'adhesion' is quite good for revealing the characteristics of a design, but I don't tackle the a, e or s until I have first established the rhythm of two faux words: nihilim and millioon. [The latter was used by my first typography instructor when manually kerning headlines: he would first kern the faux word millioon at the intended size, print it, tape it to the top of his monitor and use it as a visual reference when spacing the headline.] I really like nihilim because it is easy to produce and gives a good indication of harmony across some fundamental shapes and their spacing. If that faux word sings, there is a good chance that you'll be able to make the rest of the design join the chorus.

hrant's picture

I personally don't get how not having a descender is OK.

Clearly "hp" is best place to start. :->


dezcom's picture

I don't know Hrant, perhaps hhp is better than just hp ;-)

Bendy's picture

>I personally don't get how not having a descender is OK.

I think it could be to do with character frequencies. /adhesion/ is used as it contains the most common letters so can build many words quickly. I usually add /trlu/ as they're easy and permit more possibilities. Descending letters are less frequent in English; the other thing is they're all different shapes, so take more time to draw. Clearly for Armenian you'd want to start with ascenders and descenders, as those are the more active zones.

hrant's picture

Frequency is certainly highly important*, as is mixing in a good spectrum of shapes, but determining a font's vertical proportions is a key early step so including a single descending letter (probably "p") seems important. So what about "aphesion". :-)

* http://themicrofoundry.com/image/s_rome1-4.gif


John Hudson's picture

In the Latin script, the depth of descenders is something to be considered in proportion to what is happening above the baseline, which is more complex and has a far greater overall impact on the typeface, involving as it does as least two different strongly information bearing zones (x-height, cap height and possible independent ascender height). The only factor that in any way complicates the descender depth is the density of the g, and even then the norm is for that density to be optically adjusted to fit the letter within proportions that look balanced to the (generally slightly longer) ascender height. So while I agree that determining a type's vertical proportions is a key step, the role of descenders within that determination is relatively minor, and overall the determination of horizontal proportions in the x-height and cap height is massively more important.

hrant's picture

I think in the overall/"long-term" texture of a setting the descenders actually play a bigger role than that, especially for short measures (read: minimal leading). And although I myself do tend to be predictable in terms of the relationship between ascenders and descenders, this is clearly not true in the aggregate of all designers. For example when looking at Ruse (by Gerrit Noordzij) noticing that the descenders are fully equal to the ascenders is quite important I think.

> (x-height, cap height and possible independent ascender height)

Well, "adhesion" doesn't have a cap either. BTW, to me descenders are more important than caps.

But you're right about the "g" of course (and it's something I myself have brought up often). So "aghesion" it is. :->


abattis's picture

I was just writing up my view on this topic at the weekend for the www.craftingtype.com workshop's blog at http://workturn.tumblr.com and the below is taken mostly from what I wrote :-)

nihilim and millioon! A tasty morsel of knowledge there! haha :-)

I agree that adhesion is not, in fact, what I personally recommend. I just note that's what Reading used when I was there. As I said above, I think a type designer should do their own analysis of 'key shapes' that are particular to their design, to figure out which should be done first to establish parameters of the shape system, and which should be done after to follow the parameters already defined.

So, looking at nihilim and millioon, long words of few characters: n i h l m o

I would say that the first one to draw is the o, and the n.

The first letter to draw is the ‘o.’ This lowercase letters is at the heart of the European ‘latin’ writing system, containing the “DNA” for everything else. The ‘o’ establishes the most fundamental proportion, of width to height. Since its shape is pure curve, you can clearly imagine the pen or writing tool used to make this shape - how fast it was used, what kind of nib and ink it had - and see the variation between thick and thin pen strokes. These attributes of the mark-marking tool will inform the drawing of every shape.

The second letter to draw is the ‘n,’ and the writing tool informs the way the arch joins the left stem to the right stem, and gives the two stems an equal width.

With these two together in a row of repeating letters, a clear horizontal rhythm between the black forms and the white counter forms emerges. The inter-letter spacing of these two is symmetrical, and harmonises easily with the large white interior shapes (or “counters.”)

So with just these two, you can prototype the basic rhythm of text.

With them in place, you can easily build out the rest of the characters in Johns words if you wish - h and l, especially if the arch and right stem is drawn as a seperate bezier contour - are very easy, no more drawing required. Just move the left stem stroke entry up, and for l remove the right contour. For the i just remove that right contour and add a dot. The 'm' is a little more complicated because you can't just copy and paste the arch, it needs fiddling to join in a way that looks right. Then you can set up the left and and right bearings for all of those the same as the 'n,' and see if you hear a tune.

Looking at words like nihilim and millioon in the metrics window is useful until you do; but its a small step from there to pop them into a text generator, and set a large amount of text to see the effect in a textblock.

But actually, I wouldn't recommend building out those characters, because they are so similar. If you decide to radically change something in the n, you have to go and redo all that. And probably when you move on to less similar, more different forms, you'll want to harmonise what you settle on for those with the curves in these initial letters. My next suggestion? The 'two story' a. Its neck curve at the top and belly curve at the bottom are sure to have an impact on the 'n' (and the rest, if you made 'em.)

hrant's picture

Very good advice... for a text face.

For a display face I would say start with things like the "a" and "g", because nobody will spend time gawking at an "o".


John Hudson's picture

While I recall beginning the design of Gabriola with the lowercase a, it needs to be borne in mind that one can't finalise that letter -- or any others -- until the harmonic proportions of n and o and their spacing are worked out. So for a display face I recommend working up some characteristic shapes that will give the face its flavour and them setting them aside while one concentrates on the foundation.

John Hudson's picture

Dave, I tend to do i n o first, and my first test word is, hence, 'onion'. Only fairly recently did I remenber that 'onion' was also the practice word my mother's calligraphy teachers had used to introduce Johnston's foundational hand.

I suspect my inclusion of the lowercase i in this initial set explains why my dots tend to echo the outer shape of the o so clearly.

hrant's picture

until the harmonic proportions of n and o and their spacing are worked out.

Certainly things are more iterative than that. Especially in a display face (but really to some extent even in a text face) something like the "a" can often affect what even the "n" and "o" end up looking like. No single letter is the boss.


abattis's picture

hhp: I think we 3 are all in agreement here :)

jh: o n i... hmm, I like the idea that the i presents the opportunity to repeat the o shape, and thus immediately exercise shape systematization thinking in the design process, so I now present my current thinking on what to draw first (for, as Hrant said, a text type)

o n i a e s H O A R g k w WM 83

Syndicate content Syndicate content