Do ligatures improve readability?

ebensorkin's picture

Do ligatures improve readability? When I ask this I am thinking of ff fl fi etc rather than st or ct the later being a little distracting I think. I ask this because I had been thinking as an article of faith you might say that certain ligatures ( assuming a ligature design that is not showy or attention getting ) simply did make better word forms and hence generally speaking better & more pleasant immersive reading.

But think about it again I realize that all I have is my gut for this.

What do you guys think?

BTW: Here are some threads & links with related info/ideas. They don't actually deal withis question directly but they are ligature related.

Ligatures in OpenType: Discretionary vs. Standard
Usage of Ligatures in Corporate Font and Business Letters?
OT Ligatures > calt or liga?
Ligatures

https://listserv.heanet.ie/cgi-bin/wa?A2=ind9604&L=typo-l&T=0&P=3048

hrant's picture

I don't think it would be crazy to try to measure readability outside the lab. It would take longer, and some really oldschool -I'm talking ancient- cleverness.

BTW, Kevin and I actually seem to be in synch concerning comfort/speed.

> How does asserting ontological primacy of comfort help us?

For one thing it would stop certain people from trying to sway a possible productive consensus by saying that we're wasting our time because we're going on about finishing a book half an hour sooner and that's dumb.

Progress is in your head, and people can play head games with you.

hhp

ebensorkin's picture

For those who want to get more practical about this Notan stuff I have started a new thread to post & discuss examples of typographic Notan here:

http://typophile.com/node/29362

enne_son's picture

"It seems to me that speed can never be ignored. I can see a very clear distinction between efficiency (speed at which letter shapes/word forms/word clusters are correctly recognized) and efficacy (speed at which comprehension occurs), but speed is a component of both." [Erik]

Right, speed can never be ignored. In science it's a measure, or an index. And as an index it's a useful tool to indicate if the variation of the variable that is being varied makes a difference. I'm not trying to unseat the value and validity of that.

Relative to your "I can see..." I might want to add: ...and both are a component of speed.

degregorio's picture

http://letritas.blogspot.com/2006/05/10-consejos-al-trabajar-con-ligadur...

En mi humilde opinión, hay claves que pueden potenciar la legibilidad, pero en una ligadura la estética es muy importante.

In my humble opinion, there are elements that can improve the legibility, but in ligature the aesthetics is very important.

Rob O. Font's picture

Eben:
> David, Ligatures don’t cause speed to increase - they avoid slowing the reading speed

Uh oh, Hrant.
> I can’t believe I have to point out that this is completely bonkers. There is
> no “overall design” - it’s all the overall design. After all this abstractive
> progress we’ve been making, please don’t regress into textbook classicist
> compartmentalization

I'm not compartmentalizing idly. In order to answer the question originally posed, ligatures had first to be defined as "required", and then re-compartmented with other typography used in the interest of defeating distraction while assisting comprehend-ability. I will never use "speed" again: Ligatures don’t cause comprehend-ability to increase - they avoid retarding it.

They are not alone, A-Z, Dotless-i, $, -, and more, are used in the same overallness, or something. I'm hoping, that Hrant's point is — from the first line of the first letter of the letter drawing, to the last minute detail of the last detail of the composition is the "all", and any disturbance of the all is as good as many disturbances?

hrant's picture

What I was opposing was the perception that there's this thing that only uses a prefab set of discrete parts and is read at some sort of maximum, and anything added just has to be careful not to mess things up. This could only be true for computers, not humans. Otherwise we should be using Morse Code.

hhp

ebensorkin's picture

Ligatures don’t cause comprehend-ability to increase - they avoid retarding it. David, this is a sweeter way of putting it. It's more lucid & less baggage laden. I would amend this further in this way: 'Well designed ligatures included in typefaces in those cases where they are actually needed don’t cause comprehend-ability to increase - they avoid retarding it.'

To get past Hrant's objection that you are creating a model of reading with the statement you might offer the corrolary that 'In those typefaces that need ligatures you would see comprehend-ability to increase when well designed ligatures are included."

and any disturbance of the all is as good as many disturbances?
I am sure you meant something by this but I think the langauage actually written can't get me closer to what you meant. I would be willing to say that any distubance of the 'all' is relevant. Is that what you meant? Hrant, was this all on the page in your mind or does it include the possible sound of a child in the room?

read at some sort of maximum

I think you are absolutely right to insist on avoiding that 'model' of reading.

I think what I was getting at is that any design ( for anything ) has a built in 'maximal' potential use ( defined in whatever way you like ) but that that maximum is just potential. It doesn't mean that it is desirable to reach per se, or even 'reach for' in real life use. That also does not deny that achieving certain testable maximums might not enhance comfort. But your main point if I get you, I agree with - a test ( even a theoretical one) is not a valid theoretical model of reading in itself.

enne_son's picture

‘Well designed ligatures included in typefaces in those cases where they are actually needed don’t cause comprehend-ability to increase - they avoid retarding it.’

Eben, why think mainly in terms of comprehension here? Why not think in perceptual processing terms of the avoidance of retarding the efficiency of visual wordform resolution and efficacy of appropriate salience mapping to that end?

ebensorkin's picture

When ligatures work ( or don't) I think the impacts aught to be thought about perceptual processing terms to be sure, and in terms of comprehend-ability and in terms of simple measures such as speed, and ... many many things. There are lots of ways of slicing this.

So as far as I can tell I don't disagree with you on that.

So; far from advocating one particular way of slicing this - my point had to do with the unusualy contingent or accutely provisional context in which ligatures are designed. That being, ligatures may be needed: badly, a little, or not at all. In contrast, an 'A' or 'a' of some kind is always needed. And ligatures themselves are suited to their font's actual needs (or not) by degrees. Even though I liked David's axiom very much in some ways - these factors seemed left out of it.

That's the problem with saying something pithy & axionmatic about ligatures. All the factors count and there are almost too many of them.

hrant's picture

> "In those typefaces that need ligatures you would see
> comprehend-ability to increase when well designed
> ligatures are included."

Ah, but here's the -or actually just one- rub:
A text font, especially a bookish one, is itself not well-designed (for example I might predict that the "f" is too shy, or that the spacing is too loose) if it doesn't need ligatures! It might be a necessary "technical" compromise, but that makes it not "maximal" as this line of reasoning assumes.

--

Eben, knowing your "plans" for ligature integration, I don't understand how you can be so cavalier about accepting "negativist" opinions about them.

hhp

ebensorkin's picture

if it doesn’t need ligatures! I think the point I was making is separate because I was thinking ( assumptions at work - look out! ) of display & text at that time where as you are always in text mode. ;-) Also that phrase does not include references to maximality per se -just to improvement.

But to address your point, if I get you: I have not firmly concluded that what your saying about text faces is utterly true. But my very strong suspicion is that in spirit what you are is quite correct. Partly it's because I know I have alot more looking and study to do before I could possibly have seen for myself that this is so. But in addition to that I am still keen on the idea that ligatures are just one kind of contextual alteration that could be applied to improve text faces.

They are the most obvious because they seem to break the 'rule' that glyphs don't touch in Latin. They might also be the most gross in terms of the size or scope of the alteration. I have not assembled a good list of contextual alterations yet so again, I am working on it.

But again, given the options of ligatures ( or no ) in a text face with and assuming the options we are conventionally aware of, I think I would simply agree. That partly because I think ligatures achieve a better notanic relationship than the design solutions created to avoid needling ligatures do.

I am also starting to characterise what I think is going on there - I think ligatures offer a richer or more complex notanic relationship, and that is maybe one of the reasons they are valuable to a text face.

But in terms of my own theory with specific emphasis on text faces; the ideas fighting for equity in my mind now are wheather the ligature is perhaps great but flawed approach. Maybe in a 'perfect' font a ligature is a little too inflexible a solution. Maybe ligatures are a kind of protrotype alteration that hints at a better way.

The rival idea is that ligatures are just one of a broader range of partially undiscovered/documented pro-notanic alterations that are possible the 'rightness' of which are purely need/context based. As of today the later idea is winning.

The underlying question that separates these two ideas the degree to which a type designer can ever hope to control the way in which the face is used. Obviously there are lots of instances in which fonts are basically abused. But in general there is a trend towards font awareness and appreciation, care, respect, etc. And in the case of a special, perhaps bespoke face for immersive reading, [ Hi Tiff! there's your word ;-) ] maybe the type designer could start to expand their influence to issues of leading, word space & so on. In the case where the type designer succeeds in that the place of ligatures is stronger. If you are designing with robustness or 'abuse resistance' in mind this is less true.

Anyway, thats the way the shades of grey are swirling for me at this time.

William Berkson's picture

Here's what I wanted to say all along, which also may be what Hrant is driving at above.

Ligatures are I think in principle a liability for reading in a separated script like latin script. However, the overhanging f is a nice differentiating feature in seriffed faces. It is enough of a plus that when you subtract out the slight negativity of introducting ligatures to avoid crashes, it still comes out a plus. So ligatures are a kind of compromise that is overall beneficial, but a compromise none the less.

So f-ligatures, good. Other ligatures (st, ct etc.) bad in text as they are not necessary to compensate for problem crashes. Of course, in display you can have Zapfino or whatever, but that is a horse of a different color.

ebensorkin's picture

Other ligatures (st, ct etc.) bad.

At the time they were invented the st ligature made sense of course because it was for the long s we no longer use. People probably had afffection or familarity for the form for a long time so it lingered beyond the point of visual utility - because it had been made into a cultural-visual cue. But then the eye prevailed.

I don't have a similar idea about the ct. Does it stem from a blackletter utility perhaps?

Rob O. Font's picture

"Ligatures are I think in principle a liability for reading in a separated script like latin script"
This is just an unfortunate mix of terms. :)
Ligatures are in principle a liability for reading in separated styles of the latin script.
Making room for, absence of ligatures being a liability for reading in ligated styles (connecting scripts), of the latin script, as such absences can be mistaken for a space, or make connections ungainly.

Rob O. Font's picture

I? also jus red this:
"What I was opposing was the perception that there’s this thing that only uses a prefab set of discrete parts and is read at some sort of maximum, and anything added just has to be careful not to mess things up. This could only be true for computers, not humans. Otherwise we should be using Morse Code"

Oh? I believe for each instance of use, there is a prefab set of discrete parts that may be composed and read at a maximum, for a given audience, with range of audience maxima enabled by dynamic control over some of the discrete parts. This is true 4 humans as we do use morse code if that is the perfect set of discrete parts that may be read at a maximum for a given audience. For most uses, however, it is so low in the "tone of voice" department, we use at least Arial. Yet this Arial m or whatever our computer shows youthis text in, comes to you via litt'le blinking lights, eventually exciting other litt'le lights, much closer to morse than Snell Roundhand, e.g. to ligate to my previous post.

hrant's picture

Eben, your first post of 11/16 was very good.

Latin is only a "separated" script formally. That's fine for teaching it, not fine (and not real) for reading it. Not using ligatures is what's a[n arbitrary] compromise. Even an "r" touching a "y" is a ligature, and that's a common, conscious, pragmatically intelligent "compromise". But really, when you take into account notan, everything is a ligature.

> ... for each instance of use ...

Fine, but then we disagree about what instances actually apply to when/whom.

hhp

ebensorkin's picture

very good Thanks.

BTW. I was noticing some thin ct & st ligatures the other day. They were not so thin as to appear anemic but at the same time they were far less obtrusive than most of the others I have seen. I thought, I could read that. This loooks RIGHT. That would be pretty comfortable! I could live with that. Now who's font was it... I'd love to post an example. I will if I see it again. I think I found it on Flickr somehow. But the point is it really brought it home to me that the nominal combination for a given ligature isn't destiny/fate and there really is room left in which to take things further.

enne_son's picture

Maybe in a ‘perfect’ font a ligature is a little too inflexible a solution. Maybe ligatures are a kind of protrotype alteration that hints at a better way. [Eben]

Eben, how does what you are after differ from a systematic and contextually sensitive ligation that approaches the condition of cursive writing or Arabic connected scripts?

ebensorkin's picture

contextually sensitive ligation , cursive etc.

Here is what I was getting at: Let's take the Wy design Nick showed us a while back. ( I tried to dig it up to link to & failed. Sorry. ) He could have made a ligature but he didn't. He made contextual alteration where the h has a shorter serif on the left side so it sits with the Cap w nicely. Did you see it? If he had made a ligature it would have been designed with the assumption that you have a certain range of letter spacing & no more. It's not utterly brittle kind of thing - but compared to non-ligating letter designs, ligatures are more brittle. By brittle I mean when failure does occur - it's BAD and quite obvious. In contrast, contextual alteration permits far more letterspacing before letterspacing causes a failure of equivalent severity and that failure is probably near the point of failure for the type face as a whole. So non-ligating contextual alteration might have a benfit that trumps ligatures.* If so, it would be yet another reason to dig into it with gusto! :-)

But now I will undermine that arguement a little. The subtlety of very high quality notanic interaction of glyphs probably also depends on letterspacing not getting out of hand. The ratios of tollerance before the 'point' of the design is lost maybe nearly identical to that of ligatures. So, if the point of the design is to be finely tuned and optimized it may be that the kind of robustness I had been talking about is not relevant to contextual alteration for text faces.

On the other hand it might well be that each case has it's own dynamics and it's a very mixed situation. That's why I am keen to collect lots of samples and watch what's happening. But every experiment needs a hypothesis or two! ;-)

* Hrant of course, argues that any near-ideal text face will need ligatures and so the point of avoiding them is mute anyway.

In thinking about differences between what I am after & cursive or Arabic I am kind of knocked out by the enormity of the question. I don't know how much letterspacing if any is acceptable in Arabic fonts. That would be something I would need to know. I can imagine there is a little bit but I bet that in that case & in the case of scripts there is a built in brittleness of a different kind than you would see in a face with only occasional connections. How is that for a start? Maybe you have a different point to make. Do you?

hrant's picture

> So non-ligating contextual alteration might have a benfit that trumps ligatures.

Unless you look at it from a notanic perspective instead of a black one, and realize that, in the end, non-touching contextual alternates are just another way to ligate. That's why Legato is called Legato btw.

> I don’t know how much letterspacing if any is acceptable in Arabic fonts.

Reversing the black-white assumptions here as well, I would propose that the Arabic kasheeda (the extension of the horizontal line, usually to help justify a line, although quite often for aesthetics) is on some level equivalent to Latin letterspacing.

hhp

AzizMostafa's picture

> I don’t know how much letterspacing if any is acceptable in Arabic fonts.
Arabic writing is cursive (linked). That means that the same character has up to 4 different graphic forms, according to whether it is linked with the previous character, with the following character, with both, or with no other character.

As in Roman text, and depending on each font, Arabic consonants can form «ligatures». In a manner analogous to what happens when you write «œ» instead of a separate «o» and «e», 2 or 3 Arabic consonants can merge into one synthetic glyph with its own length, four position-representations, and kerning behaviour.

Since Arabic characters are linked together, you can’t change the letterspacing (also known as positive tracking) simply by adding short spaces between characters as in Roman text: the links would display as non-continuous ones. In this case, short pieces of link (called kashidas) between Arabic characters are automatically inserted. In other words: there are 2 ways of performing full justification in Arabic: either by magnifying the spaces between words as in Roman, or by automatically adding small kashidas after each linked character to make the words longer.

In addition to letterspacing and automatic kashida, there exist 2 other means of modifying the horizontal size of a word:
• Adjusting the width of characters
• Entering a manual kashida-character from keyboard, as you would do when using primitive word-processors.

In general, using the automatic kashida settings is a far better solution: after a text-reformating, if some kashidas become unnecessary, elastic style-kashidas will automatically disappear, while keyboard kashidas will have to be removed manually.

My font has the additional feature of making Calligraphic full-justification with the help of prolonged overlapping swashed final forms.

William Berkson's picture

I do find Nick Shinn's idea of contextual adjustment of characters--instead of just kerning--intriguing. Nick may have already implemented this in a font--he will tell us, I hope.

Aziz's reference to the kashida in Arabic script is also apt and interesting. In Hebrew also, in spite of the fact that it is a separated script (even in cursive), traditionally scribes would extend letters to fill out lines. In latin script, written with lower case, scribes, if I remember correctly, would fill out the line by extended swash terminals, rather than by stretching a character.

The reason for the difference in scribal practice, I think, is a difference in structure of the glyphs. In Arabic and Hebrew the horizontals are thick and the verticals thin--the reverse of latin characters. The horizontals also play a much more prominent role.

If you consider the latin lower case, there are no pure horizontals. The f and t cross bars are paired with arches, and the e cross bar is connected with a circle. The r could be a candidate also, but it is a partial arch also.

As opposed to the latin lower case, the resh in Hebrew has a single straight, thick stroke to the left--and it typically is a glyph stretched to fill out a line for justification. For this reason, I am a bit skeptical about how much contextual alteration in the lower case can be done. Of course I think Nick is suggesting subtile changes to help color, so this might lead to improvements, even if small ones.

Currently what comes closest to kashidas for latin script is the stretching or squashing of the whole character by up to 2% to justify a line--a capacity of InDesign.

The latin caps are a different story, as you have the 'H-LEFT' which have pure horizontals. And here I think there is more opportunity of contextual alterations, eg of the L in the LA combination. Similarly, you could shorten one or both arms of the T in the Th combination, which is common in English.

hrant's picture

The idea of non-ligating contextual alternates is pretty much as old as type itself.
And wasn't it in this thread that Raph referred to ATF's alternate "gy"?

> Aziz’s reference to the kashida

Your condition seems to be deteriorating by the thread...

> I am a bit skeptical about how much contextual
> alteration in the lower case can be done.

Nah, you just don't want the letters to change.
One obvious place is the "r": making it gently ascend, as needed.

BTW, Latin has something very useful in terms of
justification that Arabic/Hebrew don't: the hyphen!

hhp

William Berkson's picture

>Your condition seems to be deteriorating

Do you really have to flame me when I don't even address you or discuss your views?

>Latin has something very useful in terms of
justification that Arabic/Hebrew don’t: the hyphen!

One of my no doubt rigid and mentally distorted beliefs is that Hebrew does have a hyphen, the makaf, which is used along the tops of letters.

ebensorkin's picture

Nick may have already implemented this - William

He said that he had when he showed us the example here on typophile. Did you not see that? Can anybody point me to it. For some reason I can't find it.... But I am glad that you see that the idea holds some potential! It seems to me that the greatest potential for improvement in type design today lays down that path.

For this reason, I am a bit skeptical about how much contextual alteration in the lower case can be done - William

It seems like the wy might be one place to look... But that's another thread. However, I absolutely agree with you that the characteristics of the letters themselves must color the ways in which alteration can be accomplished. But that also shows why there will be variety. The forms have a huge range of variety. So while we could probably develop semi-predictive theory about where to check for opportunity for successful alteration based on inherent structural chacteristics I am more interested in looking for examples and then analysing what I have for patterns later on. The human eye & mind are pretty wonderfully adaptive. I have a feeling that like the L example on the other thread we will find that style & a host of factors and circumstances that make it possible to adjust letters in a myriad of surprising & delightful ways.

So, if you see examples of pro-color or pro-notanic adjustment would you point me or send the examples? I am starting a collection which I will share for free; (I hope) via a searchable database driven web site. Unless of course, there are so few that that is overkill. But we shall see. My best guess is that there will be a TON.

kashidas - Aziz

Aziz, thank you for explaining about arabic connections & kashidas. I was wondering, have you ever seen an arabic typeface that has extended connections so the overlap is exagerated? I have an idea that doing that would maybe allow for some additional spacing without using kashidas. Maybe this is naive. What reasons might there be for avoiding that solution?

I was thinking that for searchable text you would maybe not want to have kashidas present. Of course you could (maybe) write your search engine so that it tosses kashidas out... Do you see what I am getting at? Are kashidas used very much or is it something that is rare?

That’s why Legato is called Legato btw. - Hrant

You mean a the non-litteral definition
legato
adj : (music) without breaks between notes; smooth and connected;
"a legato passage" [syn: smooth] [ant: staccato]
adv : connecting the notes; in music; "play this legato, please"
[ant: staccato]

Yes, okay. I remember reading that on Evert's site. (I wonder if it's still in the google cache ) But do you see the distictions I was getting at? I am not sure that the distictions I was making preclude agreeing with you that contextual alterations are just another way to ligate. I have said & will say again that what we recognise as ligatures are in fact just a subset of solutions to a given problem whose disproportionate mental prominance can only be explained by our thrill in observing the obvious 'rule breaking' they engage in. Ooooh! They touch! Still, it this disproportionately emphasized visual distinction does also have a practical/fuctional impact re: letterspacing that raises it's own questions. I don't see the conflict or contradiction there. I would like it if you addressed them more specifically! And if the two ideas are somehow opposed, please explain why that is.

enne_son's picture

"Your condition seems to be deteriorating..." [Hrant]

Give it a rest Hrant. Why must you infect so many threads with this kind of invective.

hrant's picture

I didn't know about the makaf! Thank you Aziz.

> Are kashidas used very much or is it something that is rare?

The kasheeda is quite common.
But you can't just put it anywhere - there are rules, although I
don't know them, and they might actually not be fully formalized.

> You mean a the non-litteral definition

I mean the reason Evert called it Legato.

> Ooooh! They touch!

This never happens to an immersed layman (unless something has gone really wrong).

hhp

hrant's picture

> have you ever seen an arabic typeface that has
> extended connections so the overlap is exagerated?

I don't get this.

BTW, something else I've seen in Arabic to help justification
is a special wide and flat "kaaf" (although it's not common).

> Why must you infect so many threads with this kind of invective.

Press release: "I'm sorry."
But it's nice to see that you care, Aziz.

hhp

thacket's picture

Greetings to All,

This is a very interesting discussion. I am not a typophile, but I am a great admirer of functional and timeless beauty; qualities that most typefaces certainly possess. By circumstances beyond my control, I have been pressed into typographical service to publish some technical papers for my company. As a result, I have been doing a little research into readability, legibility and comprehension (which in my newbie ignorance I have determined to be seperate, but overlapping issues). I am not, nor will I ever be, a professional typophile, but I have read all of my life (end user alert! talk slow and use small words!). Now you have all of the information to studiously ignore my uninformed opinions as you wish (I have been married 32 years; plenty of practice having my opinions ignored).

I said all of that to say this. Since I have read just about anything printed, I have been exposed to many fonts and typefaces with and without ligatures. The one thing I have noticed is that when the flow of my reading is interrupted because I perceive a hitch in spacing in the word structure, one of the reasons (there are others) is because of both the absence and presence of ligatures. This bump in the reading flow can occur for me at even small point sizes (8). The factors that come into play for me are:

1. If a ligature is present and I bump over it in text, it is usually because the ligature looks cumbersome; the font does not lend itself well to construction of proper ligatures or the ligature structure was poorly implemented in that particular font.

2. The opposite is also true; if a ligature is absent and I notice it, the font could probably benefit from a well constructed ligature or better general construction.

These are just an amateur's opinions; ligatures make reading for me better when they fit well with the font and make it worse when they don't and the opposite is also true. This kind of reminds me of the Ford vs Chevy discussions that we have here at work (warning! lowbrow content!). They both have strengths and weaknesses, but both must be used for what they are best suited and that is ususally strictly in the eye of the beholder (or the end user).

Thanks for listening.

All the Best,
John

P.S. - I have started using the font Baramond which, I believe is derived from Garamond Antigua. The implemention of ligatures in this font I find pleasing.

ebensorkin's picture

Thanks for your post John!

Syndicate content Syndicate content