Special fi ligature in Turkish?

charles ellertson's picture

We have a book in where the designer chose Bembo as the text font. This book has a spattering of Turkish words; both the dotted and undoted i are used. Language tags were NOT used by either author or editor.

Now an obvious design solution for this sort of text is to use a font like Sabon, where not using the fi ligature causes no problem. Set everything without fi & ffi ligatures, no need to sort between Turkish & English. But that's not how current designers think; she picked Bembo.

Bembo without ligatures doesn't really solve the problem for Turkish; the terminal of the "f" half blots out the dot on the "i". You would have to positive kern the "i" quite a bit to make the dot clear, and the resulting space would give rise to a different confusion.

For Bembo anyway, it isn't that hard to make a special fi ligature where the dot over the "i" rides a bit lower, and the terminal of the "f" doesn't swoop down quite so far. The dot on the "i" clearly shows, so we can have a "f_i.dotted" and a regular "f" and undotted i, which solves the problem nicely, visually. The only wrinkle is that with the sequence fii, the dots on the two i's wouldn't be quite the same height.

The author or editor will have to mark affected words in proof & we can make the switch.

What I wonder though, is whether or not this will appear odd to anyone who regularly reads Turkish -- will it still be seen as an error, or "unfortunate"?



Alessandro Segalini's picture

Nice issue, Charles Ellertson.
Is there a word with 'fii' in Turkish ?
I think a Turkish Bembo would be accomodated changing the top of 'f f '——no ligatures, e.g.,

charles ellertson's picture

Well, my Bembo has been pretty heavily modified. When we were still running repro --early-mid 1990s, I took a look at a couple of books, one a fine-print book from Steinhour press, one an old Penguin paperback from college days, & got a feel for the range of how Bembo printed in metal. First thing you noticed is that whoever made the PostScript Type 1 (3?) font shortened all the ascenders & descenders. Second thing you noticed was the contrast of the letters was all wrong when printed offset. So our Bembo got one large adjustment when we were running repro, and a second adjustment when DTP became prominant.

I don't know how to post a picture here. I can easily make a PDF, but most sites only allow you to post a gif or jpeg. Dunno how to do that with type.

Anyway, I suppose the taller ascenders would help with your solution, but also allow my proposed solution. I guess to give an answer you'd have to see it.


yader's picture

Hi guys,

there are some occasions where the sequence fii becomes important. for example the turkish word fiil (http://www.pratiksozluk.com/sozluk.php?word=fiil&dictionary=1) could be used in the mentioned text and it would be very unhappy for your eye, if the two i-dots differ in there height. It would look like an error. Maybe it could work, if you only modify the stem/terminal of the f, so you won't have two different versions of the dotted i. I also want to mention that the ligature of f and the dotted i is looking like the combination of f and the undotted i for the the turkish eyes. The best solution however would be to use a different typeface, if possible ;-)

p.s. sorry for the bad english.

yader's picture

Here is a site, where you can see that the the "old" Bembo Type was planned and made with two different ascenders to fix this kind of problems, but i can't find a typefoundry which has this kind of bembo, very interesting though:

And here are some examples of that font in use:


charles ellertson's picture

As I said, my Bembo is modified with rather higher ascenders. There is no problem with making a ligature having the i dotted (wish I could show it). But I hear you --if two i's appear together, they better have the same height dots. The difference in height is so small that it can't be noticed between different words, only if next to each other. I can fix that, too.

The Bembo Book you show gives back some of the needed weight & contrast reduction, but it lost something, too.

In my world, the comp only gets to set the font the designer picks.

Alessandro Segalini's picture

Evet, "fiili olarak" gibi, sağol.
Dear Charles Ellertson, I didn't change the ascender measure in my example, I would not either, no need for a ligature glyph for this matter, or do you ; I'd like to take a look at your.pdf, if you cannot post it here (editing your first post), you can find my email linking my avatar, then website.

charles ellertson's picture

Alessandro - I sent you a small .gif file, which I couldn't manage to upload here.

Miguel Sousa's picture

Charles, to post pictures follow these instructions:

Nick Shinn's picture

I also want to mention that the ligature of f and the dotted i is looking like the combination of f and the undotted i for the the turkish eyes.

This was mentioned in a recent Build thread in Typophile, but I can't find it.

Nick Shinn's picture

I'm not convinced that f_i ligatures are always the best solution for a typeface, or that there is only one way of handling such a ligature.

For Worldwide, a news text face, it seemed to me that it was better for the components to keep their "spots", but joining the crossbar made things a bit less busy in that crucial area.

charles ellertson's picture

OK, I think I can upload the image:

Left to tight: (1) regular f_i ligature, (2) special Turkish f_i ligature, (3) f + i (no ligature) in the recut Bembo font.

Below, f_i.turkish ligature followed by an "i" with a dot slightly lowered to match the lig. The dot isn't lowered much; in 11-point settings an "i" in a adjacent word won't have a noticeable height difference.

kentlew's picture

Charles -- congrats on successfully uploading an image ;-)

It seems to me that if you're going to go ahead and lower the tittle on the i in a special cut of Bembo, then you don't really need an f_i ligature of any kind, since the f and i will coexist just fine without one.

Or are you thinking that you'll leave the i alone, and construct a special, contextual i.alt that will substitute after the f_i.turkish lig?

I'm not sure I'm following you.

-- K.

charles ellertson's picture


If you look closely, the terminal on the f in the ligature is different (shorter) as well -- shorter than the regular "f", and of course, quite a bit shorter than the f in the ligature. It is closer to a f.terminal that we use when the f is followed by a space. I'm reluctant to change the font (database) by lowering the dot on the i globally at this point; maybe later. There isn't that much Turkish in the book in question.

BTW, below an image shoing (left to right) a word in Bembo scanned in from a book printed by Steinhour, the same word from a printed book (CTP) with standard 1995 or so Bembo, and the word scanned in from a book set in our Bembo, also printed CTP. Note they are all scanned in from printed text.

Nick Shinn's picture

I suspect that the *high* position of the tittle in faces such as Bembo came about for the following reason:

In the first place, it would have been quite possible to set an f followed by an i -- the i having a normally positioned tittle, as per Charles' fix--with the arm of the f kerning over the body of the i. However, this would have resulted in a lot of breakage in the reuse of type, as the "beard" of the arm and the tittle are awfully close.

Therefore, an f_i ligature was designed, with the arm of the "f" covering the position where the tittle might be, so it didn't look as if it was missing, just obscured.

Except, the tittle was relatively low, and the arm had to bend way down to hide its supposed position, and that looked odd.

Therefore, the tittle was moved up, to make the f_i ligature more convincing.

Just a theory.

Nick Shinn's picture

BTW Charles, you don't have to upload an image with every post now :-)

charles ellertson's picture

Don't worry. I do most of my posting at home; all of the resources are at work.

cuttlefish's picture

Nick: Here is the Build thread you mentioned:

and a noteworthy point therefrom:

cuttlefish wrote:
So then would it be advisable to design a special substitute non-interfering Turkish f when the regular f tends toward accent interference?

k.l. wrote:
Why not! But I’d switch priorities: make the less sweeping ‘f’ the default, and substitute by the sweeping ‘f’ if non-ascending or non-accent glyphs follow. Fewer headaches when kerning …
It’s the non-accented glyphs that are the exceptions in an extended glyph set.

Syndicate content Syndicate content