They are just a relic of blackletter writing/printing where they were mandatory. When we finally got rid of blackletter, many serif and sans serif typefaces also got them, sort of as compromise for everyone who grew up with blackletter. But I don't think they belong in a non-blackletter typeface.
They represent a distinct sound, but it doesn't make a difference if you use a ligature or not.

Hi Eben I'm not even confident enough in the type world to try commenting on German ligatures, I'll let the professionals handle that, however I have designed several versions of a ch and ck ligature. This was my latest attempt…

(most of these were done purely for fun).

Here's some text of mine containing ch and ck ligatures.

this thread also covers this topic:
German c-ligs

Eben, these aren't at all digraphs. The letter c is not so common in German. I believe that, in true German words (i.e., not loan words from Latin, Greek, French, English, etc.), it never comes alone, but rather always followed by either an h or a k. So it made sense to write & set ch and ck together, I guess. I do not believe that there were any exceptions to this rule, except perhaps when mixing foreign language words into the text (church would look mighty weird indeed wit two closely spaced and/or ligated ch combinations).

1. What Ralf said. In addition to that: I saw veteran typographers applying tighter spacing to sch, ch and ck – in digital typesetting. Thus, not via ligatures, but customized kerning tables and the like.

2. All instances. There might be exceptions with loanwords/names, but I can’t think of even a single one.

3. I assume this is the quote?

In German, ch, sch and ck are pronounced as one sound, as ck is in English. The c always looks too far from the h or the s, and in metal we had ligatures for these (as for tz by the way), also a time-saving device. I cannot kern the c for these combinations, as they would look bad in all the other languages. — Erik Spiekermann

4. Slightly tighter spacing (and even that is a controversial issue, see Ralf’s opinion). Certainly no touching.

5. Yes, of course a ‘blotchier’ colour would be the downside, if you only alter the spacing.
I once saw a short ‘c’, similar to this (quick and bad) simulation:

F

Okay so there is no true digraph here. And to be clear I am not talking about blackletter which has it's own considerations but a Roman serif fonts and/or a Sans made for broadest uses : text, information/signage, headlines, & so on.

So as to the question of if it would be desirable to set up language specific kerning in order to accommodate the most likely use of the c in German; or if it would be better to have a language specific (german) liga feature for German. It seem like the answwer is no. I am inclined to agree with Ralph because if you have a good kern for the combination in question that kern should be based on optical considerations. If you wanted to make a special touching on non-touching liga or calt combination then it seem like the consideration would be similar eg does it read well, is it distracting, is it right optically speaking & so on. Reading the thread that Paul linked to it looks like Karsten isn't too keen on the idea of giving them special treatment. Adam says "Since the "c_h" and "c_k" ligatures are of historic interest only, I recommend putting them in the "hlig" feature, and also in the "dlig" feature, but NOT in the "liga" feature." Which resolves it for me. For me the optical aspect dominates, but also I can't disagree with Karsten Adam & Ralf all at once with some really compelling reason.

Andy, thanks for showing us your work.

Paul thanks for the link!

Eben, you left out a third scenario: a language specific (calt) feature that substitutes c with a slightly narrower, more tightly-fitted (on the right side) version for German ch and ck combinations.
any thoughts from native German readers on this approach?

Unless the fonts are made for historians I wouln't give the digraphs any special treatment. Actually, the majority of German readers might even think there's something wrong with the font, because this simply isn't used anymore for some decades now.
Adam mentioned another important point in the other thread. As you know we can build endless words:
http://en.wikipedia.org/wiki/Rinderkennzeichnungs-_und_Rindfleischetiket...

So, occasionally such ligatures might be wrong, because the two letters don't stand for the sound, but belong to separate parts of a compound word. So any automatic replacement would fail. So I don't think it's a good idea to have it in any standard feature like liga,calt,clig, ...
Anyone who wants to use is, has to do it manually. This is always the case for setting blackletter texts in German.

Ralf

If you're setting up a language specific kern, for ch it would be prudent to include it for Spanish as well. It's not 100% standard to do so, but some of the larger publishers I've noticed place them a tiny bit closer.

«El futuro es una línea tan fina que apenas nos damos cuenta de pintarla nosotros mismos». (La Luz Oscura, por Javier Guerrero)

Ralph makes the point I would have based on Adam's remarks except that but it seems like an occurance of c+? in german would be on one side of a word part or another not like the fl example. Still, right or wrong I think the optical consideration is the primary issue or "trump card" and should win. Again, I am not speaking about blackletter which I don't have an opinion about yet. And if you felt strongly about it you could put in a discretionary lig/alt form that was not tied to a region/language.

Still, right or wrong I think the optical consideration is the primary issue or “trump card” and should win.

Speaking of Trump, would it not be correct, if one were setting German text in a "retro-antiqua-letterpress" style, and the face were Trump Mediäval, to reproduce all the artefacts of that genre? This from a 1967 Linotype specimen.

An older German type setter would love this!
Sure, in this context it's okay to use it in this ways.

But still: I guess most Germans under the age of 30 have never seen this kind of type setting. So for them it would rather be distracting than helpful.

When we finally got rid of blackletter, many serif and sans serif typefaces also got them, sort of as compromise for everyone who grew up with blackletter.

Futura, as one of the faces designed to do away with black letter, had an interesting solution to this problem. It did include ch and ck composite characters in its original release set.
But the main feature was the vertical termination to the ends of the arms of both the cap and lower case c, which allowed for tighter right hand character space. The paired characters were done away with c. 1933, but the C and c with their elegant cross-language solution remained.

Charles: thanks, that’s exactly what I tried to describe with my ‘short c’. You put it more eloquently, and with historic background.

Yes, there is no reason to dismiss the idea altogether especially if you can find a solution that is optically pleasing like the example above. And I agree that implementation as an alternate would be the way to go. Deciding to make one would come down to the audience the font is intended for. Or maybe for bragging rights... ;-)

Eben, <ch> and <ck>are are digraphs in German (and <sch> is a trigraph). Digraph simply means by definition that more than one character is used to represent one sound. <ch> represents the phoneme /x/ (phonetic realisations: [x] and [ç] in literary German) and <ck> represents /kː/, or to be more precise a /k/ marking the preceding vowel to be short.
It seems to me from your earlier point that you partially mix the terminology ligature and digraph. Digraph means one sound represented by more than one character (often but not always considered one letter), ligature means the visual combination of two characters.
Digraphs may or may not be written with ligatures depending on orthography and/or tradition, while ligatures can be applied for digraphs but for simplex graphemes as well.

Examples
+digraph, +ligature: German <ch>, <ck> in blackletter tradition, Hungarian sz up to the early 19th century
+digraph, -ligature: spanish <ch>, German <ch>, <ck> in antiqua tradition, all Hungarian digraphs (<cs>, <zs>, <gy>, etc. incl. <sz>) since the 19th century
-digraph, +ligature: <fi>, <fl> in English
-digraph, -ligature: most letter combinations in most Latin languages, like <om> in "combination" in English

[edit: lesser/greater signs corrected]

«<ch> represents the phoneme /x/ (phonetic realisations: [x] and [ç] in literary German)»

I’m curious as to why you specified literary German here. To the best of my knowledge, [x] and [ç] (and minor allophonic variations thereof) are also the phonetic realisations of /x/ in vernacular German, no?

Look, German vernacular is diffracted into a multitude of dialects. Several dialects realise <ch> as [ʃ], some as [k]. These sounds do contrast in literary german (and are spelled <sch> and <k> respectively). e.g. Kirche 'church' ~ Kirsche 'cherry' might sound alike in certain dialects, not in Standard High German.

My understanding - possibly false - was that a digraph had an entry in dictionaries and phone books which was proof that people thought of the digraph as a unit. But I can see where you could use your definition, and how on some level it is a more useful one in general. From a type making point of view the way something is spoken is interesting but not perhaps crucial. For type making the issue is what sort of form or forms does it need to take to be accepted; and does this need a code point of it's own; and how do you use it as a typographer. I say his not as refutation but as an example of how a term takes on more than one meaning in different groups of people. Probably when speaking of digraphs in the future I should specify that I am using it according to your (liguistic) rules since they are more specific.

Ligature as we commonly use it just means touching. But historically "ligature" could and did also mean a single piece of lead type with one or more glyph on it touching or not. You could have the word "The" even. And the ch & & ck were exactly that - two glyphs on one peice of metal type. So in this sense the "digraphs" in question were by one acceptable definition of ligatures - ligatures.

Also, would you remind me : does “Sz” have it's own place in the dictionary? Maybe you could start a thread about Sz! :-)

> Ligature as we commonly use it just means touching. But historically “ligature” could and did also mean a single piece of lead type with one or more glyph on it touching or not. You could have the word “The” even.

Eben -- I have a different understanding of the historical use of "ligature" in relation to type. The word derives from Latin and means "tied" more than "touching." As such, ligature was used only to refer to sorts where the letters were actually connected or tied -- fi fl st ct, etc.

Sorts containing two or more letters (for fitting purposes, say) would be more properly called "logotypes." From the glossary of a printer's guide published by the Graphic Arts Company in Boston in 1924:

Logotype. Two or more letters, or a whole word, cast on one body.

Nowadays, of course, the term "logotype" has slightly different connotations.

-- K.

Eben,

There's also a difference between letter and digraph (and character).

¶ Character is usually the smallest composing entity of a writing system.
Letter is what shows up in dictionaries and alphabet listings.
¶ Digraphs are as specified above character combinations to designate phonetic or phonemic elements of the language different from the meaning of the single characters. ¶ They may or may not be treated as letters.
Whether something is a digraph, one can determine from the mapping sound <-> character sequence. Whether it's a letter, that's a "higher protocol" convention not determinable from the first information.

Consequently, character is a property of a writing system (e.g. Latin "alphabet"), while digraphs and letters are concepts of a particular language's orthography (e.g. English, German, ...).

¶ Thus, Hungarian cs, dz, gy, ly, ny, sz, ty, zs, are digraphs and letters of the Hungarian alphabet. dzs is a trigraph and a letter there. ch is a digraph, but not a letter in Hungarian (it's used in foreign words, is an allograph in that sense of Hungarian h, but not counted as letter, sorted as c+h in dictionaries).
¶ In German, ch, ck are digraphs, but are not considered as letters. sch is a trigraph but not a letter either in German.*, **
¶ In Polish dz, dż, ch is a digraph, but not a letter AFAIK. But Slovak, where dz, dž, ch are digraphs alike, they are considered separate letters as well.
¶ French aux is a trigraph (standing for /oː/, among a lot of other character sequences), but not a letter...
¶ English gh, ch, sh, etc. are digraphs, but not considered letters.
¶!!! Such conventions can change even. ch and ll were digraphs and letters in Spanish. They are invariably digraphs, but they are not considered letters any more.

* In German, the diphtongs' writing (except for au) can be seen as digraphs as well, as in German ei, eu, äu, ie are pronounced not what the individually denoted vowels would sound in a diphthong. (<e> stands for [ɛ] or [e], <i> for [i], but <ei> is pronounced [aj], not [ɛj] or [ej]; etc.).
** Analytically, ck additionally could be seen as a surface realisation of kk; one could imagine an underlying kk encoding with "contextual alternate" of k > c under the condition of being followed by a: _k. But th

PS: the letter sz has a very interesting history in Hungarian printing, and I have some very interesting information on that. However, I'm very careful not to draw too early conclusions or to publish incomplete research. (Additionally, the historical distributions/rules of long-s and s in Hungarian printing history is also very interesting, though even more complex). I'd not like to fall into the same mistake as diacritics.typo.cz does, i.e. publishing wrong or premature results.

——— I hope that terribly long post does not overexplain things and I hope equally I don't seem like an egghead now :-)

That was an interesting read, Szabolcs – thanks!
And don’t worry, on Typophile, eggheads are most welcome. :-)

Some more nitpicking:

You wrote, <ch> represents the phoneme /x/ (phonetic realisations: [x] and [ç] in German language. It can also be [k], as in Dachs, Ochse, Wechsel etc. It still is a phonetic digraph, though.

!! Such conventions can change even. ch and ll were digraphs and letters in Spanish. They are invariably digraphs, but they are not considered letters any more.

I hate to nitpick, as this thread has been most informative, but ch and ll are still considered letters in Spanish. Also, do you have an example of the ligatured form of SZ? Did it connect the top strokes or ?

«El futuro es una línea tan fina que apenas nos damos cuenta de pintarla nosotros mismos». (La Luz Oscura, por Javier Guerrero)

«Also, do you have an example of the ligatured form of SZ? Did it connect the top strokes or ?»

I’d love to see this as well.

And as a small aside, rather off-topic, I’m afraid: if you write letters as capitals when referring to them as letters (i.e., an A, to T’s, etc.), would you write SZ or Sz? The ‘title’ for the letter in my Hungarian dictionary says “Sz”, so I’d guess that would be more proper? And then SZ is only used in all-caps?

Eben — I have a different understanding of the historical use of “ligature” in relation to type. The word derives from Latin and means “tied” more than “touching.” As such, ligature was used only to refer to sorts where the letters were actually connected or tied — fi fl st ct, etc.

That is definitely the conventional idea. Of course you could also say that the letters were tied by the metal below even if is not a visible connection.. ;-) I will try to find the place/s that I have read the use I am suggesting. In the meantime and off the top of my head I know John Hudson has used the meaning I gave. ( Not that you should bow to him. ) Personally I find that it is a pretty useful definition as I have no other word to describe it other than "Logotype" which seems to be a Linotype specific word. That is to say even less well or clearly understood than "ligature".

One thing is sure and that is that Type Jargon is almost never as agreed upon as you might prefer. If we could invent one & agree to it that would be ideal, but until I am convinced otherwise I will probably continue with "non-touching metal ligature" or something like that.

Szabolcs, Ralph, Matthew, Thanks!

Also, I would like to hear a premature idea about sz. As long as it is explained it's an unfinished idea, I think it's okay to say how it looks to you today. At least then I can begin to think about it.

Matthew, I won't argue here with you, as you seem to be native (?) Spanish speaker.
However, my information was?, that in a recent reform CH and LL were ripped their letter status, and also consequent are sorted as C+H and L+L in newer dictionaries, while they sorted as separate letters earlier.
Doublechecking the information now, I see that there has been a reform in sorting (indead sorting them as simple character combinations, but still considering them an own letter of the alphabet. Sorry for that misinformation.

Oisín: I'd mostly use italics to refer to letters as themselves or in bold face (in texts where italics is used for indicating sound values). As the idea is that the letter should stand out of the text, if I'm lazy to type the tags or if only plain text is permitted, I would use the capital form of all constituting components (i.e. SH, SZ, CH, LL, GY) independently of capitalisation practices of letters in the language. Here capitalisation has the function of making something standing out.
When doing so I understand that digraphs considered letters are sometimes "fully" capitalised, sometimes not.
Dutch IJ is capitalised with both components at the beginning of a sentence or in proper names, Hungarian, Spanish digraph-letters are not, only in all-caps setting.
My name (consisting of sz-a-b-o-l-cs) is such written Szabolcs or all caps SZABOLCS (not *SzABOLCs). (Chili, Llamar,...). In that sense the digraph-letters behave in a lowercase-titlecase-uppercase triad, but as they are composed of individual characters, in in most encoding schemes e.g. Unicode, etc. such a distinction is not necessary.

Concerning the sz-ligature.

I wrote about the sz-ligature in Hungarian used up to the (approx) beginning of the 19th century and that woke your interest. The grapheme of the sound (IPA) [s] in Hungarian has a very complicated history, if you view it from a typographic point of view, if you ask me (orthographically, approx. from the 15-16th century it was fixed as <sz>, previously it was also orthographically adventurous). So that ligature was only one typographic realisation. It is built by the long-s and a normal z ligated. It is very different from the german sharp-s (ß), which is an longs_(long)z ligature in Blackletter, but a longs_s ligature in Antiqua (or in some Antiqua forms it's a longs_longz ligature, but never a longs_normalz lig).
I have scans at home which I can share later. However, this is only a typographic curiosity, as it's long deprecated...