lcaron and dcaron

How should one adjust the spacing between these Czech and Slovak glyphs and whatever letter follows them? It's easy enough to tuck an x-height glyph a bit beneath the háček (caron) to even out the letter spacing. However, how should one kern the following glyph when it has an ascender?

I've noticed that most supposedly higher-quality fonts don't bother, leaving way too much space or having the ascender crash into the háček.  Is there some Czech/Slovak convention I'm missing or am I just noticing inattention to detail in those fonts?


  • James PuckettJames Puckett Posts: 1,859
    I usually design them so vowels can tuck under them and then add positive kerning so they don’t crash into stuff. But I’ve also done it the other way around.

    ďa ďä ďe ďě ďi ďo ďǒ ďu ďk ďA ďI ďO ďU ď' ď" ď” ď“ ď! ď? ď) ď] ď} ď*
    ťa ťi ťo ťu ťk ťA ťI ťO ťU ť" ť” ť“ ť! ť? ť) ť] ť} ť*
    Ľa Ľi Ľo Ľk Ľu ĽA ĽI ĽO ĽU ĽT ĽV Ľ' Ľ” Ľ“ Ľ! Ľ? Ľ) Ľ} Ľ] Ľ*

  • I usually use contextual alternates to save myself some of the kerning trouble. For instance, Cormorant's tall consonants retract their top serif when they follow a caron.alt:
    Sometimes instead I make alternates of /dcaron/ and /lcaron/ with wide RSB that get cycled in before tall consonants.
  • Adam JagoszAdam Jagosz Posts: 683
    I might be wrong but I always understood that the háček is supposed to look like it's been added to the letter without modifying its advance width (even when it does). Not introduce a noticeable gap. Of course, it needs to look like it's been tucked onto l and not before k in Veľká, but if you see it leaves a huge gap in most online fonts? I say — Objection, hearsay.
  • Nick ShinnNick Shinn Posts: 1,867
    edited April 28

    OK then, not hearsay, but a quickly sourced example!

    We may both be right Adam, with factors like weight and point size playing a role too.
    By “plenty of sidebearing” I meant be wary of making a negative sidebearing which encroaches into the accent. 
    Who amongst us would not be sorely tempted to kern the above?
    I might have mentioned that it’s also a good practice to put the háček extremely close to the letter.
  • Adam JagoszAdam Jagosz Posts: 683
    edited April 28
    From The Insects Project (Filip Blažek):

    The caron in ď and ť should create as little white space as possible in situations where the letter is followed by another letter with an ascender, such as in the word loďka (row‑boat), or punctuation: buď! (be!).
    There are no specific remarks on the kerning between with non-ascending letters, but the examples seem rather tight and cozy.
    What I meant by “hearsay” is that I don’t think most popular and renowned fonts out there give much love to any of the European languages east of Germany. Just because everyone seems to do things a certain way doesn’t mean it’s the right way. It is not as much about not being tempted to kern these as about not being bothered.
  • Nick ShinnNick Shinn Posts: 1,867
    edited April 28
    Thanks Adam, you’ve convinced me.
    So how about this as a best practice:

    The advance width of a letter should be the same with and without the vertical háček.
    But positive kerning may be added when the háček-accented character is followed by an ascender (and exclam, question, quote, parenthesis, etc.)

    That’s a lot of characters to put in a kerning class, but if one were to make the accented character wider, one would still have to make a large (negative) kerning class of x-height characters and punctuation anyway.
    This method will fail when there is no kerning support (where ever the vertical háček is followed by an ascender etc.), more badly than the gappy alternative, but kerning is near universal these days.

    In very bold weights, it may also be necessary to add positive kerning when followed by an x-height character, so the háček doesn’t look like it’s above that character. Therefore, to avoid this, and having to create both negative and positive kerning classes, make the accented character width wider, according to it being followed by an x-height character. 

    Some of my types have questionable vertical háček characters, but this one isn’t too bad. What’s significant is indeed the advance width, but also how close the accent is to its ascender. Here (Pratt Nova Text) I should perhaps have positioned the accent closer to the ascender in the Bold—I was trying to avoid clogging and spottiness—but I don’t mind the more open effect.
  • John SavardJohn Savard Posts: 986
    Upon being told, "No, this isn't an apostrophe, it's an alternate form of the caron", I'd be tempted to inflict this (the version on the right-hand side) on the long-suffering Czech people:
    because to make sure it is understood as a caron, obviously the other side of it has to be visible!
    But then, I am the pedestrian sort, who would conclude that if Roman/Antiqua type causes problems for German orthography, the solution is to use the long s in Roman type for German, so that spelling in Roman type can be identical to Fraktur spelling!
  • Way to add more demons to the typographic nightmare, John!
  • As for the shape, I'm mostly going by @Rainer Erich Scheichelbauer's recommendation that it should look like the broken-off right side of a caron, and thus be straight rather than curved like an apostrophe. I'm not sure where he gets his information from, though!
  • Alex VisiAlex Visi Posts: 173
    edited April 29
    In my view a good practice is:
    - lcaron/dcaron’s advance width should ignore the mark;
    - try designing the mark in a way that avoids collapsing with lowercase letters (except the ascending ones);
    - kern the letters with ascenders (ďk, etc) and do leave the ugly giant gap that you probably don’t like. From my [not very extensive though] research, that gap is expected.

    Also, an interesting treatment example:

  • Vasil StanevVasil Stanev Posts: 678
    edited April 29
    I very much think @Nick Shinn has nailed it down in terms of kerning and design. My two cents:
     On this channel: and elsewhere, including grammars and textbooks, I have seen the hacek written out as a free floating caron on the right-hand side of the /l and the /t. It has been my personal experience in our profession, that the form of some of the diacritics in some languages is not strictly ortographycally codified. It seems that the more important thing is that there is some sort of additional mark, often simplified, on the letter to distinguish it from its other variants. E.g., in Slovak, the -t' is the ending of verb infinitives.The mark distinguishes it from its hard variant, the /t. We had a similar discussion I remember about the cedilla some time ago on the board - in handwriting it is often simplified to a simple stroke.
    Attached are some examples from a personal copy of a Bulgarian 1975 Czech (Not Slovak!) textbook :) You can perhaps see very well the difference between handwritten and print form.
  • Adam JagoszAdam Jagosz Posts: 683
    edited April 29
    Personally, for most projects I prefer designing ď and ľ (and ť when the design needs it) wider than d l t and add negative kerning against lowercase letters without ascenders. Why? This way I don't have to kern against ’‘”“]})$&#@ and any other ascending character in the font. Not to mention characters from other fonts, against which I cannot provide built-in kerning.
    Even though there seem to be less ascending letters than non-ascending ones, I think the overall set of characters is more predictable in the latter case — I typically only consider a-z + accents, period, comma, underscore, ellipsis, hyphens and dashes. Sometimes semicolon and colon.
  • Alex VisiAlex Visi Posts: 173
    edited April 30

    Even though there seem to be less ascending letters than non-ascending ones, I think the overall set of characters is more predictable in the latter case — I typically only consider a-z + accents, period, comma, underscore, ellipsis, hyphens and dashes. Sometimes semicolon and colon.
    I wonder if you (or anyone here) have counted the kerning pairs in both ways? To me, it seems like there are fewer pairs with ascending symbols. Sure, there are many non-alphabetic symbols, but a lot of them either don’t need kerning (currencies, percent, selection, etc) or would be in ‘lowercase’ category in this sense (punctuation, many math symbols, etc).
  • Thomas PhinneyThomas Phinney Posts: 2,289
    It’s not just “which option needs more kerning pairs/classes” but also whether you think overlapping glyphs are a bigger problem than extra space.

    I definitely believe this, so all other things being equal, I prefer @Adam Jagosz’s solution—because the effect of things that are missed is a significantly less severe problem, in my book.
  • Alex VisiAlex Visi Posts: 173
    edited May 1
    In Peter Biľak’s fonts, the spacing of standalone ľ’s (< nice!) appear to be oriented to lowercase letters, with kerning to ascenders (below). To me, a gap seems to be a bigger problem than an overlap since it may be confused with a word space. Although, if the mark gets hidden inside of the next letter altogether, is a problem too. Then the optimal solution is space it halfway and kern both ways :#

  • John SavardJohn Savard Posts: 986
    edited May 1
    It is true that too much spacing can be confused with a word space, creating a problem.
    However, letters overprinting each other seems to be a catastrophic failure of the print system, as that should be absolultely impossible.
    I would have thought that orienting spacing to the case when the following letter has an ascender, and including an adjustment for the case when letters can fit closer, is also simply following the literal meaning of the word "kern" - to modify a letter, so that it can fit more closely to another letter than would be possible without special measures.
    Overlap is catastrophic, as it gives away the "secret" that the letters were formed digitally instead of in the time-honored way with lead type.
    Of course, though, this is a bad way of thinking. If one thinks of digital type only as an inherently inferior way to imitate lead type as closely as possible, then that closes off one's thinking towards doing anything novel or creative with the enhanced possibilities of the digital type medium. And, indeed, with digital type, potentially almost anything is possible - even if, at the moment, the available software framework isn't even suitable to handling such obvious things as properly supporting nastaliq!
  • André G. IsaakAndré G. Isaak Posts: 606
    However, letters overprinting each other seems to be a catastrophic failure of the print system, as that should be absolultely impossible.
    This seems like a strange thing to say. There are certainly cases where a designer might want letters to overlap, and numerous fonts rely on this.
  • Nick ShinnNick Shinn Posts: 1,867
    This is what I love about design!
    Conflicting criteria must be resolved in a way that is both efficient and attractive.
    Fonts are like gigantic puzzles.
Sign In or Register to comment.