“French” apostrophe?

I don’t know how many have noticed this before, but the Arno Pro fonts have a character variant for uni2019 (quoteright) which is registered for {locl} substitution under languagesystem latn FRA. This alternate glyph has increased sidebearings on both sides.

That is to say, when a setting is language-tagged as French, the apostrophe is spaced noticeably looser (+30 units each side for Text Regular style):

Now, I can understand that designers of different nationalities might have slightly different preferences for overall fitting, influenced by their native typesetting traditions.

But is le goût Français for l’apostrophe really so radically different from others as to warrant such a localized distinction in general-purpose fonts?

Arno is the only instance where I’ve seen this done. Is Slimbach just being overly fussy here, or is he the avant-garde of something important?

I’m particularly interested to hear the perspective of native French designers or, better yet, French typographers.



  • Nice catch. It does seem a bit pedantic, but as a reader of French I like it.

    > But is le goût Français for l’apostrophe really so radically different from others as to warrant such a localized distinction in general-purpose fonts?

    It might be less a matter of 
    goût and more a matter of simply a designer judging what a reader of French needs. I might emulate it.
  • Cool find! Does it also have question and exclamation marks with extra spacing? I've noticed French writers sometimes like to add a space before these. 
  • joeclarkjoeclark Posts: 122
    Does it also have question and exclamation marks with extra spacing? I've noticed French writers sometimes like to add a space before these. 
    Even Microsoft Word for Windows includes such spaces (at least inside quotation marks), but, being dumb as a mule as Windows always is, the character used is nonbreaking space (cf. Pierre Igot).
  • Maybe because thinner spaces are breaking so using a non-breaking full space is a lesser evil? You don't want a question/exclamation/quote mark breaking to a new line.
  • Kent LewKent Lew Posts: 905
    No, Stephen, there is no localization for other French punctuation. But I don’t want to stray from my specific question about the apostrophe and into a discussion of French punctuation in general. Not yet, anyway.

    Hrant, thanks for sharing JBM’s comments.

    I can understand the objection to a too-tight L-apostophe pair in French context. In my opinion, L-quoteright is often kerned too tightly in text faces, in general.

    L’Allemande is a classic test case in my own workflow.

    I agree that Minion Pro is too tight, in this regard. But is that Arno default setting still so crowded as to warrant a separate French apostrophe? I lean toward the feeling that it is overly pedantic, in this case.

    But what do I know? That’s why I’m asking. Perhaps I will be persuaded otherwise.

    I’ll look forward to more responses from French type folks.

  • Yªssin BªggªrYªssin Bªggªr Posts: 73
    edited March 2017
    Seconding what JBM seems to say, because of apostrophe kerning, there are a lot of problematic situations with french, such as "L'a" where things get too close. I've seen the problem with Minion too. This requires triplet kerning, which isn't always easy to implement, depending the software you're using, the kerning has to be edited manually. Maybe this is some sort of workaround…?

    * Glyphs has a nifty way to deal with it: https://glyphsapp.com/tutorials/contextual-kerning

  • Hi Kent,

    Slimbach isn't overly fussy. It's just that Adobe love too much the tightness in their fonts.

    Minion Pro is unusable in French with the kerning provided by the fonts (and in Italian too, I presume). Compare "L’étoile" and "L’utile" in Minion Pro to see what I mean:


    French users have protested, Adobe has conceded a contextual apostrophe in Arno Pro (still nothing for Minion). I agree, it wasn't there that was the most urgent need.
  • Minion 3 is in the works...

    > Adobe love too much the tightness in their fonts.

    Tightness is very important in a text font.
  • Concerning Arno, I don't feel the default kerning for "L'heure d'aujourd'hui" in Arno is unacceptable though the French kerning looks more congenial  to me. There are situations however where the standard kerning (the kerning I am getting with Xelatex) is crowded


  • Tightness is very important in a text font.
    I tend to agree. But English and French (and Italian, to a lesser extend) don't have the same use of the apostrophe. French uses apostrophes extensively, and the kerning of the apostrophe in Minion Pro (and in Arno Pro in some cases, like Michel Boyer has noticed) is uniquely based on its English use.

    These fonts are supposed to cover a lot of languages, and kerning is only done with English in mind...

  • Kent LewKent Lew Posts: 905
    Yassin, Thomas, and Michel — Thank you for your perspectives.

    Not all English-speaking type designers kern only with English in mind, btw. ;-)

    Being mindful that there are frequent combinations in French that need to be considered from a French perspective, I am still trying to uncover whether there is a compelling reason to address them with a) a languagesystem-specific solution, and b) with a separate, wider-fitted quoteright.

    Humor me while I continue to explore this out loud.

    It seems to me that the specific issues of greatest concern to French users regarding the apostrophe are:

        1) The L-quoteright combination, given the importance of the French definite article(s), and

        2) the apostrophe not crowding on neighboring accented letters (more frequently those following, perhaps).

    With respect to 2) — If quoteright-agrave or quoteright-acircumflex appear too crowded for French, wouldn’t they be equally crowded for, say, Portuguese or Welsh or Yoruba? Is there something actually different about the French language in this regard?

    My point is: Is this really a language-specific issue? Or should type designers just pay more attention to these combinations and include class-kerning exceptions as necessary (even if the pair is not crashing, per se)? Is it just a matter of broader consideration and better exposure to good examples?

    With respect to 1) — As I said, I think the L-quoteright combination is often kerned too tightly. Even in a non-French context, there are combinations like L’. and L’, to consider. (Again, Minion Pro fails on these counts too).

    Given that single-quoted all-caps settings of words ending in L are probably relatively uncommon, wouldn’t it make more sense to kern the L-quoteright combination for the more useful French context by default?

    Does it really require an alternate “French” apostrophe to resolve this?

    Turning the question around: should French type designers be considering including a separate, more tightly fitted apostrophe for non-French use, to be registered under {locl} languagesystem latn ENG? ;-)
  • Hi Kent,

    Thanks for your interest! It's very frustrating sometimes to be unable to use the kerning defined in a font, if you don't want that your French resembles Slovak (ď, Ľ, ľ).

    But the problem is much worst than a bad "L’" combination (and "L’utile" is correct in Minion — "L’étoile" is horrible): the elision is so much frequent https://en.wikipedia.org/wiki/Elision_(French)

    I don’t what the best technical solutions are, but I like your idea to provide an exception for English. ;)

  • Hi,

    If  “L’utile” is correct and “L’étoile” is horrible, the problem is not just the kerning of the apostrophe against the letter L  but implies also the kerning of the letters following the apostrophe.

    In Minion Pro (sfnt revision 2,015) the letter L is in a left kerning class for which the right kerning class containing quoteright has a kerning of -135. Maybe that could be reduced.

    The problem with “L’étoile” comes more from the fact that quoteright is in a left kerning class for which the letters c, d, e, o and variants are in a right kerning class with a kerning of -138. So all those letters when following the apostrophe will be moved left that number of units and cause the same disagreement. On the other hand, the kerning of u against quoteright is -34, and, if it looks fine, that may imply that replacing the -138 above by -34 would help. The kerning of  a and its accented variants following quoteright is -80, that of  i is -53, and that of y is -32.

    Maybe I am naive (I am certainly inexperienced) but considering the fact that quoteright and quotedblright make up a left kerning class by themselves, it seems to me that adjusting  their kerning against just three or four right kerning classes should solve the problem. I almost never use Minion and never took the time to try, but that would be the first thing I would do, if I had no choice (with LaTeX, I have other options). 

    Are those choices language independent? I don’t know. Is there any reason for a -138 kerning between quoteright and letters in the class e, o, d etc (those having a round shape on the left)? Was that just dictated by a geometric argument, independently of any linguistic consideration? I wonder.

    As for the idea of having an alternative quoteright, why not?  I would put it in the same left and right kerning classes as quoteright but I would personally put it in a style, not as a localized form (I assume GSUBs are executed before GPOS, I have not looked at those things in a long time). I will not comment on whether it is a need or a luxury.

    I’d also like to know what French typemakers think about it.

  • Christian ThalmannChristian Thalmann Posts: 1,896
    edited March 2017
    It doesn't seem to be the acute's fault either — «L’e» looks just as bad. And since it was implied earlier on that Minion is geared towards English: I'd say «we’d» also looks pretty shoddy.

  • attarattar Posts: 209
    > with a separate, wider-fitted quoteright.

    That seems worth exploring to me.
  • Adam JagoszAdam Jagosz Posts: 688
    edited March 2017
    I've been waiting for someone more qualified to say this, but in vain.

    Don't you think kerning is a general issue that is inherent to the design and not language dependent? Imagine someone wants to publish a fantasy novel including words and sentences in a fantasy language. No one can predict what sequences of letters and punctuation marks the author may design for that language. Firstly, all punctuation should be kerned against all letters that might virtually need it. Secondly, any introduced kerning classes should be checked for causing contextual problems like that. Period. Checking against known frequent pairs is only a means to detect general problems, not the sole purpose.

    As some of you have stated this issue impacts not only French, but also Italian and Portuguese. However no one suggests making a PTG and ITA locl features. And how can you be sure if there aren't another 10 languages impacted by this error?
  • Kent LewKent Lew Posts: 905
    Thomas — I do realize that the matter of French elision involves much more than just d’ and l’.

    The main thrust of my argument above was that I don’t think that fitting for French needs to be at odds with fitting for English (or any other language). Not such that a separate quoteright glyph is needed.

    It just needs to be approached with broader awareness and care.

    Michel — I think we all need to stop looking at Minion Pro as a frame of reference. I think we can all agree that the current version is a horrible example with regard to the apostrophe — in so many ways, not just French, as Christian’s examples point out.

    I am not challenging the idea that French benefits from more open fitting of apostrophe with accented characters.

    My question is: If one can accommodate all the French issues with appropriate kerning, does one really need a different apostrophe glyph? Are these issues any different in French than in any other language involving the same characters?

    Maybe I’m being stubborn, and it’s just easier to put in a special, wider quoteright. As some have said, Why not?

  • Kent LewKent Lew Posts: 905
    Thinking about this more, and for the sake of argument, let’s assume that a localized, more widely fitted quoteright, à la Arno, is in fact a desirable solution to these issues (and not just an overcompensation for angry complaints about past bad decisions ;-).

    If the only difference is to have an overall looser fit by adding more space to the sidebearings, then it seems to me that, rather than maintaining a separate alternate glyph, one could simply include a bit of localized glyph positioning:
    feature locl {<br>    script latn;<br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; language FRA;<br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; lookup looser_apostrophe {<br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; pos quoteright <30 0 60 0>;<br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; } looser_apostrophe;<br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; language ITA;<br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; lookup looser_apostrophe;<br>} locl;<br>
    Seems like a more elegant solution to me. (Again, assuming there’s a problem that actually needs to be solved. ;-)

  • Kent

    I think there are two different questions involved and one concerns kerning classes. Let me come back to Minion, not because it is a representative example, but because Adobe made public the source files for Minion Pro Caption. The file features.kern to be used by makeotf contains the line
    It then appears clearly that the kerning was based on geometry. On the other hand the class @ROUND_LC_RIGHT contains vowels and consonants. In French and Portuguese, so far as I know, only a vowel and h can follow an apostrophe. In English you have "I'd", "I'll", "we'd", "we'll" and I must confess that I am not shocked by the "we'd" Christian got.

    I think the class could advantageously be split and vowels be given a smaller kern value.

    That does not make irrelevant what you are proposing.
  • Kent LewKent Lew Posts: 905
    Michel —

    What you point out is a general caution to be considered when creating class kerning. It is easy to be overzealous with classes. There are various arguments for different philosophies of lumping or splitting.

    But I do not think the issue here is strictly (or even largely) one of whether round consonants and round vowels should be classed together.

    I would argue in the case of your Minion example that it is more the overall value that is in error, not necessarily the class composition (although perhaps that also).

    I think that most folks here would agree that both the round vowels and the round consonants in that Minion example deserve smaller values. I might suggest that they could be the same smaller value.

    Would you say that the English we’d and o’clock should be different from the French c’est?

    If indeed so, then perhaps you can make a case for splitting round consonants from round vowel classes.

    But I think a more pertinent question would be, Should the English literary e’en be any different from the French c’est?

    If you say Yes, then I suppose we have an argument for a distinctly different French apostrophe.
  • Nina StössingerNina Stössinger Posts: 151
    edited March 2017
    I’m intrigued to hear why one would want to kern vowels and consonants differently if they share the same profile. That seems to imply some kind of fundamental philosophical difference in how these categories of letters ought to behave — for their phonetics, across languages and orthographies, really?
  • Michel BoyerMichel Boyer Posts: 120
    edited March 2017
    I can't give any reference to justify my position. I just took Source Serif Pro and made it so that after an apostrophe, vowels get more spacing (the kern for e is changed from -89 to -40). The first two lines are the original font, the next are the modified version. I'll leave others decide but I prefer my last line.

  • Michel BoyerMichel Boyer Posts: 120
    edited March 2017
    Typo corrected.
  • Kent LewKent Lew Posts: 905
    Michel — You say that in your second examples “vowels get more spacing,” but in fact it appears to me that the d and c have come along for the ride with the class as well.

    I, too, prefer your second set. For both English and French — except that I’d loosen the kerning with all of those straights on the left as well (or the fitting, as the case may be).

    Perhaps the issue here is really le goût d’Adobe?! ;-)

    Something representative of my own taste:

    Here the round vowels and consonants are in the same class, English and French alike; but the quoteright-ecircumflex is an exception to the class value (+35 looser in this case).

    But tell me, would you think the c’est and l’être should have the same absolute value?

    If you say Yes, then again perhaps we have a good argument for a language-specific overall adjustment, rather than just some judicious kerning exceptions.

  • but in fact it appears to me that the d and c have come along for the ride with the class as well.

    My bad. I had not changed back some value to -89. Now done.

    More space on the left looks better for me too. As for the actual values I don't see why they should be determined by equations. 

  • Kent Lew said:

    Something representative of my own taste:

    Here the round vowels and consonants are in the same class, English and French alike; but the quoteright-ecircumflex is an exception to the class value (+35 looser in this case).
    Truly, after having tasted Adobe, I prefer le goût de Kent. ;)

    All joking aside, it's a good argument for your demonstration: with some exceptions, the kerning of the apostrophe should be treated the same in English and in French.

  • Kent LewKent Lew Posts: 905
    Michel — When you replaced your image with a changed version, you put the first two paragraphs of my subsequent comment into a false context. :-(

    For the record, I no longer care so much for the English in the second example. I feel that such different values between the consonants and the vowels is incongruous. But, as I implied, I personally feel that the default setting for English is a bit snug. Acceptable in that context perhaps, but I agree that it is not comfortable for French.

    I suppose that if a type designer does feel strongly that such a snug setting for English is the desired default in a font, then supplying an alternative for French & Italian use may be strongly advised.

    And perhaps even if the default is perfectly acceptable across the board, a slightly more generous alternative might still be un peu plus agréable for French (as you indicated for the Arno example)?:

    As for the actual values I don't see why they should be determined by equations.
    Not by equations. I provided a number only as a frame of reference to draw out the difference in my handling of the accented character.

    As C.H. Griffith once said to W.A. Dwiggins: The eye is the final arbiter.

    However, the eye comes with the bias of experience and exposure. As a type designer, sometimes it is useful to learn to see through the eyes of another. ;-)

    Which is what conversations like this help me to do. Hopefully others as well. So, for that I thank you, Michel et Thomas.

  • the eye comes with the bias of experience and exposure.

    A sobering reality to always keep in mind.
  • Michel BoyerMichel Boyer Posts: 120
    edited March 2017
    Michel — When you replaced your image with a changed version, you put the first two paragraphs of my subsequent comment into a false context. :-( 

    Sorry and the original is at home, I can't post it back here.

    My eye tells me that the spacing after the apostrophe in "L'être" is too tight but what bothers me is that I am not quite sure why. I think my personal rule is about this: if my eye tells me there is less space to the right than to the left of the apostrophe, there is a problem. If there is more space to the right, I don't mind.  

    Notice that the apostrophe is not part of the word. According the "Le Ramat de la typographie", you must write "Le mois d'octobre" to emphasize the month; Here « d'» is not emphasized. The apostrophe replaces the letter e in the preceding word (here de).

Sign In or Register to comment.