Specific diacritic designs depending on language



  • Speaking of the tittles on the lowercase "i", the i in añadió has a diamond tittle, while the two "i"s in diciendo have round ones in the illustration of the Real Academia edition of Don Quixote. Of course, occasional letters from the wrong font were a feature of early typography.
    Probably just a printing and/or scanning artefact.
  • John SavardJohn Savard Posts: 885
    I suspect a printing artifact. This larger image from volume 4 of the same edition shows that the normal form of the tittles in the typeface is round, and as well the form of the caret in that typeface, which is a bit unusual, is illustrated:

  • That circumflex is quite... quixotic.
  • Kent LewKent Lew Posts: 905
    That’s just an artifact of Google-book digitization and that particular scaled rendering. I wouldn’t read anything to definite into it.

  • Thomas PhinneyThomas Phinney Posts: 2,165
    edited July 2017
    David B: So, an axis for accent steepness? I like it!
  • An idea that's been kicked around recently, especially useful to vary based on size.
  • "So, an axis for accent steepness?"

    If you wish Thomas.:)

    I prefer not to define any instances specific to variables the user does not know the meaning of. The eight I've defined are intended as UI hidden, aka construction axes, from which and beyond the registered axes, one can do what they like.
  • The two lines used as umlaut in German are a lowercase ‘e’ written in German handwriting (Sütterlin). The ‘e’ is only two slanted strokes. Fraktur typefaces always had the two slanted stokes but when switching to Roman types, typeseters used the dieresis as they where considered close enough. The Hungarians still use stokes. To call it double acute is a westernization. So a script typeface with strokes for Umlauts should be fine in Hungary. 
  • @Georg Seifert Hungarian alphabet has both diareses and double acutes, ö and ő.
  • John SavardJohn Savard Posts: 885
    edited October 2017
    While Hungarian has both ő and ű, an a with a similar accent is not used there, and so one would have to be newly drawn. While Hungarians might not recognize the use of ű instead of ü in quotations from German, I would think that the important question is whether Germans, being familiar with Fraktur, would prefer such a design if available.

    In this era of digital typography, I'm surprised that someone hasn't tried it if it is what German readers would find preferable and more natural.

    Strangely enough, Unicode does include ȁ, ȅ, ȉ, ȍ and ȕ; I don't know which language uses these characters. A Google search, however, allowed me to discover that no language uses them normally, but the double grave accent is used as a tone indication for Croatian and Slovene when these languages are undergoing academic discussion.
  • Personally I'd think in German you'd have strictly vertical accents on stroke-based tittle versions of ä, ö and ü. When they are slanted like in the Hungarian umlauts it only appears so from cursive writing. For example these look distinctly wrong to my eye for German: űben, őstlich, ...

  • Georg SeifertGeorg Seifert Posts: 653
    edited October 2017
    All Fraktur typefaces have slanted stoked umlauts. Do you have an example with vertical strokes (in a none display typeface). It is either dieresis or "Hungarian" umlauts. Your example looks strange because strokes were never used on roman types.
  • From an 1893 ATF specimen book, this sample of Fraktur

    appears to indicate that the umlaut was composed of marks not dissimilar to the dot (or tittle) on the lower-case i. So the shape was somewhat elongated because other dots were elongated in Fraktur, but the shape was indeed intended to be a dot, and so the use of a round dot for it with Roman typefaces is entirely appropriate.

    Of course, there are contemporary digital Fraktur typefaces, no doubt available in sharper resolution, so I could be all wrong.
  • There are two kinds of variant letters. Some language- or locale-dependent forms are suitable for general use in that given locale, but there are also "idiographic" or "idiolectic" cases, where a form is adapted to match a particular local preference, but the users in that locale don't perceive that localized form as "neutral within their locale" but as one that possesses a strong local flavor of some kind. 

    For example, any roman typeface may have German umlauts that are built from base AEU or aeu letters while the accent is a small superscripted e. Contemporary Germans would recognize such umlaut vowels as such, but they would perceive those forms as historicizing, not as neutral. 

    Ever since the advent of digital type in the 1990s, we have a tend for visual convergence and globalization of Latin type. There are many readers now who were born in the 1990s and who were never exposed to pre-digital typesetting results. Their schoolbooks, the newspaper and books they read, websites and mobile phones they grew up with since they were children — they were all set using digital fonts. 

    Lots of those digital fonts were early digital fonts that did not have any localized forms. And for the younger readers, they were the ONLY ones they grew up with, so they're necessarily "right". 

    So when we introduce strongly-flavored localized forms, older readers may prove them as "finally they got it right", bit younger readers may see them as actually more odd than the globalized forms, or at best historicizing and old-fashioned. 
  • Of course there are also circular fashions (think hipsters) and political climate. For example today in countries such as Poland, Hungary but also many others, there is a right-wing trend and intense talk of national heritage and so on.

    In a climate like this, typographic forms that are historicizing, that are "specifically adapted for the unique flavor of a certain language", that emphasize distinctness of a given local culture rather than convergence within a broader context, become more sought after.

    This may be unfortunate if a type designer realizes that her/his noble goals of catering for local quirkiness which stems from love and respect for typographic detail or from exploratory curiosity ultimately ends up as a tool for nationalistic and populist tendencies — but it's fact. Designers of digital revivals of blackletter types will know what I'm talking about. 
  • The illustration in the Wikipedia article on Fraktur - and the typeface used for a German translation of the Kalevala - do show the umlaut as a double acute. So it is common in Fraktur, even if not absolutely universal.
  • Nick ShinnNick Shinn Posts: 1,802
    I’ve started putting alternate Ä, Ö and Ü with lowered umlauts in fonts, with a <deu> language tag.

    I had previously put them in a stylistic set.
  • In most cases, I'd still do that in a stylistic set, Nick. I mean, it is an especially German thing, but even there it isn't the "normal" setting. (Of course, if it's a funky strange display font, that might be a fine time to make it the norm for German.)
  • Vertical compression of diacritical marks over capital letters is not just a German thing. It's widely used in all European languages for all-caps setting in display context, to minimize the leading. It's suitable for any language, really, but just for all-caps, not mixed-case, and for larger sizes. 
  • Kent LewKent Lew Posts: 905
    So, perhaps registration in a {case} feature would make reasonable sense?
  • Nick ShinnNick Shinn Posts: 1,802
    If you look at German foundry and type-house specimens of the mid 20th century, you will find quite a few typefaces with low umlauts on the caps. I will post some images next week.

    A classic example of low cap accents: in the 1960s, Paris Match headlines treated acute, grave and circumflex in the same manner—a thin horizontal line.

    I suspect many languages would do away with diacritics, if that wouldn’t appear to be me-tooing English.
  • Ray LarabieRay Larabie Posts: 1,140
    When I see how diacritical marks are treated in 20th century display typography, it makes today's typefaces look absolutely timid. OTVAR headroom/legroom axes might open the doors to some interesting diacritical compression.
  • Kent: for dynamically composed accented characters, yes, 'case' based alternate diacritics would make sense. Of course, for precomposed combinations, the type designer can just use the alternate accent for the caps. Adobe's been doing this on most of their in-house designs for over a decade now.
  • For the extra-compressed (vertically) caps diacritics, I'd use the "titl" feature. "case" is sensible for forms that differ naturally, i.e. when mixed case is used. 
  • Nick ShinnNick Shinn Posts: 1,802
    Here is just one spread from the Linotype-Schriften specimen, 1967.
    It’s full of many faces with special treatment of the accented German capitals.

  • John HudsonJohn Hudson Posts: 2,242
    edited October 2017
    We think that besides width, weight and size axis, and beyond into the parametric axes, (and working with composite technology already in the standard), there is separate need for interoperable accent control. So, we envisage independent "above" and "below" cluster parametrics for x and y, opaque and transparent.

    I like the idea of some axes especially for marks, affecting them independently of base glyphs. I can imagine a few ways in which this could be used.

    I'm mostly thinking in terms of diacritics being composed on-the-fly using mark attachment, not precomposed and composite diacritic glyphs. So the transparent aspect would be variation of anchor positions, allowing users to move marks closer to or further away from the base glyph. That suggests a possible distinction — and hence independent controls — for distance mark-to-base and distance mark-to-mark. So it would be possible to e.g. reduce the distance between two stacked marks while increasing the distance from the base glyph. I can imagine this being an attractive feature for fine-tuning Arabic display settings, for example.

    This creates a separately addressable design space for accents, linked if it's so desired to other axes, but within that addition space, the font developer can create "culture-specific" instances. Here, the term culture ranges from a publication to an entire language.

    That's the intent of the OTL 'language system' tags, although they're mostly interpreted as linking to a language. So we already have a whole infrastructure in OT for addressing script and language system, in GSUB, GPOS, BASE and elsewhere. At the moment, the only way into that infrastructure for variable fonts is through existing mechanisms of glyph substitution. GSUB was extended in OT 1.8 to enable variable behaviour of substitutions within the design space, but its still predicated on GSUB switching GIDs. There isn't any way, at the moment, to link an instance or a range on an axis for a single glyph to different language systems.

    I've been pondering this a lot since a long conversation with Sam in Montreal: thinking through different scenarios in which language specific variable behaviour would be useful and wondering how we might leverage the existing infrastructure.
  • That’s all great but the infrastructure of line layout and variables were intended to be implemented together, not Separated by 25 years and artificial barriers of the technologies that grew in between. So it’s always interesting to see what’s going wrong just because of the past politics, and how little has changed in “round two.”

    In any case, if the system cannot deal with the indescrete nature of some typography, it can hardly be considered “whole”, and even if I can only do something with JavaScript, if it is based on axes that help users to better typography, then it must be more wholesome than doing nothing.
  • Maxim ZhukovMaxim Zhukov Posts: 74
    edited October 2017
    Nick Shinn:
    A classic example of low cap accents: in the 1960s, Paris Match headlines treated acute, grave and circumflex in the same manner—a thin horizontal line.
    So did Elle.

    Who launched that fashion? Hollenstein?

Sign In or Register to comment.