Serifed Schwa

Nick Shinn
Nick Shinn Posts: 2,216
edited April 2021 in Font Technology

Being derived directly from minuscule /e, the sans serif Schwas, upper and lower case, almost design themselves, but the serifed versions are problematic. Lacking much in the way of precedent, I’ve explored a variety of strategies, which I’d like to share for discussion. I’d be interested in hearing from any native Azeri speaker or people familiar with reading that or other languages which use the character—and in seeing how other type drawers have designed it.

(The type specimens feature the name of Azerbaijan’s president, in caps and U&lc.)

1. Scaled and rotated minuscule /e.  Font: Pratt Display, a newspaper face.
The capital /Ə looks perfect in the U&lc name, echoing the /e, but in all caps does it not perhaps looks a little out of place, as if it has strayed from a unicase font? This is the nature of the beast: should one accommodate the majuscule or the minuscule style for the cap /Ə, or try to compromise?

2. Divergent majuscule and minuscule style.  Font: Goodchild, a book ‘Jenson’.
The capital /Ə has a serif, like a flipped /C, and its crossbar aligns with that of /E etc. The lower case /ə is a rotated /e, but it doesn’t look quite right when its large open counter is compared with that of /a, which presumably has the correct ductus; perhaps I should have tweaked it a bit. Is the different angle of cap/lc crossbars a bit much, or cool? 

3. Divergent majuscule and minuscule style.  Font: Bodoni Egyptian.
The same tack as 2, but the quite different ductus of /a and /ə makes for a better effect, and the horizontal crossbar of /e and /ə is not as disruptive as Goodchild’s Venetian /e and /ə.

4. Cap style applied to both cases.  Font: Scotch Modern text cut.
(The opposite strategy to 1.) I determined that the lower case form looked too odd as a cap, so although based on a rotated /C, I added a serif. Then I made the lower case echo this, which is consistent with /S to /s; but in retrospect, I wonder.

5. Divergent majuscule and minuscule style.  Font: Richler, a book face.
Here the similar ‘superelipse’ form of /Ə and /ə serves to harmonize the schwas, despite /Ə having a centred crossbar, and a serif, unlike /ə. However, for the same reason, /a and /ə may be too similar in body text.

6. Divergent majuscule and minuscule style.  Font: Buslingthorpe, a tiny-x’d transitional.
The difference in angle of stress between the cases makes radically different treatments of the schwas appropriate. This should apply to more normally proportioned transitional styles. The transitional style of Latin typeface seems well suited to Azerbaijani.


7. Fanciful.  Font: Fontesque.


With the exception of #1, I’ve adapted the upper case /Ə to cap style rather than upsizing the /ə. It’s not necessarily the best method, but it’s what I’ve gravitated towards up to now.

Tagged:
«1

Comments

  • Andreas Stötzner
    Andreas Stötzner Posts: 798
    Minuscule: a rotated e will do in most cases, eventually shifting the bar a little (when the e is garamondish with high bar). After rotating, the right and the left side need their weights getting balanced. A serif? Maybe; in doubt: no. Perhaps that serif is better reserved for the small capital?

    Majuscle: on top definitely a terminal (serif) with reference to C, G, S. The bar by all means aligned with E, H. Since the prominent horizontal makes it look very broad, the glyph needs a bit of tweaking towards ‘narrower’.

    In fancy faces the leeway for variation is bigger of course, as always. The capital Schwa under 7. is lovely.

    I just stumbled over this one: https://de.wikipedia.org/wiki/Gottscheerisch

  • Craig Eliason
    Craig Eliason Posts: 1,440
    Really good thread!
    I find that however you finish it off, starting from a scaled up l/c schwa nearly always makes a cap that looks too wide. (Still unsure whether that’s because my brain can’t unsee it’s lowercase-e-ness and thus judges the whole thing too big.)
    It does seem to me that more squarish curves on the right side of the glyph serves the cap well and the lowercase poorly (because of the /a disambiguation issue). I’d pondered the terminals issue before but not the tension of the curves. 
  • Craig Eliason
    Craig Eliason Posts: 1,440
    Looking at other similar-but-not-identical letters between cases (I imagine Ss or Ww would be good candidates) should be helpful for discerning the approprriate "caseness" of a given style.
  • Igor Petrovic
    Igor Petrovic Posts: 307
    Recently I was wondering about the italic form of schwa.



    By quick check of a few typefaces at hand seems that—for lowercase—those that are "oblique" keep the bar horizontal, while those who have "true" italics make the curl. The same treatment that e gets. However, if the later, lowercase schwa looks even more like the italic double storey a.

    On the other side, for caps seems there is no consensus, and that true italics sometimes keep the horizontal as well.

    What people here think about this?
  • Nick Shinn
    Nick Shinn Posts: 2,216
    Yes Igor, similarity between /ə and /a in body text is an important Schwa issue, especially in two-storey /a italics. That Recursive tactic is clever!

    On the other hand, Western readers of Futura etc. manage OK with its similarity between /a and /o, so, absent opinion from native Azeri typographers and readers, we’re rather working in the dark.

    But perhaps disambiguation is more important for Azeri readers, as they have had to contend with both Cyrillic and Latin scripts, not to mention Arabic, so they may need all the help they can get in deciphering typography.
  • Adam Jagosz
    Adam Jagosz Posts: 689
    edited April 2021
    I remember seeing some almost-Azeri (Turkish perhaps) designer's post on IG stating that the schwa should have a serif (@Andreas Stötzner which would rather imply an “in doubt: yes”—but perhaps more so for Azeri usage than phonetics?) and that the bar should not ride too low (i.e. rotated Garamond /e is a no-go).
  • Nick Shinn
    Nick Shinn Posts: 2,216
    There is one other character structurally similar to /Ə: uni042D, the Cyrillic ‘Backwards E’, /Э.
  • Ray Larabie
    Ray Larabie Posts: 1,435
    edited March 2022
    @Tural Alisoy
    Is the following true? If you're using the ə at 01DD (Latin Extended B ) you also need to include it at 0259 (IPA Extensions). Azerbaijani text usually uses the 0259 encoding.
  • John Hudson
    John Hudson Posts: 3,225
    @""Ray Larabie"
    U+01DD and U+0259 have different case mappings. U+0259 maps to the uppercase Ə U+018F , which is why it is used for Azeri. U+01DD is an African character which maps to the uppercase Ǝ U+018E.
  • Igor Freiberger
    Igor Freiberger Posts: 279
    edited March 2022
    This is how I draw the Schwa/schwa. Some fonts, especially from ParaType, use the schwa as the rotated e (uni01DD). I'm not sure if this is feasible. In this sample, I use this approach (schwa design for the rotated e) as an experiment since the e arm seems a bit too long in uni01DD.


  • Tural Alisoy
    Tural Alisoy Posts: 52
    Ray Larabie said:
    @Tural Alisoy
    Is the following true? If you're using the ə at 01DD (Latin Extended B ) you also need to include it at 0259 (IPA Extensions). Azerbaijani text usually uses the 0259 encoding.
    We use encoding U-018F and u-0259. We do not use encoding 01DD. 
  • Tural Alisoy
    Tural Alisoy Posts: 52
    edited March 2022
    @Igor Freiberger I think, the midline of the true "ə" shown in comparison should be raised a little higher. 
  • Nick Shinn
    Nick Shinn Posts: 2,216
    Thanks for your valuable comments, Tural!
  • Tural Alisoy
    Tural Alisoy Posts: 52
    My new post about Schwa. https://taft.work/schwaedesign
  • Igor Petrovic
    Igor Petrovic Posts: 307
    Tural Alisoy said:
    My new post about Schwa. https://taft.work/schwaedesign
    Bookmarked, thanks Tural! 
  • Denis Moyogo Jacquerye
    Denis Moyogo Jacquerye Posts: 197
    edited October 11
    The statement ‘always necessary to use the upper terminal of "c", "s", and sometimes "a" letters’ should seriously be nuanced.

    There’s plenty of evidence the lowercase is typically designed without such upper terminal. I guess one can make sense of it when designing it like Cyrillic э with a top serif, but that would be more like a small capital.
    It’s not clear where the top drop terminal form comes from. There’s nothing inherently wrong with it but it’s rather unusual and may work better with a reversed contrast, like for reversed c ↄ, when there is oblique contrast.

    For the capital it does make more sense to have a top similar to Э, but the large lowercase form is very common nonetheless. In the 1920s Э was used as the capital form in some cases in Azerbaijani.

    FWIW Monotype had a few shapes for lowercase and capital schwa in metal type, typically based on a turned-e. There’s at least one variant with a serif on the entry stroke of the capital. It’s not clear whether there was a distinction between Latin Ə and ə  and Cyrillic Ә and ә.


    For the italic, it can be based on a turned italic e or a slanted e. It’s a matter of preference or necessity.

    In general, it’s more important that the letter form fits well with the rest of the font.
  • Igor Freiberger
    Igor Freiberger Posts: 279
    edited October 11
    @Denis Moyogo Jacquerye  Just to make it more specific: When you say "there’s plenty of evidence", are you referring to digital typefaces, metal type or manuscript stuff?

    I look at digital fonts made until mid-2010s with a bit of care because they are mostly bad regarding languages outside Western Europe (of course, for Latin fonts). But I believe you have sources far more robust than digital fonts —like the metal samples you show above.
  • Tural Alisoy
    Tural Alisoy Posts: 52
    @John Hudson
    Thank you for your comment. I also marked the letter "a" depending on the style of the font. Yeah, always this may not be the right option. But the other two letters "c" and "s" will help you form the terminal of the schwa. In the options you show on the screenshot, I would recommend that the middle line be raised a little higher. In general, I don't know how the upper terminals of c and s are in your style, but it looks good in the version you show now. And, At one point, I think it was overlooked because of the schwa's resemblance to the letter "e". Therefore, it was overlooked in "IPA".
  • John Savard
    John Savard Posts: 1,135
    This is an interesting discussion. For showing how words are pronounced in dictionaries, a schwa being an upside-down "e" is perfectly suitable. As a letter in the alphabet of a real language, as noted here, all kinds of questions arise: the reversed stress, what should the capital form be, how does one italicize it, and so on.
    But there's one disturbing thing here: apparently there are only two codepoints for schwa in Unicode. One is specifically for a Pan-African alphabet. So the character there should obey the rules of writing in that alphabet. But the other one is for IPA.
    This means that a font designer will have to choose if that character follows the rules for writing texts in IPA or the rules of writing in Azeri.
    Of course, we now have an OpenType feature which allows the language being used to be indicated, which saves adding two additional codepoints to Unicode - non-IPA general phonetic character, and as a character within the Azeri alphabet. (Of course doing it that way would mean that, say, the Bulgarian alphabet would be completely separate from the Russian alphabet, since there are letterform differences in the italics.)
    Forcing people to use advanced OpenType features may be unavoidable, but it does cause problems.
  • Denis Moyogo Jacquerye
    Denis Moyogo Jacquerye Posts: 197
    edited October 12
    One is specifically for a Pan-African alphabet.

    It's more complicated. Ə ə was used in the 1927 African Alphabet and is used in Cameroon for example, Ǝ ǝ was used in the 1978 African Reference Alphabet and is used in Nigeria for example. Some languages even have used both forms interchangeably.

    For context Ə ə was used in Azerbaijani in the 1920s and 1930s, the similar Cyrillic letter was then used in the Cyrillic alphabet and the 1992 Latin alphabet used the letter again as we know.

    IPA uses the same Unicode character as Azerbaijani.
  • Nick Shinn
    Nick Shinn Posts: 2,216
    edited October 13
    While it’s true that local type designers can be gatekeepers who protect their native scripts from the depredations of foreign designers by advising “correct” forms, it does strike me as strange when I compare the orthodoxy this promulgates with the liberties that Western designers take with the mainstream of the Latin alphabet. For instance, we have two-bowl g’s with ears shooting off all over the place, plus single-bowl g swiped from italic, not to mention the Souvenir-Kabel version with its bizarrely attached tail. That’s not how you do it, but don’t tell Morris Benton and Rudolf Koch!

    However, the orthodoxy of letter-forms in non-Western alphabets is deceptive, as it is represented by the mass of foreign-designed types which are obliged to include such characters, and therefore inclined to follow the pattern of the gate-kept default.

    On the other hand, native designers of non-Western alphabets are more likely to take creative liberties with the appearance of their language. This much I deduced from an exhibition of student work I saw when I was in Saint Petersburg, Russia, a few years ago for an ATypI conference, and also by comparing the standard model of Greek type taught and advised in the West with designs from the well-respected Greek foundry Parachute, which break many of the “rules”.
  • John Savard
    John Savard Posts: 1,135
    Nick Shinn said:
    On the other hand, native designers of non-Western alphabets are more likely to take creative liberties with the appearance of their language.
    Well, it is absolutely true that people who speak languages other than those which use the Latin alphabet have as much right to quirky display typefaces as we do.
    But native speakers - or, should I say, native readers - know what the rules are, and they know which ones can be broken and when. A foreign type designer who is just trying to design a plain text typeface for everyday reading, but because he doesn't know the rules, does things with certain letters that are extreme and bizarre... will find his typeface not adopted much.
    It's hard enough for a non-native speaker to learn all the rules and apply them; to learn in addition their relative importance, their contexts, and so on so as to be able to safely bend them or break them requires even more learning - and this is so ambitious that there aren't yet resources to be consulted to help with this.
    That doesn't mean it's impossible for the non-native; one can work with native speakers, and one can also study display faces for the script in question. But it is an order of magnitude more difficult.
  • Nick Shinn
    Nick Shinn Posts: 2,216
    @John Savard

    and they know which ones can be broken and when.

    Most of the “rules” are subtle conventions which, when contradicted, upset the typographic cognoscenti but have no effect on readability for the lay person.

    (Which is not to deny the importance of maintaining diversity, heritage and cultural identity.)

    These “rules” are always about what an individual character should look like, considered in isolation, which is at odds with how typefaces function—as systems in which different characters fit together, side by side. Even in this thread, posts by John Hudson, Igor and Tural do not show the schwas in actual words. Tural’s posts of historical pages are, however, brilliant!

Categories