alerque
Consider an alternative ‹ĥ› with the circumflex not below the ascender height
Describe the bug
‹ĥ› is wrong ugly.
Screenshots / logs
Additional context The first ‹ĥ› is how Libertinus renders it now. The second ‹ĥ› is how most fonts render it. The third ‹ĥ› is an acceptable, if rare, alternative.
‹ĥ› is used in Esperanto. I'm the closest thing to a native Esperanto speaker that exists. My Wikipedia username contains an ‹ĥ›, so I've seen many ‹ĥ›'s in all kinds of fonts. Libertinus' ‹ĥ› is wrong.
"But Fred, Esperanto is a made up language, surely we can make up a glyph", some might say. No. Zamenhof was the one who could have chosen Libertinus' ‹ĥ›, but he didn't. Now Esperanto is old enough that there are ways of doing things and the ‹ĥ› is clearly wrong.
What other language(s) if any use h with a circumflex? Since this isn't a language specific combination it seems like we need some research and precedent from other places it may appear as well before changing it up. The scope of who else besides Esperanto would be affected should be ascertained first. Do you have any knowledge of that?
Ĥ ĥ are also used in Bassari language in Senegal.
This is what ĥ looks like in the law that sets the official Bassari orthographic rules in Senegal:
I don’t think the form is normative, it just happens to be the form in the font used in the Journal officiel de la République du Sénégal in which laws are published.
That said, the second and third forms in the OP screenshot are the most common forms, generally speaking.
I guess if one digs enough, examples of ĥ with the circumflex low on the right of the ascender can be found.
Here one I found after going through a dozen of Zamenhof’s early Esperanto books on https://www.onb.ac.at/bibliothek/sammlungen/plansprachen/digitale-medien/ludwik-l-zamenhof/ (namely Zamenhof, Ludwik Lazar (1889), Dr. Esperanto’s international language, Warszawa, page 35):
This however seems to be anecdotal as the other two forms are much more common.
@moyogo That's interesting. The ŭ in use there is also very strange indeed. I would caution you though, just because something has happened once in history doesn't make it right. It's certainly news to me that such an old source has this ĥ, but I maintain that this is just an eccentricity of a single book publisher and not representative of modern Esperanto usage.
There's no right or wrong on any of the three forms. They can all be seen throughout the history of Esperanto typesetting, though admittedly the present version is the least common. I'd generally refrain from categorising in right and wrong upon such questions as this is about design which is about aesthetics, a highly subjective matter. It'd be wrong if it could too easily be mistaken for a different glyph but that's not the case here.
I very much respect you @georgd and love your EB Garamond font but it looks wrong to me. I've owned books in Esperanto, contributed to Esperanto Wikipedia, et cetera. I'm not sure why the weight of my experience is being discounted by a single scanned page.
Now, if you just object to the word "wrong", fine...? Given it happened once, and early too, perhaps it's a rare alternative. So, I'll edit my title...
@ctrlcctrlv thank you.
The categorisation as "wrong" is indeed in the foreground of my comment. (Just for the record: in my eye this variant is not ugly – I even find it more pleasing than the more usual variants.)
About the facts: you might visit the Esperanto Museum here in Vienna (https://www.onb.ac.at/eo/museen/esperantomuzeo) and find that the incriminated variant hasn’t been used only once. Yet I do confirm that it’s really rare.
Usually I wouldn’t, but in this case I do recommend to reconsider the design choice. As Esperanto is usually acquired as second/third/ ... language and reading practice in it is mostly not as advanced as in the first language, design variations like this might affect readability, especially in a text font where the glyphs shouldn’t stand out positively or negatively. That’s much less the case for ‘more established’ glyphs like Ö or ç where lots of variants don’t pose a problem.
@ctrlcctrlv Nobody is discounting the weight of your experience or even the weight of your opinion — but it doesn't make you automatically right. In fact I think it's pretty clear that your initial argument was an overstatement of your case and also overreached a bit. My first comment addressed the over-reach (by asking about other languages even though your report made it sound like Esperanto was the only factor involved) but I will also say that the overstatement in your choice of words and tone is not endearing your argument. I will still consider this, but please back off from the dogmatic assertion of right and wrong in regard to something that at least can be a matter of style and preference. It might be best for the Esperanto ecosystem to not rock the boat and stick as closely as possible to the most common design choices, but before we go there lets actually understand what decision needs to be made.
A non-exhaustive set of questions I would expect to have answered before changing this:
- What other languages are effected?
- Why does Pragmatica Esperanto (one of three fonts showcasing Esperanto's h-circumflex on Wikipedia) also place the circumflex above the shoulder / beside the ascender instead of above the height? If an Esperanto focused font made that choice, it likely isn't 100% wrong and might even be good.
- What other fonts are there that made this choice and is there any paper trail on their reasoning?
- Why did one of Zamenhof's early books go to press with this design? Because they didn't have a choice? Because the author liked it that way? Did he later express a reason for a later change or is this something he still accepts?
- What other letters use similar diacritic positioning in Libertinus right now?
- What if any languages are affected by stacked diacritics in this position?
@moyogo Thanks for the references—both good finds.
The reference for usage in Bassari should probably be added to Wikipedia! How did you find that anyway?
Scriptsource lists it as h with tilde (https://scriptsource.org/cms/scripts/page.php?item_id=wrSys_detail_sym&key=bsc-Latn-SN) so, what @moyogo found is perhaps a stopgap as ĥ being available whereas proper support for arbitrary letter/mark combinations is still rare: h̃
The fact that a lot of fonts place the circumflex in ⟨ĥ⟩ in the mathematical middle, not attached to anything and floating in mid-air, is so disappointing. Please don’t use that positioning choice in Libertinus. (And system fonts are not precisely exemplary of what a comprehensive font’s diacritic design should be; OP seemed to imply that since most fonts go that way, then that’s what it should be. IOW: never look at Arial for inspo.)
@fitojb It’s not just Arial.
At a quick glance:
- form 1 (low, next to ascender):
- Libertinus Serif
- Pragmatica Esperanto
- form 2 (centered above the whole glyph):
- Brill
- STIX2 Text
- Segoe UI
- Lucida Grande
- Libertinus Sans
- Noto Serif
- Source Sans Pro
- Source Serif Pro
- form 3 (centered above ascender):
- Andika
- Charis SIL
- Doulos SIL
- Gentium Plus
- Castoro
- Sylfaen
- SF Pro
- Noto Sans
I doubt this trend is not reflected in Esperanto documents published in the past couple of decades.
The point I should have clarified when I provided the example with the form matching the current design is that it was one out of the dozen I went through. So, while it does occur, it is not the most common, even in the early days of Esperanto.
The Esperanto language Wikipedia page for h-circumflex states (Google Translation):
The use of a roof-shaped accent, the hat, to change the grapheme H into Ĥ is an invention of L. L. Zamenhof. Its lower case has at least three typographic variations: with the hat above the bar of the h, with the hat above the arch of the h ("on the knee"), or with the hat centrally above both.
No mention or even suggestion is made that would lead me to believe the form is considered wrong en Esperanto.
On that page besides the sample showing all three, the other sample uses Linux Libertine O and evidences the position this font apparently inherited: lower above the shoulder of the h.
Given the availability of other high quality fonts of both Serif and Sans variants that use each of the other style choices and this font being probably the leading one using the 3rd alternative I think we're probably on good ground preserving this stylistic choice.
The one thing that does actually look wrong to me is that Libertinus Sans should make the same choice as it's Serif cousin.
I realize that's literally the opposite outcome being advocated by the OP, but the evidence isn't adding up yet. I'm willing to leave this open for a while in hopes of hearing from more Esperanto speakers.
I am sorry if you felt like I oversold. I did not do so intentionally.
To my memory, I first started speaking complete Esperanto sentences at 10.
I had an Esperantist in my family (grandfather), so it's possible I knew quite a few words before then, but I only remember being "quite good at it" at 10.
I was given this book by my grandfather, already well loved and had many of his notes...
It was destroyed in a household accident. He replaced it for me with the newer edition, since my dad was too poor to and didn't see the point.
Naturally, I've read the ''Fundamento'' and the ''Unua Libro'', both in PDF form. I am a member of the Universala Esperanto-Asocio, and received its newsletter for years. You've reminded me I need to inform them of my new address.
I've also always been interested in fonts. If I would have seen the Libertinus-style ‹ĥ›, I certainly would have noticed. Depending on when I saw it, I either would have posted on lernu.net forum, ##esperanto on Freenode, or Esperanto Facebook group some joke like...
ĈIMOMENTA NOVAĴO! Ĉu rompita presilo... aŭ, aĵo tre sinistra!? ‹h›on, ni, intervjuis. Ria mesaĝ': “Tuj, relokigu mian ĉapelon tuj, barbaro! Tuj!” Xerox, komento demandita antaŭ du tagoj—ankoraŭ ne. BREAKING NEWS! A broken printer...or something more sinster? ‹h› in an exclusive interview: ‘Put my hat back this instant, you barbarian! Right now!” Xerox, asked for comment two days ago; as yet we've received none.
Really, the jokes write themselves, and this would not be a missed opportunity, especially if I saw it in a print source. Certainly I'd circle it and not think it a computer glitch. :-)
So, if I oversold, it was because I thought I'd seen enough ‹ĥ›'s in my life to know the allowable variation in the glyph. Apparently not.
Also, @fitojb, sorry to inform you, I actually find the normative ‹ĥ› with the ĉapelo¹ centered, more visually appealing, perhaps because I'm much more used to it. It depends on if we're making type that will look most natural/not strange to Esperanto speakers, or if we're making decorative type that will break design barriers, in my opinion. I think for a book face we should use normative forms that will not surprise readers.
However, seeing as @alerque gave your comment 👀, he seems to agree. I much prefer either form to the current ‹ĥ›, so, I won't complain beyond this comment if it that's what is implemented.
¹ Literally, hat. This is really what we call it, I don't know why; Vikipedio says we can also call it a tegmento, lit. roof, which I find cute!
I also have asked other Esperanto speakers to comment, on Twitter:
https://twitter.com/HW_BEAT_THAT/status/1280127676326371329
Other #Esperanto speakers are asked to kindly give their opinions on the correct shape of the letter ‹ĥ› here: https://github.com/alerque/libertinus/issues/329
And IRC:
09:08:08 josomebody1 saluton
09:08:23 -- josomebody1 is now known as josomebody
09:19:57 copypaste kiu estas la ĝusta formo de la nobla ‹ĥ›o? bv. komentu TIE → https://github.com/alerque/libertinus/issues/329 ←
(bv. ne ĉi-kanale, eble ili ne min kredos haha, ĉar ververe ŝajnas ili jam(!!) ne min ŝatas :D) se vi tempon havas. antaŭdankon.
09:20:06 sham2 Saluton al ĉiuj!
09:20:07 copypaste josomebody: ^ :-)
[124] [irc/freenode] 5:##esperanto{142}{+cnt}
I knew that this was brought up already: https://sourceforge.net/p/linuxlibertine/feature-requests/167
Thanks for digging that up @georgd. I kind of figured there must be some back story but I'd only searched through this project's old issues.
For the record, I am not currently inclined to change the design. In fact I'm currently keeping the issue open to track fixing the Sans family to match the current Serif family design. This default design choice could certainly be evaluated if it could be demonstrated that there was a heavy majority of users that preferred a different default, but as long as their is a significant minority that likes our current rendering I think it's okay to be one of two fonts catering to the underdog variant. Evidence suggests it isn't outright wrong — it's a valid if less common atheistic choice. Even if we did change the default we'd need to keep the current rendering as an alternate.
That being said I might be willing to accept a contribution which added a non-default alternate with the circumflex anchor above the ascender. I don't think I'm interested in the other variant style at all (floating a mile in the air above the ascender height but centered over the whole glyph).
By the way one of my favorite glyphs in Libertinus Serif is the stylistic alternate h:
\begin{document}
\font[family=Libertinus Serif,size=48pt,features=+salt,style=Italic]
Wash your hands!
\end{document}
I guess it depends on what this font is supposed to be.
Is it supposed to be an experimental typeface, where the glyphs are all supposed to be breaking design barriers, or is it supposed to be a text face, engineered to cause the least surprise to readers, because surprising readers slows them down.
I can't decide that for you. I guess I thought it was the latter, but you meant it to be the former. (Well, maybe @khaledhosny didn't mean it that way, but you do.)
Oh and by the way, nobody who actually speaks the language likes the current design.
- Me
- the guy from 2013
- @chemoelectric on Twittter [1]
Have fun with your display typeface, guys.
@ctrlcctrlv I’m sorry, there are some misunderstandings here.
- The "guy from 2013" found the floating version ugly, the ascender-top version the nicest and the over the shoulder version not bad. At the time, Linux Libertine had floating circumflexes on h in all italics and semibold and the shoulder-top only in regular and bold. These haven’t been consolidated yet.
- The design decision was not met by @khaledhosny but by Philipp H. Poll.
For the record, I am not currently inclined to change the design. In fact I'm currently keeping the issue open to track fixing the Sans family to match the current Serif family design [...]
The thing is, there is no single "current Serif family design". Italic, bold italic, semibold and semibold italic, as well as the whole sans family are showing the "floating" variant, so apparently Philipp himself at a rather early stage already gave up the shoulder-top variant which is only present in regular and bold. @chemoelectric points to another issue that I wouldn’t have detected on my own: I wouldn’t put the shoulder-top version in a mathematical text.
@moyogo That's interesting. The ŭ in use there is also very strange indeed. I would caution you though, just because something has happened once in history doesn't make it right. It's certainly news to me that such an old source has this ĥ, but I maintain that this is just an eccentricity of a single book publisher and not representative of modern Esperanto u
I guess if one digs enough, examples of ĥ with the circumflex low on the right of the ascender can be found.
Here one I found after going through a dozen of Zamenhof’s early Esperanto books on https://www.onb.ac.at/bibliothek/sammlungen/plansprachen/digitale-medien/ludwik-l-zamenhof/ (namely Zamenhof, Ludwik Lazar (1889), Dr. Esperanto’s international language, Warszawa, page 35):
This however seems to be anecdotal as the other two forms are much more common.
I would ignore this document. It is too strange; it is in English but has German quote marks and fails to capitalize ‘Scotch’. And it is not an attempt at handsome printing.
As someone else hints at, my taste is dictated largely by what one would expect to see in a mathematical text. But also there are many examples among commercially available fonts and they favor one or the other design with circumflex high. Storm seems generally to favor something midway between the two placements. I suspect the horizontal placement is simply done to taste, as I would do it.
My own feeling is that whatever the notable type producers are doing is what should be considered ‘proper’ Esperanto, because that is what one would see. There is no ‘tradition’ of a national type style. Esperanto style doesn’t even dictate a preferred form of the quote marks--all kinds are used (but extra space as in French printing is discouraged).
Generally, this is more about the fact that Libertinus’ anchors on "bdhk" is not the most common one, and will seem odd to many users, regardless of the language. The only exception to this is probably Gaelic Irish, where it is quite common, if not more common, to have the dot above next to the ascender instead of above it.
It’s always possible to find recent examples for any variants but the question should really be what is the more appropriate design. Which is what the original post tried to say.
Personally, I wouldn’t favour the current design unless there’s an aim to save vertical space, at least for the non-Gaelic Irish letters.
I thought of the vertical space saving argument as well and can't really support it as not much is saved. The accented uppercase letters require almost as much vertical space and the <ĥ> is really rare in Esperanto.
Generally, this is more about the fact that Libertinus’ anchors on "bdhk" is not the most common one, and will seem odd to many users, regardless of the language. The only exception to this is probably Gaelic Irish, where it is quite common, if not more common, to have the dot above next to the ascender instead of above it. ... Personally, I wouldn’t favour the current design unless there’s an aim to save vertical space, at least for the non-Gaelic Irish letters.
My (Old-) Irish lessons lie far away in the past so I might be mistaken. But h with dot above doesn't make sense to me from a Gaelic perspective: The dot above marks lenition, where plosives and /m/ become fricatives and /s/ > /h/ — /h/ doesn't fit in here. Modern Irish orthography replaces the dot with a postponed
So, I think that's no case to worry about here.
‹ĥ› is not that rare, but you are right to an extent, some people have advocated replacing ‹ĥ› with ‹k›. However, ‹q› is also rare, and no one uses that as an excuse to make it look weird.
By the way guys, it's not really an excuse to appeal to Gaelic as a reason to maintain the status quo, as of fontforge/fontforge#3645. I should know, as that was my pull request.
You can give ḣ a custom decomposition, so from:
h + ◌̇
To
h+ uni0307.lowanchor
(Give uni0307.lowanchor a different anchor class name.)
Note that all characters that will be referred to by fontforge/fontforge#3645 user decompositions must be encoded into the PUA, as only by hex can they be referred as yet due to complicated internal FontForge issue I never got around to solving. If you want, you can null the PUA char's encoding in post-processing via e.g. ttx or what have you (sfdLib I think).
The reason I opened this bug was I was getting ready to write a paper...
And right when I was, something I was talking to @skef about reminded me this font exists.
But, I remembered, I'll probably want to put my Wikipedia username on it, and the ‹ĥ› is simply unacceptable, something I'd not want to turn in unless some bug that wasn't my fault and that I didn't notice caused it. (Printer...something something...PostScript...I don't know. Let's say an esoteric bug in a print-to-PDF driver where it misunderstands anchor positions, and for some reason my document wasn't using the precombined character, or it got NFD normalized along the way to the driver because some developer loved combining characters. Stranger things have happened.)
Anyway, in this discussion we've determined all speakers who've spoken up don't like it and the original author of this font gave up on it and started putting it in the right the correct an æsthetically appealing position in later fonts in the series.
Pretty much everyone seems to agree it should change now, at least for non-Gaelic. I think the only remaining issue is whether form 2 or form 3 will be used. Unfortunately for me, most of you seem to like form 3. I still think form 2 should be default, but...it is an allowable variation, I'll admit, not like form 1.
So I'd even be OK with implementing form 3 if it means form 1 goes.
(Well, except @fitojb, but they literally just seem to just be trolling at this point. Downvoting everything I write. This isn't Reddit, up boats don't matter. Look how many down boats I got here @fitojb...and yet consensus came around to my suggestion. Find something better to do with your time than clicking thumbs down on every innocuous comment. Try making an actual argument, perhaps. Or just leaving it to people who care because they actually know this language and would use this glyph.)
Mission accomplished, I suppose. Who's going to be writing the PR? I don't mind doing it if it's okay for me to take a while because I have a lot of other things to do...but I did write the feature I explained in the post above exactly for this situation, so I already know exactly how to use it. ┐(´～｀)┌
Since I've been brought into this, I should note that I like the aesthetics of the current glyph.
