Palimpsest, contemporary text font

Updated description: this is a contemporary serif family I'm working on since January 2010. It was initially just for personal use, but the project grow up to embrace complete Latin, Cyrillic and Greek scripts, besides phonetic alphabets, arrows and dingbats.

It's a text font and the family is planned to have several weights in roman and italic versions. I believe it will be released in early 2013.

Goals:

1. it must be suitable for books and magazines, with more contrast than contemporary typefaces like Meta Serif, Greta Text or Stuart.

2. it must be clean and legible, with high x-height, generous counterspaces and reduced ornamental elements.

3. it must produce a strong horizontal impression, with well defined lines like the ones achieved with Guardian Text, Utopia or Charter, for example.

4. it must be adequate to academic, legal and technical texts with harmony between usual text, numeric data and acronyms.

Many glyphs have alternate designs, as |a|b|f|g|q|y|W|Ж|К|. Target sizes are 9 to 32pt. Small caps are taller than usual. Petite caps are also included. Numbers, currency symbols and related signs are available in 10 variations. Palimpsest is a codename.

Comments are very welcome.

Historic
2010.01: lowercase a-f, h-j, l-u.
2010.02: lowercase complete, marks and basic diacrits.
2010.03: uppercase complete, punctation, European languages.
2010.04: small caps, numbers, non European languages.
2010.05: figurine variations, bullets, dingbats, fine-tuning.
2010.06: many symbols and additional language support added.
2010.07: numbers and currencies in ten styles.
2010.08: general revision, mathematical operators, symbols.
2010.09: combining diacritics and precomposed glyphs.
2010.10: spacing, African languages, dingbats, fractions.
2010.11: support for many non-European languages and IPA.
2010.12: complete revision, UPM set to 2048.
2011.01: dozens of glyphs added to increase language support.
2011.02: first Cyrillic and Greek drafts.
2011.03: linguistic additions. Glyph naming revision.

Workflow
Initial lowercase design was made in Illustrator CS4.
Further development was done completely in FL Studio 5.
Final work will need FL Studio 6 due to the large number of glyphs.

AttachmentSize
Version085.pdf309.55 KB
Especials.pdf22.25 KB
Version0968.pdf271.59 KB
David Sudweeks's picture

Igor,
Nice start. So far it's feeling very tight. Look very closely at the negative spaces in between the letters and remember to design those too.
Why start in Illustrator?

Igor Freiberger's picture

Thanks David. I'll follow your advice and observe in detail the negative space. I'm designing in Illustrator because I know this app well while I'm a novice in FL.

David Sudweeks's picture

While it can at times require all your patience to go from an environment where you are comfortable and in control—to one where you are again a novice—in this case it's worth it. Here's why: FontLab requires you to make certain decisions about point type and placement while drawing; the same decisions that Illustrator more or less makes for you. FontLab gets you in the habit of drawing glyphs using the simplest construction possible. This is important for making malleable glyphs that you can adjust as context requires, also keep in mind rasterizing, hinting, and interpolation should you choose to interpolate.
Another reason for using FontLab for drawing that isn't obvious at all until you do it, is its silent emphasis on thinking about type as a system of interdependence. The decisions about how to space a glyph are best made while the glyph is still being drawn—in the context of the other glyphs around it. As you use it, you'll grow to appreciate its handling of curves, one of the prominent nice things about the software.

Igor Freiberger's picture

Thanks again, David. You convinced me. Actually, I already tried to keep the vectors simple at IL, with both nodes and handles at integer positions. Handles are also almost all orthogonal (0º/90º).

When possible, I tried to follow a 4x4pt grid because 4 is the minor common divisor between the two systems of type measure (1000x2048) and this may help a later version destined to screen with TT hinting. Anyway, this is just a strange idea I had, not sure it causes practical results.

But it must be better to do the entire job in FL for the reasons you did explain so clearly. As I'm a hobbyist type designer, the extra time FL may demand now is manageable and surely it will pays off later. Thank you very much.

Jean Paul Beumer's picture

Hi Igor,
Looking very nice indeed! Cudos!

I find that the lower and possibly upper diagonal on /k/ feels a bit too thin, especially compared to other diagonals.
I think /x/ leans a bit too far to the right, maybe broadening the base to the right will fix that.
/w/ stands out a bit because the middel part doesn't reach x-height.
Maybe /,/ and the middle part of /s/ are a bit too heavy?
The finial on /y/ looks a bit out of the ordinary...

I like the fact you've designed a very subtle mix of sans and serif, like in /a/ and /f/. It reminds me of Stone Informal, although that's more a mix of roman with cursive.

Finally, let me say that I admire the systematic way you have planned this project. This reflects on your design. I just guess you're a systematic guy!

Hoped to be of any help!
Kind regards,
Jean Paul

Igor Freiberger's picture

Jean Paul,

what an eye! You did mention exactly the glyphs more difficult to me, besides /g/. I'll post a new version soon and the issues you pointed will be addressed, although I still not know what to do with the /y/ terminal. As I'm avoiding round terminals, this /y/ needs a non-round shape.

Actually, I try to be systematic but sometimes the chaos prevails. :-)

You're right, this design tends to be rational and systematic. The glyphs themselves are not particularly beautiful as the purpose is to produce a clear and even colour. Now the challenge is to follow the planned agenda.

andrevv's picture

These letters go together really really well, great job on your consistency.

Really love your the alternate /a/.
The /e/ looks a tad bit wide (mostly the top part).
Right serif on the /f/ might need to be extended to balance it a bit.
Your /r/ is too light for my liking.
Might want to try more undershoot on the /u/.
Base of the /t/ looks too flat. Maybe curve it up a bit?
Descender on the /y/ seems shorter than the others.

Can’t wait to see some updates!

Igor Freiberger's picture

Updated version, now using FontLab and with lowercase diacritics added (almost all already available, a few ones still need revision). All the changes in glyph design were small, except in the middle of /w/. I also inverted the alternates.

Changes
/e/ botton curve matches /c/, top curve a bit tighter.
/f/ larger right serif.
/k/ ticker diagonals.
/r/ terminal and arm are thicker.
/s/ central portion a bit thinner.
/w/ higher middle, near to x-height.
/x/ right upper diagonal repositioned.
/y/ descender slight bigger, less round and using overshoot.
Marks a bit smaller.

The PDF sample is available in the first message.

Jean Paul Beumer's picture

Hi Igor,
How is FontLab working for you? Is it as easy as Illustrator? What version are you working in? I'm seriously considering buying FontLab...

Jean Paul Beumer's picture

I fail to see the difference between regular /k/ and alternate /k/... Furthermore /w/ is missing from the alphabet. I still think the lower diagonal of /k/ should be heavier, as well as the arm of /r/. Try extending the lower right serif on /r/. /w/ is much better like this. I feel as if /t/ is too narrow.

Those text samples, can you make them aligned to the left in stead of justified? It helps with looking at spacing.

To be honest, I like the alternates better than the regulars, but that's personal...

Igor Freiberger's picture

Sorry! The PDF has some mistakes. It's now updated, with the correct alternative /k/ and text aligned to the left. I'll do more tests about the questions you pointed in /r/k/t/.

Actually, I'm always in doubt between these alternates and regular glyphs. :-)

About FL, I sent you a personal message.

Tomi from Suomi's picture

Hello, Igor-

A formidable start.

But you should really check your control points after bringing your glyphs from AI to FontLab; while AI has pretty much limitless scale, FontLab works (by default) with 1000 x 1000 scale. And a shift of just one unit could look bad in small sizes.

I would suggest you learn to draw glyphs in FontLab instead of AI.

This is just my opinion, and I'm pretty anal about these things.

Anyway, a very nice start.

Igor Freiberger's picture

@ Andrew: sorry, I forgot to let a public thank you, just sent a private message. But your kind words are very important, a great input for this project!

@ Tomi: wow, to hear this from the man behind FS #1 is really great! Thank you very much.

I'm still migrating from IL to FL. Now, part of the draw proccess is made in IL, part in FL. I still feel FL lacks some resources, as a way to scale design to a given measure (in pt, not in %). But I see this flow causes additional work and may produce errors. Time to take a deeper journey in FL. :-)

Bendy's picture

Hi Igor,
This has a really nice feel to it. I love the style of the serifs and think this is going to work really well.
Here are my observations, take with plenty of salt ;)

I think the lower terminals of c and e (and perhaps t) are a little undecided.
g looks a little wide I think.
Ear of g and arm of r (and bottom of y) could be stronger, and r perhaps wider.
Some counters are slightly bumpy; I think this is just a matter of refining what you have rather than changing the shapes.
I love your acute and grave! Tilde looks a little shallow, dieresis has the dots perhaps too wide? Cedilla isn't quite there yet. Comma could be a more interesting shape: something like the quotes or acute?? I think I'd rotate the quotes slightly too.

Have you started drawing the caps and numerals? I'd really recommend doing those before doing too much more on the lowercase. I find that solutions to uppercase often help resolve difficulties in the lowercase.

gusrejc1989's picture

Wow, how did you learn and create this very magnificent, elegant and glamorous font? It ssems you're typograph. But I'm not and won't be typograph. I'll be foreign languages teacher and translator and graphic designer and system analist because I love foreign languages and computer and Internet. I come searching and testing the fonts who interest more me. I'm very fanatic by fonts. You're really very intelligent and creative! Will you public the font on font shops sites? When?

Igor Freiberger's picture

Thanks Bendy! Another great feedback.

I immediately agreed with most of your observations. This shows how important is to have a fresh eye critisizing the work. After hours doing and redoing a glyph, you becomes very used to it and tends to think it's OK.

Now I'm in a crossroads. I just began to design uppercases and agree it must be better to have them to solve problems already found in lowercases. But I'm also in doubt about the weight of the font.

While designing in IL and FL, it looked OK for the regular version (main stem=84). But the generated OT prints lighter than I want (and also is lighter in InDesign and PDF text samples). This difference may be due to hinting. As the auto tool in FL is not enough, I did manual hinting to /a/n/ glyphs with optimal results.

Thus I think the next step is to hint at least some glyphs to get real tests and then choose the general weight. After that, I'd create the UC set. What do you think?

Anyway, the next update will took a while because I'm migrating the whole project to FL.

Igor Freiberger's picture

Thanks for your kind words, Gustavo. Actually, this begun as a project just for personal use. But it's growing and may be released. Anyway, there is a large track until it becomes usable.

At a glance, type design seems to do with intelligence and creativity. But it's a highly technical work which demands a lot of hard work. Hinting and kerning are precision jobs made mostly by hand. The same about different weights.

Hope you manage your time in order to achieve all these goals. To like fonts is very useful to a translator/foreign language teacher because you can understand the OT resources and language support a font can hold.

gusrejc1989's picture

Hello Freiberger!

To like fonts is very useful to a translator/foreign language teacher because you can understand the OT resources and language support a font can hold.

I agree with you. I always need some fonts for pleasing my sight.

Would you give me your font family? (lol) I'm kidding (joking). I prefer to buy than to ask for because I know every typopgraph isn't generous and doesn't like to give or lend some fonts to anyone. I'm sorry for having said it, but I'm very reasonable.

Igor Freiberger's picture

Gustavo,

Type designers cannot be generous with their work. They make a living from that!

It's like to ask a taxi driver to give you a free trip because you love his car. Or for me to ask you a free book translation in a near future because your translations are good and I'm a literary addict.

It's not a question of be generous or not, but simply to be paid by his/her work. This is why you'll get some adverse responses when you ask a font for free in a forum where most of participants are type professionals.

Happily, there are lots of good free fonts available nowadays. Take a look on this thread: http://www.typophile.com/node/67693. Ex Ljbris fonts are especially good and are getting big success. You can also create an account at MyFonts and download good fonts for free.

Anyway, you give a good idea. I'd like to offer one style/weight of this font for free when (and if) it would be ready. But it's not a generous act, it's just a marketing plan! :-)

gusrejc1989's picture

Hello Freiberger!

Good evening!
Here it's very cold and autumn is arriving. Wow! I love cold days and winter. Winter is very comfortable with me to study for university entrance exams!

It's like to ask a taxi driver to give you a free trip because you love his car. Or for me to ask you a free book translation in a near future because your translations are good and I'm a literary addict.

If you were taxi driver and a woman asked you to go right to where she works, if you were in love with her, would you give her a free trip? (lol)
As for the free translation, 4 years ago, at my school, the couple searched for me at the school inspector room, they asked me to translate their papers into Portuguese, when I translated and finished, they asked me how much they might pay me, I was so generous and I told them — they didn't need to pay me because it was a free translation, so they were only two papers.

It's not a question of be generous or not, but simply to be paid by his/her work. This is why you'll get some adverse responses when you ask a font for free in a forum where most of participants are type professionals

I agree with you, so Tomi from Suomi was very negative with me when I asked a free download of Neo Sans, one of my most favourite fonts. I confess I don't like or I'm afraid of asking the people type professional, I mean, typographic designers because they would be egoist and negative with me.

Happily, there are lots of good free fonts available nowadays. Take a look on this thread: http://www.typophile.com/node/67693. Ex Ljbris fonts are especially good and are getting big success. You can also create an account at MyFonts and download good fonts for free.

It's true. But I don't like few right fonts that they offered at Best Free Fonts post. I prefer Genitium, Galette and Titilium. I've already created an account at MyFonts before I came here by first time. There's only 423 free and legal fonts at MyFonts and I've already downloaded some of them. Only more or less of 5 of 423 fonts are good, other less of 423 aren't good or are worse and ugly. At Fonts.com and Linotype.com, there's no free downloads, all the fonts are paid.

Bendy's picture

> Only more or less of 5 of 423 fonts are good, other less of 423 aren't good or are worse and ugly. At Fonts.com and Linotype.com, there's no free downloads, all the fonts are paid.

Of course. If someone has spent the time making a font beautiful, you should expect to pay them for their time. I'm not sure you appreciate the work involved in creating high quality fonts. It's not a matter of type designers being egotistical — fonts are their work and their value needs to be recognised.

gusrejc1989's picture

Freiberger, why didn't you answer my post?

Igor Freiberger's picture

Gustavo, I already said what I think about type designers to be generous. We don't agree on this. And it's an off-topic question.

Jennifer E. Dahl's picture

I mean, typographic designers because they would be egoist and negative with me.

Perhaps there is a language barrier here. I'd like to believe you don't actually mean what you are saying.

That said...I realize that this will cause a riot...but really?!?!?

If you are serious about design, typography and type design (!!!) you are speaking to your future peers... is this really the impression you want to make???

Wow, I've officially joined the ranks of annoyed typophilers.
That's what I get for being an addict.

Igor Freiberger's picture

NEW VERSION, NOW WITH UPPERCASE ALPHABET.

The updated PDF (first post) shows version .065 with the whole character set and also sample texts in various sizes.

Please note the font has no kerning by now, just sidebar spacing. Hinting was made with AFDK autohint tool.

I'm unsure about some glyphs, especially /B/R/K/S/.

After some improvement in lowercase and uppercase, I plan to create remanining glyphs (numbers, marks, dingbats, auxiliary characters, currencies, small caps and petite caps) until June. By August I hope to have the regular font kerned and hinted.

Jean Paul Beumer's picture

Hi Igor,
You're going great here I think! I'll take a closer in-depth look this week.

In an earlier post you said you were struggeling with /y/. Last week I might have come up with a possible solution: why not make the terminal on /y/ like you have on /j/?

JanekZ's picture

Hi Igor,
Great job! Nice and readable.
"Comments are very welcome." So I have some doubts (the nuts and bolts ;) Your acute, Polish kreska and Hungarian umlaut have the same slope. I don't think that's right (in particular Polish kreska ought to be steeper than French acute, and French acute steeper than Czech čarka). Hehe, what a mess...
In lslash 'slash' might be shorter (a notch) on the left side (your choice).
Ogoneks in UC Aogonek and Eogonek should be shifted left a bit.
Ogonek in lc eogonek should be shifted right a bit.
BTW your ogonek looks somewhat anaemic;)
There are some tips here: http://typophile.com/node/66141, especially the last is valuable.
Best wishes

Igor Freiberger's picture

@ Jean Paul: thanks again! I never thought about this /y/ possibility. It surely deserves a try. I already made a 0.66 version with better /R/S/Æ/Œ/ and other small improvements.

@ JanekZ: I was trying to use just one ogonek design, but this proved to be wrong. As you pointed, the /e/o/ connections must be different than /a/i/u/ and UC ones. Also, lc and UC positions and weights differ. Nothing better than a native-speaking to see these errors. Thanks a lot!

And I need to learn these subtle differences between acute, kreska and umlaut. If the font does not becomes good, at least I will discover so much about diacritics. :-)

JanekZ's picture

To be precise: "I was trying to use just one ogonek design" - that's OK. The only problem is in connections. Another useful thread: http://typophile.com/node/39060
BTW you should check the design of ogoneks and theirs connections in Lithuanian and Native American languages...
Athabaskan language studies http://books.google.com/books?id=G6J7Tki_WBcC&pg=PA349&lpg=PA349&dq=Clay...
English/Navajo Glossary of Legal Terms: http://www.navajocourts.org/
Samples from site: http://www.statemaster.com/encyclopedia/Ogonek#Languages_that_use_the_og...
Example in Polish:
Wół go pyta: „Panie chrząszczu,
Po co pan tak brzęczy w gąszczu?”
— Jan Brzechwa, Chrząszcz

Example in Lithuanian:
Lydėdami gęstančią žarą vėlai
Pakilo į dangų margi sakalai
— Vincas Mykolaitis-Putinas, Margi sakalai

Example in Älvdalen Dalecarlian:
"Ja, eð war įe plåg að gęslkallum, dar eð war slaik uondlostjyner i gęslun."
— Vikar Margit Andersdotter, I fäbodlivet i gamla tider.

Don't ask me how to pronounce these texts ;-)
http://en.wikipedia.org/wiki/Chrząszcz, click "The Polish original"

Best wishes

Igor Freiberger's picture

Thanks again, JanekZ. A lot of very precise information. Never thought ogonek has so many issues.

What I mean when said just one ogonek is to have just one combining glyph in the font. So, I can insert it as a component in composites as agonek, ogonek etc. This is not possible as connections are different.

Anyway, I have a remaining question: how to have a combining ogonek (Unicode 0328) to work generically if connections are variable? I already have uppercase and small caps combining diacritics, but these does not addresses the ogonek issue.

Jongseong's picture

I'll point out very quickly that on the round letters, the ogoneks are too far to the left and the cedillas are too far to the right. Perhaps you can slightly modify the shapes as well as the position to make them look more centred.

Igor Freiberger's picture

Updated version 0.73.

Main changes:
01. added numbers (lining, oldstyle, small caps and petite caps).
02. included complete small caps set.
03. added support to Vietnamese and other non-European languages.
04. improved diacritics, especially Polish and Lithuanian ones.

The PDF now has sample texts in English, Portuguese, German, Spanish, French, Dutch, Polish and Finnish.

The numbers still need a lot of adjustment –their stems aren't normalized and doesn't fits well by now. I changed the connections to ogoneks, especially in /A/. Some characters are still lacking, as Schwa or Pinyin translitaration glyphs.

I'd like to hear about the general feeling the font has. Is it consistent? Are the stems adequate when upper and lowercase are side-by-side?

I plan to add four other weights, one lighter than this and three bolder. Each one will have its own bold, italic and bold italic styles. So the family will actually holds 10 weights in five subfamilies. Many months of work to go.

Please be aware this version still has no kerning.

And a special thanks to JanekZ and Jongseong for the observations.

Frode Bo Helland's picture

This looks lovely.

• The 'a/A-ring-acute' isn’t really necessary.
• The 'a' bowl in your 'æ' looks a little wierd to me. Have you considered letting the inner counter follow the 'e' curve instead?
• I really like the looks of your lowercase 'g', but I keep thinking it might benefit from a little more air in between the eye and the tail.

Igor Freiberger's picture

Thanks, Frode. It's always amazing to hear from native speaking about the glyphs I don't know very well. Here is an /ae/ update, what do you think?

Frode Bo Helland's picture


A crude example, but I’m talking about something more like this.

Igor Freiberger's picture

Thanks again! Now I get exactly the criteria.

Igor Freiberger's picture

I did post a new PDF. Now the version is 0.85, with 10 figurine variations: lining, oldstyle, small caps, petite caps and uppercase, each one in proportional and tabular sets.

Numbers plays an essential role in this font (as already said in the first post). It's my aim to keep an even and horizontal effect. Actually, I believe numbers are distracting in most fonts. Lining figurines tends to be very tall and oldstyle tends to seem misaligned in a technical text with large number and acronym usage.

I changed the /r/ terminal and begun to include ligatures. There are also arrows, bullets and asterisks. Punctuation is complete and varies to uppercase and small caps. I also included an alternate /3/.

Petite caps are incomplete so they are not in this specimen. Still lacking: some dingbats, chess figurines, swashes and indexes (negative type inside geometric contour).

This version is still unkerned. This will be the last job, to be done in June.

Igor Freiberger's picture

An online flash demo is available at this page.
It's still somewhat crude sample, but it gives an idea.

Bendy's picture

I think the alternate g could be braver. The gap in the tail will disappear too much.
Otherwise this is looking great, I don't have any more suggestions for the moment! :)

Igor Freiberger's picture

Thank you very much, Bendy. I changed a bit the alternate /g/.
I'd appreciate any comment about the lining numbers and the alternate /g/ set:

Igor Freiberger's picture

I'd like to hear from Typophilers about these special characters (basically, Vietnamese diacritics and some European ones I don't know very well). There are also /m/+candabrindu, two ampersands and some others.

You can see a large set of this characters, including all Vietnamese ones, in a new PDF I included in first post (Especials.pdf).

The regular weight will be ready late June. I plan to release a free basic version at MyFonts in early July (this is, with just the basic Western 256 characters). What do you think?

1985's picture

Well done on your hard work, that's all I'd like to add.

jcrippen's picture

If you add U+0331 Combining Macron Below and make it work with lowercase g then I’ll be your friend for life. The Tlingit orthography requires K, k, X, x, G, and g with U+0331 and there are precious few fonts that support that diacritic in any reasonable way. I note you’ve got U+1E34 Latin Capital Letter K with Line Below and its lowercase counterpart. Those, along with the other characters “with Line Below” decompose into the base Latin forms and U+0331. U+0331 gets used in a few other Native American orthographies, like in Kwakʼwala (née Kwakiutl) and Coast Tsimshian where it occurs below A and a.

The lowercase h with circumflex and with caron could have alternate forms where the diacritic is shifted to the right and lowered so that it sits above the leg like the dot above. You could use the caron and circumflex for lowercase since it is narrower.

Do you plan on implementing IPA at some point? If you do, let me know and I can advise. As a linguist I’m a daily user and I’ve seen a lot of poor designs for IPA letters.

If you implement U+0332 Combining Low Line then make sure that adjacent pairs connect to each other. It’s supposed to be a textual equivalent of underlining. It should *not* act like U+0331 and be independent. It’s amazing how many font designers seem to have completely ignored the quite explicit instruction in the Unicode standard: “• connects on left and right”.

Hope to see more soon! Doing the all the Latin diacritics is somewhat of a thankless job, but I assure you that people like me really do appreciate good work.

Igor Freiberger's picture

1985: thank you very much.

James: your input is very valuable. I have interest to include correctly designed glyphs for many languages, but most of them are unknown for me.

Actually, I first planned to include just Western and Central European support. Then I begun to understand these diacritics from Polish, Welsh, Sami and so on. Later I begun to include more and more language support, as Latin Azerbaijani, Quíchua and African, besides extensive Maths symbols. Now I'm considering if the font will have Greek and Cyrillic. It never stops! :-)

U+0331 and U+0332 are already in the font. U+0332 is made from underline so its length must be OK. I know these combining diacritics needs to be positioned with GPOS, but this is still obscure to me.

How must the macron bellow be positioned with g? I made a try:

I saw a font with an OT feature to replace combining diacritics + glyphs with precomposed glyphs. Is this useful for linguists? Which combinations must be replaced to avoid bad diacritical positioning? Any advice on this is very welcome.

I'll surely include these alternates you suggested.

IPA is not planned, but I'm open to ideas in this area. If I add this, must I include all glyphs from the Unicode IPA table?

Actually, IPA and all these languages support makes sense just if the overall design is handsome and legible – and I still unsure if this font will become good enough to a commercial release. Although it seems a fair idea to add another option to the field, this must be a very well done option –otherwise it will be better to use other available fonts.

Thanks again. Further comments woud be very much appreciated.

jcrippen's picture

The macron should be positioned below the bottom of the lower bowl, and it should not touch the descender at all. It may have to stick out below the lower boundary of the font, but people who need diacritics below letters with descenders will need greater leading anyway so usually overlap with following lines isn’t a problem. The version you’ve shown is instead appropriate for U+10E5 Latin Small Letter G with Stroke which doesn’t involve a macron below.

A problem with U+0331 that many fonts have is that the diacritic is not the same distance from the baseline as it is in the “... With Line Below” characters. The two are supposed to be canonically equivalent according to the Unicode standard, so the diacritic should look exactly the same whether it’s part of a precomposed character or whether it’s added to some base character. Since U+0331 is a horizontal line it’s very striking when the distance from the baseline is different from one letter to the next, as in this example of Mac OS X’s current versions of Palatino and Times Roman as well as Victor Gaultney’s Gentium which gets it right. (The screenshot was taken from Mac OS X’s TextEdit.) The example also shows where the macron below should appear beneath the lower bowl of the lowercase g.

If you implement IPA you should definitely do all of the symbols in the IPA extensions block, as well as the spacing modifier letters and the combining diacritical marks. The diacritics need to be tested for positioning with every IPA symbol. Check the IPA chart to see how the basic diacritics are used, but note that any symbol can have more than one diacritic depending on the narrowness of the transcription. You don’t really have to worry about the ligating tone letters (U+02E5 ... U+02E9), people who want them will use one of the SIL fonts (Doulos SIL, Charis SIL). I recommend not doing any of the “Overlay” diacritics since they’re not supposed to be used according to official Unicode doctrine. You might just use the spots as placeholders for the overlay glyphs used in constructing other characters.

If you want to do a really good job on diacritics with Latin letters and phonetic transcriptions, I recommend looking at the three fonts from SIL, namely Gentium, Doulos SIL, and Charis SIL. Those serve as great examples of how to properly handle a wide variety of diacritics in different combinations for a very large repertoire of languages and transcription systems. The requirements are quite different from ordinary European orthographic needs, and the SIL fonts are carefully designed to fit the needs of linguists and many minority language orthographies.

jcrippen's picture

I forgot to mention that IPA also requires a couple of lowercase Greek letters. The absolute necessities are U+03B2 Greek Small Letter Beta, U+03B8 Greek Small Letter Theta, and U+03C7 Greek Small Letter Chi. Other ones that get used frequently in linguistics are uppercase and lowercase gamma, uppercase and lowercase delta, uppercase and lowercase pi, uppercase theta, uppercase and lowercase sigma, uppercase and lowercase omega, lowercase alpha, lowercase lambda (counterpart to U+019B Latin Small Letter Lambda with Stroke), and lowercase rho.

The “with Palatal Hook” letters in the Phonetic Extensions Supplement are not used anymore, but they may be necessary for quoting older linguistic work. The superscripts in that same block are definitely in use, as are the diacritics in the Combining Diacritical Marks Supplement just afterwards.

Syndicate content Syndicate content