Greek in the Times.

Imagine my surprise when I got to this paragraph in the NY Times obituary “Rev. Robert Palladino, Scribe Who Shaped Apple’s Fonts, Dies at 83“:

The word “calligraphy” is born of Classical Greek κάλλος (kallos, “beautiful”) and γράφω (grapho, “write”). Though he was by all accounts too courtly to have said so, it would doubtless have pained Father Palladino — whom Mr. Jobs consulted on the design of the Mac’s Greek letters — to see the flagrant unloveliness of the only Greek font at this newspaper’s disposal.

And later on, the obit (by Margalit Fox) refers to “Father Edward Catich, an eminent calligrapher and paleographer (< Greek πάλαι, palai, “long ago,” + γράφω).” I’ve often wondered why in this age of Unicode and computer fonts, more newspapers and magazines don’t print words in other alphabets (though at least some are starting to put the proper accents on foreign words in Latin script); up with this sort of thing!


  1. I’ve often wondered this too, so I asked the editor of The Economist a while back. I received this reply:

    “The problem is cost. We print the Economist in a specially designed typeface where each letter was individually designed. At the time (I think some 20 years ago) it was decided to include only french, german and spanish diacritics as these were the main international languages. In those days computer software was in its infancy and only a handful of specialist typesetters in western countries could do the east European diacritics.

    If we add a character, each one has to be specially designed. It costs a few hundred euros a time. And we would need it in upper and lower case for normal, bold, bold italic, italic and small caps, so ten different versions for each character. That makes it a few thousand per letter. Then where do you stop? I’d like all the CEE languages, plus scandinavians and Italians. But what about Turkish? Vietnamese? The full character set has 280 (assuming you want the lot). so 2800 x (at a minimum) 150 euros….. I am not sure that’s a good use of editorial resources. Is that really worth it? Another option would be to ditch our proprietory type face and go to a standard one. But I would have a hard job persuading our board that we junk something that is elegant and distinctive.

    We did add the Portuguese characters a few years ago, but it was expensive and tiresome

    The other problem is using the characters correctly. We have enough problem getting the french accents and german umlauts used correctly (see the numerous complaints when we get it wrong).”

    I imagine many major publications use custom-designed fonts so they’d be faced with the same problem.

  2. Interesting, I had no idea. Thanks!

  3. If they are paying €150 per glyph, they are being gouged relentlessly, I suspect. I’ll ask a few fontographers.

  4. We did add the Portuguese characters a few years ago, but it was expensive and tiresome…

    The only Portuguese characters not shared with French or Spanish are ã, õ. If they found adding them expensive and tiresome, they must be easy to discourage.

    The other problem is using the characters correctly. We have enough problem getting the french accents and german umlauts used correctly…

    For God’s sake, are they the editors of a major newspaper or junior school kids?

  5. David Marjanović says

    I’m surprised that a one-time expense of a few thousand € is considered prohibitive by a big international newspaper.

  6. All major newspapers that I know about are facing huge financial struggles — that certainly includes the Washington Post and the New York Times. Spending thousands of dollars on something not directly related to newsgathering is going to be a hard sell.

    For God’s sake, are they the editors of a major newspaper or junior school kids?

    The Washington Post has laid off a significant number of copy editors in recent years, and the effects show. More typos, misspelled words, misused words… Newspapers are put together at great speed and under strict deadlines. It’s a struggle to keep the English straight. Asking the same copy editors to deal with funny letters and unfamiliar accents in a variety of foreign words is only going to create more problems.

  7. This is seriously tricky business. It’s not realistic to expect a reporter, much less a pressed-for-time copy editor, to understand and apply the vagaries of non-English diacrtics. I’ve had many run-ins with Israeli graphic designers over the proper use of hyphens, n-dashes and m-dashes — mostly because the commonplace (American) English approach is simply not used in most Hebrew publications.

    Bluebook? Impossible in an alphabet lacking upper and lower cases and therefore small-caps too.

  8. This sounds ridiculous to me.
    Type design can be very lucrative; but there are also compatibly sized typefaces in Greek, Cyrillic, &c. which can be used harmoniously with different (even the most “modern” and “customised”) proprietary typefaces. [I’m avoiding the popular word “font” for technical reasons. Or peevish ones.]
    Sadly, the people I know who design proprietary type families are not linguistically oriented. I’ll call this to their attention anyway.
    Oh, I am so sorry to hear Bob Palladino is dead. Who’s left, I wonder, really.

  9. Maybe I’m missing something– but the Economist editor’s claim that they can’t use non-Latin alphabets because the fonts would be too expensive doesn’t make much sense to me. After all, you’re talking about non-Latin fonts, so why would they necessarily conflict with the Economist’s Latin fonts?

    Or… is he saying they don’t use Unicode for their text encoding? If that’s the case, they have much bigger problems than choosing a set of fonts for non-Latin alphabets.

  10. & it looks fine to me in the online version. Difficult to believe there’ll be some wild discord in the printed copy.

  11. The Economist editor isn’t saying that they can’t use non-latin alphabets, he’s saying that they can’t add more diacritics to their latin typeface, which currently only includes those needed for German, French, Spanish, and Portuguese. It’s a shame that type design was parochial in the past; I don’t necessarily see that improving. Anyone know of any good counterexamples?

  12. I wish people didn’t take the have-a-go-Granddad approach to Greek – it is really not that hard to find a friendly classicist. κἀλλος means ‘beauty’, καλός means ‘beautiful’, the NYT is constantly badgering me to subscribe, perhaps they could simply offer me a free subscription in return for quick answers when they want to use Greek in an article.

  13. type design was parochial

    Fine typography is inherently parochial, because you can’t use the same ó for Spanish and Polish, or at least you shouldn’t (the Polish accent is steeper and shorter). Interlingual fonts are necessarily compromises. But that said, there are some nice fonts with large coverage, like Everson Mono and SIL Gentium.

  14. It would be correct, though, to say that καλλι- means ‘beautiful’ as the first member of compounds, corresponding to the free-standing adjective κάλλος.

  15. because you can’t use the same ó for Spanish and Polish, or at least you shouldn’t (the Polish accent is steeper and shorter).

    That one’s new to me. (I do know about Germans positioning diereses differently over capitals.) But as far as I know, the universal practice online is to gloss over differences of this sort.

    On a personal level, I’ve tried and failed to find a good serif typeface for interlingual use (including at least extended Latin, IPA, Greek and Cyrillic, and Hebrew and Arabic if possible). Gentium Plus is nice, but despite a five-year-old promise it still lacks a bold face. Charis is nice and has a bold face, but despite its name it lacks Greek. Google’s Noto Serif would fit the bill, but it looks a little underbaked in places – and in particular, its treatment of IPA stress and length marks is atrocious. It seems like I’m being forced to accept Times New Roman, a typeface that no esthete would be caught dead with.

  16. Speaking of interlinguality, I just stumbled across the Common Turkic Alphabet; seems like a bit of a shame that that didn’t catch on (probably lots to be said about Soviet policies there). There were two versions of it, one based on the Latin alphabet and one on Cyrillic. The Latin one is a superset of the Turkish alphabet, so to adopt it you wouldn’t have to change anything about the one Turkic language that had then, and still has, by far the greatest existing volume of written materials.

  17. marie-lucie says

    Speaking of fonts: I asked a fellow linguist to send me an article, which would include a fair amount of phonetic characters. When I read it on the screen, everything looked fine, but I wanted a printed copy, and once on paper, all the examples containing different characters (including a story with 3 interlinear equivalents) looked like they were covered with rolls of barbed wire! on the edges it is possible to see that each character is displayed as a much larger (4 or 5 cm tall) outline (I mean not filled with ink), but next to each other in a word or line the characters overlap in such density that it is impossible to see through them (they also cover the lines written in ordinary characters). Of course I could recover the hidden lines from the screen display, but I want a paper version that I can write on without having to copy all the lines by hand. Any advice? At the moment I don’t have phonetic fonts on my (still new) computer, or software that helps you organize lines for analysis.

  18. marie-lucie says

    p.s. I tried to contact that colleague and another one too, but have not had any answers from them.

  19. David Marjanović says

    It seems like I’m being forced to accept Times New Roman, a typeface that no esthete would be caught dead with.

    Why? Is it too mainstream? 🙂 The one complaint I have about Times New Roman is that it puts the dot on the i very high, but all other dots (äöüż) extremely low. They should all be at the same height (and size).

    The Latin one is a superset of the Turkish alphabet

    …Oh, you mean the new one, not the one from early Soviet times.

  20. Oh, right, the chart I was looking at wasn’t for the old one. Okay. But I think it would be hard to persuade small languages to change orthography yet again nowadays, right? After the messes they went through under Soviet language policy.

  21. Adding a character or several to a font is not that difficult but time consuming. The new letters have to be added to the ‘classes’, the kern has to be checked that it does not clash with other letters, example: the diacritic ‘grave’ may clash with the capital T before, All this has to be carefully checked. Then the font has to be remade and tested in all applications. The font comes in 3 or 4 variations. That is where the costs accrue.

  22. David Marjanović says

    Those that have changed their spelling systems – by no means all small – have largely done so on the order of dictators, in particular Islam Karimov and Türkmenbaşy. I don’t know how popular the results are.

  23. “Why? Is it too mainstream? :-)”

    Yeah, I think that’s part of it!

    Times New Roman has become sort of standard typeface, I don’t know why. Is it because the words ‘Times’ and ‘Roman’ connote respectability, trustworthiness? Is it perceived as being “classic”?

    TNR was created for The Times in the early 20th century sometime, the 30s I think. Essentially it’s an old-fashioned newspaper typeface. It probably served The Times pretty well back in the day (for some technical reason, no doubt) but why on earth use this not-all-that-well-designed typeface today? It shows its age and not in a good way, in my opinion.

  24. Marie-Lucie: You could try saving the file as .txt and print that out at, say 1.5 line spacing for clarity and room to amend it. I think the phonetic characters will come out as little vertical rectangles (meaning character unknown). Then copy the phonetic characters from the screen by hand to your printed copy, over-writing the rectangles This may be one way to solve the problem, if time-consuming, inelegant and boring.

  25. Hooray! Thanks, David!
    I, however am a “rebel” & happy to use Times.
    It’s just a modern shibboleth, I think. Gerrit Noordzij sneered at Stanley Morison so “everyone” decided he is passé.
    I wouldn’t use Comic Sans, though. Even if it had a whole IPA set.

  26. some technical reason

    A very straightforward one. Newspaper fonts have to be condensed (with narrow letters) and yet legible at small point sizes, to cram all the text into what is fundamentally far too little paper for it. Consequently, the x-height (a vertical length related to a font, traditionally the height of the letter x) was raised, which accounts for the high i-dot and j-dot that David complained of. Vertical strokes are wider than normal for readability, whereas their intersections are thinner than normal for efficiency.

    The result was a font that should have worked well for newspapers, but didn’t catch on widely, because it required more ink-per-page and higher-quality paper than most newspapers were willing to spend on “tomorrow’s fish-wrap”. Instead it caught on for magazines and then for books, much to the surprise of the Times itself and the font’s designer, Stanley Morison, who mocked his own work thus: “[…] by the vice of Mam­mon and the mis­ery of the machine, it is big­oted and nar­row, mean and puri­tan.” Precisely because it was so popular for a time, it was one of the first fonts created for each successive new typesetting technology.

    As such, it has now become the default serif font, the one you use if you have no reason to use anything else and no interest in doing so, the aesthetic void. In addition, its italic is crappy, a bog-standard Monotype italic with no real relationship to the roman, which is perhaps why the font is called “Times Roman” and not just “Times”, though the italic is “Times Italic”, not “Times Roman Italic”. The “New” in the name reflects the fact that after the Times’s one-year exclusive license ended, Linotype and Monotype (the Microsoft and Apple of their day) gave it slightly different names, “Times Roman” and “Times New Roman” respectively, and those names were licensed by Apple and Microsoft respectively.

  27. pratchett2 says

    It’s worth noting here (and probably no coincidence) that Margalit Fox has also written the excellent *The Riddle of the Labyrinth*, about Alice Kober and the decoding of Linear B (the book has some asides about font choices in representing classical languages).

  28. There are notable differences between Times and Times New Roman. Check out the miniscule e, for example.

  29. Helen, I was irritated by the κάλλος / καλός mistake too. But you don’t even need a friendly classicist. For $4.99 (if you have an iPhone) you can get the marvellous Ancient Greek app by Paul Hudson, with the big Liddell and Scott lexicon and lots of texts in Greek, many with English translations. Highly recommended.

  30. vrai.cabecou says

    Graham Asher: And that comment illustrates why newspapers don’t want to deal with non-Romance languages. When in doubt, leave it out.

  31. David Marjanović says

    Times New Roman has become sort of standard typeface, I don’t know why.

    Because it was the default font in Microsoft Word for Windows for many years. That’s why.

    See also: Arial.

    (The Mac versions used Times and Helvetica instead.)

    TNR was created for The Times in the early 20th century sometime, the 30s I think. Essentially it’s an old-fashioned newspaper typeface.

    Surprises me; the fact that it keeps the figures in the line instead of letting them hang below (0 not like o, 3 not like mirrored ع…) looks rather newfangled to me.

    There are notable differences between Times and Times New Roman.

    Oh yes.

  32. I always had the impression the Microsoft Word defaulted to Arial simply because it was the first font in alphabetical order among the Windows standard fonts at the time. [Of course, this couldn’t have been the only factor. If the first font in order had been something hideously hard to read (unlike Arial, which is merely annoying when read over long stretches), they doubtless would have changed it. And having the default as a sans serif font was probably also influenced by the use of Helvetica as the system font on Macs.]

  33. marie-lucie says

    Thanks for the tip, Paul!

  34. David Marjanović says

    MS Word defaulted to Times New Roman 10 pt (which is rather small) all the way up to the recent change to Calibri. Powerpoint also defaulted to Times New Roman (which is about the worst possible choice for projection: the thin strokes just disappear) till maybe 2007, when it changed to Arial (a good choice); now it defaults to Calibri, too.

  35. marie-lucie says

    David: Thanks for this info. I read something in the last few days that made me think Colibri was something old, so I was surprised today when I opened a document I had written some time ago in TNR and it said Colibri. I changed it back to TNR and it looked all right except that one page (out of about 10) was all mixed up (lines overlapping, chunks of text moved around, etc). What happened? What should I do? Does this have to do with the problem I wrote about earlier?

  36. David Marjanović says

    I have no idea.

  37. Arial was the default font in Word for Windows from at least version 1.1 (the Windows 3.0 version), probably through to the Windows 95 or Windows 97 era.

  38. Please allow me to recommend The Elements of Typographic Style by Robert Bringhurst.
    I think any of you who bother to have a look will enjoy it.

  39. I believe it’s been recommended here before, and the author himself showed up (rather irritated) in this thread.

  40. marie-lucie says

    If I remember correctly, the thread was not at all about typography.

  41. True, but it’s memorable in terms of Bringhurst. He was referred to as “doyen of sensitive and sane typographers” by Noetica here (and where has Noetica gotten to?) and as a Canadian “who could both provide the content of a book and design it at a high level” in this post.

  42. marie-lucie says

    Yes, he could really “make a book” from beginning to end.

  43. I am only astounded that I hadn’t remembered that whole thread. Maybe I was offline for some reason. Very enlightening. It’s a good book.

  44. David Marjanović says

    Arial was the default font in Word for Windows from at least version 1.1 (the Windows 3.0 version), probably through to the Windows 95 or Windows 97 era.

    The default font in Word 6.0 for Windows (e.g. 3.1) was Times New Roman (10 pt). Arial was the default sans-serif font of sorts, but not the default font.

  45. @David Marjanović: What country was that in? I am absolutely certain that the American Word for Windows that I used and installed on so many computers defaulted to Arial. It wouldn’t even make sense to talk about it having a default sans serif font, since the fonts were just chosen from a single long list.

  46. David Marjanović says

    Austria, though I really don’t think this made a difference. Whenever I made a new file in Word 6.0 and several later versions, or even in PowerPoint, it was in Times New Roman by default.

    It wouldn’t even make sense to talk about it having a default sans serif font, since the fonts were just chosen from a single long list.

    True; but Helvetica wasn’t in the list, except in the Mac version (which I later saw in France, and which defaulted to Times rather than Times New Roman, IIRC).

  47. The current version of the Economist typeface was designed in 2001 by Erik Spiekermann and Ole Schäfer. That’s ages ago in terms of digital font design, and the tools and standards would be far different today. It has become much easier to design letters with diacritics, which once had to be created individually. So the pay-for-each-glyph model would have made more sense back then. Today, generating the necessary glyphs when you already have the base letters and diacritics and just have to combine them is much easier and painless than it was 15 years ago. A different pricing model might be more realistic, like paying for each additional diacritic (ring, caron, ogonek) with a smaller price increase depending on the number of new glyphs.

    As for the cost issue, I’m pretty sure that the Economist typeface is for exclusive use, meaning that there is no extra income from selling directly to users or licensing to other parties (such as for inclusion in software). Furthermore, there is no incentive for the designers to increase the glyph coverage to attract new customers. Of course they are going to charge more for it. If we forget diacritics for the moment and consider that adding new glyphs might mean designing from scratch (e.g. thorn, eth, new currency symbols), then consider that €150 is less than what a freelance designer might charge for a day’s work. I’m with David Marjanović on this one, that a big international newspaper shouldn’t be worrying about a few thousand euros to update its bespoke typeface.

  48. Well, if their copy editors don’t know when to use them anyhow, I suppose it’s pointless to have them.

  49. When I come across an article from The Independent, I always marvel at their upper-case u with the lower right serif. I mean, I wouldn’t say it’s ugly, but just by virtue of its uniqueness I think it draws attention to itself in a way that may not be intended. But who knows, maybe it’s a point of identity for them – like The New Yörker with their diereses.

  50. Looks a bit Armenian, now that you draw my attention to it.

  51. Armenian always looks to me like the sort of Latin script people would construct who don’t actually know any, having just a vague idea of what it looks like.

  52. Looks a bit Armenian
    … or a bit like 以.

  53. January First-of-May says

    @John Cowan – isn’t that what happened with Cherokee?

  54. Sequoyah did have Latin letters to look at, though he did not know what they meant. But the specific shapes of the Cherokee script are the responsibility of Elias Boudinot (aka Buck Watie, the brother of the famous Confederate general Stand Watie), who was bi-literate. It’s quite likely that in establishing the printed forms of Cherokee letters, he made heavy use of the sorts he already had as a newspaper printer in approximating Sequoyah’s handwritten glyphs.

  55. Multilingual fonts remain a constant source of frustration for me. I’ve just discovered that if I put a Hebrew text in Noto Sans Hebrew, the letters themselves will use the correct Noto Sans glyphs, but everything else – numbers, punctuation, etc. – will use glyphs taken from Lucida Grande, the default system font. And if I put the same text in basic Noto Sans, the numbers and punctuation will use the correct Noto Sans glyphs, but the letters will use glyphs taken from Lucida Grande.

    How is this even a thing?

  56. Depends on the operating system and to some extent the application (some applications have their own font renderers).


  1. […] Hat starts a discussion about the cost of designing […]

Speak Your Mind