Beautiful!

But what is that bit about the first Urdu typeface becoming available in 1997??? This cannot be true. Isn't Urdu the main printed language in Pakistan?

what a terrific article. i would love to sit in that room for those three hours and see it come together.

cut/pasted from my blog:

Part of me wants to praise their skill and dedication, while another part of me keeps paraphrasing Jan Tschichold, who said something along the lines of “What the world needs is not more handmade books, but more well-designed mass-produced books.”

But I think that there could be a bigger future for this sort of operation than most people realize. As printing costs continue to cascade downward and niche marketing becomes more popular, expensive short press run magazines are popping up like weeds in the cities of the US and Europe. Who’s to say that gorgeously hand-composed magazines could not take off in the parts of the world where stunning calligraphy is still revered? Imagine a fashion magazine hand-composed, printed on exotic papers and sold in Jeddah and Dubai. Or a fanfic/dojinshi magazine decorated with kanji and sold in Tokyo, Kyoto, and Osaka. I would love to see an American design magazine hire Marian Bantjes to treat an entire issue.

@Dunwich Type: What if I tell you that that newspaper is both handmade and mass-produced?

>“What the world needs is not more handmade books, but more well-designed mass-produced books.”

I don't think that a hand-written newspaper with a (relatively, for that country) small circulation is endangering the production of well-designed, mass-produced books! Whether those books be in Urdu, or any other language. Calligraphy and tradition DO have their place. I'm assuming here that Urdu readers in India have other newspapers to choose from…

There was a nice BBC article about this process a while back.
Can't find it though. :-(

BTW Dan, Urdu is complex enough that it did take forever to get
good typesetting (as opposed to just typing) implementation for it.

hhp

1997 probably relates to Arabic OpenType fonts? Of course the Linotype Qalmi and Monotype's Noori pre-date these as true Nastaliq style typefaces for other technologies - it's a shame that neither is available in OpenType form.

Anyhow Urdu Nastaliq is hard - the number of glyphs, and OpenType schemes needed put it up there as an expensive and complicated proposition - also balancing performance with quality is a difficult, er balancing act. The calligraphers have set the bar pretty high.

Cheers, Si

Hi Si,

In 1997 OpenType technology was still in its infancy, and certainly not capable of handling nastaliq.

The fact that OpenType has still difficulty handling it, has nothing to do with calligraphers setting the bar high. Its rather that OpenType was developed on the basis of artists' impressions rather on the basis of than analysis of Arabic scripts. In other threads I have already (well, much later than your post above, I realize) pointed out that there is no intrinsic difficulty in nastaliq as such. To the contrary, it seems to be simplified Arabic when compared with Naskh - in its pre-typographic form, I mean, not minimalist typographic developments of naskh.

Thomas Milo
DecoType
www.decotype.com

In 1997, they said that the first "widely circulated Urdu computer font" arrived.

By that, I would take it that there were metal fonts, matrices for Linotype and Monotype, and so on, for Urdu long before then. And there could even have been Adobe Type 1 fonts too - by "widely circulated", one can read 'cheap' or 'free', and that isn't so hard to believe.

Tom wrote: The fact that OpenType has still difficulty handling it, has nothing to do with calligraphers setting the bar high.

I agree. I think one can also be quite precise in just what aspect of nastaliq is most difficult for OpenType. The actual shaping of letter groups (what Tom calls fusions) is really quite easy, especially if one avoids using ligatures and instead uses a set of contextual letter shapes with GPOS cursive attachment positioning. This is not difficult, either from a design perspective nor from a layout programming perspective. The thing that is very difficult in OpenType -- is, in fact, the greatest weakness of the OpenType Layout model -- is the interaction of spacing and mark positioning. Shaping the letter groups is easy: positioning the letter groups relative to each other and then placing the dots and other marks relative to the letter groups is where the whole thing becomes very difficult indeed. The problem isn't particular to OpenType, but to any font layout technology that has inherited concepts of sidebearings and kerning from metal type. Simply put, this paradigm doesn't work for nastaliq (it doesn't work well for any Arabic style, but is particularly bad for the strongly diagonal styles).

Agree with Hudson. Nastaliq is just not "possible" with only opentype. One is forcrd to use alternative tools.

Agree with Hudson. Nastaliq is just not "possible" with only opentype.

To clarify, I don't think it is not possible, just very difficult in ways that expose a fundamental weakness of the OT GPOS model.

http://www.flickr.com/photos/68427900@N00/4255226770/

Sorry, I don't know how to make pictures on typophile, but the above link is the front entrance to the newspaper in question.

It's been 2 years since the wired article and the paper was still going strong when we visited in December. When I was there it was two sisters writing under a yellowy bulb ... the press takes up most of the room, along with the usual detritus that accompanies a working press.

The work and the people are completely fascinating.

And a few more pictures from December, including views of the printed page: http://www.flickr.com/photos/ultrasparky/tags/musalman/

Reviving this old thread :)

Last year I learnt that the famous Bulaq press in Cairo had finished their first Nastaliq typeface in 1831 (their first Naskh was finished in 1822), and by the time of Vienna International Exhibition of 1873 they had at least 3 Nastaliq typefaces. The samples I've are small and not clear enough, but the Nastaliq looks very well executed and "authentic".

Previously I thought the first Nastaliq was developed much later than this, and I was fascinated by this "discovery".

Hi Khaled,

This is a great find! All these fonts look very authentic end testify of what I call knowledge of "script grammar". To my knowledge, the nasta`liq (or rather, the Ottoman ta`liq) is created by the Armenian-Ottoman Ohannis Mühendisoğlu. The naskh looks like the typeface created by Ibrahim Müteferrika.

I downloaded Nastal'iq Navees, the open-source Nastaliq font, and I've been able to look at the glyphs in Fontforge. While I don't think my knowledge in this area is equal to porting that font so that it can be used outside of the old Macintosh GX system, I have been also hunting up information on the attempts to capture Nastaliq in metal type.

Given that Pakistan is a developing country, even a very imperfect attempt at a Nastaliq font that's free is likely to find many grateful users, and it will also help to reduce piracy of the quality products from WinSoft and Deco Type.

Oh, wait: there does appear to be one in existance other than Nastal'iq Navees, at this link which appears legitimate and not a link to anything pirated.

But that font was made in Iran, which may be a problem for some. Here is another font, for "basic" Nastal'iq (actually Naksh-i-tal'iq, I've learned in today's web searches...), whatever that may be.

While I have not found information or samples of Fort William College's Nastal'iq font... and I have not even tried to search for details on the "miserable failure" of the attempt to create a Nastal'iq typewriter... still, however inadequate a Simplified Nastal'iq on the model of what Simplified Arabic did to Naksh would be, still, it may be at least a useful starting point on which some would find it possible to build.

Of course, today, a number of excellent commercial fonts fully implementing Nastal'iq exist, so (finally!) it isn't fair to say that Urdu is not being served by the computer revolution.

Oh - and the Government of Pakistan has even seen the need, and created a font, called Pak Nastaleeq, to serve this need. It, like the other free Nastal'iq fonts listed above, is a TrueType font, not even OpenType... but also it's still in beta.

But, hey, it works good in Windows, even if it apparently has problems in Ubuntu...

Hi quadibloc,

Try setting Ubuntu's "autohinting" to "none".

There's also the "Urdu Typesetting" font in Windows 8.

Incidentally, I looked at the information about Pak Nastaleeq, and while it does include some simplifications, it still has about 200 ligatures, so it bids fair to be reasonably impressive.

EDIT: I misremembered what I had read. There are 200 glyphs in the font, but no ligatures.

There shouldn't be any ligatures. Ligatures are a dumb way to implement Arabic, and especially nastaliq.

@John Hudson:
Ligatures are a dumb way to implement Arabic, and especially nastaliq.

I hadn't been thinking of ligatures as a method of implementation of the unique features of the Nastaliq script.

But I can see where the idea could have occurred to you that this was what I was talking about.

Recently, I found this image

of the Arabic type due to Smith and Hallock in http://a book on Google Books: on page 134 of a book titled "The Fourth Exhibition of the Massachusetts Charitable Mechanic Association, at Quincy Hall, in the City of Boston, September 16, 1844".

This illustration of the type shows how successive letters that move the baseline up or down are still individual types. So words need to be built up with short pieces of leading. In the metal type era, this was a serious issue.

Hence, for naksh, if not nastal`iq, ligatures did become the preferred method of implementing the calligraphic niceties:

as I learned from The Traditional Arabic Typecase Extended to the Unicode Set of Glyphs by Yannis Haralambous.

And, yes, he may have chosen the "dumb" way to upgrade typesetting for Urdu and Farsi to what was available for Arabic, since fiddling around with leading is no longer an issue in the computer age. But I'm more inclined to praise his efforts as a start, to establish what the minimal things that need to be done are, as part of the progress towards doing them right.

Of course, by now, what with the efforts of people like Thomas Milo, the Arabic script is, or is close to being, a solved problem, I suspect: excuses to accept substandard typography on technical grounds are no longer valid, even if there is still room for progress.

> Ligatures are a dumb way to implement Arabic

Unless one implements a (much) better application of ligation. :-)

hhp

Further on this topic:

The paper by Yannis Haralambous on the Cairo or ’Almatab‘ al’amarya typecase does note the purposes that the ligatures serve, dividing them into a limited number of classes - about twenty - and so it can serve as a guide to a non-ligature implementation.

What I really would have loved to have found somewhere would have been detailed explanations of the font designed by Mühendis-oğlu and the "American Arabic" of Smith and Hallock - so that a comparison would be possible.

Because, of course, a proper analysis of the problem requires one not only to make contact with the Arab calligraphic tradition, but also to systematically survey the previous typographic efforts to determine what portions of that tradition are considered most significant.

It might be that some calligraphic niceties actually inhibit readability instead of promoting it, and so effort might be being put into the equivalent of repeating Gutenberg's mistake of preferring blackletter to Roman.

Also, while there has been discussion of "Simplified Arabic", where only two versions of each letter, rather than four, are used as being the ultimate debasement of the script, it has seemed to me that this merely displaces the join between two letters to always be on the first letter of a joined pair - which works on a typewriter, but not on metal type - and thus does not result in anything being lost.

That is, it doesn't result in anything additional being lost as against typical ordinary fonts which incorporate the drastic simplification of having only four forms for each letter, and having an unchanging baseline - that is, those which follow the lead of the "Yakout" font for Linotype machines.

Hrant: Unless one implements a (much) better application of ligation.

In explaining the implementation in the Brill types to the client, I coined the term 'ligaturing' to make it clear that what is involved is an action, not a thing. A ligature, in contrast, is a technical solution to joining behaviour involving the display of multiple characters as a single glyph. And that is what I consider dumb in the implementation of Arabic script, and increasingly for other scripts, including Latin: the only time ligatures make sense are when the representation of the multiple characters is not decomposable into linking contextual and positioned forms, e.g. in some of the Byzantine Greek combinations where the characters are represented by what amounts to a new form. In Arabic, ligatures are particularly dumb because a ligature system cannot capture the range of joining behaviours represented in the script without tens of thousands of ligature glyphs (the old Monotype nastaliq had c. 22,000 and still failed in some situations), whereas the natural efficiency of the writing system means that all those behaviours can be captured with just a couple of hundred well designed glyphs.

Tom Milo's ACE engine definitely does it best. Recent OpenType fonts that have followed his lead -- avoiding ligatures, breaking the script down into achigraphemes, shaping joining behaviour on the fly, and contextually adjusting the position of dots and marks -- include some very nice design work, and some innovative solutions within the strictures of the OpenType Layout architecture, but they still don't come close to the flexibility provided by the combination of the ACE engine, ACE fonts and the Tasmeem UI. They are, despite the beauty of their designs, hampered by a technology that wasn't designed based on a thorough analysis of the Arabic script, and which involves serious bottlenecks.

I'll be talking about one such project at ISType in Istanbul this summer: and will discuss the design achievements, as well as both what OpenType has been able to do well and where it runs into problems.

In this connection, the "American Arabic" of Smith and Hallock did implement Naksh by casting individual characters as "linking contextual and positioned forms".

For metal type, this has its own problems, making Arabic typesetting an exercise in creative leading, like typesetting mathematical formulas. Which is why the Cairo typecase, with a mere 470 characters, ended up being a more successful approach to quality Naksh typesetting.

In any case, though, the Arabic New Testament typeset in the Smith and Hallock types - it's the translation by van Dyck - is on the Internet Archive. The combined one with the entire New Testament is defective and won't open, but volumes 1, 2 and 3 as separate volumes are apparently good:

http://archive.org/details/ArabicNewTestament1867VanDyckOriginalVol1

http://archive.org/details/ArabicNewTestament1867VanDyckOriginal_vol2

http://archive.org/details/ArabicNewTestament1867VanDyckOriginalVol3

I've only downloaded volume 1, though, so I can't be sure about the other two.

Thank you very much for posting these PDFs of the van Dyck Bible. I had not seen this before, and it is very interesting. It appears to use a combination of contextual forms, arranged at varying heights, and ligatures.

I just posted the URLs. Someone else uploaded the PDFs.

BTW, there was a nicely detailed article about Hallock (including his Arabic efforts) in an issue of the APHA journal (which I have somewhere in the garage). The man was a true innovator.

hhp

Sorry, yes, I meant posting the links.

Thanks for the tip, Hrant. That would presumably be Printing History 45 (Volume 23, No. 1) 2003; J.F. Coakley, ‘Homan Hallock, Punchcutter’.

Relevant to this topic, I happened to bump into, at the local public library, a few weeks ago, a copy of the Daily Jang from Lahore, Pakistan. This newspaper was typeset, not handwritten, but that didn't stop its Urdu text from being in the Nastaliq style.