Adobe Devanagari Font

[Member Joh…]

hey guys.

soo many discussions thanks for sharing this information...

thanks guys...

[Member Uli…]

Mr. Hudson:

> I will write to James Agenbroad, whom I know from Unicode circles

I would be pleased, if you could pass along my best wishes to Mr. Agenbroad.

I looked up my old files and saw that ten years ago, he send me by airmail his "Difficult Characters" list, which he published in 1991 in the Bulletin 38 of the International Association of Orientalist Librarians.

I hope, Mr. Agenbroad is in good health and that he is enjoying his retirement.

Tell him that my comprehensive Sanskrit handbook is now in its fifth edition.

[Member hra…]

Doesn't Fiona have an advanced degree in Sanskrit? I think that even predates her work in type.

I'll alert her to this thread (although I suspect John already has).

hhp

[Member Uli…]

Mr. Hudson:

Having stated above on 4 Sep 2012:

"Charles Wilkins (1808), Alix Desgranges (1845), M.R. Kale (1894), Richard Fick (1922), A.A. MacDonell (1926), H.M. Lambert (1953), Michael Coulson (1973) and Madhav M. Deshpande (1997) invented the conjunct consonant "bn". If "bn" were no invention, then a Sanskrit word would exist containing "bn". But there does not exist any such Sanskrit word. Therefore "bn" has been invented. Yet "bn" could occur in a foreign-language word, e.g. in "abnormal", inserted into Neo-Sanskrit texts. But serious scientific research on Sanskrit conjunct consonants must dismiss such "abnormalities"."

it seems to me that you and Mrs. Fiona are trying to find an attestation for the ligature "bn" in Sanskrit in order to prove that I am wrong.

I should like to mention that you are not the first ones who tried to do that.

However, it seems to me that Mrs. Fiona did not learn Sanskrit, otherwise she would not have said that

अब्नाभः ="abnābhaḥ"

was given in the Monier-Williams dictionary. The usage of this dictionary is tricky and error-prone for non-Sanskritists. On page 60, middle column, the main entry starts with "ab-ja" which is a compound in itself composed of "ab" (from "ap") + "ja". A few lines below, you see "-nābha" (and also "-netra").

The compound here must be build as "abja-nābha", and not as "ab-nābha", as was thought by Mrs. Fiona. But the correct compound "abja-nābha" does not contain the ligature "bn".

Theoretically or "exempli gratia" or "just for fun", it would be possible to construct a Sanskrit compound word containing the ligature "bn", for example the compound

अब्निर्वाण (= abnirvāṇa = "the nirvana in the water")

invented by me here exempli gratia. But you will never find any attestation for this compound.

I do not deny that it is theoretically possible to find the combination "bn" in any old Sanskrit text, but none was found so far.

The statistics mathematician and font lover Luc Devroye would say that from the probability point of view, the mathematical likelihood that a Sanskritist will ever need the ligature "bn" in a font for typesetting a real Sanskrit word is converging to "zero", "nil" or "nought".

For example, by analyzing Panini's grammar, I found exactly 114 additional artificial ligatures pertaining to Panini's acronym-like grammatical elements. But I included none of these artificial acronym ligatures into my official ligature list here

http://www.sanskritweb.net/itrans/itmanual2003.pdf (page 29 seq.)

because none of these artificial acronym-like ligatures could be found in any ordinary Sanskrit text.

Additional note for scholars: "abja-nābha" occurs in the Bhagavata-Purana repeatedly. Hence it is an attested compound word.

[Member Joh…]

Uli, I am certainly not setting out to prove that you are wrong. And neither I nor Fiona singled out the bn conjunct ligature for attention. I am whittling away at a list of dubious conjuncts and this one happened to be unique in that we found it used in Lambert's book and apparently in Monier-Williams, and I posted a message about it here not to prove you wrong but because I wanted your opinion on it. Why? Because I value your opinion on matters Sanskrit, and you can usually explain oddities in a reasonable and convincing manner. In other words, I asked because I consider you likely to be right, not wrong.

[Member Joh…]

Hrant: Doesn't Fiona have an advanced degree in Sanskrit?

A postgraduate diploma in Sanskrit and a PhD in Indian palaeography, both from SOAS.

One of the things I have noticed is that grammars of the writing of Sanskrit, from Panini onwards, have a tendency to fill out, to variable degrees, the possibilities of the writing system independent of its application to the actual language. Hence, as Uli notes, one can find many listings of conjuncts that are not attested in actual texts. Some of these 'systematic inclusions' may end up being used in foreign loanwords, or in the orthographies of other languages adopting the script, but will never occur in Sanskrit. This is why Uli's work analysing frequency in proofread Sanskrit texts is so useful.

[Member ken…]

One of the things I have noticed is that grammars of the writing of Sanskrit, from Panini onwards, have a tendency to fill out, to variable degrees, the possibilities of the writing system independent of its application to the actual language.

In this way, the high degree of order and rationality of the Sanskrit grammar and sandhi strikes me as akin to the Periodic Table of the elements in that it seems to hypothesize and predict specific possibilities that do not however exist naturally — only in the laboratory environment and for short durations. ;-)

[Member Rai…]

I have stepped over this discussion on Adobe Devanagari only recently and would like to add a few comments. But first let me say, while I admit that there is place for improvement, Adobe Devanagari was exactly what I needed and came right in time for a bilingual edition of Hindi short stories accompanied by German translations I am working at.

(1) Most of the joint letters (ligatures) Uli took offense at are in actual use in Hindi, Marathi and Nepali. Both in novels and newspapers it is quite common to transliterate words and even complete sentences from English into Devanagari. It took me only a few seconds to think for possible candidates:
क्स्प्र kspr: express (extremely common, there are hundreds of trains with that name, no "ink-spray" necessary)
क्स्प्ल kspl: explain
also things like न्फ़्र nfr: conference (which happily have been included)
Please, Adobe, don’t remove these ligatures whatever Sanskritists may think about them, there is much practical publishing experience behind most of the list.
ब्न is common in Nepali (e.g. डुब्नु drown)

(2) Adobe Devanagari goes a long way even in Sanskrit and Vedic. Modern editions don’t have all the traditional ligatures used in manuscripts as they tend to turn out clumsy in print.
It would be fine some time to have Samavedic accents too (so far only Rigvedic accents are available).
Accent combinations for transliteration of Vedic (ā́ ā̀ etc) are also missing.
But then, even Rome wasn’t built in one day.

(3) Language codes and stylistic sets should be enabled for much more alternatives than they are now. InDesign CS6 has a language tag for Hindi, but not for Sanskrit (on the internet, you can find a clumsy way to get it, but I did not try).

(4) For the time being, while the font claims to be primarily made for modern Hindi, the ligatures you get by default (letter + halant + letter) are quite often the traditional rather than the modern forms, the latter preferring half letters. To get theses, you have to type an additional non width joinder. Compare:
क + क : क्क (old style) क्‍क (modern) visible only in Adobe Devanagari
न + न : न्न (old style) न्‍न (modern) visible only in Adobe Devanagari
श + व : श्व (old style) श्‍व (modern)
It would be a good idea to handle this issue by different stylistic sets and/or language tags.
I would appreciate to have Sanskrit, Hindi (traditional ligatures), Hindi (modern ligatures), Marathi & Nepali (for the latter two languages, automatical creation of e.g. र्‍य ry and the like where needed would be very helpful)

I don’t think that John’s commentary on this point is correct. The modern half forms are really ‘modern’ (in Hindi!), not just a compromise with dated technology: they are used in contemporary print because they are more legible than the traditional manuscript ligatures and therefore even recommended for e.g. text books. The comparison with the old Linotype-f is misleading; rather one should consider the whole paraphernalia of Greek ligatures used in Byzantine manuscripts which hardly any font for classical Greek does include as they are no longer used in print, again for the simple reason of legibility (Claude Garamond did use them, and there are two or three fonts around that have a lot of them – but in a stylistic set you have to turn on if you want to have them!)

(5) The short i-matra extends by default over the complete consonant combination it precedes:
कि but क्ष्मि (only visible in Adobe Devanagari). This seems sometimes a bit to much, so it would be fine if there were an option to turn the feature off or reduce the it to a more moderate size – say by an option like swash letters or titling caps (But then, Adobe can be adamant, as with the Th-ligature in all of Slimbach’s fonts).

ps.: I hope the Devanagari samples are readable, as I don’t have the time to prepare and include graphics. If not, copy the text and paste it into a Unicode-savvy Word processor and formate the whole section with Adobe Devanagari, Devanagari MT, Kokila or any other new Devanagari font.

[Member gas…]

Since I've missed a lot of fun, here I go.

@Michel Boyer
For a Hindi corpora http://dsal.uchicago.edu/dictionaries/platts/ might do, but that's not pure Hindi. But for a pre-modern font as per John Hudson it might do. Great you're "familiar" with the LaTeX font package, because nobody else seem to be in the discussion deep enough.
I must say I'm impressed with your .pdfs.
A small specimen I put together:
Adobe Devanagari for a Sanskrit Dhatupatha.
I compiled a list of Sanskrit 1379 ligatures based on 15 Sanskrit grammars after reading https://typography.guru/forums/topic/105460-forwarding. Graphical variants which are introduced by the authors quoted are treated as separate ligatures, up to 79. As many as 991 are unique to only 1 source (a minimum). 8 are met in 14 out of 15 sources (a maximum). The list is even bigger than Mr. Boyer's Samyoga Table. But it's not about size. Because errors and ligatures go together a long way.
1 991
2 92
3 41
4 23
5 16
6 17
7 25
8 21
9 27
10 38
11 35
12 31
13 14
14 8

@Uli
We owe much to Ulrich. His work is known and used wider even than Ernst Tremel's. I would love to see a version of his "entirely fictitious ligature" list for Sanskrit. Nothing remained the same
Rigveda edition by Prof. R. L. Kashyap and Prof. S. Sadagopan published in printed book form in 1998
after he started his research on devanagari typography, of which a culmination was his replica of Bombay's Nirnayasagar's bold type called Sanskrit 2003. One would love to see a "Sanskrit 2013" (ten years after) with OpenType variants included for ligatures and vowels.

Even though I have personal reasons for not loving Ulrich Stiehl, I must admit that he knows more about the technical part of a Sanskrit font than any German knows. I could say that more than any European, but there is Mihas Bayaryn, so you can't be #1 until Mihas is alive. And seems he is here to stay. A pitty all http://www.sanskritweb.net/temporary/ URLs have died.

These rarest ligatures are only covered by our own highly specialized Itranslator fonts "Sanskrit2003.tt", "Chandas.ttf" and "Siddhanda.ttf" downloadable here
http://www.sanskritweb.net/itrans/

Siddhanda.ttf is and I hope never will downloadable at Ulrich's website. Chandas.ttf indeed is,
but as per Mihas (author of the font), he treats Chandas as a draft for Siddhanta.

by analyzing Panini's grammar, I found exactly 114 additional artificial ligatures pertaining to Panini's acronym-like grammatical elements
As said it would be interesting to see the context and the ligatures, so one could check how they where printed in the 1887 edition printed at Drugulin. I have a list of Drugulin's devanagari from an archive with 425 elements, including ligatures (one would have to add - only). Still no better Sanskrit typography was ever printed that the one of Drugulin.

Same as Ulrich I deal with Sanskrit on a daily basis. And as Indians do not make Sanskrit fonts, Germans and Russians have to do so. It might sound strange, but even the original Nirnaya is based on a Berlin-made font from Uengern brothers. Schlegel was a German if I'm not wrong. Frutiger was an Indian only a bit - so there is not much of Indian in the history of Indian fonts. Indians are above fonts. They care about the text, to preserve the manuscripts. Errors? Typoes? Who cares. Virama-ligatures in a Sanskrit conference thesis? So be it. "We" are above it. Oh, really?!

After we had years to fight with "Mangal", which was a makeshift font including only the "commonest conjuncts" (using Ulrich's expression), I understand why Ulrich does not feels comfortable with the Adobe Devanagari (John Hudson's) policy. I understand that Adobe's policy is above Mr. Hudson's. But not having Marathi or Nepali in the next edition would not have any excuses, I guess. Sanskrit is Huge. Marathi is just a feature compared to the complexity of Sanskrit ligatures.

@John Hudson
Would it ever be possible to make at least a few screencasts of your work with VOLT showing the font process working with devanagari? There is next to nothing documented and a video would be something appreciated indeed.
Can you explain the basis of this subset? Is it based on frequency, or on a particular set of texts?
When Ulrich asked you, he meant, that nothing of it is used. It's strange indeed. If you take something for granted (a list not of your own), you have to have samples in what words do they occur. For example, one can not make a universal Sanskrit font, every font will work only with a limited (small or huge) amount of texts. At GRETIL, with which most of the Sanskritists are working nowadays, there are around 600 texts.
But Ulrich's list is not a list of ligatures in all GRETIL's 600 texts (as he states in different wording himself). He has dropped out the grammatical literature with it's anubandhas. So you can not print an Indian grammar or might experience issues - because grammar is not a text he would want to include. Mihas' font Siddhanta is the first one to include the grammarians only ligatures as well, because Mihas himself is rooted deep in the Paninian system. I'm working on a Dhatupatha Concordance and must say, that from the technological point of view no devanagari font comes even close. I'm not a fan of it's thin as hair stylistics, I'm used to the bold types.

Coulson, Michael. Teach yourself Sanskrit.
Monier-Williams. A practical grammar of the Sanskrit Language.
Vasu, S.C. Aṣṭādhyāyī of Pāṇini.
Agenbroad, J.E. 'Difficult characters: a collection of Devanagari conjunct consonants' (in Bulletin 38 of the International Association of Orientalist Librarians).
Coulson's list a very small one. It has 1 variant for every ligature, which means it is a poorly made list. Why on earth one would want to take it in consideration at all? Monier-Williams has around 170 000 Sanskrit words (even if put together with other Sanskrit dictionaries it would be around 200 000 so a relatively small number). Vasu's Aṣṭādhyāyī is the most authoritative edition of Pāṇini. But are the anubandhas from Dhatupatha included? Never seen Agenbroad's list, so can't comment on it. Is there a scan of it available somewhere, gentlemen? Mr. Hudson, when you wrote "get a copy of the journal with his original collection" - did you have any luck with it?

@quadibloc
It does not matter that "OCR software is oriented around the Latin alphabet". The Cologne project had Indians working (not Frenchmen) on the proofreading but there are thousands of errors left. I've been submitting my five cents every few months, but it would take around 50 years to fix all of them at such a rate. And there are several editions of the OCR texts from Cologne - not always newer is better, new errors arive as well (because of converting to the new markup as well). Anyway - nobody else has done even 1/10 of they've done so you can't say a single bad word about them. One heels only gratitude and can't say enough "thank you"
After all those OCRs where done http://www.indsenz.com/int/index.php was released. It was unreliable in the beginning. But as of 2013 it has a very small error rate and is the best OCR software for devanagari. ABBYY Fine Reader (a trained one) does not comes even close. No other (inlcuding Indian coders) comes close. I've explored the OCR market, I know what I speak about.

@Rainer
(1) agreed.
(2) Samavedic accents from Adobe - you made my day. They can't make a feature for a single text if they have no features for hundreds of other (more popular) texts. That would not make much logic. Rome wasn’t built in one day - but the building of a devanagari at Adobe is only in the plans, as regarding Sanskrit. I must say I'm impressed by the look and feel of the font except the details as screenshoted http://samskrtam.ru/adobe-devanagari-font/
(3) "language tag for Hindi, but not for Sanskrit" - indeed.
(4) "font claims to be primarily made for modern Hindi, the ligatures you get by default (letter + halant + letter) are quite often the traditional rather than the modern forms" - so if it's Hindi, be it Hindi by default. Kill every Sanskritism and we'll accept that we should not even try to print a Sanskrit text. If it's in between - that's worse.
(5) As per short i-matra that "extends by default over the complete consonant combination it precedes" - I fully agree. There must be a way to turn it off. Sometimes it can become too long. Adobe's Devanagari is not the first one to have it, but when it's too long is looks ugly in every print, be it old or new.
(6) As per "modern half forms are really ‘modern’ (in Hindi!), not just a compromise with dated technology" - that's an interesting and new insight.

If one only could add screenshots and include attachments I could show the Ligature Concordance here.