One reason I like the font David (not the only one) is its slant. Here is with Maxim Iorsh's David CLM:
Hmm. Did not even see that "Évariste Galois" was inverted.
Hrant, I think I misunderstood....
Whoops. Looks like I forgot to close an italics tag in my last message. Sorry about that. Perhaps one of the administrators can go in and fix it?
Speaking of italics, :)
Hebrew doesn't have them. But people want and would use them.
It is not true that "italics that are leaning right is acceptable". Left leaning italics are actually preferable.
John, you can't derive any logical preferences for Hebrew from Tzvika Rosenberg of MasterFont's actions. He does what ever is easier. If the default for slant is -12, he'll make -12 right leaning italics.
Tzika's views on Hebrew typography reject tradition (an important factor in typography but not in rocket science), and only come from his "pupick", as they say. For example, he told me that a chirik, or shva, under a vov, daled, or reish, should be off centered, so the reader does not confuse them with a final nuhn. This is logical, but it opposes centuries old tradition.
Since few italics has been made to lean right, so then left leaning italics are only tolerated.
Why should italics incline to any specific direction?
Italics are meant to follow the direction and movement of the eye.
Here are three examples of a true italic for David-Regular, and two versions of Diamond-Script.
David, I don't think you misunderstood Hrant's question. But perhaps for clarity we should talk in terms of left-slant and right-slant rather than forward- and back-slant, since the latter terms are relative and the former are absolute. For further clarity, left-slant is defined as having the top of a stroke being to the left of the bottom \, and right-slant as having the top to the right of the bottom /.
So Hrant was asking about right-slanted letters in Hebrew, and wondering whether Hebrew readers mind this.
As I understand it, there remains some debate about the 'correct' direction of slant for Hebrew 'italics'. I seem to recall that Zvi Narkiss decided that the letters should slant to the left. I see that Masterfont offer a number of families with options, i.e. secondary fonts with left- and right-slant, so that users can select which they prefer (but presumably one is treated is default 'italic' style and associated with the italic text function in apps).
Personally, I never use Hebrew 'italics', but if I were to I think the decision about slant direction would depend on a number of factors. When asked to make italic versions of Adobe Hebrew -- primarily for technical reasons -- I decided that, given the target users of that family, facilitation of bilingual text was the most important factor. So Adobe Hebrew Italic has a right-slant to harmonise with the Latin italic. Apart from other considerations, this decision avoids the very problematic situation that arises from common punctuation characters, shared by both scripts, needing left-slant variants that may or may not be accessed by applications. [Adobe Arabic Italic has a left-slant, and there are environments in which the punctuation variants for Latin do not work, so you end up with right-slant Latin italic letters with left-slant punctuation. Ack!]
While I agree with David that the impulse to right-slant Hebrew 'italics' in modern typographic practice probably originates from copying Latin -- for either stylistic or technical reasons --, my palaeography books show lots of examples of Hebrew writing with a right-slant, as well as examples with a left-slant. In both cases, the slant appears to be a result of writing speed, as with the development of the Latin italic, and the direction is more likely the result of the writing habits of the individual scribe than a conscious stylistic development. There are, of course, plenty of people whose Latin-script handwriting slants to the left, but this never translated into normal typographic use.
Here is one example of right-slanted Hebrew writing, selected at random from Manuscrits médiévaux en caractères hébraïques. This example is from central Italy, from a document dated 1433, now in the Bibliothèque nationale, Paris. The same document also shows a more formal, upright writing in the larger size.
This example is instructive, I think, because one can see how such a slant would develop very naturally from writing quickly, with the down-strokes made quickly, with a flick. If you consider that the hand will be moving on, to the left, to begin writing the next letter, a flick in this direction on the down-stroke is much more natural than a movement to the right to produce a left-slanted letter. [This also accounts for the characteristic exit stroke of Latin italics: a flick up and to the right toward the starting point of the next letter.]
Diamond Left and Diamond Right look inspired by hand writing (might be they be called "script fonts"?). On a more prosaic tone, here is another font from Culmus (click "Fancy fonts" and scroll down) called "Ktav Yad CLM Medium Italic".
I wonder if the slant approximates the average slant of actual Hebrew handwriting.
> Since few italics has been made to lean right, so then left leaning italics are only tolerated.
Since few italics have been made to lean left, so then right leaning italics are only tolerated, but not what is preferred.
The truth is that any handwriting, script-like, or italic, typeface design can incline to the right or to the left. There seemingly is nothing in the design that requires the slant to be one way or the other? Is there?
A few years ago, I created a set of fonts with joining letters to teach children to form letters correctly, to write with a clear handwriting.
In one font, the alphabet appeared with each letter form separate, to show a child how the letter by itself is supposed to look. In a second font, the alphabet appeared with each letter form joined to the next letter to show a child how the letters together are supposed to look (this was very difficult as letters join each other differently, and there was no OpenType yet to address these different situations). In a third font, the alphabet appeared with each letter form joined to the next letter to show a child how the letters together are supposed to look, but very small arrows were added to each letter, to indicate how each letter should be correctly formed.
Most left-to-right languages, like English, have letters formed clockwise, while most right-to-left languages, like Hebrew, have letters formed counter-clockwise. However, slant plays no role in being determined by the rotation of the formation of a language's letters. Rather, the direction of eye movement is the determining factor.
This is why a right inclined letter is inappropriate for Hebrew, because eye movement is leftwards. In this incorrect case, the stress on one's eye movement in increased continously because of the letter's slant.
On John's example, I believe the very old beautiful Hebrew handwriting was likely from a right-handed person, as Rashi observes that most people write with their right hand. In writing Hebrew, the hand must be held somewhat awkward, as the arm covers the written text; hence, right slanted text is easier for the hand to form, due to the somewhat awkward position off the hand. A left-handed person has an easier time holding the pen to write Hebrew, by to produce right slanted text is easier than producing left slanted text.
I think that this correct. What does everyone else think?
> Hmm. Did not even see that “Évariste Galois” was inverted.
You typed this in typophile.com, on Mac or Windows, in Explorer, Firefox, or Safari?
In Adobe InDesign ME, when you type in Hebrew, and also pick the main text direction is left-to-right, and type also some words in English (inserting the letters from right-to-left), the words first appear in opposite order when you type the space character, but afterwards corrects itself with another word.
Did you type only one word, and then move the cursor to the place you thought it should be? Afterwards, when you typed the second word, then the sequence of words was reversed.
Or did you type both words one after the order, without moving the cursor?
This was typeset with XeLaTeX. I just forgot to put macros \beginL and \endL around Évariste Galois. A XeLaTeX input file is a text file with formatting commands. Changing direction when we change paragraph (with the polyglossia package) is automatic with language selection, but within a paragraph, we need to be more careful and I was not.
InDesign ME works similarly, on a paragraph by paragraph basis.
Each paragraph has a dominant language direction. However, within a paragraph, you can go in the opposite direction, but you must be careful.
The difference os that ID ME has a WYSIWYG interface, and does not have codes.
Does XeLaTex support an OpenType font, too?
InDesign ME supports an OpenType font on Mac and PC, MS Word supports it on Windows, and Mellal supports it on Mac. Publishing is only possible with InDesign ME, though.
A left-handed person has an easier time holding the pen to write Hebrew...
Have you tried to write formal Hebrew letters -- i.e. letters written with a broad-nib pen, not a ballpoint or pencil -- with your left hand, or watched a left-handed person trying to write such forms. Instinctively, because the script is written from right-to-left, one imagines that it might be easier for left-handed writers. But the only benefit for left-handers is that their hand does not obscure what has already been written: there are no benefits to the formation of individual letters; indeed, the traditional steep ductus (pen angle) requires either a very awkward angle to the arm and wrist or rotation of the page. Further, what are normally pull-strokes for right-handers become difficult push-strokes for left-handers. Try it.
> Does XeLaTex support an OpenType font, too?
Yes, XeLaTeX is meant for TrueType and OpenType. I could take John's list above and typeset it in four columns using XeLaTeX. I could also compare side by side SBL Hebrew and Adobe Hebrew with that same list (using a few unix commands and scripts to produce a .tex input file).
> Have you tried to write formal Hebrew letters
Did you ever have a look at that nice little book, Toby's The Art of Hebrew Lettering, that is digitized here on Yaronimus Maximus flickr site?
> on Yaronimus Maximus flickr site?
Where is Yaronimus Maximus flickr site?
> Where is Yaronimus Maximus flickr site?
Here is Yaronimus Maxiums photo stream and here is The Art of Hebrew Lettering. Just click on the link. I like cursiva and sfaradi (though sfaradi looks better on a photocopy I took from probably another edition years ago).
Thanks John. I was too tired; wasn't sure what I'm reading & understanding.
An example -- same book (by Emanuel Tov); Hebrew edition (second revised edition, 1997); English edition ( second revised edition, 2001):
Thank you for the links.
I saw this book when I was a child, Toby’s "The Art of Hebrew Lettering", that has many Hebrew alphabets with arrows showing how to correctly form Hebrew letters.
The "sfaradi" script caught my attention, as my research showed that many Askenazic-stlye scripts had the shapes of the aleph as this one, with the right top element facing to the left, and Sephardic-style alephs with right top element facing to the right. Here is an example:
Does anyone have older manuscripts with either Sephardic-style alephs with right top element facing to the right, or Askenazic-style scripts with the shapes of the aleph with right top element facing to the left, or otherwise?
John, I know you have and seen many older manuscrpts with Hebrew. Have you seen this? Do you think this is an accurate distictive trait?
Does anyone know if there is this combination in the Tanach?
Here is what I found in Genesis only. That is from some Karaite tanakh I fetched on the internet, then used a few unix scripts to generate a .tex file; utf8 text files are just great! Maybe worth checking though.
you probably noticed that the above text it in SBL Hebrew. That font seems to be quite slow compared to the other fonts I have used so far (with XeLaTeX on my mac). Are you able to process large texts with it in a reasonable time?
Sergey Malkin, the VOLT programmer at MS, used to say that SBL Hebrew was the most complex OpenType font he had examined. The only reason he doesn't still say it is that I've made another font (not Hebrew) that is even more complex.
Contextual GPOS lookups seem to take a longer time to process than most other lookups. The new version of SBL Hebrew I am working on uses GSUB to precombine below-base mark combinations, so that will probably speed things up somewhat. At the end of the day, though, the Bible text presents a complicate piece of typesetting, which requires some special behaviours in a font.
Israel, there are 22 occurences of final kaf with dagesh and sheva in the Masoretic text. My list of the first ten corresponds with what Michel has shown.
Thanks for the references.
I knew I saw it in a few places in Genesis: "v'yave'k'", meaning "and he cried", from the Hebrew verb for "to cry", "liv'coat".
The first two letters of the 3 letter root are each veis, the second letter of the Hebrew alphabet. Seemingly, the third letter of the 3 root letters, hei, falls away, in this case, and is represented by a dagesh point in the second letter of the 3 letter root, the final chaf. (Perhaps, the dagesh point is representative of an indirect pronoun, referring to the cheek upon which Joseph cried). So, a shva appears below, and a dagesh above it.
I wonder what makes this Hebrew text "from some Karaite tanakh".
I know the Christian Bible and the traditional Jewish tanach has some minor differences, mainly in the nikkud and taamim.
There two versions of the traditional Jewish tanach. In one version the hei and the end of a word appears which is logical, grammatically correct, and to be expected. In the other version, accepted by the Chassidic groups, the hei is replaced with an aleph, which is illogical, seemingly grammatically incorrect, and not to be expected.
However, in one of the oldest copies of a Sefer Torah scroll, which existed in the famed Maharal of Prague's synagogue, the oldest standing synagogue until Nazis destroyed it, in a certain word an aleph appeared at its conclusion and not a hei. This would indicate that the Chassidic tradition is correct, since it has historical verification.
According to Jewish mystical texts, this difference is more than just scholarly. The entire Five Books of Moses (Chumash) spells out G-d's name. The entire text, from the first letter, an aleph, until the last, a lamed, and all the letters in between, is precise. So, it's meaningful and significant whether it's an aleph or whether it's a hei.
* * *
If you notice, in my sample, both the dagesh and the shva is lifted up above the baseline. This is for purely aesthetic purposes, as having the dagesh lifted up, and the shva below, a large gap if white space exists between then, plus when taam and meteg appear, there is very little room to place them, plus there is a blatant inconsistency between the occurrence when no dagesh appears and when a dagesh appears.
Israel, I think your analysis of aleph forms is halfway correct. Yes, the construction of the flag is one of the style indicators, but it must be checked against other features of the individual typeface or writing hand: there are other features, such as the baseline shape of shin and related forms, that are probably more significant. I must also point out that I have not looked at as many Ashkenazic exemplars as Sephardic, so can't say for sure what variety of flag forms might be found in that style.
In Sephardic styles, there are two basic constructions for the flag: a single stroke down to the right and then curving to the left, and a returning stroke down to the right and then doubling back and down to the left. What is important is the construction pattern, rather than the resulting shape, which can be misleading in some cases. This is why I think you are misclassifying Adobe Hebrew as Ashkenazic. If you look at the construction of the aleph flag in Adobe Hebrew, the returning construction is evident. So although the flag shape is much more horizontal than in the traditional Sephardic styles, the stroke construction is derived from the Sephardic model. And, of course, if you look at the bottom of the shin, samekh, etc., these are very clearly Sephardic forms.
> I wonder what makes this Hebrew text “from some Karaite tanakh”.
I am not sure I understand.
Here is the story: I googled to get some utf8 Tanakh and found this Tanakh: Masoretic text with vowels in a site where it is written "World Karaite Movement Downloads". That gave me a .doc file that Word 2008 on my mac showed me as just question marks. I saved it in html format, edited the html to change the encoding header to utf8, opened that html with a browser which displayed it correctly, copied and pasted the text into an editor that allows utf8 editing to finally save a utf8 .txt file. I think that's my record in the number of steps I had to apply to get some workable .txt file from a source. After that I "massaged" a bit the .txt file.
In fact, I wanted to test a python script I had in my things to get trigrams from the utf-8 input but for some reason it skipped all vowel points and gave me nothing useful. The above output comes from a simple grep.
Thank you for the comments. I will review carefully your points.
Regarding defining the design traits of Ashkenazic and Sephardic traditions, I think we are treading upon new ground, as I have never seen this form of analysis before (even though in 2008 you would think that no one could discover something new).
This belongs in another thread, but can an MS Volt and an OpenType source file, or a font to ship, have a reference to a complex exception table for look ups?
For example, can I designate a CSUB to occur whereby one string "a" is replaced with a different string "b", except in the cases listed in the exception table.
Another question about MS Volt/OpenType support wild cards in the CSUB strings, whereby one string "a" with a wild card or more in between is replaced with a different string "b" with the same wild card or more in between?
I know the Christian Bible and the traditional Jewish tanach has some minor differences, mainly in the nikkud and taamim.
'Christian Bible' doesn't refer to anything in this context, since there is no single Hebrew text recognised as finally authoritative. Rather there are a few different lines of approach to the Jewish scriptures in Christianity:
The Greek Orthodox (and presumably Eastern Catholic) Churches still use the Septuagint (LXX) text, i.e. the Greek translation prepared by Jewish scholars in Alexandria, between the 3rd and 1st Centuries BC, and which was used by Jewish communities during the Hellenistic period. After the destruction of the Temple, rabbinical tradition restored the Hebrew version and established the Palestine canon, excising those books of the Septuagint for which Hebrew texts were not available.
The Septuagint is also the basis of the Slavonic, Armenian, Syro-Hexeplar, and Coptic versions.
Following use of Latin tanslations from the Septuagint, the Roman Catholic Church in the late 4th Century adopted Saint Jerome's Latin version translated directly from Hebrew texts. A few years ago, Pope John Paul II announced that new Catholic translations would take into account the latest scholarship of the earliest surviving Hebrew texts found at Qumran.
Today, almost all scholars studying the Hebrew scriptures -- whether Jewish or Christian -- work with the Masoretic text as preserved in the surviving Ben Asher manuscripts, the most complete of which is the Codex Leningradensis and the oldest of which is the Aleppo Codex. These Masoretic manuscripts are the oldest standard texts containing nikudot and ta'amim. Pre-masoretic texts are largely unvocalised and without cantillation. The Masoretes standardised the vocalisation and cantillation. The consonant text of different versions shows very little variations -- testimony to how carefully it was transmitted --: even the very oldest surviving texts, found at Qumran, are almost identical to that preserved in the Masoretic texts from a thousand years later.
Michel: I googled to get some utf8 Tanakh and found this Tanakh: Masoretic text with vowels in a site where it is written “World Karaite Movement Downloads”
The website may have been Karaite, but the text is not specifically Karaite. The Masoretic text is Rabbinical.
Israel, can you start a new thread to post your VOLT questions?
Re. the stylistic indicators in Sephardic and Ashkenazic traditions, this is something that Misha Beletsky discusses -- but without reference to the aleph flag, as I recall -- in his essay on Zvi Narkiss' work in the book Language Culture Type.
> XeLaTeX on my mac
I thought it was on Unix. Do mean Unix under System X, or under Sys. X 1.3 or later?
> ...the Masoretic text as preserved in the surviving Ben Asher manuscripts, the most complete of which is the Codex Leningradensis and the oldest of which is the Aleppo Codex... ...the earliest surviving Hebrew texts found at Qumran... The website may have been Karaite, but the text is not specifically Karaite. The Masoretic text is Rabbinical.
Has anyone documented the differences?
I was told by Philip Payne of Linquistics that the differences are only in few letters, nikkudot, ans taamim. Very few.
He told that there is a very old rabbi in Jerusalem that understands all the very minor differences, but like many Orthodox, he refuses to engage in a dialogue with Christians about the Bible.
Interestingly, his OpenType Biblical Hebrew fonts crashed in Adobe InDesign ME but not in MS Word for Windows, while SBL-Hebrew performed the opposite.
> I thought it was on Unix.
I am running XeLaTeX on OS X 10.4. Darwin is unix. When I work on my mac, the screen is filled with shell windows. To get a "grep"able input above, I used the standard Unix commands sed, and awk on my mac (in fact, I compiled gawk to get a bit more than awk). I also compiled XeLaTeX from the sources on my mac using the standars commands "configure", "make" and "sudo make intall". In fact, I often start the fontbook by just typing "open /applications/Fon*ook.app" intead of browsing (I could have put a script in my $HOME/bin folder). That's all oldtime unix.
Are you familiar with the page layout of a page of Talmud?
I seek to know whether XeLaTeX or InDesign ME is better more robust engine to typeset a new edition of the Talmud.
A page of Talmud consists of three parallel text blocks.
The outer left and right text blocks have two widths, a wider width at the top and bottom of the column, and the center part has a narrower width.
The center block only has one width. At the line under the depth of the center block, the two outer blocks return to wider length, which they had at the top.
I would like to create a script to automate the page layout of some 20,000 pages which are formatted in this way.
I seek to know if this is possible using XeLaTeX.
> Are you familiar with the page layout of a page of Talmud?
No but enough to think that doing the layout from scratch would not be an easy job. If you know well InDesign and think it can do the job then, in my opinion, you should stick with it. There is no better environment than the one we know.
I know LaTeX because I use it to typeset mathematics. That is a tool I have been using for years.
I have noticed (on non extensive tests) that justification of narrow columns in InDesign seems to be better than with TeX even if in principle InDesign justification is based on Knuth's algorithm. That might also be a point to take into account if your columns get narrow.
I tried Henri GH on MS Word for Office 2007 on Vista with the Book of Joshua. No crash. It used to take down XP, too! I guess UniScribe is much stronger now. The furtive patach stayed centered.
Why do think Philip's OT Biblical Hebrew crashed InDesign ME but not MS Word even on XP?
Also, Adil Allawi's Ready Set Go International could display every CSUB in Payne's fonts or SBL? But it didn't crash XP or itself.
Philip said his fonts work in Mac's Mellel, too.
Thank you for your views.
I am before typeseting over 30,000 complex pages (it's contained in 20 thick volumes of 11" x 17" pages of many different columns. It's done every 200 years of so. :) ).
As you can imagine, I want to automate the page layout as much as possible.
The question is can a script be written in InDesign ME to do the page layout, and just pour in the text data. I would like to change the typeface, and its all automatically vocalized with the correct nikkud.
I wonder if that very old rabbi in Jerusalem is Rabbi Winefeld of Shay Lemorah publishers and the Shay Lemorah commentaries.
> As you can imagine, I want to automate the page layout as much as possible.
If you look at page 7 of the article (from tugboat) Makor: Typesetting Hebrew with Omega you will see what can be done with another extension of Tex-LaTeX, namely Omega, to typeset the Talmud. The files are freely available here on CTAN. I have no idea how easy it would be to adapt those sources to some extension of TeX-LaTeX using otf fonts (there is XeTeX-XeLaTeX, there is also luaTeX available for the pc that I have never tried and that, I think, is in a more experimental state than XeTeX).
Thank you for the reference to Makor, and the fascinating article.
My concern are the fonts.
I seek the typesetting engine to fully support OpenType's context substitution features (CSUB), supported by InDesign ME (which uses its own internal marking engine on either platform) but not by MS Word (which uses UniScribe on XP and much more robust on Vista).
I noticed that in terms of precise positioning (CPOS), XeLaTeX performed, but in terms of stylistic alternatives and contextual replacement (CSUB) then XeLaTeX's performance is questionable.
For example, the ayin patach lamed combination is a good example for use of a stylistic alternative, and shin kubutz ayin patach combination of a contextual replacement.
In the first case, the ayin changes from the form where part of it dips below the baseline, and is replaced a different ayin with the form which remains above the baseline, to make room for the patach.
In the second more complicated case, the patach itself is repositioned according to the context: being under ayin (or chet or dagesh hei) and following certain consonant/vowel combinations.
For me this is new. Not only can your fonts operate in MS Word (for Office 2007) on Vista, but even the feature like the contextual replacement/positioning features of OpenType needed for the furtive patach operate correctly as well.
> in terms of stylistic alternatives and contextual replacement (CSUB) then XeLaTeX’s performance is questionable
I am afraid you are right for XeTeX on the mac with system 10.4 and otf fonts (AAT fonts seem to work correctly). I thought things were better for otf fonts under OS X 10.5. I don't know how it is on the PC. Adam Twardoch is usually up to date on those matters.
> Misha Beletsky discusses — but without reference to the aleph flag, as I recall — in his essay on Zvi Narkiss’ work in the book Language Culture Type.
> Zvi Narkiss' work
When I lived in Rochester, NY from 1989 to 1995, I saw a holocaust (or from the Warsaw ghetto) poster, announcing a concert, hand-drawn by one of the residents in Yiddish (Hebrew letters). The typeface design resembled very much Zvi Narkiss' popular Narkiss Tam sans serif design. I believe that I saw the poster at RIT's famous Cary Library.
Is the part of the article to which you refer?
Interestingly, Misha Biletsky mentions the remarkable significance the design of the Romm family printing establishment of Vilna, Lithuania, and the imoportant role it played in dominating Jewish printing for over one hundred years, without actually showing a sample of it.
He cites design flaws without going into any details. He neglects to mention the involvement of Bodoni's school of type design in its creation, as reported and published by Frederick Goudy.
For such an important type design, you would think that it would get more than a passing remark.
> Adam Twardoch
How do I contact him?
As I understand it, OTF fonts are more than fonts of other formats, such as TrueType or PostScript, with just a header added to it to make it multi-platform.
Rather, OTF fonts contain programming instructions that affect formatting and appearance. To achieve this, OTF fonts need to be rendered on a marking engine, such as Microsoft's UniScribe, Adobe InDesign's internal marking engine, and Apple's limited one.
A code-based application, such as Tex-LaTeX, uses no marking engine, and simply sends its formatting codes, however complex, to output, to a screen proofing mode, to a desktop printer, or to higher resolution device.
So, the special contribution of OTF is lost at this time. I'm sure the world won't stand still, and Tex's brilliant people will add a means to tap into the power of OTF fonts.
Adam Twardoch is here on typophile.
Concerning Apple, they have their own engine with a finite state machine. XeTeX on the mac can use two renderers, one for those Apple AAT fonts, one for Opentype fonts. The fontspec documentation gives examples of use.
Israel, this is the first I've heard of SBL Hebrew crashing Word in any version of Windows. I've never encountered such a problem, and neither I nor SBL have received any support queries about it; indeed, Word 2003 was the primary target app for SBL, and the font has been more thoroughly tested in that app, under Windows XP, than in any other environment.
Please email me an exact description of the problem (system version, incl. service packs; amount of RAM; Office version; steps taken resulting in crash)?