Hebrew Grammar Distinguish at VOLT Level

Typograph's picture

Gohebrew: this is the place for you.
first will discus what glyphs are needed for such a project, and the best way of designin them.

Seconed, We will start at ground basics decisions we must make brfore we start.

then will talk about the sheva na\nach rules on by one in depth

So Gohebrew Fill free to start the discotion

gohebrew's picture

Eli, to create a Hebrew grammar font, to use later in Volt, I would create a set of 22 extra glyphs. There are ten nikkud, and a dagesh.

So, if add two more for every one that is there now, 11 + 22 = 33.

They do not need Unicode values. Rather, they can simply be alternative glyphs for each nikkud Unicode value, and the dagesh Unicode value.

If they are placed in the FontLab file in the existing place of a font imported into Volt, adjustments to the total number of glyphs need not be made to the project in Volt, or to the FontLab file.

If they are placed in the FontLab file in the new locations of a font not yet imported into Volt, adjustments to the total number of glyphs need to be made to the project in Volt, and to the FontLab file.

I suggest that the new graphic forms of the nikkud and dagesh symbols be intuitive. Their new form should be similar to their old forms. A kal version should be simply a hollowed version of the nikkud and dagesh; a chazak version should be simply an outlined version of the nikkud and dagesh.

In the case of the already familiar kamatz katan, it should remain the same in its Unicode value location, but have the above appearance in the other locations in the font.

Similarly, I created and placed in my fonts, graphic symbols for the shvah-na: a small floating asterisk used by Lubavitch/Kehos; a small floating rafe-like used by ArtScroll and others; I added the enboldened shva, used by many Sephardim; and the circled asterisk, used by Shai LiMorah publishers.

But this is only for publishing purposes. But the shvah-na should have the above cited appearance in the other locations in the font.

I believe we do not need to create anything more in the FontLab files.

Generate a .ttf font, to use and edit in Volt.

Typograph's picture

I don't fully understand, can you show me a sample od your added glyphes??

Typograph's picture

I have 2 main reasons not to use indication of sheva over the letter.

1. Not a native symbol (people reconize the rafe for other use).
2. with taamy mikra it makes things much more complicated.

A glyph replacing the already type sheva i thing is a better approach.

1. Are you gowing to do automatic Dagesh kal and Hazaq ???
1. are you going to do automatic qamats qatan???

gohebrew's picture

I will post the 22 added glyphs soon.

In the shvah-nach and shvah-na will appear as a regular shvah. Each however will be hollowed for shvah-na, or outlined for shah-nach.

The other graphic symbols are only for printing purposes, as these publishers, Kehot, ArtScroll, Shai L'Morah, and others, are used to seeing these images. However, for Hebrew grammar purposes, the other images are to be used.

About automating text with taamei mikra, this is addressed alreay in the furtive patach.

About automating everything, the answer is yes.

At first everything. Later, some fonts will have them not automated by simply creating subsets of the Volt program.

Typograph's picture

do you have your 22 glyphes allready designed, or you are about to design them???

gohebrew's picture

Samples

Typograph's picture

the black sample will work
The third can work
but the second one is problematic in print

gohebrew's picture

The original Nikkud design for the Hadasa/Henri/Friedlander font was created by John Hudson - as you see, the Hebrew grammar graphic symbols are adaptations of the originals.

I think that the middle graphic symbols need the white space between the nikkud and the out line to be increased, in order to be legible at smaller sizes.

Typograph's picture

On That i agree.
I did't see the graphics dor dagesh kal and hazaq

gohebrew's picture

Eli, if the space is increased, and center part decreased, I think it should work.

For the hollow forms, I created them in Fontographer very easily, by going to the fourth Element menu from the left, to the fourth entry, Selection Info, and selecting Contour and Fill Properties, and turning off Fill, and clicking Stroke Outline with a 2 setting for the ems.

I could create them as filled outlines by redrawing them.

For the outlined forms, I simply copy and pasted them upon themselves. Then, I picked the Element menu again, and the ninth entry Expand Stroke, and set the Pen Width to 40. Perhaps, it should be 60 or even higher.

gohebrew's picture

Dagesh kal and hazaq would be similar. Look at the English symbol's chart. It would be the same thing but smaller.

Doo we have the meteg's rules defined in Chomsky's book, Bill?

gohebrew's picture

Lesson One
=======

First, each nikkud and dagesh must be defined as a Glyph Group by itself.

Second, each kal etc. nikkud and dagesh must be defined as a Glyph Group by itself, and each chazak etc. nikkud and dagesh must be defined as a Glyph Group by itself.

Another dagesh symbol is required to represent a mappik. I suggest this is a dagesh outlined with two circles. It must be defined as a Glyph Group by itself, too.

Volt processes earlier GPOS look-ups, and afterwards Volt processes later GPOS look-ups. So, the sequence or order of the GPOS look-ups is important. The order of the Glyph Groups is not important.

Typograph's picture

You should call me for clearifyng some issues;
But I donw' see the logice in your grouping;

your basic groups should be

gohebrew's picture

Excuse me, Eli, for not calling. I've been more out than in today.

Yes, the Glyph Group definitions are incorrect.

I can not address this now, as I will be busy for the next few weeks.

Typograph's picture

Ok, So lets start from the begining.
what are the ingredients that we will need at the font level?
Israel, You are talking about 3 sets od glyphes
1. the normal sheva,qamats,dagesh.
2. shva-nach,qamats qatan, dagesh kal
3. sheva na, qamats qatan, dagesh hazak.

We will need more than that.
first, a hataf qamats is always qatan, so we need another hatafqamats qatan.
second, in general the shevas of the hatafim are NA, so we need another set of hatafim that have sheva na'a attachet to them.

Now will need another set of this glyphes
Sheva NACH, Sheva-NA, Qamats, Qamats Qatan, Dagesh Kal and Dagesh Hazaq
This set of glyphes we need for later use in volt not as mark glyphs but SIMPLE or BASE GLYPHES.

The reason for this is, because i can't sub a mark with mark (shva nach to Na) with PROCESS MARKS = NONE
In cases i want to check words without mark sequence in the context section.
So First i sub the diacritics to their equivlent BASE GLYPHES then in the next lookup I sub the shva na to nach as base glypes
and finally switch them back NA-BASE to NA-MARK and NACH-BASE to NACH-MARK

Until here is every thing clear???

gohebrew's picture

> We will need more than that.
first, a hataf qamats is always qatan, so we need another hatafqamats qatan.

Shai L'Morah publishers of Jerusalem (Rabbi Yehuda Winefeld Shlita) already documented this in the entire Tanach etc. according to the Minchat Shai.

Prof. Dotan was surprised, as he never saw this distinction before. Do all views accept this, David?

gohebrew's picture

> second, in general the shevas of the hatafim are NA, so we need another set of hatafim that have sheva na'a attachet to them.

Eli, what are hatifim?

Typograph's picture

Now why is this important????
building a project like this will take you to the limits of VOLT's Code block.
it is very important to know how to optimize your project with the least amounts of anchers, kerns, lookups and lines of context.

most projects in other languages can be a bit dirty, they may become complex but do not reach volts limits (not realy Volt's foult but OT)

specially when using context as dictionary.

Word that are built around PAALA פעלה, have a pattern of
qamatsshevaqamats+he
in these cases you will never know if the qamats is qatan and the sheva is nach, or the qamats is gadol and the sheva is na.
unless i have a mapik at the last He or BACHLAM Letters before but plain you can never know.

So.....
We build a dictionary for known cases
Chochma
Orma
Hogla
hozka

and so on and so on.
doing this with PROCESS MARKS=ALL will jam the intier project

gohebrew's picture

> Now will need another set of this glyphes
Sheva NACH, Sheva-NA, Qamats, Qamats Qatan, Dagesh Kal and Dagesh Hazaq
This set of glyphes we need for later use in volt not as mark glyphs but SIMPLE or BASE GLYPHES.

I think they should be standard in Unicode. If they are there, all Jewish publishers will use them in their prayerbooks and Hebrew language primers, plus the Orthodox publishers will start featuring them in their books of Psalms and Pirkei Avot etc., and the secular publishers will feature them in their books of poetry.

Typograph's picture

HATAF QAMATS, HATAF PATAH, HATAF SEGOL
(caps lock on >> SHFT 1,2,3)

gohebrew's picture

> Now why is this important????
building a project like this will take you to the limits of VOLT's Code block.

In discussing this with John Hedson, I don't think so.

Raphael's experience was because of poorly written code.

What we speaking about: a dozen or two extra glyphs - the same in GPOS look-ups?

If the code is written well and tight, there should be no problem.

Typograph's picture

Israel, you don't think so???? you make me laugh!!!! you have no idea what we are talking about... do you????

We are talking about hundrets of lookups & thousands of lines of context!!!!

Well, try checking all the cases with PROCESS MARKS=ALL
meaning you have to check each word with all of the conditions of NIKKUD + TAAMIM And also Dageshim in the middle and on shin the also shindot or sindot.

Israel, trust me your project wont compile before you know it.
oh, And israel... how do you know if the code of Ariye Marzel was clean or dirty???
do you have any idea of the requirements and demands of this particular project????

Typograph's picture

It makes no difference if you spoke with hudson or not, if you'r not clear on what you'r doing and how, trust me, your project will fail to compile.

Typograph's picture

Now lets talk abut DAGESHIM.
There are two ways to go about it.
1. using a preset of dagesh kal and a preset of dagesh hazak which ar positioned at the font level.
2. positioning Dagesh kal and Hazaq with volt.

Approach 2 was my approach, because this way the user has full control over the posioning via Indesigns Diacritic Postioning.
But this approach will damnd much more code at the VOLT level.
It also allows me to use a Dagesh hazaq axactly the same as the normal dagesh, for those who don't want the dagesh distinguish but do want the sheva na and qamats qatan
but i still need to know what is dagesh kal and what is hazaq...

using approach 1, might be better for you. it makes coding life much simpler healtier and happier.

After we make this desitions will talk further.

gohebrew's picture

I am clear very much. I am listening very carefully to you and David, more than you think.

I know that Volt can handle this complex a tast, as I ran it through more complex "programs". Processing time is longer to handle very tenious look-ups. Who cares? It's a couple extra seconds on a slow laptop.

gohebrew's picture

> We are talking about hundrets of lookups & thousands of lines of context!!!!

I know. A mean less than two dozen more.

If the 'program' is only nikkud etc and not taamei mikra, it's very short. Can it run with the taamei mikra too?

Typograph's picture

Israel, Its not a matter of computer resources, it's a matter of OT's phisical limitation of code block.

VOLT will give you an unknown compilation error.
as long Volt compiles the project, do as many of lookups you want.

Typograph's picture

my project supports perfect placment nikud and taamim and abjustments when needed.
(I test the font on the intier TANACH fo DIBUKIM) + all the sheva na-nach functions.
If you wish to know more about my project
here is a link to the tutorial movie of the fonts functions

http://vimeo.com/3945578

gohebrew's picture

> Now lets talk abut

We need 3 sets of nikkud and dagesh. The two are alternate versions of one.

One OT font via GSUB substitutes or exchanges the basic set of nikkud and dagest into the two alternative sets, either the second or the third.

Except as you did, all shvah-nas are used only to replace the standard shvah.

Then through later GSUB again, but afterwards, the shvah-nas are replaced selectively with shvah-nachs.

Not many GSUBs, I think two or three, and a few Glyph Groups.

I need to know what causes the kal or chazak state of the dageshes and each nikkud.

gohebrew's picture

> Israel, Its not a matter of computer resources, it's a matter of OT's phisical limitation of code block.

John said that there is no upwards limit.

Typograph's picture

so he says, if i add 3 more lines of context, my project will not compile, no matter what context i put in. i hav to give up one function for another to work.

Ask Ariye Marzel and see what he has to say about it...
the fact that jhon hudson never had problems compiling his projects, does not mean that there are no limits to OT lookups.

I built the project twice, the first time i learned volts limits, the built it again working around these limitations.

so....
you can either learn from my experiance or try it out your self.

gohebrew's picture

ישר כחך על סרטך

From a design point of view, I liked very much one of the early fonts you designed, which had many qualities similar to Friendlaender's Hadasa.

david h's picture

> Prof. Dotan was surprised, as he never saw this distinction before.
> Do all views accept this, David?

Israel,

Accept what -- the distinction (graphic distinction?) ?

Typograph's picture

I toled him earlier in this discution that all HATAF-QAMATS are allways Qatan, so Israel ansawred that Prof.Dotan never heared of such thing, so he is asking your opinion if all HATAF-QAMATS are really Qatan.

Typograph's picture

Now lets examin some of the basic rules of sheva na and nach.
A sheva na continues a sound and a sheva nach stops the sound.

1. A sheva under the first letter of a word will allways be NA
This sheva is instead of Segol so it sounds Like E
also logical it would be imposible to stop a sound right at the begining of a word.

2. A sheva under a letter with a dagesh (any dagesh) Will be a sheva na.

3. Two letters rhe same like bet bet or lamed lamed the sheva under the first letter of the same will be a NA.

4. two shevas one after the other the first will be nach, the second na.
But if the two shevas are at the last two letters of the word then they both are NACH.

5. A sheva after a Tenua Qetana Is NACH after Tenua Gedola Is NA.
here is where we have to know if a qamats is tnua gedola or qetana or in other words Qamats rachav Or Qamats Qatan.

here mainly is where a Meteg comes in misuse.
Text use a meteg instead of a mafsiq so meteg is used also for this and also for mileeil and milera.
A qamats with a meteg should be rahav and the coming sheva a NA
with out a meteg the qamats is qatan and the following sheva is nach.
but this is not a ruling rule.

all this about the qamats qatan is very much in general
we will examin qamats qatan more in depth later

there are more rules with taamim
like uf i have a sheva with a taam maamid so even if before i have a tenua gedola the sheva is nach.

Typograph's picture

whan i say misuse of meteg i mean this
There is a taam called SOF PASUK, there is no Sof pasuk in unicode only meteg.
this meteg is used also in text with nikud and withaout taamim.
the use of the meteg is generaly used For grammer, for mileail and milera and for taamim also not only sof pasuk, shmuel riarchi uses it in is own ways.

but lets just say that the main use of meteg is for grammer.

Typograph's picture

Now here is what i am saying why you can't have the font insert metagim in text automaticly.
in alot of cases i am depending on metagim to distinguish between qamats gadol and sheva na & qamats qatan and sheva nach.

So as i said before, forget automatic insertion of metagim in text, and dageshim in BEGED KEFET for that matter. anyways a font should not insert unexisting charachters.

Typograph's picture

Oh, BTW, regarding meteg.
in my font i allow a hidden meteg by pressing SHFT+7 (&) to have cases functioning as if there was a meteg but hidden (just a 0 spaced character like Null)

raphaelfreeman's picture

gohebrew: sorry but you are wrong. There was nothing wrong with the code. It is a known (by Microsoft) limitation in VOLT. You just can't compile it. There is no plans to make it possible to do.

Please don't blame bad programming.

gohebrew's picture

Raphael,

2 experts in Volt say the opposite.
You claim not true, based upon an unreferenced Microsoft document. Please cite it.

My own experience confirms the former opinions. Length is not the cause of errors; poor written code is.

Your claim is like my suggestion that a document shows Koren was more than inspired when he made his famous design. You did not accept it on hearsay, but wanted to see tangible proof.

Please bring tangible proof.

===

What is your view about the new graphic forms for Hebrew grammar symbols?

Solid or filled-in forms for standard or existing forms of nikkud and dagesh
Hollow or unfilled forms for kal, or katan forms of nikkud and dagesh
Outlined with filled-in forms for chazak or gadol forms of nikkud and dagesh

gohebrew's picture

Eli,

Prof. Dotan viewed Rabbi Winefeld of Shai L'Morah publisher's symbol for hataf-kamatz-katan, and remarked: "What's this?"

It was as if he agreed with you.

Apparently, Rabbi Winefeld sees a distinction that some hataf-kamatzes are like kamatz-katan, and some are not.

gohebrew's picture

> 2. A sheva under a letter with a dagesh (any dagesh) Will be a sheva na.

This is a practical rule. A simple GPOS to write.

===========
------------------------
*******************
+++++++++++

gohebrew's picture

> 3. Two letters rhe same like bet bet or lamed lamed the sheva under the first letter of the same will be a NA.

ibid

===========
------------------------
*******************
+++++++++++

gohebrew's picture

> 5. A sheva after a Tenua Qetana Is NACH after Tenua Gedola Is NA.
here is where we have to know if a qamats is tnua gedola or qetana or in other words Qamats rachav Or Qamats Qatan.

SO, EVERY SHVAH AFTER A KAMATZ IS EITHER NA OR NACH, IF THE KAMATZ BEFORE IS EITHER Qamats rachav Or Qamats Qatan.

How do we now if it's either a Qamats rachav Or Qamats Qatan? If that kamatz is either a Tenua Qetana or a Tenua Gedola.

If Tenua Qetana, then it's a kamatz rachav, and the subsequent shvah is na,

or if Tenua Qetana, then it's a kamatz katan, and the subsequent shvah is nach.

Right?

gohebrew's picture

First, I must digest this.

AFTERWARDS, I'LL GET INTO METEG.

AT STAGE ONE I WASN'T GONNA DO THE METEG.

Typograph's picture

Qamats qatan = Tenua Qetana the following sheva is nach.
Qamats Gadol = Tenua Gedola The following Sheva Id NA
So I need to write here alot about qamats qatan in depth.

for now i am waiting for questions you ,ight have or unclear issues u want clearfication
after i fill that we are on solid ground, W'll continue on.

raphaelfreeman's picture

re coding: I'm not sure how I can prove anything.

I'll tell you what. You go ahead and try and do it. If you succeed let me know and I'll say bad programming, if you fail, then you'll know. But the only two people in the world (to the best of my knowledge) that has tried to do it, haven't managed 100%. Now, it depends on the font. It would work with Hadasa because the font is very wide and there will be very few collisions, but with a font with very narrow widths, it's harder due to the sheer number of substitutions.

Regarding your ideas for hollow or outline nikud. I personally don't like this approach for 3 reasons:

  • it's not used in any other typeface
  • it will be impossible to print
  • it's not intuitive as to what on earth it means
  • It could be useful in a book on Hebrew grammar with the specific words printed at 24pt, but for general purpose use, I'm not even sure the point of it.

  • gohebrew's picture

    Raphael,

    My goal is to create a 2 different Hebrew grammar 'program's in Volt.

    The 'program's are identical to each other, except thar one 'program' includes only GSUB look-ups which mark the shvah-na, and the kamatz-katan. This a general purpose 'program'; the results are Volt-generated fonts, in any design, intended to be used commercially by anyone; only the shvah-na, and the kamatz-katan, are displayed, and printed.

    The second 'program' includes GSUB look-ups which mark every nikkud and dagesh. This NOT a general purpose 'program'; the results are Volt-generated fonts, in only certain designs, intended to be used educationally by achools (and perhaps publishers of educational books); every nikkud and dagesh, are displayed, and printed.

    ===

    You write that there are two opposing factors why this may not work:

    1. There are a large number of look-ups which will 'crash' the computer, as every nikkud and dagesh must be compiled for substitutions.

    2. Except for Hadasa, Hebrew letters are too narrow to avoid collisions.

    Which it?

    It seems that these are two separate issues.

    The latter factor seems to an issue which can be addesed in planned out design. Hence, it is not an issue.

    The former factor is not substantiated by you, as you can not cite this MicroSoft document. Furthermore, my two Volt experts suggest that there is no upward limit to processing a Volt font in InDesign CS/x ME.

    Now, you say, "let us see".

    ===

    >> Regarding your ideas for hollow or outline nikud. I personally don't like this approach for 4 reasons:

    1) it's not used in any other typeface
    2) it will be impossible to print
    3) it's not intuitive as to what on earth it means
    4) It could be useful in a book on Hebrew grammar with the specific words printed at 24pt, but for general purpose use, I'm not even sure the point of it.

    ===

    > 1) it's not used in any other typeface

    This is why exactly that I am not introducing any new design. A hollow or not filled-in letter is the same design that a person is used to see. But now, it is merely hollow or not filled in.

    > 2) it will be impossible to print

    At small types, like 12 pt and below, this is true for desktop printers. At larger sizes, like 14 pt and larger, this is possible. Try and see for yourself. Highlight some text in InDesign, and apply the Swatches to it in white (paper) near the top of its pop-up window. Then, go to swatches and select the text outline, apply the .5 or .75 pt.. Then, go back to Swatches, and apply black to the text outline. Now print and see.

    > 3) it's not intuitive as to what on earth it means

    I disagree. Let us not argue just for argument's sake.

    The design is exactly the same, but hollows for pronouncing or intending differently, or outlined for pronouncing or intending differently.

    > 4) It could be useful in a book on Hebrew grammar with the specific words printed at 24pt, but for general purpose use, I'm not even sure the point of it.

    The Volt fonts intended for ue by everyone are not radically new. The shvah-na and kamatz-katan have been printed in books for over a 100 years, and used by anyone who prays during that time with those books.

    There are over 100,000 religious children today who pray daily. Most don't have an inkling how to pronounce differently a shvah-nach and a shvah-na. Isn't it time for something to be done?

    david h's picture

    > There are over 100,000 religious children today who pray daily. Most don't have
    > an inkling how to pronounce...

    Israel,

    What is the right pronunciation -- the name of God אֱלוֹהַּ ?

    gohebrew's picture

    E-l lo ah
    or
    E-l lo ha - with emphasis on the last syllable, if the dagesh creates an indirect pronoun, as in "his G-d" (if the reference is to the Creator, or to any false god, then it would be "his god"

    I never understood the first possibility, since John taught me about the furtive patach.

    I think that I am a fine example of an am-haretz, no?

    Syndicate content Syndicate content