Thursday, March 31, 2011

Shibboleth: New version 0.9b... and typing in Greek

I have posted about the free and very helpful Shibboleth program from Logos. (HERE with a full description when first released and HERE in relation to typing in biblical Hebrew.) Logos has now announced an update to it, and you can download the program HERE. As described on the Logos web site:
Shibboleth is a tool for typing Unicode text in ancient scripts. It was designed to help people unfamiliar with a script easily enter the correct characters, and then copy text to the clipboard in Unicode or another format.

While a keyboard layout is provided for several scripts, the emphasis is on helping the user recognize and select the proper characters. To that end, user input is shown in both typed and rendered format, with multiple font options, and all of the characters for each script are selectable from a well organized palette on the right side of the application window.
 There have been a number of improvements to this new version, but of most importance to me:
  • New Rich Text copy mode that preserves font choice and doesn't put XML tags around the text.
What this means is that you can now type your text using the helpful and logically laid out keyboards or clicking on the symbols, then hit Copy, then paste directly into any RTF word processor without having to remove XML tags. You have a few choices of fonts you can use, including the new SBL Hebrew and SBL Greek. Excellent! With this new version, you can even type in Akkadian, Hittite and Old Persian Cuneiform, and Egyptian Hieroglyphs!

So far, so good, but I still have some quibbles with the Greek...

The main problem is that Shibboleth renders polytonic Greek with combining characters rather than with the precomposed ones. What does this mean?
UPDATE: Be sure to read the comment by Vincent Setterholm of Logos. He has correctly surmised that I am using Word 2007. The issues I describe below are in part due to the Word 2007 handles the glyphs and in part may be due to the font (especially Palatino Linotype). Shibboleth should NOT be the problem if you are using Word 2010.
What happens is that when type characters that have breathing marks and accents, there are two ways of creating the character you want. One way is to start with the vowel, then add the marks using glyphs that don't take their own space. In Unicode, each glyph has its own 4 character code, so as you can pick up from the first line in the graphic below, an η is 03B7, the iota subscript is 0345, the rough breathing is 0314, and the acute is 0301. 
(Want to see for yourself? In MSWord, place the cursor after any character and hit Alt-X. Hit it again to return it to its normal appearance. Or, get the BabelMap program which maps every single Unicode character.) 

The other way a character is formed is if the input method is smart enough to figure out when you are all done with the character which one you want and provide that specific character which has been precomposed in the Unicode system. As you can see in the second line above, that very same character that takes 4 glyphs to form in the first way is represented by a single character in the second.

So, what difference does this make? To look at it, not much. To look closely at it, a little. Since those placeholder glyphs have to work with all vowels, their location has to be estimated a bit to fit properly. With the precomposed characters, the character with all its glyphs can be formed exactly how you want it to look. So, you will have to look very closely, but the top line is the character with all the glyphs added. The bottom line is the precomposed one.
You might be able to discern in the second row that the acute accent
and rough breathing mark are just a bit more centered and tucked down.

No big deal you say... But part of the beauty of Unicode is that you should be able to use any polytonic Unicode Greek font and the character displayed should be exactly the same. (This was a major problem with the old TrueType fonts where designers placed different Greek characters on different keys.) The problem with polytonic Greek is those placeholder glyphs whose location relative to the vowel is not fixed. So, using combining characters, if you switch the Greek font, you may end up with an ugly mess.









As you can see, what may look quite attractive in SBL Greek may not look so good in Cardo, Gentium, or Palatino Linotype. If you just composing documents for your own use and only use the same font, this may not be important, but if you plan to send that document to someone else or to a publisher who may not have the same Unicode font as you, then it is a problem.

So again, thank you very much to Logos for sharing Shibboleth for free. Hey, how else other than Shibboleth are you going to easily type in Egyptian hieroglyphs?!
UPDATE: In light of Setterholm's comment, if you use Word 2007 or Google Docs or even WordPad, you will probably want to use one of the solutions given below for getting the precomposed characters.
Some notes:
How else do you get properly precomposed Greek characters? (The following solutions do work properly for Word 2007.) As far as I know, you can use:
  • Tavultesoft Keyman (a wonderful program for $19) integrates into your system so that you can type in brilliant Greek and Hebrew right in the program. No need to copy/paste...
  • Tavultesoft also has a free online notepad for typing in any language you want. If you want to use Greek, you have four different keyboards to choose from. (Where is that psi?) I find "Greek Classical" to be the most usable.
  • The free online composer at TypeGreek.com is very nice, and note that it is even smart enough that if you type a sigma followed by a space or punctuation, it will convert it into a final sigma. One problem with TypeGreek: if you are typing and don't know what character you need, you hit the "Alphabet Key." This provides the layouts, but when you go back to the previous page where you were typing, it will be all gone.
  • The free online composer at GreekInputter2. The keyboard is a bit less intuitive, but you can display the layouts so that they are visible while you type in the box.
What else?
  • There is a similar issue of markings with biblical Hebrew, but there really aren't that many precomposed characters. The Hebrew Unicode system does rely more on combining characters, and it is not really a problem.
  • Logos (and BibleWorks and Accordance) use the precomposed characters in their Greek texts. That is, if you copy a chunk of text in one of these programs and paste it into your word processor, you will get the precomposed forms, and any Unicode Greek font you use will look fine.

2 comments:

  1. You're missing a key piece of information in your critique with regards to precomposed glyphs, etc. All of Microsoft's rendering technologies (WPF, Office, Uniscribe) will automatically substitute a string of uncomposed Greek glyphs into the precomposed form IF all the standalone glyphs are present in the font. If all the glyphs are not present, then the font substitution takes over first and you're seeing glyphs from more than one font displayed and they won't compose properly (you can test this by highlighting a whole string pasted from Shibboleth with Palatino Linotype and you'll see that the font info in Word shows up blank because the font for the highlighted text is mixed). So in your example, there has to be a glyph present at 0345, 0314 and 0301 - if those are all present, then Microsoft displays 1f95, but if even one is missing, then font substitution kicks in and things look ugly.

    Now Office 2007 in particular broke a lot of those glyph look-ups, and I'd hazard a guess that you're using Word 2007 to get the Gentium image in your blog post, because Gentium has all the necessary characters to form properly. If you were to run the same example in Word 2010, these tables have been fixed and Gentium looks grand.

    There are some Greek fonts that don't have all the stand alone glyphs needed to make the substitution tables work well. A lot of times the font designer just doesn't know this particular fact about the Windows environment and if you ask nice, they'll add the missing glyphs. We probably should remove Palatino Linotype from the list of Greek options in Shibboleth since it doesn't paste out well in Word with the new RTF copy.

    ReplyDelete
  2. Thank you for that clarification, Vincent. You are indeed correct that I am using Word 2007. I have added an UPDATE to the post to draw attention to your comment. I am not able to test using Shibboleth with Word 2010, but I did try it with Google Docs and with WordPad. Both of them revert to the combining forms. (Using one of the other solutions I suggest does provide the precomposed forms.)

    ReplyDelete