Technical Details about GreekKeys Unicode 2008

November 11, 2015: sales of GreekKeys 2008 ceased on October 28. The updated product GreekKeys 2015 is now available at the web site of the Society for Classical Studies (free to SCS members, $20 for non-members, free or discounted for those who purchased GreekKeys 2008 on or after August 1, 2014). Click here for the new sales and support site.

Problems of Unicode | Alternative Methods of Keyboarding Unicode | Characters in the font New Athena Unicode | Private Use Area Code Points in APA Unicode Fonts

Problems of Unicode

The Problem of Precomposed vs. Decomposed

Ancient Greek as used by scholars in modern printing and word-processing employs not only the plain letters of the Greek alphabet, but also letters that have been modified by diacritic or editorial marks of various kinds, such a breathing signs, accents, diaeresis, macron or breve symbol, and sublinear dot. Theoretically, if you want an omega with rough breathing, this is represented by two characters: U+03c9 (lowercase omega) and U+0314 (combining rough breathing). A combining character has no width of its own, but places itself in the same space as the previously entered character. Similarly, if you want a dotted lowercase beta, then this needs to be entered as two codepoints, U+03b2 (the beta) and U+0323 (the combining dot below diacritic). Or, if you want alpha with macron and smooth breathing and acute accent, this entails four Unicode characters: U+03b1 (alpha), U+0304 (combining macron), U+0313 (combining smooth breathing), U+0301 (combining acute accent). All of these are examples of composition of a complex character from several code points; such input is called decomposed. If you actually enter these codepoints with many fonts and applications, the result may be unattractive: the breathing might not be centered over the omega; the dot might not be centered under the beta or might intersect the descender; the diacritics over the alpha might overlap or be stacked vertically; in some fonts a diacritics may even appear incorrectly in its own space after the letter.

The solution to this difficulty is the definition of precomposed complex characters in Unicode, as an interim measure until decomposed representation is better supported. Unfortunately, only the first example, omega with rough breathing, is recognized in the Unicode standard: the code point U+1f61 supplies this precomposed character. For the other two, the choice is between assigning a Private Use Area (PUA) code point to the precomposed character (which entails that there may be conflicts in the code point selected and that most fonts will lack that character anyway) or trying to work with the decomposed input. Currently, maximum cross-platform compatibility and compatibility with users who do not have the very latest software suggests the continued use of PUA code points. But decomposed input can provide good-looking input if used with a modern font that contains additional information about "ligatures" (the substitution of a single well-designed glyph for the combination of two or more code points) and the needed precomposed glyphs.

Because of the differences between Windows and Mac OS X, it has not been possible to make the keyboarding of the more specialized characters completely consistent. Nevertheless, if a Windows user follows the deadkey method with GreekKeys 2008, the input will be identical to what is produced by a Mac user for all the routine polytonic Greek combinations. Only the more complex combinations involving macron and breve plus other diacritics pose problems, since they cannot be handled in the same way in both keyboards. With Windows keyboard, alpha with macron and smooth and acute has to be input as a series of decomposed code points (if you use a font with the needed OpenType feature definitions, such as are found in GreekKeys 2008 Unicode fonts). In Mac OS X, the same combination is entered as a PUA code point (unless you make use of the experimental decomposed input instead of the regular GreekKeys Unicode input). If you are going to work cross-platform with specialized characters, please read Appendix III about compatibility issues and workarounds.

The Problem of Duplicate Code Points

Because at an early stage the Unicode standard provided only for monotonic Greek and the characters for polytonic Greek were added later, the current standard contains duplication of a number of characters on the now-abandoned assumption that the tonos printed on modern Greek vowels is distinct from the acute accent printed on ancient Greek vowels. (There have in fact been some fonts that make these two accents appear slightly differently.) This is now a deprecated distinction, and it is recommended that the duplicate characters be unified under the code point in the Greek block and that the duplicate code points in Greek Extended not be used. When consistency of typography really matters, it is advisable to avoid the use for Greek of fonts in which there is a visible difference; for instance, an older version of Apple's Lucida Grande had a significant difference between tonos and acute, the version in 10.4 still had a slight difference, but in 10.5 there is no difference. Many older Microsoft fonts show a difference between tonos and acute, but newer versions do not. Similarly, punctuation symbols have been unified, so that use of the specifically Greek middle dot and Greek question mark are now deprecated, with the standard code points in the first two blocks of Unicode preferred. In the GreekKeys fonts, the glyphs used for these two code points have always been identical in appearance. The GreekKeys Unicode inputs in GreekKeys 2005 and 2008 both use the recommended codepoints for the accented letters and the semicolon/question mark, but the Greek colon/middle dot was still input as U+0387 in GreekKeys 2005 while it is input as U+00b7 in GreekKeys 2008. Again, the glyphs have the same appearance.

Alternative Methods of Keyboarding Unicode

Having a font with the needed characters encoded as Unicode is only half of the battle for users. The other half is having a convenient means to type the needed characters, one that will operate at a systemwide level and thus be available in all modern applications. GreekKeys 2008 provides such a means for Windows as well as for Mac OS X (whereas GreekKeys 2005 was for Mac only). This is an advance, but there will still be some areas of cross-platform discrepancies that the user of specialized characters will need to be aware of, and there may be many code points of interest that are not accessible on either the regular GreekKeys Unicode keyboard or the GreekKeys Unicode symbol keyboards. Thus, it is important to know about the other resources built into the OSes and applications for dealing with Unicode code points not provided for by standard keyboards.

Additional Ways to Input Unicode in Windows

(1) In MS Word (from at least the 2002 version forward) you can enter any character for which you know the Unicode hex code point by entering the four or five hexadecimal digits and then pressing ALT-x. The digits will then convert to the character you want. Of course, you need to know the code point to do this.

(2) There is a System Tool (Programs:Accessories:System Tools) called Character Map. This is a small application displaying the characters in a font in Unicode code point order. First set the font, and then select a character, and click the Copy button so that you can paste it into a document of your choice. If you know the code point, you can enable Advanced View and enter the code point to locate the character in the table. Character Map has limitations: SMP (five-digit) code points are not available for selection under either XP or Vista; under XP this utility shows only the BMP (four-digit) code points that were present in Unicode 4.0, and ignores Unicode 4.1 and 5.0 and pipeline characters that may actually be present in the font; under Vista, however, it shows all BMP code points present in the selected font, including pipeline characters.

(3) Both Word and OpenOffice Writer have menu commands for inserting unusual characters. OpenOffice has the superior functionality (tested January 2008 with version 2.3.1). The command is under the Insert menu and is called Special Characters…. Set the font to New Athena Unicode. The table of characters in OpenOffice will then show all the BMP and SMP code points present in the font. Click on as many characters as you want and the string of characters will be shown at the bottom of the window; when you click OK, the string is inserted in the document and the Special Characters window closes. [This is not consistent: on one of the two XP machines I tested the window showed the presence of SMP code points but only with empty squares in place of the actual characters, and if these are inserted in the document, the squares cannot be changed by changing the font. With Vista there is perhaps no problem.]

In Word the command is Symbol… under the Insert menu. With Word 2003 under XP, this command has the same limitation to the ranges of BMP characters present in Unicode 4.0 as the Windows Character Map (but under Vista this will show later additions that are within those 4.0 blocks, such as archaic heta). Thus, Coptic, for instance, is not visible. There is, however, a field in which you can enter a BMP code point for the characters that are not displayed, and it will then temporarily show the character in the first position in the table (not in its Unicode order), and it is thus possible to insert it into the document. In Word 2003 the Insert Symbol… command cannot deal with SMP characters. In Word 2007, however, Insert Symbol… does show all the characters present in the font, at least on one computer running XP. Because of the inconsistency of various versions of Windows and Word, some users may encounter results different from those described here.

Additional Ways to Input Unicode in Mac OS X

The Character Palette can be activated in System Preferences: International: Input Menu. (Or for 10.6 and later, the name has been changed to Character Viewer, and the location for activation also has new terminology: System Preference: Languages & Text: Input Sources.) If you know the Unicode code point for a character or symbol you want that is not directly accessible from the GreekKeys Unicode input, you can locate it there by searching for that string and (in most cases) insert it in a document. You can also search by part of the character's name, if you know how it is described in Unicode. If you do not know the name or code point of the character, bu tonly its appearance.you can scan the codes tables, but this may be quite inefficient.

Another way to enter a character when you know the Unicode code point is to use the Unicode Hex input that comes with OS X. Activate this input in System Preferences: International: Input Menu. To use it in a document, change to Unicode Hex in the input menu, and then hold down the option key while typing the four hexadecimal digits of any BMP code point. (You cannot enter SMP (five-digit) code points with Unicode Hex.) When you do not use the option key, Unicode Hex behaves like a US English input.

Characters in the font New Athena Unicode

For a complete list of characters and code points in New Athena Unicode font, see the document NAUcmap3_2.pdf in the folder Unicode Fonts of the GreekKeys 2008 package. Or go to the font download page to obtain the text version of this document corresponding to the latest revision of the font.

Private Use Area Code Points in APA Unicode Fonts

A consortium of scholarly font producers who work with Unicode polytonic Greek has agreed to use the same Private Use Area (PUA) code points to refer to special characters that are not recognized by Unicode (and most of which are not likely ever to receive official Unicode code points). See the proposal document written by Juan-José Marcos that is downloadable from the GreekKeys support site. In addition to the APA Unicode fonts, this convention is followed in the fonts of Ralph Hancock’s Antioch package, in Juan-José Marcos’ ALPHABETUM, and in David Perry’s Cardo. Because of this agreement, users should be able to shift between the various fonts that comply with this convention. System fonts and commercial fonts will not contain these characters and may contain conflicting characters that use the same PUA code points. Users should be aware that documents containing these PUA characters may not be completely readable for other users who lack a necessary font or for publishers.

A paper by Juan-José Marcos setting forth the proposal for this agreement is available here with his permission: PUA_coordination_usage.pdf.

Here are two tables, the first with the characters grouped alphabetically by base character, the second with the same data ordered instead by PUA code point. If the character name is capitalized, the reference is to the capital letter.

By code point:

Other PUA code points (some of which are no longer needed, since the character has now been accepted in the Unicode standard):

Additional PUA code points used in New Athena Unicode for the alternative forms of capital vowels with iota and for some metrical symbols, and to support transliteration of Demotic Egyptian (these have not be agreed upon by the consortium that recognizes the code points in the previous lists):

Additional ligature combinations for unusual combinations needed in manuscript descriptions, including one convention of cryptographic writing:

GreekKeys Support by Donald Mastronarde, © American Philological Association 2008.