Guide to Unicode Greek
All authors publishing with Dumbarton Oaks are required to submit text that is Unicode compliant. This policy applies to the submission of any text in any language, not simply polytonic Greek. For some of our authors, particularly those who work only in western European languages, this poses no problem, since the Latin alphabet has been de facto Unicode compliant for decades. For those of you who use Greek, however, the prospect of creating Unicode-compliant Greek may be daunting. You may ask, why convert, especially if a particular font has suited your needs for some time. Given the frequency with which computing changes, it is sensible to wonder how long this standard will endure, and how complicated it is to configure your computer.
This guide is intended to answer these questions, by explaining the importance and benefit of the standard, and by providing instructions on how to set up your computer to be Unicode compliant. Only the most essential information about Unicode Greek is presented here. Suggestions for further reading can be found at the end of this guide.
What is Unicode Greek?
To understand the need for Unicode Greek it helps to understand something of the development of codes and the telegraph. The earliest telegraphs transmitted and received electronic pulses, which were transcribed onto long paper strips. As the voltage changed, either a pen wiggled across the paper, or a set of needles or stamps punctured or indented it, creating a visual pattern representing the message. (The audio component-the clicks and clacks we associate with telegraphs-was a later development.) To interpret the message it was necessary that both sender and recipient use the same code, listing the assignments of pulses to letters or numbers. These code tables were the prototypes of the code tables used in computers, to convert sequences of electrical impulses into various assigned characters and symbols.
The initial codes consisted of a limited set of characters. International Morse Code, for instance, has only fifty-one characters: the twenty-six letters of the English alphabet (assumed to be uppercase), the ten digits, and fifteen signs of punctuation. All the earliest telegraph codes possessed a limited number of characters. The smaller the character set the more cost effective it was, since smaller sets reduced the number of dots and dashes, and therefore the time, needed to encode, transmit, and decode a message.
As technology developed so did communication needs, and the telegraph prototypes gave way to more expansive code tables that included both upper and lowercase letters, as well as commands, such as "carriage return" or "end of transmission," meant to instruct the receiving device about the format or shape of the transmission. At first there was little uniformity in the development of new codes. The number and variety of coding systems developed over the first century of telecommunications compelled the International Organization of Standards (ISO) to develop a single standard for coding telecommunications. Thus was born in the 1960s the American Standard Code for Information Interchange (ASCII), a 128-character code that included upper and lowercase letters, the digits, standard punctuation, and commonly used command codes. (The number 128 is significant for its compatibility with binary systems, because it is 27.)
Since computers work on a principle similar to telegraphs, it was natural for the earliest computers to take advantage of these standardized character sets. Consequently IBM and Apple each made ASCII the basis for the character sets of their new computers. These new computers, being binary, transmitted information in packets of 8 (23) bits, so 256 (28) was the natural size for its code table. Since the 128-character set excluded a number of other letters, symbols, and commands that could and should be represented, both IBM and Apple developed a 256-character set. But, because the two systems were developed independently, the upper 128 characters did not correspond to each other. Thus, the initial attempts to expand the ASCII character set were marked by inconsistency.
The character sets computers used through the mid-1990s catered almost exclusively to the Latin alphabet. Both PC and Macintosh upper character sets (spaces 128–255) assigned some slots to Greek letters, but these were intended to serve, not speakers of modern Greek or scholars dealing with ancient and Byzantine Greek texts, but mathematicians who needed to write equations. Thus, throughout the 1980s and 1990s anyone who wanted to use computers to work with Greek or other alphabets had to invent creative ways of getting around the Latin alphabet.
Most often, the way to circumvent the problem was simply to change the font. By all appearances, this seemed to imitate what had been done in printing in the past. In handset or letterpress printing, several languages in a single font-here, literally a set of lead blocks or molds that were inked and pressed onto the paper-could be mixed at will. In computers, these physical blocks are replaced, not by the computer's printer, as might be expected, but by something immaterial: computer code. This code is a description of a typeface, a way of presenting a particular form of a letter, number, or other character on the computer screen or the printed page. The earliest method then for working with Greek text on a computer was to use either the Macintosh or IBM 256-character set, but with a specially created Greek font. In this technique, when Latin letters were typed, what appeared was a Greek letter the font designer assigned to that place in the character set. For instance, by pressing the ell key, a lambda would appear. The underlying data was still an English letter (in this case, ell), but it looked Greek (that is, like a lambda).
This technique, although still used widely today, has for several reasons been less than satisfactory.
First, the assignment of the various letters of the Latin alphabet to that of the Greek has been capricious. For example, some fonts assign to chi the x, others, the c. Some fonts use the left parenthesis for a rough breathing, others, the J. Some fonts assign precomposed combinations of vowels and their diacritical marks to arbitrary, hard-to-remember places in the upper character set. Other fonts split vowels and their individual accents. Other examples are legion. In the end, every font of this sort requires the user to learn a new keyboard configuration.
Second, like other types of intellectual property, fonts are copyrighted and cannot be shared without the designer's approval. Many of the best fonts are expensive, a reflection of how much time and effort goes into creating a beautiful font. Anyone who wishes to share a Greek text with someone else requires the recipient to have the same font. The recipient, then, is placed in the awkward position of having either to break the law or buy the font.
Third, it is difficult or impossible, because of accents and breathings, to search through text using conventional word processors or internet browsers. If you are searching for ἤδη (ēdē) do you enter the key combination hdh, hjvdh, hvjdh, h[dh, ≥dh, or some other combination? You must know exactly how that particular font's architecture works, and hope that the word was keyed in correctly. That same search routine will probably not work on other fonts.
Fourth, even the most beautifully designed Greek font might produce shoddy typography. For instance, a rough breathing may be correctly centered over a lowercase eta, but that same breathing over an iota will be off center. An iota subscript properly centered under an omega will not be correctly positioned under an eta, which takes the iota subscript under its left leg. For many scholars such finesse has not been a concern. But for publishers who value their craft, this has contributed to erosion of standards in typography and the art of bookmaking.
Fifth, some fonts are incomplete in their repertoire of extra nonalphabetic characters. For instance, a font may have neither an obelisk nor the Greek numeral for six. This poses a problem for many scholars who deal with Greek texts. If the primary font does not have certain characters, then the editor is forced to use two or more Greek fonts, producing the potential for confusion, and awkward or ugly typography.
Sixth, exchanging information between Apple or Macintosh and IBM-compatible PC systems proves an insoluble problem. Even if both users have copies of the same font, one for the PC and the other for the Macintosh, this does not guarantee seamless transmission. If the font assigns important glyphs to the upper 128 characters, then those glyphs will be lost in transmission, since PC and Macintosh have completely different assignments to the upper character set.
These are very serious problems. And for every challenge Greek poses, there are dozens more in other alphabets and languages, such as Chinese, Arabic, and Tibetan.
Acknowledging all these problems, computer developers at Xerox and Apple in the late 1980s began to work on solutions, beginning with efforts to standardize Han Chinese. Other corporations, such as Sun, Adobe, Microsoft, and IBM, joined the effort to develop a common standard and in 1991 founded the Unicode Consortium, the nonprofit organization that supports and develops the Unicode Standard. Their intent was and is simple, to develop a character code that addresses every need in every language, a code that is universal, uniform, and unique (hence the name).
Unicode is universal in that it addresses the computing needs of all the world's languages. The initial versions of Unicode had a character map consisting of 65,536 (216; hence called a 16-bit code) points, but this was later enlarged to 1,114,112 (232; i.e., 32-bit code) points, to accommodate all possible scripts, living and historical. Under this plan each and every character in the world's writing systems can be assigned a unique, unambiguous code point.
Furthermore, Unicode is efficient, in that it has built into its structure a set of equivalencies. That is, it establishes rules that tell the computer, for instance, that the two keystrokes alpha and rough breathing is the same as a single keystroke containing the precomposed glyph ἁ. This allows some flexibility in the way Greek text is entered into the computer; the result is the same. Thus, any software that takes full advantage of Unicode will allow you to search for Greek text more easily and accurately than older software did with non-Unicode Greek fonts.
There are a number of other advantages and caveats to Unicode that require a more technical explanation, beyond the immediate scope of this presentation. In sum, Unicode makes possible for Greek texts beautiful typography, a complete set of characters, accurate searching, ease of typing, and a seamless exchange between users without any loss of information.
Will Unicode become obsolete any time soon? No. This question is asked mainly because it seems that the computing industry is abandoning one standard for another. But this is not the case. Unicode has not rendered the older system, ASCII, obsolete. ASCII is still an industry standard, but now as a proper subset of Unicode. What is being rendered obsolete is the incorrect use of ASCII. Scholars have been trying to use ASCII to do things for which it was never designed. Unicode has been designed specifically to address the needs of those who have had to make ASCII do what it was not intended to do.
In the future Unicode will expand (it is currently at version 5.1), but it will preserve every previous standard. This growth involves the inclusion of new linguistic blocks, as the Unicode Consortium attempts to develop Unicode to serve the needs of the entire world. It will be important for some Byzantinists to keep track of these developments, since occasionally new symbols are introduced for inclusion in Unicode, in Greek and in other languages.
Preparing your system to handle Unicode Greek
Unicode is merely a code, a conceptual infrastructure. To take advantage of it, you must ensure that the hardware and software you are running is Unicode compliant. Hardware normally is not a problem. Any Macintosh or IBM-compatible computer that runs Unicode-compliant software is itself Unicode compliant. The software is the most critical aspect. There are four different areas that must be attended to: the operating system, the fonts, the keyboard driver, and the word processor.
Operating system. The operating system is the software your computer initially installs, making it possible for you to run other software. Macintosh users must be running OS X. Previous versions (OS 9 and below) will not work. PC users must run Linux or Windows CE, NT 4.0, 2000, XP, or Vista. Many older operating systems, such as Windows 98, do not work. If you have an old computer that cannot run the latest software, then it is time for you to upgrade to a new system.
Fonts. As already mentioned, fonts are sets of instructions for the proper display or printout of a particular typeface. Different fonts are designed with different numbers of characters. Unicode makes it possible (but not necessary) for a font to have both the Latin alphabet and all the polytonic Greek characters. If you are running Windows XP/2000 or Macintosh's OS X, you already have Unicode-compliant fonts that reflect this capability: Lucida Sans, Palatino Linotype, Tahoma, and Arial Unicode MS (all PC) as well as Lucida Grande (Mac). The most recent versions of Macintosh's operating system, OS 10.4–5, includes two new preinstalled fonts with polytonic Greek characters: Helvetica and Times.
There are many other Unicode-compliant Greek fonts available, many for free. In our publications we have oftentimes used Kadmos (now renamed KadmosU), which is now distributed by the American Philological Association as part of GreekKeys 2008. This same package includes other beautiful polytonic Unicode Greek fonts, including New Athena Unicode, which is available for free download. At the end of this document are links that will let you obtain many other Unicode Greek fonts.
In the past, Macintosh and PC fonts could not be interchanged. With the latest operating systems this difficulty has been overcome. Any font that works on a Mac (OS X) will now work on a PC (Windows XP and higher), and vice versa, thanks to a new standard of font called OpenType, which is completely Unicode compliant. If you are working with an OpenType font you are using Unicode (provided that the font was competently designed). If you are using a TrueType font or PostScript font, you might not be working with a Unicode-compliant font. If you are uncertain if the font you are using is Unicode compliant and want to check, contact the publications department.
Keyboard driver. Although Unicode has expanded its character set to over one million characters, our keyboards still have little more than a hundred keys. In the days of the 256-character set, this was convenient, since every character was only one or two keystrokes away. (For instance, to get a capital A, you must hold the shift key down while pressing a.) In the age of Unicode, where dozens of different languages have their own text blocks, and some languages, such as Han Chinese, have tens of thousands of individual characters, the use of the keyboard is not as straightforward.
Every computer uses a keyboard driver to interpret the keys being typed. Keyboard drivers are built into the operating system, and are selected by the user to indicate to the computer what language block should be typed, and even what physical keyboard is being used. (German keyboards, for instance, switch the y and the z, from an Anglophone perspective; other national keyboards, such as Russian, bear little resemblance to the Qwerty arrangement most familiar to many computer users.)
Macintosh OS X includes dozens of possible keyboard drivers as a standard part of the operating system. Unfortunately, prior to OS 10.4 (Tiger), polytonic Greek was not one of the options. For those not yet running Tiger (and those who run it, but are unhappy with the configuration of its polytonic Greek keyboard), third-party software is required to tell the computer that you want to type in polytonic Greek. This same software provides its own method of what keystrokes correspond to what letters, accents, or breathing marks. We recommend SophoKeys, a free keyboard driver. The instructions on how to download and install this file are provided here.
For Windows XP and Vista, there is a Polytonic Greek keyboard driver already built into the operating system, but you must activate it yourself. There are very detailed instructions with pictures provided by Microsoft here. Unfortunately, this driver's assignment of letters and diacriticals is difficult for many people to memorize and use. You may find this free keyboard driver easier and more intuitive. Other Windows keyboards are listed here. One other alternative is MultiKey. (Bear in mind, MultiKey is not a true universal keyboard driver, since it works only for Microsoft Word and Classical Text Editor.)
One of the benefits of using some of these keyboards is that they are context sensitive. That is, the drivers are able to convert key combinations into single glyphs on the fly, and determine what sort of shape a character should take based on position. For instance, an alpha will change its shape as you continue to add diacriticals to it, but still consist of only one character. Also, if you end a word with a sigma, then press the space bar, the closed sigma automatically turns into a final sigma.
Bear in mind that the keyboard driver is a means to an end, and as long as the end result is Unicode-compliant polytonic Greek, it does not matter what method you use. Experiment with different keyboard drivers, and use the one you find most comfortable.
You may, in fact, find a visual keyboard more to your liking. Try the Character Map (preinstalled on Windows-based machines) or the Character Viewer (preinstalled on Macintosh). Alternatively, try the HTML-based keyboard for Classical Greek.
Word processors. Not all word processors are Unicode-compliant, and even those that are, may not support the entire range of Unicode. There are only a few word processors that take full advantage of Unicode. WordPerfect is not at all compatible with Unicode, and we strongly discourage its use. Fortunately, one of the most widely used word processors, Microsoft Word (Mac or PC versions published after 1996), is fully Unicode compliant. We recommend authors use Microsoft Word, Open Office, or one of the programs listed here.
After you attend to the four components listed above your system should be capable of handling polytonic, Unicode-compatible Greek. Be sure to read carefully the documentation that comes with the software you have chosen to use, to familiarize yourself with its conventions and requirements.
Converting text to Unicode Greek
Many authors have already committed large amounts of Greek to a font that is not Unicode compliant. Oftentimes this text needs to be submitted for publication at Dumbarton Oaks. Does this Greek need to be retyped?
No. There are resources that allow you to convert your preexisting Greek into a Unicode-compliant format. You may be able to make this conversion on your own. There is a website that provides conversion of a few fonts. There is a program that facilitates other kinds of conversion for Mac OS). Some utilities for the PC are listed here.
If none of these resources work, do not despair. There probably is a way to make the conversion. Contact the publications office, present a sample of your text, and we will suggest a way to handle the material.
For further reading
If, at this point, you wish to learn more, or investigate any of the issues above more in depth, here are several of the many resources available on the internet:
- Introduction to Unicode Greek, written for Biblical scholars; has many useful links and explanations applicable to classicists and Byzantinists.
- Extensive scholarly articles regarding the implementation of Greek Unicode.
- Official site for Greekkeys. Through this site new forms of various Unicode fonts such as KadmosU, BosporosU, and AtticaU are distributed. KadmosU is noteworthy for its beauty. These fonts are especially helpful for Copticists, since they include the recently standardized Coptic text block.
- Discussions of Unicode issues for Macintosh users.
- List of some of the Unicode Greek fonts that are available
- Very good explanation of Greek Unicode; the free Unicode font distributed here incorporates Greek, Hebrew, and Latin.
- Extensive discussion about Unicode and Greek, addressing the intersection of Unicode and specific points of Greek philology, such as the two types of qoppa.
- A long page listing available Unicode Greek fonts. See the subpage on how to configure your web browser to view Unicode Greek on the internet.
- The official website of the Unicode Consortium, with a plethora of information about the Unicode standard. Of interest to Byzantinists will be their section on Greek.