Ligatures

 
Home > Greek > Unicode > Stories

It is, of course, against the design principles of Unicode to encode ligatures as separate characters—as I've already expounded with the example of fi. If an exception is made, it is usually going to be because of some backwards compatibility issue. For both the ligatures that have ended up in Greek Unicode, that is not the case: they are additions subsequent to Unicode 2.0.

In that case, one would have to argue that these are not mere ligatures, but autonomous signs. The case for kai is somewhat suspect; for the drachma, it is ludicrously shaky. But we're stuck with them.

1. Kai

U+03D7 Greek Kai Symbol [ϗ], U+03CF Greek Capital Kai Symbol [Ϗ]

In the bad old days—starting with the papyri, going berserk in the late Middle Ages, and not letting up until the 19th century—Greek was written and printed in lettering that was more shorthand than anything else. (In fact, the technical term for many such signs is tachygraphical—meaning shorthand: that's where they originated.) There was a profusion of ligatures and abbreviations, particularly for inflectional endings and common letter pairs, but also for certain frequent words. Greek wasn't different to Latin in this regard; but whereas Western printing, to some extent under the influence of Roman epigraphy, wound back the obsession with ligatures quite early, it took until the French Enlightenment to do so in Greek typography.

Now that it has happened, though, it has happened all the way: mainstream typographical practice in Modern Greece has no ligatures comparable to Latin ff or fi. The stigma ligature died out in the 19th century; OU is the only ligature that survives in some use in Greece today—and that use is marginal.

There were also mediaeval symbols which served as abbreviations for entire words (Thompson 1912:84-85): , amazingly, stand for ἐστί "is", γάρ "because", δέ "but", ὅτι "that", εἶναι "to be". These, thankfully, did not stick around. The one abbreviation glyph which did stick around into the 19th century is that for καί, "and". An abbreviation for the same word has stuck around in various variants of Latin script (U+0026 Ampersand, &; U+204A Tironian Sign Et, ⁊), and it makes sense for it to do so; even if the word isn't that long to begin with, it is extremely frequent (1 out of 17 words in the TLG corpus).

There were three different stages in the development of the glyph. In the papyri, before the abbreviation system went fully berserk, the abbreviation appeared as kappa with an abbreviation prime: κʹ, κ`, κˏ. But it also appeared as a Latin S (Thompson 1912:81), originating in a tachygraphic form of κ, . This is the form that prospered in the middle ages, with the attendant variants: . No use putting those in your font, though: only palaeographers will recognise them. The form that took hold in the Renaissance, and which the Unicode reference glyph uses (relying on Everson's proposal), reverts to the Hellenistic model: it is a kappa with a tail.

In the 20th century, the ligatures OU and kai have been primarily handwritten or painted; one of the main places the kai ligature is still seen (decreasingly) is in shop signage, as informal lettering—whence Everson's notorious reference to shop fronts:

Apart from its frequent appearance in early Greek typography, I recently noticed that this sign is in common use in contemporary Athens. Many, many shops use the character & to denote καί on their shop signs, but a considerable number of them eschew this Latin sign and prefer to use ϗ. As a foreigner, although admittedly one with some typographical knowledge, I found the use of & to look rather odd in Greek text, and was pleased to discover that ϗ was preferred by many. It was only after I had seen it on many public shop signs that I also discovered that ϗ was a standard typographical sign in Greek printing.

What needs to be stressed here is this this usage in modern Greece is informal, and confined to places like shopfronts; it's analogous to the use of Nite in the US. The Latin ampersand, even if we admit its disharmony with Greek text (and that judgement cannot be purely aesthetic), has been in use in Greece at least since the 1920s. (Accordingly, the Greek ligature has more traditional connotations than the ampersand; in fact, the only place I saw the Greek ligature on my 2004 trip to Greece was on recent tombstones.)

Fittingly, one of the few online presences of the symbols is in a transliteration of a Greek cemetery.

Compared to English, the use of either sign is quite rare in Modern Greece, and does not make it to printed documents. In fact, the glyph Everson presents (which reflects Renaissance practice) may give an impression to the contrary, that there is a print tradition of ϗ in Modern Greece. There is not: the sign is usually hand-daubed, and therefore more Sans-Serif than anything else: . (The glyph Everson uses in Everson Mono reflects this.) Whether this means the character belongs in a repertoire intended for print-like use is open to debate.

Since the character has not been available in 8-bit fonts, one will also see κ used in the role, especially in the abbreviation κ Σία "& Co." A search on www.alltheweb.com yields 46 hits for και Σία, 2 for κ Σία—and 890 for & Σία. (Searches in 2005.)

The use of the abbreviation in Early Modern Greek typography is neither here nor there, either; Early Modern Greek is awash in ligatures and abbreviations, and noone would seriously suggest that Unicode needs to represent them as distinct codepoints. The desktop printing package and font that can take a word like εὐγάλωμεν "produce" and spit out (or something like that) will be a prodigious system indeed. (Good thing noone's attempted that for the past two centuries.) But that's not Unicode's concern, and if Unicode is not obligated to replicate the εὐ- and -μεν ligatures in Glytzounis' 1568 arithmetic manual, neither is it obligated to reproduce its καί abbreviation. Everson takes pains to point out that ϗ is an abbreviation and not a ligature; but that in itself does not make ϗ merit a codepoint, any more than the abbreviations for ἐστί, γάρ or ὅτι do.

Everson's proposal goes on to introduce a note of codepoint activism:

I believe that if ϗ were found in coded character sets, Greeks would prefer it to &, and I suspect that the prevalence of the Latin & on Greek signage today is solely an artifact of limited 8-bit technology.

The Latin character turning up on everybody's keyboards would have certainly helped things along, although as I've said the Latin ampersand has been used in Greek for many decades. I will admit, when I found that the Greek kai was included in Unicode, I exclaimed that it was utterly cool; so Everson's activism has had its desired effect on me. It's not clear that there has been much take-up though: a search for all ϗ online yielded 42 hits in 2005—all in discussions about Greek Unicode or as random binary characters. I'd hate to think my online bibliography is the first document on the web to have used them as they were intended, because I have used them there as an act of whimsy. (Even in 2008, usage is quite rare; e.g. in an online publication of a 1656 work (James Harrington, The Oceana and Other Works) with Greek formatted with the ligatures of the time: ϖρὸ της βȣλῆς ϗ ϖρὸ τῆς ἐϰϰλησίας.

That's kappa-symbol, pi-symbol, and letter ou: the quote is reproducing glyphs and not letters, which is naughty.

This character was proposed by Michael Everson in May 1998, and adopted in Unicode 3.0. A capital counterpart was added in Unicode 5.1.0, April 2008; it can be used in principle, although I am not aware of well established usage.

Pretty much the same ligature turns up in Coptic, as U+2CE4 Coptic Symbol Kai, ⳤ.

2. Abbreviation Primes

U+02B9 Modifier Letter Prime [ʹ]; U+02B9 Modifier Letter Double Prime [ʺ]

Already in the first century BC (Thompson 1912:78-79), abbreviations were indicated with an acute or grave spacing accent (which Thompson calls an "abbreviating dash"), after an initial letter; e.g. μ´ = μέν "on the one hand", μ` = μετά "with". Though in mediaeval use the abbreviating prime was overtaken by other devices—primary amongst them the slash—the prime kept being used, especially in ligatures that served as abbreviations of whole words. As a result, it was usual in Renaissance usage of the kai ligature, which usually surfaced as ϗʹ or ϗ̀. (Why Everson's proposal of ϗ̀ as a distinct codepoint from ϗ was rejected by Unicode is an exercise left to the reader; the answer is right on your screen.)

The abbreviations that the prime marked are gone, but there is one remnant of it in the abbreviation of the notoriously long Greek surnames. The surname prefix Χατζη- in particular can be abbreviated as Χʺ; so Hatzipanagiotou goes from Χατζηπαναγιώτου to Χʺπαναγιώτου. This device is not seen much; it appears in older handwritten documents, and rarely on TV credits, where space is at a premium.

3. Drachma

U+20AF Drachma Sign [₯]

To my memory, the symbol for the now defunct drachma has never been anything but the abbreviation δρ. or δρχ. It is of course already possible to represent the abbreviation δρχ in Unicode using existing Unicode codepoints. In that regard, the drachma sign is unlike the other 17 currency signs in Unicode 4.0—although some glyph realisations of U+20A3 French Franc Sign, ₣, and U+20A7 Peseta Sign, ₧, are also squashed up abbreviations, and others can be realised readily with overstrike glyphs. Even if the cursive form of the glyph was current in the 19th century, it could still be composed straightforwardly with a simple font switch. So one might wonder why the codepoint was adopted.

The answer is that this was an ELOT idea, and when ELOT wants something, Unicode is obliged to comply. This is the sum total of the justification given:

The creation of the EURO SIGN has necessitated the creation of a unique DRACHMA SIGN for use in banking, administration, and for general purposes in Greece and countries trading with Greece both inside and outside the European Union, especially during the transitional period when both the drachma and the euro are in use.

And when asked whether the characters already exist:

The glyph looks like script capital DELTA and small RHO but the symbol is intended for unitary use in collocation with the EURO SIGN.

Allow me to paraphrase this uncharitably. For over 150 years, the drachma is written as an abbreviation, with a delta and a rho and usually a chi. Two years before the drachma ceases to exist, ELOT decides that if the Euro gets to have a single glyph, so should the drachma—even though that single glyph is a delta followed by a rho, and it has a graphical form that has not been seen for at least 50 years, and possibly ever. Why it is so pressing that the drachma has a single glyph now that it would be shown next to € is never made clear. National pride? DM remained good enough for the Germans. Sorting? Surely that's an issue for spreadsheet implementers, not Unicode. Visual display? The two character saving doesn't solve the problem of labels now having two prices on them instead of one. And the whole shebang gets adopted into Unicode, where it will reside as a codepoint for centuries, in September 1999—15 months before the drachma is abolished, and the rationale for the existence of the codepoint (coocurrence with the Euro) ceases to apply.

Riiight.

To be fair, as Alexandros Diamantidis reminds me, some price tags did print delta-rho in a single space, which might be counted as a ligature—with the delta either above or to the top left of the rho:

Of course, this is still a ligature and not a single character; but if we're going to have the codepoint, the price tag ligature has the advantage of having existed within living memory.

This character was proposed by ELOT in January 1999, and adopted in Unicode 3.0.

4. Rho With Stroke

U+03FC Greek Rho With Stroke Symbol [ϼ]

Among the signs proposed by the TLG in November 2003, and accepted in Unicode 4.1, March 2005, is rho with stroke. The letter is used in abbreviations ending in rho -- most prominent in the abbreviation γϼ, for the weight unit γράμμα. It is also used by Sextus Julius Africanus as a symbol for 50 mnae.

Like the abbreviating prime, the slash through a letter was a common way of indicating abbreviation; it can also be seen in various other Greek signs for measurements adopted in Unicode. There is a case to be made that rho slash is equivalent to U+03C1 Greek Small Letter Rho U+0337 Combining Short Solidus Overlay, ρ̷.

Nick Nicholas, opoudjis [AT] optusnet . com . au
Created: 2003-09-14; Last revision: 2005-04-09
URL: http://www.opoudjis.net/unicode/ligatures.html