From: Jukka K. Korpela (firstname.lastname@example.org)
Date: Sun Nov 04 2007 - 02:35:55 CST
James Kass wrote:
> There is a difference between my example and a "true ligature", but
> I wanted to call attention to the similarity. Making a typographic
> ligature of two letters and then employing it as a logo does not alter
> the fact that it is, essentially, a typographic ligature.
There is still a difference between typographic ligation, which is generally
applied without considering the meaning and context (e.g., representing "fi"
as a ligature no matter where it appears in the text) but with due
consideration of fonts*), and an expressive ligature, which is used for
particular reasons related to meaning and context, such as a logo ligature.
*) In some fonts, the "fi" combination needs no ligation, and ligation might
even be quite improper (e.g., when monospace font is used). In some fonts,
it may be ligated without any visible effect (i.e., the fi ligature is
identical in shape to the two letters "f" and "i" in succession). And in
some fonts, ligation changes the shape and might be typographically
Typographic ligation is generally not a character-level issue at all.
Expressive ligatures may be, and and they may even be coded as separate
characters (though company logos won't, as a policy matter).
> It was noted on this list several years ago (if I recall correctly)
> that "Encyclopædia Britannica" always expressed its own name using the
> a-e ligature, but was happy to refer to other encyclopedias without
> the ligature. Although Unicode has a dedicated character for the
> a-e ligature, there should be nothing conceptually wrong with
> storing or exchanging that title as "Encyclopaedia Britannica"
> (using: a, ZWJ, e).
Unicode has a dedicated character for LATIN SMALL LETTER AE.
It's not just the name; the Unicode name is, as we know, in principle just
an identifier, and some Unicode names are outright misleading if read as
English text. It's the defined properties that matter. What matters here is
the decompositions, namely the absence of any: there is no canonical or
compatibility decomposition for LATIN SMALL LETTER AE.
Thus, by changing "æ" to "a", ZWJ, "e" you change the identity of the data
as characters, just as you would by changing "w" to "v", ZWJ, "v". Some
characters (including "w", "ñ", and "ß") have _originated_ as ligatures but
turned into characters.
With "æ", there's the particular problem that in some languages, it is
definitely a separate character that is not decomposable into anything,
whereas in some usage, mostly in writing words of Latin origin, it is a
ligature of "a" and "e". In the latter role, it's somewhere between a
typographic ligature and an expressive ligature. When encoding data, you
need to make up your mind and write it either as LATIN SMALL LETTER AE
(thinking that this may, in addition to other use, be used to represent the
ligature) or a "a" followed by "e".
Using ZWJ between the letters in the latter choice is more or less
illusionary. You would be asking for typographic ligature (or cursive
joining) in general, not the established form of LATIN SMALL LETTER AE in
particular. Your request might well be denied, because implementations need
not perform ligation, or they might implement it for some character
> If ZWJ requests a more joined form of two characters if
> the system can provide one, and the user desires to represent
> the AP SYMBOL in plain text, and the user inserts a ZWJ
> between "A" and "P", and the system can provide a more
> joined form of A+P, and that more joined form happens
> to match the desired appearance, is there a problem?
There need not be any problem. The ZWJ character is meant to be used for
suggestions on ligation in exceptional situations where ligation cannot be
handled at some other level.
But ZWJ just asks for ligation or cursive joining, not any _particular_ kind
of ligation. Thus, you might end up with something that is clearly not just
the letters "AP" in normal presentation and clearly not the commonly used AP
Jukka K. Korpela ("Yucca")
This archive was generated by hypermail 2.1.5 : Sun Nov 04 2007 - 02:36:26 CST