Re: [css3-text] Greek and UPPERCASE

Lea Verou wrote:

> In greek, we have accent marks that denote in which syllable you put the 
> stress on when pronnouncing a word. For example, in the word καλημέρα, 
> the accent mark is on the 6th character. However, when words are written 
> in all UPPERCASE, those accent marks are removed (for example, ΚΑΛΗΜΕΡΑ 
> for the same word, not ΚΑΛΗΜΈΡΑ). The only case where an accent mark is 
> included in a capital letter, is when it's the first letter of the word 
> (e.g. when it's a name or it's at the beginning of a sentence). This 
> rule is not followed by the aforementioned CSS rules, as they just use 
> the corresponding uppercase unicode character, making those declarations 
> completely unusable in greek websites.

This is something that may be handled at the glyph level, using 
contextual substitution lookups in the font. The rule is not quite as 
simple as you describe: if an accent falls on the first letter of a two 
vowel sequence, indicating that this is not pronounced as a diphthong 
but as two separate vowel sounds, then when the accent is removed in 
all-cap setting a dialytika (diaeresis) must be added to the second 
vowel to prevent confusion with the diphthong.

Handling this at the glyph display level, rather than in character 
transformation, has some benefits. Firstly, it is a convention that came 
about after the practice of writing accents and breathing marks to the 
left of uppercase letters developed, and I have yet to find examples of 
that practice prior to the 19th Century. This means that although it is 
the norm for all modern Greek typography, it was not the norm for most 
of the history of accented Greek text, in which accents were frequently 
written above Greek uppercase letters (or even to the right!). There 
also exist examples of smallcap Greek letters used as transcription of 
uncial texts, again with marks above the letters rather than suppressed. 
These variant typographic treatments are possible if the handling of 
marks in all caps and smallcaps is made at the glyph level, e.g. using 
the OpenType <calt> feature, which users can then apply or disable as 
they wish. Also, I am not sure that a case transformation that strips 
marks at the character level is losslessly reversible, whereas if it is 
handled at the glyph level the character string is preserved.

JH

Received on Tuesday, 24 May 2011 23:35:41 UTC