Re: Ancient Greek (symbols versus letters and duplicate letters)

From: Nick Nicholas (
Date: Sun Apr 13 2003 - 11:27:10 EDT

  • Next message: Chris Pratley: "RE: Variant Glyph Display"

    This is way too late, and Ken's already pretty much addressed it, but I
    thought I might as well chime in.

    Clearly the inventory of codepoints in Unicode is intended to be emic,
    so the diversity of glyph variants of antiquity are not proper to it.
    Where some typographical tradition does make a consistent
    differentiation between two glyphs, though, they are candidates for
    separate codepoints.

    While normal sigma and lunate sigma are indeed mere glyph variants in
    most contexts --- the lunate being early mediaeval, and not much seen
    in modern typography --- there is a context where it is used routinely:
    editions of old texts where the editor is being deliberately agnostic
    about whether the sigma is final or not. Often, the editor merely
    carries that through as a stylistic choice to the rest of the text; but
    given that in antiquity text was written without spaces, there is a
    semantic loading to the choice of non-final and final sigma which the
    lunate obviates. (e.g. if you're publishing a papyrus, you often put on
    facing pages what the papyrus actually has --- with lunates, since the
    papyrus give no indication of word breaks, so it would be presumptuous
    to put in medials and finals --- and what you think it says --- again
    with lunates, so that the two facing pages look consistent, if nothing
    else.) So the lunate sigma is distinct in modern practice and necessary.

    The lunate capital becomes necessary for completeness and consistency
    (all-caps rendering of ancient texts, say). It might be possible to
    insist that those editors using lunate capital in title case are merely
    using an affectation, and that lunate should be the same codepoint as
    the capital normal sigma. But I think that makes matters much more
    complicated than they need be.

    As I describe on , while there
    are cases in typographical traditions where the form of sigma used is
    not predictable (medial word-finally, final word-medially), they are
    almost all deviant and rare enough that they can be ignored. So in most
    cases, even the medial/final distinction is unnecessary (just as the
    medial/final beta distinction would be --- if anyone still observed
    it...) The one exception is abbreviation, because period is
    semantically overloaded.

    This is not to say that all that has been included in Unicode has been
    included well; Haralambous in his '99 paper doubts that mathematicians
    could use upsilon-hook or cursive kappa with semantic differentiation
    from their normal forms, and I doubt it too. But that's clearly a
    legacy issue now, and nothing much can be done about it, even if the
    variants end up not being used distinctly by mathematicians.

    The story with the koppa is that the lightning-bolt koppa is used in
    Modern Greek as a numeral not only in legislation, but pretty much
    anywhere English would use a Roman numeral -- notably in page numbering
    of introductions, and enumeration of points. So the context one would
    likeliest find both numeric and q-like koppa would be a Modern Greek
    scholarly work on Archaic Greek inscriptions. Not extremely frequent,
    but it can happen. And it is true that most Greeks would have no idea
    that the lightning-bolt koppa is the same character as the q-like
    koppa, since outside of at most a mumbled sentence in the Classical
    stream at high school, most will have never seen a q-like koppa at all.

    Thanks to everyone for saying all this before me. :-)

    Life                Dr Nick Nicholas, Dept of French & Italian Studies
    Is a knife          University of Melbourne, Australia
    Whose wife
    Is a scythe                            
    --- Zoe Velonis, Aged 14 1/2.

    This archive was generated by hypermail 2.1.5 : Sun Apr 13 2003 - 12:08:25 EDT