Re: What justification for separately encoding two forms of lowercase  sigma

From: verdy_p (
Date: Mon Sep 07 2009 - 19:28:54 CDT

  • Next message: verdy_p: "Re: Request clarification on disunification based on different character properties"

    "Michael Everson" wrote:
    > On 23 Aug 2009, at 14:08, David Starner wrote:
    > > Greek is a simple script. You have a choice of duplicating one
    > > character, like has always been done for Greek keyboards, or you can
    > > force all the Greek users to learn how to use characters like ZWJ--in
    > > fact forcing them to think in Unicode terms--so one character can be
    > > saved and the encoding is arguably theoretically a bit cleaner. It's
    > > not worth the trade-off.
    > This is a splendid argument for the disunification of some Greek
    > characters for Latin IPA. :-)

    Why that? Latin also has its own contextual forms, even if they are now deprecated in most modern usages ; think
    about the long form of the small s (historically used in initial and medial positions and preferred in most cases)
    and the current "normal" form of the small s (historically correct only at final positions of morphemes), it is
    exactly similar to the case of the two forms for small sigma.

    It also exhibit the same caveats about the impossibility to determine algorithmically how to choose between the two
    forms (in modern Latin, the choice has been made and changed the historical preference).

    But there still remains some ambiguities about the choice of the letter forms for small a (once you start making
    distinctions in them, by borrowing the Greek small alpha into Latin as a distinct letter), and this impacts the
    still common cursive style used by Latin hand writing (even though the usage of this cursive style is slowly
    decreasing now, given that many schools do not teach the cursive forms of Latin letters before more advanced
    education levels, and that many people are now so much exposed to numeric texts rendered on screens that they have
    completely forgotten how to use the cursive style correctly; there might even be countries where the cursive style
    of Latin is not taught at all in schools).

    There still remains however, even within people that now use the non cursive style exclusively, a lot of people that
    use only the form without the long harm above, and thzt just draw a short vertival stroke, or draw their small a
    letters like small alphas, with a single curved stroke instead of two with the usual forms used on printed and
    displayed texts; they don't perceive the difference as significant enough to justify them to change the way they
    handwrite their texts on paper: the single curved stroke is just faster/simpler for them to draw manually if they
    don't use the cursive style.

    But once you start thinking about the cursive handwritten Latin style (which is still popular, including in typed
    documents like ads, logos and trademarks, title decorations), there are a lot of contextual forms for many letters
    (in fact for most consonnants). The long s form is just one of them that may have disappeared today, even if it
    remains in typed books. You think that they are ligatures, but it's not entirely true given that the contextual
    forms are simple local adaptation to the cursive joining, and the fact that ligatures are often more complex, and
    are created by:

    - simplifications of this contextual cursive forms (to ease their drawing and produce more "fluid" calligaphy
    without repositioning the pen),
    - with additional strokes or strokes merged together (such as the horizontal stroke in the middle of two successive
    'f' or 't',
    - or the continued strokes in the ligatures like 'ct',
    - or strokes that are drawn in the opposite direction,
    - or strokes that disappear completely.

    The presence of additional diacritics may also change the distribution of ligatures as these diacritics have to be
    dranw sometime by breaking the current cursive line often in the middle of the same word, diabling the formation of
    the ligature at this breaking position.

    Consider the two contextual letter forms of the lowercase 'l' (also in cursive style) that can appear after a vowel
    or a 'p/c/d/h...' and compare it to the alternate form after a lowercase 'b' : the joining type and position plays a
    role similar to the joining types in Arabic (which is still usually written and printed in cursive style). In fact
    the cursive contexutal interactions between letters in Latin are probably even more complex than in Arabic, even if
    the "kernel" part of the Latin letters remains generally identifiable.

    Consider also the difference between the final and non-final forms of cursive letters with a leg (f, g, j, p, y, z):
    in final positions, the legs are often left open below the base line, and typically drawn below the previous
    letters, without creating a loop to join the next letter, or a simpler form is used without any leg (z for example).

    What happens in Greek is then not exceptional, and happens in Latin too. This is not an argument for IPA as Latin
    (even if it has been historically specified as accepting the cursive style, something abandonned today).

    Conclusion: Really IPA may have been created to harmonize with Latin, but the reality is that today, IPA only has to
    harmonize with itself only, i.e. as a script separate from Latin too... So it does not borrow, by itself, Greek
    letters into Latin, but only within its own script.

    This archive was generated by hypermail 2.1.5 : Mon Sep 07 2009 - 19:31:14 CDT