Re: New property for reordrant dependent vowels reordering?

From: Vinod Kumar (
Date: Tue Sep 20 2005 - 04:45:09 CDT

  • Next message: Vinod Kumar: "Re: New property for reordrant dependent vowels reordering?"

    > Kent Karlsson Sun Sep 04 2005 - 04:56:34 CDT
    > Then there are two orthographic syllables here, per definition.

    R15 (Chap 9.1 TUS4): When the dependent vowel I_vs is used to override
      the inherent vowel of a syllable, it is always written to the extreme
      left of the orthographic syllable. If the orthographic syllable contains
      a consonant cluster, then this vowel is always depicted to the left
      of that cluster.

    The important word here is ortho. The rule R15 has to be interpreted
    and modified in terms of the mathematical meaning of being orthogonal
    or independent of each other. From my familiarity with Indic scripts,
    and experience in designing and implementing the shaping engine
    and OpenType tables for the Saral series of fonts for all nine Indic scripts,
    I modify and interpret the Unicode rule R15 for glyphs as follows:

    First the orthogonal glyph cluster is defined.

    Orthogonal glyph cluster:
      An orthogonal glyph cluster is the minimal sequence of glyphs
      that can occur stand-alone in displayed text between two spaces.

    Example: Each of <Virama-Glyph>, <I_vs-Glyph>, <Half-TTA-Glyph> are
      not valid as stand alone forms in text. Each of these is not an
      orthogonal glyph cluster.
      Some orthogonal glyph clusters would be
      <TT-MA-Glyph>, <Half-TTA-Glyph MA-Glyph>,
      <TTA-Glyph Virama-Glyph> and <MA-Glyph>.
      <TTA-Glyph Virama-Glyph MA-Glyph> is not an orthogonal glyph cluster.
      Even though it can stand alone between two spaces, its subsequences
      <TTA-Glyph Virama-Glyph> and <MA-Glyph>
      can stand alone and so it is not a minimal sequence.
      The importance of an orthogonal glyph cluster is that it can be
      produced, pronounced and interpreted independent of other
      glyph clusters.

    (Modified R15) When the dependent vowel I_vs is used to override
      the inherent vowel of a syllable, it is always written to the
      left of the previous orthogonal glyph cluster.

    I_vs repositioning using orthogonal glyph clusters

      The character sequence <TTA Virama MA> could take several forms
      in the glyph domain, depending on how rich the font is or because
      the writer chooses (using ZWJ, ZWNJ) a particular form.

                   3 Forms of Visual syllable for <TTA Virama MA>

      Full Visual syllable Orthogonal glyph clusters
      <TT-MA-Glyph> <TT-MA-Glyph>
      <Half-TTA-Glyph MA-Glyph> <Half-TTA-Glyph MA-Glyph>
      <TTA-Glyph Virama-Glyph MA-Glyph> <TTA-Glyph Virama-Glyph >,
      The explicit halant form (third) of the visual syllable consists of
      two orthogonal glyph clusters. The other two consist of a single
      orthogonal glyph cluster.

      Now consider the full syllable <TTA Virama MA I_vs>.
      The correct reordering of I_vs for the 3 Forms will be

      <I_vs-Glyph TT-MA-Glyph>
      <I_vs-Glyph Half-TTA-Glyph MA-Glyph>
      <TTA-Glyph Virama-Glyph I_vs-Glyph MA-Glyph>

      Look at the third form. The I_vs-Glyph should not move to left of
      the first orthogonal glyph cluster <TTA-Glyph Virama-Glyph>
      but to the left of the previous orthogonal glyph cluster <MA-Glyph>.
      A form like <I_vs-Glyph TTA-Glyph Virama-Glyph MA-Glyph> is
      incorrect. I will not be surprised if such incorrect forms have been
      created first by buggy shaping software. Then the mistake is
      propagated by other influential
      people as "two orthographies" or as "two writing conventions"
      and implemented in software like Windows Vista!

      Sometimes, people say that text display should be driven by convention
      and not by logic. I could not find any instance in Devanagari text
      (searching in several pages of a Jnaneswari exposition in Marathi)
      where the I_vs-Glyph crosses to the left of the Virama-Glyph.
      I could find only two syllables with explicit halant consonants
      and in both cases there was no I_vs in the syllable.
      Prof R K Joshi too does not recollect any syllable
      with explicit Halant and a left moving I_vs that crosses it.
      In Malayalam text, I could locate (again in a Jnaneswari exposition)
      several instances of E-matra (or Ee-matra) that moves left
      but each one of them stopped before crossing the Chandrakala (halant)
      in the syllable's form.

      When conventional instances cannot be seen, one has to fall back
      on logic to deduce forms that can be justified and defended.
      The form
      <TTA-Glyph Virama-Glyph I_vs-Glyph MA-Glyph> will be pronounced
      properly as TT followed by MI, maybe with an inordinate delay
      between TT and MI. The incorrect form
      <I_vs-Glyph TTA-Glyph Virama-Glyph MA-Glyph>
      tends to be pronounced as TTI followed by MA at first glance.
      Later the Virama will be noticed and the pronouncer may scratch
      her head wondering how an animal like <I_vs-Glyph TTA-Glyph
      Virama-Glyph> should sound. Such a confusion does not arise with
      <I_vs-Glyph Half-TTA-Glyph> because <Half-TTA-Glyph> is not a stand
      alone symbol and has to be seen together with the succeeding glyph.

    Vinod Kumar
    Project IndiX

    This archive was generated by hypermail 2.1.5 : Tue Sep 20 2005 - 04:45:47 CDT