Re: Authography and Grammar

From: Richard Wordingham (
Date: Sun Jun 26 2005 - 13:08:05 CDT

  • Next message: Sinnathurai Srivas: "Re: Tamil sha (U+0BB6) - deprecate it?"

    Michael Everson wrote:
    > From: suzanne mccarthy <>

    >>Work on early Tamil Brahmi by Iravatham Mahadevan has also shown
    >>that Tamil Brahmi was not an abugida like the other Brahmi scripts.
    >>It did not have an inherent a.


    > The page states that "MA KA NA" is read "MA KA N". So what? Modern Hindi
    > does the same; that does not mean that Devanagari has ceased to be an
    > abudiga, and it does not mean that Tamil is not an abugida either.

    The article actually says, ' It is the recognition of the absence of the
    inherent vowel a (short) in the early phases, e.g. ma, ka, na with strokes
    or medial vowel notations, which are actually to be read as ma, ka, n ...',
    which looks muddled to me. There is a clearer earlier statement:

    ' These are reflected in the development of the Tamil-Brahmi in three stages
    (TB I, II and III): Stage I when the inherent a (short-medial vowel) was
    absent in the consonants and the strokes (vowel notations) were used for
    both the short and long medial a, and hence the need for the reading of
    consonants with reference to context and position; Stage II when the stroke
    for medial a marked only the long a; and Stage III when the use of
    diacritics like the pulli was introduced for basic consonants and for
    avoiding ligatures for consonant clusters (as in Simhala-Brahmi). '

    Even that is not totally clear.

    However, the history makes plenty of sense. Early Brahmi did not
    distinguish final consonants from consonant plus short /a/. Disregarding
    consonants clusters, which were treated separately, final consonants were
    rare enought that the ambiguity did not matter and it was more efficient to
    use an inherent vowel for Prakrits. This does not match the phonetics of
    Tamil, so perhaps it is not surprising that it was felt better to mark any
    /a/ - TB I. Prakrit/Sanskrit practice then comes to prevail - TB II. The
    pulli is invented (I don't know how related to the equivalent virama it is),
    and pulli/virama makes the evolving Brahmic system unambiguous. North
    Indian and Khmer then shed final short vowels (or at least, both shed final
    /a/) without modifying the spelling, and the ambiguity well known from Hindi
    and Thai can set in.

    >>It is also interesting to note that Isaac Taylor 1883 in The
    >>Alphabet represented Tamil as an alphabet. The consonant plus pulli
    >>was shown as the basic unit unlike all other Brahmi derived scripts.

    >The pulli kills the inherent vowel of the abugida.

    >>Diderot's encyclopedia, 1750, portrays Tamil as a syllabary, once
    >>again, unique representation among all Brahmi scripts.

    >My copy does no such thing. Plate XV shows Grantha with inherent vowels and
    >plate XX shows Tamil with inherent vowels. That it also draws out the full
    >set of letters with their vowel matras is a
    presentation issue; in Plate XVII for Devanagari and in Plate XXI for Thai,
    the letters and vowel matras are also shown. The use of the word "syllabary"
    on that page does not signify that Tamil is not an
    abugida (a term unknown to Diderot).

    Would you two care to resolve whether electrons are waves or particles? The
    abugida view predicts most of the shapes, the syllabary view explains the
    difficulty some have in typing Tamil in phonetic order. Did you notice that
    N. Ganesan just cited the forms with pulli when he wanted to talk about the

    >>Part of the confusion seems to come back to the fact that the pulli
    >>has been called "virama" in Unicode although they do not do the same

    > I disagree.

    Surely most of the confusion here arises from the fact that a visual
    virama/pulli is read as 'ends in a consonant' rather than 'inherent vowel
    removed', and thus it does not feel related to the concept of a conjunct.
    That's why I like the 'coeng' of Khmer. Using the non-Unicode language, the
    dual role of graphic character and control function assigned to VIRAMA feels
    wrong to many.


    This archive was generated by hypermail 2.1.5 : Sun Jun 26 2005 - 13:10:15 CDT