L2/08-078

Subject: Tamil issues
Source: Mark Davis
Date: 2008-01-29

It is is clear that many people would have wanted Tamil to be encoded more along the lines of the Ethiopic model, with complete clusters, instead of following the ISCII model. While the decision to do that is long since past, it would be helpful to provide a clearer description of how Tamil clusters function, with a code chart showing the clusters. Here is a draft proposal for addition of text to http://www.unicode.org/versions/Unicode5.1.0/  along those lines. (This would take wordsmithing by the editorial committee, so the goal here is just the basic content.) I also suggest moving the named sequences for Tamil from provisional to accepted in Unicode 5.1.

Tamil is less complex than some of the other Indic scripts, and both conceptually and in processing can be treated as an atomic set of elements: consonants, stand-alone vowels, and syllables. The following chart shows these atomic elements, with the corresponding Unicode code points. They are also represented as Named Sequences in the Unicode Character Database, which can also be derived from the chart as described below.

In implementations such as natural language processing, where it may be useful to treat these as single code points for ease of processing, they can be mapped to a segment of the Private Use Area.

Key

• Column 1 shows the ASCII representation of the vowel names.
• Row 1 shows the ASCII representation of the consonant names.
• Column 2 contains the consonants. Their Unicode Named Sequence names are formed by concatenating "TAMIL CONSONANT " to the ASCII value in Column 1.
• Row 2 contains the stand-alone vowels. They have Unicode character names corresponding to the concatenation of "TAMIL LETTER " to the ASCII value in Row 1.
• Column 3 has Unicode character names corresponding to the concatenation of "TAMIL LETTER " to the value in Row Column 1.
• The other values (Row R, Column C) have Unicode Named Sequence names formed by concatenating "TAMIL SYLLABLE " to the ASCII value for Row R and the ASCII value in Column C.
  • For example, கூ => "TAMIL SYLLABLE " + "K" + "UU" => TAMIL SYLLABLE KUU.