A number of Indic scripts have a danda and double danda which are punctuation characters. Unicode 4.0 includes U+0964 । DEVANAGARI DANDA and U+0965 ॥ DEVANAGARI DOUBLE DANDA but does not include similar characters for the other scripts. The question is whether those two characters are intended to be used with the other scripts, or whether new characters need to be encoded for the other scripts.
The Indic Working Group did not arrive at a consensus on this question. The next two sections present the arguments for and against the encoding of script-specific characters. The final two sections are alternative proposals, one of which should be selected by the UTC.
F1: this pattern has already been followed for other scripts, such as Myanmar and Hanunoo. which have their own danda. It has also been followed in the "core" Indic scripts for characters such as U+0950 ॐ DEVANAGARI OM / U+0AD0 ૐ GUJARATI OM.
Rebuttal: Actually, the dandas encoded in the Hanunoo block are not specific to that script, but are shared by the Philippine scripts: they appear under the header “Generic Punctuation for Philippine scripts”, and have generic names: U+1735 PHILIPPINE SINGLE PUNCTUATION and U+1736 PHILIPPINE DOUBLE PUNCTUATION; the names are not HANUNOO ... So the pattern already in place is one of sharing among closely related scripts. In fact, if separate characters are encoded for the Indic scripts, then it would be awkward to not also encode separate characters for the Philipinne scripts.
F2: the typographic behaviour of the danda and the double danda changes across the scripts. Not only are the shapes somewhat different, but the positioning is also different.
Rebuttal: in a font that supports a single script, this is not a problem at all: whether the glyphs are mapped from the Devanagari block or the Bengali block has no impact. Font technologies which support mutliple scripts also support some mechanism by which the same code point can result in different glyphs, possibly with different layout behavior.
F3: in multi-script text, handling an occurrence of a danda between characters of different scripts (not an uncommon occurrence) requires some care, so that it is rendered using the same font as the text it belongs to.
Rebuttal: this is not that difficult: the danda goes with the text that is before it. And anyway, the problem is not specific to the danda, it also applies to other punctuation marks (comma, period, etc), so a solution, which is applicable to the danda, has to exist anyway.
F4: users are often confused when they don't see the danda and double danda in the code chart for the Bengali (or Gurmukhi, ...) block.
Rebuttal: for most users, the code charts are not the first tool they do or should use. The layout of their keyboard is much more relevant to them, and it will show the danda and the double danda. We can assume that users who do need to dig in the code charts are also willing to read, e.g., chapter 9.
F5: users can be mislead when they look at the Devanagari code chart, and see "DEVEVANAGARI DANDA"; they can easily interpret the name as restricting the use of the character to Devanagari.
Rebuttal: this can easily be addressed by an annotation of the character, right next to the name, such as "not script specific despite its name, also use with the other Indic scripts.", as well as text in chapter 9.
A1: this pattern (shared characters) is already being followed for other punctuation characters which are used across the scripts: e.g. period, comma, question mark,
A2: the shared nature of U+0964 and U+0965 has already been recognized, in data and in implementations. Thus, encoding script specific characters would amount to a disunification, will the problems this usually causes. Even if there is today a small amount of data that would suffer from disunification, because of fairly long time it takes for the standard process, the amount of data with problematic representation would be significant by the time the new characters can be used.
Rebuttal: Unicode will be used for centuries. Ten years of Devanagari-encoded dandas in Oriya text is of little consequence. The disunification costs for this fix are insignificant compared to the benefits.
The proposal is to encode the following characters (“P+” is used instead of “U+” to emphasize that those characters are proposed):
The names are those proposed by the TDIL. Alternatively, the model currently in place for Devanagari could be followed: “<script> DANDA” and “<script> DOUBLE DANDA”, with annotations using the names above.
The properties of those characters would similar to those of the Devanagari equivalents:
In addition, the script of U+0964 and U+0965 should be changed from Zyyy to Deva.
Note that this proposal does not include dandas for Tamil, Kannada, Malayalam nor Sinhala.
Here are examples of both dandas in Bengali, from Learn Bengali yourself, Bidhu Bhusan Das Gupta, Calcutta: Das Gupta Prakashan, 3rd edition, page 147. The danda occurs twice (vertical line at the end of the second and fourth line) and the double danda occurs at the end of the last line.
Here are examples of the Gurmukhi danda, from a newspaper, reprinted in Writing Systems of the World, Akira Nakahishi, Rutland: Charles E. Tuttle Company, 1980. The danda is the vertical line which occurs, among other positions, at the very end of the text.
Here are examples of the Oriya danda, from a newspaper, also reprinted in Writing Systems of the World. Again, the danda appears at the end of the text.
Put the two characters U+0964 DEVANAGARI DANDA and U+0965 DEVANAGARI DOUBLE DANDA, under a new header “Generic Punctuation for Indic scripts” (on the model of the Philippine equivalents). Futhermore, add an annotation to each character, to compensate for the “DEVANAGARI” in their names.
Add some text in chapter 9, in the introduction of the chapter and/or in sections 9.1 Devanagari, 9.2 Bengali, 9.3 Gurmukhi, 9.4 Gujarati, 9.5 Oriya and 9.7 Telugu, indicating the that U+0964 and U+0965 are the appropriate characters to use for the dandas of those scripts.
Note that the danda characters U+104A MYANMAR SIGN LITTLE SECTION and U+104B MYANMAR SIGN SECTION have the script Zyyy. If they are specific to Myanmar, they should probably have the script Mymr.
Author: Indic Working Group
| Revision | Date | Comments |
| November 8, 2004 | Added the examples |