ISO/IEC JTC1/SC2/WG2 N_______
Date: 1997-06-13
This is an unofficial HTML version of a document submitted to WG2.

Title: Mapping of Sinhala between ISO/IEC 10646 and SLS 1134

Source: Michael Everson, Everson Gunn Teoranta (IE)
Status: Expert Contribution
Action: For consideration by JTC1/SC2/WG2

In addition to the basic mapping table, information is given on the encoding of conjuncts in Sinhala and Pali with ISO/IEC 10646.

Equivalence table

Archaic Sinhala numerals in UCS data cannot be encoded in SLS 1134 (see clause 4.11). The SHORT CODE is also used to generate secondary forms of the vowel symbols U and UU. I am not sure if this is necessary in a 10646 context, if selection of the long or short form is based on the preceding consonant.

Coding of Sinhala syllables and conjuncts

The documentation system used for transliterating throughout this standard is a linear method and all entries will display on the screen as they are entered. But in the other method which could be developed on the phonetic basis of the characters, some entries carrying combinations will be stored, if necessary modified and displayed as they are to be documented, for example E + KA + AA (= KO) and KA + O (= KO).

Some vowels wich can be consructed using relevant consonant modifiers are not included. But for users who are interested in having the complete set of vowels, spaces are still reserved to include those characters, in accordance with the alphabetical order.

ISO/IEC 10646-encoded Sinhala is based on the Brahmic harmonization, which means that the underlying encoding is phonetically-based. The advantages of Brahmic harmonization are many: improved facility for transliterating Pali to and from Sinhala script, greater ease of portability for software (providing a wider market for Sri Lankan programmers), simplification of string comparisons for historical comparative linguistics, etc. The mapping table presented here will work easily if an SLS 1134 text has been encoded according to this phonetic principle. If it has been encoded according to the visual principle, reliable transfer can also be achieved, though the mapping table will need to be considerably larger. Further information on coding of SLS 1134 texts will be welcome.

Of course despite the fact that Sinhala is coded according to the phonetic principle in ISO/IEC 10646, input of characters could still be achieved according to the visual principle if Sri Lankan input preferences require it thus (the same can be said of Burmese).

Correct formation of conjuncts is somewhat more complex in Sinhala than in other Brahmic scripts, but it is very important because conjuncts are formed differently depending on whether a word is Sinhala or Pali. From Gunasekare:

It is important to observe that letters in Elu [Sinhala] words are generally written separately, while the rule is reversed in Pali and Sanskrit. It is therefore incorrect to write the Elu words AT-TA atta, 'branch,' VIS-SA vissa, 'twenty,' as A-TTA, VI-SSA; nor is it thought proper to write the Sanskrit words SA-RVA sarva, 'all,' VII-RYYA víryya, 'strength,' A-NTA-RAA-YA antaráya 'danger,' as SAR-VA, VIIR-YA, AN-TA-RAA-YA. [§19]

• BUD-DHO > BA + U + DA + VIRAMA + ZWNJ + DHA + O (zero-width non-joiner forces virama to display and defines a syllable break)
  
• BU-DDHO > BA + U + DA + VIRAMA + ZWJ + DHA + O (zero-width joiner forms normal kerned conjuncts; both consonants in same syllable)
  
• BU-[DDH]O > BA + U + DA + VIRAMA + DHA + O (special conjunct formed; both consonants in same syllable.)
  
• DHAM-MO > DHA + MA + VIRAMA + ZWNJ + MA + O (zero-width non-joiner forces virama to display and defines a syllable break)
  
• DHA-MMO > DHA + MA + VIRAMA + ZWJ + MA + O (zero-width joiner forms normal kerned conjuncts; both consonants in same syllable)
  

In SLS 1134, BUD-DHO is coded BA + U + DA + LINK + E + DHA + AA (or BA + U + DA + LINK + DHA + O). In clause 1.3 it appears to be claimed that BU-DDHA cannot be coded with SLS 1134, but this seems to be contradicted by clause 4.6. I suspect my reading of the standard; clause 1.3 states that it "does not make provisions to produce double linked characters". It seems to me that the three examples here can be encoded thus:

• BUD-DHO > BA + U + DA + AL-LAKUNA + DHA + O
  
• BU-DDHO > BA + U + DA + LINK + DHA + O
  
• BU-[DDH]O > BA + U + DA + SHORT + DHA + O
  

The glyphs for the Pali special conjuncts [CC]A and [BB]A happen to look identical to those for the base letters DDA and NNYA; they should always be encoded with their constituent consonants and VIRAMA or with SHORT.

Sorting

Further information from SLSI and other Sri Lankan experts on the sorting of texts with both Sinhala and Pali or Sanskrit words (vis à vis the conjuncts discussed in this paper and with reference to clause 1 note 2 of SLS 1134) will be gratefully received.