Proposal for a Kannada Script Root Zone Label Generation ... · dynasty, in the 6th century A.D.,...

ProposalforaKannadaScriptRootZoneLabelGenerationRuleset(LGR)

1


LGRVersion:3.0Date:2019-03-06Documentversion:2.6Authors:Neo-BrahmiGenerationPanel[NBGP]

1. GeneralInformation/Overview/Abstract ThepurposeofthisdocumentistogiveanoverviewoftheproposedKannadaLGRintheXMLformatandtherationalebehindthedesigndecisionstaken.Itincludesadiscussionofrelevantfeaturesofthescript,thecommunitiesorlanguagesusingit,theprocessandmethodologyusedandinformationonthecontributors.TheformalspecificationoftheLGRcanbefoundintheaccompanyingXMLdocument: proposal-kannada-lgr-06mar19-en.xmlLabelsfortestingcanbefoundintheaccompanyingtextdocument: kannada-test-labels-06mar19-en.txt

2. ScriptforwhichtheLGRisProposedISO15924Code:KndaISO15924N°:345ISO15924EnglishName:KannadaLatintransliterationofthenativescriptname:Nativenameofthescript:ಕನ#ಡ

MaximalStartingRepertoire(MSR)version:MSR-4SomelanguagesusingthescriptandtheirISO639-3codes:Kannada(kan),Tulu(tcy),Beary,Konkani(kok),Havyaka,Kodava(kfa)


2

3. BackgroundonScriptandPrincipalLanguagesUsingIt

3.1Kannadalanguage

KannadaisoneofthescheduledlanguagesofIndia.ItisspokenpredominantlybythepeopleofKarnatakaStateofIndia.ItisoneofthemajorlanguagesamongtheDravidianlanguages. Kannada is also spoken by significant linguisticminorities in the states ofAndhraPradesh,Telangana,TamilNadu,Maharashtra,Kerala,Goaandabroad.Asperscholars,Kannadawasaspokenlanguageduringthe3rdcenturyB.C.Ptolemy,ascholarfromAlexandria,inhisTheGeographywrittenduringthefirsthalfofthesecondcenturyA.D.mentionssomeKannadawords.PtolemyspeaksofmanyplacesinKarnatakasuchasKalgeris(identifiedasKalkeri),Modogoulla(Mudugal),Badamios(Badami)andsoon.AllthesearenotonlyplacesinKarnataka,butarealsonamesofKannadaorigin.ThefamousHalmidiRecordoftheKadambas,whichisaninscriptionofthe5thcenturyA.D.,istheoldestavailableevidenceofKannadalanguagewritteninthepre-OldKannadascript.KappeArabhatta’sRecordatBadami (700A.D.)has the firstKannadapoem in%&ಪ( tripadi metre. The oldest available literary work in Kannada is ಕ)*ಾಜ-ಾಗ/ –

Kavirajamarga,abookonpoeticsbelongingto9thcentury.ThisworkspeaksofsomeearlierpoetsinKannada.Hence,Kannadamusthavebeenafullydevelopedlanguagebythe5thorthe6thcenturyA.D.andmusthavebeenaspokenlanguageforatleastafewcenturiesearlier.Kannadaisattestedepigraphicallyforaboutoneandahalfmillennia,andliteraryOldKannadaflourishedinthe6th-centuryGangadynastyandduringthe9th-century Rashtrakuta Dynasty. Kannada has an unbroken literary history of over athousandyears.

3.2EvolutionofKannadascript

TheKannadalanguageiswrittenusingtheKannadascript,whichevolvedfromthe5th-centuryKadambascript.TheoldestformofKannadascriptbeginsin3rdcenturyB.C.Thefirst popular andwell-knownKannada scriptwas calledKadamba script used by theKadambadynastyduring5thcenturyA.D.Buhler,thefamousepigraphistsaysthattheKadamba script is the earliest formof the present dayKannada script. DuringGangadynasty, in the 6th centuryA.D., the script used is known asAdi Ganga script,whichresemblesKadambascript.During6-7thcenturyA.D.,theChalukyasofBadamiusedascriptwhich isnowcalledBadamiChalukya scriptbyhistorians.Rashtrakutawas thenextfamousdynasty,whichruledduring8-10thcenturyA.D.andthescriptusedduringthosetimeisreferredtoasRashtrakutascript.ThescriptusedbytheKalyanaChalukyarulersiscalledKalyanaChalukyascript.Itcanbeseenintherecordsof10-12centuryAD.Cursivewritingwas started during the13th century byHoysala kings.They built thedecorativecursivewayofwritingbasedonthescriptofKalyanaChalukyas.Inscriptions


3

atBeluruandHalebeeduhavetextwrittenusingthiskindofscript.TheVijaynagarkingsruledduringthe14-16thcenturyA.D.didnotmakeanymajormodificationstothescript.The lastdynastyofKarnataka, thekingsofMysoredevelopedwhat isknownasModiscript.ItiscalledModiscriptor0ೕ2ಬರಹ(Modibaraha).Mostofthepublicrecords

that were written during the period of the Mysore kings are in the Modi script. NoinscriptionswerewrittenintheModiscriptasthisstyleisdifficulttoinscribeonastone.Thismaybeconsideredthelatestdevelopedformofthescript,andistaughtevennowinschoolsascursivewritingforKannada.

Figure1:EvolutionofKannadascriptfrom3rdcenturyB.C.to18thcenturyA.D.

(fromhttps://karnatakaitihasaacademy.org/karnataka-history/evolution-of-kannada-script/)

3.3Languagesconsidered

ApartfromtheKannadalanguage,otherlanguagesthatusetheKannadascriptare-Tulu,Kodava(Coorgi),Konkani,Havyaka,Sanketi,Beary(byaari),Arebaase,Koraga,etc.Tuluhaditsownscript,whichisnotinmuchusenowadayseventhoughlotofeffortsarebeingdoneof late to revive theTulu script.TheKonkani language iswritten inDevanagari,Roman,andMalayalamscriptsalso.


4

3.4StructureofwrittenKannada

ThestructureofKannadaissimilartootherIndianlanguages,especiallytoTelugu.TheheartofthewritingsystemistheAkshar.TheKannadaalphabetisknownasaksharamaleorvarnamale.Themodernalphabetcontains49characters.ThishasbeenarrivedatbyremovingtwocharactersthataremainlyusedtowriteclassicalKannadatexts.Thesetwocharacterswereinusejustabout50yearsago.Characterscombinetoformcompoundcharacters called as samyuktakshara (conjuncts). These compound characters havedistinctdisplayforms.Thetotalnumberofsuchcombinationswillbeabout650,000.Thebasiccharactersinvarnamaleareclassifiedintothreemaincategories.Theyare-swara(vowels),vyanjana(consonants)andyogavahas.

3.4.1Swaras(vowels) Therearethirteenvowels

Letter Diacritic ISOnotation

ಅ N/A a

ಆ ◌ಾ ā

ಇ ◌ i

ಈ ◌ೕ ī

ಉ ◌ು u

ಊ ◌ೂ ū

ಋ ◌ೃ rū

ಎ ◌ e

ಏ ◌ೕ ē

ಐ ◌ೖ ai

ಒ ◌ೂ o

ಓ ◌ೂೕ ō

ಔ ◌ au

Table1:KannadaSwaras(vowels)

(fromhttps://en.wikipedia.org/wiki/Kannada_alphabet)

Whenavowelfollowsaconsonant,itiswrittenwithadiacriticratherthanasaseparateletter.Sometimesthesearereferredtoasvowelsignsormatras.Vowelsignsormatrasareattachedonlytoconsonants.


5

3.4.2Yogavahas(Semiconsonants) TheYōgavāha(part-vowel,partconsonant)includetwoletters:

1. Theanusvara:ಅಂ(aṁ)

2. Thevisarga:ಅಃ(aḥ)

3.4.3.Vyanjanas(consonants)Two categories of consonant characters (Vyanjanas) are defined in Kannada: thestructured consonants (Vargīya Vyañjana) and the unstructured consonants (AvargīyaVyañjana.The structured consonants are classified according to where the tongue touches thepalate of themouth and are classified accordingly into five structured groups. Theseconsonantsareshownhere

voiceless voicelessaspirate

voiced voicedaspirate

nasal

Velars ಕ(ka) ಖ(kha) ಗ(ga) ಘ(gha) ಙ(ṅa)

Palatals ಚ(ca) ಛ(cha) ಜ(ja) ಝ(jha) ಞ(ña)

Retroflex ಟ(ṭa) ಠ(ṭha) ಡ(ḍa) ಢ(ḍha) ಣ(ṇa)

Dentals ತ(ta) ಥ(tha) ದ(da) ಧ(dha) ನ(na)

Labials ಪ(pa) ಫ(pha) ಬ(ba) ಭ(bha) ಮ(ma)

Table2:KannadaConsonants

(fromhttps://en.wikipedia.org/wiki/Kannada_alphabet)

The unstructured consonants are consonants that do not fall into any of the abovestructures:ಯ(ya),ರ(ra),ಱ(ṟa)(obsolete),ಲ(la),ವ(va),ಶ(śa),ಷ(ṣa),ಸ(sa),ಹ(ha),ಳ

(ḷa),ೞ (ḻ) (obsolete).From this list, the twoobsolete characters (ಱ andೞ)havebeen

removedinmodernvarnamalebringingthetotalnumberofcharactersto49.


6

3.4.4Implicitvowelಅ(a)inconsonants

All consonants (vyanjanas) in Kannada when written as ಕ (ka),ಖ (kha),ಗ (ga), etc.

containanimplicitvowelಅ(a).Theconsonantsh,i,j,etc.,areshownafterremoving

theimplicitvowelಅ(a).Infact,manygrammarbooksonKannadalisttheconsonantsby

removing the implicit ಅ (a). The Unicode character U+0CCD, which is the Kannada

equivalent of Devanagari’s Halant U+094D (or VIRAMA as Unicode calls it), followsconsonantstoremovetheimplicitಅ(a).Halantcanonlyfollowaconsonantandnoother

characters.Avowelsign(matra)followingtheconsonantreplacestheimplicitvowelbyadifferentvowel.

3.4.5ConjunctsKannada is known to have a large number of conjuncts which are nothing butcombinationofconsonantsandvowelsigns(matras).Thesearealsoknownassyllables.Different types of consonant and vowel sign combinations are possible. They are thefollowing:

• Consonant+Vowelsign, e.g.,ಕ(ka, U+0C95)+ ◌ೂ (U+0CCA,matraofvowelಒ) =oೂ

• Consonant+Halant+Consonant, e.g.,ಕ (ka, U+0C95)+ Halant(U+0CCD)+ ಕ (ka, U+0C95)=ಕr

• Consonant+Halant+Consonant+Vowelsign, e.g. ಕ (ka, U+0C95)+ Halant(U+0CCD)+ ಕ (ka, U+0C95)+ ◌ೂ (U+0CCA,matra

ofvowelಒ) =oೂr

• Consonant+Halant+Consonant+Halant+Consonant e.g.,ಷ (ṣa, U+0CB7)+ Halant(U+0CCD)+ಟ (ṭa,U+0C9F)+Halant(U+0CCD)+ ರ

(ra, U+0CB0) =ಷs

• Consonant+Halant+Consonant+Halant+Consonant + Vowelsign

e.g.,ಷ (ṣa, U+0CB7)+ Halant(U+0CCD)+ ಟ (ṭa,U+0C9F)+Halant(U+0CCD)+ ರ

(ra, U+0CB0) +◌ೂ (U+0CCA,matraofvowelಒ) =tೂs

Conjunctsclusterhavingmorethan3consonantsinonesyllablearenormallynotseeninKannada.


7

3.4.6Purevowelsinthemiddleofaword

InKannada,itiscommontohavewordsstartingwithavowel.Sometimes,oflate,peoplearewritingwordshavingpurevowelsinthemiddleofaword.Thiskindofwritingwasoriginally normally not seen in Kannada. This kind of writing has been made arequirementtowritethewordsimportedfromotherlanguages,especiallyfromEnglish.Linguisticallythisisnotinvalidandhencecanbeallowed.

3.4.7IllegalcombinationsTherearesomecombinationswhichareinvalidasperKannadagrammar.Theyarelistedbelow:

3.4.7.1 Havingtwoormoreconsecutivevowelsigns(matras).3.4.7.2 Havingavowelsign(matra)afteravowel.3.4.7.3 Havingavowelsign(matra)afteraYōgavāha(anusvaraorvisarga).3.4.7.4 HavingaHalantafteravowelorvowelsign(matra).3.4.7.5 HavingaYōgavāhaafteraHalant.

For3.4.7.4therecouldbecasesinvolvingmulti-worddomainswhereVmayneedtobeallowedtofollowanH.ThisisthecasewheretwodifferentwordsarejoinedtogetherfirstofwhichendswithaHalantandthesecondwordbeginswithaVowel.SomesectionsofthelinguisticcommunityrequiretheexplicitpresenceofHforfullrepresentationofthesoundintended.However,byandlarge,theformofthefirstwordwithoutaHisconsideredsufficientforthefullrepresentationofthesoundintendedforthefirstword.Thisisauniquesituationnecessitatedbythelackofhyphen,spaceortheZeroWidthNon-joinercharacterinthepermissiblesetofcharactersintheRootzonerepertoire.Otherwise,VisneverrequiredtobeallowedtofollowanH.However,permittingthismaycreateaperceptualsimilaritybetweentwolabels(withandwithoutH)formajorityofthelinguisticcommunity,hencethisisexplicitlyprohibitedbytheNBGP.Dependingontheprevailingrequirementsbythecommunity,afutureNBGPmayconsiderrevisitingthisrule.

4. OverallDevelopmentProcessandMethodologyNeo-BrahmiGenerationPanel(NBGP)hasbeenformedbymembershavingexperiencein linguistics and computational linguistics. Under the Neo-Brahmi Generation Panel,thereareninescriptsbelongingtoseparateUnicodeblocks.EachofthesescriptshasbeenaseparateLGR;however,theNeo-BrahmiGPensuresthatthefundamentalphilosophybehindbuildingthoseLGRsareallinsyncwithallotherBrahmi-derivedscripts.


8

NBGPconsideredallthelanguageswithEGIDSscale1to4andfoundthatKannadascriptis being used for Kannada, Tulu, Beary, Konkani, Havyaka, Kodava, among otherlanguages.

4.1GuidingPrinciples

TheNBGPadoptsfollowingbroadprinciplesforselectionofcode-pointsinthecode-pointrepertoireacrosstheboardforallthescriptswithinitsambit.

4.1.1Inclusionprinciples:

4.1.1.1Modernusage:Every character proposed should be in the everyday usage of a particular linguisticcommunity.Thecharacterswhichhavebeenencoded intheUnicode for transcriptionpurposesonlyorforarchivalpurposeswillnotbeconsideredforinclusioninthecode-pointrepertoire.

4.1.1.2Unambiguoususe:Everycharacterproposedshouldhaveunambiguousunderstandingamongthelinguisticaboutitsusageinthelanguage.

4.1.2Exclusionprinciples:ThemainexclusionprincipleisthatofAcknowledgementofEnvironmentalLimitations.These comprise of protocols or standards which are pre-requisites to the LabelGenerationRulesets.Allfurtherprinciplesareinfactsubsumedundertheselimitationsbuthavebeenspeltoutseparatelyforthesakeofclarity.

4.1.2.1ExternallimitsonScope:Thecodepointrepertoire forrootzonebeingaveryspecialcase,uptheladder in theprotocolhierarchies,thecanvasofavailablecharactersforselectionasapartoftheRootZonecodepointrepertoireisalreadyconstrainedbyvariousprotocollayersbeneathit.Followingthreemainprotocols/standardsactassuccessivefilters:i.TheUnicodeChart:Outofallthecharactersthatareneededbythegivenscript,ifthecharacterinquestionisnotencodedinUnicode,itcannotbeincorporatedinthecodepointrepertoire.Suchcasesarequiterare,giventheelaborateandexhaustivecharacterinclusioneffortsmadebyUnicodeConsortium.ii.IDNAProtocol:Unicode being the character encoding standard for providing the maximum possiblerepresentationofagivenscript/language,ithasencodedasfaraspossibleallthepossible


9

charactersneededbythescript.However,theDomainnamebeingaspecializedcase,itisgovernedbyanadditionalprotocolknownasIDNA(InternationalizedDomainNamesinApplications).TheIDNAprotocolintroducesexclusionofsomecharactersoutofUnicoderepertoirefrombeingpartofthedomainnames.

Example: Kannada sign CANDRABINDU ! (U+0C81) is not allowed to be part of thedomainname.

iii.MaximalStartingRepertoire:TheRoot-zoneLGRbeingarepertoireofthecharacterswhicharegoingtobeusedforcreationoftherootzoneTLDs,whichinturnareanevenmorespecializedcaseofdomainnames,theROOTLGRprocedureintroducesadditionalexclusionsonIDNAallowedsetofcharacters.Example:KannadaSignAVAGRAHA"ऽ"(U+0CBD)evenifallowedbyIDNAprotocol,isnotpermittedintheRootZoneRepertoireasperthe[MSR].Tosumup,therestrictionsstartoffbyadmittingonlysuchcharactersasarepartofthecode-blockofthegivenscript/language.ThisisfurthernarroweddownbytheIDNA2008Protocol and finally an additional filter in the form of Maximal Starting Repertoirerestrictsthecharactersetassociatedwiththegivenlanguageevenmore.

4.1.2.2NoRareandObsoleteCharacters:There are characterswhich have been added toUnicode to accommodate rare formsespeciallylikeKANNADALETTERVOCALICL"ಌ"(U+0C8C)aswellasitsmatraforms"◌"

(U+0CE2). All such characters will not be included. This is in consonance with theConservatismprincipleaslaiddownintheRootZoneLGRprocedure.

4.2MethodologytoincorporatethefeedbackreceivedthroughPublicCommentprocess

TheKannadascriptLGRproposalwaspublishedforpubliccommenttoallowthosewhohadnotparticipatedintheNBGPtomaketheirviewsknown.TheNBGPanalyzedallcommentsreceivedtofinalizetheproposal.Theanalysisofpubliccommentscanbeaccessedonlinegivenat[111].

5. RepertoireSection5.1providesthesectionofthe[MSR]applicabletotheKannadascriptonwhichtheKannadacodepointrepertoireisbased.Section5.2detailsthecodepointrepertoire


10

thattheNeo-BrahmiGenerationPanel[NBGP]proposestobeincludedintheKannadaLGR.

5.1KannadasectionofMaximalStartingRepertoire[MSR]Version4

Figure2:KannadaCodePagefrom[MSR]

Colorconvention1:Allcharactersthatareincludedinthe[MSR]-YellowbackgroundPVALIDinIDNA2008butexcludedfromthe[MSR]-PinkishbackgroundNotPVALIDinIDNA2008-Whitebackground

1This document needs to be printed in color for this to be read correctly.


11

5.2Codepointrepertoire

GivenbelowistherepertoireforKannadabasedonUnicodecharacterset.

Sr.No.

UnicodeCodePoint

Glyph CharacterName Category Reference

1 0C82 ◌ಂ KANNADASIGNANUSVARA

Anusvara 110

2 0C83 ◌ಃ KANNADASIGNVISARGA

Visarga 110

3 0C85 ಅ KANNADALETTERA Vowel 110

4 0C86 ಆ KANNADALETTERAA Vowel 110

5 0C87 ಇ KANNADALETTERI Vowel 110

6 0C88 ಈ KANNADALETTERII Vowel 110

7 0C89 ಉ KANNADALETTERU Vowel 110

8 0C8A ಊ KANNADALETTERUU Vowel 110

9 0C8B ಋ KANNADALETTERVOCALICR

Vowel 110

10 0C8E ಎ KANNADALETTERE Vowel 110

11 0C8F ಏ KANNADALETTEREE Vowel 110

12 0C90 ಐ KANNADALETTERAI Vowel 110

13 0C92 ಒ KANNADALETTERO Vowel 110

14 0C93 ಓ KANNADALETTEROO Vowel 110

15 0C94 ಔ KANNADALETTERAU Vowel 110

16 0C95 ಕ KANNADALETTERKA Consonant 110

17 0C96 ಖ KANNADALETTERKHA

Consonant 110


12

Sr.No.

UnicodeCodePoint


18 0C97 ಗ KANNADALETTERGA Consonant 110

19 0C98 ಘ KANNADALETTERGHA

Consonant 110

20 0C99 ಙ KANNADALETTERNGA Consonant 110

21 0C9A ಚ KANNADALETTERCA Consonant 110

22 0C9B ಛ KANNADALETTERCHA Consonant 110

23 0C9C ಜ KANNADALETTERJA Consonant 110

24 0C9D ಝ KANNADALETTERJHA Consonant 110

25 0C9E ಞ KANNADALETTERNYA Consonant 110

26 0C9F ಟ KANNADALETTERTTA Consonant 110

27 0CA0 ಠ KANNADALETTERTTHA

Consonant 110

28 0CA1 ಡ KANNADALETTERDDA

Consonant 110

29 0CA2 ಢ KANNADALETTERDDHA

Consonant 110

30 0CA3 ಣ KANNADALETTERNNA

Consonant 110

31 0CA4 ತ KANNADALETTERTA Consonant 110

32 0CA5 ಥ KANNADALETTERTHA Consonant 110

33 0CA6 ದ KANNADALETTERDA Consonant 110

34 0CA7 ಧ KANNADALETTERDHA

Consonant 110

35 0CA8 ನ KANNADALETTERNA Consonant 110

36 0CAA ಪ KANNADALETTERPA Consonant 110


13

Sr.No.

UnicodeCodePoint


37 0CAB ಫ KANNADALETTERPHA Consonant 110

38 0CAC ಬ KANNADALETTERBA Consonant 110

39 0CAD ಭ KANNADALETTERBHA

Consonant 110

40 0CAE ಮ KANNADALETTERMA Consonant 110

41 0CAF ಯ KANNADALETTERYA Consonant 110

42 0CB0 ರ KANNADALETTERRA Consonant 110

43 0CB2 ಲ KANNADALETTERLA Consonant 110

44 0CB3 ಳ KANNADALETTERLLA Consonant 110

45 0CB5 ವ KANNADALETTERVA Consonant 110

46 0CB6 ಶ KANNADALETTERSHA Consonant 110

47 0CB7 ಷ KANNADALETTERSSA Consonant 110

48 0CB8 ಸ KANNADALETTERSA Consonant 110

49 0CB9 ಹ KANNADALETTERHA Consonant 110

50 0CBE ◌ಾ KANNADAVOWELSIGNAA

Matra 110

51 0CBF ◌ KANNADAVOWELSIGNI

Matra 110

52 0CC0 ◌ೕ KANNADAVOWELSIGNII

Matra 110

53 0CC1 ◌ು KANNADAVOWELSIGNU

Matra 110

54 0CC2 ◌ೂ KANNADAVOWELSIGNUU

Matra 110

55 0CC3 ◌ೃ KANNADAVOWELSIGNVOCALICR

Matra 110


14

Sr.No.

UnicodeCodePoint


56 0CC6 ◌ KANNADAVOWELSIGNE

Matra 110

57 0CC7 ◌ೕ KANNADAVOWELSIGNEE

Matra 110

58 0CC8 ◌ೖ KANNADAVOWELSIGNAI

Matra 110

59 0CCA ◌ೂ KANNADAVOWELSIGNO

Matra 110

60 0CCB ◌ೂೕ KANNADAVOWELSIGNOO

Matra 110

61 0CCC ◌ KANNADAVOWELSIGNAU

Matra 110

62 0CCD ◌ KANNADASIGNVIRAMA

Halant/Virama

110

Table3:Codepointrepertoire

5.3Codepointsnotincluded Followingcodepointshavenotbeenincludedintherepertoire.

Sr.No.

UnicodeCodePoint

Glyph CharacterName ReasonforExclusion

1. 0C8C ಌ KANNADALETTERVOCALICL

NotusedinKannada

2. 0CB1 ಱ KANNADALETTERRRA Obsoletecharacter,notusedinmodernKannada

3. 0CBC ◌ KANNADASIGNNUKTA DoesnotbelongtoKannada,notneededinLGR

4. 0CC4 ◌ೄ KANNADAVOWELSIGNVOCALICRR

NotusedinKannada

5. 0CD5 ◌ೕ KANNADALENGTHMARK Notinuse

6. 0CD6 ◌ೖ KANNADAAILENGTHMARK

Notinuse

Table4:Codepointnotincluded


15

6. Variants

6.1In-scriptvariants

TherearenovariantswithintheKannadascript,whentheformationofalabelisgovernedbytheWholeLabelEvaluationrulesinsection7.

6.2Cross-scriptvariantsanalysis

SomecharactersofKannadalookthesameassomecharactersinDevanagari,Gujarati,Telugu,MalayalamandSinhala.Thetablesbelowlistthem.

6.2.1Cross-scriptvariantsforKannadaandTeluguanalysisTheTeluguandKannadacodepointsetsinTable5arecross-scriptvariantcodepoints.

VariantSet TeluguCodePoint KannadaCodePoint

1 ◌ం (0C02) ◌ಂ (0C82)

2 ◌ః (0C03) ◌ಃ (0C83)

3 అ (0C05) ಅ (0C85)

4 ఆ (0C06) ಆ (0C86)

5 ఇ (0C07) ಇ (0C87)

6 ఈ (0C08) ಈ (0C88)

7 ఐ (0C10) ಐ (0C90)

8 ఒ (0C12) ಒ (0C92)

9 ఓ (0C13) ಓ (0C93)

10 ఔ (0C14) ಔ (0C94)

11 ఖ (0C16) ಖ (0C96)

12 గ (0C17) ಗ (0C97)

13 జ (0C1C) ಜ (0C9C)

14 ఝ (0C1D) ಝ (0C9D)

15 ఞ (0C1E) ಞ (0C9E)

16 ట (0C1F) ಟ (0C9F)


16

Table5:TeluguandKannadacodepointanalysis

6.2.2Cross-scriptvariantsforKannadaandDevanagarianalysisVisargaistheonlypotentialvariantcodepointthatexhibitsshapesimilaritybetweentheKannadaandDevanagariscripts.However,asthisisacombiningmarkandtherearenoothervariantcodepointsbetweenthetwolanguages,itisnotdefinedasavariantcodepoint.

DevanagariCodePoint KannadaCodePoint

◌ः (0903) ◌ಃ (0C83)

Table6:DevanagariandKannadacodepointanalysis

17 ఠ (0C20) ಠ (0CA0)

18 డ (0C21) ಡ (0CA1)

19 ఢ (0C22) ಢ (0CA2)

20 ణ (0C23) ಣ (0CA3)

21 థ (0C25) ಥ (0CA5)

22 ద (0C26) ದ (0CA6)

23 ధ (0C27) ಧ (0CA7)

24 న (0C28) ನ (0CA8)

25 బ (0C2C) ಬ (0CAC)

26 భ (0C2D) ಭ (0CAD)

27 మ (0C2E) ಮ (0CAE)

28 య (0C2F) ಯ (0CAF)

29 ర (0C30) ರ (0CB0)

30 ల (0C32) ಲ (0CB2)

31 ళ (0C33) ಳ (0CB3)

32 ◌ (0C3F) ◌ (0CBF)

33 ◌ు (0C41) ◌ು (0CC1)

34 ◌ృ (0C43) ◌ೃ (0CC3)


17

6.2.3Cross-scriptvariantsforKannadaandGujaratianalysisVisargaistheonlypotentialvariantcodepointthatexhibitsshapesimilaritybetweentheKannadaandGujaratiscripts.However,asthisisacombiningmarkandtherearenoothervariantcodepointsbetweenthetwolanguages,itisnotdefinedasavariantcodepoint.

GujaratiCodePoint KannadaCodePoint

◌ઃ (0A83) ◌ಃ (0C83)

Table7:GujaratiandKannadacodepointanalysis

6.2.4Cross-scriptvariantsforKannadaandMalayalamanalysisAnusvaraandVisargaaretheonlypotentialvariantcodepointsthatexhibitshapesimilaritybetweentheKannadaandMalayalamscripts.However,astheyarecombiningmarksandtherearenoothervariantcodepointsbetweenthetwolanguages,theyarenotdefinedasvariantcodepoints.

KannadaCodePoint MalayalamCodePoint

◌ಂ (0C82) ◌ം (0D02)

◌ಃ (0C83) ◌ഃ (0D03)

Table8:KannadaandMalayalamcodepointanalysis

6.2.5Cross-scriptvariantsforKannadaandSinhalaanalysisInitiallyNBGPconsideredthreevariantsetsbetweenKannadascriptandSinhalascriptasshowninTable9.

VariantSet KannadaCodePoint SinhalaCodePoint 1 ◌ಂ (0C82) ◌ං (0D82)

2 ◌ಃ (0C83) ◌ඃ (0D83)

3 ರ (0CB0) ර(0DBB)

Table9:KannadaandSinhalacandidatevariantsetsTheSinhalaGenerationPanel(SinhalaGP)disagreedthatರ (0CB0)andර(0DBB)arevariantcodepoints.TheSinhalaGPalsodisagreedthatර(0DBB)andtherelevantTelugucodepointర (0C30),arevariantcodepoints.AsafollowupSinhalaGPandNBGPmetanddiscussedthesethreecodepoints.


18

KannadaCodePoint TeluguCodePoint SinhalaCodePoint

ರ (0CB0) ర (0C30) ර(0DBB)

Table10:Kannada,Telugu,andSinhalacandidatevariantsetThetwoGPsagreedthatthevariantrelationshipdoesnotexist.BasedontheconsensusofthetwoGPs,thissetofcharactershasbeenremovedfromthevariantsets.Therefore,AnusvaraandVisargaaretheonlypotentialvariantcodepointsthatexhibitshapesimilaritybetweentheKannadaandSinhalascripts.However,astheyarecombiningmarksandtherearenoothervariantcodepointsbetweenthetwolanguages,theyarenotdefinedasvariantcodepoints.

VariantSet KannadaCodePoint SinhalaCodePoint 1 ◌ಂ (0C82) ◌ං (0D82)

2 ◌ಃ (0C83) ◌ඃ (0D83)

Table11:KannadaandSinhalacodepointanalysis

7. WholeLabelEvaluationRules(WLE)ThestructureofwrittenKannadaisprovidedinsection3.4.usingthefollowingvariablesordefinitions:

V → VowelM→ MatraC → ConsonantH → Halant/ViramaB → AnusvaraX → Visarga

Rulesforformingaksharoraclusterofoneunitaregivenbelow:

1. Rule1:HmustbeprecededbyC2. Rule2:MmustbeprecededbyC3. Rule3:BmustbeprecededbyC,VorM4. Rule4:XmustbeprecededbyC,VorM5. Rule5:VcannotbeprecededbyH

8. Contributors1.Dr.U.B.Pavanaja,[email protected].,[email protected],[email protected]


19

9. References[MSR] IntegrationPanel,"MaximalStartingRepertoire—MSR-4Overviewand

Rationale",7February2019https://www.icann.org/en/system/files/files/msr-4-overview-25jan19-en.pdf(Accessedon18February2019)

[NBGP]Neo-BrahmiGenerationPanel[101] AComparativeGrammaroftheDravidianorSouthIndianFamilyofLanguages

byRobertCaldwell,Trubner&Co,London,SecondEdition,1875[102] EvolutionofKannadascript-https://karnatakaitihasaacademy.org/karnataka-

history/evolution-of-kannada-script/(Accessedon18February2019)[103] HistoryoftheKannadaScriptandLanguage-

https://bookstalkist.com/history-of-the-kannada-script-and-language/(Accessedon18February2019)

[104] HistoryoftheKannadaLiterature-http://kamat.com/kalranga/kar/literature/history1.htm(Accessedon18February2019)

[105] Kannadaalphabet-https://en.wikipedia.org/wiki/Kannada_alphabet(Accessedon18February2019)

[106] AboutKannadalanguage-https://en.wikipedia.org/wiki/Kannada(Accessedon18February2019)

[107] EthnologueentryaboutKannada-http://www.ethnologue.com/19/language/kan/(Accessedon18February2019)

[108] OLACresourcesinandabouttheKannadalanguage-http://www.language-archives.org/language/kan(Accessedon18February2019)

[109] EncyclopaediaBritannicaentryaboutKannada-https://www.britannica.com/topic/Kannada-language(Accessedon18February2019)

[110] ಕನ#ಡ ಮಧzಮ {ಾzಕರಣ, %ೕ.ನಂ. }&ೕಕಂಠಯz, ~ೕ�ಾ ಬುh ��, �ೖಸೂರು, ೨೦೦೧

(KannadaMadhyamaVyakarana(meansAnIntermediateKannadaGrammar),T.

N.Sreekantaiya,GeethaBookHouse,Mysore,2001.)[111] PubliccommentfeedbackforKannada,Telugu,OriyaScriptLGRProposals,

https://docs.google.com/document/d/1m9MbBfNBQZAFc9SOYpt0lgeeyM3N-DsUP173J4Vb948(Accessedon18February2019)


20

AppendixI:ConfusableCodePointsAnalysis

A-1.KannadaandTelugu

SomecodepointsforKannadaandTeluguscriptwerediscussedandtheNBGPconcludedthatthesearenotconfusablecodepoints.Thetablebelowlistsalldiscussedcodepointsandtheirresolution.

Ser.No.

KannadaCharacter

Unicodecodepoint

Telugucharacter

Unicodecodepoint

NBGPResolution

1 ಎ 0C8E ఎ 0C0E Distinguishable

2 ಘ 0C98 ఘ 0C18 Distinguishable

3 ಙ 0C99 ఙ 0C19 Distinguishable

4 ಚ 0C9A చ 0C1A Distinguishable

5 ಛ 0C9B ఛ 0C1B Distinguishable

6 ಪ 0CAA ప 0C2A Distinguishable

7 ಫ 0CAB ఫ 0C2B Distinguishable

8 ವ 0CB5 వ 0C35 Similar

9 ಶ 0CB6 శ 0C36 Similar

110 ಷ 0CB7 ష 0C37 Distinguishable

11 ಸ 0CB8 స 0C38 Similar

12 ◌ 0CCC ◌ 0C4C Distinguishable

TableA-1:NBGPresolutionofTeluguandKannadacodepoints


21

A-2.KannadaandDevanagari

ThefollowingtabledefinesKannadaandDevanagaricodepointswhichareconfusable.

DevanagariCodePoint KannadaCodePoint

◌ः (0903) ◌ಃ (0C83)

TableA-2:DevanagariandKannadaconfusablecodepoints

A-3.KannadaandGujarati

ThefollowingtabledefinesKannadaandGujaraticodepointswhichareconfusable.

GujaratiCodePoint KannadaCodePoint

◌ઃ (0A83) ◌ಃ (0C83)

TableA-3:GujaratiandKannadaconfusablecodepoints

A-4.KannadaandMalayalam

ThefollowingtabledefinesKannadaandMalayalamcodepointswhichareconfusable.

KannadaCodePoint MalayalamCodePoint

◌ಂ (0C82) ◌ം (0D02)

◌ಃ (0C83) ◌ഃ (0D03)

TableA-4:KannadaandMalayalamconfusablecodepoints

A-5.KannadaandSinhala

ThefollowingtabledefinesKannadaandSinhalacodepointswhichareconfusable.

KannadaCodePoint SinhalaCodePoint

◌ಂ (0C82) ◌ං (0D82)

◌ಃ (0C83) ◌ඃ (0D83)

TableA-5:KannadaandSinhalaconfusablecodepoints


22

AppendixII:Zero WidthJoinerandZeroWidthNonJoinerinKannadadomainnames Zero Width Joiner (ZWJ) and Zero Width Non Joiner (ZWNJ) have special roles inKannada.ZWNJisusedinsequenceslikeConsonant(C)+Halant(U+0CCD)+ConsonantkindwherethesecondCshouldnottakethevattuformafter(below)thefirstconsonant.Example:1.ಜ(U+0C9C) + ◌ +(U+0CCD)+ ಕ(U+0C95) = ಜr–withoutusingZWNJ 2.ಜ(U+0C9C) + ◌ +(U+0CCD)+ ZWNJ (U+200C) +ಕ(U+0C95) = � ಕ –usingZWNJ ExampleofthewordnotusingZWNJ–*ಾಜುr-ಾ� (Rajkumar)ExampleofwordusingZWNJ–*ಾ� ಕು-ಾ� (Rajkumar)Bothwordsmeanthesame.Buttheirpronunciationsareslightlydifferent.ThesecondformdoesnotoriginallybelongtoKannada,butnowadaysduetotheinfluenceofEnglishandHindi,somepeopleareusingtheform.ButthisformisneededtowritemanyEnglishwordsinKannadalikesoftware(�ಾ�� {ೕ�,usingZWNJ). Thewordsoftwarewillbecome�ಾ��ೕ� ifZWNJisnotused.ZWJ has a unique purpose in Kannadawhere conjuncts are formedwith consonantswherethefirstconsonantisರ (Ra)(U+0CB0).WhenconjunctisformedwithRa+Halant+<consonant>,thedefaultformissecond<consonant>followedbyarkavattu,whichisKannadaequivalentoftherephofDevanagari.Example:ರ (U+0CB0)+ ◌ (U+0CCD)+ ಕ(U+0C95)= ಕ/

OriginalKannadadidnothavethisarkavattuorreph.ItwaslikeanyotherC1+Halant+C2wheretheC2takesthevattuform,evenwhenC1=ರ (Ra).Togetthevattuformforರ + Halant+<consonant>,ZWJisused.Example:ರ (U+0CB0) + ◌ (U+0CCD) + ZWJ (U+200D) + ಕ (U+0C95) = ರ r –Microsoft’simplementationರ (U+0CB0) + ZWJ (U+200D) +◌ (U+0CCD) +ಕ (U+0C95) =ರ r – as perUnicode’sdefinition.BrowsersFirefoxandChromeusethisrenderingrule.TheoriginalKannadaform(notusingarkavattuorreph)isneededwhenthefirstletteroftheworditselfisರ (Ra,U+0CB0)followedbyHalantandaconsonant.Example- ರ (U+0CB0)+ Halant (U+0CCD)+ ಯ (U+0CAF) = ಯ/ as in�ಾ/� (rallywritteninKannada).ThisiswrongasperthewritingstylefollowedinKannada.Herethe


23

secondconsonantಯ (Ya,U+0CAF)musttakethevattuform.TogetthatonemustuseZWJasexplainedabove.CorrectformofrallywritteninKannadawillbe*ಾ z�makinguseofZWJ.OnesuchwordIcanthinkofwillbethecompanyRallisIndia.WhentheygoforregisteringdomainnameinKannada, ifZWJ isnotallowed indomainname, theywillhavetogofor�ಾ/�ೕ�.�ಾರತ,whichisthewrongforminKannada.Theyshouldwriteitas*ಾ z�ೕ�.�ಾರತ bymakinguseofZWJ.ThusbothZWJandZWNJareneededforhavingproperdomainnamesinKannada.HowtoavoidduplicatedomainnamesinvolvingZWJandZWNJ?ZWJandZWNJareusedmainlytohavetwodisplayformsofwhatislinguisticallysamewordor combinationof characters inKannada.WhenZWJandZWNJsareallowed indomainnames forKannada, itwill create twodomainnameswhichhave twodisplayformsbut linguistically theyaresame.TomakethebrowsersandDNSstreat themasequal,wehavetoignoreZWJandZWNJsforcomparingtwowords.Thismethodologyisfollowedbythespell-checkusedinMicrosoftWord.Thesamephilosophycanbeappliedherealso.AcceptingZWJandZWNJindomainnamescreatesconfusiontoamajorityofthelinguisticcommunityandjoinercharactersareprohibitedfortheRootZone,hencethisisexplicitlyprohibitedbytheNBGP.


24

AppendixIII:NBGPCross-scriptVariantInclusionPolicy

If,inanytwogivenscripts,allthepotentialcross-scriptvariantsconsistofdependent(e.g.VowelSigns,Anusvara,Visarga,Chandrabinduetc.)charactersONLY,thenthatentiresetcanbeignoredandnocross-scriptvariantsbeproposedbetweenthosetwoscripts.

If,inanytwogivenscripts,thereisATLEASTONEnon-dependent(e.g.Consonant,Voweletc.)cross-scriptvariantcharacter/sequencepresent,allthepotentialcross-scriptvariantsshouldbeconsideredandproposedbetweenthetwoscripts.Thiscross-scriptanalysishasbeenrestrictedtothescriptsthathavedescendedfromtheBrahmiasmostofthemsharesimilarusagepatterns.Byandlarge,allofthesescriptshaveacommonsetofcharactersthatexistedinBrahmiscriptandbearthesameidentities.However,asthescriptsbranchedoutfromtheBrahmi,dependingonvariousfactors,theshapesofthecharacterschanged.Thischangeintheshapewasnotuniformacrossallthecharactersandthescripts.Somecharactersshapesdidchangesignificantly,whereassomeofthemstillretainedsimilarity.Thecross-scriptsimilarityanalysisalsoaimstoidentifysuchcaseswherethesamecharacterretainedalmostthesameshapedespitebeingpartofthedifferentscripts.Thesesetofcharactersarevariantsofeachotherintruesensethanmerelyofco-incidentalvisualsimilarity.Sincehavingsuchlabelsisarealisticpossibilityandthecorrespondinglabelslookalmostexactlyalike,NBGPhasproposedthemasblockedvariants.

NBGPacknowledgestheconcernthatthisshapeisquitegenericandmayhaveparallelsinotherscriptsnotunderitsambit.However,asNBGPdoesnothaveanyexposureaboutactualusageofthose characters in those particular scripts, NBGP desisted from including them in theanalysis.AsNBGPhasalreadyconsideredalltherelatedscriptsunderthecross-scriptvariantanalysis,thesimilarityofthecharactersbelongingtoNBGPscriptswithotherscriptsnotundertheNBGPambit,maybeofamereco-incidentalvisualnature.

Additionally,thisconcernisnotlimitedtothesetwocharactersbutforallthecharactersinallthescriptsunderthescopeoftheRootLGRprocedure.CarryingoutthisanalysiscanpracticallybedoneonlywiththeGenerationPanelsthatexistwhiletheNBGPisactive.ThisstillleavesoutthosescriptsoutofthescopewhichmaynothaveaGenerationPanelestablishedyet.Hence,carryingoutthisexerciseinentiretyisquiteimpracticable.Thisconundrumcanberesolvedifallsuchcasesarehandledbythe"StringSimilarityAssessmentPanel"ofICANN.

Proposal for a Kannada Script Root Zone Label Generation ... · dynasty, in the 6th century A.D.,...

Documents

Transcript of Proposal for a Kannada Script Root Zone Label Generation ... · dynasty, in the 6th century A.D.,...