i排版

InternationalWebSites

Internationalization&UnicodeConference1SanJose,USA,September2008

TheInternationalizationTagSet

Internationalization&UnicodeConference2SanJose,USA,September2008

InternationalWebSites

Thegoalofthistaskforceisthecreationofadocumentaboutrequirements

flJltlidiththlilikCSSSVGd

Internationalization&UnicodeConference3SanJose,USA,September2008

forgeneralJapaneselayoutrealizedwithtechnologieslikeCSS,SVGand

umentwillbemainlybasedonastandardforJapanese

layout,r,itwilladdressalsoareaswhicharenot

umentiscurrentlyindraftstageandis

beingdevelopedfurtherbytheJapaneseparticipantsinthetaskforce.

Thetaskforcewillgatherandintegratefeedbackfromtheparticipating

workinggroupsabouttheneedandtechnicalfeasabilityofvarious

requirements.

TheoutputofthetaskforcewillbeanEnglishdocument(aWorkingGroup

Note)criptionwillnot

sumedthattheparticipating

workinggroups(seebelow)willdecideindependentlyhow(andif)to

implementtherequirementsintheirspecifications.

InternationalWebSites

Internationalization&UnicodeConference4SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference5SanJose,USA,September2008

InternationalWebSites

TheKihonhanmenistheareaofthepagethatcontainsthemainbodyof

thJttItd'tildhddft

Internationalization&UnicodeConference6SanJose,USA,September2008

n'tincludepageheadersandfooters.

Thedocumentintheslideisverticallysetandtypicalofthelayoutofa

Japanesenovel.

Inordertounderstandwhythekihonhanmenisofinterest,wemustfirst

understandsomeofthecharacteristicsofJapanesecharactersandtheir

use.

InternationalWebSites

Japanesekanji,hiraganaandkatakanacharactersnormallysitinsmall

bthtllthiThildtlfft

Internationalization&UnicodeConference7SanJose,USA,September2008

adstoaveryregulareffect.

EvencharacterssuchastheJapaneseperiodandcommashownonthe

slidesitwithinasquareboxofthesamesize.

Therearenospacesbetweencharactersintheslideabove.

InternationalWebSites

ould

bdthhthttbthtilldhitllh

Internationalization&UnicodeConference8SanJose,USA,September2008

bedrawnthroughthetextbothverticallyandhorizontally,asshown.

InternationalWebSites

Inreality,edLatincharacters,rules

bththtbidlittdtdll

Internationalization&UnicodeConference9SanJose,USA,September2008

aboutwhatcharacterscanbeginorendaline,etc,tendtoproducesmall

heless,

wherecharacterscomeadriftfromthegridcells,thereisastrongtendency

achievedusing

ridlayoutisnotrecoveredbefore,thelast

characterboxonalinewillendflushwiththebottomofthegrid.

InternationalWebSites

Actually,setextusuallyhas

itliitillifhlfhtbiidth

Internationalization&UnicodeConference10SanJose,USA,September2008

someinterlinespacing,typicallyrangingfromhalfacharacterboxinwidth,

forshortlines,toafullcharacterboxforlongerlines.

Thistendencyforcharactersonapagetofittoagridmakesthekihon

hanmenalittlespecialinacoupleofways,aswewillnowsee.

InternationalWebSites

AtypicallyWesternwayofcreatingthemaintextregiononapageinvolves

jtifitifiFlttildti

Internationalization&UnicodeConference11SanJose,USA,September2008

mple,tocreateasimpledocumentin

XSL-FO,afterspecifyingtheprintableareaofthepageusingpagemargins,

youspecifythesize,withintheprintablearea,ofthemainbodyregionby

specifyingitsmargins.(Thosemarginscanthenbeusedforpageheaders

andfooters.)

Thekeypointisthatyoudon'tusuallyspecifythesizeofthebodyregion

eofthebodyregioniswhateverisleftoverafteryouhave

rwords,youspecifythesizeoftheblank

space,thenjustflowthetextintotheremainingspace.

InternationalWebSites

Becausethekihonhanmenisagrid,Japanesepeoplethinkofpagelayout

lihtldifftl

Internationalization&UnicodeConference12SanJose,USA,September2008

slightlydifferently.

Theirnaturalinclinationistofirstdefinethesizeofthemaintextarea(ie.

thekihonhanmen).Onlyafterthatarethemarginsconsidered.

InternationalWebSites

Forapagewithasinglecolumn,definingthesizeofthekihonhanmen

ilhihtiifithlthfthli

Internationalization&UnicodeConference13SanJose,USA,September2008

involveschoosingacharactersize,specifyingthelengthofthelinesasa

numberofcharacters,thenspecifyingtheotherdimensionofthetextarea

asthenumberoflinesperpageandthesizeoftheinterlinegap.(There

arepreferredsizesandratiosforcharacters,linelengthsandlinesper

page.)

Fordocumentswithmultiplecolumns,thesizeofthegapbetweenthe

columnsisalsotakenintoaccount.

InternationalWebSites

Strategiesfordeterminingmargins.

Internationalization&UnicodeConference14SanJose,USA,September2008

Thecommondefaultistocenterthekihonhanmeninthebodyregion.

Otheralternativesinclude:

yingthetopspace(forhorizontallysetpages)orthebottom(for

verticallysetpages),andcenteringthesides.

yingthesizeoftheguttertotheside,andcenteringthetopand

bottom.

reof1and2.

Insummary,tospecifythepositionofthemaintextflowinasimple

documentusingkihonhanmenyouwouldspecifythedimensionsofthe

maintextregionitself,andzerototwomargins.

TherehasbeendiscussionattheW3Caboutmakingiteasiertospecifythe

bodyregioninthiswayinXSL-FO2.0.

InternationalWebSites

Anothersignificantcharacteristicofthekihonhanmenasatypeofbody

ibktit'idlikt

Internationalization&UnicodeConference15SanJose,USA,September2008

regiongoesbacktoit'sgrid-likenature.

Elementspositionedwithinthekihonhanmentendtosnaptothelogical

gridthatunderliesthisregion.

Forexample,thefirstlineofaparagraphstendstobeindented,andthe

usualindentationisonecharacterbox.

Headingsinthedocumentalsotendtobepositionedrelativetothegrid,

eventhoughthesizeofthecharactersinthetitleislargerthanthatused

forthelogicalgridofthekihonhanmen.

Figuresandsurroundingtexttendtobepositionedrelativetothekihon

hanmengrid.

InternationalWebSites

Internationalization&UnicodeConference16SanJose,USA,September2008

InternationalWebSites

Youmaythinkthatyouwouldbeabletoswitchbetweenhorizontaland

tiltttthflikfCSSithIhthtll

Internationalization&UnicodeConference17SanJose,USA,September2008

aveapagethatallowsyou

todothat,usingsomeJavaScriptthatchangesthewriting-modeproperty

onablockofJapanesetext(see

/scripts/samples/-requiresInternetExplorer

5.5+).

Therearesomedifferencesbetweentextinverticalandhorizontallayouts

thatcanbedealtwithautomatically,butthereareothersthatarenotso

easytohandle.

Thechangeshownontheslideisbroughtaboutbysimplychangingthe

valueofthewriting-modepropertyontheenclosingblockfromlr-tb(left-to-

rightblockflow,top-to-bottomlineprogression)totb-rl(top-to-bottom

blockflow,right-to-leftlineprogression).

InternationalWebSites

ThisslideshowssomethingsthatalreadyworkinInternetExplorer,using

thiilith

Internationalization&UnicodeConference18SanJose,USA,September2008

thissimpleswitch.

TheJapanesecommaandfull-stopmovefromthebottomleftofthe

achievedbychoosingavertical-

ractercodesremainthesame.

Soundelongationmarksandpunctuationsuchasparenthesesarerotated.

RunsofLatintextflowdownthepageinverticallines.

Rubytextmovesfromabovethehorizontallinetotherightofthevertical

line.

InternationalWebSites

Unfortunately,thereareotheraspectsofthetextthatshouldhave

hdbtdid'tFlbldhlkdbtt

Internationalization&UnicodeConference19SanJose,USA,September2008

changed,butdidn'mple,somenumberswouldhavelookedbetter

askanjicharacters,ratherthanWesterndigits,inverticaltext;other

numbersshouldrunhorizontallywithintheverticallineprogression;and

someLatintextthatisusedasanacronymwouldlookbetterifthe

characterswerenotrotated.

nbeworkedaround,others

'stakealookatsomeofthecommondifferences

betweenhorizontallyandverticallysetJapanesetext.

InternationalWebSites

ALatin-scriptacronymorsingleletteristypicallyrenderedwithoutrotation

itilttItlfllidthhhhitlttill

Internationalization&UnicodeConference20SanJose,USA,September2008

usesafull-widthshape,whereashorizontaltextwill

typicallyuseproportionally-spacedcharacters.

Shorttwo-digitnumbersareoftenhorizontallysetwithintheverticalline

(thisiscalledtate-chuu-yoko).

Inbothofthesecases,youwouldneedtohavemarkupinthehorizontal

texttoautomaticallyachievethedesiredresultwhentheswitchtovertical

isflipped.

InternationalWebSites

Punctuationcharacterstendtobeuseddifferentlyinhorizontalandvertical

tt

Internationalization&UnicodeConference21SanJose,USA,September2008

text.

DuetoWesterninfluence,JapaneseauthorssometimesuseWestern

altext,ontheotherhand,should

onlyusetheideographiccommaandfullstop.

InternationalWebSites

Similarly,therearesomedifferencesintheusageofquotationmarksin

hitltilttFlhitlttftdbl

Internationalization&UnicodeConference22SanJose,USA,September2008

mple,horizontaltextoftenusesdouble

orsinglequotationmarksasanalternativetoleftandrightcornerbrackets.

Verticaltext,ontheotherhand,doesn'tusethequotationmarks(except

sometimesforembeddedLatinruns);itusescornerbracketsordouble

primemarksarenotusedforhorizontaltext.

drighttortoiseshellbrackets(︹

︺)areaverticalequivalentofleftandrightsquarebracket(［］).Square

bracketsshouldbeusedinhorizontaltextexceptforspecialcases.

InternationalWebSites

Ideographicnumeralsweretraditionallyusedinverticallysettext,instead

fWtAbil(Rdbdbl

Internationalization&UnicodeConference23SanJose,USA,September2008

ofWestern-Arabicnumerals.(Roadnumbersandcarnumbersareexamples

ofexceptions).However,newspapersandotherpublicationshavebeen

adoptingWestern-Arabicnumeralsmoreinverticalwritingmode.

Inverticallysettext,symbolsforunitsareusuallydescribedwithkatakana

characters,eg.センチメートル(centimeter)orセンチ(abbreviatedformfor

centimeter).Inhorizontallysettext,theInternationalSystemofUnits(SI)is

usuallyused,eg."cm".

InternationalWebSites

Whereanewpageisstartedandthecurrentpageisn'tfulloftext,there

bdifftttifbliltht

Internationalization&UnicodeConference24SanJose,USA,September2008

maybedifferentstrategiesforbalancingcolumnsonthecurrentpage.

Typically,textcolumnsinverticallysettextendwithoutanybalancing.

Columnsinhorizontallysettext,however,aretypicallybalancedsothatthe

twocolumnsarethesamelength.

InternationalWebSites

Therearealsodifferencesinpagelayoutdetailsbetweenhorizontallyand

tilltthttttillhid

Internationalization&UnicodeConference25SanJose,USA,September2008

verticallysetpagesthatarenotautomaticallyachieved.

Forexample,whenpositioningrunningheadsandpagenumbers

horizontallywithreferencetokihonhanmeninverticalwritingmode,the

amountofverticalspacebetweentheedgeofkihonhanmenandthe

runningheadisonefullwidthcharactersizebasedonthekihonhanmen

nhanmenishorizontallyset,youneedtotakemore

verticalspacethanthecharactersizeinkihonhanmen.

InternationalWebSites

Typographicconventionssuchaswarichuaremuchmorecommonin

tilltttthhitl

Internationalization&UnicodeConference26SanJose,USA,September2008

verticallysettextthanhorizontal.

InternationalWebSites

Internationalization&UnicodeConference27SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference28SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference29SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference30SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference31SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference32SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference33SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference34SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference35SanJose,USA,September2008

RichardIshida35

Japaneseisamultisyllabiclanguagethatisverydifferentthatmono

llbitlChibtlkiitfthithJ

Internationalization&UnicodeConference36SanJose,USA,September2008

syllabic,tonalChinese,butlackingascriptoftheirown,theJapanese

isproved

notadequatetothetasktheyaddedtwophoneticscripts(firsthiragana

andthenkatakana).Today,hiraganaisusedtoexpressthingslikeverband

adjectiveinflections(endings)(lackinginchinese)andtowritewordsthat

lackedChinesecharacter(s).And,katakanaisusedtoexpresswords

importedinthe19thcenturyorlaterandforpropernamesand/oremphasis.

Finally,thelatinalphabetisusedtowriteJapaneseinaformthatforeigners

mightunderstand.

EnglishiscomposedofwordsderivedfromAnglo-Saxon(pig,sheep)andNorman

French(porkmutton)SimilarlyJapanesehaswordsfromthenativespoken

Internationalization&UnicodeConference37SanJose,USA,September2008

French(pork,mutton).Similarly,Japanesehaswordsfromthenativespoken

langauge,wordsimportedfromChina,andwordsimportedfromWestern(orother

non-chinese)countries.

llyasingle

kanjiisassignedtoeachJapaneseword,independentlyofhowmanysyllablesis

ve“higashi”(3syllables)for“east”and“yama”(2syllables)for

“mountain”.SinceatthetimetheChinesescript(Kanji)wasadoptedforwriting

Japanese,ittypicallyhadwords,inChinese,forallthethingsJapanesehadwords

for,therearealsoSino-Japanesepronunciationsforthesamekanjithatrepresent

ereis“Tō”for“east”and“san”for“nese

“readings”ofthekanjicharactersarenormallymono-syllabic(buttheyhavelostthe

associatedChinesetone).

SinceChinesewasmoresophisticatedthatJapaneseatthetimeofborrowingof

Kanji,therearewordsthathavemultiplekanjiintheir“spelling”.Therecompound

wordsarecalled“jukugo”ividualscharactersin“futon”:fu

(character737)means“spread”or“cloth”and“ton”(character243)means“group

ortroupe”.Sotheliteralcombinationwouldbe“groupofcloth”whichisanOK

descriptionofa“mattress”whichiswhat“futon”meansasacompoundword.

AnotherexampleinJapaneseisthepairShinto(神道Shintō)meaningthe

“path/wayofthegods”and

ThewordShinto,fromtheoriginalChineseShendao(神道),[1]

combines

twokanji:"shen"(神"shen")(Chineseloanwordsusuallyretaintheir

Chinesepronunciation,hence“shin”not“kami”theJapanesefor“gods”),

meaninggodsorspirits;and"dao"(道"dao"),meaningaphilosophicalway

orpath.

Therubysays,“ku”“n”“shi”(ha)“wa”(shi)(te)“do”“u”(ze)(zu)

Internationalization&UnicodeConference38SanJose,USA,September2008

wheretherubyisquotedandthebasekanaisinparentheses

‘kun’(character479)isanhonorific

‘shi’(character61)meanschildorperson

sotogethertheymean“wiseman”,“manofvirtueorhighrank”

“wa”(character277)means‘peace’,“harmony”,or‘Japan”

“dou”(character185)seemstomean“same”(butnormallywithoutthe“u”

part)

Theimaginarybox(inwhichthecharacteriscentered)isdrawntomore

llhthitiifthbhtltitthb

Internationalization&UnicodeConference39SanJose,USA,September2008

clearlyshowthepositioningoftherubycharactersrelativetothebase

character.

Thehiraganaseemstosay,“tonde”

Therearetwoapproachestopositioningtherubycharacterswithrespectto

thbhtIfthiddbfhtith

Internationalization&UnicodeConference40SanJose,USA,September2008

eisanoddnumberofcharacters,asisthecase

here,thenthecentercharactercanbealignedwiththecenterofthebase

called,“nakatsuki”alignment.

Alternatively,theinitialcharacterofthesequenceofrubycharacterscanbe

called,“katatsuki’

proachwasmoreoftenusedwith(hot)metaltypesetting

andwithdigitaltextlayout,thenakatsukiapproachismorepopular.

Notethatiftherearetworubycharactersattachedtothebasecharacter,

thenbothapproachesgivethesameresultsaslongasthesizeoftheruby

charactersare½thesizeofthebasecharacters.

Therubyannotationis:“shi’“yu”“n”(?)

Internationalization&UnicodeConference41SanJose,USA,September2008

Goingfromlefttoright:

Thefirstsequencehaskanabeforeandaftertheannotatedkanjisooverlap

isallowedoneitherside,buttheinkatatsukithefirstrubycharacteris

normallyalignedwiththestartedgeofthebasecharacters

Thesecondsequenceasakanjicharacter(hand)followingtheannotated

kanjisotheonlyallowableoverlapiswiththehiraganacharacterthatis

beforetheannotatedbasecharacter

Thethirdsequencehasakanjibeforetheannotatedbasecharacterand

kanaafterwords,sothenormalkatatsukiruleworks(actuallyisforced)in

thiscase.

Inthefourthandfinalsequence,therearekanjicharactersbeforeandafter

theannotatedbasecharacterand,therefore,itisnecessaryto(a)use

nakatsukialignmentand(b)introducespacebetweentheannotatedbase

akatsuki

alignmentmeansthattheextraspaceisdistributedequallyonbothsidesof

theannotatedbasecharacter.

InternationalWebSites

Internationalization&UnicodeConference42SanJose,USA,September2008

Inthelefthandset,thekanjiisgivenanEnglishmeaning,“editor”.Note

thtthltihtfttd/jtifidthllldb

Internationalization&UnicodeConference43SanJose,USA,September2008

thatthelatincharactersareformatted/justifiedasthenormallywouldbe

ratherthanspreadingthemacrossthewhole“width”ofthekanjibase

charactersequence.

Intherighthandset,thepronunciationoftheEnglishword,“editor”isgiven

inkatakana,whichisevenlydistributedacrossthe“width”ofthebase

(romanji)characters.

TheexamplehereisintwolinesofJapaneseandisreadfromrighttoleft

till

Internationalization&UnicodeConference44SanJose,USA,September2008

vertically

Thefirstjukugo,theoneonthetopright,hasonlythreerubycharactersin

itsannotation:oneonthefirstbasecharaacterandtwoonthesecondbase

character

Thesecondjukugo,theoneonthebottomright,hastworubycharacaters

oneachofitstwobasecharacters

Thethirdjukugo,theoneontheleft,hasthreerubycharactersonitsfirst

basecharacter,butonlyonerubycharacteronitssecondbasecharacter.

Becausetherubycannotoverlapakanjicharacter,theremustbeaspace

,thisissomewhatugly.

so,nextslide

Inthisslide,wehaverepeatedthemonorubysettingofjukugoruby,but

hthtbtiflttiltfttithjkb

Internationalization&UnicodeConference45SanJose,USA,September2008

showthatamorebeautifulsettingresultsfromtreatingthejukugorubyas

ifitweregroupruby.

Herearetwoexamplesofjukugotextwithattachedrubyannotations.

Internationalization&UnicodeConference46SanJose,USA,September2008

Inthefirstexample,theleftone,thefirstbasecharacterhasthree

attachedrubycharactersandthesecondbasecharacteronlyhasone

attachedrubycharacter.

Inthesecondexample,therightone,thefirstbasecharacterhasonlyone

attachedrubycharacterandthesecondbasecharacteronlyhasthree

attachedrubycharacter

Inbothcases,thefirstbasecharacteranditsattachedrubycharaactersare

showninblueandthesecondbasecharactersofthejukugoandits

attachedrubycharactersareshowninblack.

Thepairofexamplesontheleftshowasettingofbothexamlesusingthe

ondpairofexamplesshowthattherulesfor

ugo

atitdoesnotmatter

whichcharacterhas3rubycharactersattachedaslongastheother

characterhasonlyonerubyattached.

So,whathappensifthejukugobasecharactershavemorethatatotalof4

rubycharactersattached?

Asforgroupormonoruby,theoverlapsareonlyallowediftheadjacent

httkjithtithithl

Internationalization&UnicodeConference47SanJose,USA,September2008

charactersarenotkanji;thatisthecaseintheseexamples.

Onemajordifferencebetweenjukugoandgrouprubyisthatalinebreak

tithibFjkbthth

Internationalization&UnicodeConference48SanJose,USA,September2008

ugoruby,ontheother

hand,linebreakscanoccurandthelayoutrevertstothemonorubylayout

mpleshowsthebehaviorofour

previous1-3and3-1exampleswhenalinebreakoccurswithinthejukugo

basecharactersequence.

notethatwhenthereare3ormoremonorubycharactersatthestartor

endofaline,thenthebasecharactersmaybemoveddowntogiveroom,

withintheKihonhanmen,reotherwaysto

settherubycharactersinthatsituation,butthatisbeyondthescopeofthis

talk.

InternationalWebSites

Internationalization&UnicodeConference49SanJose,USA,September2008

RichardIshida49

InternationalWebSites

Internationalization&UnicodeConference50SanJose,USA,September2008

InternationalWebSites

Internationalization&UnicodeConference51SanJose,USA,September2008