WPC7  0}a 0xޥ 0j 0j 0z 0} 07 0Pg 0 0 0 H 0w 00[ 0B 0U 0" 0 0 08 0 0 0o 0G! 0!U9@" %" 1m" 0y@# 0# 1$ 0% & 0=& 0F& 0P' 1m' 0(( 0!V,Ew- 0}y- 0x- 0n. 02. 0,/U7U7e7 0b7 0M<UBU6B 0&NB 0&tB 0&B 0CB  C 0TC bD lDU*:vD CzDfDaD BD C)Df&Ea(E B *IRIfbIadIfxIazIIIfIaII 0J 1mK 0XL 0jL 07M 0N 0O 0EO 07QU8URUNRU FRU2!S 09SS 0DTT;YZfo_fq_fs_fu_fw_fy_f{_f}_f_f_f___a_ B__cd sffyff{f}f 0"f g gfgfgfgfgfgfgfgfgfgfgfgfgfgfgfgfgfgfgfgggagU8haGh[hkh}hahah3hjajakk "ka.kaBkVkahka|k kakakAkooo/oa=oaQoeoasoaooaoaooaoooapa%p9paKpa_pspapfppp %rk+rt ttttttt tauau *u:uaLu`upuauau uauau uau uavU(v#Dv'Sw zzzzzz N|X|\|ad|ax|U@|UN|!p!p1UN}U@h}UN}UN}D~˂!p!p!p!p!V! f!p!p!f!!p! !!p6!!p6!!p6!p!p!pZ!p!:!!pZ!!Z!!z!p !pz! !z !p !p!p!p!!P!p!@P$!0$!p$!0*!P*!p+!p0!5!6!p7!pP=!B!pPC!pH!0N!pN!0T!pT!0Z!Z!pP[!p`!0f!f! 3h!Sl!ct!su!v!w!#!p!ps!p!S!!s!p!ps!p! !p!p!! !!pX!p!p8!p!p!p!p!ph!p!p>!p!p!p!p !pn!p!pN!p"!p.(!p-!p3!p~8! =!pI!pO!T!pU!Z!p[!p`!pe!pgk!pp!pGv!p{!p'!p!p!p!p!pt!p!pT!pĭ!p4!p!!! 4!p!pT!p! 4!pT!!T )"  "E..0h..hz :Right Par[6]Right-Aligned Paragraph Numbers..` ..`  .. ..h..h..2(a)3  Ԁh..h0..z :Right Par[7]Right-Aligned Paragraph Numbers..` ..`  .. ..h..h....2i)3  Ԁ..0..z :Right Par[8]Right-Aligned Paragraph Numbers..` ..`  .. ..h..h....p..2a)3  Ԁ..0p..pVX8Document[1]Document Style  @..8  2I.3  Ԁ     Ԉ l2:Technical[5]Technical Document Style.. 2(1)3  Ԁ. l2:Technical[6]Technical Document Style.. 2(a)3  Ԁ. l/$:Technical[2]Technical Document Style 2A.3  Ԁ   .. l, :Technical[3]Technical Document Style 21.3  Ԁ   .. l( :Technical[4]Technical Document Style 2a.3  Ԁ   .. l:/:Technical[1]Technical Document Style  2I.3  Ԁ     .. l1:Technical[7]Technical Document Style.. 2i)3  Ԁ. l1:Technical[8]Technical Document Style.. 2a)3  Ԁ. \  `(Times NewRomanTTX35;AGOW_e11a.i.(1)(a)(i)1)a)R&8BibliogrphyBibliography0....jo4Tech InitInitialize Technical StyleS   1 .1 .1 .1 .1 .1 .1 .1 S CuyTechnicalTechnical Document Style11.11.1.11.1.1.11.1.1.1.11.1.1.1.1.11.1.1.1.1.1.11.1.1.1.1.1.1.1fp2Doc InitInitialize Document Style   S $    I. 1. A. a.(1)(a) i) a)S!"#($0 ($0 0 (($0 0 0   A_ekqwDocumentDocument StyleI.1.A.a.(1)(a)i)a)x?t2PleadingHeader for Numbered Pleading Paper &  '(  XX6A('*d66A('*d6\f '1\f '2\f '3\f '4\f '5\f '6\f '7\f '8\f '9  '10  '11  '12  '13  '14  '15  '16  '17  '18  '19  '20  '21  '22  '23  '24  '25  '26  '27  '28  .+(2$ {!7OXXd7      C  $#|}~C-- W p4Heading 2Underlined Heading Flush LeftV4Heading 1Centered Heading@..  b* 8Bullet ListIndented Bullet List0..0` ..` 20Exampler~XF:Squib ReviewRequest for review of squib-.*\ `-,|..LinguisticInquiry..SquibsandDiscussion..Editors:..|..|ChisatoKitagawa..|..|JohnMcCarthy(mccarthy@cs.umass.edu)..|..|MargaretSpeasDepartmentofLinguisticsUniversityofMassachusettsAmherst,MA01003USA-/<p`(-..|..|"US ,  0 July21,1994!  8.,4 <DL84..4TheenclosedmanuscripthasbeensubmittedforpublicationintheSquibsandDiscussionsectionofLinguisticInquiry.Wewouldappreciateitifyoucouldlookoverthemanuscriptandsendusabriefreviewofit.Inordertoensurethatwewillbeabletorespondtotheauthorwithinareasonableamountoftime,werequestthatyousendusyourreviewwithinthreeweeks.Ifyouarenotabletoreviewitwithinthreeweeks,pleasereturnitimmediatelyand,ifpossible,suggestanalternatereviewer.4..4Thankyouinadvanceforyourhelp.Forourconvenience,pleaseenclosethisletterwithyourreview.4..4____Publishasis.4..4____Donotpublish.4..4____Publishwithminorrevisions.4..4____Publishonlywithmajorrevisions.4..4____Inappropriateasasquib;subjectmatterrequiresgreaterlength.4..4Pleaseencloseadditionalcommentsonaseparatesheet.4..4 ..  .. <..<....D..DSincerely,4..4 ..  .. <..<....D..DJohnJ.McCarthy-/<p`(-*\ `<p`(<~bF:Squib Letter-.*\ `-,|..LinguisticInquiry..SquibsandDiscussion..Editors:..|..|ChisatoKitagawa..|..|JohnMcCarthy(mccarthy@cs.umass.edu)..|..|MargaretSpeasDepartmentofLinguisticsUniversityofMassachusettsAmherst,MA01003USA-/<p`(-..|..|"US ,  0 July21,1994!  8.,4 <DL84..4 ..  .. <..<....D..DSincerely,4..4 ..  .. <..<....D..DJohnJ.McCarthy-/<p`(-T _.LetterLinguistics Letter-/<p`(-   =    -3)\> `-UNIVERSITYOFMASSACHUSETTS(#..c-41 q`-DepartmentofLinguistics-3)\> `-ATAMHERST-41 q`-SouthCollege(#..wmccarthy@cs.umass.eduAmherst,MA01003(413)5456830-/<p`(-       (#.."US ,  0 July21,1994!  Ԉ  4..4 ..  .. <..<....D..D..Sincerely,4..4 ..  .. <..<....D..D..JohnJ.McCarthy4..4 ..  .. <..<....D..D..ProfessorofLinguistics-/<p`(-    )\> `1 q`&$1&$2&$3($    !( !2$ {!7OXXd7      C  $#|}~C-- W Ӏ  0  MUTZMUKI<6X9`("Courier NewTT dTable_A ddTable_B&0 d d dKAGITABLTable_CTable_DKUGINVOITable_ETable_F   %  C  $#|}~C--  9  .  ݀Itgoeswithoutsayingthatunderlyingrepresentationsbythemselvescontinuetobecrucialindeterminingtheclassofsurfacerepresentationsgeneratedbyanoptimalitytheoreticgrammarforagivenlanguage:unpredictablelexicalinformationisbydefinitionirreducible.FAITable_GTable_HTOMTable_I Table_JTOMBO5TOMBO10 Table_K Table_L< 9(*Courier NewBoldTTTOMBO11 Table_M Table_NTOMBO12Table_OTable_PTOMBO13Table_QTable_RLEXOPTLEXOPT2Table_S  '  C  $#|}~C--  ToappearinLinguisticInquiry26.4($7OXXd7      C  $#|}~C--}~#337=CIQYag1.a.i.(1)(a)(i)1)a)(0;$(2  0  3  )($0 2  A  3  0` (#(#($0 0` (#(#2  1  3  0 ` (#` (#(E$0 0` (#(#0 ` ` (#` (#2  a  3  0  (# (#($0 0` (#(#0 ` ` (#` (#0  (# (#0h(#(#2(3  0$h(#h(#($0 0` (#(#0 ` ` (#` (#0  (# (#0h(#(#0h#h(#h(#2  i  3  0)(#(#($0 0` (#(#0 ` ` (#` (#0  (# (#0h(#(#0h#h(#h(#0((#(#2  a  3  0pc.(#(#A Helv Regular9 6Times New Roman Regular9.Courier New RegularAHelv Bold(2 $ {!7OXXd7      C  $#|}~C-- W   0  .(&$  1     %  C  $#|}~C--  8  .  ݀ForanyFillFeatviolationassociatedwithasingleFillLinkviolation(andsimilarly,for  anyParseFeatviolationassociatedwithasingleParseLinkviolation),wehaveadoptedthe |] policyofincludingonlyonemarkintheFaithcolumn(seee.g.(  MUKI27  bc)and( KAGITABL28  a)).Besides  expositoryconvenience,thispracticeavoidsanunwarrantedreificationofassociationlines(cf.alsothefeaturesasattributesviewascribedtoJ.PierrehumbertinMcCarthy&Prince1994b).ThepointoftheFaithLinkconstraints,asseparatefromFaithFeat,istomilitateagainstthe    establishment(ornonparsing)ofadditionalmultipleassociations,seeforexample( TOM31  c)below. M".  Initsfocusonspreadingconfigurations,itcorrespondstothe*SpreadconstraintofIt=,Mester #  &Padgett1993. 1  %  C  $#|}~C--  7  .  ݀SeealsoMcCarthy1993forasomewhatdifferenttheoryappealingdirectlytomorphemicaffiliation(MSeg). y  %  C  $#|}~C--  6  .  ݀HerewefollowPS1993inassumingthatnoelementmaybeliterallyremovedfromtheinput:i.e.,thegeneratorfulfillstheconditionthatinputsareliterallycontainedinallassociatedoutputs(amonotonicityrequirementdubbed"Containment"inMcCarthy&Prince1993:20).Hencethenotionfailuretoparse,incontrasttoremoval,ofanelement.ThereareanumberofalternativepossibilitiesforencodinginputoutputrelationsinOptimalityTheory(seeforexampleIt=,Mester,&Padgett1993),Containmentisonlyoneoftheavailableoptions.(SeeMcCarthy&Prince1994bforanexplicitalternativetheoryusingacorrespondencerelationholdingbetweeneachoutputcandidateandtheinput).Asfaraswecansee,little,ifanything,ofthesubstanceofouranalysisandtheorydependsontheprecisemethodofencodinginput/outputdisparities,andwehavehereadoptedthestandardContainmentviewforeaseofexposition. concl10concl11TABLE AP Pd   %  C  $#|}~C--  12  .  ݀Forexample,ithasbeenproposedthatvoicedobstruentscanbedistinguishedfromvoicedsonorantsbyreferringtotheaerodynamicfactthattheproductionoftheformer,butnotthelatter,goesregularlyhandinhandwithanexpansionofthepharynx(seeSteriade1994,andthedevelopmentinPater1994).Butthisdoesnottakeintoaccount,asJohnKingston(personalcommunication)remindsus,thattheproductionofvoicingontheobstruentportionofintervocalicNCclusters"i.e.,preciselytheclustersunderdiscussioninthispaper"involveslittleexpansionofthepharynx,ifany(sincetheoralstopportionisofextremelyshortduration).Ifso,areferenceto expandedpharynxinthiscontextamountstoaphonologicaldiacriticforobstruentvoicing.  '  C  $#|}~C--  ToappearinLinguisticInquiry26.4   %  C  $#|}~C--  13  .  ݀Consider,forexample,thesetofgrammarsthatwouldresultfromthefreepermutationofvisibilityclearancesamongthephonologicalrulespositedinCalabreses(1994)restatementofIt=,Mester,&Padgetts(1993)optimalitytheoreticanalysisofvoicinginJapanese.>1(3 $ {!7OXXd7      C  $#|}~C-- W >TABI SONVOI10SONVOI2Table_A<6X9`( Courier10cpiTable_BSONCONDAPPROX10ATTable_CTable_D )  %  C  $#|}~C--  10  .  ݀Forsyllablestructurereasons,othersonorantsconsonantsbesidesnasalscannotprecedeobstruentsintheoutput.Whensuchclustersariseininputformsthroughmorphemeconcatenation,theyarereparsedasgeminates:/kaw+te/kattebuying,/tor+te/totte - taking,etc.,withoutanytraceofvoicing.LOCAL1Table_ETable_FLOCAL2AMENDTable_GTable_HMUKWRONGTable_ITable_JNABETable_KTable_L 7  %  C  $#|}~C--  11  .  ݀ItisperhapssignificantthatonlyNCclustersshareplacefeaturesalso.Butthenotionthat"likethingsinteract"cannotbeinterpretedtomeansimplythatsegmentsalreadysharingfeaturelinkagearemorelikelytosharemore"thesimilarity(e.g.instrictureorsonority)neednotinvolvelinkage(seePadgett1991andPierrehumbert1993).Thelastpointfurtherindicatesthatthesimilarityideacannotbefullygeometrized,eitherbyclassnodesrepresentations(Clements1985"seePadgett1994fordiscussion)orbymeansofdependencytheoreticstructures(Mester1986,Selkirk1988;seeIt=&Mester1994b).LOCAL4ORDER9ORDER10LOCAL5Table_MTable_NLOCAL6Table_OTable_PLOCAL7Table_QTable_RRANKDIFFTable_SUDEUDETABL1Table_TTable_UUDETABL2Table_VTable_WUDETABL3Table_X4 |  %  C  $#|}~C--  3  .  ݀Itisconceivablethattheserestrictionson>>shouldbeloosenedinsomerespects(e.g.,  fromatotaltoapartialorder),seePS1993fordiscussion.H> a  %  C  $#|}~C--  4  .  ݀AsPS1993:197198note,thiskindofinfinityisnogreatliability;atheoryofgrammarrequiresthatthenotionbestsatisfactorbewelldefined,andnotthatitbeconstrainedbya M!] priorinotionsofcomputability.Furthermore,thesourcesofinfinityarefewinnumber(mainlyepenthesis,theinsertionofnonunderlyingmaterial)andcanbecontrolledbyasuitableheuristic.REDTVOIFGLICCONNASVOILICVOI3LICVOINASCONDKAMITable_ATable_BLICRANKKAMITABLTable_CRENDAKUKAGIKAKITable_DTable_ETOMPOTable_FTable_GTOMBOTable_HBOZOTable_I#2`HelvHP LaserJet IIIostScriptHP3.PRS,\,\,,HzI- 0   %  C  $#|}~C--  5  .  ݀Giventhepresenceofredundantspecificationsinthephoneticoutput,e.g.voicingof[m]inkami,anyaccountofJapanesenecessarilydistinguishesthephonologicalandphoneticlevels. u(] Pursuingthispoint,wemightsupposethatphoneticsdiffersfromphonologyinthatLicenseis ) lowerontheconstrainthierarchyintheformer.Whilesuchstatementscaptureanintuitionabout 2+ thephonology/phoneticsdistinction,itremainstobeseenwhetheritisreasonabletoemploythevocabularyofOTitselfwithreferencetophonetics.ONNASHINDOI6ATEFORMS   %  C  $#|}~C--  2  .  ݀Clustersoftheform{w/r}+CaresyllablestructurallyimpossibleinJapanese,buttheydo  ariseunderlyinglyintheverbalmorphology,asshownin(  MITE8  b,c).MITE9AKATABLE ATABLE A\  `(Times NewRomanTT9 6Times New Roman Regular9 6Times New Roman Regular\  `(Times NewRomanTT9 6Times New Roman Regular9 6Times New Roman Regular }  %  C  $#|}~C--  ڀ  .  ݀ThisworkwassupportedinpartbyfacultyreasearchgrantsfundedbytheUniversityofCaliforniaatSantaCruztoeachofthethreeauthors.EarlierversionsofthispaperwerepresentedatTrilateralPhonologyWeekendI,RutgersOptimalityWorkshopI,OsakaPhonology  Workshop,andinseminarsatUMass/AmherstandUC/SantaCruz.Wewouldliketothankthe '  participantsattheseoccasions,aswellasthetwoLIreaders,forstimulatingquestionsandhelpfulsuggestions.Fordetailedcommentsonanearlierversion,wearegratefultoEricBakovia,StuartDavis,BruceHayes,SharonInkelas,JohnKingston,OrhanOrgun,LisaSelkirk,JoeStemberger,DoncaSteriade,MoiraYip,andCherylZoll.SpecialthanksareduetoJohnMcCarthy,MireNChiosin,AlanPrince,andPaulSmolenskyforsignificanthelpatvariouspointsinourrelentless,butalwaysunfulfilled,pursuitofperfection.   %  C  $#|}~C--  1  .  ݀Themorphosyntacticdistributionofthevoicingmorphemeisanentirelyseparateissue:besidesmorphemeclass(Yamatovs.nonYamato),theargumentstructureofthewholecompoundplaysadecisiveroleinthiscontext,seeUribeEtxebarria1992forrecentdiscussion. Qm 3|G"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^;C]ddCCCdCCCCddddddddddCCȲdxN`xoȐCCCddCdoYoYFdo8Co8odooYNCodddYdddddddddCddddddddodddddϐYYYYYN8N8N8N8oddddooooddoddddxodddYYYYoYYYYddddddooN8N8N8N8p`o888N8ooodd┐YYYoNoNoNoNCCCooooooȐdYYYo8oYoNCddodoYN88d4CCddCC/NdddCdUUddddddFddddFCCssd44ddxxddd~ooCsdFdsd岲dCCȐxȲCddodȐȅdCdYdsȐ`ȐȐȮxȐUvŐdȐddCCCCؐxoxoYNYYYN8YooYdYxxdxddYYxoxxxxNYN88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNYYYYYYYYYYYYYYYYYYYYYY888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxxdYC"u,ae^;C`ddCCCdCCCCddddddddddCCȰdxxxsCYoxxdoxxooCCCddCddYdY8dd88Y8ddddLL8dYYYLYdYddddddCdddddddddxdxdxdxdxdYxYxYxYxYC8C8C8C8dddddddddoYxddddoYdxdxdxdxdYYYYdxYxYxYxYddddddddC8C8C8C8dYYo8o8o8p8p8dxddddxLxLxLdLdLdLdLo8o8o8ddddddoYoLpLoLdo8dxLdLo8oYoYdddxYC88d4CCddCC/NdddCd]]ddddddFddddFCCddd88ddxxdddkddCddFddd尰dCCȐxȰCdxdodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCxxxoxoYxLoYdYC8YooYdYxxdxddoYoYxxoxxxxxxLdC88YoYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxLLLLLLLdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;LhddCCCdCCCCddddddddddCCȰdxLdxxoxxxCCCddCddYdYFdo88d8odddLL8oYdYLdddddddddCddddddddddddddYYYYYL8L8L8L8oddddoooozYddddxYdxddddYYYYdYYYYddddddooL8L8L8L8kddx8x8x8xHx8oooddԔLLLoLpLoLoLx8x8x8ooooooxYxLxLxLdx8oLoLx8xYxYddoYL88d4CCddCC/NdddLdYYddddddCddddCCCkkd88ddzzdddsssCkdCd~d尰dCCȐzȲCddodȐȅdCdYdsȐ`ȐȐȮzȐUvŐdȐddCCCC؅zozoYNzYYYN8YooYdYzzdzddzYzYzozzzzNYN88YzYYzzYzzzzCCCdddddddzzzzzzzzzzzzzzzzzzzzzzNNNNNNNYYYYYYYYYYYYYYYYYYYYYY888888888888YYYYYYYYYYYYYYYYYYYzzzzzzzzzzzzzzzzzzzzzzzdYC"u,ae^/5MPP|555P5555PPPPPPPPPP55P``kt`\tt5GkYkt`t`PYt``YY555PP5PPGPG,PP,,G,tPPPP==,PGkGG=GPGPPPPPP5PPPPPPPPP`P`P`P`P`PkkG`G`G`G`G5,5,5,5,kPtPtPtPtPtPtPtPtPYG`PtPtPtPYGtP`P`P`P`PkGkGkGkGtP`G`G`G`GtPtPtPtPtPtPtPtP5-5,5,5,PGkGY,Y,Y,Z-Z-kPk`kPkPtPtPk`=`=`=P=P=P=P=Y,Y,Y,tPtPtPtPtPtPkYGY=Z=Y=tPY,kP`=P=Y,YGYGtPtPtP`G5,,tP455PP55/NtttPPP5PJJPPPPPP8PPPP855PPP,,PP``yPPtPVyyPP5PP8ttPtPttP卍P55t`tttttttttt5P`PYPtkP5PtGP\t|fttttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5ttttttttttttt```Y`YkG`=YGtPtG5,kGkYYkGkPtGt``P`PPYGYGk``Y|``````=tP5,,tGYGG``G````555PPPPPPP``````````````````````=======PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5V& 8Document[8]Document Style0..8` ..` V8Document[4]Document Style.. . "*,Zh^2ACd2LLd2B26dddddddddd66b{LWymA6AddbjWmcAam75h6mmmlLLAmZXYWdddd2ddddddddddddddm5bbbbbWccccL7L7L7L7mmmmmmmmmYbmmmYnxmbbbWWWWmccccaaaaaammL7L7L7L7lW5h6R6h6mmmmmLLLmLmLmLmL]AAmmmmmmYWWWm6mLmLAYYmmcL77m/dddd/NuddAbeeddIQdBRmmyB2]]d==dd{pr2]Bjdh!ddddxxxxAnAAdhwwwsow_Wsbrm\sWsSssssssssssssssssssssssWWWWWWWssssssssssssssssssssss."u'>E^(45{P(==P(5(+PPPPPPPPPP++N{mqoc=Fvf{a{Wiz{zr4+4PPNUFWO4NW,+S+WWWV==4WHlFGEPPPP(PPPPPPPPPPPPPPW+{N{N{N{N{N{qFoOoOoOoO=,=,=,=,{WWWWWWWWWzG{NWWWzGX`W{N{N{NqFqFqFqFWoOoOoOoONNNNNNWW=,=,=,=,{VF+vSf+fBf+Sf+{W{n{W{WWW{={={=W=W=W=W=iKi4i4WWWWWWlzGrErErEWf+{W{=W=i4zGzGnWWoO=,,W/PPPP/Nnl^PP4NQQPk~P;AP5BWWa5(KKP11}PP}bZ[(K5UshPhPPPP````4X44PxS___f\Y_iLE\zN[lWjJzgz\E\zB\\\\\\\\\\\\\\\\\\\\\\EEEEEEE\\\\\\\\\\\\\\\\\\\\\\."u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"46^ ^GPoxxPPPxPPPPxxxxxxxxxxPPkP]PPPxxPkxkxkPxxCCxCxxxxY]CxxxxkkxkxxxxxxPxxxxxxxxxkkkkkkkkkkPCPCPCPCxxxxxxxxxxkxxxxxxkkkkkkkxkkkkxxxxxxxxPCPCPCPC]xCCCCCxxxxxYYY][]]CCCxxxxxxxkjkxCxY]CxxxxxkPCCx4PPxxPP/NxxxPkbbxxxxxxTxxxxTPP||x>>xxxxxP|x!Tx}xxPPPxxxxPxkxofxkkPPPPk]kxkPCkkxkxxxkk]xPCCkkkkPPPxxxxxxx]]]]]]]xxxxxxxxxxxxxxxxxxxxxxCCCCCCCCCCCCkkkkkkkkkkkkkkkkkkkxkP"AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^CCd||ȘCCC|CCCC||||||||||CChȔ|so8okv|ȐxCCCddCkskskHss00k0ssssFdHsooo`YdY||||||C||||||d|xkkkkkĨksksksksk80808080sssssssssoksssosvskkkkkkksskskskskssssssss80808080`okk0k0k0lHl0sssssȵFFFdddd|H|H|HssssssȬox`x`x`sk0sFd|Hoossssk800s4CCddCC/Nd||Choo||~QQ|LLdd|LCCdd~JJ~~zzdddzzCddLdjd dCCȐvȐCd|o~ȐȔdC~d~sȐkȐȐȤzȐUvŐdȐYYCCCCؔo~kodks]zUsk80dhoo~UssvkvvY~]k`os]s800skkkoCCC~~~~~~~oooooooooooooooooooooo]]]]]]]ssssssssssssssssssssss000000000000ssssssskkkkkkkkkkkksdYC"u,ae^PPxPPPPPPPPP|𲕩Cح𭤐PPPxxPV:::ΊTxV΅skxkPxʩC:C:C:C:C:C:C:C:s:::V:ښTTTxxxxVVVΤsss:TxVC::4PPxxPP/NxP|bb[[xx[PPxxYY𗗓xxxޓPxx[xx xPP𭭭PxxPx𭭭𭭭ŠfŭxkkPPPPزxofC:x|څfkoĩsoC::PPPooooooo::::::::::::xkPV 8Document[6]Document Style8..V 8Document[5]Document Style0.."u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^/5JPP555P5555PPPPPPPPPP55Ptkttk`||>M|kt|`|tYkttttk555PP5PYGYG8PY,5Y,YPYYG>5YPtPPGPPPPPPPPP5PPPPPPPPYtPtPtPtPtPttGkGkGkGkG>,>,>,>,tY|P|P|P|PtYtYtYtYtPtPtY|P|PtPtP`YtPtPtPtGtGtGtGtYkGkGkGkG|P|P|P|P|P|P|Y|Y>,>,>,>,ZM|Yk,k,k,k>k,tYttYtY|P|PvtGtGtGY>Y>Y>Y>k5k5k5tYtYtYtYtYtYttPkGkGkGtYk,tYtGY>k5tPtPtY|PtYkG>,,|P455PP55/NtttPPP5PDDPPPPPP8PPPP855\\P))PP``yPPtPeYY5\P8ttPt\ttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|nttttttttttMttttttk`|tD_tttttttPtttttttPttttPtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`kY`YkGk>kG|GtG>,|GkYYtGkP|Gt``P`PPkGtGk`tY|```t`k>|G>,,|GtGG``G````555PPPPPPP``````````````````````>>>>>>>GGGGGGGGGGGGGGGGGGGGGG,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG```````````````````````PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYCV/8Document[2]Document Style 2A.3  Ԁ   V& 8Document[7]Document Style0..0` ..` "u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "/6=Dz/)@."u,ae^;C]ddCCCdCCCCddddddddddCCȲdxN`xoȐCCCddCdoYoYFdo8Co8odooYNCodddYdddddddddCddddddddodddddϐYYYYYN8N8N8N8oddddooooddoddddxodddYYYYoYYYYddddddooN8N8N8N8p`o888N8ooodd┐YYYoNoNoNoNCCCooooooȐdYYYo8oYoNCddodoYN88d4CCddCC/NdddCdUUddddddFddddFCCssd44ddxxddd~ooCsdFdsd岲dCCȐxȲCddodȐȅdCdYdsȐ`ȐȐȮxȐUvŐdȐddCCCCؐxoxoYNYYYN8YooYdYxxdxddYYxoxxxxNYN88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNYYYYYYYYYYYYYYYYYYYYYY888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxxdYC"$0 W "$  "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"AIP[rﳈE?0404/Nmpϟ@2BP@LL "/6=Dz/)@."u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL zU :Right Par[1]Right-Aligned Paragraph Numbers..2I.3  Ԁ..0.."u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL zh :Right Par[2]Right-Aligned Paragraph Numbers..` ..2A.3  Ԁ..0` ..` "u'>E^2AEd2MMd2A25dddddddddd66aŁxMb|t~V5VddfkWnY?aq;>k;qdnlUUCo[_YWdddd2ddddddd !dddddddn>fffffWYYYYM;M;M;M;qddddooooYfoddYe|nfffWWWWnYYYYaaaaaaqqM;M;M;M;wb>k;];mFI<PIMFLq*!,<=63=0JF<<:;<:<]01100ee;:7o  nn^^hlw0.[eCePTBubbbbKKNN O%l 006;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;fO+r+oss"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^/5JPP555P5555PPPPPPPPPP55Ptkttk`||>M|kt|`|tYkttttk555PP5PYGYG8PY,5Y,YPYYG>5YPtPPGPPPPPPPPP5PPPPPPPPYtPtPtPtPtPttGkGkGkGkG>,>,>,>,tY|P|P|P|PtYtYtYtYtPtPtY|P|PtPtP`YtPtPtPtGtGtGtGtYkGkGkGkG|P|P|P|P|P|P|Y|Y>,>,>,>,ZM|Yk,k,k,k>k,tYttYtY|P|PvtGtGtGY>Y>Y>Y>k5k5k5tYtYtYtYtYtYttPkGkGkGtYk,tYtGY>k5tPtPtY|PtYkG>,,|P455PP55/NtttPPP5PDDPPPPPP8PPPP855\\P))PP``yPPtPeYY5\P8ttPt\ttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|nttttttttttMttttttk`|tD_tttttttPtttttttPttttPtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`kY`YkGk>kG|GtG>,|GkYYtGkP|Gt``P`PPkGtGk`tY|```t`k>|G>,,|GtGG``G````555PPPPPPP``````````````````````>>>>>>>GGGGGGGGGGGGGGGGGGGGGG,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG```````````````````````PG5"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5V? 8Document[3]Document Style.. 21.3  Ԁ   "u,ae^9@Y``@@@`@@@@``````````@@Uyyr@Kyngy|@@@``@U`U`U@``55`5````GK5````UU`U``````@`````````UUUUUUyUyUyUyU@5@5@5@5``````````U`````n`UUUUUUU`yUyUyUyU````````@5@5@5@5jK`y5y5y5w6w6`o````GGGgKhIgKgKy5y5y5```````|U{T|U`y5`GgKy5`````yU@55`4@@``@@/N```@UNN``````C````C@@cc`22``ss```rjj@c`C`d`媪`@@s@``j``@`U`n|YsRrŊ`UU@@@@؊syjsjUyK|U`U@5UjjU`Usn`s``yUUsjssssyK`@55UUUssUssss@@@```````ssssssssssssssssssssssKKKKKKK``````````````````````555555555555UUUUUUUUUUUUUUUUUUUssssssssssssssssssssssn`U@"46^ ^GPoxxPPPxPPPPxxxxxxxxxxPPx]sPPPxxPxkkTxCPCxk]PxxxkxxxxxxxxxPxxxxxxxxxxxxxkkkkk]C]C]C]Cxxxxxxxxxxxxxkkkkkkkkxxxxxx]C]C]C]CsCCC]Cxxkkk]]]]PPPxkkkCk]Pxxxk]CCx4PPxxPP/NxxxPxffxxxxxxTxxxxTPPx>>xxxxxPxTxxxPPPxxxxPxkxsfxxxPPPPk]kkk]Ckkxkxxxkk]k]CCkkkkPPPxxxxxxx]]]]]]]kkkkkkkkkkkkkkkkkkkkkkCCCCCCCCCCCCkkkkkkkkkkkkkkkkkkkxkP"u,ae^9@Y``@@@`@@@@``````````@@`sK\sj@@@``@`jUjUC`j5@j5j`jjUK@j```U`````````@````````j`````NJUUUUUK5K5K5K5j````jjjj``j````sj```UUUUjUUUU``````jjK5K5K5K5k\j555K5jjj``َUUUjKjKjKjK@@@jjjjjj`UUUj5jUjK@``j`jUK55`4@@``@@/N```@`RR``````C````C@@nn`22``ss```yjj@n`C`n`媪`@@s@``j``@`U`n\sRrŊ```@@@@؊sjsjUKUUUK5UjjU`Uss`s``UUsjssssKUK55UUUssUssss@@@```````ssssssssssssssssssssssKKKKKKKUUUUUUUUUUUUUUUUUUUUUU555555555555UUUUUUUUUUUUUUUUUUUsssssssssssssssssssssss`U@"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"AIP[rﳈE?0404/Nmpϟ@2BP@LL "46^ ^GPoxxPPPxPPPPxxxxxxxxxxPPkP]PPPxxPkxkxkPxxCCxCxxxxY]CxxxxkkxkxxxxxxPxxxxxxxxxkkkkkkkkkkPCPCPCPCxxxxxxxxxxkxxxxxxkkkkkkkxkkkkxxxxxxxxPCPCPCPC]xCCCCCxxxxxYYY][]]CCCxxxxxxxkjkxCxY]CxxxxxkPCCx4PPxxPP/NxxxPkbbxxxxxxTxxxxTPP||x>>xxxxxP|x!Tx}xxPPPxxxxPxkxofxkkPPPPk]kxkPCkkxkxxxkk]xPCCkkkkPPPxxxxxxx]]]]]]]xxxxxxxxxxxxxxxxxxxxxxCCCCCCCCCCCCkkkkkkkkkkkkkkkkkkkxkP"AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲdxN`xoȐCCCddCdoYoYFdo8Co8odooYNCodddYdddddddddCddddddddodddddϐYYYYYN8N8N8N8oddddooooddoddddxodddYYYYoYYYYddddddooN8N8N8N8p`o888N8ooodd┐YYYoNoNoNoNCCCooooooȐdYYYo8oYoNCddodoYN88d4CCddCC/NdddCdUUddddddFddddFCCssd44ddxxddd~ooCsdFdsd岲dCCȐxȲCddodȐȅdCdYdsȐ`ȐȐȮxȐUvŐdȐddCCCCؐxoxoYNYYYN8YooYdYxxdxddYYxoxxxxNYN88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNYYYYYYYYYYYYYYYYYYYYYY888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxxdYC"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5z{ :Right Par[3]Right-Aligned Paragraph Numbers..` ..`  ..P 21.3  Ԁ` ..` 0 .. "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^/5JPP555P5555PPPPPPPPPP55Ptkttk`||>M|kt|`|tYkttttk555PP5PYGYG8PY,5Y,YPYYG>5YPtPPGPPPPPPPPP5PPPPPPPPYtPtPtPtPtPttGkGkGkGkG>,>,>,>,tY|P|P|P|PtYtYtYtYtPtPtY|P|PtPtP`YtPtPtPtGtGtGtGtYkGkGkGkG|P|P|P|P|P|P|Y|Y>,>,>,>,ZM|Yk,k,k,k>k,tYttYtY|P|PvtGtGtGY>Y>Y>Y>k5k5k5tYtYtYtYtYtYttPkGkGkGtYk,tYtGY>k5tPtPtY|PtYkG>,,|P455PP55/NtttPPP5PDDPPPPPP8PPPP855\\P))PP``yPPtPeYY5\P8ttPt\ttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|nttttttttttMttttttk`|tD_tttttttPtttttttPttttPtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`kY`YkGk>kG|GtG>,|GkYYtGkP|Gt``P`PPkGtGk`tY|```t`k>|G>,,|GtGG``G````555PPPPPPP``````````````````````>>>>>>>GGGGGGGGGGGGGGGGGGGGGG,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG```````````````````````PG5"46^ ^GPoxxPPPxPPPPxxxxxxxxxxPPx]sPPPxxPxkkTxCPCxk]PxxxkxxxxxxxxxPxxxxxxxxxxxxxkkkkk]C]C]C]Cxxxxxxxxxxxxxkkkkkkkkxxxxxx]C]C]C]CsCCC]Cxxkkk]]]]PPPxkkkCk]Pxxxk]CCx4PPxxPP/NxxxPxffxxxxxxTxxxxTPPx>>xxxxxPxTxxxPPPxxxxPxkxsfxxxPPPPk]kkk]Ckkxkxxxkk]k]CCkkkkPPPxxxxxxx]]]]]]]kkkkkkkkkkkkkkkkkkkkkkCCCCCCCCCCCCkkkkkkkkkkkkkkkkkkkxkPz :Right Par[4]Right-Aligned Paragraph Numbers..` ..`  .. .. 2a.3  Ԁ .. 0.."u,ae^9@Y``@@@`@@@@``````````@@`sK\sj@@@``@`jUjUC`j5@j5j`jjUK@j```U`````````@````````j`````NJUUUUUK5K5K5K5j````jjjj``j````sj```UUUUjUUUU``````jjK5K5K5K5k\j555K5jjj``َUUUjKjKjKjK@@@jjjjjj`UUUj5jUjK@``j`jUK55`4@@``@@/N```@`RR``````C````C@@nn`22``ss```yjj@n`C`n`媪`@@s@``j``@`U`n\sRrŊ```@@@@؊sjsjUKUUUK5UjjU`Uss`s``UUsjssssKUK55UUUssUssss@@@```````ssssssssssssssssssssssKKKKKKKUUUUUUUUUUUUUUUUUUUUUU555555555555UUUUUUUUUUUUUUUUUUUsssssssssssssssssssssss`U@"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"AIP[rﳈE?0404/Nmpϟ@2BP@LL "u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^&+;@@rd+++@r++++@@@@@@@@@@++r9r]QX_QL]_+2ZQt_]J]VEQ_]y]]S+++@@+9@9@9+@@$$@$d@@@@02$@@]@@99@9@@@@@@+@@@@@@@@@]9]9]9]9]9rVX9Q9Q9Q9Q9+$+$+$+$_@]@]@]@]@_@_@_@_@]@]9_@]@]@]@_@J@]9]9]9X9X9X9X9_@Q9Q9Q9Q9]@]@]@]@]@]@_@_@+$+$+$+$^G2Z@Q$Q$Q$P$P$_@_J_@_@]@]@r]V0V0V0E2E1E2E2Q$Q$Q$_@_@_@_@_@_@y]]@S9R8S9_@Q$_@V0E2Q$]@]@_@]@_@Q9+$$]@4++@@++/N]]]@@@+944@@q@@@rr@-@@@@r-++CC@r!!@@MMa@@]@LiiGGr+C@y-]]@]C]]rr@rrr@++]Mr]]]]]]]]]]+@]@G@r]V@+@]9@J]dS]]]]]]]]]];]]]]]]iVMd]7L]]]]]]]@]]]]]]]9]]]]9]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]+]]]+]]]+]]]+]]]]]]]]]]]]]]MQGMGV9Q2S9_@]9+$Z9VGtG_9V@]9]MJ@M@@Q9]9VM]GdMMM]MQ2_@+$$]9]99MM9MMMM+++@@@@@@@MMMMMMMMMMMMMMMMMMMMMM2222222@@@@@@@@@@@@@@@@@@@@@@$$$$$$$$$$$$9999999999999999999MMMMMMMMMMMMMMMMMMMMMMJ@9+"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYCz :Right Par[5]Right-Aligned Paragraph Numbers..` ..`  .. ..h..2(1)3  Ԁ..0h..h"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"AIP[rﳈE?0404/Nmpϟ@2BP@LL "AIP[rﳈE?0404/Nmpϟ@2BP@LL "46^ ^wd7ddd!(!(k(!! (!z!!ooo7o!!!!!!d oooo(!!!!((((!!(!!!(!!!(!!!!!!((oooo$ooopp((((!!d!   ooo((((((z!!(o( o!!(!(oo!4/N!!!_ddddhh0!HHdz!!!!!dddd4d!d!!!!!!!!!!!d! !!7!!!!!!!!!!!!!!!!H 7!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! (!o k( !!! !7!(oo!!oooooooooooo"u,ae^;C]ddCCCdCCCCddddddddddCCȲdxN`xoȐCCCddCdoYoYFdo8Co8odooYNCodddYdddddddddCddddddddodddddϐYYYYYN8N8N8N8oddddooooddoddddxodddYYYYoYYYYddddddooN8N8N8N8p`o888N8ooodd┐YYYoNoNoNoNCCCooooooȐdYYYo8oYoNCddodoYN88d4CCddCC/NdddCdUUddddddFddddFCCssd44ddxxddd~ooCsdFdsd岲dCCȐxȲCddodȐȅdCdYdsȐ`ȐȐȮxȐUvŐdȐddCCCCؐxoxoYNYYYN8YooYdYxxdxddYYxoxxxxNYN88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNYYYYYYYYYYYYYYYYYYYYYY888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxxdYC"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u,ae^/5JPP|555P5555PPPPPPPPPP55Gtenve_tv5>qevt\tkVevttth555PP5GPGPG5PP,,P,|PPPP;>,PPtPPGGPGPPPPPP5PPPPPPPPPtGtGtGtGtGknGeGeGeGeG5,5,5,5,vPtPtPtPtPvPvPvPvPtPtGvPtPtPtPvP\PtGtGtGnGnGnGnGvPeGeGeGeGtPtPtPtPtPtPvPvP5,5,5,5,uY>qPe,e,e,c-c-vPv]vPvPtPtPtk;k;k;V>V=V>V>e,e,e,vPvPvPvPvPvPttPhGfFhGvPe,vPk;V>e,tPtPvPtPvPeG5,,tP455PP55/NtttPPP5GAAPPPPPP8PPPP855SSP))PP``yPPtP_YY5SP8ttPtSttP厎P55t`tttttttttt5PtPYPtkP5PtGP\t|httttttttttJttttttk`|tD_tttttttPtttttttGttttGtttttttttttttttttttttttttttttttttttttttttttt5ttt5ttt5ttt5tttttttttttttt`eY`YkGe>hGvPtG5,qGkYYvGkPtGt`\P`PPeGtGk`tY|```t`e>vP5,,tGtGG``G````555PPPPPPP``````````````````````>>>>>>>PPPPPPPPPPPPPPPPPPPPPP,,,,,,,,,,,,GGGGGGGGGGGGGGGGGGG``````````````````````\PG5"u'>E^2;Ed2OOd2A27dddddddddd66b}zJY~ymxwqhwO7Oddd_RcP9Yj95`3j^b_MK7jT}YSMdddd2ddddddd dddddddb5}d}d}d}d}dRPPPPJ9J9J9J9j^^^^jjjjhS}dd^^hS_wb}d}d}dRRRRcPPPPYYYYYYjjJ9J9J9J9nY5`~3~T~3e5jjj^^ɉMMMmKmKmKmKxWx7x7jjjjjj}hSwMwMwMc~3jMmKx7hShS^jPJ99^/dddd/Nudd;bkkdxdKFdBOmmyB2__dAAddqoo2_Bjvdh!ddddxxxxAnAAdsȟ~wweYwbuug~YwY~~~~~~~~~~~~~~~~~~~~~~YYYYYYYwwwwwwwwwwwwwwwwwwwwww."u,ae^;C]ddCCCdCCCCddddddddddCCȲY~~vCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYddddddCdddddddddYYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdYYYYYYYd~Y~Y~Y~YddddddddC8C8C8C8oNd~8~8~8|8|8dtddddJJJkNlLkNkN~8~8~8dddddddYXYd~8dJkN~8ddddd~YC88d4CCddCC/NdddCYQQddddddFddddFCChhd44ddxxdddvooChdFdhd岲dCCȐxȲCddodȐȅdCdYdsȐ]ȐȐȤxȐUvŐdȐYYCCCCؐx~oxoY~NYdYC8YooYdYxsdxdd~YYxoxxxx~NdC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYC"fxdd/dIIkkkEEkH@ddH///IIId//ddddddk``d+IdddȅHSik"AIP[rﳈE?0404/Nmpϟ@2BP@LL  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~) p`CG Times,_ p CG TimesBold)4&_ p`$CG TimesItalic*:*0 p.CG TimesBold Italic*4\L  p`0Dutch 801RomanSpeedo) p`CG Times<6X9`+Courier<6X9`+Courier) p`CG Times<6X9`+CourierB<6X9`+Courier<6X9`+Courier&/*Ap`Univers<6X9`+Courier< 9+CourierBold<6X9`+Courier<6X9`+Courier<6X9`+Courier<6X9`+Courier<6X9`+Courier<6X9`+Courier<6X9`+Courier<6X9`+Courier< 9+(CourierBoldLegal,_ p CG TimesBold<6X`#"Line DrawFull26X`#(Line DrawFull-LG<6X9`+Courier<6X9`+ CourierLegal<6X9`+Courier<6X9`+CourierB<6X9`+Courier<6X9`+Courier<6X9`+Courier<6X9`+Courier<6X9`+Courier<6X9`+Courier<6X9`+Courier+ * p.Dutch 801BoldSpeedoPb>Ap`6WP IconicSymbolsAType 1<6X9`+Courier<6X9`+Courier<6X9`+Courier3Ha`' Line Printer,_ p CG TimesBold) p`CG Timesu<6X9`+Courier<6X9`+CourierB,_ p CG TimesBoldw,_ p CG TimesBold<6X9`+Courier<6X9`+Courier<6X9`+Courier,<6X9`+Courier1,_ p CG TimesBold(X w  p<Dutch 801Bold ItalicSpeedoGr Ap`>WP TypographicSymbolsType 1<6X9`+"CourierItalicT8RCR,,XRw"qq{q7WCW,,1XW 6RCR,,jXR{Ts-B5B,,Bqg4TCT,,XTnQQu+B5B,,jB{47UCU,,XUs]Ht,D5D,,DsT 8RCR,,7XRqST!s-B5B,,7Bq?xxx,,hXxITs4ddd,,dT78RCR,,XRq?{5TDbPb,,bqf'&/?xxx,,XxR)D CTs-B5B,,BqT5s4ddd,,&dT8RCR,,XRq? { 6?xxx,,UXx? T/^C^,,X^\ T@9qPq,,q\Ts-B5B,,Bq+T\s-B5B,,Bq~ t,F5F,,1F{T=s4ddd,,dT?xxx,,Xx8.T8RCR,,XRq :?xxx,,Xx?t Ts4ddd,,Hd ?xxx,,wXx? jT~s4ddd,,d T8RCR,,XRq`T;s-B5B,,Bqa?xxx,,Xx?VTqs4ddd,,dTv8RCR,,XRqLTls-B5B,,BqTfs-B5B,,Bq$d?xxx,,3Xx? T[s4ddd,,bdBW?xxx,,Xx97WCW,,XW%Y?xxx,, XxBS?ddd,,FXdT8RCR,,XRqP?xxx,,Xx .N?xxx,,XxTE8RCR,,XRqTOs-B5B,,BqTs4ddd,,d?xxx,,Xx w.T8RCR,,XRqTs-B5B,,BqT s4ddd,,MdTH8RCR,,XRqmTs-B5B,,Bqk?xxx,,|Xx c.T8RCR,,XRqYTs4ddd,,dT,s-B5B,,Bqc?xxx,,Xx O.Ts4ddd,, dT8RCR,,XRqETis-B5B,,Bqd?xxx,,8Xx? ;C.Ts4ddd,,gdD5UCU,,XU"1 DC:D (@fPf,,fT %CX՘"'!T L,P?xxx,,,XxR) $#.y%k,,T GI@@@T8RCR,,XRq#Ts-B5B,,Bq#T8RCR,,XRq?p&T[s4ddd,,[d!y?xxx,,Xx?z% (k+HHH,,H1t,F5F,,FTs-B5B,,Bq $T6O@O,,?Oq?BhPh,,h'5S@S,,?Sf'T8RCR,,+ XRqR)*>+H*Ts4ddd,,\ d$TDbPb,,+ bq4?xxx,, XxR)z% K6*-.Txs-B5B,,+ Bq\(Ts-B5B,,+ Bq+7WCW,, XW_4T8RCR,,+ XRq*T8RCR,,+ XRqR),Ts-B5B,,+ Bq4,T8RCR,,+ XRq-Ts-B5B,,+ Bq*-T8RCR,,+ XRq.Ts-B5B,,+ Bq .T8RCR,,+ XRqR)/Ts-B5B,,+ Bq/Ti8RCR,,+ XRq0Ts-B5B,,+ Bq 0T"8RCR,,+ XRqR)>+}1Tbs-B5B,,+ Bq1T'8RCR,,+ XRqs2Ts-B5B,,+ Bq1T8RCR,,+ XRqR)3>+i3Ts-B5B,,+ Bq2Ts-B5B,,+ Bq5t,F5F,, F>+BhPh,, hU5g5S@S,, ?S4TP8RCR,,+ XRq3Ts4ddd,,, d)TG8RCR,,+ XRqR)7$?xxx,,[ Xx b@=.T,s-B5B,,+ Bq6T!8RCR,,+ XRqR)8TOs-B5B,,+ Bq78T=8RCR,,+ XRq9Ts-B5B,,+ Bq-9T78RCR,,+ XRq;:TCs-B5B,,+ Bq#:T"s-B5B,,+ Bq;T.\$5+5,,+ 5q;T8RCR,,+ XRqTMs-B5B,,+ Bq=T8RCR,,+ XRqR)l?Tfs-B5B,,+ Bq>T8RCR,,+ XRqBANBTRs4ddd,, dA7?xxx,, XxXAR) ?.T` p,,+ qa7WCW,, XWTSs-B5B,,+ Bq?Ts-B5B,,+ BqT6OCO,,! XO T 8RCR,,+ XRq @#Cv X3?xxx,, XxrUQ~uic!& {uwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwww {!7nQXXd7      C  $#|}~C-- W 7w"XXdXXd7  zCF:;CF:;%  NC:LicensingandUnderspecificationinOptimalityTheory   &  ڀ       a   JunkoIt=,ArminMester,andJayePadgettUniversityofCalifornia,SantaCruz 1Introduction   t Itisacommonobservationthatredundantphonologicalfeaturesaremostlyinert,neithertriggeringphonologicalrulesnorinterferingwiththeworkingsofcontrastivefeatures.Considerredundantsonorantvoicingvs.distinctiveobstruentvoicing:Inlanguagesenforcingabanonmultipleoccurrencesof[voice]withincertaindomains"Japanese,orProtoIndoEuropean(inthestandardreconstruction,seeHayward1989:45,Garrett1991:793803)"thisconstraintmanifestsitselfinthefactthatavoicedobstruentnevercooccurswithanothervoicedobstruent(withinthesameroot).Suchvoicingconstraintsleavesonorantsunaffected:Althoughvoiced,theyfreelycooccurwitheachotherandwithvoicedobstruents;theirredundantvoicingisapparentlyinvisibletotheconstraint.  Thislinkbetweenphonologicalcontrastivenessandactivityontheonehandandphonologicalredundancyandinactivityontheotherhasplayedakeyroleinmodernphonology.Inavarietyofrecenttheoreticalapproaches(beginningwithKiparsky1982;seeSteriade1994forcomprehensivereferences),phonologicalinactivityhasbeenformallyexpressedasfeatureunderspecification:unlikevoicinginobstruents,voicinginsonorantsisliterallynotspecifiedinthephonology.Thisunderspecificationofredundantfeaturesholdsofunderlyingrepresentationsandpersistsintothephonologicalderivation.  Therecanbelittledoubtthattheoriesofunderspecification,asdevelopedoverthelastdecade(whetherradical,contrastiveorother),havebroughtanewdepthofexplanationtoanumberofareasofphonologicalanalysis(forexample,morphemestructureconstraints,harmonysystems,assimilationanddissimilation).Buildingonthisnotionofunderspecification,LexicalPhonology(Kiparsky1982etsqq.)developedaprincipledtypologyofphonologicalrulesthatwentfarbeyondearliermodelsandfoundanaturalplacewithinacomprehensiveviewofthephonologicalgrammarasamultilevelderivationalsystem.  Whiletheconceptofunderspecificationhasthusplayedapivotalrole,seriousanalystsdidnotfailtonotetheexistenceofcertainproblemsandparadoxesthreateningtheconceptual -': andempiricalbasisofunderspecificationtheory.ThispointhasbeenmadeforcefullybyMcCarthy&Taub(1992)regardingthe(under)specificationofthefeature[coronal]inEnglishconsonants(asthedefaultmemberofthecontrastiveconsonantalplacecategory):Thoughseveralargumentsfortheunderspecificationof[coronal]havebeenadduced,basedonfactsofdistribution(Yip1991,Davis1991)orassimilation(Avery&Rice1989),numerousgeneralizationsholdingofcoronalsduringtheearlyderivationrequirereferenceto[coronal],andhencespecification.Thegeneralizationsinquestionholdoftheentireclassofcoronals,includingbothmarked(e.g.[y,])andunmarkedmembers(e.g.[t,s]).Herearetwoexamples ;  (seeMcCarthy&Taub1992forothers):(i)allinitialcoronalC+[yu]sequencesareprohibitedinAmericanEnglish(*[yyu],*[tyu](Borowsky1986);(ii)anyOCPbasedaccountofthe E  [U]~Halternationfoundintheplural,possessive,andpasttensesuffixesinEnglish(hit[s], R  hide[z],kiss[Uz],wish[Uz],etc.)mustbuildontheagreementinbothplaceandmannerbetween  theconsonantsinquestion,andhencepresupposesthatthefeature[coronal]isspecifiedinallcoronals(Borowsky1986,1987).Thoughonemightentertainsomehopethatsuchfactscouldbeaccountedforbyjudiciouslyorderingdefaultspecificationof[coronal]betweentherelevantstatements,suchattemptedorderingsolutionsareunlikelytohaveaprincipledbasis,nottomentionthedifficultyofmeetingthesimplecriterionofdescriptiveadequacy(orderingparadoxes,etc.).  Thecoronalityproblemcalls[coronal](andmorebroadly,radical)underspecificationintoquestion;infact,itispreciselyfactsofthiskindthathavefiguredprominentlyintheargumentationforcontrastive(orrestricted)underspecificationtheory(i.e.underspecificationofredundantfeatures,butnotofdefaultvaluesofcontrastivefeatures,seeSteriade1987,Clements1987,&Mester&It=1989).Furtherreflectionandinvestigationreveals,however,thatthedemarcationlinebetweenphonologicallyactivefeaturesandphonologicallyinactivefeaturesdoesnotfallpreciselywherecontrastiveunderspecificationtheorydrawsit;underspecificationparadoxesinvolvingonlyredundantfeaturesarenotdifficulttofind(althoughtheyarehardlyevermentionedinthestandardliterature).  Onecaseofthiskind,involvingnasalsand[voice]inJapanese,wasbroughtforthanddiscussedintheappendixofIt=&Mester(1986),wheretheproblemliesinthepresenceofredundantvaluesinsomecontextsbutnotinothers.Inthispaper,wereconsiderthesefactsand ?,': showthatsuchunderspecificationparadoxesarenotminorproblems,butratherresultfromanincorrecttheorythatviewsphonologicalconstraintsasabsoluteandinviolablewellformednessconditions,relativizedintermsofindividualgrammars( parametersettings")andtomodules/levelswithinagrammar("turningoff").Incontrast,OptimalityTheory(Prince&Smolensky1993,seealsoMcCarthy&Prince1993),whichweadoptinthisstudyasaconceptualframework,holdsthatallphonologicalconstraintsare uniformlypresent inall    grammarsandatalllevels,butareinprinciple violable (constraintviolationisalwaysminimal,  0  withinagrammarconsistingofahierarchyofrankedandviolableuniversalconstraints,seesection3belowforasummary).Ourgeneralgoalistoshowhowanoptimalitytheoreticconceptionofphonologyovercomessomeofthelimitationsofthetraditionalwaysoftreatingredundancyandpredictability.Weharnessoptimalitytheoreticinsightstobreathenewlifeintotheuseofunderspecification.Inaccordancewiththeoutputorientedframeworkweadopt,conventionalUnderspecificationTheoryinthesenseoflegislationoffeatureminimizationatunderlyingandintermediaterepresentationsiseschewed.Butdifferentfromattemptstosimplygiveuponphonologicalunderspecification(andinsteadappealtofeatureproliferation,visibilityparametrization,orrepairandrescuedevices,seeCalabrese1993,Mohanan1991,Steriade1994forvariousproposals),thetheorydevelopedherecontinuestograntasignificantroletophonologicalunderspecification"notasaninputpropertybutasanemergentoutputproperty 1" enforcedbyfeaturelicensingconstraints.SincethereisnosequentialphonologicalderivationinOptimalityTheory,thereisalsonosenseinwhich(partsof)thephonologicalderivationcouldbecharacterizedbyunderspecification.Theresultingconceptiondivergessignificantlyfromthedichotomybetweenuniformspecificationanduniformunderspecificationthatcharacterizedthestandardtheory.Inoptimalitytheoreticterms,specificationornonspecificationofafeatureisjustoneaspectoftheselectionoftheoptimaloutputforanygiveninputbythegrammar;thisselectiondependsonthestructureoftheinput,anddifferentresultsarepossible,andindeedexpected,fordifferentlystructuredinputs.Specificationaluniformityacrossalloutputsatacertainlevelisnotanoperativeconsideration.  Asfurthergoalsofthepaper,wehopetoshedsomelightonthepropertiesofpostnasal voicingandofrulesspreadingredundantfeaturesingeneral,developanotionoffeature *3&8 licensing,andprovideinsightsintotheworkingsofOptimalityTheoryitself,focussingonissues  relatedtolexiconoptimizationandconstraintfamilydispersion.  Thispaperisstructuredasfollows.Afterapresentationofthenasalvoicingproblem(section2),wedevelopanoptimalitytheoreticconceptionoffeaturelicensing,illustratingandmotivatingitbymeansofananalysisofJapaneseNCvoicing(section3).Weturnnexttosomeissuesregardinginputfeaturespecificationandlexiconoptimization(section4)andtakeupquestionsrelatedtomultiplelinking(section5),developingouranalysisalongtheway.Weconcludethepaperwithasummaryofthemainresultsandadiscussionofanumberofopenissues(section6). & t 2TheProblem:VoicingandNasalsinJapanese  U  JapaneseexhibitsamorphophonemicalternationknownasRendaku(sequentialvoicing):undercompounding,asin( 11  ),aninitialobstruentinthesecondmemberbecomesvoiced(Martin'tU>%1952:4849,McCawley1968:8687).ThealternationisconfinedtotheYamato(native)stratumofthelexicon,and,thoughproductive,haslexicalexceptions.IntheterminologyofclassicalLexicalPhonology(e.g.Kiparsky1985),Rendakuhasallthecharacteristicsofanearlylexicalprocess.3 ,XXm wpXX3& 9 (2  1  3  )    a.ori k ami 8 [ ori+ g ami    paperfolding 0"   b.oo s umoo 8 [ oo+ z umoo  grandsumotournament ?$   c.yama t era 8 [ yama+ d era  mountaintemple &   d.mizu t eppoo 8 [ mizu+ d eppoo  waterpistol N(   e.mizu h ana 8 [ mizu+ b ana  runningnose ]!* H ,XXm <X ,XXm wpXH   ` (h

  4  Westatethisconstraintinformallyin( 6A6  ):(2  6  3  )     Anasalmustsharethefeature[voice]withafollowingconsonant   Speakingprocedurally,nasalsspread[voice]toafollowingobstruent.Thisholdsnotonlyasamorphemestructureconstraint,asillustratedin( SHINDOI5  ),butisinstantiatedinalternationsaswell,asseeninexamplesinvolvinggerundivetein( TEFORMS7  ).  &  * X,XX hX*(2  7  3  )    a. ` /yom+te/  8 h yo nd e  reading g   b. ` /in+te/  8 h i nd e  dying'B v Thereisnocomparablevoicingafterothersonorants,neitheraftervowels(  MITE8  a)norafterothernonnasalsonorantconsonantsliker(  MITE8  b,c).  &  2       2 & z (2  8  3  )    a. ` /mi+te/  8 h mi t e  seeing     b. ` /tor+te/  8 h to tt e  taking A"   c. ` /kaw+te/  8 h ka tt e  buying P$ 'zE  o[X` hp x (#%'0*,.8135@8:<H?AXoThebasicanalysisofNCvoicinginIt=&Mester(1986:6971)takesastandardviewofsuchfacts(cf.Kiparsky1982);inbrief:& t 0  !Theredundant[voice]ofnasalsisunderspecified( 99  a). (#(# 0  !Atsomepointduringthephonologicalderivation,nasalsreceive[voice]( 99  b). (#(# 0  ![voice]spreadsrightward( 99  c)bymeansofaruleindependentlynecessarywithintheanalysis'ti!H. (#(# & u (2  9  3  )  ԀNCVoicing:a.  Underlying: b.  NVoicing  c.  NCVoicing :(#4   NC `  8  NC h   8  NC )$6    `     h      *%7    `    V h     V'u&UJ i+&8 Ї1Ko[X` hp x (#%'0*,.8135@8:<H?AXoHowever,It=&Mester(1986:6971)alsoshowthattheseassumptionsleadtoanunderspecificationparadox,apointlatertakenupinBorowsky(1986:34)andTaub(1988).Nasals(togetherwithothersonorants)mustnotbespecifiedfor[voice]whenRendakuand j LymansLawapply,asseenin( 11  )above.ThisimpliesthatRendakuandLymansLawmusttakeplacebeforetheassignmentofredundant[voice]tonasals( 99  b).Bytransitivity,NCvoicing( 99  c),whichmustfollow( 99  b),shouldalsonotbevisibletoLymansLaw.However,thefactsshowtheopposite:NCvoicingisvisibletoLymansLawandblocksRendaku,asillustrated  1  bycompoundslikeirooto k a g aelaymansidea( AKA10  )(*irooto g a g ae). A  &  (2   10  3  )    Theunderspecificationparadoxforserialderivations:      ` hayaganefirebellvs.irootokagaelaymansidea Q  *hA dh,dd#,dd ,dd +  " a  "a. Underlying: ," ,/haya+kane/ % %/irooto+kakae/ %"r %Rendaku: %  %haya+ g ane % %irooto+ g akae %  %N&NCVoicing: "m "haya+ga n e  m irooto+ga g ae %m %Output: %  % 3haya+gane % %*irooto+gagae&    &*hA dddd#dd dd Ah,dd#,dd ,dd +  " a! "b. Underlying: ,"u" ,/haya+kane/ %u" %/irooto+kakae/ %u" %N&NCVoicing: %B$  %haya+ka n e %B$ %irooto+ka g ae %B$  %Rendaku: "p& "(blockedbyOCP)  ( (blockedbyOCP) %p& %Output: %J!*  %*haya+kane %J!* % 3irooto+kagae &J!*    &' Q  Thederivationsin( AKA10  )bringoutthecontradictionbycontrastingaformwithaplainmedialnasal,whichrequiresRendakutoprecedeNVoicing( AKA10  a),withaformcontainingamedialNCcluster,whichrequirestheoppositeorder( AKA10  b).Wehave,then,aparadox:NasalsinNCclustersmustbespecifiedfor[voice]justwhenothernasalsmustnotbe.Ablanketapproachinvokingaredundancyrule[+nasal][+voice],whetherbystipulationtoadomainorbytheRedundancyRuleOrderingConstraint(Archangeli1984,Archangeli&Pulleyblank1986),failsbecauseitpaintswithtoowideabrush,requiringthatallnasalsreceive[voice],if +>': anydo.Instead,nasalsappeartoreceive[voice]onlywherenecessarytotriggerNCvoicing.Wecouldofcourseencodethisfactintheredundancyruleitself,bymakingthepresenceofafollowingobstruentaconditionontheinsertionof[voice].Suchasolutionislittlemorethanastatementoftheproblem,sinceitremainsacoincidencethatinsertiontakesplacejustinthecontextwhereanotherrulewilllaterapplytospread[voice]ontothefollowingconsonant.(Whynotinsert[voice]onfinalnasals?Oronnasalsbeforehighvowels?etc.)Asaminimalconditionofadequacy,werequirethattheanalysisattempttoestablishsomerelationbetweentheNCvoicingphenomenonandthepresenceofredundant[voice]inN.  Inresponsetothisproblem,It=&Mester1986proposealevelorderedsolution,inwhichnasalvoicingisinsertedatalevelpriortoothersonorantvoicing,notingatthesametimethatthisproposal,althoughdescriptivelyadequate,hasseriousdrawbacksandinanycasecannotserveasageneralanswertounderspecificationparadoxes.AninterestingnewideaappearsinBorowsky(1986:34),relyingonanunorthodoxassumptionaboutthefeaturestructureoflinkedNCclusters.Althoughtheaccountsucceedsinderivingthepostnasalvoicingfactswithoutadditionalmachinery,Taubs(1988:70!73)comprehensivediscussionshowsthattheparadoxitselfstillremains.However,oneofthecentralingredientsofBorowskysaccount"theideatorelatepostnasalvoicingtosegmentmarkednessconditions(nasalsmustbevoiced,andobstruentscanaccommodatevoicing)"ispreserved,albeitinadifferentform,inouroptimalitytheoreticaccountbelow.Otherproposalhavetriedtocircumventtheparadoxbyspittingthedistinctivefeature[voice]intoseveralsubfeatureswhichcanbeseparatelyappealedto(seeRice1993for[spontaneousvoicing],Steriade1994for[expandedpharynx]).Butpreciselybecausesuchanalysesdifferentiateinafundamentalwaybetweensonorantandobstruentvoicing,voicinginteractionsbetweenthetwosegmentclasses,asinNCvoicing,come ", asanunpleasantsurprise,tobeaccommodatedbyspecialpleading(seesection6forfurtherdiscussion).Wewillarguethatthetroubleliesnotwith[voice],butwiththetraditionalviewoftheorganizationofphonology;andthechallengeistoresolvetheparadoxwithoutdestroyingtheunityandintegrityofthedistinctivefeature[voice].  Takingafreshlookattheproblem,thesolutiontobedevelopedbuildsonanunderstandingofredundantfeature(under)specificationintermsoflicensing,advancedinthe */&8 settingofOptimalityTheory.Thissolution,whileconceptuallyverydifferentandwithsomewhat ?,': differentempiricalconsequences,sharesonecrucialpointwiththeearlieranalysisofIt=&Mester1986:Itreliesonthefactthatwithinspecificgrammars,generalconstraintscanbedecomposedintoafamilyofseparatemorespecificconditions:Nasalvoicingcanbeseparated j fromothersonorantvoicing(cf.thesimilardecompositionof[ATR]/heightconstraintsforvowelsinArchangeli&Pulleyblankinpress),employinganotionofwhatwewillrefertoas ' w constraintfamilydispersion(firstproposedinPrince&Smolensky1993,henceforthPS1993).     &  3FeatureLicensingandRedundancy  A   3.1OptimalitytheoreticBackground    ThebasictenetsofOptimalityTheorywhichwillbeimportanttoourtheoryoffeaturelicensingareconstraintbasedoutputselection,constraintuniversalityandranking,andtheprincipleof'Aiminimalviolation(seePS1993foracomprehensiveandmoreformalpresentation,aswellasMcCarthy&Prince1993,1994andreferencesthereinforfurtherdetails).[i]  Constraintbasedoutputselection:Theoutputofphonologyormorphologyisdetermined o bywellformednessconstraintsthatselectamongsomecandidatesetofforms(basedonsomeunderlyingrepresentationasinput),consideredinparallel.Asaresult,therearenorulesorrepairstrategies,andnoserialderivation.Candidatesetsareinprincipleinfiniteandmaximallyinclusive,leavingthejobofwinnowingoutformstothewellformednessconstraints.[ii]  Universalityandranking:Thesetofconstraints bԀisprovidedbyUniversalGrammar C$ (={C1,C2,...,Cn});anindividualgrammar G isobtainedbyimposingastrictdominance & order>L>ontheelementsofo[X` " hp x (#%'0*,.8135@8:<H?AXoӀ( G =(;>>)).{[X` hp x (#%'0*,.8135@8:<H?AX" hX{Inthestandardconception,therelation>b">  R( isastrictlinearorder,i.e.itistotal,transitive,andasymmetric:foranytwoconstraintsCiand a!* Cj,eitherCi> >CjorCj>>Ci,butnotboth;andifCi>>CjandCj>>Ck,thenCi>#> ", Ck.  &  3      ׀Thewellformednessconstraintsinformthesubstanceofsomepartoftheoretical !$q. phonology,e.g.prosodicphonology,featuregeometry,thetheoryofautosegmentaloperations,etc.[iii]  Minimalviolation:Allconstraintsareinprincipleviolable.Rankingandviolabilityare ;(#4 thekeycharacteristicsofOptimalityTheorythatsetitapartfromotherconceptionsofgrammar(liketheminimalistprogramoutlinedforsyntaxinChomsky1993).Inparticular,theseconcepts *H&8 arenotfoundinothernonderivationalconstraintbasedtheoriesofphonology,suchasHarmonicPhonology(Goldsmith1990,1991,1992)andDeclarativePhonology(Scobbie1991)(thoughcf.Paradis1988a,bforadifferentkindofconstraintviolabilityandconstraintrankinginderivationalterms,seePS1993:214!219fordiscussion).Andarguablythesetwopropertiesofrankingandviolabilityarethecrucialelementsresponsibleforthetheorysexplanatorysuccesses,notonlyinphonology,butalsoinotherareasofgrammar,includingprosodicmorphology(McCarthy&Prince1993),allomorphy(Mester1994)andsyntax(Grimshaw,toappear).  Theoutputofanoptimalitytheoreticgrammar,asnoted,isnottheresultofaderivation,asinstandardphonologicaltheorizing.Rather,phonologicaladjudicatingproceedsinparallel:everypossiblecandidateisconsideredforagiveninputform.  &  4      ׀Theactualoutputform"the R  optimalform"isthe(often,butnotnecessarily,unique)memberofthecandidatesetthatbestsatisfiestheconstrainthierarchy.Wewillseeinthenextsection,withconcreteexamples,howbestsatisfactioncanbedeterminedinastraightforwardwayfortableaurepresentationsofcandidatesets(seePS1993:6876foraformaldefinitionintermsofHarmonicOrdering).    ThissketchofOptimalityTheory(OT)isincompleteinvariousrespects,butsufficienttosetforthourbasichypothesesaboutredundancyandunderspecification,whichisintrinsicallylinkedtoaconceptionofconstraintsasrankedandviolable.Ourgoalistoexplainthespecificationalbehaviorofredundantfeatures(like[voice]innasals)astheresultoftheantagonisticinteractionoftwofamiliesofconstraints,onefavoringrichnessofphoneticspecifications,theotheronefavoringtheopposite.Bothgroupsofconstraintsarepartofthedominancehierarchythatconstitutesthegrammarofalanguage.  Ontheonehand,therearethefamiliarphoneticrealizationalconstraints(involvinggrounding,intheterminologyofArchangeli&Pulleyblankinpress),whererepresentationsarerequiredtoberichlyspecifiedforphoneticallyrequiredordesirableproperties.Theseincludetheredundantpropertiesforeachsegmentclass;theymaybephysicallyinherent,orservetoenhancecontrasts,orinotherwaysbefavored(seeStevens,Keyser&Kawasaki1986).Thus,sonorantsarevoiced,back(nonlow)vowelsareround,highvowelsare[+ATR],etc.  Ontheotherhand,thereistheworldoffeatureminimizationanddistinctiveness:languagesareinventorywiseparsimonious,buildingonafewoperativephoneticdistinctions. 9,': Themanyotherdistinctionsthatformpartofthespectrumofuniversalphoneticcapabilitiesofhumansplaynoroleintheorganizationoftheinventory.  Thesetwofactors"onefavoringspecificationalabundance,theotheronefavoringspecificationalparsimony"areantagonistic.Standardunderspecificationtheories,withtheirapparatusoffeatureminimization,markingconventions,defaultandredundancyrules,orderings,etc.,canbeviewedasparticularwaysofadjudicatingbetweenthem.Facedwithantagonisticprinciples,atheorysubscribingtotheviewthatgrammaticalconstraintsarealwaysinviolableattheirpointofapplicationonlyhastheoptionofassigningthetwoantagonisticprinciplesdifferentpartsofthederivation(earlyvs.late)astheirrespectivedomains,suchthateachofthemcanholdtruewithinitsparticulardomain.Itisthisattempttofindthesolutioninderivationaldifferentiationthatleadsstraightintounderspecificationparadoxes.  OptimalityTheory,ontheotherhand,canadjudicatebetweenthetwokindsofconstraintsbyrankingthemdirectlywithrespecttoeachother,withouthavingtocreateaspecialderivationalstageorlevelduringwhichredundantfeaturesareabsolutelyprohibited.Acrucialaspectofoursolutiontothefeaturespecificationproblemisthatitbuildsonasmallgroupofsimpleconstraints,allindependentlymotivatedandplausiblecandidatesforUniversalGrammar. & t 3.2UnderspecificationasanOutputProperty  :$ Considertheconnection,oftenmerelyimplicitinpracticeandyetclearinprinciple,betweenfeatureredundancyandfeatureunderspecification.BuildingonaproposalmadeinPadgett't:(1991:568),welinkthetwobymeansofthenotionoflicensing.Thehypothesisformulatedin Q!* ( RED11  ),asaprincipleofuniversalgrammar,isanexplicitstatementofthisconnection(cf.theredundancyrulesandmarkingconditionsofKiparsky1985,Archangeli&Pulleyblank1986,andothers).&  (2   11  3  )    LicensingCancellation: h If FDG ,then (FG) . &"2 0  0` (#(#0 ` (#` (#0 (# (#0h(#(#"IfthespecificationFimpliesthespecificationG,thenitisnotthecasethatFlicensesG."'& h(#h(# Thestatementin( RED11  )isintendedasfollows:Supposeagrammarcontainsasegmentstructureconditionexpressingthesegmentinternalredundancythatasegmentthathasthefeature F also >,': hasthefeature G .Thenthefeature G isnotlicensedwithinasegmentcontaining F .For  example,giventheredundancyimplication[son]D[voice],asonorantsegmentdoesnotlicense[voice].The[voice]featureislicensedwhenlinkedtoobstruents(  TVOI12  a),butnotwhenlinkedtosonorants(  TVOI12  b).&  (2   12  3    )a. ` licensed[voice]: h b.  unlicensed[voice]: & v    ` [d]    h   [n]       `     h    l      ` [voice]   h   [voice]'& G U     OuruseofthenotionlicenseandoftheLicensingCancellation( RED11  )differsinacrucialwayfrommorefamiliarformulationsintermsofcooccurrencestatementslike*[F,G],e.g.*[son,voice].Unlikethelatterconstraint,LicensingCancellationdoesnotimplyanyincompatibilitybetweenthetwofeatures.Thedifferencecanbebestappreciatedinascenario > asin( FG13  ),wherebothsegmentsS1andS2arespecifiedfor[G],butonlyS1,andnotS2,is  specifiedfor[F].&  (2   13  3  )    S1󀀀S2 X Ѐ   A   [F][G]'XՐGiven( RED11  ),S1doesnotlicense[G].[G]isneverthelesslicensed"byS2,aslongasS2contains L nofeature,[F]orother,thatimpliesthepresenceof[G];thereforetherepresentationiswellformed.Ontheotherhand,[G]wouldviolateaconstraint*[F,G],regardlessofanydoublelinking.Thisuseoflicensing(cf.alsoSteriade1994onindirectlicensing)extendstothefeaturaldomainlicensingaccountsfamiliarfromotherareas,e.g.codalinking(Goldsmith1990,It=&Mester1993,Lombardi1991,buildingonIt=1986,1989;cf.alsoBorowsky1986,Hayes1986,Schein&Steriade1986).  Licensingcancellationsdefinespecialsituationsthatcontrastwiththedefaultscenariowhereallfeaturesproperlyincorporatedintofeaturegeometricalrepresentationsarelicensed.Thus,[voice]linkedtoanobstruentroot,wherenoredundancyconditionleadstoacancellationoflicensing,isipsofactolicensed(  TVOI12  a).Therelationshipbetweengeneralfeaturelicensingand particularlicensingcancellationsisanexampleofthegeneralcase/specialcasescenariofamiliar )K%4 fromtheElsewhereCondition(Kiparsky1973)aswellasmorerecentworkinOT(thePaninianrelationofPS1993).  Thus,aredundancyimplication[son]D[voice],givenLicensingCancellation,meansthatthefeature[voice]isnotlicensedinasonorantsegment(  TVOI12  b).Theabsenceof[voice]isforcedbytherelevantmemberofthefamilyoffeaturelicensingconstraintsschematicallycharacterizedin( LICCON14  ).(2   14  3  )  0  License(,):Thephonologicalfeature,mustbelicensed.; (#(# Inthecontextofourdiscussion,therelevantredundancyimplicationinJapanese(sharedbyallnaturallanguages)istheonerepeatedin( NASVOI15  )(wefollowtheconventioninOTofabbreviatingconstraintnamesforuseinconstrainttableaus).Thisconditioninfactrepresentsawholefamilyofsonorantvoicingconditions(governingnasals,liquids,vocoids,andothersonorants).(2  15  3  )  Ԁ  SonVoi [son]D[voice]      ToillustratetheroleoffeaturelicensinginouraccountofJapanese,consideranexamplelikekamipaper.Therelevantmembersoftheconstraintfamiliesinvolvedarethevoice -"  -" licensingconstraintLicense(Voice)( LICVOI16  ),where,=[voice],andthenasalversionofthe =$ sonorantvoicingconditionNasVoi( NASCOND17  ). & (2  16  3  )    License(Voice)  I.e.,thefeature[voice]mustbelicensed. T!* (2  17  3    )  NasVoi   I.e.,[nasal]D[voice] ", SinceLicense(Voice)istheonlymemberofthefamilyofLicensingconstraintstoplayarole k% 0 inouranalysis,wewillrefertothisconstraintsimplyasLicense.Focussingonthenasal[m] &"2 inkami,wehavetwopotentialcandidaterepresentations:onewiththenasalspecifiedforvoicing %(u#4  (  KAMI18  a),andonewiththenasalunspecified(  KAMI18  b). )$6 &   (2  18  3  )  *h dddd#dd dd Ah,dd ,td ,Xtd ,tt +  ,"  S,a. %AAA !q %kamìT    Z Ѐ V  C  C  0A!,AAA 08 4AAAA!!qA 4*License !q 3NasVoi󀀀1'%C  S AAAA 1*y dddd td Xtd tt y,dd ,td ,Xtd ,tt +  ,"   S,b. %AAA F   %kami 0A!F  AAA 08 4AAAA!F  A 43License F   *NasVoi󀀀1'% h   S AAAA 1' +(  KAMI18  a)fulfillsNasVoi,butviolatesLicense.(  KAMI18  b),ontheotherhand,violatesNasVoi,but n  fulfillsLicense.Thesituationthuslookslikeastandoff,witheachrepresentationincurringone   constraintviolation,whilefulfillingtheotherconstraint.ThisisthekindofconstraintconflictthatOptimalityTheoryresolves,withitsnotionofconstraintdomination(seetheearlierdiscussioninsection3.1).Leavingdetailstothemoretechnicalexpositionsoftheformaltheory(seeespeciallyPS1993),thecentralideaisthatconstraintsarerankedinastrictdominationorder,withhigherconstraintstakingabsoluteprecedenceoverlowerconstraints.Forthecaseathand,wehypothesizethattheconditionLicenseisrankedabovethevoicingcondition I NasVoi.InthenotationofOT,thisiswrittenasin( LICRANK19  ). V (2  19  3    )0  License󀀀>>NasVoi`#(#(# Giventhisconstraintranking,(  KAMI18  b)isthewinningcandidate,withthenasalunspecifiedforvoicing.TheOTstyletableau,whichillustratesthecrucialrankinggraphicallyintermsof columnsorderedfromlefttoright,isgivenin( KAMITABL20  ). "+ & f  (2  20  3  )  =X XXXw"  *y dddd td Xtd tt y,dd ,dd ,d",td"+  ,"  @x, %a" %Candidate *Ea *License 0A!aE 0NasVoi FE9!a x @ A Fa.Xq XX=XԀ+=XXXXqԀ 5E(E 5kamì 7EE( aE 7 6EA>!EE 6* HA;#" EA> Hb. 5A(a A 5kamì ̀V 7AE( A A 7Ѐπ*! -AA>) y AE -1'%a    AA> 1nXq XX=X'f Intableaux,a*inrowi,columnjindicatesthatcandidateiviolatesconstraintj.Candidate f  ( KAMITABL20  b),forexample,violatesLicense.Evaluationofthecandidatesproceedsrecursivelyby v  constraint,untilallcandidatesbesidesthewinningcandidateareeliminated.Inthissimplecase,thedecisionisalreadyreachedinthefirstround.*!denotesthecrucialfailureofagivencandidate.Hereisthesenseinwhichtherearenotradeoffs:eventhemostsplendidfulfillmentofsubordinateconstraintscannotrescueacandidatethathasslippedwithrespecttoahigherrankingconstraintintheeyesofthetheory.Inthisway,thewinningcandidateemergesasoptimal,astatusdenotedbythepointinghand(+).Theoptimalcandidateistheonlywellformed  candidate,andisselectedastheoutput.Allothercandidatesareillformed.Thoughviolationsaftercriticaldecisionpointsarerecordedforcompleteness,theseresultshavenobearingontheoutcome,afactindicatedbyshadingofthecorrespondingcells.  Theconstraintconfigurationin( KAMITABL20  )isthecoreofourproposalinthispaper,whichwillbefurtherfleshedoutincrucialrespectsasweproceed.Ingeneral(withoneimportantexception,tobediscussedbelow),itsettlesexactlyonthedegreeoffeature(under)specificationamplysupportedinpastworkonJapanesePhonologyandelsewhere.Moreconcretely,( KAMITABL20  a)istheempiricallyjustifiedunderspecifiedrepresentationforRendakuandLymansLaw,aswehaveseen:[m]doesnotcarry[voice],henceRendakuvoicingisnotblockedandappliestotheinitial[k]in( RENDAKU21  a).&  (2  21  3  )    a.orikami 8  ori+ g ami,*ori+ k ami p foldingpaper &/"2 Ѐ  b.nisekagi 8  *nise+ g agi,nise+ k agi p fakekey >(#4    c.aokaki 8  ao+ g aki,*ao+ k aki   p greenpersimmon'& )$6   Incontrast,aformlikekagikey( RENDAKU21  b),withdistinctivevoicingon/g/,doesnothave   theunspecifiedrepresentation.Thefactthat[voice]isdistinctiveforobstruentsmeansthattherecannotbearedundancyconditionforobstruentvoicing(sincetheunpredictablecannotbepredicted).Consequently,LicensingCancellationdoesnottakeeffect,andthevoicingonthemedial/g/fulfillsLicense( KAGIKAKI22  a).NasVoiisvacuouslysatisfiedsincetheformhasnonasal ' w segment.Thevoicingfeaturelicensedbythemedialobstruent[g]manifestsitselfbyblockingRendakuvoicingin( RENDAKU21  b).Themedialobstruentinkakipersimmon( RENDAKU21  c)likewisesatisfies  1  LicenseandNasVoi( KAGIKAKI22  b). A  &  (2  22  3  )  *y dddd dd d"td"y,dd ,td ,Xtd ,tt +  ,"   S,a. %AAA _  %kagìT   H  Ѐ V  1  1  0A!AAA 08 4AAAA!_ A 43License _  3NasVoi󀀀1'%1  S AAAA 1*y dddd td Xtd tt y,dd ,td ,Xtd ,tt +  ,"  j,b. %AAA 4 %kaki 0A!4AAA 08 4AAAA!4A 43License 4 3NasVoi󀀀1'%V  j AAAA 1' 0  Turningnowtothepostnasalvoicingfactspresentedinsection2,wehavetoaccountforthefactthatthereare(YamatoJapanese)rootsliketombodragonfly,butnorootslike  ! *tompo,andforalternationslike/yom+te/yonde,not*yonte.FocussingontheNCcluster i# (andabstractingawayfromplaceassimilation),considerfirstthefollowingtwocandidatesthatareparallelto(  KAMI18  ):&  (2  23  3  )     3!) *y dddd td Xtd tt y,dd ,td ,Xtd ,tt +  ," "l* S,a. %AAA "+ %tompò ̀V 0A!v$-AAA 08 4AAAA!"+A 4*License "+ 3NasVoi󀀀1'%v$-  S AAAA 1*y dddd td Xtd tt y,dd ,td ,Xtd ,tt +  ," X&!0 S,b. %AAA &0"1 %tompo 0A!&0"1AAA 08 4AAAA!&0"1A 43License &0"1 *NasVoi󀀀1'%($3  S AAAA 1 ' 3! )$5 Everythingelsebeingequal,thesamereasoningappliedfor(  KAMI18  )shouldleadtotheconclusion that(  TOMPO23  b)issuperior.Noteverythingisequal,though,sinceinthiscasethereisanotherrelevantcandidate(  TOMBO24  ),withdoublylinkedvoicing.&  (2  24  3  )  *y dddd td Xtd tt y,dd ,td ,Xtd ,tt +  ,"   S, %AAA 8  %tombò< Ԁ| ! q  ЀV 0A! Z AAA 08 4AAAA!8 A 43License 8  3NasVoi󀀀1'% Z   S AAAA 1'Thefeature[voice],thoughnotlicensedbythenasalrootnode,isherelicensedbytheobstruentrootnode,andhencefulfillsLicense.Thislastcandidaterepresentation,whichfulfillsboth   VoiceLicensingandtheredundancyconditionthatnasalsshouldbevoiced,isthereforesuperiortobothoftherepresentationsin(  TOMPO23  ),eachofwhichrespectsonlyoneoftheconditions.Thisismoreperspicuouslybroughtoutintableau(  BOZO25  ).& ^ (2  25  3  )  =XXXXq*y dddd td Xtd tt y,dd ,dd ,d",td"+  ," j @x, %" %Candidate *E *License 0A!E 0NasVoi FE9! x @ A Fa. 5E(RE 5tompo 7EE(E 7 0EA!REE 0*! AA4R EA Ab. 5A(2 A 5tompò ̀V 7AE(r"A 7*! -AA>!AE - FA9!2   AA> )Fc.XqXX=X+=XX+XXqԀ 5A($A 5tombòX\XX=X$ =XX! XX\Ԁ % ЀV 7AE(&A 7 0AA!$AE 0Ѐπ1'%% )  AA 1'^ 1XqXX=X  Recallinwhatrespectfeaturelicensingdiffersfromstatementslike*[nasal,voice]:in !* ourterms,thenasalsegmentisinnosenseincompatiblewithlinked[voice].Rather,itmerelyfailstograntlicensing;suchlicensinghoweverisconferredbytheobstruentrootnode.FeaturelicensingexplainswhyonlydoublylinkedNCnasalsare(redundantly)voiced.InitscrucialrejectionofthenotionthatavoicingconditionlikeNasVoiisnoforceatallatsomeearly 'S"2 levelofthephonology,thispictureofunderspecificationisspecifictoanaccountinthespirit ofOT. ) %6   Intheabsenceofevidencetothecontrary,weassumethatLicenseisundominated.   GiventhepossibilityenvisionedwithinOTofderivingnewgrammarsbyrerankingconstraints,wemightexpecttofindsomelanguagewiththerankingNasVoi>>License,whichentails j redundant[voice]specification.(Wemightsimilarlyfindsucharankingbetweenanyredundancyconditionandtherelatedlicensingcondition,inJapaneseorelsewhere).However,theissueiscomplicatedbyanevidentneedforatheoryofrankingmarkednessitself,giventheexistence    ofapparentlyrobustlyundominatedconstraintsacrosslanguages(seee.g.PS1993andMcCarthy&Prince1993).   &  5      ׀ >    Theresultoftheconsiderationsinthissectionisthatfeaturespreadingconfigurationsareadvantageous:Clustersdoublylinkedfor[voice],suchasYamatoJapaneseNCclusters,areoptimalbecausetheysimultaneouslyfulfillbothconstraints"featurelicensingandthenasalvoicingcondition.Everythingelsebeingequal,norepresentationfulfillingonlyoneoftheseconstraintswhileviolatingtheotheronecouldbesuperior.  But"iseverythingelsetrulyequal?Whatisthecostexactedbyfeaturespreadingitself,comparedtorepresentationswithouttheextraassociations?Inmostrulebasedtheoriesofphonology,featurespreadingisascribedtotheoperationofaphonologicalrule,spreadingisnotautomatic(seePulleyblank1986foranextendedargumenttothiseffect).Itseemsreasonabletohypothesize,then,thatmultiplylinkedconfigurationsmustexactsomeprice.Howshouldtheextraassociationlinesbecounted,andhowshouldthesemarksbeweighedagainstthemarksincurredforotherconstraints?  Suchquestionsareimportantnotjustfortheparticularanalysisbeingpursuedhere,butforaproperunderstandingofOptimalityTheoryitselfasageneraltheoryofphonologyextendingbeyondprosodicphonology(syllableandfootparsing,etc.)intotheareaof(auto)segmentalphenomena,wherespreadinganddelinkinghavefiguredveryprominentlyinpastanalyses.Seekingananswertothesequestions,thenextsectiondevelopsthestandardtheoryoffaithfulnessconstraintsandexplicatesthenatureofinputsinatheorythatadoptsaprincipleoflexiconoptimization(PS1993).  )$6  & t  4InputOutputDisparities:FaithfulnessandLexiconOptimization     InOptimalityTheory,muchoftheroleofthetraditionalphonologicalderivation,asasequenceofoperationsmodifyinginputsinastepbystepfashion,istakenoverbytheGenerator't1(Gen).Roughlyspeaking,theresultsofphonologicaloperationsareincludedinthevastclassofimprovisationsthatGenproducesforagiveninputastheassociatedcandidateset,whichisthenevaluatedbytheconstrainthierarchyofaparticulargrammar.Centraltoanenterpriseofthiskindareconstraintsthatregulateinput/outputdisparities.Thisrolefallstotheconstraints  .  ParseandFillintroducedinPS1993,knowncollectivelyastheFaithfulnessfamilyof ;  constraints.  Faithfulnessfavorsminimaldeviation:Parsemilitatesagainstunderparsing(deletion: E  afailuretoparseaninputelementintheoutput),whereasFillmilitatesagainstoverparsing R  (insertion,afailuretofillaphonologicalpositionwithunderlyingmaterial,i.e.theappearanceofoutputelementswhicharenotpartoftheinput).  &  6      ׀Sincewearehereconcernedwithelements  \ atthefeaturallevel(andnotjustwithwholesegments)andwithassociationrelationsbetweenelementsoffeaturestructure,itisincumbentuponustoaddresstheissueofFaithfulnessinamoregeneralway.Faithfulness,understoodasabanondisparitiesbetweeninputandoutput,mustcoverbothsubstanceandstructure.Besidesmacroelementslikewholesegments,Faithfulnessgovernsindividualfeaturesaswellasassociationrelationsbetweenfeatures. -"   BuildingonPS1993andonourearlieroptimalitytheoreticwork(It=,Mester,&Padgett1993),aswellasthatofothers(McCarthy&Prince1993,Myers1993,Archangeli&Pulleyblank1993),  &  7      ׀weextendtheFaithfulnessconstraintsasin(  MUTZ26  ).Sincetheindividual G( constraintsshownbelowneednotbedistinguishedintherankinghierarchytobediscussed,wegroupthefamilyofFaithfulnessconstraintsintoasinglecollectiveconstraintFaith. ", &  (2  26  3   ; ) Faith (FeatureFaithfulness) k% 0   ParseFeat:   Allinputfeaturesareparsed. &"2   FillFeat:   Allfeaturesarepartoftheinput. '(w#4   ParseLink:   Allinputassociationrelationsarekept. )$6    FillLink:   Allassociationrelationsarepartoftheinput.'k%) *1&8   Gencaninsertnewstructure/features,andfailtoparseinputstructure/features,atthe priceofaccruingFaithviolations.Theseconstraintsassignthecostcorrespondingtoanappeal  ] toautosegmentaloperationslikespreading,insertionordeletioninarulebasedderivationalanalysis.  WewillpresentlyseethatFaithisdominatedbyLicenseandNasVoi:Asa $ t consequence,Faithisviolatedunderthepressuretofulfilloneofthesehigherranking    constraints.FirstletusseeamoreelementaryroleforFaith:thisconstraintmilitatesagainst  .  thesheergratuitousinsertionordeletionofafeaturelike[voice].  Considerfirsttheform/kaki/persimmonin(  MUKI27  ),withtwounderlyinglyvoicelessobstruents.Asmembersofthecandidatesetassociatedwiththisinput,Gensubmitstherepresentationsin(  MUKI27  ),withandwithoutinserted[voice].(Hereandthroughout,insertedelementsarebolded.ThedashedlineinthetableauindicatesthatthereisnoconstraintrankingargumentregardingFaithsofar.).  \ &  (2  27  3   < )0  Input:i(#(# =XX+XXq*y?@ dddd dd d"td"y,Bdd ,tt +  ," R `, 'AAAA * '/kaki/1'%*  ` AAAA 1*WCD ddBdd tt ?@W,dd ,dd ,d",td",td"+  ,"  @x, %*z % *E*z *License 0A!*zE 0NasVoi 0F!*zA 0Faith FE9!*z x @ F Fa.+ 5E(!E 5kaki 7EE(*z#E 7 0EA!!EE 0 0EF!!EA 0 AA4! EF Ab. 5A(z%A 5gaki  B& V 7AE( Z'A 7 0AA!z%AE 0 0AF!z%AA 0*! FA9!B&  AF Fc. 5A(Z )A 5kagì  "!r* ЀV 7AE(!:+A 7 0AA!Z )AE 0 0AF!Z )AA 0*! FA9!"!r*  AF Fd. 5A(:#-A 5gagi  󀀀   $R.  V 󀀀 V 7AE($ /A 7 0AA!:#-AE 0 0AF!:#-AA 0*!*1'%$R.   AF 1XqXX=X'iAllcandidatesfulfillthelicensingconditionandtheredundancycondition(see( KAGIKAKI22  )above);andthecandidate(  MUKI27  a)thatisfaithfultotheinputwins.  &  8      ׀Inadditiontotheconstraintsdepictedhere, L(#5 candidate(  MUKI27  d)withtwoinserted[voice],violatestheOCPover[voice](seesection2).Ontheotherhand,givenunderlying/kagi/key,thecandidate( KAGITABL28  a)withavoicelessmedialobstruent, +V&9 isunfaithful,sinceithaslefttheunderlying[voice]featureunparsed(indicatedbycrossingout).& V (2  28  3  ) I 0  Input: ` j(#(# =XX~ XXq*JK dddd dd d"td"td"CD,Cdd ,tt +  ," S ,Ѐ 'AAAA + '/kagi/̀ ̀V1'%k    AAAA 1*WLM ddCdd tt JKW,dd ,dd ,d",td",td"+  ," C   @x, %   % *E   *License 0A!  E 0NasVoi 0F!  A 0Faith FE9!   x @ F Fa. 5E(+ { E 5kakì  ̀ V 7EE( E 7 0EA!+ { EE 0Ѐ 7EF( C "EA 7*! HA;# C " EF Hb.+ 5A( [ A 5kagì ̀V 7AE( A 7Ѐπ 0AA!# AE 0 0AF! [ AA 01'% [    AF 1XqX X=X'V jThusallelsebeingequal,Faithmakesthechoiceinfavorofthecandidatewhichremains  faithfultotheinput. N   Letusnowreturntoformslikekamipaper.Assuminganinputinwhichthenasalis z unspecifiedfor[voice],Faithwouldseemtointeractinnointerestingwaywithouranalysis, * as(  NVOI29  )suggests.Underspecified/m/againemergesasthewinner;thevoicespecifiedcandidateonlylooksworseforthepointlessinsertionofunlicensed[voice].&  (2  29  3   O )0  Input:=XX~ XXqA#(#(# *PQ dddd dd d"td"td"LM,Cdd ,tt +  ," *$ (, l 'AAAA b% '/kami/1'%*z&  ( AAAA 1*WRS ddCdd tt PQW,dd ,dd ,d",td",td"+  ," R( @x, %z) % *Ez) *License 0A!z)E 0NasVoi 0F!z)A 0Faith FE9!z) x @ F Fa.+ 5E( :+E 5kamì̀ 7EE(z"-E 7 6EA>! :+EE 6Ѐ* =EF>(!,"EA> = HA;#!," EF> %Hb. 5A(#/A 5kamì   $0 Ѐ V 7AE(Z% 1A 7Ѐπ*! -AA>$0AE - -AF>#/AA> -*1'%$0 %  AF> 1 XqX X=X' t   Faithfulnessconsiderationsbecomemoreinteresting,however,whenwetakeinto '"5 accountalternativeinputrepresentations.Supposethatthenasalisspecifiedfor[voice] (,$7 underlyingly,asin( FAI30  ).Forthisinputtosurfacewith[voice]unspecified,asrequiredfor <*%9 Japanese(asevidencedbytheRendaku/LymansLawevidenceinsection2,cf.ori g ami,etc.) +&; [voice]mustgounparsed,inviolationofFaith.Thisoutput,unfaithfultotheinput,isensured  providedFaithisoutrankedbyLicense. ` & V (2  30  3   U )0  Input: ` m(#(# =X!XXXq*VW dddd dd d"td"td"RS,Cdd ,tt +  ," V ,Ѐ 'AAAA . '/kami/̀ ̀V1'%n    AAAA 1*WXY ddCdd tt VWW,dd ,dd ,d",td",td"+  ," F   @x, %   % *E   *License 0A!  E 0NasVoi 0F!  A 0Faith FE9!   x @ F Fa.+ 5E(. ~ E 5kamì  ̀ V 7EE( E 7 6EA>!. ~ EE 6Ѐ* =EF>( F "EA> =* HA;# F " EF> Hb. 5A(^ A 5kamì ̀V 7AE( A 7Ѐ*! -AA>& AE - -AF>^ AA> -1'%^    AF> 1mXq#XX=X'V ma  Traditionalunderspecificationistthinking,withitsinputfocussedfeatureminimizationprogram,mightbebaffledattheideaofevenconsideringthepossibilityofinputrepresentationsasin( FAI30  ).ButwithinOptimalityTheory,thiskindoflegislationoninputshasnodirectcounterpart,andoneiscompelledtoaskwhat,ifanything,governsthedegreeofspecificationininputrepresentations.Thenotionunderlyingrepresentationtakesonasomewhatnewstatusinatheoryinwhichtheoutputisdeterminedbytheparallelconsiderationof(potentiallylimitless)candidatesets(seePS1993:175196).Centraltotheoptimalitytheoreticenterpriseisthehypothesisthatexplanationcanbeachievedthroughoutputconstraintsalone.Therefore, N% neitherunderspecification,noranythingelse,canbemeaningfullyrequiredofinputs(inanycase,notdirectly;seePS1993:4951forfurtherdiscussiononMParse).  [)   Thestrongest,andthereforemostinteresting,hypothesisisthatoutputfocussedconstraintsarenotonlynecessary,butalsosufficient.Thismeansthatthegrammarcontainsnoseparateconstraintsystemgoverninginputswhichcouldenforceaparticulardegreeof(under)specification.Rather,ineverycasetheoutputconstrainthierarchyitselfmustbeabletoforcethecorrectoutcome,irrespectiveofthedegreeofinput(under)specification. T &  9      ׀Otherviews &,"3 arecertainlypossible,butintheabsenceofanyevidencewarrantingamorecomplextheorywithanadditionallevelofinputconstraints,thestrongerhypothesismustbemaintained.Inourterms,underspecificationeffectsareoutputeffectsanddependsolelyontherankingLicense *C&9 >>Faith.Underspecification,then,isanemergentpropertyoftheoutput.Whetherredundant  voicingisunderlyinglypresent( FAI30  )ornot(  NVOI29  ),theoutputsconvergeonthesameoutputcore ` (parsedsubstructure),with/m/notlinkedtovoice.(ThisargumentismodelledontheapproachtosegmentinventoriesinPS1993:178185).  Considernowtheconstrainttableauin( TOM31  ),whichevaluatescandidatesbuiltontheinput/tompo/(fortheJapaneseformtombodragonfly).Asnotedearlier,weassumethecandidate    settoincludeanyautosegmentalimprovisationontheinput"insertionandnonparsingoffeatures,etc.SinceourfocusisonthemedialNCclusterandthefeature[voice],weindicateonlythepresenceorabsenceofrelevantvoicingspecificationsintheinputandinthecandidateforms.& 6 (2  31  3  ) [ 0  Input:=X&XXXq[ (#(# *\] dddd dd d"td"td"XY,Bdd ,itt +  ," D `, l 'AAAA  '/tompo/1'%  ` AAAA 1*W^_ ddBdd itt \]W,dd ,dd ,d",td",td"+  ," l @x, %4 % *E4 *License 0A!4E 0NasVoi 0A!4A 0Faith FE9!4 x @ A Fa. 5E(TE 5tompo 7EE(4E 7 0EA!TEE 0*! 6EA>!TEA 6Ѐ HA;#T" EA> THb. 5A(4 A 5tompò  󀀀 L! Ѐ V 7AE("A 7*! -AA>L!AE - -AA>4 AA> -Ѐ̀* FA9!" T AA> TFc. 5A(d$A 5tombò   ,% Ѐ V 7AE(&A 7 0AA!d$AE 0*! 6AA>!,%AA 6* MA@(,%" T AA> TMd.+ 5A(D(A 5tombò    ) Ѐ<  V 󀀀 7AE(!*A 7 0AA!D(AE 0Ѐπ -AA> )AA -** FA9! ) T AA> TFe. 5A(",,A 5tombò   󀀀 #- Ѐ VV 7AE(l$.A 7*! -AA>#-AE -Ѐπ -AA>#-AA> -̀**1'%l$. T  AA> 1(Xq,XBX=X'6[(Thetwocandidates( TOM31  b,e)failfornowfamiliarreasons,fallingvictimtoLicense.Candidates &!2 ( TOM31  a,c)areeliminatedbyNasVoi.Thedoublylinkedcandidate( TOM31  d)emergesas '>#4 optimal"thoughviolatingfaithfulness"sinceitsatisfiesbothhigherrankedconstraints.HereweseethecrucialrankingofNasVoi>>Faith:Werethisrankingreversed,thefaithfulcandidate *%8 ( TOM31  a)(havingpassedlicensing)wouldemergethewinner.InsteaddominantNasVoiforcesthe ,U': insertionof[voice]anddoublelinking,resultinginmultipleviolationsofFaith.Thus,post    nasalvoicingdependsontheassumptionthatinsertingandspreading[voice]costslessthanviolatingtherequirementthatnasalsbevoiced.  Turningnowtothequestionofinputrepresentations,wewillusethesameexampleand  considerseveralalternativeinputrepresentationsfortheJapaneseformtombo.Aspotential ' w alternativeinputs,wechoosethefiverepresentationsin( TOMBO532  )identicalwiththeoutputcandidates    in( TOM31  a-e).Anticipatingthemainresult,wewillseethatallofthesepossibleinputsconvergeonanoutputcorewithdoublelinkingof[voice].&  (2  32  3  ) ` 0  =X-X$XXqa. ` b.  c.  d. p e. (#(#   tompo tompo h tombo  tombo  tombo R     `     h      p     j     `  V   h V    $,V p   VV 9 Xq/XX=X'b:n:  Wehavealreadyevaluatedthecandidatesetassociatedwithinput( TOMBO532  a)intableau( TOM31  )above.Belowweestablishthatdoublelinkingfor[voice]isthefateofalltheotherinputsaswell.ThisresultfollowsfromthefactthatFaithisoutrankedbybothLicenseandNasVoi. T Theselatterconstraintswillalwaysconspireinthewaywehaveseentoeffectdoublelinking,regardlessoftheamountoffeatureinsertionornonparsingrequired.Thusinthefollowingtableauforinput( TOMBO532  b),Faithtoleratestheinsertionofanotherassociationlineontheoptimal k   candidate( TOMBO1033  d).Thefaithfulcandidate( TOMBO1033  b),ontheotherhand,fatallyviolatesLicense. " &   (2   33  3  ) a =X-XXXqԀ  *bc dddd dd d"td"td"^_,Bdd ,tt +  ,"  , l Input: 'AAAA 9 'Ѐ/tompo/̀ ̀V1'%   AAAA 1*Wde ddBdd tt bcW,dd ,d ,dd",td",td"+  ," Q  @x, .E  .Candidate 2( E 2License #A   #NasVoi 0A! A 0Faith FE9!  x @ A Fa. ;EE(9  E ;tompò  󀀀̀ V 󀀀 5E(  EE 5 0EA!9  E 0*! 6EA>! Q EA 6Ѐπ* HA;# Q " EA> Hb. ;AE(i A ;tompò ̀V 5A( AE 5*! -AA>1 A -Ѐπ -AA>1 AA> -̀ FA9!   AA> Fc. ;AE(I A ;tombò      Ѐ V fXXX=XV=X2X XfXԀ 5A( AE 5 0AA!I A 0*! 6AA>! AA 6Ѐπ** MA@( "  AA> Md.+ ;AE()A ;tombò     ЀV 5A(qAE 5 0AA!)A 0Ѐπ -AA>AA -* FA9!  AA> Fe. ;AE(A ;tombò     ЀV V 5A(QAE 5*! -AA>A -Ѐπ -AA>AA> -̀*1'%   AA> 1AXq3XX=X'@  Intableau( TOMBO1134  )(forinput( TOMBO532  c))theestablishmentofanewassociationlineissimilarly requiredin( TOMBO1134  d).Inthiscase,thefaithfulcandidate( TOMBO1134  c)incursafatalviolationofNasVoi. #" &   (2!  34  3  ) g =X2XXXq*hi dddd d dd"td"td"de,dd ,tt +  ,"  , l Input: 'AAAA 9 '/tombo/̀ ̀V1'%   AAAA 1*Wjk dddd tt hiW,dd ,d ,dd",td",td"+  ," Q  @x, 'E   'ЀCandidate +! E +License #A   #NasVoi 0A! A 0Faith FE9!  x @ A Fa. 4EE!9  E 4tompò  ̀ V .E!  EE .  7EA(9  2E 7*! 6EA>! Q EA 6Ѐ̀* HA;# Q " EA> Hb. 4AE!i A 4tompoπ    1  Ѐ V V .A! AE .*! -AA>1 A -Ѐπ -AA>1 AA> -̀** FA9!q   AA> Fc. 4AE! A 4tombò ̀V .A!QAE . 0AA! A 0*! 6AA>! AA 6Ѐπ MA@( "  AA> Md.+ 4AE!A 4tombò   i ЀV .A!1AE . 0AA!A 0Ѐπ -AA>iAA -* FA9!i  AA> Fe. 4AE!A 4tomboπ   I Ѐ V V .A!aAE .*! -AA>IA -Ѐπ -AA>IAA> -̀*1'%a   AA> 1eNXq6XX=X'ON  Inthenexttableau(forinput( TOMBO532  d))weseeahappymeetingoftheneedsofallofourconstraintsatonce:inadditiontofulfillingbothLicenseandNasVoi,theoptimalform( TOMBO1235  d) # isalsodevoutlyfaithfultotheinput,afactthatwillbecomesignificantinourdiscussionof lexiconoptimizationbelow. M' &   (2"  35  3  ) l Ԁ=X8XXXq*mn dddd d dd"td"td"jk,Bdd ,itt +  ,"  , l Input: 'AAAA 9 '/tombo/̀ 9 Ѐ$ V1'%   AAAA 1*Wop ddBdd itt mnW,dd ,d ,dd",td",td"+  ," X  @x, .E  .Candidate 2( E 2License #A   #NasVoi 0A! A 0Faith FE9!  x @ A dFa. ;EE(@  E ;tompò<     X  Ѐ V 5E( ' EE 5 0EA!@  E 0*! 6EA>! X EA 6Ѐ̀** QAA@( X " d EA> TQb. ;AAE('w AA ;tompoπ$   󀀀 ?  Ѐ$ V 9AA( AAE 9*! -AAA>? AA -Ѐπ -AAA>? AAA> -̀* C9!   T AAA> 4xCc. .E^  .tombò   &  ЀV 2( E 2 #A ^  #*! 6A>!& A 6Ѐπ* MA@(& " x 4 A> DMd.+ ;AE(EA ;tombò$    Ѐ< V 5A(AE 5 0AA!EA 0Ѐπ -AA> AA - FA9!  D AA> Fe. ;AE(,A ;tomboπ     Ѐ V $l V 5A(tAE 5*! -AA>A -Ѐπ -AA>AA> -̀**1'%4   AA> 1z[Xq9XX=X'd[  Thelastinputtoconsider,( TOMBO532  e),startsoffwithseparate[voice]specificationsonthemembersoftheNCcluster;again,theoptimalform( TOMBO1336  d)showsdoublelinkinginstead,ata costtoFaith. c% &   (2#  36  3  ) q Ԁ*rs dddd d dd"td"td"op,Bdd ,itt +  ,"  ,=X<XcXXq l Input: 'AAAA 9 '/tombo/̀  ̀VV1'%   AAAA 1*Wtu ddBdd itt rsW,dd ,d ,dd",td",td"+  ," Q  @x, .E  .Candidate 2( E 2License #A   #NasVoi 0A! A 0Faith FE9!  x @ A Fa. ;EE(9  E ;tompò   ̀ V V 󀀀 5E(  EE 5 0EA!9  E 0*! 6EA>! Q EA 6Ѐ̀** HA;# Q " EA> Hb. ;AE(i A ;tompoπ   󀀀πV V 5A( AE 5*! -AA>1 A -Ѐπ -AA>1 AA> -̀* FA9!   AA> Fd. ;AE(I A ;tombò  ̀ V V 5A( AE 5 0AA!I A 0*! 6AA>! AA 6Ѐπ* MA@( "  AA> Mc.+ ;AE()A ;tombò   <     ЀV V 5A(qAE 5 0AA!)A 0Ѐπ -AA>AA -** FA9!  AA> Fe. ;AE(A ;tomboπ  ̀VV 5A(QAE 5*! -AA>A -Ѐπ -AA>AA> -̀1'%Q   AA> 1hXq>XX=X'h  Thisexperimentwithvariouspossibleinputsbearsoutourearlierclaimthatthereisnoneedforaseparatetheoryofunderlyingfeatureminimization:Theconstrainthierarchyitselfforcesthecorrectoutput,irrespectiveofspecificationintheinput.Thismeansthatthereisno 0$ grammaticalimperativeagainstevenaredundantlyspecifiedinputformasin( TOMBO1235  )or( FAI30  )above. &   However,theremaywellbelearnabilityfactorsrestrictingthechoiceoftheunderlying =( form,requiringthattheproperunderlyingrepresentationbeinferable,asthesimplestchoice,fromtheconstrainthierarchy.Thelanguagelearnerhasathis/herdisposalthestrategyofLexiconOptimization(Stampeanoccultation;seePS1993:192,196forafullstatement):& t (2$  37  3  ) v Ԁ  LexiconOptimization: &"2 0  Ofseveralpotentialinputswhoseoutputsallconvergeonthesamephoneticform,chooseastherealinputtheonewhoseoutputisthemostharmonic.'td%w (#(#   *(&8 Inordertodevelopthesesomewhatabstractconsiderationsintoaconcreteanalyticalmethod, weproposethetableaudestableauxtechniquein( LEXOPT238  ).TakinguparemarkinPS(1993,192),wecompareeachofthewinningoutputsseenaboveforharmonicstatus,eachinrelationtothecorrespondinginput.Thetableaudestableaux( LEXOPT238  )assemblestheinputoutputpairingsestablishedinthetableaus( TOM31  )!( TOMBO1336  ),withthesetofviolationmarksforeachconstraint.Allofthewinningoutputsaredoublylinkedfor[voice],satisfyingbothLicenseandNasVoi;they    differonlyinviolationsoflowrankedFaith.Asshown,thesuperhandchooses( LEXOPT238  d)asthe  .  optimalinput(i.e.,theinputassociatedwiththemostharmonicofthedifferentoutputs).&  (2%  38  3  ) w ԀTableaudestableaux:evaluatingoutputsofthedifferentinputs=X@XXXq*xy dddd d dd"td"td"tu,Add ,Add ,td ,d ,dd",td",td"+  ," I  @x, %  %Input *E  * ;E( E ;Output 2( E 2License #A   #NasVoi 0A! A 0Faith FE9!  x @ A tFa. 5E(1E 5/tompo/ 7EE(1E 7XBXX=X+#=XۇXXX# 1  ;EE(VEE ;tombò$  $n   I Ѐ V 󀀀 5E(EE 5 0EA!1E 0 0EA!1EA 0*!* MA@(I" t EA Mb. 5A(iA 5/tompo/̀ ̀V 7AE(A 7XBX@X=X+#=XۇXXX# i  ;AE(>AE ;Ѐtombò   󀀀 1 ЀV 5A(AE 5 0AA!iA 0 0AA!iAA 0*! JAA9!1  AA Jc. 9AA(QAA 9/tombo/̀ πV 9AEA(AA 9XBX@X=X+#=XۇXXXX# Q  ;AAE(&vAEA ;tombò    ЀV 9AA(AAE 9 2AAA!QAA 2 2AAA!QAAA 2*! C9!   AAA 4xCd.XBX@X=X+#=XFXXX#Ԁ %1! %/tombo/̀ " Ѐ$9 V *Ex# *XBX!X=X+#=XۇXXX# 1!  ;E(V"E ;tombò< " ЀV 2(x#E 2 #A 1! # 0A!1!A 0 MA@(1!" x 4 A Me. 5A(%A 5/tombo/̀  ̀VV 7AE(X 'A 7XBX@X=X+#=XFXXX,#Ԁ ;AE(%AE ;tombò     & ЀV V 5A(` 'AE 5 0AA!%A 0 0AA!%AA 0*!*1'%&   AA 1e}XqIX!X=X'Y}  WithLexiconOptimizationandtableaudestableauxasaguideline,thelearnerchoosestheinputformthatmapsontoanoutputinthewayleastoffensivetothegrammarofrankedconstraints.Forourpurposes,thismeansthatthelearnerwillchoosetheinputleadingtothefewestfaithfulnessviolations,towit,theinputbearingdoublelinkingofredundant[voice].Thisconclusion,ifcorrect,pointsupevenmoredramaticallyourbasicconclusion:thereisnorequirementofunderlyingfeatureminimization.  Torecapitulatethissectionandtheoverallanalysis,werelyonthreebroadcategoriesofconstraints"redundancyconditions,derivativelicensingrestrictions,andconstraintsagainst ,`'9 input/outputdisparities.Eachoftheseislargelyakintonotionswidelyheldinmodernphonology.Ourapproachtounderspecificationisnovelintwoimportantrespects,however.First,ourinterpretationoflicensingputsnopenaltyonfeaturecooccurrenceperse;rather,afeatureismerelyrequiredtobelicensedinsomefashion.Second,werelycruciallyonthe  rankingandviolabilityofconstraints,notionsthatformthebackboneofOptimalityTheory.  Continuinginthislastvein,itmayseemodd,eveninthecontextofOT,toproposesimultaneousconstraints,onedemandingthatnasalsbevoiced(NasVoi),andtheotherblocking  1  licensingof[voice]bynasalsegments(License).Yetasnotedattheoutsetofthissection,these A  arejustfamiliarnotionsinanewguise,andtheymustberegardedastwosidesofacoin:theimportofNasVoiisthatvoicingisinasenseinherentinnasals(Stevens,Keyser&Kawasaki K  1986).Yetitissurelytheveryredundancyofsuchvoicingthatalsoentailsitsphonologicalinertness.Whiletheantagonismbetweenthesetwoconstraintshasledpasttheoriestoassignthemtocomplementarylevelsofthegrammar(i.e.underspecificationearlyandpresenceofredundantvalueslate),wefindanewpossibilitywithinOT:redundantspecificationisnotsimplyirrelevantuntillateinthederivation,butratheriscruciallyaconstantdemand.ThefactsofJapaneseprovidestrikingsupportforthisview.&  5SonorantVoicingandMultipleLinkingConditions 6" 5.1Theproblem  C$ AsmallstepbackfromYamatoNCclustersrevealssomehightlyrelevantquestionslurkingnearby.Ouranalysishassofarnotaddressedanimportantlimitationoftheprocess:Itisonly'Inasalsonorantsthattriggervoicingonfollowingobstruents(/yom+te/yondereading Z!* (*yonte),but/mi+te/miteseeing(*mide)).Insinglingoutnasalsfromallsonorantsas " , voicingtriggers,YamatoJapaneseisfollowingapatternparalleledinotherlanguages,includingZoque(Penutian;Wonderly1951),Mwera(Bantu;Harries1950,KenstowiczandKisseberth1977),amongothers.Whilecomforting,suchcrosslinguisticsupportisnotenoughtoanswercertainquestionsofamorefundamentalnature:  (i)NonuniversalityofNCvoicing:Whydon'twefindpostnasalvoicinginalllanguages,includingEnglish(impossible,*imbossible)?EvenwithinJapanese,NCvoicingisrestrictedto *;&8 thenative(Yamato)stratum(aswitnessedbywellknownnonYamatoexamplesliketempuraor  shinkansen). `   (ii)Nasalsaspriviledgedvoicingtriggers:Ifvoicingspreadtakesplaceinordertoallowforthepresenceofredundantvoicing(byfindingalicenser),whyshouldthisberestrictedtonasals?Whydon'tnonnasalsonorants,sayvowels,spreadtheirredundant[voice]inaparallelway?  Toillustratethesecondquestion,considertheconsequencesofreplacingtheredundancyconditionNasVoiinourcurrentanalysisbythemoregeneralSonVoi( SONVOI1039  )requiringall A  sonorants,notjustnasals,tobevoiced.(2&  39  3  )    SonVoi:[son]D[voice] X  SonVoiexpressesageneraltruthaboutsegmentstructure,andisoperativewithintheconstraint & systemofalllanguages,includingJapanese.Foraninputlike/aki/(forakiautumn),replacing 3 NasVoibySonVoiintheanalysisresultsintheincorrectmultiplylinkedoutput*agi,shown  in( SONVOI240  d).Moregenerally,allsonorantsarewronglypredictedtocausevoicinginneighbouring C  obstruents. S  &   (2'  40  3  )  * ddAdd Add td d dd"td"td"xy,Bdd ,td ,td +  ,"  J, l Input: %AAA  Z %XFX#XXq/aki/XqKX!XX 0A!!qAAA 0Ѐautumn1'%!q  J A 1*h ddBdd td td h,dd ,dd ,d",td",td"+  ," < ax,XFX $XXq %d " %Candidate *Ed  *License 0A!d E 0SonVoi 0A!d A 0Faith#XqKX!XXX# FE9!d  x a A (FXMX $XXqa.!!  $  intended    winner: 2    5E(  E 5aki 7EE( $ E 7 0EA! $ EE 0*!* -EA> $ EA -#XqKX$XXۼ# FA9! $  ( EA> "FXMX $XXqb. 5A(h A 5aki  󀀀   0   V 󀀀 V 7AE( A 7*!* -AA>0 AE - -AA>h AA> -**#XqKX$XXn# JAA9!0  " AA> 2JXMX $XXqc. 9AA(H AA 9agì     Ѐ V 9AEA( AA 9Ѐπ 2AAA! AEA 2*!* -AAA> AAA -*#XqKX$XX # C9!   2 AAA> xCXMX $XXqd. %( %agì     Ѐ V *Eh * 0A!(E 0*! -A>A -**#XqKX$XX۪# FA9! x  A> "FXMX $XXqe.#XqKX$XX#XBX $XXq+#XqKXXXd#XMX $XXq  wrong - winner: 5A(sA 5agì     Ѐ V 7AE(PA 7 0AA!AE 0 -AA>AA -***#XqKX$XX۪#1'% "  AA> 1'Aswewillsee,thetwoquestionsposedabove(restrictiontonasalsandnonuniversality)arecloselyrelated;bothwillrequireappealtoconstraintsonthemultiplelinkingoffeatures.& v 5.2ANonviableApproach:SonVoiDecomposition  & Onepotentialsolution,attractiveatfirstsightbutultimatelynotworkable,reliesonthedecompositionofSonVoiintoauniversallyrankedfamilyofsonorantvoicingconditions,asin H!* 'vK( SONCOND41  )(seeIt=,Mester&Padgett1993foradetailedexpositionofthiskindofapproach).&  (2(  41  3  )  Ԁ  SonorantVoicingConditionsandConjecturedRanking: _% 0   NasVoi: [+nasal]D[voice] & "2   ApproxVoi: [+approximant]D[voice] (i#4   VocVoi: [consonantal]D[voice] v)$6    NasVoi>x >ApproxVoi>">VocVoi'$ *#&8 Thenotionofafamilyofconstraintswithintrinsicrankingisoneofthecentralanalyticaltools ofOptimalityTheory.Theindividualmembersofafamilyofconstraintsstandinparticulardominancerelationswithrespecttoeachother(givenuniversally),butneednotallbeadjacentintheconstraintrankingofaparticularlanguage(cf.PS1993onsyllablesonority).Thepossibilitythatotherconstraints"mostimportantly,membersoftheFaithfulness $ t family"mightinterveneatcertainjuncturesintheoverallhierarchyconstitutesanimportantareaofvariationbetweenphonologies.  Forthecaseathand,themostdirectattackontheproblemofdistinguishingnasalvoicingfromothersonorantvoicingistorankFaithstrategicallybetweenNasVoiand   ApproxVoi/VocVoi,asin( APPROX1042  ). E  (2)  42  3    )  NasVoi> > "'"Faith "'"Ԁ>>ApproxVoi>W>VocVoi  TheoccurrenceofpostnasalvoicingfollowsfromthedominanceofNasVoioverFaith,which  ishenceviolatedinordertoensurecompliancewithNasVoi.Othersonorantredundancy C conditionsarerankedbelowFaithandcannotcommandviolationsofit.Thisrankingyieldsthe P correctoptimaloutputforaki,asshownin( AT43  ).(Forourdiscussionhereandbelow,the   distinctionbetweenApproxVoiandVocVoiisnotrelevant,andwewillsimplifythe  ]"  presentationbyrestrictingourattentiontothelatter.) j$ &   o[X` hp x (#%'0*,.8135@8:<H?AXo(2*  43  3  )      * dddd dd d"td"td",Bdd ,td ,td +  ,"  J, l Input: %AAA  Z %XMX $XXq/aki/XqTX$XX 0A!!qAAA 0Ѐautumn1'%!q  J A 1=XMX\(XXq*h ddBdd td td h,dd ,dd ,d",td",td",td"+  ," < @x, %d " %Candidate *Ed  *License 0A!d E 0NasVoi 0A!d A 0Faith 0A!d A 0VocVoi FE9!d  x @ A 8Fa.XqTX$X=X򀀀+=XWX\(XXq 5E( $ E 5aki 7EE(  E 7 0EA! $ EE 0 0EA! $ EA 0 -EA> $ EA -** FA9! $  8 EA> Fb. 5A( < A 5aki  󀀀       V 󀀀 V 7AE(| A 7*!* -AA>  AE - -AA> < AA> -** -AA>  AA> - FA9!    AA> Fc. 5A( A 5agì      Ѐ V 7AE(d A 7 0AA! AE 0 0AA! AA 0*!** -AA> AA -1'%    AA> 1o'XqTX)X=XInsertionofvocalic[voice],andthesharingofthisfeaturewithalicenser,asin( AT43  c),isno X longerpossible;inspiteoftheconditiondemandingthatvowelsbevoiced,thedominantpositionofFaithentailsthattheoptimalformisunspecified.  &  10      ׀ b   ThoughthedecompositionofSonVoiintothesemoreparticularconditionshasnoclear o independentsupportwithinJapanesephonology,itisanaturalstepwithinOTthatcanbeexpectedittobecalledforinothercases.Moreworrisomeistheconjecturedintrinsicrankingofthisconstraintfamily:Withitsimplicationthatthevoicingofanasalisinitselfofmoreconsequencethanthevoicingofothersonorants(suchasvowels),itlacksanyobviousgroundinginphoneticfactsormarkednessconsiderations.  ThemostseriousproblemwiththeSonVoidecompositionanalysis,however,isthefact ( thatitleadstounwantedinsertionandlinkageof[voice]informscontaininganasalanda nonadjacent obstruent,suchasmakifirewood.NothingsaidsofardistinguishessuchNVC W",  W", formsfromNCformsliketombodragonflyanalysedinsection4,andtableau( LOCAL144  )illustrates #. howtheSonVoidecompositionapproachwronglypointsto*magi,withinserted[voice]spread %g 0  tothemedialobstruent,astheoptimaloutputforunderlying/maki/. w&!2 &   (2+  44  3  )  ԀWrongcandidateselected:* dddd dd d"td"td"td",Bdd ,td ,td +  ,"  J, l Input: %AAA C %XWX\(XXq/maki/Xq[X)XX 0A! ZAAA 0Ѐfirewood1'%C  J A 1*h ddBdd td td h,dd ,dd ,d",td",td",td"+  ,"  % ax, %M " %Candidate *EM  *License 0A!M E 0NasVoi 0A!M A 0Faith 0A!M A 0VocVoi FE9!M  x a A (F=XWX+XXqa.!!  .  intended    winner: <    5E( " E 5maki 7EE( . E 7 0EA! . EE 0*! -EA> . EA - -EA> . EA> -** FA9! .  ( EA> Fb. 5A("r A 5maki   :   V 7AE( A 7*! -AA>: AE - -AA>: AA> -* -AA>: AA> -** FA9!:   AA> Fc.Xq\X)X=XԀ+=XWX4,XXq R  wrong c winner: 5A(A 5magi      V 7AE(A 7Ѐπ 0AA!AE 0 -AA>AA -** -AA>AA> -**1'%   AA> 1^Xq\X)X=X'H  Lookingbeyondsuchindividualexamples,theanalysisin( LOCAL144  )makesthepredictionthatnasalscannotcooccurwithvoicelessobstruentsinthesameform,apredictionpatentlywrongnotonlyforJapanese,butinalllikelihoodalsouniversally.Onesourceoftheproblemliesinthefactthattheaccountdoesnotrecognizethatpostnasalvoicingisrestrictedtostrictlyadjacentsegments,arestrictionwhichseemsuniversal(weknowofnocasesthatcrossvowels).Whatguaranteesthiskindoflocality?FollowingKiparsky(1981),Levergood(1984)andArchangeli&Pulleyblank(inpress),amongothers,weposittheconstraintin( LOCAL245  )againstgappedconfigurations(cf.alsothe*EmbedconstraintinSmolensky1993). J' & ; (2,  45  3  )  ԀNoGap !* XWX4,XXq   ` *Xq^X)XX y"+ XWX*-XXq   ` Xq^X)XXԀwhereisapotentialbeareroffeatureF H#, XWX*-XXq   ` F 1$- Xq^X)XX'; JVAstheamendedtableau(  AMEND46  )shows,thenewconstraintNoGap(undominated,andunranked %2!/ withrespecttoLicense)rulesoutthepreviouswinner(  AMEND46  c),sinceavowelisapotentialbearer ?'"1 of[voice]"beingapotentialbearerisamatteroffeaturestructure,notredundancy.Hence,the correctform(  AMEND46  a)isselectedinstead.(Inordertoletthecrucialrankingsstandoutinthe )I%5 tableaus,theconventionaldashedlinesbetweenthenoncruciallyrankedconstraintcolumnshavebeenomittedfromhereon.)&  (2-  46  3  )  * dddd dd d"td"td"td",Bdd ,td ,td +  ," S J, l Input: %AAA   %XWX*-XXq/maki/Xq`X)XX 0A!+AAA 0Ѐfirewood1'%   J A 1*h ddBdd td td h,dd ,dd ,d",td",td",td",td"+  ,"    ax, % n " %Candidate *E n  *NoGap 0F! n E 0Licens  n    n  e 0A! L F 0NasVoi 0A! n A 0Faith 0A! n A 0VocVoi FE9! n  x a A F=XWX .XXqa.XqbX)X=XԀ+=XWX/XXqԀ e   5E(v E 5maki 7EE(e E 7 (Ee EE ( 0EA!e E 0*! -EA>e EA - -EA>e EA> -** FA9!e   EA> Fb. 5A( A 5maki  󀀀    V 7AE(V A 7 (A AE (*! -AA> A - -AA> AA> -* -AA> AA> -** FA9!   AA> Fc. 5A(A 5maki    n  V 7AE(>A 7*! (AnAE ( -AA>nA - -AA>nAA> -** -AA>nAA> -**1'%n   AA> 1[XqbX)X=X'jO  However,aconsiderationoffurthercandidatesrevealsthatNoGapisnotsufficientto c solvetheexcessivevoicingproblem.Sincevowelscanbear[voice],therearecandidateswherespreadingproceedsthroughthevowel,insteadofskippingit.Thisleadsstraighttoanotherwayofwronglyderivingthemediallyvoiced*magiinsteadofmakishownin( MUKWRONG47  ).HereNoGap z! succeedsindisqualifying( MUKWRONG47  c),butispowerlessagainsttheungappedbutstillexcessivelyvoiced candidate( MUKWRONG47  d),whichwinsover( MUKWRONG47  a)becauseofNasVoi. 7% &   (2.  47  3  )  ԀWrongcandidateselected:* dddd dd d"td"td"td"td",Bdd ,td ,td +  ,"  J, l Input: %AAA  Z %XWX/XXq/maki/XqdX)XX 0A!!qAAA 0Ѐfirewood1'% Z  J A 1*h ddBdd td td h,dd ,dd ,d",td",td",td",td"+  ," < ax, %d " %Candidate *Ed  *NoGap 0F!d E 0Licens d   d  e 0A!B F 0NasVoi 0A!d A 0Faith 0A!d A 0VocVoi FE9!d  x a A (F=XWX 0XXqa.!!    intended s   winner:  Y   5E(  E 5maki 7EE(  E 7 (E  EE ( 0EA!  E 0*! -EA>  EA - -EA>  EA> -** FA9!   ( EA> Fb. 5A(? A 5maki  󀀀    V 7AE( A 7 (A? AE (*! -AA> A - -AA> AA> -* -AA> AA> -** FA9!   AA> Fc. 5A( A 5maki       V 7AE(gA 7*! (A AE ( -AA> A - -AA> AA> -** -AA> AA> -** FA9!   AA> Fd.XqfX)X=XԀ+=XWX1XXq  wrong  winner: 5A(^A 5magi      V 󀀀 7AE(OA 7 (AAE ( 0AA!A 0 0AA!AA 0*** -AA>AA -* FA9!  AA> Fe. 5A(A 5magi     g  V 󀀀 7AE(7A 7 (AAE (Ѐππ 0AA!/A 0 0AA!gAA 0****! -AA>gAA -1'%g   AA> 1XqfX)X=X'}  TheSonVoidecompositionanalysis,besidesleadingtodescriptivelyincorrectresults, \! alsoentailserratic(under)specificationpatternsinsurfaceformswheneveravoicedobstruentispresentwhichcanserveasalicensinganchorfor[voice].Thisisillustratedbyformswithan f%  underlying voicedobstruentsuchasnabepot(  NABE48  ). s' & K  (2/  48  3  )  ԀQuestionableresult:* dddd dd d"td"td"td"td",Bdd ,td ,td +  ,"  , l Input: %AAA C %XWX1XXq/nabe/  Z Ѐ ̀VXqhX)XX 0A!AAA 0Ѐpot1'% Z   A 1*h ddBdd td td h,dd ,dd ,d",td",td",td",td"+  ," [  ax, % # " %Candidate *E #  *NoGap 0F! # E 0Licens  #    #  e 0A!  F 0NasVoi 0A! # A 0Faith 0A! # A 0VocVoi FE9! #  x a A (F=XWX1XXqa.!!  j  intended  2  winner: x   5E(^ E 5nabè ̀V 7EE( E 7 (E j EE ( 0EA! j E 0*! -EA> 2 EA - -EA> j EA> -** FA9! 2  ( EA> Fb.XqjX)X=XԀ+=XWX2XXq ^   5A(o A 5nabe   &v  ЀV 7AE(> A 7 (A^ AE ( 0AA!&v A 0 0AA!&v AA 0** -AA>&v AA -* FA9!&v   AA> Fc. 5A(VA 5nabè     V ЀV 7AE(&A 7 (A>AE ( 0AA!VA 0 0AA!VAA 0***! -AA>VAA -1'%V   AA> 1XqjX)X=X'KThevoicingspecificationpatternin(  NABE48  )canonlybedescribedasarbitrary:Thewinningcandidateshowsvoicingonvowels,butonlywhentheylieonthepathbetweenavoicecravingnasalandavoicedobstruent(i.e.,wherenecessarytofulfillbothNasVoiandNoGap).  Elsewhere(e.g.,forthefinalvowelinnabe(  NABE48  b)),Faithpreventsvoicingonvowels.Thislast b! pointraisesafurtherdisturbingquestion:SinceFaithisnowcruciallyinvolvedinmakingthe r# finaldecisionregardingthevoicingspecificationsofvowels,theissueofdifferentdegreesofinputspecification(ofredundantvalues)raisesitsheadagain,subvertingtheresultsofsection4.  Thecombinedforceoftheseconsiderationscompelsustopursueanalternativeapproachwhichshiftsthemainemphasisinouranalysisawayfromsegmentfocusedvoicingconstraintstowardssyntagmaticlinkageconstraints.Thebasicideabehindthenewapproachisthefollowing:NCsequences,incontrasttoVCsequences,sharevoicingnotbecausenasalvoicing %M!1 issomehowstrongerthanvocalicvoicing,butbecausefeaturelinkageislessmarkedinNC ]'"3 sequencesthaninVCsequences.Wewillseethatthislineofanalysisenrichesourunderstandingofpostnasalvoicingontheonehand,andconnectswithaninterestinglineofphonologicalresearchontheother. w+&9 Ї& t 5.3NCLinkage     Thetwoproblemsdiscussedabove"thelocalityissue,andtheconfinementofvoicingeffectstoNCclusters"turnouttohaveoneandthesamesolution.Thekeyliesinthenotion,to'tphraseitinformally,thatlikethingsinteract.Thisideahasbeenvoicedintheliteraturebefore  (seee.g.Hutcheson1973,Kiparsky1988),mostclearlyinworkcenteringonOCPeffects(McCarthy1986,1988,Mester1986,Selkirk1988,1993,Padgett1991,Pierrehumbert1993).Theupshotofthislatterworkisthatdissimilatoryeffectsinvolvingplaceofarticulationaremorelikelybetweensegmentsthatarealikeinotherfeatures,notablyminorplacefeaturesandfeaturesofstrictureorsonority.WorkofSelkirk(1988,1993),NChiosin&Padgett(1993),McCarthy(1993),andPadgett(1994)findspromiseinextendingthisideatovariousassimilatoryeffectsaswell.  Inthiscontext,letussupposethatlinkageof[voice]betweennasalsandobstruentsislessmarkedthansuchlinkagebetweenliquids/glides/vowelsandobstruents,becausenasalsandobstruentsarealreadymoresimilarinstrictureandhencemorepronetointeract.  &  11      ׀We l implementthisideabymeansoftheNoLinkfamilyofconstraintsin( LOCAL449  )"withtheadded  caveatthattheindividualconstraintsshouldbetakennotasirreducibleprinciples,butratherasreflectionsofadeepergeneralization(seetheworkscitedabove;thelettersV,G,L,N,Cin( LOCAL449  )denotevowels,glides,liquids,nasalsandobstruents,respectively).&  (20  49  3  )  ԀConstraintfamily:NoLink &   NoVCLink>" >NoGCLink>U>NoLCLink>v>NoNCLink'= G( AsconjecturedconcerningthefamilyofSonVoiconstraintsdiscussedearlier(see( SONCOND41  )insection ", 5.2),thisfamilyisrankedintrinsicallyanduniversally;thisrankingentailsthegeneralizationthatinteractionislessmarkedbetweensegmentsmoresimilarinstricture/sonority.GiventhisdecompositionofNoLink,andthepossibilityotherconstraintstointervenebetweenanypair &"2  ofsubconstraints,weproposethefollowingscenarioforJapanese: %(u#4 &   (21  50  3  )InterventionofSonVoi:     `  NoLinkFamily Z   ...>>NoLCLink>+ >" l,|,'|" SonVoi",| l,'|"Ԁ >>NoNCLink>>...'; ,| Q  Forourpurposes,onlytheseparationofNoNCLinkfromtherestofthefamilyis v  relevant.AsinthediscussionofSonVoidecompositionintheprevioussection,wefocus  #  exclusivelyonthecontrastbetweenVCandNC.Intheseterms,theimportantsubhierarchyoftheconstraintsystemofYamatoJapaneseistheonein( ORDER951  ).& J (22  51  3    )  NoVCLink #         SonVoi     x   NoNCLink'JO V [Thiscorrectlyentailspostnasalvoicingontheonehand,andthefailureofotherpostsonorantvoicingsontheother.Combinedwiththeotherconstraintsthatplayaroleintheanalysis(NoGap,License,andFaith),weendupwiththepartialorderingofconstraintsin( ORDER1052  ).For ' presentpurposes,weareassumingthatNoVCLinkisundominated,onaparwithLicenseand 4 NoGap.(WithintheoverallgrammarofJapanese,NoVCLinkisdominatedbyvarious  constraints"enforcingpalatalizationandothervowelconsonantinteractioneffects"thatarenotrelevanthere,seeIt=&Mester1994a,bfordiscussion).& ' (23  52  3    )   ` License  NoVCLink    NoGap   !&    ` ___________ _________________    `    ` SonVoi A$)    ` ___________   `     &X!+    ` Faith   NoNCLink'' # &A",  #Asanillustrationoftheanalysis,considerfirsttheinput/in+te/dyingin( LOCAL553  ). ($. &   (24  53  3  )  * dddd dd d"td"td"td"td",Bdd ,td ,rtd +  ,"  J, l Input: %AAA  Z %XWX2XXq/in+te/XqpX)XX 0A!!qAAA 0Ѐdie+Gerund1'% Z  J A 1=XWX5XXq*h ddBdd td rtd h,dd ,dd ,t",td",td",tt",dd"+  ," < x, %d  %Candidate *Ed  *License 0F!d E 0NoVC d  Link 0A!, |F 0SonVoi 4AF!d A 4NoNC d  Link +!, |AF +Faith 9E,d  x  89a. 5E(  E 5inte 1EE"d  E 1 (E  EE ( 0EA!  E 0***! 0EA>  EA 0 -E>  EA> - FA9!   8 E> Fb. 5A(  A 5intè H   󀀀 |  ЀH  V 1AE"D A 1*! (A| AE ( -AA>  A -** 0AA>| AA> 0 -A>  AA> -* FA9!|   A> Fc.XqrX)X=XԀ+=XsX6XXq 5A( A 5indeπ H    \  ЀH  V 󀀀 1AE",| A 1Ѐπ (A\ AE ( 0AA! A 0** 0AA>\ AA 0* -A>\ AA> -**1'%\    A> 1&XqrXA7X=X'&Since( LOCAL553  )involvesonlyNClinkage,littlehaschangedfrombefore.Theimportantpointisthat Q SonVoiiscruciallyrankedoverNoNCLink;weretheoppositerankingcorrect,ourinput  wouldsurfaceunchangedas*inte.Tableau( LOCAL553  )shouldbecomparedto( LOCAL654  ),wheretheinput  [ form/aki/containingtheVCsequence/ak/isshownrunningthesameconstraintgauntlet.&  (25  54  3  )  * dddd dd t"td"td"tt"dd",Bdd ,Htd ,td +  ,"   J, l Input: %AAA "r" %XsX6XXq/aki/XquXA7XX 0A!9!AAA 0Ѐautumn1'%9!  J A 1*h ddBdd Htd td h,dd ,dd ,t",td",td",tt",dd"+  ," T% Jx, %|& %Candidate *E|& *License 0F!|&E 0NoVC |& Link 0A!Z'F 0SonVoi 4AF!|&A 4NoNC |& Link +!Z'AF +Faith 9E,|& x J 89a.+=XsX78XXq 5E( 0)E 5aki 1EE"!*E 1 .E! 0)EE . -EA> 0)E -** 0EA> 0)EA> 0 -E> 0)EA> - FA9! 0) 8 E> Fb. 5A("H,A 5aki  󀀀 #-  V 1AE"$.A 1*! %A#-AE % -AA>"H,A -* 0AA>#-AA> 0 -A>"H,AA> -* FA9!#-  A> Fc. 5A(%(!0A 5agi  󀀀 &!1  V 󀀀 1AE"p'"2A 1Ѐπ .A!&!1AE .*! -AA>&!1A -* 0AA>&!1AA> 0 -A>&!1AA> -**1'%&!1   A> 14XqwXA7X=X'y4Asusual,sonorantvoicingmayariseonlyiflicensedbyanobstruentlink;hencethequickdemiseofcandidate( LOCAL654  b).Candidate( LOCAL654  c)lookssuperficiallysimilartothewinningcandidate *B&8 ( LOCAL553  c).ButlinkageofavoweltoanobstruentispreciselywhatisprohibitedbyNoVCLink;  unlikeNoNCLink,thisconstraintranksaboveSonVoi,andsopostvocalicvoicingis  ] impossible,eliminating( LOCAL654  c).Theremainingcandidate( LOCAL654  a),inspiteofitstwoSonVoi m violations,istheoptimalform.WenoteforcompletenessthatnoevidenceheremotivatesarankingbetweenFaithandNoNCLink(wewillshortlyturntoaconsiderationoftheroleof ' w Faithfulnessinouranalysis).      RecallfromtheprevioussectionthattheSonVoidecompositionanalysissufferedfrom  1  anexcessivevoicingprobleminformscontainingbothanasalandanobstruent,whetheradjacentornot.WewillnowshowthatthisNClinkageanalysisisfreeoftheseproblems,andalreadycontainsthesolutiontothelocalityproblembroughtupearlier:TheverysameprohibitiononVClinkagewhichwasoperativein( LOCAL654  )akialsopreventslongdistancepostnasal U  voicingin( LOCAL755  )makifirewood.Thepreviouslyproblematiccandidate( LOCAL755  d),wherevoicing  spreadhassweptthroughthevowel,nowcruciallyviolatesNoVCLink.Thevowelskipping e candidate( LOCAL755  e)continuestoviolatetheconstraintagainstgappedconfigurationspositedabovein( LOCAL245  ).Asdesired,theoptimalcandidate( LOCAL755  a)mirrorstheinputinnotcontainingany[voice] specification. ,| &   (26  55  3  )  * dddd dd t"td"td"tt"dd",Bdd ,td ,td +  ,"  J, l Input: %AAA  Z %XsX-9XXq/maki/XqyXA7XX 0A!!qAAA 0Ѐfirewood1'% Z  J A 1*h ddBdd td td h,dd ,dd ,:t",:dd",:td",td",tt",dd"+  ," < x, %d  %Candidate .EFd  .No d  Gap +! lEF +Lic d  ense #F  l #No d   d  VC  l Link 0A! $ F 0SonVoi 4AF!d A 4NoNC d  Link +!B AF +Faith 9E,d  x  89XsX#:XXqa.XBXA7XXԀ+=XsXXX 5E(4  E 5maki 7EE( L E 7 5E(4  "EE 5 5E(4  "E 5 =EA>(4  "E =*** @EA>+4  "EA> @ =E>(4  "EA> = MA@(4  " 8 E> Mb. 5A(L A 5maki   d   V 7AE(, A 7 5A(L "AE 5*! 5A(d "A 5 =AA>(L "A =** @AA>+d "AA> @ =A>(L "AA> =* MA@(d "  A> Mc. 5A(,| A 5maki   D   V 󀀀 7AE(A 7 5A(,| "AE 5Ѐ*! 5A(D "A 5 =AA>(D "A =* @AA>+D "AA> @ =A>(D "AA> =** MA@(D "  A> Md. 5A(dA 5magi   ,  V 󀀀 7AE(A 7 5A(d"AE 5Ѐπ 5A(,"A 5*! =AA>(,"A =* @AA>+,"AA> @* =A>(,"AA> =*** MA@(,"  A> Me. 5A(LA 5magi     V 󀀀 7AE(A 7*! 5A("AE 5Ѐππ 5A(T"A 5 =AA>("A =** @AA>+"AA> @* =A>("AA> =**8.,"   A> 8HXq}XA7X=X'G  OurunderstandingoftheopacityofinterveningvowelsechoesworkstemmingfromKiparsky(1981),inwhichopaquesegmentsareviewedasnontriggersoftherelevantprocess. &# Inthepresentaccount,however,thisbehaviorisnottheresultofprocessspecificstipulationsontargetandtrigger,butratherfollowsdirectlyfromtherankingofquitegeneralandindependentlymotivatedmarkednessconditions.  Tosumupthediscussionsofar:BesidesauniversalSonVoiconstraint,ouranalysis M"+ makesuseofauniversallyrankedNoLinkconstrainthierarchy,whoseexistenceisstrongly #- supportedbytheempiricaltypologyofdissimilationandassimilationpatternsacrosslanguages.TheonlyrelevantparametertobesetwithinthephonologyofYamatoJapaneseconcernstherankingofthevoicingconstraintwithrespecttotheinteractionconstraintsystem.HereouranalysisinsertsSonVoiimmediatelyaboveNoNCLinkwithintheNoLinkhierarchy.This )n$5  singlemoveaccountssimultaneouslyforthreeclassesoffacts:(i)thepresenceofpostnasal {*%7 voicing,(ii)theabsenceofotherpostsonorantvoicing,and(iii)thestrictlylocalcharacterofpostnasalvoicing.  Itisnowasimplemattertoanswertheotherquestionposedattheoutset:Whyaretherelanguageswithoutpostnasalvoicing?Theanswerissimplythatinsuchlanguages,NoNCLink  joinsotherNoLinkconditionsinoutrankingSonVoi"henceSonVoicannotcommandthe $ t violationofNoNCLink.Weillustratethetwopossibilitiesin( RANKDIFF56  ).    & t (27  56  3  )  ԀRankingdifferencesbetweenNoNCLinkandSonVoi ;  * dddd dd :t":dd":td"td"tt"dd", d ,` td ,` t +  1EE"   1Ranking  0EA! EE 0Outcome 4EAE!@ EA 4Example HAE9!@   EAE qHSonVoi>>NoNCLink ^   0AA!kAE 0e.g./...mp.../+ ^ [...mb...] 4AAE!kAA 4YamatoJapanese,Zoque FAEE5!^ q AAE FNoNCLink>>SonVoi  2AAE!,AEE 2e.g./...mp.../+ , [...mp...] 4AAEE!9AAE 4NonYamatoJapanese,English,etc.1'%9 0   AAEE 1't ;`Moregenerally,andexploitingtheentirehypothesizedNoLinkfamily,wepredictthefollowing  postsonorantmarkednesspattern,derivedbyrankingSonVoisuccessivelyhigherinthe  \! hierarchy:& . (28  57  3  )Postsonorantvoicingpatterns  Nopostsonorantvoicing h   (SonVoirankedbelowentireNoLinkhierarchy) #s'   Postnasalvoicing h      p :  )   Postnasal/liquidvoicing h      p : !-+   Postnasal/liquid/glidevoicing     p : :#-   Postsonorantvoicing h   (SonVoirankedaboveentireNoLinkhierarchy)'.f $/ Whetherthishierarchyiscorrectinallrespects(e.g.,initsexclusionofpostliquidvoicingwithoutpostnasalvoicing"barringofcoursethepossibleinterferenceofotherconstraintswearenotconsideringhere),remainsadomainoffutureresearch.However,theworkcitedearlier onsimilarityininteractionsuggeststhatmarkednesspredictionsliketheseareontherighttrack.  *[%7   WiththeincorporationoftheNoLinkfamilyinourgrammar,itbecomesimportantto   clarifytherelativerolesofFaithfulnessandNoLinkinouraccount.Clearlyfeature  ] Faithfulnessisanindispensableelementoftheoverallgrammar,NCvoicingaside.Itis j Faithfulnessthatprohibitsunmotivateddeparturesfromunderlyingformingeneral,  preventinge.g./aki/fromsurfacingas*agibyaspontaneousepenthesis[voice]ontothe $ t obstruent.Closertoourconcerns,itistherankingofFaithbelowSonVoithatallows/in+te/    dyinginJapanesetosurfaceasinde"iftherankingwerereversed,wewouldhaveagrammar  1  withoutNCvoicing,andtheoutcomewouldbeinte.ThelowrankingofFaithinaddition A  ensuresthatamonomorphemicJapaneseformliketombodragonflysurfaceswithNCvoicing   irrespectiveofthefeature(under)specificationintheinput,animportantresultdiscussedat Q  lengthinsection4.TurningthentoNoLink,wesawimmediatelyabovein( RANKDIFF56  )thatthe a  rerankingoftheinteractionconstraintNoNCLinkwithrespecttoSonVoihasthesameeffect  fortheoccurrenceofNCvoicing.Whatthenaretherelativefunctionsoftheseconstraints?  DifferentfromFaithfulnessconstraints,whosedomainistheinputoutputrelation,the x NoLinkconstraintsdealwithoutputwellformednessquestionsinfeaturestructure"hencetheir % indifferencetotheprovenanceoffeaturelinkage(underlyingorinserted).Thisdifferenceisless 2 obviousthanothertraitsofNoLinkwehavefocussedupon,e.g.itsroleinbarringpostvocalic   voicingornonlocalpostnasalvoicing,effectsthatFaithalonedoesnotspeaktoinanycase. ?"   ThemainresultofouranalysisfortherelativerankingofFaithfulnesswithrespect L$ toNoVCLinkisthefollowing:NoVCLink>>SonVoi( LOCAL755  a/d);SonVoi>>Faith & ( LOCAL553  c/a).HencebytransitivityNoVCLink>>Faith.Thisderivedrankingisofsignificance  V( forcaseswhereNoLinkandFaithfulnessconflict.Continuingthelexiconoptimization c!* investigationofsection4,letusconsidertwocandidateinputsfortheformudearm,onewith ", sparse[voice]specifications( UDE58  a),andonewithfullylinked[voice]( UDE58  b).&  (29  58  3    )a. ` XsXEE - AA4 EA> Ab. 5A(s!A 5udè     ;" ЀV 7AE( [#A 7*!* -AA>;"AE -**1'%;"   AA> 1XqXA7X=X' ~   & & H  (2;  60  3  )  *y dddd dd d"td"y,Bdd ,Htd ,td +  ,"  , l Input: %AAA  Z %XsX>XXq/ude/ !q Ѐ   9 ЀVXqXA7XX 0A!AAA 0Ѐarm1'%!q   A 1*hZ ddBdd Htd td h,dd ,dd ,d",td"+  ," y  Jx, % A %Candidate *E A *NoVC  A Link 0A!  E 0Faith FE9! A x J A Fa.+=XsX>XXq 5E(U  E 5udè      m  ЀV 7EE( < E 7 -EA>U  EE -** AA4 m  EA> Ab. 5A(< A 5udè  T  ЀV 7AE(# A 7*!* -AA>T AE -1'%T    AA> 1~XqXA7X=X'HReinterpretingthisconvergenceofparsedoutputstructuresasdetermininglexicallyoptimalinputs,thetableaudestableauxschemeofsection4,appliedto( UDETABL159  )and( UDETABL260  ),settlesmattersinfavorof( UDE58  a)astherealinput,asshownin( UDETABL361  ).& 5 (2<  61  3  )  *yH dddd dd d"td"Zy,dd ,dd ,xd",dd",d",td"+  ," H Jx,XsX?XXq %p %Input *Ep * +!pE +Output #E p #NoVC p Link 0A!8E 0Faith#XqXA7XX# FE9!p x J A F=XۇX?XXqXۇX@X=Xa.+ 5E( E 5udè ̀V 7EE(8"E 7+ .E! EE .ude V 0EE!8"E 0 0EA! EE 0#=XۇX@XXÑ# AA4  EA AXۇX@X=Xb. 5A($A 5udè  P% ЀV 7AE(o&A 7+ .A!$AE .ude     P% V 0AE!o&A 0 0AA!P%AE 0*!*#=XۇX@XXs#1'%P%   AA 1 bXqX@X=X'5 V 7(  & '  6Conclusion:SummaryandOutlook   Recapitulatingourmainproposalsinthispaper,wefirstturntothethreelevelhierarchyofconstraintsrepeatedin( concl1062  ).(2=  62  3    )  Overallrankingofconstraints:    License NoVCLink h   NoGap p      ___________ _________________     SonVoi I     ___________   `   `    Faith `  NoNCLink'' ĕ I  IntermsoftheconflictbasedrankinglogicofOptimalityTheory,theconstraintsystemin( concl1062  )hasastraightforwardsetup:Betweenatoplayerconsistingoftheundominated(andinouranalysisunviolated)constraintsLicense,NoVCLink,andNoGapandabottomlayer z consistingofthedominated(andfrequentlyviolated)constraintsFaithandNoNCLink,we ' findthesegmentmarkednessconstraintSonVoi.Thissimplerankingscenarioisresponsiblefor 4 theintricatespecification/underspecificationbehaviorofvoicinginJapanese:Besidessolvingthevoicingparadox,itaccountsfortheappearanceofNCvoicing,forthelocalityoftheinteraction,andfortheabsenceofotherpostsonorantvoicinginJapanese.Andattheleveloftypology,thefreedomofindividualgrammarstorerankSonVoiwithrespecttothesubstantivelyfixed   hierarchyofNoLinkconstraintsimposesplausiblelimitsonthecrosslinguisticvariationspace U" forvoicinginteractions.  Comparedtoearlierwork,theanalysispresentedhereconstitutesasignificantsteptowardsaprinciplebasedtheory,eliminatingrulestipulations.Thesixconstraintsin( concl1062  )areallindependentlymotivatedandarguablyuniversal"ifnotintheexactformulationproposedhere,thenatleastinspirit.Assubstantiveelementsofphonologicaltheory,theyareparticularmembersoftheconstraintfamiliesshownin( concl1163  )governinginput/outputrelations,enforcingdomainconnectedness(togetherwithothercontiguityeffects),regulatingfeaturelicensing,and controllingthemarkednessofsegmentsandsegmentallinks. (=$2 &   (2>  63  3  )  ԀSubstantivetypologyofconstraints*GA dddd dd xd"dd"d"td"H,dd ,dd ,dd ,dd ,dd +  *E  ] *ConstraintFamilies 9/ q"E 9Input/OutputDisparity( Faith) :0!   :intrinsicallyranked:@p9 ,">" ,Licensing :0!> :DomainConnectedness 1'!K mm 1 !   !SegmentMarkedness %   %LinkageMarkedness g        !   !ParseFeat A  FillFeat   ParseLink +  FillLink %(x  %SonVoi %A  %NoVCLink A  NoNCLink %  %License(Voi) %A  %NoGap A   '}  Takenasawhole,theanalysisprovidesstrongsupportforthebasictenetsofOptimalityTheory,asdevelopedinPS1993:Grammaticalconstraintsarerankedandviolable,andthereisnoserialderivation.OnecouldenvisionanaccountofJapaneseinaserialphonologyinvokingconstraintviolationandrepair.Forexample,wemightsaythattheredundancyrule[nasal][voice],applyingearly,targetsallnasals,violatingaconstraintagainstredundantfeaturespecification.Theseviolationscouldthenrepaired,wherepossible:spreadingtoaneighboringobstruentgrantslicensing;otherwise,[voice]mustdelinkagain.AsPS1993pointout,suchrepairandrescuescenarioshavethedo,exceptwhen"unlessqualitysignallingahiddenappealtorankingandviolability,theveryfactorsthatareelevatedtothelevelofprincipleinOptimalityTheory.Inotherwords,suchanalyseswouldmerelyreplicatetheoptimalitytheoreticaccountinastipulativeandinexplicitway.  Asfarasconventionalunderspecificationtheoryisconcerned,itisreplacedinourproposalbyatheoryoffeaturelicensingwhichbuildsonearlierworkonlicensing(e.g.It=1986,Goldsmith1990,Lombardi1991,It=&Mester1993)andmarkingconditions(Kiparsky1985,Padgett1991,56!58)(seealsoSteriade1994foranindependentlydevelopedconceptofindirectlicensing).Redundantfeaturespecificationisgovernedbyafamilyoflicensingconstraints"License(,)"andnotbythemorefamiliarmarking/featurecooccurrenceconditions. c*%8 Ourworkthusmilitatesagainsttheuseoffeaturecooccurrencerestrictionslike*[+son,+voice] +': toensureunderspecification,sincetheyimplyasimpleincompatibilitybetweentherelevantfeatures.Aswehaveseen,thereisnosuchincompatibility;rather,thereisonlyafailureoflicensing.Onlythisnotionoflicensingcanilluminatethefreeoccurrenceofredundantfeaturesindoublylinkedstructures.Weareofcoursenotarguingagainstfeaturecooccurrencerestrictionsingeneral,whichareneededtoruleoutcooccurrencesofantagonisticfeatures,such $ t as*[+son,voice](or,inatheorywithprivative[voice],*[+son,+spreadglottis],seeMester&It=1989,Lombardi1991).AntagonisticfeaturecooccurrencerestrictionsareindependentofLicensingTheory(forexample,thereisnosenseinwhichdoublelinkingrendersvoicelessnessinsonorantswellformed)andplayanimportantroleinconstrainingthederivationofsegmentsbyassimilation,seeKiparsky(1985),Pulleyblank(1989),Cohn(1989),Archangeli&Pulleyblank(inpress),Padgett(1991:48!63,toappear).  Inanothervein,wenotethepossibleimplicationsofourresultsforabroaderconceptionofunderspecification.Theprogramdevelopedhereforredundantfeaturesmightbeprofitablyextendedtoincludeaninvestigationofthespecificationalbehaviorofunmarked(butnot l redundant)valuesofcontrastivefeatures,therebyaddressingtheperennialtopicofdegreeofunderspecification"full,contrastiveorradical.Thus,weanticipatethattheopposingcontrastiveandradicalviewpointsmayrevealthemselvesasreificationsoftwodifferentorderingsofantagonisticconstraints.Asanexample,considertheplacefeature[coronal],perhapsthebestinvestigatedcaseinpoint(seeParadis&Prunet1991andreferencestherein).Wemightattribute[coronal]behaviortoaconflictbetweensomekindoflicensingconstraintontheonehand(cf.*PL/CorinPS1993,181,andKiparsky1994),andconstraintsagainstemptyPlacenodesand J( unparsedfeaturesontheotherhand(FillPlace,ParseFeature,ibid.).Ifsuchanapproach W!* provesfruitful,thecontrastbetweenradicalandcontrastiveviewsofunderspecificationreducestoacontrastinranking,withempiricalconsequencestobepursued.Inasimilarvein,redundantfeaturespecificationisdeterminedinourviewbyaconflictbetweenlicensingconstraintsmilitatingagainstthepresenceofredundantfeaturesandconstraintsrequiringthepresenceofthoseverysamefeatures.Therankingofthese(familiesof)constraintswithrespecttoeachothertipsthescaleseitherinfavorofspecification,orofunderspecification.  Finally,letusreturnonelasttimetotheissueofredundantfeatureunderspecificationinoutputs.Futureinvestigationswillhavetodevoteseriousthoughttoproposals(asinSteriade ?,': 1994)thatallowthephonologydirectaccesstoavastrangeofphoneticproperties,andinthiswayareabletomakeuseofamultitudeofdistinctionsinphonologicaldescriptionsthatarenotavailablewithinthelimitedfeaturesetofJakobsonianphonology.Thereisnodoubtthatcarefulattentionmustbepaidtothephoneticinterpretationofphonologicalrepresentations.However,itseemstousthatadirectimportationofacoustic,physiological,andaerodynamicfactorsintophonology,whilemakingcertaingeneralizationsmoreeasilystatable(likeanyexpansionofthedescriptivevocabulary),mightleadtoalossofexplanation.  &  12      ׀Proposalsrevivingtheideaof  .   phonologicalprojections(Halle&Vergnaud1978)intheformofrulespecificvisibilityparametrizationhavelittletroubleinmakingfullyspecifiedphonologicalrepresentationscompatiblewiththefactsbydeclaring,forcertainprocesses,thatonlymarked,oronlycontrastive,featuresmaybeaccessed(Calabrese1994)"adescriptivesuccessboughtatthepriceofrulebyrulestipulations,therebyimplicitlyorexplicitlygivinguponthesearchformorefundamentalphonologicalprinciples.  &  13        \   WhiletheparticularfactsdiscussedthroughoutthispaperarespecifictoJapanese,ourapproachtoredundantfeature(under)specificationisbasedonageneraltheoryoffeaturemarkednessandlicensingappliestoabroadclassofothercases.FactslikethoseofNCvoicinghavenotgenerallyinformedthediscussionofunderspecificationintheliterature.Thisispartlyduetothedifficultyinidentifyingsuchfacts,andnottotheirrarity(forfeaturesdifferentfromvoicing,wereferthereadertothebriefanalysisofTurkishroundingeffectspresentedinIt=,Mester,&Padgett(1993,18!28)andthemorefullydevelopedtreatmentinNChiosin&Padgett1993).WehavechosenYamatoJapaneseasatestinggroundforourlicensingbasedtheoryoffeaturespecification,withitsentailmentofunderspecificationasanemergentoutputproperty,becauseitisparticularlysuitedfortheempiricalinvestigationofphonologicalvoicing"themorphophonemicprocessofRendakuallowsaccesstoinformationaboutphonologicalspecificationandunderspecificationwhichisnotusuallyavailableinotherlanguages.  ObservationsaboutJapanesevoicingspecifications,likeallobservations,areultimatelytheorybound"heretheyaremoldedbytheautosegmental/OCPbasedviewofRendaku, LymansLaw,andVoicingSpreadinIt=&Mester(1986),itselfpredicatedonthefundamental *,&8 assumptionthatphonologicalrepresentationsareredundancyfree.Withinoptimalitytheoreticapproacheschampioningfullspecification(seee.g.Smolensky1993forastepinthisdirection),theprojectofcapturingthecentralphonologicalpropertiesofJapanesevoicingstrictlyinmarkednesstermshasremainedsofarelusive,butisclearlyworthpursuing"weawaitcompetitivealternativeswhichareabletomatchtheunderspecificationbasedaccountintermsofeleganceanddepthofexplanation.& w References  ;  0   Archangeli,Diana.1984.UnderspecificationinYawelmaniPhonologyandMorphology.   Doctoraldissertation,MIT,Cambridge,Mass. (#(# 'w;0   Archangeli,Diana,&DouglasPulleyblank.1986.TheContentandStructureofPhonological U  Representations.Ms.,UniversityofArizonaandUniversityofSouthernCalifornia.(#(# 0   Archangeli,Diana,&DouglasPulleyblank.1993.Optimality,GroundingTheory,andRuleParameters.HandoutfromtalkpresentedatRutgersOptimalityWorkshopI,Rutgers r University,NewBrunswick,N.J. (#(# 0   Archangeli,Diana,&DouglasPulleyblank.inpress.GroundedPhonology.Cambridge,Mass: / MITPress. (#(# 0   Avery,Peter,&KerenRice.1989.SegmentStructureandCoronalUnderspecification.Phonology6.2:179!200.I$(#(# 0   Borowsky,Toni.1986.TopicsinEnglishPhonology.Doctoraldissertation.Universityof & Massachusetts,Amherst. (#(# 0   Borowsky,Toni.1987.AntigeminationinEnglishPhonology.LinguisticInquiry18:671!678.f!*(#(# 0   Calabrese,Andrea.1994.MarkingStatements,CompexityandSimplifcationProcedures.Ms.,HarvardUniversity. (#(# 0   Cho,YoungMeeY.1990.ATypologyofVoicingAssimilation.ProceedingsofWCCFL9, % 0 141!155. (#(# 0   Chomsky,Noam,&MorrisHalle.1968.TheSoundPatternofEnglish.NewYork:Harper& =(#4  Row.)$6(#(# 0   Chomsky,Noam.1993.AMinimalistProgramforSyntax.InTheViewfromBuilding20.   EssaysinLinguisticsinHonorofSylvainBromberger,ed.KennethHaleandSamuelJay ` Keyser,1!52.Cambridge,Mass.,andLondon,England:MITPress. (#(# Clements,G.N.1985.Thegeometryofphonologicalfeatures.Phonology2:225--252.  0   Clements,G.N.1988.Towardasubstantivetheoryoffeaturespecification.InProceedingsof - } NELS18,79!93.GLSA,UniversityofMassachusetts,Amherst.  (#(# 0   Clements,G.N.1991.PlaceofArticulationinConsonantsandVowels:AUnifiedTheory.In  =  L'ArchitectureetlaGomtriedesreprsentationsphonologiques.ed.B.Laks&A. J  Rialland.Paris:EditionsduC.N.R.S. (#(# 0   Cohn,Abigail.1989.PhoneticEvidenceforConfigurationConstraints.InProceedingsofNELS W  19,63!77.GLSA,UniversityofMassachusetts,Amherst.g (#(# 0   Davis,Stuart.1991.CoronalsandthePhonotacticsofNonadjacentConsonantsinEnglish.InParadis&Prunet1991,49!60. (#(# 0   Garrett,Andrew.1991.IndoEuropeanReconstructionandHistoricalMethodologies.Reviewarticle.Language67:790!804..(#(# 0   Goldsmith,John.1990.AutosegmentalandMetricalPhonology.Oxford:Blackwell.>(#(# 0   Goldsmith,John.1991.PhonologyasanIntelligentSystem.InBridgesbetweenPsychologyand   Linguistics:ASwarthmoreFestschriftforLilaGleitman,ed.D.J.Napoli&J.A.Kegl, N" 247!267.HillsdaleNJ:LawrenceErlbaumAssociates. (#(# 0   Goldsmith,John.1992.HarmonicPhonology.InTheLastPhonologicalRule:Reflectionson  & ConstraintsandDerivationsinPhonology,ed.JohnGoldsmith,21!60.Chicago:  k( UniversityofChicagoPress. (#(# 0   Grimshaw,Jane.toappear.MinimalProjection,Heads,&Optimality.ToappearinLinguistic "(, Inquiry.[EarlierversiondistributedasTechnicalReport#3,RutgersUniversityCenter 8$. forCognitiveScience,September1993.] (#(# 0   Harries,Lyndon.1950.AGrammarofMwera.Johannesburg,SouthAfrica:Universityof &E"2 WitwatersrandPress. (#(#  Hayes,Bruce1986.InalterabilityinCVphonology.Language62:321!251. )%6 0   Hayward,K.M.1989.TheIndoEuropeanLanguageandtheHistoryofitsSpeakers:The TheoriesofGamkrelidzeandIvanov.Lingua78:37!86. ](#(# 0   Hutcheson,James.1973.RemarksontheNatureofCompleteConsonantalAssimilations.InPapersfromtheNinthRegionalMeetingoftheChicagoLinguisticsSociety,215!222.  ChicagoLinguisticsSociety,UniversityofChicago,Chicago,Ill. (#(# 0   It=,Junko.1986.SyllableTheoryinProsodicPhonology.Doctoraldissertation,Universityof    Massachusetts,Amherst.[PublishedbyGarland,NewYork,1988.] (#(# 0   It=,Junko.1989.AProsodicTheoryofEpenthesis.NaturalLanguageandLinguisticTheory D  7:217!259. (#(# 0   It=,Junko,&ArminMester.1986.ThePhonologyofVoicinginJapanese.LinguisticInquiry Q  17:49!73. (#(# 0   It=,Junko,&ArminMester.1993.LicensedSegmentsandSafePaths.CanadianJournalof  Linguistics38.2,197-213.n(#(# 0   It=,Junko,ArminMester&JayePadgett.1993.LicensingandRedundancy:Underspecification ~ TheoryinOptimalityTheory.Reportno.LRC!93!07,LinguisticsResearchCenter,UC . SantaCruz. (#(# 0   It=,Junko,&ArminMester.1994a.JapanesePhonology.ToappearinHandbookof   PhonologicalTheory,ed.JohnGoldsmith.Oxford:Blackwell.K"(#(# 0   It=,Junko,&ArminMester.1994b.FaithfulnessandInventories:AnOptimalityTheoreticApproachtoSublexicalDomains.Ms.,UniversityofCalifornia,SantaCruz.[ToappearinUniversityofMassachusettsOccasionalPapers.GLSA,Amherst.] e((#(# 0   Kenstowicz,Michael,andCharlesKisseberth.1977.TopicsinPhonologicalTheory.NewYork: u!* AcademicPress. (#(# 0   Kiparsky,Paul.1973. ElsewhereinPhonology.InAFestschriftforMorrisHalle,ed. 2$. StephenR.AndersonandPaulKiparsky,93!106.NewYork:HoltRinehartandWinston. (#(# Kiparsky,Paul.1981.Vowelharmony.Ms,MIT,Cambridge,Mass.0   Kiparsky,Paul.1982.LexicalPhonologyandMorphology.InLinguisticsintheMorningCalm )$6 1,ed.I.S.Yang,3!91.Seoul,Korea:Hanshin# +Y&8(#(# #0   Kiparsky,Paul.1985.SomeConsequencesofLexicalPhonology.PhonologyYearbook2:  82!138. (#(# 0   Kiparsky,Paul.1988.Phonologicalchange.InLinguistics:TheCambridgeSurveyvol.1,ed. m FritzNewmeyer,363!415.Cambridge:CambridgeUniversityPress. (#(# 0   Kiparsky,Paul.1994.RemarksonMarkedness.HandoutfromtalkpresentedatTrilateral * z PhonologyWeekend,UniversityofCalifornia,SantaCruz.  (#(# Leben,Will.1973.SuprasegmentalPhonology.Doctoraldissertation,MIT,Cambridge,Mass.  :  0   Levergood,Barbara.1984.RulegovernedVowelHarmonyandtheStrictCycle.InProceedings J  ofNELS14,275!293.GLSA,UniversityofMassachusetts,Amherst. (#(# 0   Lombardi,Linda.1991.LaryngealFeaturesandLaryngealNeutralization.Doctoral  Z  dissertation,UniversityofMassachusetts,Amherst. (#(# 0   Martin,Samuel.E.1952.MorphophonemicsofStandardColloquialJapanese.Supplementto  Language.LanguageDissertationNo.47.'w(#(# 0   McCarthy,JohnJ.1986.OCPEffects:GeminationandAntigemination.LinguisticInquiry  17:207!263. (#(# 0   McCarthy,JohnJ.1993.TheParallelAdvantage:Containment,Consistency,andAlignment.HandoutfromtalkpresentedatRutgersOptimalityWorkshopI,RutgersUniversity,New   Brunswick,N.J. (#(# 0   McCarthy,JohnJ.,&AlanS.Prince.1993.ProsodicMorphologyI:ConstraintInteractionand ^$ Satisfaction.Ms.,UniversityofMassachusetts,Amherst,andRutgersUniversity.[To & appear,MITPress,Cambridge,Mass.] (#(# 0   McCarthy,JohnJ.,&AlanS.Prince.1994a.TheEmergenceoftheUnmarked.Toappearin {!* ProceedingsofNELS24.GLSA,UniversityofMassachusetts,Amherst."(,(#(# 0   McCarthy,JohnJ.,&AlanS.Prince.1994b.AnOverviewofProsodicMorphology.HandoutfrompaperpresentedattheUtrechtWorkshoponProsodicMorphology,Utrecht % 0 University,TheNetherlands. (#(# 0   McCarthy,JohnJ.,&AlisonTaub.1992.ReviewofParadis&Prunet1991.Phonology9.2: R(#4  363!370.)%6(#(# 0   McCawley,JamesD.1968.ThePhonologicalComponentofaGrammarofJapanese.The   Hague:Mouton. (#(# 0   Mester,Armin.1986.StudiesinTierStructure.Doctoraldissertation,Universityof m Massachusetts,Amherst.[PublishedbyGarland,NewYork,1988.] (#(# 0   Mester,Armin.1994.TheQuantitativeTrocheeinLatin.NaturalLanguageandLinguistic * z Theory12:1!61.  (#(# 0   Mester,Armin,&JunkoIt=.1989.FeaturePredictabilityandUnderspecifcation.Language65:  :  258!293. (#(# 0   Mohanan,K.P.1991.OntheBasesofRadicalUnderspecification.NaturalLanguageand   LinguisticTheory9:285!325.W (#(# Myers,Scott.1994.OCPEffectsinOptimalityTheory.Ms.,UniversityofTexas,Austin.0   NChiosin,Mire,&JayePadgett.1993.OntheNatureofConsonantVowelInteraction.PaperreadattheFirstHollandInstituteofLinguisticsPhonologyConference,Leiden. !q [Alsoreportno.LRC!93!09,LinguisticsResearchCenter,UCSantaCruz.] (#(# 0   Padgett,Jaye.1991.StrictureinFeatureGeometry.Doctoraldissertation,Universityof . Massachusetts,Amherst.[Toappear,CSLI/UniversityofChicagoPress.] (#(# 0   Padgett,Jaye.toappear.StrictureandNasalPlaceAssimilation.NaturalLanguageand   LinguisticTheory.K"(#(# 0   Padgett,Jaye.1994.OntheBasesofInteraction.HandoutfromtalkpresentedatTrilateral [$ PhonologyWeekend,StanfordUniversity. &(#(# Paradis,Carole.1988a.OnConstraintsandRepairStrategies.TheLinguisticReview6:71!97.  k( 0   Paradis,Carole.1988b.TowardsaTheoryofConstraintViolations.McGillWorkingPapersin {!* Linguistics5,1!44.DepartmentofLinguistics,McGillUniversity,Montreal."+,(#(# 0   Paradis,Carole,&JeanFran oisPrunet,eds.1991.PhoneticsandPhonology2:Thespecial ;$. StatusofCoronal.InternalandExternalEvidence.SanDiego:AcademicPress.% 0(#(# 0   Pater,Joe.1994.AProsodicAnalysisofIndonesian/MalayNasalSubstitution.Ms.,McGillUniversity,Montreal.[PaperpresentedattheUtrechtWorkshoponProsodic X(#4  Morphology,UtrechtUniversity,TheNetherlands.]#)%6(#(# #0   Pierrehumbert,Janet.1993.DissimilarityintheArabicverbalroots.InProceedingsofNELS   23,367!381.GSLA,UniversityofMassachusetts,Amherst.`(#(# 0   Prince,AlanS.,&PaulSmolensky.1993.OptimalityTheory.ConstraintInteractionin p GenerativeGrammar.Ms.,RutgersUniversity,NewBrunswick,andUniversityof   Colorado,Boulder.[Forthcoming,MITPress,Cambridge,Mass.] (#(# 0   Pulleyblank,Douglas.1986.ToneinLexicalPhonology.Dordrecht:D.ReidelPublishing    Company. (#(# 0   Pulleyblank,Douglas.1989.PatternsofFeatureCooccurrence:TheCaseofNasality.InCoyoteWorkingPapers9,98!115. (#(# 0   Rice,Keren.1993.Areexaminationofthefeature[sonorant]:Thestatusof sonorantobstruents.Language69:308!344.d (#(# 0   Sagey,Elizabeth.1986.TheRepresentationofFeaturesandRelationsinNonLinearPhonology.  Doctoraldissertation,MIT,Cambridge,Mass. (#(# 0   Scobbie,JamesM.1991.AttributeValuePhonology.Doctoraldissertation,Universityof  Edinburgh,Edinburgh,UK. (#(# Schein,Barry&DoncaSteriade.1986.OnGeminates.LinguisticInquiry17.4,691!744. > 0   Selkirk,ElisabethO.1988.Dependency,PlaceandtheNotion"Tier".Ms,UniversityofMassachusetts,Amherst. (#(# 0   Selkirk,ElisabethO.1993.LabialRelations.ms,UniversityofMassachusetts,Amherst. (#(# 0   Smolensky,Paul.1993.Optimality,Markedness,andUnderspecification.HandoutfromtalkpresentedatRutgersOptimalityWorkshopI,RutgersUniversity,NewBrunswick,N.J. b((#(# 0   Steriade,Donca.1987.RedundantValues.InPapersfromtheTwentythirdRegionalMeeting r!* oftheChicagoLinguisticSociety:ParasessiononAutosegmentalandMetricalPhonology, "", 339!362.ChicagoLinguisticsSociety,UniversityofChicago,Chicago,Ill. (#(# 0   Steriade,Donca.1994.UnderspecificationandMarkedness.ToappearinHandbookof % 0 PhonologicalTheory,ed.JohnGoldsmith.Oxford:Blackwell.&?"2(#(# 0   Stevens,Kenneth,SamuelJ.Keyser&H.Kawasaki.1986.TowardsaPhoneticandPhonologicalTheoryofRedundantFeatures.InSymposiumonInvarianceofSpeech )$6 Processes,ed.J.Perkell&D.H.Klatt.Hillside,NJ.:LawrenceErlbaumAssoc.# +\&8(#(# #0   Taub,Alison.1988.TheUnderspecificationof[voice]andtheStatusofNasalConsonantClustersinJapanesePhonology.Ms.,UniversityofMassachusetts,Amherst. (#(# 0   Trubetzkoy,NikolaiS.1939.GrundzgederPhonologie.TraveauxduCerclelinquistiquede j Prague.7.[Reprinted1958,Vandenhoeck&Ruprecht,Gttingen,Germany.] (#(# 0   UribeEtxebarria,Myriam.1992.ASyntaxticapproachtoMorphophonologicalPropertiesofCompounds:OnJapaneseRendakuand(De)geminationPhenomenainDravidianLanguages.Ms.,UniversityofConnecticut. (#(# 0   Wonderly,W.L.1951/52.ZoqueI,II,III,IV,V,VI.InternationalJournalofAmerican >  %  Linguistics17,1951,1!9,105!123,137!162,235!251.InternationalJournalof   AmericanLinguistics18,1952,35!48,189!202.N (#(# 0   Yip,Moira.1991.Coronals,ConsonantClusters,andtheCodaCondition.InParadis&Prunet1991:61!78. (#(# >A(' -xd> u %  It=,Mester,Padgett    h ito@ling.ucsc.edu : DepartmentofLinguistics   h mester@ling.ucsc.edu G  AdlaiE.StevensonCollege   h padgett@ling.ucsc.edu " UniversityofCaliforniaSantaCruz,CA95064  ^& Notes