WPC w9&TQ}dy!㼣6ҁ^37)wwH$66QܺEpd8Oݟh:ؙ% NZ=;t DԤ65Owu+ pq4sP{3gprUZQ#^뒳B}NF~ִ$jdZ8@خ #BJO3Zu?2DEoj7j5o2b> 0C> B!?UON>?b?b?U0?f?a? B*? B! @f)@f+@a-@ 0A@f@a@ B @FACKDImKUNQ7aQU0SfSaSfSaSSfUaUfUUNUUNAVaVViWbYns[U+N!^fo^aq^f^UN^a^f^a^L^Kd 1eene 0rnUgNnIoUNp]pa#qUN7q 0qd;rwfexagxf{x4}x AYxx 0Fyay D+ z5z 0szfn{ap{{f:a<PfBaD,X*#ÆU*I 0LsfffÇfŇfLJfɇfˇf͇ hχaՇa hY h\nbaЎaaaϑaFaJ hN hT hZ h` hf hl hr hx h~ h h h h h,ž h hR hxF h hğ h,ʟ hV& 8Document[8]Document Style0..8` ..` V8Document[4]Document Style.. . V 8Document[6]Document Style8..V 8Document[5]Document Style0..V/8Document[2]Document Style 2A.3  Ԁ   V& 8Document[7]Document Style0..0` ..` zU :Right Par[1]Right-Aligned Paragraph Numbers..2I.3  Ԁ..0..zh :Right Par[2]Right-Aligned Paragraph Numbers..` ..2A.3  Ԁ..0` ..` V?8Document[3]Document Style.. 21.3  Ԁ   z{ :Right Par[3]Right-Aligned Paragraph Numbers..` ..`  ..P 21.3  Ԁ` ..` 0 .. z :Right Par[4]Right-Aligned Paragraph Numbers..` ..`  .. .. 2a.3  Ԁ .. 0..z :Right Par[5]Right-Aligned Paragraph Numbers..` ..`  .. ..h..2(1)3  Ԁ..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:0: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. <6X9`("Courier NewTTX\  `(Times NewRomanTT'   (Takako Kawasaki(0Takako Kawasaki .   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 ' z ((  qXXXXX6P)'*d66P)'*d6HH 1HH 2HH 3HH 4HH 5HH 6HH 7HH 8HH 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28  .+(e2 $ {!XXXXXm)XXX   p4Heading 2Underlined Heading Flush LeftV4Heading 1Centered Heading@..  b* 8Bullet ListIndented Bullet List0..0` ..` B& $1Document Style0..8` ..` B $2Document Style.. . B $3Document Style8..B $4Document Style0..B-$5Document Style 23  Ԁ   B& $6Document Style0..0` ..` dS $7Right-Aligned Paragraph Numbers..23  Ԁ..0..df $8Right-Aligned Paragraph Numbers..` ..` 23  Ԁ..0` ..` B=$9Document Style.. 23  Ԁ   fy &10Right-Aligned Paragraph Numbers..` ..`  .. 23  Ԁ` ..` 0 .. f &11Right-Aligned Paragraph Numbers..` ..`  .. ..23  Ԁ .. 0..f &12Right-Aligned Paragraph Numbers..` ..`  .. ..h..h23  Ԁ..0h..hf &13Right-Aligned Paragraph Numbers..` ..`  .. ..h..h..23  Ԁh..h0..f &14Right-Aligned Paragraph Numbers..` ..`  .. ..h..h....23  Ԁ..0..f &15Right-Aligned Paragraph Numbers..` ..`  .. ..h..h....p..p23  Ԁ..0p..pDV!&16Document Style  @..  23  Ԁ     Ԉ X/&17Technical Document Style.. 23  Ԁ. X/&18Technical Document Style.. 23  Ԁ. X-(&19Technical Document Style 23  Ԁ   .. X*$&20Technical Document Style 23  Ԁ   .. X)$&21Technical Document Style 23  Ԁ   .. X83&22Technical Document Style  23  Ԁ     .. X/&23Technical Document Style.. 23  Ԁ. X/&24Technical Document Style.. 23  Ԁ. 5G `2Goudy Old Style ATTTT- -($     d(9 Z 6Times New Roman Regular- -- -:\ Y `SymbolTTTable_A&0 d d dTable_A(f3Ǜ$ {!XXXXXm)XXX   Table_Ad < $XXXXX9XXX      7    ݀UKUS.,%$%XX9AnimportantstepforwardismadebyKager(1989),whoassumesstrictlybisyllabicfootconstruction,  sothatthelightsyllablesaresimplyleftunparsed,ratherthanstressedandthendestressed.US.,UK.,I%$%%%$ԀKager'spremisethat  (H)feetareimpossibleprefiguresBurzio's(1994)theory,thoughtheirapproachestodealingwithapparentcasesof(H)feetarequitedifferent.WhereasBurziopositsemptysyllablestructure,KagerinvokesaWeighttoStressprinciplethatgrantsstresstounfootedheavysyllables.#%$%%I%$# 9 $XXXXX9XXX      2    %$%XX9ԀBurzio(1994)avoidsthecomplicationsofthesepatternsofnonuniformitybydenyingthatthe  syllablesin(2a)and(4a)areinfactstressed,inconformancewithhisbasictheoreticalpremisethatafootmadeupofasingleheavysyllableisuniversallyillformed(apremisewhichshouldrightlybeattributedtoKager1989:129).Thispremiseisatoddswithmostotherworkinmetricaltheory,andrequiresanumberofequallyidiosyncraticassumptionstomaintain(e.g.thatwordslikebandanahaveanullinitialvowel,thatvowel ` reductionisfarmorecontextsensitivethanusuallyassumed).ThisisnotnecesarilyacriticismofBurzio'sextremelythoroughaccountofEnglishstress;itisinternallyconsistent,andcontainsanumberofimportantdescriptiveandtheoreticaladvancesthathavebeenmadeuseofhere.Theseidiosyncraciesdo,however,seriouslyimpedeanyattempttosystematicallycompareBurzio'sanalysiswithotheranalysesofEnglish,aswellassituateitinthewidercrosslinguistictypologyofstresssystems(seee.g.Hayes1981,HalleandVergnaud1987,Idsardi1992,Hayes1995).  $XXXXX9XXX      4    ݀IX9XXX9ԀI%$%XIX9Headshipandstressarenotnecessarilycoextensive,eveninatheorythatrecognizestheroleof  constituencyindeterminingprominenceonthegrid(Kager1989,P&S1993).However,stressednonheadswillnotfigureinthepresentanalysis(thoughseeAppendixA).#%$%%I%$# c $XXXXX9XXX      9    ݀%$%XX9Thefactsofprimarystressretractiondointroducesomecomplications(seeespeciallyKager1989for  thoroughdiscussionandreferences).Whenthereisfinalstress,themainstressusuallyoccursonthenextfoottotheleft.ThisispresumablyaneffectofNonfinality.Withnoelaboration,thepresentaccountwouldpredict t thatmainstressshouldlandonthenextlegitimatefoot,eitheraheavysyllable(e.g.stalct7te),orpairoflights(e.g.actylne).However,mainstressoftenendsupfurthertotheleft,skippingtheheavysyllable(e.g.dsignte)or LL sequence(e.g.ctamarn).Onewaytocapturethesecasesof'strong'and'long'retraction L wouldbetodemoteAlignHeadintheranking,andtouseaconstraintdemandingthealignmentoftheright  ( edgeoftheProsodicWordwithafoothead,oramoraictrochee(cf.McCarthy1995),toplacetherightmoststress.Asbothstressretraction,andtheweight-to-stressbehavioroffinalsyllablesarequitecomplex,andrifewithexceptionality,IabstractfromtheseaspectsoftheEnglishstresssysteminthepresentanalysis.(9 Z 6Times New Roman Regular - $XXXXX9XXX      *    ݀ThankstoEricBakovia,EdwardFlemming,HeatherGoad,RenKager,GregLamontagne,  JohnMcCarthy,andAlanPrinceforhelpfuldiscussion.IwouldliketoparticularlyacknowledgeGlynePiggottforencouragingmetotakeonEnglishstressonthefirstplace,andforhisperceptivecommentsonthevariousanalysesthatfollowed.ThisresearchwassupportedbySSHRCCfellowship752932773.:\ q `SymbolTTTable_BTable_B  $XXXXX9XXX      5    %$%XX9ԀAswillbediscussedin3,thesewordsdohavevariantpronunciationsinwhichthepretonicvowels  arereduced.Inthisrespecttoo,longvowelsparallel%s's(e.g.theunderlinedsyllableinrpresenttioncanbe  eitherstressedornot).Table_B(9 Z 6Times New Roman Regular(9 Z 6Times New Roman RegularTable_BLevel 1Level 2Level 3Level 4Level 5  {!XXXXX9XXX     򀀀  3  %$%XX9KenyonandKnott1953givesannexation,adaptation,andaffectationwitheitherafullora  reducedvowelinthesecondsyllable.Seethesection3fordiscussionofvariationinthisdomain. X $XXXXX9XXX      6    @XX9Ԁ%$%@Alsoneededhere,andanywhereelsethatFtBinisinvokedtoruleoutanonfinalmonomoraicfoot,  isahighrankingOnsetconstraint,whichwouldruleoutbn.na,and,ifambisyllabicityispermitted,a  constrainttoruleoutambisyllabicityacrossafootboundary.SeeMcCarthyandPrince(1993b),andPater(1994)forrelevantdiscussion.\  `(Times NewRomanTTlarXXP\  P6QXP!L q`(Times NewRomanTTlarXX]L APeQXP\  `(Times NewRomanTTlard\  PQP\  `(Times NewRomanTTlarC\  PQP(9 Z 6Times New Roman RegularTable_A(9 Z 6Times New Roman RegularTable_ATable_A B $XXXXX9XXX      10    ݀IX9XXX9I%$%XIX9ChomskyandHalle(1968:fn.64)claimthatinformationdoesnotpreservethestressofitsstem  because"informationisnotthenominalizedformofinform,butratherasinglenounpresumablyrepresented  as/inform+At+iVn/.Thuswecannothavephraseslike*hisinformationofmyfriendaboutthelecturerelated t toheinformedmyfriendaboutthelecture,aswehavehisrelaxationoftheconditionsrelatedtoherelaxed L theconditions."Note,however,thattheirconservationoftheforest,andhislamentationofthelosscanbe $t relatedtotheyconservedtheforest,andhelamentedtheloss,eventhoughthepretonicsyllablesoflamentation L andconservationarereduced.Arejectionoftheideathatthereisacorrelationbetweenthemorphosyntactic  $ factsofnominalization,andstresspreservation,isimplicitinHalleandVergnaud(1987:251)#%$%%I%$##X9X%%$#ԂLevel 1Level 2Level 3Level 4Level 5  CRight ParRight-Aligned Paragraph NumbersI.A.1.a.(1)(a)i)a)3|U!<6X9`("Courier NewTTXXx6X@DQX@\  `(Times NewRomanTTXXP\  P6QXP5G `2Goudy Old Style ATTTTXXS5G P}QXP\  `(Times NewRomanTTTT^\  P6QP\  `(Times NewRomanTTTTd\  P6QP\  `(Times NewRomanTTTTC\  P6QP:\ Y `SymbolTTmanTTTTXXt\  P[AXP\  `(Times NewRomanTTTTy\  P6QP<6X9`("Courier NewTTTTTTd6X@DQ@<6X9`("Courier NewTTTTTTZ6X@DQ@<6X9`("Courier NewTTTTTT6X@DQ@)4 k" `,SILManuscriptIPATTTTa4 k" PqQP<6X9`("Courier NewTTTTTTT6X@DQ@(J$XqXXXXmOXXXq  (9 Z 6Times New Roman RegularLevel 1Level 2Level 3Level 4Level 5&O Z 6Times New Roman RegularLevel 1Level 2Level 3Level 4Level 5Table_A(9 Z 6Times New Roman Regular(f(3$ {!XXXXXm)XXX       0   Z $XXXXX9XXX      18    %$%XX9ԀOnewayofgeneratingtheseexceptionalmonomoraicmainstressesistohavealexicallyspecific  versionofNonfinalitythatoutranksFtBin.LikethelexicallyspecificStressWell,thisconstraintwould  applyprimarily,butnotexclusively,towordscontainingboundaffixes!inthiscasethesuffixesite,oid,and ` ode(e.g.Semite,cathode,lithoid!seeLibermanandPrince1977:305).Thiswouldnotinterferewitha 8 FtBinbasedanalysisofwordminimality,sincewhentheinputismonosyllabic,Nonfinalityisrendered ` inactivebythedominanceofLex Pr,thatis,bytherequirementthatlexicalwordsbeprosodified(Princeand 8 Smolensky1993).Aninterestingrelatedobservationisthatwordsofthisshape(i.e. LH )areregularlyend   stressed(Oehrle1971;LibermanandPrince1977:299).ThissuggeststhattheregularpatternisforFtBinto   dominateNonfinality,andthattheseconstraintsareactivelycompeting.numberLevel 1Level 2Level 3Level 4Level 5Table_BCVTM << deUULevel 1Level 2Level 3Level 4Level 5(n$ (  1  ) Table_F'dxd1, 2, 3,Level 1Level 2Level 3Level 4Level 5B1h21, 2, 3,Numbers  .0 Table_F  $XXXXX9XXX      19    ݀%$%XX9Oneconsequenceofthestressbasedversionofthisanalysis,though,isthatitwouldrequirean  extensionofthetheoryofprosodicfaithfulnessinMcCarthy(1995)tohandledirectpreservationofprominence.InMcCarthy(1995),prominenceisonlypreservedindirectly,throughfaithfulnesstotheprosodicroleofsegments.Insofarasthepatternsof'stresspreservation'discussedinthispaperareidiosyncratictoEnglish,whichcanonlybeestablishedthroughfurtherresearch,itmaybeinfactpreferabletotreatthemasvowelqualitypreservationinstead,andleavethetheoryofprosodicfaithfulnessalone.ThepredictionthiswouldmakeisthatEnglishlikestresspreservationshouldonlyoccurinlanguageswithvowelreduction.Table_B  $XXXXX9XXX      8    ݀%$%XX9SUKUS.,eeCohnandMcCarthy(1994)andKenstowicz(1994)foranalysesofIndonesianinwhichaniUS.,UK.,UKUS.,ambic  footisformedtoplacestressrightmostintherootUS.,UK.,:,eventhoughtrochaicfeetarepreferredelsewhere.Table_B " $XXXXX9XXX      20    ݀%$%XX9Afullaccountofwordslikethosein(69)willalsohavetodealwiththecomplicationsinducedby  theinstancesofapparentmonomoraicfeet(  cAustcian)andvowellengthening( dA mstcity)thatseemtooccur  asalternativestoParse%violations(seeKager1989:171forfurtherexamples).     $XXXXX9XXX      1    ݀%$%XX9Itakethestandardviewthatasyllablewithareducedvowelorasyllabicsonorantconsonantis  unstressed,andonewithafullvowelisstressed,modulotheeffectsofwordfinality(seeBurzio1994foranotherstance).Iwillnotdiscussfinerdistinctionsbetweenlevelsofstressthansecondary,primary,andthecompletelackofstressbecauseneithersyllableweight,norstemstress,seemtodeterminewhetherasyllablehastertiaryorsecondarystress(onthenondeterminationofthesecondary/tertiarydistinctionbystemstress,seeHalleandVergnaud1987andKager1989;cf.Kiparsky1979).PostScript PrinterPSCRIPT ,,,, 0Hoo#6Ȍ & \  `Times>b$Large Circle0 MTable_ATable_AQ O {!XXXXX9XXX   ݀    11  %$%XX9Lexicalexceptionstoleftalignmentalsooccur:apAthe;sis,ApAllinris,Epmin;ndas(Halleand  Kenstowicz1991:492).Bytreatingtheseasbearingalexicalstress,wemustassume,contraHalleandKenstowicz,thatthevowelinitialityofthesewordsisacoincidence.Inconnectionwiththis,notethatnotallvowelinitialLLLpretonicstringslackinitialstress(e.g.bracadbra),andthatthereisatleastoneexception L toleftalignmentthathasanonset(analternatepronunciationofNavrtil;vaattestedtobyAlanPrincein $t personalcommunication).R d $XXXXX9XXX      12    @XX9Ԁ%$%@Lexicallyspecificconstraintsappeartobeformallyequivalenttotheconstraintdomains,or  co-grammars,proposedinIt=andMester(1995a&b),anddiscussedinInkelas,Orgun,andZoll(1994),andPater(1994).IpreferthetackofproliferatingconstraintsoverthatofproliferatinggrammarsbecauseIbelievethatitgivesaclearerviewofthelimitsthatalanguageimposesonreranking,andespeciallybecausetheproliferationoflexicallyspecificconstraintsseemsindependentlynecessary(e.g.Alignmentconstraintsthatexpresstheprosodicsubcategorizationrequirementsofspecificmorphemes;seeMcCarthyandPrince1993b).Table_A  $XXXXX9XXX      13    ݀%$%XX9Itisinfactnotentirelyclearwhetherthisdenialconstitutesanexplanation.Burzio(1994)claimsthat  condensationlikeformsareinstancesofvowelqualitypreservation,ratherthanthestresspreservationseenin  imaginationetal.However,hedoesnotprovideaformalaccountofvowelqualitypreservation,andhence t providesnoreasonwhyvowelqualitypreservationshouldbevariable,andstresspreservationconsistent.Table_ATable_ATable_A  $XXXXX9XXX      14    @XX9Ԁ%$%@LibermanandPrince(1977:285)notewhatmightbeanothertwocasesofthistype:"Inthewords  concave,convex,theprefixretainsstress;curiously,inthederivativesconcavity,convexityitseemseasily  destressable."However,inKenyonandKnott1953,thestemsandthederivativesareequallygivenwithbothstressedandstresslessinitialsyllables,whileinWebster's1981,bothconcaveandconcavityhaveonlystressed 8 initials.Here,aswellasfortherestofthespecialcasesdiscussedin3,carefulstudyofthepronunciationsofnativespeakerswouldbeextremelyinformative(seeinthisregardthenextfootnote).     $XXXXX9XXX      15    @XX9ԀX9X@Ԁ%$%XX9Iassumethatvariationinthepronunciationofindividualwordshereandelsewhereinthissection  isduetointerspeakervariation.Thattwocompetingproductivepatternswouldproduceagreatdealofvariationistobeexpected.Ifcloserstudyrevealsthatthisvariationoccurswithinindividualspeakers,thenonemightappealtothe"floatingconstraint"formalismproposedbyReynolds(1995),whichisquiteconsistentwiththeapproachtolexicallybasedvariationtakenhere.Table_A < $XXXXX9XXX      16    ݀@XX9Ԁ%$%@IexcludefromthislistwordslikeVermont,andBerlin,sinceasLibermanandPrince(1977:284)  note,*[v rmnt]and*[b rl0n]areimpossibleinEnglish.Theremustthereforebesomeindependentconstraint  thatforcescoalescencebetween/ /and/r/inthisenvironment,makingtheseineffectlightsyllables.Table_A @ $XXXXX9XXX      17    @XX9Ԁ%$%@Kager(1989:125)alsocitesindignationasacounterexample,thoughhistranscriptionprovidesa  stressonthepretonicsyllable.ThestressedversionisinagreementwithKenyonandKnott(1953),andWebster's(1981).Thisexamplehighlightsthedifficultiesinattributingstressorstresselessnessto[I]notethat ` ` [ ]and[iy]oftenreducetosomethinglike[I],ratherthatschwa.Thepresentstudyfollowspreviousstudiesof  Englishstressinabstractingfromthedifficultiesofaccountingforvariousrealizationsofthereducedvowel.U_adfoqtCCCCCCCCCCCCCCCC {!XXXXX9XXX     VXX9%$%VPDecember1995CommentsWelcome:emailpaterjv@qucdn.queensu.ca #V%%$#  @V @Onthenonuniformityofweighttostressand L ЀstresspreservationeffectsinEnglish#@ f#X9X@*{!] G  *       ` JoePater,McGillUniversity  79XXdd7F *{!Introduction  L  SinceChomskyandHalle1968,ithasbeenagreedthatsyllableweightplaysadeterminingroleinmainstressplacementinEnglish.Nouns,forexample,arestressedonthepenultimatesyllableifitisheavy,whereeitheralongvowel(1  a),oracodaconsonant(1  b)makesasyllableheavy.Whenthepenultislight(1  c),stressisantepenultimate.EXlaXXX9#X9XXEXla#|X9XXX9 (1  )   a.ar;mabalalikahitushorzonarnaMinnes;taangna     b.agndautnsilappndixplacntasyn;psisamlgam  c.CnadaAmricacnemarsenalanlysisjvelinvnison  Therelationshipofsyllableweighttosecondarystressislessstraightforward,andsohasbeenthesubjectofmoredispute.AsHalleandKenstowicz(1991)emphasize,itistosomeextentIX9XX|X9Ԁarbitrary p whether#|X9XXIX9b#heavysyllablesbearsecondarystressornot., G  1      ׀IX9XX|X9Correspondingtothestressedheavysyllables \ in(2a)aretheunstressedheaviesin(2b).#|X9XXIX9]#gX9XX|X9 H ! ,XXX! (2)   a. ` 7ncrntion  #|X9XXgX9& #X9XX|X9Astnttion  #|X9XXX9 #X9XX|X9UKUS.,ch7mpnze p #|X9XXX9E #US.,UK.,d X9XX|X9HlicrnssusrAdomAntde#|X9XXX9 #  p   X9XX|X9b. ` Pnnsylvnia#|X9XXX9a #X9XX|X9UKUS.,  #|X9XXX9 #X9XX|X9US.,UK., rpercssion#|X9XXX9$ #UKUS.,X9XX|X9  #|X9XXX9 #X9XX|X9US.,UK., srendpity#|X9XXX9 #X9XX|X9UKUS., p K7limanjro#|X9XXX9_ #US.,UK.,~ UKUS.,X9XX|X9  Nbuchadnzzar#|X9XXX9 #US.,UK.,  ` Althoughtheexistenceofthesenearminimalpairsdoesindicatethatthelexiconhasaroletoplayinthestressingofheavysyllables,theweighttosecondarystressrelationshipisnotentirelyarbitrary.Acleardemonstrationofthiscomesfromthefactthatsyllableweightdoesunequivocallydeterminesecondarystressplacementinsomeenvironments(moresubtleargumentsagainstsimplyarbitraryweighttostressforeventhewordsin(2)arepresentedin1andIX9XX|X9Ԁ#|X9XXIX9m#3).Forexample,IX9XX|X9the"initial  dactyl"effect#|X9XXIX9#IX9XX|X9(3a;seePrince1983:49)#|X9XXIX97#,inwhichaternarystringoflightsyllablesreceivesinitial  stress,isblockedwhenthesecondsyllableisheavy,asin(3b)IX9XX|X9(#|X9XXIX9&#IX9XX|X9ChomskyandHalle1968:114and  manysubsequentothers).#|X9XXIX9i#Ԁ   X9XX|X9(3)0  a. ` Ttamag;uchibracadbraKlamaz;oW7nnepesukeeWpakontadlicatssent(#(# #|X9XXX9# X9XX|X9  b.  ` MonAngahlaValncinnesgerAntophliaBelchistn d #|X9XXX9#IX9XX|X9Stresspreservationyieldsa#|X9XXIX9#Ԁstrikinglysimilarpatternofsometimesarbitrary,andsometimes <  unequivocaldeterminationofsecondarystressplacement.Whetherornotstemstressispreservedonmedialpretonicheavysyllablesisbasicallyarbitrary,asacomparisonofthewordsin(4a)withthosein(4b)bearsout. (4)  a.IX9XX|X9 ` CEUS.,dvntgeousugmnttionuthntcitycAndmntioncAndnstionUS.,CE., (     b.CEUS., ` 7nformtionUS.,CE.,QԀCEUS.,lmenttionUS.,CE.,ԀCEUS.,ԀcAnservtionUS.,CE.,Ԁfr#|X9XXIX9i#IX9XX|X9CEUS.,#|X9XXIX9G#US.,CE.,fIX9XX|X9gment#|X9XXIX9#IX9XX|X9CEUS.,#|X9XXIX9#US.,CE.,"IX9XX|X9tiontr#|X9XXIX9r#IX9XX|X9CEUS.,#|X9XXIX9#US.,CE.,IX9XX|X9nsport#|X9XXIX9+#IX9XX|X9CEUS.,#|X9XXIX9y#US.,CE.,IX9XX|X9tion   However,stresspreservation,likeweighttostress,consistentlyoverridesthepreferenceforinitialdactyls(Hammond1989#|X9XXIX9#IX9XX|X9,Burzio1994#|X9XXIX9#IX9XX|X9):   CEUS., (5)   #|X9XXIX9#X9XX|X9US.,CE.,Iaccr#|X9XXX9#IX9XX|X9CEUS.,US.,CE.,CEUS.,dittionimgintionor7ginlitymed7cinlitydiv7sibltyphenAmen;logy d US.,CE.,B#|X9XXIX9#ThisbriefsketchofEnglishsecondarystressissufficienttoshowthatprinciplesofweighttostress, @ andstresspreservationarenotuniformlyactiveorinactive,onoroff.NoristheirapplicationdeterminedpurelylexicallyIX9XX|X9Ԁ(cf.HalleandKenstowicz1991)#|X9XXIX9%#;suchanaccountwouldpredictarbitrary h variationintheapplicationoftheseprinciplestothewordsin(3b)and(5),wherenoneinfactexists.  Totheextentthatithasbeendealtwith,thisnonuniformity(Prince1993)ofweighttostress @ andstresspreservationeffectshascreatedtremendouscomplicationsinprioranalysesofEnglishstress.Y G  2      ׀IX9XX|X9Forexample,inthestandardtreatmentofweighttostressnonuniformity,institutedby  LibermanandPrince1977andHayes1982(seealsoHalle1973),primarystressisfirstassignedinaquantitysensitivefashion,thensecondarystressassignmentproceedswithoutregardtoquantity,followedbyasetofveryspecificquantitysensitivedestressingrules(cf.Kager1989).#|X9XXIX9s!#Ԁ    Nonuniformityisproblematicbecauseoftheusuallyimplicittenetoffullsatisfaction,orinviolability,whichclaimsthatwhenalinguisticprincipleisinforce,itisneverviolated.IX9XX|X9Atheory d  basedonthistenetoftenhaslittletosayaboutaprinciplethatisonlysatisfiedincertaincontexts. P! Underfullsatisfaction,nonuniformityinthedatatendstoleadtothefollowingtheoreticalconsequences:#|X9XXIX9 $#  IX9XX|X9&  0     0` (#(#  i.theproliferationofotherwiseunmotivatedderivationalstagesorlevelsbetweenwhichtheprinciple(rule/constraint/condition)isturnedonandoff ` (#` (#   ii.alackofgeneralityinconstraintorruleformulation:nonuniformityissimplystipulated  iii.descriptivegaps:nonuniformityissimplyignored't%Aswillbedetailedbelow,theparticularcasesofnonuniformityexaminedinthispaperhaveinfactinspiredinstancesofeachoftheselessthanelegantanalyticmoves.#|X9XXIX9}%#  L    AsbroughtoutmostforcefullybyPrince(1993),andMcCarthy(1995),nonuniformityisanexpectedconsequenceofOptimalityTheoreticconstraintrankingandviolability,insteadofanunwelcomeburden.ThebasicdistinguishingcharacteristicofOptimalityTheory(PrinceandSmolensky1993!henceforthP&S1993)isthatitabandonsfullsatisfactioninfavorofminimalviolation:aconstraintisviolatedonlytotheextentnecessarytosatisfyahigherrankingone.Inatheorybasedonminimalviolation,nonuniformityreceivesanabsolutelydirecttreatment.Aconstraintisviolatedinaparticularenvironmentbecauseitssatisfactionwouldconflictwiththesatisfactionofahigherrankedconstraint.Ifinanothercontextthehigherrankedconstraintmakesnoconflictingdemands,thelowerrankedoneisobeyed.   H   Theexistenceofnonuniformitythusprovidescompellingargumentsagainstfullsatisfaction,andinfavorofminimalviolation.Inthispaper,IcloselyexaminethemassivelynonuniformeffectsofsyllableweightandstemstresspreservationonsecondarystressinEnglish,andshowthatconstraintrankingpermitsexplanatorygainsonanumberoffronts:0     0` (#(#  i.stressassignmentisintegratedintoasinglelevel ` (#` (#   ii.constraintsareformulatedinaverygeneralfashion  iii.previouslyinexplicabledataaremadesenseof p     Thepaperisstructuredasfollows.InIX9XX|X91#|X9XXIX9.#Ԁtheusual,productivepatternsofweighttosecondarystress  areaccountedforintermsofasmallsetofrankedconstraints.Section2containsabriefdiscussionofprimarystressplacement,demonstratingthatquantitysensitiveprimarystressplacementiscompatiblewiththerelativelylowrankoftheWeighttoStressconstraintthatthefactsof D! secondarystressplacementrequire.Withanaccountoftheusualdistributionofstressinhand,IthenturntothedataforwhichHalleandKenstowicz(1991)invokelexicallyconditionedquantitysensitivity!lexicalexceptions,andapparentcasesofcyclicstresspreservation.Inthisthirdsection,IpointoutseveralempiricalshortcomingsoftheHalleandKenstowiczanalysis,andIX9XX|X9argueforan $D # accountoftheseinstancesofspecialsecondarystressbased#|X9XXIX91#IX9XX|X9both#|X9XXIX92#IX9XX|X9on#|X9XXIX92#IX9XX|X9s#|X9XXIX93#IX9XX|X9tresspreservation,formalized %0!$ asprosodicfaithfulness,andon#|X9XXIX9X3#IX9XX|X9lexicallyspecificconstraintrankin#|X9XXIX93#IX9XX|X9g#|X9XXIX9T4#.IX9XX|X9CEUS.,Byinterspersingprosodic &"% faithfulnessconstraints,andlexicallyspecificconstraints,intothehierarchyestablishedforregular stress,thesameprinciplesthatdetermineregularprosodificationcanbeusedtostraightforwardlyand (#' P    !P    P    !preciselycapturethedistributionoflexical,andstembasedstress.Asaspurtofurtherresearch,!P    P    !r#|X9XXIX94#US.,CE.,4& L esidualdata,andissues,arediscussedinappendices.    1.Weighttosecondarystress  ` 1.1The'Arab'rule L  Iwillstartwithaparticularlycomplexcaseofnonuniformitywhichhastothispointdefiedprincipledanalysis,andwhichcaninfactbereducedtotherankingofahandfulofbasicmetricalconstraints.IX9XX|X9Kager(1989)citesdatasuchasthefollowingtoshowthat#|X9XXIX98#IX9XX|X9anobstruentfinalsyllable#|X9XXIX9u9#IX9XX|X9(%o)#|X9XXIX99#IX9XX|X9is  ` unstressedifitisprecededbyalightsyllable(6a),butstressedifprecededbyaheavysyllable(6b):& A (6)   a. (l%o) 0 lexndernaptUxisrithmtic(adj.)cAllectnia.picttus.rechttum $  rsigntionsrrepttiousMzatlnmAlybdnum  (# (# ' :  b. (h)(%o) 0 dlcttioncAutch;uc7ncAgnto(alt.)t7cktckt;eT7mbckt;o  (# (# #|X9XXIX9#:#IX9XX|X9Posttonically,thisbehaviorof#|X9XXIX9<#IX9XX|X9%o's#|X9XXIX9=#IX9XX|X9produces#|X9XXIX9l=#IX9XX|X9Ԁwhathasbeencalledthe'Arab'rule,duetothe   covarianceoffinalsyllablestresswiththelengthoftheinitialvowelintwoidiolecticalpronunciationsof'Arab'(i.e.#|X9XXIX9=#IX9XX|X9[#|X9XXIX9>#IX9XX|X9]rabvs.#|X9XXIX9>#IX9XX|X9Ԁ[#|X9XXIX9H?#IX9XX|X9ey]rb;seeFidelholtz1967;Ross1972)#|X9XXIX9?#IX9XX|X9. d #|X9XXIX9?#IX9XX|X9  Alsorelevantherearethefollowingnearminimalpairs: P  (7)   a. nnextion  dapttion  #|X9XXIX9M@#IX9XX|X9#|X9XXIX97A#IX9XX|X9ffect#|X9XXIX9zA#IX9XX|X9#|X9XXIX9A#IX9XX|X9tion p  (x Ѐ  b. 7ndxtion  rlxtion  #|X9XXIX9B#IX9XX|X9#|X9XXIX9B#IX9XX|X9xp#|X9XXIX9B#IX9XX|X9#|X9XXIX9AC#IX9XX|X9ct#|X9XXIX9C#IX9XX|X9#|X9XXIX9C#IX9XX|X9tion p  h Though#|X9XXIX9 D#IX9XX|X9thestemsoftheverticallyalignedwords#|X9XXIX9D#IX9XX|X9displaynearlyidenticalstresspatterns,thederived D wordswithinitiallightsyllables(7a)canhavestresslessmedialheavysyllables,whilethosewithheavyinitials(7b)musthavestressonbothofthefirsttwosyllables.k G  3      ׀  #|X9XXIX9D#  IX9XX|X9ԀWordsinwhichthefirsttwosyllablesareparsedas (l%o) violateWeighttoStress,since  theheavysyllableisunstressed.#|X9XXIX9F#ԀPrinceandSmolensky(1993:53)statethisconstraintasin(8).   (8) WeighttoStress    HeavysyllablesareprominentinfootstructureandonthegridThisconstraintrequiresthataheavysyllablebeboththeheadofafoot,anddominatedbyanaccentualgridmarkindicatingstress.ZT G  4      ׀UnderthepremisesofOptimalityTheory,inwhichconstraint 4#!  violationmustbecompelledbysomehigherrankingconstraint(s),Englishmusthaveatleastone  $p" otherconstraintrankedaboveIX9XX|X9WeighttoStress#|X9XXIX9'J#,whichthisparsingmanagestosatisfy,andwhich  wouldbeviolatedifIX9XX|X9WeighttoStress#|X9XXIX9J#Ԁwererespected.     OnesuchconstraintisIX9X|X9Parse#|X9XXIX9K#LXXX|X9%#|X9XXLXK#,whichdemandsexhaustivityofconstituentconstructionX9XX|X9Ԁ(see  McCarthy1993,Prince1993,#|X9XXX9RL#X9XX|X9Mester1994,#|X9XXX9L#X9XX|X9CabrandKenstowicz1995,andespeciallyP&S1993  onLatinforcloselyrelatedParse#|X9XXX9M#LXXX|X9%#|X9XXLXM#X9XX|X9Ԁeffectsinotherlanguages)#|X9XXX9N#.X9XX|X9IncontrastwithHalleandVergnaud x (1987),HalleandKenstowicz(1991),andearlierwork,Exhaustivityisrecognizedasaviolable,ratherthananinviolableprincipleinmostrecentworkinmetricalphonology,evenoutsideofOptimalityTheory(seeKager1989,Idsardi1992,HalleandIdsardi1995,Hayes1995,andespeciallyBurzio1994,whereitplaysanactiveroleindeterminingthewellformednessofmetrifications).#|X9XXX9qN#IX9X|X9Parse#|X9XXIX9sP#LXXX|X9%#|X9XXLX߾P#Ԁcanbestatedasin(9). , |   (9) ParseSyllable(ParseLXXX|X9%X9XXLX): l    SyllablesmustbelongtofeetSyllablesnotparsedbyfeetareassumedtobeparsedbytheProsodicWord(It=andMester1992,McCarthyandPrince1993a).  IfParseLXXXX9%X9XXLXԀisrankedaboveWeighttoStress,thefullyparsedIX9XXX9 (l%o) #X9XXIX9S#ԀispreferredtoIX9XXX9 l(%o) ,   inwhichtheheavysyllableistheheadofthefootandbearsstress#X9XXIX9oS#.Asthingsstandthough,nothing $ discriminatesagainsttheparsing (l)(%o) ,wherethetwosyllablesformseparatefeet,fulfillingthe  requirementsofbothconstraints.WhatrulesthisoutisFootBinarity(P&S1993:47;seealso  McCarthyandPrince1986and1993a):&  (10) FootBinarity(FtBin): t   Feetarebinaryatsomelevelofanalysis(LXXXX9%,%X9XXLX). d 'UTherequirementthatafootmustcontainatleasttwomoras(LXXXX9%X9XXLX)disallowsthe (l) footneededforthe  T (l)(%o) parsing,sincealightsyllablecontainsbutonemora. X   Furthermore,allofFtBin,ParseLXXXX9%,#X9XXLXW#andWeighttoStresscouldbesatisfiedbyparsing H  (l%o) asasinglerightheadedfoot(e.g.Alxnder).However,thiswouldviolateaconstraint L demandingleftheadedfeet@XX9,#X9X@Y#whichcouldbeformulatedaseitheraligningtheheadsyllableofafoot <  withitsleftedge(cf.McCarthyandPrince1993b)orasRhythmicType=Trochaic(P&S1993,McCarthyandPrince1993a).Astheexactformulationisofnoconsequencehere,Iwillsimplycalltheconstraint'Troch'. #P    WithFtBin,ParseLXXXX9%,IX9XXLXandTroch#LXXXIX9Z#ԀX9XXLXrankedaboveWeighttoStress,leftheaded (l%o) is #<! chosenasoptimal,insteadof (l)(%o) , l(%o) ,orrightheaded (l%o) .Thetableauin(11)illustratesthis $@ " resultforthepretonicstringAlexofAlexander(asprimarystressplacementisdiscussedin2,the %0!# mainstressparsingisleftunindicateduntilthen).Constraintsseparatedbyasolidlineinthetableau &"$ arerankedwithrespecttooneanother,andthoseseparatedbyadashedlineareunranked.Constraintviolationsareindicatedbyanasterisk,andanexclamationmarkshowstheviolationthatrulesoutaparticularcandidate.Theoptimalcandidate,whichisthegrammaticalform,receivesa checkmark. |*%( &     (11)FtBin,ParseLXXXX9%#X9XXLX:_#\X9XXX9,#X9XX\X9_#IX9XXX9Troch#X9XXIX9_#ԀNWeighttoStress    _*sMN ddd Xdd Xdd X(#(#s,rB@,3RB@,eRB@,RB@,Rr@+  /OOO "r 20C/Input: R Alexander 9OPO(>COOO C9FtBin 9OHO(RC"OPO C9ParseLXXX9%X9XXLX 9OHO(RC"OHO C9Troch ;OPOO(RC"OHO C;Weight R ЄtoStress OPO@(> C" 00 2 OPOO <<COrX9XXX9( lex)|nderrXrX93#rX9Xrc# 7PP(  CPO C7 7PH(  C"PP C7 7PH(  C"PH C7 ;PPO(  C"PH C;rXrX9*#rX9Xrd# OPO@( C" <<  PPO <<COA(lx)|nderX9XXrX9 7PP(n CPO C7 7PH(n C"PP C7XX9*!#X9Xf# 7PH(n C"PH C7 ;PPO(n C"PH C; RPOC(n C" <<  PPO 5<<CR~X9XXX9( )(lx)|nder :PP+F CPO C:~X~X9*!#~X9X~xg# :PH+F C"PP C: :PH+F C"PH C: >PPO+F C"PH C> QPOO@(F C" << 5 PPO 5<<CQgX9XX~X9(Alx)|nder 9PPO(n CPOO C9 9PHO(n C"PPO C9 9PHO(n C"PHO C9gXgX9*!#gX9Xgi# ;PPOO(n C"PHO C;8.,n C" << 5  PPOO 8' ^Sincetheheavysyllableintheoptimalformisunstressed,thiscandidateviolatesWeighttoStress.   Asshownbytheothercandidates,thesatisfactionofX9XgX9WeighttoStress#gX9XXX9k#Ԁrequirestheviolationof   oneofthehigherrankingconstraints!FtBin,Parse-LXXXgX9%,hX9XXLXorTroch#LXXXhX9l#.nX9XXLX ~    Whentheinitialsyllableisheavy,stressonthepretonicsyllableiscorrectlygenerated,astableau(12)shows.& \ (12)Tableauforh%o  F$ *Q ddrB@3RB@eRB@RB@Rr@MN(#(#,rB@,3RB@,yRB@,yRB@,Rr@+  /OOO 2% 20C/Input: b& Timbucktoo 9OPO(N'COOO C9FtBin 9OHO(b(C"OPO C9Parse% 9OHO(b)C"OHO C9Troch ;OPOO(b*C"OHO C;Weight b+ ЄtoStress OPO@(N,C" 00 2 OPOO <<CO|X9XXnX9(T7mbuck)|t;o 7PP(-CPO C7 7PH(.C"PP C7 7PH(/C"PH C7 ;PPO(0C"PH C;|X|X9*!#|X9X|zq# OPO@(1C" <<  PPO <<COTim(bck)|t;o#nX9XX|X9fp# 7PP(~2CPO C7 7PH(~3C"PP C7nXnX9*!#nX9Xnr# 7PH(~4C"PH C7 ;PPO(~5C"PH C;nXnX9*#nX9Xnfs# OPO@(~6C" <<  PPO <<CO(Timbck)|t;o 7PP(V7CPO C7 7PH(V8C"PP C7 7PH(V9C"PH C7nXnX9*!#nX9Xnt# ;PPO(V:C"PH C;nXnX9*#nX9Xn-u# QPOO@(V;C" <<  PPO 5<<CQ|X9XXnX9(T7m)(bck)|t;o|X|X9Ԁ3#|X9X|u# 9PPO(. ~<CPOO C9 9PHO(. ~=C"PPO C9 9PHO(. ~>C"PHO C9 ;PPOO(. ~?C"PHO C;8.,. ~@C" << 5  PPOO 8#nX9XX|X9u#'\ FmThedifferencehereisthattheinitialsyllablecontainsacodaconsonant,sothatitcanbeparsedalone "A asabimoraicfoot,withoutcausingaviolationofFtBin.ThiseliminatestheconflictbetweenIX9XnX9FtBin#nX9XXIX9~x# #B andIX9XnX9WeighttoStress#nX9XXIX9x#Ԁsatisfaction,allowingthelowerrankedIX9XnX9WeighttoStress#nX9XXIX9ey#Ԁtochoosethe $C candidatewithheavysyllablestress.Thus,thiscomplexcaseofnonuniformity,inwhichtheweightoftheprecedingsyllabledetermineswhethera IX9XXnX9%o getsstressedornot#nX9XXIX9sz#,isreducedtotherankingof f&!E threeextremelywellmotivatedconstraintsaboveIX9XnX9WeighttoStress.#nX9XXIX9?{#ԀThedifficultythiscaseposes V'"F foratheorybasedonfullsatisfactionisattestedtobythefactthatinpreOptimalityTheoreticanalyses,thisgeneralizationhaseitherbeenleftunaccountedfor(seePrince1985:486foranexplicitdiscussionoftheinabilityofthencurrenttheoriestocopewithit),orsimplystipulated(amongstmetricaltheorists,seeHayes1982:256andKager1989).   +B'K 1.2Sonorantfinalsyllables  k}AstheobstruentspecificformulationoftheArabruleimplies,sonorantfinalsyllablesIX9XXnX9(%s)behave  differently.Thestressingofa#nX9XXIX92~#Ԁ%sdoesnotdependontheweightoftheprecedingsyllable.Thisis  attestedtobythedatain(13),inwhichIX9XXnX9pretonic%s'sprecededbybothlightandheavysyllablesare t uniformlyunstressed#nX9XXIX9H#. `  (13)   a. (l%s) 0 ffirmtionlmenttiond7sserttionrpercssionsrendpitys7multneous 8  trantlla  (# (#   b. (h%s) 0 cAmpenstion7nformtionsurptionPnnsylvniaMAzambque  d gArgonz;la h cAnsulttion T  (# (# Withjusttheconstraintsintroducedabove,thestring hIX9XXnX9%s #nX9XhIX9#Ԁwouldbetreatedlike hIX9XXnX9%o ,#nX9XXIX9n#IX9XXnX9Ԁ#nX9XXIX9#andparsedas ,   (h)(%s) .However,theproductivityofpretonicIX9XXnX9 (h%s#nX9XXIX9h#) isdemonstratednotonlybythevastnumerical   superiorityIX9XXnX9of (h%s) #nX9XXIX9#ԀoverIX9XXnX9 (h)(%s#nX9XXIX9y#) (Kager1989:123),butalsobytheexistenceofderivedwordsin   whichasyllablethatisstressedinthestembecomesstressless,soastoconformtothe IX9XXnX9(h%s)#nX9XXIX9#Ԁ pattern   (e.g.7n for mtionandcAn sul ttionfromin f;r mandcon sl t!seefurther3).     Toruleout IX9XXnX9(h)(%s#nX9XXIX9#IX9XXnX9) #nX9XXIX9#,theremustbeanactiveconstraintthatdisfavoursstressonthepretonic  syllable.Suchaconstraintcanbederivedfromthe"StressWell"environmentofHalleandVergnaud1987:238,whichisusedtotargetstressedsyllablesadjacenttothemainstressfordestressingandshortening(seealsoLibermanandPrince1977:285andinterveningworkonEnglishstressforsimilarnotions).Iassumetheformulationin(14).&  (14)   StressWell h    ` NostressedsyllablemaybeadjacenttotheheadsyllableoftheProsodicWord'Thisconstraintmayberegardedasaslightlymorespecificinstantiationofthegeneralprohibitionagainstadjacentstresses,orstressclash(Prince1983,Hammond1984).TheevidencefromwithinEnglishforthisspecificformulationisthatadjacentstressespersearewelltolerated.Wordslike  T7cAnder;gashownotendencytowardbecomingclashless;IX9XXnX9examplesparallelto*T7conder;ga#nX9XXIX9h#Ԁare  infactcompletelyunattested(seeIX9XXnX93#nX9XXIX9#IX9XXnX9.3)#nX9XXIX9Z#.Furtherexemplificationofthestrongdispreferencefor  pretonicstress,andthelackofaparallelintoleranceofmereadjacency,isprovidedinIX9XXnX93#nX9XXIX9G#.2. |    Itisnoteasytomarshallcrosslinguisticevidenceforthisconstraint,asitseffectsareoftenindistinguishablefromsimple*Clash.However,Hayes(1995:157)notesthatMaithilihas T"  specificallypretonicshortening,whichcouldbereducedtothecombinedeffectsofStressWelland @#! WeighttoStress,aspretonicshorteningwouldresultinthesatisfactionofbothofthese ,$|" constraints.&    '&BeforeproceedingtorankIX9XXnX9StressWell#nX9XXIX9ӏ#Ԁsoastoproduce IX9XXnX9(h%s#nX9XXIX9:#) ,itmustbeensuredthatits &T!$ introductiondoesnotaltertheresultsobtainedforobstruentfinalsyllables.Thefactthata%ois &D"% usuallystressedwhenadjacenttoaprimarystress,indefianceofIX9XXnX9StressWell#nX9XIX9#,canbeattributedto '0#& therankingofWeighttoStressaboveIX9XXnX9StressWell:#nX9XXIX94# ($'   )%(  &   (15)WeighttoStressNIX9XXnX9StressWell#nX9XIX9#   *RS ddrB@3RB@yRB@yRB@Rr@Q(#(#,rB@,JRr@,JRr@+  /OOO  20C/Input:  Timbucktoo ;OPOP(COOO C;Weight  ЄtoStress >OOO+C"OPOP C>IX9XXnX9StressWell#nX9XXIX9N# OPO@(C" 00 2 OOO <<CO|X9XXnX9(T7mbuck)|t#nX9XX|X9#IX9XXnX9;#nX9XXIX9?#|X9XXnX9o ;PPP(`CPO C;|X|X9*!#|X9X|# >PO+`C"PPP C>#nX9XX|X9# QPOO@(` C" <<  PO 5<<CQ|X9XXnX9(T7m)(bck)|t#nX9XX|X9ٗ#IX9XXnX9;#nX9XXIX9+#|X9XXnX9o|X|X9Ԁ3#|X9X|# ;PPOP( 8 CPOO C; >POO+ 8 C"PPOP C>|X|X9*#|X9X|Q#8., 8 C" << 5  POO 8#nX9XX|X9q#X9XXnX9IX9XXX9'ThistableaushowsthattheheavysyllablestresslessnessrequiredtosatisfyStressWellleadstoa p   violationofthehigherrankedWeighttoStress,andisthusruledout.#X9XXIX9# \    TheeasewithwhichaIX9XXX9%s#X9XXIX9f#ԀisunstressedrelativetoIX9XXX9a%omayatfirstseemunexpected,giventhe H  crosslinguisticgeneralizationthatifasubsetofconsonantsismoraic,itiscomposedofthesonorants,ratherthantheobstruents(Prince1985;seePrince1983:57,Zec1988onthetypologicalfacts).However,thesheddingofamoraisnottheonlymeansbywhichthedemandsofWeight-to-Stresscanbeescaped.#X9XXIX9Λ#IX9XXX9ThekeyhereisthelongstandingobservationthatinEnglish#X9XXIX9#, H  syllablefinalsonorantsareincorporatedintothenucleuswhenunstressed(seerecentlyLibermanandPrince1977:299,Travis1983,PiggottandSingh1985;Kager1989:166).Assuch,aIX9XXX9%s#X9XXIX9#Ԁincursno   Weight-to-Stressviolation,andIX9XX9StressWell#X9XXIX9#Ԁisfreetochoosepretonicstresslessness.   IX9XXX9 &  (16)Tableauforh#X9XXIX9#IX9XXX9%s#X9XXIX9y#IX9XXX9   *stu ddrB@JRr@JRr@RS(#(#s,rB@,JRr@,JRr@+  /OOO  20C/Input:  Pennsylvania ;OPOP(COOO C;Weight  ЄtoStress >OOO+C"OPOP C>#X9XXIX9Ơ#IX9XXX9StressWell#X9XIX9#IX9XX9 OPO@(C" 00 2 OOO <<CO#X9XXIX9>#IX9XXX9(Pnnsl)|vnia#X9XXIX9ң#IX9XXX9IXIX93#IX9XIC##X9XXIX9$#IX9XXX9 ;PPP(DCPO C; >PO+D C"PPP C>#X9XXIX9#IX9XXX9 OPO@(D!C" <<  PO 5<<CO#X9XXIX9g#IX9XXX9(Pnnsyl)|vniaIXIX9Ԁ#IX9XI&# ;PPP("CPO C;IXIX9*!#IX9XI# >PO+#C"PPP C>#X9XXIX9#IX9XXX9 QPOO@($C" << 5 PO 5<<CQ#X9XXIX9J#IX9XXX9(Pnn)(s)yWdd0l)|vniaIXIX9Ԁ#IX9XI&# ;PPOP(%CPOO C; >POO+&C"PPOP C>IXIX9*!#IX9XI#8.,'C" << 5  POO 8#X9XXIX9ݧ#IX9XXX9' 2FollowingLibermanandPrince(1977),#X9XXIX9#ԀasonorantnucleusisindicatedbytheIX9XXX9omissionofthevowel , |( inthespelling.#X9XXIX9%#Ԁ !h)   Fromthisperspective,thegreaterattractionofstresstoIX9XXX9%o's#X9XXIX9#ԀthantoIX9XXX9%s#X9XXIX9W#'scanbederivedfrom "T* thecrosslinguisticgeneralizationthatsonorantsmakebetternucleithandoobstruents(P&S8).Toexpressthisgeneralization,Iwillinvokeapairofconstraints,andauniversallyfixedrankingbetweenthem(cf.P&SIX9XXX98#X9XXIX9#).If*SonNucisviolatedbyanuclearsonorantconsonant,and*ObsNucbya $ - nuclearobstruent,thenthefixedrankingIX9XX9*ObsNuc#X9XXIX9#Ԁ IX9XXX9N #X9XXIX9#IX9XX9*SonNuccreatesauniversaldispreferencefor %!. #X9XXIX9.#obstruentnucleirelativetosonorants.SolongasIX9XX9*ObsNuc#X9XXIX9#dominatesWeighttoStressand &!/ IX9XXX9StressWell#X9XIX9q#IX9XX9,obstruentswillre#X9XXIX9#sistintegrationintothenucleus,evenatthecostofviolatingthelatter '"0 twoconstraints.WithIX9XX9*SonNuc#X9XXIX9#Ԁrankedbeneaththeseconstraints,theywillcontinuetocompel |(#1 sonorantnuclei.Thetableauin(17)showshowthisrankingstopsobstruentsfrombehavinglike sonorantsinthefaceofaIX9XXX9StressWell#X9XIX9#violation. T*%3 & <#  IX9XXX9 (17)#X9XXIX9A#IX9XXX9*ObsNuc#X9XXIX9#IX9XXX9NStressWell #X9XIX9۲#IX9XX9  *sop ddrB@JRr@JRr@tu(#(#s,rB@,RB@,rRB@,Rr@+  /OOO  20C/Input:  Timbucktoo 9OPO(COOO C9*ObsNuc ;OPOP(C"OPO C;Weight  ЄtoStress >OOO+C"OPOP C>#X9XXIX92#IX9XXX9StressWell#X9XIX9#IX9XX9 OPO@(C" 00 2 OOO <<CO#X9XXIX9 #IX9XXX9(T7mbck)|t;o 7PP(L CPO C7IXIX9*!#IX9XI# ;PPP(L C"PP C; >PO+L C"PPP C>#X9XXIX9#IX9XXX9 OPO@(L C" <<  PO <<CO(T#X9XXIX9#IX9XXX97#X9XXIX9{#IX9XXX9mbuck)|t;o 7PP( $ CPO C7 ;PPP( $C"PP C;IXIX9*!#IX9XI_# >PO+ $C"PPP C> QPOO@( $C" <<  PO 5<<CQ#X9XXIX9#IX9XXX9(T7m)(bck)|t;oIXIX9Ԁ3#IX9XI# 9PPO( CPOO C9 ;PPOP( C"PPO C; >POO+ C"PPOP C>IXIX9*#IX9XI#8., C" << 5  POO 8#X9XXIX9V#IX9XXX9Theranking#X9XIX9"#IX9XX9*ObsNuc#X9XXIX9s#IX9XXX9NStressWel#X9XIX9#IX9XX9lrulesoutanuclearobstruent,asinthetopcandidate,infavor 4  ofpretonicstress,asinthefinal,optimalone.Thefinalcandidatewouldremainoptimalif#X9XIX9#IX9XX9*ObsNuc  p  #X9XXIX9#IX9XXX9andWeighttoStresswereunranked,sinceallthatisnecessarytoruleoutthesecondcandidate  \  istherankingof#X9XXIX9q#IX9XX9WeighttoStress#X9XIX9C#IX9XXX9NStressWel#X9XIX9#IX9XX9l#X9XIX9#IX9XXX9.#X9XXIX92#IX9XXX9Thenecessarydominanceof#X9XIX9v#IX9XX9*ObsNuc#X9XXIX9#IX9XXX9Ԁis H  displayedonlyinasituationinwhich#X9XIX9"#IX9XX9WeighttoStress#X9XXIX9#IX9XXX9isviolatedbytheoptimalcandidate: 4  (18)#X9XIX9#IX9XX9*ObsNuc#X9XXIX9#IX9XXX9N#X9XIX9#IX9XX9WeighttoStress#X9XXIX9#IX9XXX9    *UV ddrB@RB@rRB@Rr@op(#(#,rB@,RB@,rRB@+  /OOO  20C/Input: ( Alexander 9OPO(COOO C9*ObsNuc ;OPOO((C"OPO C;Weight (  ЄtoStress#X9XXIX9p#IX9XXX9 OPO@(!C" 00 2 OPOO <<CO( lx)|#X9XXIX9#IX9XXX9#X9XXIX9I#IX9XXX9nder 7PP(l"CPO C7IXIX9*!#IX9XI# ;PPO(l#C"PP C;#X9XXIX9#IX9XXX9 QPOO@(l$C" <<  PPO 5<<CQ( lex)|#X9XXIX9#IX9XXX9#X9XXIX9&#IX9XXX9nder#X9XXIX9i#IX9XXX9ԀIXIX93#IX9XI##X9XXIX9#IX9XXX9 9PPO(D%CPOO C9 ;PPOO(D&C"PPO C;IXIX9*#IX9XI#8.,D'C" << 5  PPOO 8S#X9XXIX9:#IX9XXX9onorantnuclei,ontheotherhand,continuetobepreferredoverpretonicstress,since#X9XXIX9m#IX9XXX9Ԁ#X9XXIX9#IX9XX9StressWell |( #X9XXIX9L#IX9XXX9dominates#X9XXIX9#IX9XXX9Ԁ#X9XXIX9#IX9XXX9*SonNuc#X9XXIX9>#IX9XXX9: h) #X9XXIX9# (19)IX9XXX9StressWell#X9XXIX9#IX9XXX9N#X9XXIX9C#IX9XXX9Ԁ#X9XXIX9#*SonNuc  @+ IX9XXX9*sqs ddrB@RB@rRB@UV(#(#s,rB@,JRr@,JRr@+  /OOO , |, 20C/Input: \ - Pennsylvania ;OPOP(H!.COOO C;#X9XXIX9#IX9XXX9StressWell#X9XXIX9#IX9XXX9 >OOO+\ /C"OPOP C>#X9XXIX9#IX9XX9*SonNuc#X9XXIX9n#IX9XXX9 OPO@(\ 0C" 00 2 OOO <<CO#X9XXIX9#IX9XXX9(Pnnsl)|#X9XXIX9K#IX9XXX9v#X9XXIX9#IX9XXX9#X9XXIX9#IX9XXX9nia#X9XXIX9#IX9XXX9IXIX93#IX9XI##X9XXIX9c#IX9XXX9 ;PPP("1CPO C; >PO+"2C"PPP C>#X9XXIX9#IX9XXX9IXIX9*#IX9XI# QPOO@("3C" <<  PO 5<<CQ#X9XXIX9#IX9XXX9(Pnn)(s)yWdd0l)|#X9XXIX9|#IX9XXX9v#X9XXIX9#IX9XXX9#X9XXIX9&#IX9XXX9niaIXIX9Ԁ#IX9XI# ;PPOP(x$4CPOO C;#X9XXIX9i#IX9XXX9IXIX9*!#IX9XIK##X9XXIX9,#IX9XXX9 >POO+x$5C"PPOP C>8.,x$6C" << 5  POO 8#X9XXIX9#'<#*Beforeturningtofurtherdataandanalysis,itisworthnotingthecontributionofthetheoretical 'P"7 assumptionofParallelism(seeP&S1993,McCarthy1993,CohnandMcCarthy1994,McCarthy '<#8 andPrince1995)tothistreatmentoftheIX9XXX9%s/#X9XXIX9#IX9XXX9%oasymmetry.I#X9XXIX9#tappearstobecrucialthatsyllabification (($9 andstressassignmentbeevaluatedinparallel,ratherthanestablishedandevaluatedinsequence(seeP&S1993:3.2forargumentstothiseffectforTongan).Whetherthesecondsyllableina hIX9XXX9%#X9XXIX9K#Ԅ  *&; pretonicsequenceisunstresseddependsuponwhetherthesyllablefinalconsonantcanbeparsedas +&< anucleus.Whetherasonorantisparsedasanucleusinturndependsuponwhetheritisunstressed.Thissortofinterdependencebetweenthewellformednessofstressandsyllablestructureisextremelydifficulttoexpressinatheoryinwhichsyllabificationderivationallyprecedesstressplacement.IX9XXX91.3Otherpretonicpatterns1.3.1Oddparitystrings L  #X9XXIX9Q#IX9XXX9Combiningtherankingsmotivatedthusfar,thecompletehierarchystandsasfollows: 8   (20) #X9XXIX9#IX9XXX9ԀFtBin,Parse%,Troch,*ObsNucNWeighttoStressNStressWellN*SonNuc#X9XXIX9#IX9XXX9  ` #X9XXIX9m#JF7XXdXXd7IX9XXX9Aninterestingandencouragingpropertyofthishierarchyisthatitneedsverylittleembellishmentto  <  generatethecorrectresultsforoddparitypretonicstringsofsyllables.Inaccountingforthestressingofbisyllabicstrings,wehaveestablishedthatParse#X9XIX9 #^X!XX9%#X9X!X^Xm#X9XXX9ԀdominatesWeighttoStress   (4( lex)ander;*A(lx)ander)andthatWeighttoStressdominatesStressWell   (4(T7m)(bck)too;*(T7mbuck)too).Thismeansthatbytransitivity,Parse%dominates   StressWell,whichleadstothepredictionthatasinglepretonicsyllableshouldbeparsed,evenat   theexpenseofaStressWellviolation.    Thispredictionisborneoutinthedifferencebetweenasinglepretonic%s,andonethatis x precededbyanothersyllable.Aswesawinthelastsubsection,whena%sfollowsanothersyllable, d lightorheavy,itisparsedastheweakmemberofabisyllabicfoot,inobediencetoStressWell. P However,whenthereisbutasinglesyllable,#X9XXX9#IX9XXX9Parse#X9XIX9#^X!XX9%#X9X!X^X#IX9XXX9forces#X9XXIX9$#X9XXX9ittobeparsedalone,andstressed,in < contraventionofStressWell.Illustrativedataappearin(21),andanillustrativetableauin(22). @  (21)  bn dna Nn tcket pAn t;on cn ten cn trion cn tnkerous h  (22)Tableaufor%s  D *s_` ddrB@JRr@JRr@qs(#(#s,FB@,B@,erB@,rB@,r@+  3OOO" 0 llC3Input: L bandana 9OPO(8COOO C9Parse L ! 4  4 !#X9XX9m#^X!XX9%#X9X!X^X#X9XXX9 9OPO(8C"OPO C9Weight L ! 4  4 !toStress 9OPO(8 C"OPO C9Stress L! Well ;OPOO(8"C"OPO C;*Son L# Adj OPO@(8$C" ll  OPOO L<CO#X9XXX9c#IX9XXX9(bn)|dna#X9XXIX9G#IX9XXX9IXIX93#IX9XI##X9XXIX9#IX9XXX9 7PP(D%CPO C7#X9XXIX9#X9XXX9 7PP(D&C"PP C7 7PP(D'C"PP C7XX9*#X9X!# ;PPO(D(C"PP C; OPO@(D)C" L<  PPO L<CObn|dna#X9XXX9#IX9XXX9 7PP(!,*CPO C7#X9XXIX9#X9XXX9XX9*!#X9X# 7PP(!,+C"PP C7 7PP(!,,C"PP C7XX9 ;PPO(!,-C"PP C;*#X9Xc# QPOO@(!,.C" L<  PPO LlCQ#X9XXX9#IX9XXX9ban|dnaIXIX9Ԁ#IX9XI|#IXIX9 9PPO(#/CPOO C9*! 9PPO(#0C"PPO C9* 9PPO(#1C"PPO C9#IU##X9X# ;PPOO(#2C"PPO C;8.,#3C" Ll   PPOO 8NotethattheavailabilityofanuclearsonorantasanalternativetoaWeighttoStressviolationhas |&!4 noeffectontheoutcome,because#X9XXX9#IX9XXX9inthisinstance#X9XXIX9,#X9XXX9,Parse#X9XX9~#^X!XX9%,#X9X!X^X#IX9XXX9ratherthanWeighttoStress,isthe h'"5 constraintmotivatingheavysyllablestress.#X9XXIX9#X9XXX9 l(#6   Whena%sisthelastmemberofatrisyllabicpretonicstring,italsousuallyreceivesstress,as X)$7 theexamplesin(23)illustrate.  0+&9  &   (23)   a.(%%)(%s) 0 HlicrnssusrAdomAntdep7thecnthr;pus (# (#   b. (%%)(%s) 0 rgumnttion7nstrumnttionscramnttionsdimnttion#X9XXX9#X9XXX9lephntasis  #X9XXX9#X9XXX9'  (# (# Thewordsin(23b)arebasedonrootswithoutstressonthefinalsyllable,whichprecludesananalysisinwhichthestressisstoredunderlyingly,andpointstotheproductivityofthispatternofsecondarystressassignment(Kager1989:123).TheproductivityofpretonicstressinthisenvironmentalsofollowsfromthedominanceofParse#X9XX9|#^X!XX9%#X9X!X^XA#^X!XXX9:#X9X!X^X#X9XXX9 @   & < (24)Tableaufor%%%s  0  *ab d,dFB@B@erB@rB@r@_`(#(#,SB@,B@,erB@,rB@,r@+  3OOO"  l  llC3Input:    Halicarnassus 9OPO(t COOO C9Parse#X9XX9#^X!XX9%#X9X!X^X:#X9XXX9 9OPO(  C"OPO C9Weight    ! |    |   !toStress 9OPO(t C"OPO C9Stress   Well ;OPOO(t C"OPO C;*Son   Adj OPO@(t C" ll  OPOO L<CO#X9XXX9#IX9XXX9(#X9XXIX9g#X9XXX9Hli)(cr#X9XXX9#IX9XXX9)|n#X9XXIX9#X9XXX9#X9XXX9:#IX9XXX9ssus#X9XXIX9}#IX9XXX9IXIX93#IX9XI##X9XXIX9#IX9XXX9 7PP(h CPO C7#X9XXIX9K#X9XXX9 7PP(h C"PP C7 7PP(h C"PP C7XX9*#X9XQ# ;PPO(h C"PP C; OPO@(h C" L<  PPO L<CO#X9XXX9#IX9XXX9(#X9XXIX9A#X9XXX9Hli)car|#X9XXX9#IX9XXX9n#X9XXIX9#X9XXX9#X9XXX9#IX9XXX9ssus#X9XXIX9U#IX9XXX9 7PP(P CPO C7#X9XXIX9#X9XXX9XX9*!#X9X3# 7PP(P C"PP C7#X9XXX9#IX9XXX9IXIX9*#I##X9X# 7PP(P C"PP C7XX9 ;PPO(P C"PP C;#X9X# RPOC(P C" L<  PPO LlCR#X9XXX94#IX9XXX9(#X9XXIX9#X9XXX9Hli)cr|#X9XXX9#IX9XXX9n#X9XXIX9,#X9XXX9#X9XXX9o#IX9XXX9ssus#X9XXIX9#XX9I :PP+8CPO C:*! :PP+8C"PP C: :PP+8C"PP C:#I##X9X# >PPO+8 C"PP C>XX9*#X9X # QPOO@(8!C" Ll  PPO LlCQHa(l7cr)|n#X9XXX9 #X9XXX9#X9XXX9N #X9XXX9ssus 9PPO(P"CPOO C9XX9*!#X9X # 9PPO(P#C"PPO C9 9PPO(P$C"PPO C9 ;PPOO(P%C"PPO C;XX9*#X9X #8.,P&C" Ll   PPOO 8Heretoo,leavingasyllableunparsedisfatal,duetothedominanceof#X9XX9 #X9XX9Parse#X9XX9 #^X!XX9%#X9X!X^X #X9XXX9Ԁovertheother ' constraints.Afifthcandidate,notshowninthistableau,inwhichallofthepretonicsyllablesaregroupedintoasinglefoot,wouldviolateFtBin,sincesuchafootisneithermoraicallynorsyllabically ) binary(moreonthisbelow).  Inthisenvironment,a%obehavesinthesamewayasa%s:'<0  +  (25)   bc tria @c t;ber x trnsic cAg ntionpo thg mticni md vrsion X- ЀThisisasexpected,sincethereisnothinginthisanalysisthatdifferentiates%oand%sinrelationto 4 / Parse#X9XXX9Z #^X!XXX9%#X9X!X^X#X9XXX9.Andaswewillseeinthenextsubsection,pretoniclongvowelsarealsostressedinthis  !p0 environment,andstresslesswhenprecededbyasinglesyllable.1.3.2Pretoniclongvowels #L3 Wehaveyetexaminethestressingofpretonicopensyllables.Syllableswithunderlyinglongvowelsforthemostpartpatternwith%s's.HalleandVergnaud(1987:240)observethatlongvowelsusually %$!5 retaintheirlengthandarestressedininitialposition(26a),butsurfaceasstresslessandreducedmedially,afterbothheavyandlightsyllables(26b).  (26)  a. 0 ` pr7 vtion vA ction c7 ttion  jction gr dtion)$9` (#` (#   b.0 ` d pri vtion7n vo ctionx ci ttionr ve ltiond gra dtiont*%:` (#` (#   d+&; Thereare,however,noreportedmonomorphemicwordsthathaveapretoniclongvowelinthesame positionasthestressed%sinHli cr nssus.Giventhesmallnumberofunderivedwordswith  trisyllabicpretonicstrings,itisdifficulttoknowifthisisanaccidentalgap.Inderivedwords,atleast,wedofindsuchpretoniclongvowels(seefurther3,andAppendixB):c G  5       x  (27)   rtro gr dtionc7vi l7 ztionstndar d7 ztionpra s7 t;logy P  ЀToaccountforthepatterningoflongvowelswith%s's,theconstraintrequiringthepreservationof , | inputvowellengthintheoutputformcansimplybeplacedinthesamepositioninthehierarchyas*SonNuc:atthebottom.FollowingMcCarthy(1995)andUrbanczyk(1995),aformulationofthis  T  constraintintermsofMcCarthyandPrince's(1995)CorrespondenceTheoryoffaithfulnessisprovidedin(28). (28)   WeightIdent     0 ` Ifisbimoraic,thenf()mustbebimoraic, ` (#` (#    ` wherefisthecorrespondencerelationbetweeninputandoutput   ЀThisconstraintstatesthatifaninputvowelisbimoraic,thenthecorrespondingvowelintheoutputmustalsobebimoraic.Inthefollowingtableaux,vowellengthisindicatedbysubscriptedmoras.Whenthelongvoweledsyllableisprecededbyanothersyllable,theyareparsedtogether,soastosatisfyStressWell.ThesatisfactionofWeighttoStressbytheoptimalcandidateisevenmore ,| obvious#X9XXX9#X9XXX9thaninparallel%scases(e.g.Pnn syl vnia),#X9XXX9#X9XXX9Ԁsincethevowelsurfacesasshortandreduced. h ThatvowellengthislostinthisenvironmentindicatesthatWeightIdent,like*SonNuc,is X dominatedbyStressWell: D Ѐ (29)StressWellNWeightIdent   *de ddSB@B@erB@rB@r@ab(#(#,SB@,erB@,rB@,r@+  3OOO"  llC3Input: $t deprivation 9OPO(`COOO C9Weight $t ! $t  $t !toStress 9OPO(`C"OPO C9Stress $t Well ;OPOO(` C"OPO C;Weight $t! ! $t!  $t! !Ident OPO@(`"C" ll  OPOO L<CO#X9XXX9#IX9XXX9(depri#X9XXIX9x!#X9XXX9#X9XXX9!#IX9XXX9)|vtion#X9XXIX9 "#IX9XXX9IXIX93#IX9XIu"##X9XXIX9V"#X9XXX9 7PP(#CPO C7 7PP($C"PP C7 ;PPO(%C"PP C;XX9*#X9X## OPO@(&C" L<  PPO L<CO#X9XXX9"#IX9XXX9(depri#X9XXIX9[$#X9XXX9#X9XXX9$#IX9XXX9)|vtion#X9XXIX9$#X9XXX9 7PP(!'CPO C7XX9*!#X9X%# 7PP(!(C"PP C7XX9 ;PPO(!)C"PP C;#X9X&# QPOO@(!*C" L<  PPO LlCQ#X9XXX9=%#IX9XXX9(#X9XXIX9'#X9XXX9de)(pri)|vtionXX9#C'#I 9PPO(#+CPOO C9 9PPO(#,C"PPO C9#I'##X9X'#XX9*!#X9X(# ;PPOO(#-C"PPO C;8.,#.C" Ll   PPOO 8Again,thestressingofasinglepretonicsyllableisduetotherequirementthatitbeparsed(Parse#X9XX9u(#^X!XX9%)#X9X!X^X)#X9XXX9, <&!/ ratherthantheneedforheavysyllablestobestressed(WeighttoStress),orforvowellengthto @'"0 bepreserved(WeightIdent): ,(|#1 Ї & t (30)TableauforV:   *fi ddSB@erB@rB@r@de(#(#,'B@,9B@,erB@,rB@,r@+  3OOO"  llC3Input: X privation 9OPO(DCOOO C9Parse X ! X  X  X !#X9XX9F*#^X!XX9%#X9X!X^X-#X9XXX9 9OPO(DC"OPO C9Weight X ! X  X  X !toStress 9OPO(DC"OPO C9Stress X Well ;OPOO(D C"OPO C;Weight X  ! X   X   X  !Ident OPO@(D C" ll  OPOO L<CO#X9XXX9-#IX9XXX9(pr7#X9XXIX9A0#X9XXX9#X9XXX90#IX9XXX9)|vtion#X9XXIX90#IX9XXX9IXIX93#IX9XIC1##X9XXIX9$1#IX9XXX9 7PP( CPO C7#X9XXIX91#X9XXX9 7PP( C"PP C7 7PP(C"PP C7XX9*#X9X2# ;PPO(C"PP C; OPO@(C" L<  PPO L<COpri|vtion#X9XXX9%2#IX9XXX9 7PP( CPO C7#X9XXIX93#X9XXX9XX9*!#X9XR4# 7PP( C"PP C7XX9*#X9X4# 7PP( C"PP C7XX9 ;PPO( C"PP C;#X9XH5# QPOO@( C" L<  PPO LlCQ#X9XXX934#IX9XXX9pri#X9XXIX996#X9XXX9|#X9XXX9~6#IX9XXX9vtionIXIX9Ԁ#IX9XI6#IXIX9 9PPO( CPOO C9*! 9PPO( C"PPO C9 9PPO( C"PPO C9#I6##X9X37# ;PPOO( C"PPO C;XX9*#X9X8#8., C" Ll   PPOO 8't 6+1.3.3Pretoniclightsyllablesandleftalignment 8  Sofar,thedominanceofParse#X9XX9#8#^X!XX9%#X9X!X^X9#X9XXX9Ԁhasplayedacentralroleintheanalysis.Itcausesboththelackof $t  stressonobstruent-finalsyllableswhentheyareprecededbylightsyllables,andthepresenceofstressonlonepretonicheavies.#X9XXX9 :#X9XXX9ԀEventhisrobustlysatisfiedconstraintisnotuniformlysatisfied,however. d  Singlelightpretonicsyllables,asexemplifiedbythewordsin(31),arealmostalwaysunstressed,andhenceunparsed,undernowstandardrepresentationalassumptions(thoughcf.AppendicesAandC). (31) 0  a. l  ` 0 (#(# ba nna A mrica te rrfic ce rmic Fe llni la g;on go rlla Ja mica cre vsse" (# (#   b. (ll)l 0 Tta ma g;uchibra ca dbraKla ma z;oW7nne pe sukeeWpa ko nta # dli ca tssenLAlla pa l;oza$ (# (# SatisfactionofParse#X9XX9);#^X!XX9%#X9X!X^X>#X9XXX9intheseinstancesleadstoaviolationofFootBin,astheresultingfoot & consistsofasinglemonomoraicsyllable.l G  6      ׀WithFootBinNParse#X9XX9>#^X!XX9%#X9X!X^X4@#X9XXX9,thesesyllablesremainunparsed: '  &  (32)FootBinNParse#X9XX9}@#^X!XX9%#X9X!X^X(A#X9XXX9  t) * d,d'B@9B@erB@rB@r@fi(#(#,B@,B@,erB@+  3OOO" x* llC3Input: 4+ banana 9OPO( ,COOO C9FootBin ;OPOO(4-C"OPO C;Parse#X9XX9qA#^X!XX9%#X9X!X^XC#X9XXX9 OPO@(4.C" ll  OPOO L<CO#X9XXX9C#IX9XXX9ba|nna#X9XXIX9zD#IX9XXX9IXIX93#IX9XID##X9XXIX9D#IX9XXX9 7PP(t /CPO C7#X9XXIX9KE#X9XXX9 ;PPO(t 0C"PP C;XX9*#X9XF# QPOO@(t 1C" L<  PPO LlCQ#X9XXX9E#IX9XXX9(ba)|nnaIXIX9Ԁ#IX9XIF#IXIX9 9PPO(\"2CPOO C9*! ;PPOO(\"3C"PPO C;#IF##X9X@G#8.,\"4C" Ll   PPOO 8' d!@YettobeaccountedforistheinitialstressinTtamag;uchiandtheotherwordsin(31b).This %d 5  disruptionoftheusualpatternofalternatingstressisascribedbyMcCarthyandPrince(1993b)to &P!6 oneofafamilyofAlignconstraints,whichinthiscaseforcesthealignmentoftheleftedgeoftheprosodicwordwiththeleftedgeofafoot: (33) Align(PrWd,L,Ft,L)AlignLeft t   'Aligntheleftedgeoftheprosodicwordwiththeleftedgeofafoot.'Whentheinitialstringismadeupofthreelightsyllables,FtBindemandsthatoneofthemmust <  remainunparsed.ThisisbecauseFtBinstatessimplythatafootmustbebinary,sothatconstituents ( x mustbenotonlyminimallybinaryateitherthesyllabicormoraiclevel,butmaximallybinaryatoneoftheselevelsaswell(P&S1993:47thisrestatestheprincipleofStrictBinarityoffootsizeproposedbyPrince1985;seealsoIt=andMester1992).ThejobofAlignLeftistoensurethatthe  <  unparsedsyllableisnottheinitialone: (34)LeftAlignment    *s d,dB@B@erB@(#(#s,`@,9@ @,0@,@@@+  /OOO   +@C/Input: ,  Tatamagouchi ;OPOP(COOO C;FtBin 9OO(, C"OPOP C9Parse#X9XXX9G#IX9XXX9%#X9XXIX9]O#IX9XXX9 ;OPOO(, C"OO C;Align ,  ! ,   ,  !Left OPO@(C" @0 + OPOO  CO(Ttama)|g;uchi ;PPP(TCPO C;IXIX9*!#IX9XIP# 5P(TC"PPP C5 ;PPO(TC"P C; OPO@(TC"  PPO b CO#X9XXIX9O#IX9XXX9Ta(tma)|g;uchi#X9XXIX9R#X9XXX9 ;PPP(CPO C; 5P(C"PPP C5XX9*!#X9XR# ;PPO(C"P C;XX9*#X9X|S# MPOO<(C" b PPO  CM(Tta)ma|g;uchi#X9XXX9mR#IX9XXX9IXIX93#IX9XIaT##X9XXIX9BT#X9XXX9 ;PPOP(TCPOO C; 9PO(TC"PPOP C9XX9*! ;PPOO(TC"PO C;#X9X]U#8.,T C" 0   PPOO 8ThistableaushowsthattherankingofAlignLeftbeneathParse#X9XXX9T#IX9XXX9%#X9XXIX9V#X9XXX9Ԁiscompatiblewiththeinitial ! dactyldata.Toseewhythisrankingisnecessary#X9XXX9V#IX9XXX9,wemustconsiderotherdata,sinceleftalignment l" inTatamagouchitypewordswillobtainunderanyrankingof#X9XXIX9rW#X9XXX9thesetwoconstraints,asinspectionof X# theabovetableaushouldreveal.ThedatathatmotivatealowrankforAlignLeftarethosein(35). D$ Hereweseethatifthesecondsyllableinaternarypretonicstringisheavy,it,ratherthantheinitialsyllable,isstressed.Whiletherearefewexamplesofthispattern,itappearstobeexceptionless,andnativespeakerintuitionsarestrongontheunacceptabilityofinitialstressforthesewords: (35)   l(hl) ` MonAngahlaValncinnesgerAntophliaBelchistn  0) Sinceevena%soralongvowelinthesecondsyllableinhibitsleftalignment,AlignLeftmustbe " + rankedatthebottomofthehierarchy.ThetableauforMonongahelashowsthattherankingbetween #,  AlignLeftand*SonNuciscrucial: $- & <  (36)#X9XX9?X#IX9XX9*SonNuc#X9XXIX9y\#IX9XXX9Ԁ#X9XXIX9\#X9XXX9N#X9XXX9]#IX9XXX9ԀAlignLeft#X9XXIX9N]#X9XXX9:   * dd`@9@ @0@@@@(#(#,`@,9@ @,0@,@@,:@@,@@@+  /OOO  +@C/#X9XXX9]#X9XXX9Input:  Monongahela#X9XXX9[_#X9XXX9 ;OPOP(COOO C;FtBin 9OO(C"OPOP C9Parse#X9XXX9_#IX9XXX9%#X9XXIX9`#IX9XXX9 9OPO(C"OO C9Weight  !    !toStress 9OPO(C"OPO C9*Son   Nuc ;OPOO( C"OPO C;Align   !      !Left OPO@( C" @0 + OPOO  CO(MAnnga)|hla ;PPP(@ CPO C;IXIX9*!#IX9XIc# 5P(@C"PPP C5 7PP(@C"P C7 7PP(@C"PP C7 ;PPO(@C"PP C; OPO@(@C"  PPO b CO#X9XXIX9`#IX9XXX9(MAnon)ga|hla#X9XXIX9d#X9XXX9 ;PPP(p CPO C; 5P(p C"PPP C5XX9*#X9Xe# 7PP(p C"P C7XX9*!#X9Xf# 7PP(p C"PP C7 ;PPO(p C"PP C; OPO@(p C" b PPO  CO#X9XXX9e#IX9XXX9(MAnn)ga|hla#X9XXIX97g#X9XXX9 ;PPP( @CPO C; 5P( @C"PPP C5XX9* 7PP( @C"P C7 7PP( @C"PP C7*! ;PPO( @C"PP C;#X9Xh# MPOO<( @C"  PPO  CMMo(nAnga)|hla#X9XXX9g#IX9XXX9IXIX93#IX9XIi##X9XXIX9i#X9XXX9 ;PPOP(p CPOO C; 9PO(p  C"PPOP C9XX9* 9PPO(p !C"PO C9 9PPO(p "C"PPO C9 ;PPOO(p #C"PPO C;*#X9Xj#8.,p $C" 0   PPOO 8'< 4\ThetopmostcandidateisimmediatelyruledoutbyFtBin.Theremainingcandidatesallviolate < % #X9XX9j#X9XX9Parse#X9XXX9l#IX9XXX9%#X9XXIX9l#X9XXX9,soevaluationispassedontothenextconstraint,WeighttoStress,whichrulesoutthe ( & parsing (lh)l .With#X9XX9m#IX9XX9*SonNuc#X9XXIX9m#X9XXX9Ԁ#X9XXX9 rule)hasbeendealtwithhereintermsofconstraintrankingandminimalviolation,andcontrastingthisanalysiswithonebasedonfullsatisfaction.  Themostobvioussourceofnonuniformitycomesfromthedifferent#X9XXIX9zx#X9XXX9behaviorsofheavy )%A syllablesininitial,andmedialposition#X9XXX9|#X9XXX9.Ininitialposition,heavysyllablesareregularlystressed(e.g. *&B #X9XXX9}#IX9XXX9S)yWdd0l#X9XXIX9}#IX9XXX9v#X9XXIX9~#IX9XXX9#X9XXIX9R~#IX9XXX9nia,#X9XXIX9~#X9XXX9pr7vtion)#X9XXX9~#X9XXX9;inbetweeninitialandtonicposition,#X9XXX92#X9XXX9abstractingfrom#X9XXX9#X9XXX9Ԁ#X9XXX9#X9XXX9%o'sandthe#X9XXX92#X9XXX9ԀArabrule#X9XXX9#X9XXX9, +&C theyareusuallyunstressed#X9XXX9Հ#X9XXX9Ԁ#X9XXX9I#X9XXX9(e.g.Pnnsy#X9XXX9#IX9XXX9l#X9XXIX9݁#IX9XXX9v#X9XXIX9 #IX9XXX9#X9XXIX9f#IX9XXX9nia,#X9XXIX9#X9XXX9dprivtion).#X9XXX9#X9XXX9Ԁ#X9XXX9E#X9XXX9Herethispositionaldifferenceisderived  fromtherankingof#X9XX9#IX9XX9Parse%#X9XXIX9!#X9XXX9Ԁ #X9XXX9p#X9XXX9N #X9XXX9#X9XXX9ԀStressWell.StressWelldemandspretonicstresslessness,butis  overruledwhenitwouldleadtononexhaustiveparsing.  HalleandVergnaud(1987:240)explicitlydiscussthelongvowelcases.Theydoaccountforthemwithoutmakingtheirruleofstressdeletionrefertotheword-medialenvironment;thisisdone,however,bypositingaruleofshorteningthatappliesonlymedially.Thus,theexistenceofnonuniformityisaccountedforbystipulatingitintheformulationofarule.  Moregenerally,theabsenceofmedialheavysyllablestressisattributedtothequantityinsensitivenatureofsecondarystressassignment,asexpressedinStrongRetraction#X9XXX9#X9XXX9(Hayes1982)#X9XXX9#X9XXX9,  d andthesetofrulesHalleandVergnaud(1987)dub"theAlternator".Butaswehaveseen,inmanyenvironments,pretonicheavysyllablesareregularlystressed.Infact,theonlyenvironmentinwhichtheyareregularlyunstressedispreciselytheoneunderdiscussion:whentheyareadjacenttoboththe#X9XXX9\#X9XXX9 (  initialandtonicsyllables#X9XXX9#X9XXX9.Toproducetheperhapsmoreusualcasesofheavysyllablestress,Hayes   (1982)invokesrulesofpreandpoststressdestressingthatarequantitysensitive(followinginsomerespectsHalle1973andLibermanandPrince1977).Stressisassignedequallytobnnaand   bndna,toTtmagouchiandMAnAngahla,butremovedfromonlythelightsyllables.X G  7      ׀Herewe   havethetackofpositingaderivationalstageduringwhichtheprincipledoesnotapply,andthenturningitbackontotrimbackthemisplacedstresses.#X9XXX9U#X9XXX9Withinaderivationaltheory,suchamoveis ` bynomeansillegitimate;itisactuallytheprimemeansbywhichbothsimpledescription,anddescriptivegapscanbeavoidedindealingwithnonuniformity.However,wecansurelycountitasanadvanceifthefactscanbeaccountedformoredirectly,andatthesametimereducedtobasicuniversalprinciples.Byassumingminimalviolation,insteadoffullsatisfaction,justthissortofadvanceintheanalysisofthenonuniformweighteffectsinEnglishsecondarystressispossible.  #X9XXX9K#IX9XXX9 8 #X9XXIX9ŏ# X9XXX9UKUS.,2.Weighttoprimarystress AnevenmoreobvioussourceofnonuniformweighteffectsinEnglishcomesfromthedistinction  betweenprimarystressplacementandsecondarystressplacement.Acompleteanalysisofmainstressplacement,dealingwithcomplexitiessuchasstressretraction,andlexicallyandmorphologicallydeterminedstressing,isbeyondthescopeofthispaper(seeKager1989andBurzio1994forcomprehensivereferencesanddiscussion).However,inthissection,Iwilldiscussthebasicpatternofmainstressplacementonnounsoutlinedin(1),becausewithWeighttoStress,andrelated H! constraints,suchasWeightIdentand*SonNuc,rankedsolowinthehierarchy,onemightwonder 4"  howquantitysensitiveprimarystressplacementistobeexplained.  Toseewhythiscouldbeaproblem,recallthatinwordssuchasagnda,syn;psis,andar;ma,  $\" primarystressfallsontheheavypenultimatesyllable,butwhenthepenultislight,asinCnada,stress $H # isantepenultimate.ThatstressisantepenultimateinCanadaisusuallyexplainedbyaruleof %4!$ Extrametricality(Hayes1982),oraconstraintofNonfinality(P&S1993)thatdisallowsthefooting  ofthefinalsyllable,sothattheparsing(C#X9XXX9#X9XXX9US.,UK.,>UKUS.,#X9XXX9 #US.,UK.,DX9XXX9UKUS.,na)daoccursinsteadofCa(n#X9XXX9#X9XXX9US.,UK.,UKUS.,#X9XXX9 #US.,UK.,CUKUS.,X9XXX9da)(cf.Burzio1994).The  apparentdilemmacanbesummarizedinthefollowingquestion:IfParse-%isrankedabove  #X9XX9#X9XX9US.,UK.,UKUS.,Weight-toStress#X9XXX9#US.,UK.,˘X9XXX9UKUS.,Ԁandtheotherweightrelatedconstraints,thenwhyaren't#X9XXX9-#US.,UK.,LX9XXX9UKUS.,a(gn)da,sy(n;p)sis, t anda(r;)ma#X9XXX9ٙ#US.,UK.,UKUS.,X9XXX9Ԁmoreexhaustivelyparsed,andfootedlike#X9XXX9#X9XXX9US.,UK.,UKUS.,(Cna)da#X9XXX9#US.,UK.,LX9XXX9UKUS.,? `   #X9XXX9#X9XXX9US.,UK.,›UKUS.,McCarthyandPrince's(1993b)#X9XXX9#US.,UK.,VX9XXX9UKUS.,Ԁformulationoftheconstraintresponsibleformainstress L  placement#X9XXX9œ#X9XXX9US.,UK.,UKUS.,inEnglishallowsasimpleanswertothisquestion.Itisstatedasfollows: 8   (38) Align(PrWd,R,Head(PrWd),R)AlignHead  `   'AligntherightedgeoftheProsodicWordwith   ` therightedgeoftheheadoftheProsodicWord'#X9XXX9o#X9XXX9US.,UK.,UKUS.,AsinP&S1993,headshipistakentobetransitive,sothat'headoftheProsodicWord'isfulfillednot   onlybythefootthatbearsmainstress,butalsobythesyllablethatistheheadofthatfoot.#X9XXX9#US.,UK.,SX9XXX9UKUS.,If   #X9XX9#X9XX9US.,UK.,UKUS.,Align-Head#X9XXX9#US.,UK.,'UKUS.,X9XXX9Ԁwereundominated,theheadsyllableoftheProsodicWordwouldbetherightmost   syllableintheprosodicword,andmainstresswouldbefinal.ThatEnglishdoesnothavefinalmainstressindicatesthatthisconstraintisdominated. t   Nonfinality,formulatedas'theheadoftheProsodicWordmustnotbefinal'(P&S1993:52; ` cf.Hung1994,Buckley1995)isindirectcompetitionwith#X9XX9#X9XX9US.,UK.,UKUS.,AlignHead#X9XXX9#US.,UK.,X9XXX9UKUS.,,andwhenrankedabove L it,forcesmainstress,andthemainstressedfoot,offthefinalsyllable.Nonetheless,AlignHead 8 continuestodemandthatthemainstressedsyllablebeascloseaspossibletotherightedgeoftheword,witheachsyllableinterveningbetweentheheadsyllableandtheedgecountingasaviolation(P&S1993:57,McCarthyandPrince1993b,Prince1993).Therefore,theminimalviolationofAlign-Head,whichsatisfiesNonfinality,istohavemainstressonthepenultimatesyllable.The 8 ranking#X9XX9E#X9XX9US.,UK.,dNonfinalityN#X9XXX9#X9XXX9UKUS.,AlignHead#X9XXX9d#US.,UK.,X9XXX9UKUS.,Ԁisallthatisneededtoplacemainstressonthecorrectsyllable $ ofagenda,andtheotherwordswithbimoraicpenults:   #X9XXX9#X9XXX9US.,UK.,(39)NonfinalityN#X9XXX9#X9XX9UKUS.,AlignHead#X9XXX9]#US.,UK.,|X9XXX9   * dd`@9@ @0@@@:@@@@@(#(#,B@,B@,erB@+  3OOO"  llC3Input: @ agenda 9OPO(,COOO C9Nonfinality#X9XXX9ة#X9XXX9 ;OPOO(@C"OPO C;#X9XX9ث#X9XX9UKUS.,Align @  ! @   @  !Head#X9XXX9X#US.,UK.,wX9XXX9 OPO@(,!C" ll  OPOO L<CO#X9XXX9#X9XXX9a(g#X9XXX9#X9XXX9UKUS.,#X9XXX9#X9XXX9US.,UK.,n)daXX93#X9X# 7PP(!"CPO C7#X9XXX9G#X9XXX9 ;PPO(!#C"PP C;XX9*#X9X~# JPO;(!$C" L<  PPO L<CJ#X9XXX9$#X9XXX9(g#X9XXX9.#X9XXX9UKUS.,e#X9XXX9s#X9XXX9US.,UK.,n)da#X9XXX9Ȱ#X9XXX9 7PP(h#%CPO C7 ;PPO(h#&C"PP C;#X9XXX9(#X9XXX9XX9**#X9X# LPOO;#h#'C" L< PPO LlCL#X9XXX9ܱ#X9XXX9(gen)(d)XX9Ԁ#X9Xײ#XX9 9PPO(P% (CPOO C9*! ;PPOO(P% )C"PPO C;###X9X#8.,P% *C" Ll   PPOO 8WithbutasingleviolationofAlignHead,thefirstcandidate,withpenultimatestress,isoptimal. (X#+ Asnotedabove,acandidatelikethesecondone,withinitialstress,faresbetterintermsofexhaustivitythantheoptimalonedoes,asitleavesonlythefinalsyllableunparsed.Thisshowsthat AlignHeaddominates#X9XX9ӳ#X9XX9UKUS.,Parse-%.#X9XXX9޵#US.,UK.,X9XXX9Ԁ *&.   Withthepenultimatestressonagendaexplained,#X9XXX9[#X9XXX9UKUS.,ouroriginalquestionisreversed.Whydoes   Canadahaveantepenultimatestress?Theansweristhatpenultimatestresswouldviolateconstraints  whosehighrankwehavealreadyestablished:eitherFtBin,orTroch#X9XX9#US.,UK.,X9XX9UKUS.,. G  8      ׀Iftheseconstraints,along  withNonfinality,outrankAlignHead,antepenultimatestressisoptimal: t Ѐ `  #X9XXX9x#X9XXX9US.,UK.,(40)FtBin,Troch,NonfinalityN#X9XXX9ܹ#X9XX9UKUS.,AlignHead#X9XXX9k#US.,UK.,X9XXX9  L  *s d,dB@B@erB@(#(#s,B@,B@,B@,FB@,rB@+  3OOO" 8  llC3Input:   Canada 9OPO( COOO C9#X9XX9#X9XX9FtBin#X9XX9ۼ#X9XXX9 9OHO(  C"OPO C9#X9XXX9%#X9XX9Troch#X9XXX9#X9XXX9 9OHO(  C"OHO C9#X9XXX9#X9XXX9Nonfinality ;OPOO(  C"OHO C;#X9XX9h#X9XX9UKUS.,AlignHead#X9XXX9#US.,UK.,X9XXX9 OPO@(  C" ll  OPOO L<CO#X9XXX9t#X9XXX9(C#X9XXX9#X9XXX9#X9XXX9I#X9XXX9na)da#X9XXX9#X9XXX9XX93#X9X# 7PP(4  CPO C7 7PH(4 C"PP C7#X9XXX9#X9XXX9 7PH(4 C"PH C7#X9XXX9#X9XXX9 ;PPO(4 C"PH C;XX9**#X9X##X9XXX9B#X9XXX9 JPO;(4 C" L<  PPO L<CJCa(n#X9XXX9#X9XXX9#X9XXX9#X9XXX9da) 7PP(l CPO C7 7PH(l C"PP C7 7PH(l C"PH C7#X9XXX9#X9XXX9XX9*!###X9X#X9XXX9 ;PPO(l C"PH C;XX9*#X9X# JPO;#l C" L< PPO L<CJ#X9XXX9G#X9XXX9Ca(n#X9XXX9Q#X9XXX9#X9XXX9#X9XXX9)da#X9XXX9#X9XXX9 7PP(T CPO C7XX9*!#X9Xu# 7PH(T C"PP C7#X9XXX9#X9XXX9 7PH(T C"PH C7#X9XXX9#X9XXX9 ;PPO(T C"PH C;#X9XXX9#X9XXX9XX9*#X9X(# QPOO@(T C" L<  PPO LlCQ#X9XXX9 #X9XXX9(Ca#X9XXX9#X9XXX9n#X9XXX9$#X9XXX9#X9XXX9g#X9XXX9)daXX9Ԁ#X9X#XX9 9PPO(<CPOO C9 9PHO(<C"PPO C9*! 9PHO(<C"PHO C9 ;PPOO(<C"PHO C;###X9X#XX9*#X9X|#8.,< C" Ll   PPOO 8#X9XXX9:#X9XXX9UKUS.,"Quantitysensitive"primarystressplacementisthusachievedbysimplyplacingtheindependently ! necessaryconstraintsofNonfinalityandAlignHeadinthehierarchyestablishedforsecondary " stressplacement.[T G  9      ׀Crucially,#X9XX9#X9XX9US.,UK.,:UKUS.,AlignHead#X9XXX9#US.,UK.,X9XXX9UKUS.,ԀmustdominateParse%,sothatprimarystressdoesnot |# shifttothelefttoincorporateastraysyllable(i.e.#X9XXX9m#X9XXX9US.,UK.,3#X9XXX9N#UKUS.,X9XXX9a(gn)da,*(#X9XXX9#US.,UK.,X9XXX9#X9XXX9(#X9XXX9UKUS.,gen)da),butmustitselfbe h$ dominatedby#X9XX9k#X9XX9US.,UK.,FtBin,Troch,andNonfinality#X9XXX9#UKUS.,X9XXX9,soastoensureantepenultimatestresswithalight T% penult.Thisresultsinthehierarchyin(41).  ,|'  #X9XXX9#X9XXX9US.,UK.,&   (41)   FtBin,Troch,#X9XX9w#X9XX9Nonfinality  #X9XX9#X9XX9    󀀀 `  |   h   #X9XX9o#X9XX9UKUS.,   ` AlignHead#X9XXX9%#X9XXX9US.,UK.,D     `  |  *ObsNuc x    ` Parse%  / d    `  |  / P     ` WeighttoStress <     `  |   ` StressWell  d    ` 󀀀 |  P     ` *SonNuc,WeightIdent  <     `  |   ` AlignLeft#X9XXX9#X9XXX9UKUS.,Ԁ#X9XXX9#US.,UK.,X9XXX9UKUS.,   #X9XXX9#X9XXX9' #X9XXX9^#X9XXX9US.,UK.,Therearetwoespeciallysalientpointsaboutthenewadditionstothehierarchy.First,itcanbeseen   thatbytransitivity,AlignHeaddominatesAlignLeft.ThisisinagreementwithM&P(1993b)'s   analysisofthepreferenceforrightmostmainstressplacementoverleftalignment(asinA(mri)ca). t Second,thatquantitysensitivemainstressplacementcanbeattributedtothejointeffectsofFtBin ` andAlignHead,ratherthantoWeighttoStress,underscoresanimportantresultoftheprevious L section!theinfluenceofsyllableweightonstresscanmanifestitselfeitherdirectly,throughWeight-toStress,ormoreindirectly,throughFtBin.Therefore,weightsensitivitycannotserve $t asadiagnosticofanactiveWeighttoStressconstraint,andthelowrankingof ` Weight-to-Stressdemandedbysecondarystressplacementdoesnotimpedeanaccountofprimary L stressassignment.ThisintegrationofprimaryandsecondarystressassignmentcanbecontrastedwiththeanalysisinHalleandVergnaud1987:228,inwhichanentirelyseparaterulestratumissetupforsecondarystressassignment,whichdiffersfromprimarystressassignmentalmostexclusivelyinitslackingtheweighttostressrule(theAccentRule)thatcausesquantitysensitiveprimarystressplacement.  Inthenextsection,weturntoexceptionalheavysyllablestressandstresslessnesscausedbytheidiosyncraciesoftheEnglishlexicon,andbytheinfluenceofstemstressonderivedwords.There,onceagain,ahighrankingFtBinconstraintplaysacentralrole,thistimeincontrollingtheextentto \  whichlexicalandstemstresscandisturbtheusualpatternsofsecondarystress.Inaddition,therelativerankingsofotherconstraintsinthehierarchy,suchas#X9XX9#X9XX9AlignHead#X9XXX9#X9XXX9,StressWell,and 4"  Align-Left,findsfurthersupport.  #p!  3.Specialsecondarystress  %4!$ Thegeneralizationsaboutweighttosecondarystressoutlinedin1areupsetbytwosetsofwords:lexicalexceptions,andderivedwordsthatretainstressfromtheirstems.Inthissection,Iargueforatreatmentofthesespecialcasesintermsofexactlytwoformalmechanisms:prosodicfaithfulness,andlexicallyspecificranking.Inthisdomainaswell,minimalviolationallowsaprincipledaccountofnonuniformitythathasbeenlackingfromprioranalyses.  l+&* &   3.1Lexicaland'cyclic'stressasprosodicfaithfulness  AclassicexampleofcyclicstresspreservationiscAndnstion,whichconflictswiththeclearly  productivetendencyforapretonic%stobeunstressedandreducedinthisenvironment(see1.2).  S'inceChomskyandHalle1968,thepretonicstressincondensationhasgenerallybeenassumedto t beduetothestressonthecorrespondingsyllableofcondnse(cf.cAntemplte>cAntempltion). ` InChomskyandHalle'sanalysis,primarystressisassignedtocondenseonthefirstcycle,and L  preservedasasecondarystresswhenationisaddedonasubsequentcycle.Someadditional 8  examplesofstembasedexceptionslikecondensationareprovidedin(42). $ t  (42)   dvntgeousugmnttionuthntcitycAndmntion      L    AspointedoutbyHalleandKenstowicz(1991:460),parallellexicalexceptionsalsooccur: (43)   7ncnttion 7ncrntion h Astnttion  ch7mpnze   ЀSincetherearenoindependentstemsoftheformincnt,incrn,ostnt,orchimpn,thepretonic   stressinthesewordscannotbeduetocyclicity(thoughcf.Fidelholtz1967:7).HalleandKenstowicz(1991)drawattentiontothesecasesinproposingaradicallynoveltreatmentofcondensationlike d words:thattheytooaresimplylexicalexceptions,subjecttoalexicallyconditionedweighttostressrule.  However,usingalexicallyconditionedweighttostressruleforapparentcasesofstresspreservation,anddenyingthatthestresspatternofthestemplaysanyrole,leadstoamissedgeneralization.AsLibermanandPrince(1977:299)note,whilea%scanbeproductivelydestressed P inthisenvironment(e.g.inf;rm>7nformtion),therearenoinstancesofsuchasyllablebecoming < stressedinaderivedword.Thatis,therearenowordslikecAntempltethatbecomecAntmpltion, ( withastressedpretonicsyllable(comparergument>rgumnttion,discussedin1.2,andfurther  in3.2below).Thus,whilethepresenceofstressonthecorrespondingstemsyllabledoesnotensurestressinthisposition,alackofstressonthestemsyllabledoesguaranteestresslessness.ThisgeneralizationshowsthatcontraHalleandKenstowicz(1991),ChomskyandHalle(1968)wereinfactcorrectinassumingthatthestresspatternsofcondnseandcAntemplteinfluencethestressing t ofcAndnstionandcAntempltion. `    Theextremerarityofunderivedwordswithpretonicstresslikethosein(43),alongwithexistenceofderivedwordsthatproductivelydestress,butnonethatstress,syllablesinthisposition,demonstratesthatthispatternofsecondarystressisunproductive.Iwillnowadvanceananalysisoftheseinstancesoflexicaland"cyclic"stresspreservationthatreliesonasinglemechanismofprosodicfaithfulness,whichappliesbetweenlexicalandsurfaceforms,andbetweenmorphologicallyrelateditems(cf.McCarthy1995).Onebenefitofthisanalysisisthatitisconsistentwiththeunproductivenatureofthispattern.  Firstofall,Iassumethatthelexicalformofawordlikeincantation,orchimpanzee,includes '#& stressonthepretonicsyllable.Inorderforthisstresstobepreservedintheoutput,theremustbeafaithfulnessconstraintthatoutranksStressWell,sinceStressWellusuallyforcessuchsyllables )$( tobestressless(seethetableauin(19)above).McCarthy(1995)showsthatprosodicfaithfulnessconstraintscantakeatleasttwoformsintheCorrespondenceTheoryoffaithfulness,whichas p+&* expoundedinMcCarthyandPrince(1995),premisescorrespondencerelationshipsbetweensegmentsonly.Theedgeofthefootcanbekeptinthesameplacebyrequiringthecorrespondentofanyedgemostsegmenttobeedgemostitself(cf.Idsardi1992),orelsestresscanbekeptinthesameplacebyrequiringthecorrespondentofasegmentalheadofprosodiccategorytohavethesamerole(cf.HalleandVergnaud1987).Aseitherformulation,orindeed,anyofascoreofothers,wouldsufficeforpresentpurposes,Iwillassumetherelativelyinformal(44)(thoughseeAppendixAforsomeevidencethatEnglishrequiresaformulationslightlydifferentfromthosewhichMcCarthyproposes). (44)   StressIdent  `    ` Ifisstressed,thenf()mustbestressed  P     ` AsfortherelatedWeightIdentconstraintin(28),herefisthecorrespondencerelationbetween (  input(lexical)andoutput(surface)stringsofsegments(seeMcCarthyandPrince1995,McCarthy1995forformaldetails).TherankingStressIdentNStressWellleadstothepreservationof   underlyingstress,evenadjacenttothemainstress: (45)StressIdentNStressWell:Lexicalstresspreservation  t * d,dB@B@B@FB@rB@(#(#,rB@,JRr@,JRr@+  /OOO ` 20C/Input:  chimpnzee ;OPOP(|COOO C;StressIdent >OOO+C"OPOP C>StressWell OPO@(C" 00 2 OOO <<CO(ch7mpn)(ze) ;PPP($CPO C;XX9*!#X9X# >PO+$C"PPP C> QPOO@($C" <<  PO 5<<CQ(ch7m)(pn)(ze)XX93#X9X# ;PPOP(CPOO C; >POO+C"PPOP C>XX9*#X9Xl#8.,C" << 5  POO 8Withtheaddedassumptionthatacorrespondencerelationshipalsoexistsbetweenastemanditsderivative(Benua1995;McCarthy1995;cf.alsoBurzio1994,Steriade1994,Kenstowicztoappear),thenanunstressedpretonicsyllableincondensationalsoviolatesStressIdent,andStressIdent   \  NStressWellgeneratesstemstresspreservationaswellaslexicalstresspreservation: H  &  (46)StressIdentNStressWell:Stemstresspreservation   ! *s ddrB@JRr@JRr@(#(#s,rB@,|Rr@,JRr@+  /OOO  " 20C/Input:  <# condensation ;OPOP(!($COOO C;StressIdent >OOO+ <%C"OPOP C>StressWell OPO@( <&C" 00 2 OOO <<CO(cAndn)(s)tion ;PPP(0#'CPO C;XX9*!#X9X~# >PO+0#(C"PPP C> QPOO@(0#)C" <<  PO 5<<CQ(cAn)(dn)(s)tionXX93#X9Xg# ;PPOP(%X *CPOO C; >POO+%X +C"PPOP C>XX9*#X9X&#8.,%X ,C" << 5  POO 8'Becausestressonthesepretonicsyllablesisdrivenbyfaithfulnesstoprosodicstructureineitherthelexicon,orinthestemofaderivedword,stresswillnotemergeinthispositionwhenitisabsentunderlyingly,orinthestemofderivedword.ThisanalysisthuscapturesthesimilaritybetweenlexicalandstemstressbasedstressthatpromptedHalleandKenstowicz(1991)tousealexicallyspecificweighttostressruleforbothtypesofexceptionality,withoutsacrificingthegeneralizationthatwordslikecAntmpltiondonotexist.Forthedataconsideredthusfar,onecouldequallyprofferananalysis ,,|'2 intermsoftraditionalderivationalassignmentandpreservationofstress.However,inthenextfewsubsections,whichfocusonnonpreservation,or'unfaithfulness',theadvantageofformalizingstresspreservationintermsofaminimallyviolableconstraintwillbecomeclear:itallowsapreciseandconciseaccountofthecircumstancesunderwhichpreservationdoesanddoesnotoccur.3.1.1UnfaithfulnessI:LexicallySpecificRanking   L  Thefirstcaseofnonpreservationthatwewillexamineisdisplayedinwordslike7nformtion,thatdo 8  notpreservethestressoftheirstems.Toexplainthedestressingofthesesyllables,weneedawayofformalizingtheideathatStressIdentexertsagreaterinfluenceoncondensationthanitdoeson  ` information.Iwilldiscusstwopossibilities:morphologicalreanalysis,andlexicallyspecificranking,  L  andshowthatonlythelatterappearstobeadequate.  Anexplanationbasedonmorphologicalreanalysiswouldbethatinformationhasbeen $  (diachronically)reanalyzedasanindependentword,andnolongerstandsincorrespondencewith   inform(cf.ChomskyandHalle1968:112v G  10      ;H&V1987:251).StressIdentwouldnotapply,and   informationwouldbefreetoobeyStressWellinstead.     Thechiefflawofthisexplanationisthatitgeneratesregularizationthroughreanalysisinanotherenvironment,wherenoneactuallyoccurs.TransitivityofrankingentailsthatsinceStressWellisdominatedbyStressIdent,soisthelowerrankedAlignLeft.Theranking \ StressWellNAlignLeftproducesthepreferenceforstresspreservationoverleftalignmentthat H isexhibitedbythewordsin(47),inwhichthenonalignedleftmoststresscorrespondstoastressedsyllableinthestem(i.e.accrdit,imgine,orginal,etc.).d G  11        p  (47)   accr#X9XXX9?#IX9XXX9CEUS.,#X9XXIX9|#US.,CE.,dittionimgintionor7ginlitymed7cinlitydiv7sibltyphenAmen;logy H Thatstresspreservationisatworkherecanbeclearlyseeninthecontrastbetweencademcian,as $ derivedfromcadmic,andacdemcian,fromacdemy(Fidelholtz1967;Kager1987:170).    Thefactthatthisinstanceofstresspreservationfallsoutfromrankingsmotivatedonothergroundsprovidesawelcomeindicationthatouranalysisisontherighttrack.However,whatisof  interestatpresent,andremainstobeaccountedfor,isthecompletelackofderivedwordsthatvalueleftalignmentoverstresspreservation(Burzio1994).Ifcorrespondencecouldsimplybe'turnedoff'byreanalysis,thenonewouldexpecttofindthatjustaswordslikeinformationobeyStressWell  whenreanalyzed,wordslikeimaginationcouldbereanalyzedandleftaligned. t   Thisconundrumisavoidedifinsteadofmorphologicalreanalysisforinformation,weinvoke ` alexicallyspecificversionofStressIdentforcondensation. G  12      ׀Designatingthesetofwords, L  includingcondensation,andchimpanzee,thataresubjecttothelexicallyspecificversionof 8  StressIdentas'',thisconstraintcanbereferredtoasStressIdent.ByplacingStressIdent $ t aboveStressWell,andthegeneralversionofStressIdentbetweenStressWellandAlignLeft,  ` wegeneratelexicallybasedvariationinstresspreservationfortheinformation/condensationcases,  L  andstrictobediencetoStressIdentforimaginationandsimilarwords.Toshowwithmaximal  8  perspicuitythedifferenteffectsthisrankinghasonthesevarioussetsofwords,anexampleofeachisplacedinthesingletableauin(48): &  (48)StressIdentNStressWellNStressIdentNAlignLeft    IX9XXX9*s ddrB@|Rr@JRr@(#(#s,rB@,Rr@,kRr@,Rr@,Rr@+  /OOO   20C/Input:condensation  information  imagination ;OPOP(COOO C;Stress  Ident#X9XIX9#IX9XXX9 <OO+C"OPOP C<#X9XIX9/#Stress  WellIX9XXX9 ;OPOP(C"OO C;#X9XIX9#Stress  Ident IX9XXX9 >OOO+C"OPOP C>#X9XIX9|#AlignLeftIX9XX9 OPO@(C" 00 2 OOO <<CO(cAndn)(s)tion ;PPP(4CPO C;IXIX9*!#IX9XI # 8P+4C"PPP C8 ;PPP(4C"P C;IXIX9*#IX9XI!# >PO+4C"PPP C> QPOO@(4C" <<  PO 5<<CQ(cAn)(dn)(s)tionIXIX9Ԁ3#IX9XIh"# ;PPOP( CPOO C; <PO+  C"PPOP C<IXIX9*#IX9XI&## ;PPOP( !C"PO C; >POO+ "C"PPOP C> MO>( #C" << 5 POO <<CM(7nfr)(m)tion#X9XXIX9 #IX9XXX9IXIX93#IX9XI$##X9XXIX9d$#IX9XXX9 :P+$CO C: :P+%C"P C: :P+&C"P C:IXIX9*#IX9XI%# >PO+'C"P C> GPOO6#(C" << PO <<CG(7n)(fAr)(m)tion 9PPO(l)CPOO C9 9PPO(l*C"PPO C9IXIX9*!#IX9XI '# 9PPO(l+C"PPO C9 ;PPOO(l,C"PPO C; HO9#l-C" << PPOO <<CHi(mgi)(n)tion#X9XXIX9$#IX9XXX9IXIX93#IX9XI](##X9XXIX9>(#IX9XXX9 :P+D.CO C: :P+D/C"P C: :P+D0C"P C: >PO+D1C"P C>IXIX9*#IX9XI)# GPOO6#D2C" << PO <<CG(7ma)gi(n)tion 9PPO(l3CPOO C9IXIX9(*!)#IX9XI*# 9PPO(l4C"PPO C9 9PPO(l5C"PPO C9IXIX9*!#IX9XI^+# ;PPOO(l6C"PPO C;3)'l7C" <<  PPOO 3#X9XXIX9(#'8sAscondensationissubjecttoIX9XX9StressIdent#X9XIX9o,#,itsrankingaboveStressWellrendersIX9XXX9CEUS.,thestress !8 preservingcAndnstionoptimal#X9XXIX9,#US.,CE.,-,evenwiththeattendantStressWellviolation.When "9 IX9XX9StressIdent#X9XIX9-#doesnotapply,asininformation,therankingStressWellNStressIdentcreates |#: apreferenceforpretonicstresslessnessoverstresspreservation,leadingtothegrammaticalityofIX9XXX97#X9XXIX9:/#nformIX9XXX9CEUS.,#X9XXIX9/#US.,CE.,/tion.Finally,nomatterwhetherawordlikeimaginationistargetedbythelexicallyspecific T% < versionofStressIdentornot,stresspreservation,asinIX9XXX9imgin#X9XXIX90#IX9XXX9CEUS.,#X9XXIX90#US.,CE.,1IX9XXX9tion,#X9XXIX9j1#Ԁisalwaysmorehighlyvalued  thanleftalignment,asinIX9XXX97magin#X9XXIX92#IX9XXX9CEUS.,#X9XXIX9O2#US.,CE.,n2IX9XXX9tion,becauseofthedominanceofStressIdentoverAlignLeft.#X9XXIX92#Ԁ    TheadditiontothehierarchyoftheprosodicfaithfulnessconstraintStressIdent,anda  lexicallyspecificversionofthesameconstraint,allowsustocapturethefactthatpretonicstresspreservationissubjecttolexicalconditioning,whilestresspreservationonsyllablesnotadjacenttothemainstressoccurswithoutexception.Thiscaseofnonuniformityhasneverbeenexplainedbefore,exceptbydenyingthatIX9XXX9CEUS.,cAndnstion#X9XXIX905#US.,CE.,O5Ԁisanexampleofstresspreservation(Burzio1994:185). G  13      ׀ 8    Inthepresentanalysis,thelowerboundofpreservationisdeterminedbytherankingofStressIdentaboveAlignLeft:preservationisalwaysvaluedoverleftalignment.Thequestionto  ` beaddressednowiswhetherthereisanabsolutelimittofaithfulness.ContrarytoLibermanandPrince's(1977:286)claimthat"thereluctance...toobscuretheshapeofunfamiliarwords...caninhibitanyreductionprocessinEnglish",itturnsoutthatthereisanupperboundtostresspreservation,whichisprovidedbyaconstraintthatrestsatthetopofthehierarchyestablishedforregularstress:FootBinarity.   &  3.1.2UnfaithfulnessII:FootBinarityNStressIdent   InthevastmajorityofsituationsinwhichStressIdentconflictswithFtBin,FtBinalwaystriumphs. p '8Thiscanbeseenbothinthecompleteabsenceoflexicalstress,andintheconsistentfailureofstemstresstobepreserved,incertainenvironments.Inparticular,alightsyllableisneverstressedwhenitisthefinalmemberofbisyllabicortrisyllabicpretonicstring.LexicalstressneverturnsupontheunderlinedsyllableofwordslikeMAntebllo,orTtamag;uchi(Selkirk1984),andinderivedwords,  p thesesyllablesarealwaysdestressed(Kager1989).Correspondingtotheabsenceofwordslike*MAntblloarethefollowingalternations: H  (49)   phontic,phAnetcian;cAsmtic,cAsmetcian;path;logy,pthol;gical;specfic,spcifcity     telpathy,tlepthic;mechnic,mchanstic;philtely,ph7latlic;d7meter,d7amtricAndtothe*Ttamg;uchigap:   (50)   cadmic,cademcian;thortic,thoretcian;mthemtic,mthematcian; p   hmat;logy,hmatol;gic;m7litry,m7litarstic;6ndina,6ndianpolis x  `  Aswesawin1.3.3,thesameabsenceofstressusuallyoccursinpretoniclightsyllablesinword-initialposition,suchasinbanna.Heretoo,wehaveproductivedestressing: $#t! Ѐ (51 )  grmmar,grammrian;;rigin,orginal;mdicine,medcinal;n;vel,novlla; $L #   cvil,civlian;mjesty,majstic;pr;phecy,prophtic;mracle,mirculous &("% Wordinitially,though,wefindsomewellknownlexicalexceptions(e.g.rc;on)aswellassome  usuallyunrecognizedinstancesofstemstresspreservation(e.g.fscstic).Ireferthereadertothe  AppendixAforadiscussionofthesomewhatcomplextheoreticalchallengesthese"exceptionalexceptions"present.  In1.3.3,thelackofpretonicstressonwordslikebanna,andTtamag;uchi,wasattributed ` totherankingofFtBinaboveParse%.Theparallelblockingoflexicalandstembasedstress L  preservationdescribedherecanlikewisebeascribedtothehighrankingofFtBin,solongasit 8  dominatesStressIdent.SincethegeneralversionofStressIdentisdominatedbyStressWell, $ t whichitselfisseveralrungslowerinrankingthanFtBin,thetendencyformonomoraicsyllablesto  ` bedestressedisalreadybuiltintothehierarchy.However,thattheyarealwaysdestressedestablishes  L  anewranking:FtBindominatesthelexicallyspecificStressIdent.WithStressIdent  8  interveningbetweenFtBinandStressWell,lexicallybasedpretonicstresspreservationispossible $  forwordslikecondensationandchimpanzee,butimpossibleforwordslikephoneticianand   Montebello,ormathematicianandTatamagouchi,aswellasmajesticandbanana.Thefollowing   tableaucomparescondensationandphonetician;theotherwordtypeswouldbetreatedjustlike   phonetician:    & \ (52)FtBinNStressIdent  \ IX9XXX9* ddrB@Rr@kRr@Rr@Rr@(#(#,rB@,Rr@,kRr@,Rr@+  /OOO H 20C/Input:condensation x phonetician ;OPOP(dCOOO C;FtBin#X9XIX9)H#IX9XXX9 <OO+xC"OPOP C<#X9XIX9$J#Stress x IdentIX9XXX9 ;OPOO(dC"OO C;#X9XIX9J#Stress x WellIX9XXX9 OPO@(dC" 00 2 OPOO <<CO(cAndn)(s)tion) ;PPP( CPO C; 8P+ C"PPP C8IXIX9*!#IX9XIeL# ;PPO( C"P C;#X9XXIX9oK#IX9XXX9IXIX9Ԁ#IX9XI&M##X9XXIX9M#IX9XXX9 QPOO@( C" <<  PPO 5<<CQ(cAn)(dn)(s)tionIXIX9Ԁ3#IX9XIN# ;PPOP(CPOO C; <PO+C"PPOP C< ;PPOO( C"PO C;IXIX9*#IX9XI O# MO>(!C" << 5 PPOO <<CM(ph#X9XXIX9M#IX9XXX9A#X9XXIX9O#IX9XXX9ne)(tcian)#X9XXIX9P#IX9XXX9IXIX93#IX9XIsP##X9XXIX9TP#IX9XXX9 :P+l"CO C: :P+l#C"P C:IXIX9*#IX9XIoQ# >PO+l$C"P C> GPOO6#l%C" << PO <<CG(ph#X9XXIX9P#IX9XXX9A#X9XXIX9]R#IX9XXX9)(n#X9XXIX9R#IX9XXX9#X9XXIX9R#IX9XXX9)(tcian) 9PPO(D&CPOO C9IXIX9*!#IX9XIS# 9PPO(D'C"PPO C9 ;PPOO(D(C"PPO C;IXIX9*#IX9XIDT##X9XXIX9+S#IX9XXX93)'D)C" <<  PPOO 3#X9XXIX9T#'\ \GBecausewordslikeIX9XXX9CEUS.,cAndnstion#X9XXIX9IU#US.,CE.,hUԀincurnoviolationofFtBin,stresspreservationispossible. * However,inawordlikephonetician,stresspreservationcreatesaviolationofFtBin.As  + StressIdent-isrankedbeneathFtBin,stresspreservationisimpossibleinthiscontext. !,   Sincethisanalysisalsoappliestothedistinctionbetweenpossibleexceptionalmonomorphemicwordslikech7mpnze,andimpossibleoneslike*MAntbllo,ausefulcomparison |#. canbemadetothediscussionofsuchcasesinKager1989:140.Kageralsopositslexicalstressonthepretonicsyllableofch7mpnze.Thislexicalstressblockstheruleof'ClosedSyllableAdjunction', T% 0 whichwouldordinarilyformafootbyadjoiningthemedialsyllablewithaprecedingone.However,toexpresstheabsenceofwordslike*MAntbllo,Kagerisforcedtoinvokeabaldrestrictionagainst ,'|"2 lexicalstressonlightsyllables,presumablybecausethereisnoprincipledreasonwhylexicalstressshouldblock'ClosedSyllableAdjunction',butnot'OpenSyllableAdjunction'.WhenpreservationofunderlyingstressisformalizedasanOptimalityTheoreticconstraint,however,theextenttowhich lexicalstresscanoverridetheusualstressphonotacticsofthelanguagecanbeneatlyexpressed.This *,&6 comparisonservestoagainhighlightthedifficultiesnonuniformityposesforatheorybasedonfullsatisfaction,andtheeasewithwhichitisdealtwithunderminimalviolation. 3.1.3UnfaithfulnessIII:AlignHeadNStressIdent t TheapproachtakentorulingoutstresspreservationwhenitconflictswithFtBincarriesovertoan ` accountofacoupleofcentralfactsleftundiscusseduntilnow.Theinstancesofstresspreservationdiscussedherearewhatisreferredtointheliteratureasweakpreservation:theprimarystressofthe 8  stemcorrespondstoasecondarystressinthederivative.Weakpreservationisnotcompletefaithfulness!thesegmentbearingstressinthestemistheheadoftheProsodicWord,anditscorrespondentisbuttheheadofafoot.Aseparate,butobviouslyrelatedobservation,isthatpreservationofthestemstressdoesnotinterferewithmainstressplacement!stressispreservednotatall,insteadofforcingthedisplacementofmainstress.Similarly,lexicalstressdoesnotforcemainstressplacementtotheleft.  Iwillstartwiththesecondobservation.Anexampleofhowstresspreservationmightinterferewithmainstressplacementcanbefoundinthecaseswehavejustlookedat.IfFtBinN   StressIdent-disallows*grmmrian,thenwhatrulesout*grmmarian?Thiswouldsatisfyboth   FtBinandStressIdent.Theproblemwith*grmmarian,ofcourse,isthatmainstressistoofarto p theleft.Assumingthatthese,andotherstressplacingsuffixes,areincorporatedintotheProsodicWord(seeLamontagneandSherer1993),arankingofAlignHeadNStressIdentwillensure H thatmainstressplacementisunresponsivetothedemandsofstresspreservation: (53)AlignHeadNStressIdent   \ IX9XXX9* ddrB@Rr@kRr@Rr@(#(#,KrB@,kRr@,Rr@+  /OOO H 20C/Input:#X9XXIX9c#IX9XXX9 (x grammarian <OPO+dCOOO C<#X9XIX9d#Align (x ! (x  (x !HeadIX9XXX9 ;OPOO(dC"OPO C;#X9XIX9e#Stress (x IdentIX9XXX9 OPO@(dC" 00 2 OPOO <<COЀ#X9XXIX9f#gram(mri)anIX9XXX9Ԁ#X9XXIX9=g#IX9XXX9IXIX93#IX9XIg##X9XXIX9g#IX9XXX9 :PP+lCPO C:IXIX9**#IX9XIah# ;PPO(lC"PP C;#X9XXIX9h#IX9XXX9IXIX9*#IX9XI"i##X9XXIX9i#IX9XXX9 QPOO@(lC" <<  PPO 5<<CQIXIX9Ԁ#Ii##X9Xi#(grmma)rianIX9XXX9 <PPO+D CPOO C<IXIX9*#IX9XIj#IXIX9**#IX9XIk#IXIX9!#IX9XILk# ;PPOO(D!C"PPO C;#X9XXIX9jj#IX9XXX98.,D"C" << 5  PPOO 8#X9XXIX9k#Asdiscussedin2,thedominanceofAlignHeadbyFtBin,andNonFinality,entailsthatthe # minimalnumberofAlignHeadviolationsisthetwoincurredbytheoptimalcandidate.Thethird  $ violationrulesout*grmmarian,duetotherankingofAlignHeadoverStressIdent.Similar !% resultsobtainifonepositspreantepenultimatelexicalstress.  Thesamerankingcanaccountforthesubordinationofthepreservedstresstotheprimarystress.LetusassumethatStressIdentisagradientconstraint:itissatisfiedifthecorrespondentof h$( theheadoftheProsodicWordisitselftheheadoftheProsodicWord(i.e.strongpreservation),oneviolationiscausediftheheadoftheProsodicWordisincorrespondencewithonlytheheadofafoot(weakpreservation),whiletwoviolationsresultiftheheadoftheProsodicWordisincorrespondencewithanonhead(nonpreservation).AnattempttobettersatisfyStressIdentby (h#, makingthepreservedstresstheheadoftheProsodicWord,willautomaticallyincreasethenumberofviolationsofthehigherrankedAlignHead: )@%.   *,&/  &   (54)AlignHeadNStressIdent:Stresssubordination   IX9XXX9*s ddKrB@kRr@Rr@(#(#s,KrB@,kRr@,Rr@+  /OOO  20C/Input:#X9XXIX9q#IX9XXX9  imagination <OPO+COOO C<#X9XIX98s#Align  !    !HeadIX9XXX9 ;OPOO(C"OPO C;#X9XIX95t#Stress  IdentIX9XXX9 OPO@(C" 00 2 OPOO <<COЀ#X9XXIX9t#i(mgi)(n)tionIX9XXX9IXIX93#IX9XIu##X9XXIX9yu#IX9XXX9 :PP+`CPO C:IXIX9*#IX9XIZv# ;PPO(` C"PP C;#X9XXIX9v#IX9XXX9IXIX9*#IX9XIw##X9XXIX9v#IX9XXX9 QPOO@(` C" <<  PPO 5<<CQIXIX9Ԁ#Iw##X9Xw#i(mgi)(n)tionIX9XXX9 <PPO+ 8 CPOO C<IXIX9***#IX9XIx#IXIX9!#IX9XIy# ;PPOO( 8 C"PPO C;#X9XXIX9ex#IX9XXX98., 8 C" << 5  PPOO 8#X9XXIX9y#'qHeretheminimalAlignHeadviolationisbutasingleone,sincethevowelisbimoraic,andFtBin p  isobeyed.Ifthestemstressispreservedasaprimarystress,theadditionalAlignHeadviolations \  arefatal.Ihaveomittedacandidatewithtwomainstresses:thiswouldberuledoutbyanindependent,probablyuniversallyundominated(orGeninternal),constraintagainstjointheadship. 4  Ѐ  Towindupthediscussionofprosodicfaithfulness,itisstrikingthatthedistributionof  p  exceptionalandstembasedstressrequireshighrankingFtBinandAlignHead,alowranking  \  AlignLeft,andanintermediateStressWell,allofwhichareprovidedbytheanalysisofregular H  stress.ThisprovidesstrongsupportfortheapproachtakenhereofaccountingforthesespecialcasesbyinterspersingStressIdentconstraintsintothehierarchyofconstraintsestablishedfortheusual   patterns.Moreover,thecloseparallelsbetweenthedistributionoflexicalandstembasedstressfalloutnicelyfromtheuseofasinglemechanismofprosodicfaithfulnessforthemboth(thoughcf.AppendixB).Andaboveall,wehaveaccountedforawidearrayoffactsbysimplyaddingoneconstraint,anditslexicallyspecificinstantiation,totheindependentlymotivatedhierarchy.Theresultinghierarchyisasin(55): &  (55)  IX9XXX9Ԁ   FtBin,Troch,#X9XIX9C#Nonfinality  IX9XX9 D   󀀀 `  |   h  4 #X9XIX9€#UKUS.,   ` AlignHeadIX9XXX9US.,UK.,x  p   / |  *ObsNuc  \    򀀀/ ` Parse%  / H! #X9XXIX9#StressIdent IX9XXX9 |  / 4"   \WeighttoStress $#   \ |  $    ` StressWell !%    ` /󀀀 | "& #X9XIX9#ԀStressIdentIX9XX9Ԁ*SonNuc,WeightIdent #'    ` \ | p$(    ` Align!Left#X9XXIX9w#UKUS.,ԀUS.,UK.,N' D) \% ) Inthefollowingsubsection,weturntosomemoreexceptionstothegeneralizationsof1,whichthistimefalloutsidethedomainofprosodicfaithfulness,andsomustbegeneratedbythegrammar.Thesecasesservetostronglymotivateagrammaticalaccountofexceptionality,inadditiontoapurelylexicalone,aswellastohighlightthedifferencesbetween"productiveexceptions",andthenonproductiveonesdiscussedinthislastsection. + '0 3.2Lexicalexceptionalityaslexicallyspecificranking  Asdiscussedin1.2.1,oddparitypretonicstringscontrastwithbisyllabiconesinthatthefinalpretonicsyllableofthosestringsisusuallystressedifitisheavy,asinbndna,Hlicrnssus,and  rgumnttion.ThiswasexplainedbytherankingParse%NStressWell,whichcreatesa t preferenceforparsingthelonesyllableintoafoot,overthestresslessnessthatStressWelldemands ` ofasyllableadjacenttothemainstress.Thissectionexaminestheratherlargesetofexceptionstothispatternofheavysyllablestress.IwillprovideanaccountoftheseproductiveexceptionsintermsofalexicallyspecificrankingofStressWellNParse%,whichcreates,forthesewords,a $ t preferenceforpretonicstresslessnessoverparsing.  ThebestknownexceptionstotheusualpatternofinitialpretonicheavysyllablestressarewordsthatwerehistoricallyformedwithLatinateprefixes(seee.g.ChomskyandHalle1968:121,LibermanandPrince1977:284,andH&V1987:239).Theprefixesoftensurfaceasstresslesswhenpretonic,whethertheyendinasonorant(56a),anobstruent(56b),orhavealongvowelinother(oftenrelated)words(56c).ItwasintheirdiscussionofthesewordsthatLibermanandPrincepointedoutthespecialmarkednessofstressadjacenttothemainstress,asopposedtosimpleadjacency(cf.prsnttion,prAlAngtion,andrlxtion):    &  (56)   a.condmncondnseembrrassembrceenggeengrveenj;y \   b.CEUS.,abs;lveUS.,CE.,CEUS.,ԀUS.,CE.,CEUS.,admreUS.,CE.,ԀCEUS.,advntageUS.,CE.,IԀextrmeextnguishobsrveobstrct  c.prec;ciouspresntprol;ngrecrrentref;rmrelCEUS.,US.,CE.,x'\)Itisnotthecase,though,thattheseprefixesalwaysreduceinthepretonicenvironment.Besidesthefactthatmoresemanticallytransparentcasesofprefixation,especiallywiththeveryproductiveprefixes/pre/,/re/,/pro/and/de/,donotinvolvereduction(e.g.recover'coveragain'vs.recover 8 'getback';rebutter'butteragain'vs.rebutter'onewhorebuts',preconsciousvs.precocious!the $ consistentlongvowelsarelikelyduetoarestrictionthat'true'prefixesinEnglishmustbebimoraic;seeMcCarthyandPrince1994b),thereisagreatdealofvariationinwhetherwordswithopaqueLatinateprefixationhavestressedorstresslessinitialpretonicsyllables.Ingeneral,morecommonwordshavestresslessinitials,whilemorelearnedwordshavestressedinitials(Fidelholtz1975).Togiveasenseofthesortofvariationthatoccurs,thelistsin(57)provideexamplesofwordswithhistoricalLatinateprefixesthataretranscribedbyKenyonandKnott(1953)asstressless,stressed,orwithbothstressedandstresslessvariants.IhaveindicatedinbracketsinstancesinwhichWebster's1981disagreeswithKenyonandKnott.Aplussign(+)meansthatWebster'stranscribestheinitialsyllableofthewordasstressed,aminussign(CEUS.,US.,CE.,{)stressless,andanequalsign(=)bothstressedand  #p! stressless:  $H #  & $  (57)   CEUS., a.Stressless:   0  administer,admire,absolve,admonitory,advance,advantage,adversity(+),advise,combat(v.),combust,companion,compassion,compose,compress,compulsion,companion,concur,concern,condemn,conduct,confection,confer,conflate,conflict(v.),congressional,controller,convenient,convention,embarrass,embody,embrace,endeavour,endow,engage,enjoin,enjoy,enlarge,enlighten,entice,entire,exact,example,exceed,except,excoriate,excrete,excursion,excuse,executive,exhume,expose,express,extend,extinguish,extravagance,extraneous(+),extreme,object(v.),obsequious(=),observe,obsess(=),obstetric,obstruct,obtain,obtrude,obtuse,obvert,proceed,produce,profess,profound,project,prolong,promote,propel,propose,protect,propose (#(# US.,CE.,NCEUS.,'$ %  b.Stresslessorstressed:  $  0  abdominal,abduct(+),abhor(),absorb(),abstemious(+),abstract(v.),abstruse,absurd   (),abnormal(+),accelerate,accentuate,accept,accessible,accessory,acknowledge,adhere,   admit(),admixture(+),admonish(+),adverse(+),adverbial,concelebrate,concoct,   concordance,eccentric,emphatic,exhale(+),obscene,obscure,obverse,pronomial(+) (#(#   c.Stressed: 0 \ (# (# 0  abjure,abscise,abscissa,abscond,access(v.),admeasure,adsorb,advection,agnomen,concrete(v.),concretion(=),conglobate(=),concur(=),empiric(),emporium(=),enteric, 4 excreta,excursus,exogamy,expropriate,exsect,extrinsic,extrorse,obtest,obtund,progentior,proliferate(),proscenium,prosector,protract(=),protrude \(#(# Sincethemorecommonwordstendtobereduced,forthissetofwords,itwouldseemthatpretonicstresslessnessisproductive.Thisisconfirmedbytheexistenceofanumberofderivedwordsinwhichpretonicstresslessnessoccursonsyllablesthatarestressedinthestems: G  14          (58) IX9XXX90  ccss,accsible;dverb,advrbial;c;ncArd,conc;rdance;c;ngress,congrssional;  mphasis,emphtic;xecte,excutive,excutor#X9XXIX9#(#(# US.,CE.,ΙThepresenceoffullstressedvowelsinthestemsprecludesananalysisofthepretonicstresslessnessofthesewordsintermsoffaithfulnesstoanyprosodicorsegmentalfeature.Thesestresslessheavysyllablesmustthereforebegeneratedbythegrammar.Tocapturetheproductivityofthispattern,wecanpositalexicallyspecificversionofStressWell('nostress/mainstressadjacency'),whichIwill  #p!  refertoasStressWell ,thatdominatesParse%.Foratypicalspeaker,theset' 'wouldinclude  $\" mostofthewordsin(57a),someofthewordsin(57b),butfewofthosein(57c). G  15      ׀Theresultofthis   lexicallyspecificrankingisillustratedin(59): (59)StressWell NParse%  t IX9XXX9*s ddKrB@kRr@Rr@(#(#s,KrB@,Rr@,JRr@+  /OOO d 20C/Input:  advantage ;OPOP( COOO C;#X9XIX9T#StressWell IX9XXX9 >OOO+C"OPOP C>#X9XIX9#Parse%IX9XXX9 OPO@(C" 00 2 OOO <<CO(d)(vn)tage ;PPP( ( CPO C;IXIX9*!#IX9XIV# >PO+ ( C"PPP C>IXIX9*#IX9XI٧# QPOO@( ( C" <<  PO 5<<CQ#X9XXIX9#IX9XXX9ad(vn)tage#X9XXIX9#IX9XXX9IXIX9Ԁ3#IX9XI# ;PPOP(  CPOO C; >POO+  C"PPOP C>IXIX9**#IX9XI#8., C" << 5  POO 8Forwordsthataresubjectto#X9XIX9ݨ#StressWell IX9XXX9,pretonicstresslessnessispreferred,eventhoughthis 8  resultsinanextra#X9XIX9#Parse%IX9XXX9Ԁviolation(thatis,inadditiontotheonemadenecessarybythedominance $t  ofNonfinality).Forwordsthatarenottargetedby#X9XIX9C#StressWell IX9XXX9theextra#X9XIX95#Parse%IX9XXX9Ԁviolation#X9XXIX9#IX9XXX9 `  makespretonicstresslessnessungrammatical,sinceasshownin1.2,#X9XIX9ݬ#Parse%IX9XXX9Ԁdominatesthegeneral L  StressWellconstraint. 8 #X9XXIX9#  TheeffectsofthisrankingcanalsobeseenoutsideofthedomainofwordsbasedonLatinate $ prefixes.Firstofall,asonemightexpect,therearewordswithstresslessinitialheavysyllablesthatsuperficiallyresemblethosein(57),butdonotinfactcontainLatinateprefixes: (60)   agn;stic,conftti,conqustador,obsdian  Therearealsosomemonomorphemicwordsthatbearlessofaresemblancetothewordsin(57),butyetcanhavestresslessinitialsyllables.Allofthewordsin(61)aregivenineitherKenyonandKnott1953orWebster's1981withatleastavariantwithastresslessinitial: G  16       8  (61)   ambssador,Atlnta,Atlntic,Kentcky,Manhttan,Septmber,sincre ` OutsideofLatinateprefixationtoo,initialheavysyllablestresslessnessappearstobeproductivelygenerated.Initialreductionofsuchclosedsyllablesis"restrictedtowordsofconsiderablefrequency"(Kager1989:142,citingFidelholtz1975).Thiswouldseemanoddrestrictiononanentirelyextrasystemicpattern(whichishowithasusuallybeentreated).Furthermore,thereareexamplesofdestressinginderivation.Mostofthesecasesinvolvelongvowels(62a),butthereisatleastone "$ instanceofeachasonorantfinal(62b),andanobstruentfinalsyllable(62c)becomingdestressedinaderivedword.Again,variationrunsrampanthere. (62) 0  a.Plto,plat;nic;ph;ne,phon;logy;vcte,vaction;schma,schemtic;lgal,leglity; t dmon,dem;nic;rgal,reglity;ftal,fatlity;ph;togrph,phot;graphy (#(#   b.snse,senstion  c.spctacle,spectcularSofar,wehavelookedonlyatinitialsyllables.However,therankingStressWell NParse%  d generalizestomedialenvironmentsaswell.Recallthatinthepresentaccount,thedominanceofParse%overStressWellalsogeneratesstressonthepretonicsyllableofwordslike  <  Halicarnassus.Theproductivityofstressinthisenvironmentwasdemonstratedbywordslike (  argumentation,inwhichastresslesssyllableinthestembecomesstressed.Thereare,however,   monomorphemicwordsthatlackstressonsuchsyllables: (63) 0  K7limanjroNbuchadnzzarlecampne (#(# Wealsofindinstancesofproductivedestressinghere.LibermanandPrince(1977:298)notethatsntimentlityoptionallyoccurs,thoughsntimntalbearsastressonthecorrespondingsyllable.To P thisexamplewecanaddthosein(64a)and(64b).AllofthemappearinWebster's1981;allhavevariantswithpretonicstress.̀ (64)   a.7nstrumntal,7nstrumentlity;rcommnd,rUKUS.,commendtion;;rint,ArientUS.,UK.,BUKUS.,tion; P   represent,rpresentUS.,UK.,UKUS.,tionUS.,UK.,Ԁ  b.rtrogrde,rtrogradtion;cvil7ze,c7viliztion;stndard7ze,stndardiztion;  pras7te,prasit;logyThistooresultsfromStressWell NParse%:   (65)StressWell NParse%  x IX9XXX9*s ddKrB@Rr@JRr@(#(#s,KrB@,Rr@,JRr@+  /OOO h  20C/Input:   Kilimanjaro ;OPOP(! COOO C;#X9XIX9h#StressWell IX9XXX9 >OOO+ !C"OPOP C>#X9XIX9$#Parse%IX9XXX9 OPO@( "C" 00 2 OOO <<CO(K7li)(mn)(jro) ;PPP(",#CPO C;IXIX9*!#IX9XIs# >PO+",$C"PPP C>IXIX9*#IX9XI# QPOO@(",%C" <<  PO 5<<CQ#X9XXIX9#IX9XXX9(K7li)mn(j#X9XXIX9#IX9XXX9#X9XXIX9#IX9XXX9ro#X9XXIX9?#IX9XXX9)IXIX93#IX9XI# ;PPOP($ &CPOO C; >POO+$ 'C"PPOP C>IXIX9**#IX9XIc#8.,$ (C" << 5  POO 8#X9XXIX9#IX9XXX9Finally,asKager(1989:125)notes,thereisoneclearexceptiontothegeneralizationthatpretonic <'") obstruentfinalsyllablesretainstresswhenprecededbyaheavysyllable:d7 ag n#X9XXIX9#;IX9XXX9sis. G  17      ׀Since#X9XIX9!#Parse%IX9XXX9 ((x#* dominatesWeighttoStress,bytransitivityStressWell#X9XIX9#Ԅ IX9XXX9Ԁalsodoes.Therefore,thisexceptionally  unstressedheavysyllable,whosesuperficialresemblancetoaLatinateprefixisperhapsnotentirelyacoincidence,canalsobeattributedtothelexicallyspecificrankingof#X9XXIX9v#IX9XXX9StressWell#X9XXIX9#IX9XXX9.  #X9XXIX9#  Insum,inthissectionIhaveshownthatstresslessnessonloneheavysyllablesisproductive, t atleastforalexicallyrestrictedgroupofwords.InUKUS.,1.2.1,itwasdemonstratedthatforthebulkofthevocabulary,stressonsuchsyllablesisthenormandisproductive.StressinthiscontextisproducedbytherankingIX9XX9US.,UK., Parse%#X9XXIX9#UKUS.,ԀNUS.,UK.,4IX9XX9StressWell#X9XXIX9j#,whilestresslessnessisinducedbyalexically 8  specificrankingofIX9XXX9StressWell#X9XIX9#Ԅ UKUS.,NUS.,UK.,jԀIX9XX9Parse%#X9XXIX9#.Thedifferencebetweentheunproductiveexceptions $ t likech7mpnze,whichwereaccountedforviafaithfulness,andtheseproductiveexceptions,iswell  ` capturedbyamodelinwhichunproductiveexceptionsareencodedwithphonologicalstructureinthelexicon,andanappropriaterankingofafaithfulnessconstraint,whileproductiveexceptionsareaccountedforentirelyinthegrammar,throughreranking(Inkelas,Orgun,andZoll1994,Inkelastoappear).Thoughthereissomeoverlapbetweentheempiricalcoverageoftheseformalmechnisms,andthoughcriteriaofproductivityarenotoriouslycontentiousanddifficulttodefine,itseemsclearthatlexicallyspecificranking,andexceptionallexicalspecification,haveseparaterolestofulfill,andthatneitheronecanbeeliminatedfromthetheory.3.3Remarksonspecialsecondarystress \ Themostimportantpointabouttheanalysisofspecialsecondarystresspresentedhereisthatitdoesnotovergenerate.ThiscontrastswithIX9XXX9theaccountoflexicalvariationinHalleandKenstowicz1991#X9XXIX9#, 4 whichinvokesIX9XXX9lexicallyconditionedrulesofweighttostressandStressCopy#X9XXIX9#IX9XXX9.#X9XXIX9A#IX9XXX9ԀInsteadofcompletely  p shuttingdownweighteffects,asthelexicallyspecificweighttostressruledoes,thelexicallyspecificrankingofStressWellalterstherelativestrengthsofsomeoftherelevantconstraints.Crucially, H thisrerankingdoesnot#X9XXIX9#UKUS.,subordinateWeighttoStressor*SonNuctoAlign-Left.Therefore,the 4 grammarconstructedheredoesnotgenerateunattestedformssuchas*MAnongahla.Further,the   lexicallyspecificrankingofStressWelldoesnotfavour*T7conder;gaoverthecorrect   US.,UK.,UKUS.,T7cAnder;gaUS.,UK.,UKUS.,,sinceadjacentstressesarenotmilitatedagainstbyStressWell,onlystressadjacent  totheprimarystress.IfsecondarystressweighteffectscanbeturnedoffUS.,UK.,UKUS.,arbitrarilyUS.,UK.,UKUS.,,asinHalleandKenstowicz1991,thenthereisnoreasonwhytheseunattestedformsshouldnotbegeneratedalongsidetheperfectlyregularTtamag;uchi. l US.,UK.,UKUS.,  US.,UK.,HalleandKenstowicz's(1991)useofUKUS.,theUS.,UK.,;UKUS.,lexicallyconditionedAccentRulefor X  cAndnstion-typeUS.,UK.,kUKUS.,ԀUS.,UK.,UKUS.,wordsisinpartmotivatedbytheobservationthatstemstressisnotconsistently D! preservedinthisenvironment,asdemonstratedbywordslike7nformtion.IUS.,UK.,UKUS.,nsteadofdenyingthat 0"  stemstressplaysanyrole,US.,UK.,UKUS.,asthatanalysisimplies,thelexicallyspecificStressIdentconstraints #l! allowfaithfulnesstostemstresstovarysomewhatbetweenwords.Becausefaithfulness,ratherthansyllableweight,iscalledupontomotivatethepretonicstressinUS.,UK.,rUKUS.,cAndnstionUS.,UK.,UKUS.,,pretonicstressina $D # wordlikecAncentrtion,inwhichthestemlacksstressonthecorrespondingsyllable,isruledout. %0!$ Also,sinceStressIdentisrankedaboveAlignLeft,leftalignmentofwordslikeimgintionis &"%  impossible,thoughitispredictedtooccurundertheassumptionIX9XXX9US.,UK.,UKUS.,either#X9XXIX9y#US.,UK.,UKUS.,Ԁofalexicallyconditionedrule '#& ofStressCopy(HalleandVergnaud1987;HalleandKenstowicz1991US.,UK.,UKUS.,;cf.US.,UK.,Burzio1994:192UKUS.,),orofmorphologicalreanalysis(ChomskyandHalle1968;HalleandVergnaud1987US.,UK.,UKUS.,).   US.,UK.,kItappearsthatheretoo,itistheallornothingcharacteristicoffullsatisfactionthatisatthe  rootoftheempiricalshortcomingsoftheseprioranalyses.UKUS.,ԀUS.,UK.,kTheprecisionofthepresentaccountofthelexicalinfluencesontheweighttosecondarystressrelationshipisnotboughtatthecostofgenerality.Aswehaveseen,thelexicallyspecificrankingofStressWellprovidesaunifiedaccount L  ofanumberofexceptionalpatternsofheavysyllablestresslessness.AndbyplacingthefaithfulnessconstraintStressIdentanditslexicallyspecificcounterpartintothehierarchyestablishedforregular $ t stress,thesameprinciplesthatdetermineregularprosodificationareusedtocontroltheextenttowhichstemandunderlyingstresscanaltertheusualpatternofsecondarystress.̀ 4.FinalHierarchyandConclusions    UKUS.,Thehierarchynowstandsasin(66).Underneaththehierarchyareexamplesfromthesetsofwordstargetedbythelexicallyspecificconstraints.US.,UK.,UKUS., &  (66)US.,UK.,B  IX9XXX9Ԁ   FtBin,Troch,#X9XIX9#Nonfinality  IX9XX9 p   󀀀 `  |   h  ` #X9XIX9 #UKUS.,   ` AlignHeadIX9XXX9US.,UK., L    ` / |   8   /#X9XIX9#IX9XX9ԀStressWell#X9XIX9#Ԅ IX9XXX9 h  $t   / |  *ObsNuc `   򀀀/ ` Parse%  / L #X9XXIX92#StressIdent IX9XXX9 |  / 8   \WeighttoStress (   \ |     ` StressWell     ` /󀀀 |  #X9XIX9c#ԀStressIdentIX9XX9Ԁ*SonNuc,WeightIdent     ` \ | t    ` AlignLeft#X9XXIX9#UKUS.,ԀUS.,UK.,UKUS., `    WhereUS.,UK., UKUS.,=US.,UK.,4CEUS.,{admireUS.,CE.,bCEUS.,,companion,US.,CE.,CEUS.,entire,US.,CE.,CEUS.,exposeUS.,CE.,CEUS.,,US.,CE.,4CEUS.,presentUS.,CE.,bCEUS., , Atlanta,ManhattanUS.,CE.,CEUS.,, 8"     `  Kilimanjaro,US.,CE.,UKUS.,representUS.,UK.,QCEUS.,aUS.,CE.,UKUS.,tion...}US.,UK.,UKUS.,  AndUS.,UK., UKUS.,={chimpanzee,condensation,creativity,contestation...} $d" 'pWUS.,UK.,/UKUS.,Attheoutsetofthepaper,itwaspointedoutthatweighttostress,andstemstresspreservationunequivocallydetermineUS.,UK.,UKUS.,theplacementofsecondarystressincertainphonologicallydeterminedenvironments,whileinotherstheactivityoftheseprinciplesissubjecttothevagariesofthelexicon.US.,UK.,dUKUS.,In3.1,wealsosawthatinyetotherenvironments,thepreservationofstemstress,aswellasofunderlyingstress,isunabletoexertanyinfluenceatall.US.,UK.,BUKUS.,Throughoutthepaper,thedifficultiesthisnonuniformitycreatesforanalysesbasedonfullsatisfactionhavebeenpointedout,andhavebeenovercomethroughtheadoptionofminimalviolationandconstraintranking.Thereislittlenovelin x+&* theconstraintsthemselves;themaininnovationhereconsistsofthewaysinwhichtheyaremadetointeract.US.,UK.,UKUS.,  Acrucialaspectofthisanalysisisthattherelativerankingsoftheconstraintsarevalidatedacrossthevariousdomainsinvestigated.US.,UK.,ԀForinstance,thelowrankofAlignLeftthatismotivated t bythenonalignedleftmoststressinwordslikeMonAngahlaUKUS.,isalsorequiredtoexplainthe ` preferenceforrightmostmainstressUS.,UK.,UKUS.,(e.g.Amrica),andforUS.,UK.,UKUS.,stresspreservation(e.g.imgintion), L  overleftalignment.AdominantFtBinisanessentialingredientintheanalysisofthelackofheavy 8  syllablestressinwordslikeAlexUS.,UK.,UKUS.,US.,UK.,UKUS.,nder;aswellasintheabsenceofinitiallightsyllablestressin $ t banUS.,UK.,UKUS.,US.,UK.,UKUS.,na;itshighrankissupportedinprimarystressbytheviolationofAlignHeadthatitcompels  ` inwordswithlightpenults(e.g.Cnada),andinlexicalandstembasedstressbythenonpreservation  L  ofstressininstancesinwhichthiswouldcreateamonomoraicfoot(e.g.US.,UK.,UKUS.,*MAntblloand  8  *phAntcian).ThissortofcoherenceUS.,UK.,QUKUS.,Ԁprovidesaddedreasontobelievethatthisapproachto $  nonuniformweighttostressandstresspreservationeffectsisontherighttrack.US.,UK.,̀X9XXX9 KJ7XXdXXd7      AppendixA #X9XXX9#: Initialmonomoraicstresspreservation X9XXX9  "  Theanalysisinthetextsucceedsin#X9XXX9#X9XXX9fully#X9XXX9%#X9XXX9Ԁaccountingforthedistributionofregularand  exceptionalstressonpretonicheavysyllables.Thereis,however,aresidueofexceptionallystressedlightsyllablesthatremaintosomeextentintractabletoit.Examplesoflexicalstressarein(67): (67)   rc;onbb;onffcevm;osesttesttefftebss;on <    cffine.ssnerrtaCAllenflltio7llsionH7pp;crates   , | Andofstembasedstressin(68): (68)   fscist,fscstic;lprosy,lpr;tic;narchy,nrchic;gmma,gmmtion ,    hrald,hrldic;m;dern,mAdrnity;aron, ar;nic;cid,cdic;0taly,6tlian  rbbi,rbbnical;thic,thcian;Hllne,Hllnic;lpid,l7pdic;clnic,cl7ncian  mmmal,mmmlian;mtric,mtrcian;sm,smmtionItshouldbenotedthatmanyofthewordsinboth(67)and(68)havealternatepronunciationswithunstressedinitialsyllables.Thisisnotunlikethesituationforwordslikecondensationand h condemnation,bothofwhichappearinWebster's1981withbothfullandreducedpretonicvowels T (KenyonandKnott1953givecondensationwithonlyafullpretonicvowel,andtherestofthewords @ likeitwithfullandreducedvariants).Thepresentpatternisalsosimilartothecondensation/chimpanzeeoneinthatmonomorphemicwordswithstressinthispositionaregrossly h outnumberedbyoneswithout,andthatderivedwordsproductivelydestress,butdonotstress,thesesyllables.Allofthesefactsserveasdiagnosticsoftheunproductivenatureofthispattern,whichaswehaveseen,iswellcapturedbyananalysisbasedonfaithfulnesstounderlyingorstemstress.  Anotherbenefitofananalysisbasedonfaithfulnessisthattherestrictionofexceptionallystressedmonomoraicsyllablestoinitialpositioncanbederivedfromprinciplesactiveinthegrammarsofotherlanguages.WorkbySelkirk(1994)andBeckman(1995)(seealsoSteriade1993;Flemming1994)showsthatfaithfulnessconstraintsaresensitivetothemorphologicalorprosodicpositionofthetargetedelement.Forexample,inShona,contrastivevowelheightoccursonlyinthefirstsyllableofaroot.Beckman(1995)attributesthistoapositionsensitivefaithfulnessconstraintdemandingidentityinheightvaluesbetweencorrespondentsegmentsinrootinitialsyllables,whichisrankedhigherthanthegeneralvowelheightfaithfulnessconstraint.Giventheformalequivalencebetweenfeaturalfaithfulnessandprosodicfaithfulnessconstraints,itisnaturalthatprosodicfaithfulnessshouldalsobepositionsensitive.Inthepresentcase,byrankingapositionalStressIdentconstraintabove $d" FtBin,therestricteddistributionofmonomoraicfeetcanbegenerated. %P #   Whilesuchanalysise#X9XXX9l#X9XXX9xplainswhymonomoraicfeetoccuronlyinitially,thereareacoupleof %#IX9XXX9US.,UK.,,0     #X9XXIX9#X9XXX9UKUS.,Prince,Alan.1985.ImprovingTreeTheory.ProceedingsoftheEleventhannualmeetingofthe   BerkeleyLinguisticsSociety11 : 471490.  (#(# #X9XXX9#IX9XXX9US.,UK.,0     #X9XXIX9#X9XXX9UKUS.,Prince,Alan.1993.MinimalViolation.#X9XXX9/#IX9XXX9US.,UK.,NHandoutoftalkpresentedatROW-1,RutgersUniversity, t NewBrunswick,N.J.#X9XXIX9#UKUS.,X9XXX9.`(#(# Prince,Alan&PaulSmolensky.1993.OptimalityTheory:ConstraintInteractioninGenerative L 0  Grammar.Ms,RutgersUniversityandUniversityofColoradoatBoulder.[ToappearMIT 8 Press]. (#(# #X9XXX9w#IX9XXX9US.,UK.,e0   #X9XXIX9#X9XXX9UKUS.,Reynolds,Bill.1995.VariationinOptimalityTheory:TheFloatingConstraint.Paperpresentedat ` theAnnualMeetingoftheLinguisticSocietyofAmerica. (#(# Ross,JohnR.1972.AReanalysisofEnglishWordStress.InM.K.Brameed.Contributionsto 8 0  GenerativePhonology.Austin,Texas:UniversityofTexasPress.$(#(# #X9XXX9^#US.,UK.,}IX9XXX90     #X9XXIX9q#IX9XXX9Selkirk,Elisabeth.1984.PhonologyandSyntax:TheRelationbetweenSoundandStructure.  Cambridge,Mass:MITPress.#X9XXIX9֏#X9XXX9UKUS.,(#(# #X9XXX9#IX9XXX9US.,UK.,Ɛ0   #X9XXIX9#IX9XXX90(#(#  Selkirk,Elisabeth.1994.PositionallicensingaspositionalconditionalParsevs.AlignF;Gcat.Class  notesfromLinguistics730,UniversityofMassachusetts,Amherst.#X9XXIX9#IX9XXX9(#(# #X9XXIX9#X9XXX9UKUS.,0   Steriade,Donca.1993.Positionalneutralization.TalkpresentedatNELS24,Universityof p Massachusetts,Amherst.#X9XXX9 #XJ1XXX9US.,UK.,+CEUS.,\ (#(# 0   Steriade,Donca.1994.Conclusionsonlexicalphonology.Classnotes,UniversityofCalifornia,LosAngeles.#X9X1XXJ#US.,CE.,7UKUS.,X9XXX90" (#(# Travis,Lisa.1983.EvidencefortheGrid.ProceedingsoftheannualmeetingoftheWestCoast #h! 0  ConferenceonFormalLinguistics2:287298.$T"(#(# Webster'sNewCollegiateDictionary1981.Springfield,MA:G.C.MerriamCo. $@ # #X9XXX9;#IX9XXX9US.,UK.,)0   Withgott,MaryMargaret.1982.Segmentalevidenceforphonologicalconstituents.Ph.D. %,!$ dissertation,UniversityofTexas,Austin.0(#(#  &"%(#(# 0   Urbanczyk,Suzanne1995.DoubleReduplicationsinParallel.InBeckmanetal.,499531.#X9XXIX9Җ#X9XXX9UKUS.,'#&(#(# #X9XXX9#IX9XXX9US.,UK.,0     #X9XXIX9#