Pluralshiftedindexicalsareplural:evidencefromAmharic ChrisLaTerza†,RuthKramer‡,MorganRood‡,DustinChacón†,JenJohnson‡ †UniversityofMaryland,‡GeorgetownUniversity 1. Introduction Thispaperdocumentsandanalyzesthephenomenonoffirst-personalpluralshiftedindexi- calsinAmharic,exhibitedbelowin(1). (1) 1tS’tS’u-wotStS-u 1nn-aSänf-all-än al-u candidate-PL-DEF 1PL-win.IPFV-AUX-1PL say.PF-3PL ‘[Thecandidates]isaidthatWEiwillwin’ Our primary goal is to show that plural shifted indexicals are interpreted as semantically plural; a thesis that goes against many previous accounts on the interpretation of plural pronouns that are referentially dependent on an attitude holder nominal. We then offer modificationstoSchlenker’s(1999,2003,2012)neo-KaplanianproposalsaboutAmharic shiftedindexicalstoaccountforthesenewfacts. Thepaperisorganizedasfollows.In§2,wepresentasummaryoftherelevantempirical data to be discussed. This will include important data that has already been discussed in theliterature,aswellasnoveldatathatwehavecollectedfromseveralnativespeakersof Amharic.1In§3,weofferargumentsforthesemanticpluralityofpluralshiftedindexicals, followedbyouranalysisin§4.2Thepaperisconcludedin§5. 2. Empiricaloverview Shiftedindexicalsarearelativelyrecentdiscovery,andasurprisingoneatthatgiventhat Kaplan(1989)famouslyconjecturedthatsuchthingsdonotexist.Heobserved,correctly, 1WethankourAmharicinformantswhoprovideduswiththenewdatapresentedhere:GirmaDemeke, BirukFikadu,EzanaFisha,MelatFisha,YonHaregot,BekaleSeyum,KeduseTsegaye,andMeriemTikue. We make use of the following gloss abbreviations throughout: 1-first person, 3-third person, AUX- auxiliary, COP-copula, DEF-definiteness marker, IMPF-imperfective verb, M-masculine, NEG-negation, .O- objectmarker,PF-perfectiveverb,PL-plural,.S-subjectmarker,SG-singular. 2Thecontentof§3and§4areinanabridgedform.Foramoredetailedexpositionoftheargumentsand theformalanalysis,wereferreaderstoLaTerza(2014). © 2014byChristopherLaTerza,DustinChaco´n,JenJohnson,RuthKramer,andMorganRood JyotiIyer&LelandKusmer(eds.):NELS44,Vol.1,259–269. GLSAAmherst. 260 LaTerza,Kramer,Rood,Chacón,&Johnson that English sentences like (2) are not ambiguous with respect to the the first-personal indexicalintheembeddedclause,whichcanonlyrefertothespeakeroftheactualspeech act,notthespeakerofareportedspeechact(i.e.,John). (2) JohnsaidthatIamahero. KaplanhypothesizedthatthispropertyofEnglishindexicals,oftencalledrigiddesignation, isapropertyofindexicalscross-linguistically.Schlenker(1999)introducedAmharicdata thatfalsifythishypothesis.TheAmharicequivalentof(2),givenbelowas(3),showsthat theindexicalisambiguousinthewayKaplanthoughtwasimpossible. (3) John [dZägna nä-ññ] y1-l-all John hero COP-1SG.S 3SGM.S-say.IMPF-AUX.3SGM.S ‘Johnsaysthat{Iam,heis}ahero’ Schlenkerobservedthatnotonlycantheembeddedindexicalrefertothespeakerofthe reportedspeechact,butwhensuchreferenceisintendedtheindexicalreceivesanobliga- tory De Se interpretation. We refer readers to Lewis (1979), Perry (1979), and Schlenker (2012) especially for detailed discussion of De Se phenomena. For the purposes of this paper,itsufficestocharacterizeaDeSeattitudeasanattitudeaboutone’sselfwhereitis impossibletobemistakenaboutwhotheselfis.Certaintypesofattitudereportsinnatural language,suchasAmharicspeechreportswithshiftedindexicals,obligatorilyreportaDe Seattitude.Toseethis,considerthetruth(3)withrespecttothetwosituationsbelow;S1 andS2(contextanddatatakenfromAnand(2006),basedonSchlenker(1999),re-glossed asperourconventions). (4) S1:Johnsays“Iamahero.” S2:John,whoisacandidateintheelection,issodrunkhedoesn’trememberwho heis.HewatchesTVandseesacandidatehefindsterrific,thinkingthisguymust beahero.ThiscandidatehappenstobeJohnhimself,althoughhedoesn’trealizeit. John [dZägna nä-ññ] y1-l-all John hero COP-1SG.S 3SGM.S-say.IMPF-AUX.3SGM.S ‘Johnsaysthat{Iam,heis}ahero.’[TrueforS1,FalseforS2] InS1,Johnsayssomethingabouthimself,andheknowsforsurethatitishimselfthathe istalkingabout.InS2,Johntalksaboutathirdpersonhehasinmindthatheknowsunder somedescription,butlittledoesheknow,hehappenstobetalkingabouthimself.Inother words,inbothcasesJohntalksabouthimself,butonlyinS1doeshetalkabouthimselfDe Se.Assuch,(3)canonlybejudgedtruewithrespecttoS1,andisfalsewhenusedtoreport onS2. Pluralshiftedindexicalsareplural 261 2.1 Enterplurality Higginbotham (1981) observed that attitude reports like (5) and (6) can be true in very differenttypesofsituations. (5) JohnandMarythinkthattheyaresick. (6) JohnandMarywanttobesick. Iwilllabelthetwointerpretationsof(5)and(6)asgroupanddependent readings,which areparaphrasedbelow. (7) i. Groupreading:JohnandMaryeachthink/want:“wearesick.” ii. Dependentreading:JohnandMaryeachthink/want:“Iamsick.” Beck&Sauerland(2000)observedfurtherthatthepresenceofthedependentreadingis determinedbywhetherornotthereisapronounintheembeddedclausethatisreferentially dependentontheattitudeholdernominal.Theyprovideexampleslike(8)toshowthatthe dependentreadingitnotpossiblewithoutsuchapronoun. (8) MaxandPetersaidthatBillmarriedAnnandAmy(*respectively). We provide new data from Amharic that makes the same point: a dependent reading is not possible without a pronoun in the embedded clause that refers to the speakers of the reportedspeechact. (9) Journalist1says:“Obamawillwin.” Journalist2says:“Romneywillwin.” gazet’äñña-wotStS-u täwädadari-wotStS-u y-aSänf-allu journalist-PL-DEF candidate-PL-DEF 3PL.S-win.IMPF-AUX.3PL.S al-u say.PF-3PL.S ‘Thejournalistssaidthatthecandidateswillwin.’(Falseinabovesituationunless thecandidatesarepartofthesamegroup/party.) Asafinalpreliminaryremark,noticethatthereareisno“crossed”readingof(5)and (6),asshownbelowinEnglish(10)andinAmharic(11). (10) JohnandMarywanttobesick. Cannotmean:JohnwantsonlyMarytobesick,andMarywantsonlyJohntobe sick. 262 LaTerza,Kramer,Rood,Chacón,&Johnson (11) Obamasays:“Romneywillwin.” Romneysays:“Obamawillwin.” täwädadari-wotStS-u 1nn-aSänf-allän al-u candidate-PL-DEF 1PL.S-win.IMPF-AUX.1PL.S say.PF-3PL.S ‘Thecandidatessaidwewillwin.’(Falseinabovesituation.) Insum,Amharicpluralshiftedindexicalspossesthefollowingpropertiesthatneedto be accounted for in the semantics: (i) they are obligatorily De Se and (ii) they can have groupanddependentconstruals,butnotacrossedone.Theseissueswillbetakenupagain in§4,butfirstweturntodiscussionoftheinterpretationofpluralshiftedindexicals. 3. Essentiallypluralshiftedindexicals In this section we will argue that Amharic plural shifted indexicals are semantically plu- ral. This may seem like a trivial thesis, but much previous work on plurality in attitude reports actually treats such pronouns as singular. Below I provide rough truth conditions forasingularisttreatmentofdependentreadingof(1),assumingthatthedenotationofthe candidates is Obama and Romney. The shifted indexical is treated as a singular variable thatisboundbya(covert)universalquantifier. (12) x:x Obama,Romney [xsaidxwins] ∀ ∈{ } WeputasidetheDeSepropertiesofthereportforthetimebeing,andso(12)issimplified to not express anything about De Se content. The reason for this is that the arguments in thissectionfocusonthesingularnatureofthevariableassociatedwiththeshiftedindexical in(1),notaboutitsDeSeproperties.InlatersectionswewillreturntotheDeSeproperties ofthesereportsandshowhowtheycanbeintegratedintothesemanticsgiventheresults fromthissection.Havingsaidthat,wethinkthatitisimportanttoatleastrecognizethat theDeSepropertiesofsentenceslike(1)providesomeinitialmotivationforasingularist account.DeSeattitudesareaboutanattitudeholder’sself,andwetakethistobeafunda- mentallysingularnotion.IftheshiftedindexicalisthelocusofDeSeinformation,thenit isplausiblethatthesepronounsshouldtreatedassemanticallysingular.Furthermore,these reportsallowdependentreadings,whicharecharacterizedasreportingtheattitudesabouta singularindividual.Thisprovidesfurthersupportforsingularisttruthconditionslike(12). Despitethesereasons,wewillarguethatthesemanticvalueofthepluralshiftedindex- icalisalwayspluralwhenantecededbyapluralnounphrase.Theargumentforsemantic plurality comes from reports where the embedded clauses expresses cumulativity or reci- procity. These phenomena have been independently argued to require the presence of a localpluralnounphrase.Inthecasesthatinterestus,onlytheshiftedindexicalcanfitthis role; and therefore since these constructions are well-formed, then the shifted indexical mustbeplural.3 3Notethatsingularistaccountsmustprovideanexplanationforthepluralmorphologyonsuchpronouns. Forthis,mostpreviousanalyseshavereliedonarulelikethefollowingwhichessentiallyremovesthenumber featurefromthepronounsoastomakeitsemanticallysingular. Pluralshiftedindexicalsareplural 263 3.1 Reciprocity It is a standard assumption that reciprocal anaphors require an antecedent nominal that is both local and semantically plural; see Fiengo & Lasnik (1973) and Chomsky (1981) among others. Of particular interest to us is the requirement that the local antecedent be semanticallyplural,asillustratedby(13)4. (13) a. Thelionandthetigerkilledeachother. b. *Hekilledeachother. ConsidernowareciprocalanaphorintheembeddedclauseofanAmharicspeechreport withashiftedindexical.5 (14) 1ns1sa-wotStS-u 1nn1-ggädd1l-all-än al-u animal-PL-DEF 1PL-kill.RECIP.IPFV-AUX-1PL say.PFV-3PL ‘Theanimalssaidthatwewillkilleachother.’ Thissentencehasadependentconstrualwherewhateachanimalsaidisalongthelinesof“I willkillthatotheranimal”.Assumefurtherforthesakeofsimplicitythateachanimalitself wantstoremainalive.Ifweapplythesingulariststrategyin(12)asthetruthconditionsof (14),thentheresultwouldbeasfollows. (15) x:x thelion,thetiger [xsaidthatxwillkilleachother] ∀ ∈{ } Thesetruthconditionsareunsatisfactoryforthedependentmeaningof(14).Thereciprocal is anteceded by a singular value,the value assigned to the variable x. As such, we would expectthesameinfelicityhereasweseein(13a).Sincethereisnosuchinfelicity,wecan assumethatthepluralshiftedindexicalisinfactsemanticallyplural.6 (i) LFFeatureDeletionUnderVariableBinding(Stechow(2003)) Deletethefeaturesofallvariablesthatarebound. For us, no appeal to such a rule is required, at least for number morphology. We offer the parsimonious alternativethatthenumbermorphologyisactuallyreflectinggenuinesemanticplurality,asitoftendoeswith othernounphrases. 4Fiengo&Lasnik(1973)offerargumentsbasedonpluraliatantumnounsthatitisnotenoughforthe reciprocal’santecedenttobemerelymorphologicallyplural.Thisisevidencedbytheunacceptabilityof(i) whentalkingaboutasinglepairofscissors. i. *Thescissorsareconnectedtoeachother. 5InAmharic,reciprocityismarkedbyareciprocalverbform,notananaphorinanargumentposition. Nonetheless,theremuststillbealocalpluralantecedent. 6Thereexistso-calledscopalanalysesofreciprocals,suchasHeimetal.(1991)thatmakeduewitha singularvariableintheplaceoftheembeddedpronoun.However,severalcounterargumentshavebeenraised againstsuchanalyses;suchasWilliams(1991),Dalrympleetal.(1994),Asudeh(1998),amongothers.See LaTerza(2014)forfurtherdiscussion. 264 LaTerza,Kramer,Rood,Chacón,&Johnson 3.2 Cumulativity Cumulative interpretations of plural sentences were first observed in the semantics litera- turebyScha(1984),andarecharacterizedbyhavingpartsofonepluralityrelatedtoparts ofanotherbutnotexhaustively.Considerthefollowingsentencewhichhasaclearcumula- tiveinterpretation.7 (16) ObamaandRomneydancedwithMichelleandAnn. This sentence can be true if each candidate danced with his own wife and no one else; ObamaonlydancedwithMichelle,andRomneyonlydancedwithAnn.Thisisthecumu- lativeinterpretationof(16).Crucialforourpurposesisthatthemeaningof(16)doesnot requirethateachofObamaandRomneyhavethepropertyofdancingwithbothMichelle andAnn. Consider now this sentence embedded in an Amharic speech report with a shifted in- dexical. (17) 1tS’tS’u-wotStS-u kä-Michelle-na Ann gar 1nn1-däns-all-än candidate-PL.DEF with-Michelle-and Ann with 1PL-dance.PF-AUX-1PL al-u say.PF-3PL ‘ThecandidatessaidwewilldancewithMichelleandAnn.’ Applyingthesingulariststrategyoutlinedin(12)tothiscasegivestruthconditionsthatare toostrong;theysaythateachofthecandidateswanttodancewithbothwomen. (18) x:x Obama,Romney [xsaidthatxwilldancewithMichelleandAnn] ∀ ∈{ } Whilethisisapossibleinterpretationof(17),itisnottheonlyinterpretation.Thesetruth conditionsincorrectlypredictafalsejudgmentinacumulativesituation,whereeachcan- didate said that he wants to dance only with his wife and no one else. If the value of the shiftedindexicalwerepluralinthiscase,theseproblemswouldnotarisesincelocalplural valueslicensecumulativeinterpretations.8 7Beck&Sauerland(2000);p.350,usethephrase“cumulativeinterpretation”toreferto“...allcaseswhere asentencecontainingtwopluralshastruthconditionsweakerthanthoseofadoublydistributiveparaphrase”. Thisdefinitionwillsufficeformostofthesentencesdiscussedhere,thoughseeSchein(1993)andSternefeld (1998)forgeneralizednotionsofcumulativeinterpretationsuitableforcaseswithanarbitrarynumberof pluralnounphrases. 8Again,therehavebeentreatmentsofsimilarcasesthatpositasingularvariable;suchasBeckandSauer- land(2000).Theymakeuseofanon-lexicaltheoryofcumulativitythathasbeencriticizedbyauthorslike Schein(1993)andKratzer(2005).SeeLaTerza(2014)forfurtherdiscussionandcounterexamples. Pluralshiftedindexicalsareplural 265 4. Semanticanalysis InthissectionweprovideabriefsketchofthemodificationstoexistingtheoriesofDeSe reportsneededtoaccountforthesefacts.Wewillstartwithanoverviewofsomefundamen- talideasfromthesemanticsofpluralitywhichareneededforthemodificationstoattitude verbandindexicalsemanticsthatfollow. 4.1 Plurality We assume the basic tenets of mereological theories of plurality, as found in Link (1983) andLandman(2000),forexample.Onecoreidea,foundinalmostanytheoryofplurality, is that the denotation of predicates is closed under sum formation. Below we provide a generalprinciplethatencapsulatesthisideaforpredicatesofanyadicity.Iusethestandard notationalconventionoftreatinguppercasevariablelettersasnumber-neutral,whilelower casevariablescanonlybeassignedsingular/atomicvalues. (19) CumulativityPrinciple IfRisann-aryrelationandboth X ,...,X and Y ,...,Y areinR’sdenotation, 1 n 1 n ⟨ ⟩ ⟨ ⟩ thensois X Y ,...,X Y . 1 1 n n ⟨ % % ⟩ Thisprinciple,amongotherthings,ismeanttoaccountforinferenceslike(20). JohnkissedMary. BillkissedSue. (20) JohnandBillkissedMaryandSue. ∴ Acentralobservationaboutpluralityinnaturallanguageisthatitcomesindistributive and collective varieties. For the purposes of this paper, we follow authors such as Link (1984), Roberts (1987), and Lasersohn (1998) in taking collective predication to be ba- sic,anddistributiveinterpretationsasderivedbythepresenceofaDistributivityoperator, informallydefinedbelow. (21) TheDistributivityOperator Foranyone-placepredicatePandsumofindividualsX:DPholdsofX iffPholds ofeachatomicpartxofX. Toanticipateourproposalsomewhat,weproposethatthedistributive/collectivecontrastis atworkinattitudereportsandcanbeusedtoaccountforsomeofthepuzzlingproperties ofpluralDeSereports. 266 LaTerza,Kramer,Rood,Chacón,&Johnson 4.2 Attitudeverbsandpluralpredication We assume a context-based theory of attitude reports based on the work of Schlenker (1999) and Schlenker (2003). Formally, a context c is a tuple c ,c,c where c is the a t w a ⟨ ⟩ author/speaker of c, c is the time of c, and c is the world of c. Schlenker uses contexts t w toreplacepossibleworldsintraditionaldenotationsofattitudeverbs;seeHintikka(1969) andvonFintel&Heim(2012). (22) believe c=λp.λx. c′ DOX(x,cw)[p(c′)] ∀ ∈ (23) say c=λp.λx. c′ SAY(x,cw)[p(c′)] ! " ∀ ∈ CentraltotheseHintikkanapproachestoattitudeverbmeaningisroleofaccessibility ! " relations, which are used to determine a set of alternative contexts relative to an attitude holderanevaluationworld.Eachattitudeverbisassociatedwithacorrespondingaccessi- bilityrelation. (24) DOX(x,w)= c:ciscompatiblewithwhatxbelievesinwandxisca} { (25) SAY(x,w)= c:ciscompatiblewithwhatxsaysinwandxisca} { Our first step is to define a notion of accessibility that is suitable for a plurality of attitude holders. If we are to define a set of contexts that are accessible to Obama and Romney,therearereallytwopossiblewaysofdoingso:itiseithertheintersectionorunion ofeachsingularattitudeholder’srespectsetsgivenbysingularrelationslike(24)and(25). We believe that the union/sum operation is the appropriate choice given the existence of dependent readings. The plural SAY-accessibility relation thus has the form of (26); (27) illustratesthisaccessibilityrelationputtoworkwithourparadecase(1). (26) SAY(X,w)= c: x[x X &ATOM(x)&ciscompatiblewithwhatxsaysinwand { ∃ ≤ xisc ]} a (27) SAY(o r,w)= c:[ciscompatiblewithwhatObamasaidandObamaisca]or[c ⊕ { iscompatiblewithwhatRomneysaidandRomneyisc ]} a Wenowturntothemodificationsofattitudeverbmeanings.Westartwiththeobserva- tionthattheHintikkanapproachtoattitudeverbmeaningcanberecastwithadistributivity operator;insteadofuniversalquantificationoverasetofaccessiblecontexts,wehavedis- tributivepredicationoverthesameset.ForanyattitudeverbAVandaccessibilityrelation R,thiscanbedoneasfollows. (28) AV c=λp.λX.Dp(R(X,c )) w So far this is just a notational variant of a standard idea. However, treating attitude verb ! " meaningsasatypeofpluralpredicationopensupthepossibilitythatothertypesofplural predicationmaybeatwork.ThisisourproposalaboutDeSeattitudeverbs;specifically,the denotationofDeSeattitudeverbslackadistributivityoperator,andcollectivepredication ofcontextsisthuspossible. Pluralshiftedindexicalsareplural 267 (29) AV c=λp.λX.p(R(X,c )) DeSe w What does it mean to have collective predication of contexts? For one, it means that a ! " predicate may be collectively satisfied by a sum of contexts in a way that no singular contextcando.Forexample,ifthepredicateisλC.[theydestroyeachotherinC],itmay be satisfied by the totality of Obama and Romney’s desire contexts taken together, even though that same predicate does not hold of any single accessible context that is a part ofC. Likewise,λC.[they dance with Michelle and Ann inC] can describe the collection of Obama and Romney’s desire alternatives, even though each candidate by himself just wantstodancewithasinglewoman. Asthelaststepofourproposal,weturntothesemanticsoftheshiftedindexicalitself. WefollowSchlenkerintreatingthemeaningof(first-personal)shiftedindexicalasbeing determinedbytheauthorcoordinateoftheevaluationcontext. (30) shiftedindexical c=theauthorofc We propose for plural shifted indexicals that the context parameter is pluralized; it is a ! " sum of accessible contexts. This in turn means that the coordinates of the contexts are themselvessums.Forthepluralindexicalthen,itsvalueisapluralityofauthors. (31) pluralshiftedindexical C=theauthorsofC Withtheseproposalsinplace,wenowprovideasimplifiedderivationofthetruthconditions ! " of(1). (32) ObamaandRomneysaidWEwillwin C = saidWEwillwin C(o r) ! ⊕ " = said C(o r)(λC′. WEwillwin C′) ⊕ ! " = said C(o⊕r)(λC′.WIN(Ca′,Cw′)) ! " ! " = [λp.λX.p(SAY(X,Cw))](o⊕r)(λC′.WIN(Ca′,Cw′)) ! " = [λC′.WIN(Ca′,Cw′)](SAY(o⊕r,Cw)) = TrueiffλC′.WIN(Ca′,Cw′)holdsofthesumofObamaandRomney’s compatibleSAY-contextsinCw. = TrueiffthesumofObamaandRomney’sSAY-contextsaresuchthatthe authorsofthosecontexts(cumulatively)winintheworldsofthosecontexts. Thelastlineofthederivationrepresentsthetruthconditionsof(1)inprose.Wenowturn tohowthesetruthconditionsaresatisfiedindependentandgroupsituations,andhowthey arenotsatisfiedincrossedsituations. We propose the novel hypothesis that both the dependent and group readings can be satisfied with the same truth conditions; there is no ambiguity, but underspecified truth conditions.ThisispossiblebytheCumulativityPrinciple(19)appliedtotheinterpretation of WIN(Ca,Cw), which says that the authors of C cumulatively win in the worlds of C. 268 LaTerza,Kramer,Rood,Chacón,&Johnson This can be done for example if Obama wins on Obama worlds, and Romney wins on Romney worlds. This predicate can also be satisfied if both Obama and Romney win on Obamaworlds,andObamaandRomneywinonRomneyworlds.Thetruthconditionsfor (1)aresatisfiedif,attheminimum,theauthorswinintheirrespectiveworlds;thereisno restrictionthattheyaretheonlywinners. ThehypotheticalcrossedreadingisruledoutbyappealingnotjusttoObamaandRom- ney,buttoObamaandRomneyasauthors.Thisqualificationisindependentlyrequiredby DeSepropertiesofthereport.Inthecrossedreading,itistruethatObamaandRomneycu- mulativelywininObamaandRomney’sworlds,butitisnottruethatObamaandRomney asauthorsareonesthatarewinning(inthissituation,theauthorsarelosing).Forreasons ofspacediscussionofthisiscutshort,andagainwereferthereadertoLaTerza(2014)for furtherdetails,buttheessenceoftheclaimisthatwhatmakesthesereportobligatorilyDe Seisalsoresponsibleforthelackofcrossedreadings. 5. Conclusion In this paper we have introduced plural shifted indexicals in Amharic. We have argued that these items should be interpreted as semantically plural, contrary to many previous studiesonpluralityinattitudereports.AnanalysiswasdevelopedthatmodifiedSchlenker’s contextbasedsystembyallowingDeSeattitudeverbstorepresentcollectivepredication ofcontexts.Withthiscentralproposaltogetherwithsomepluralrevisionsofaccessibility relationsandthedenotationofshiftedindexicals,wewereabletoprovideasketchofthe truth-conditional derivations of the range of interpretations available to Amharic attitude reportswithshiftedindexicals. References Anand,Pranav.2006. Dedese. Doctoraldissertation,MIT. Asudeh,Ash.1998. Anaphoraandargumentstructure:Topicsinthesyntaxandsemantics ofreflexivesandreciprocals. Master’sthesis,UniversityofEdinburgh. Beck, Sigrid, & Uli Sauerland. 2000. Cumulation is needed: a reply to winter (2000). NaturalLanguageSemantics8:349–371. Chomsky,Noam.1981. LecturesonGovernmentandBinding. MoutondeGruyter. Dalrymple,Mary,SamMchombo,&StanleyPeters.1994. Semanticsimilaritiesandsyn- tacticcontrastsbetweenchichewˆaandenglishreciprocals. LinguisticInquiry25:145– 163. Fiengo,Robert,&HowardLasnik.1973. Thelogicalstructureofreciprocalsentencesin english. FoundationsofLanguage9:447–468. vonFintel,Kei,&IreneHeim.2012. Lecturenotesonintensionalsemantics. Spring2011 edition,MIT. Heim,Irene,HowardLasnik,&RobertMay.1991. Reciprocityandplurality. Linguistic Inquiry22:63–101.