Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
The case of a person: The person case constraint in German
(USC Thesis Other)
The case of a person: The person case constraint in German
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
TheCaseofaPerson: ThePersonCaseConstraintinGerman by ThomasMichaelBorer Dissertation Presented to the Facultyofthe Graduate School of the University of Southern California InPartialFulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY (LINGUISTICS) December2018 Contents 1 Introduction 6 1.1 VariationinthePCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.2 SwissGermanPCC-TheProblem . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.3 Sketchingouttheanalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.4 StructureoftheWork . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2 WeakandStrongPronouns 12 2.1 Thestatusofstrongandweakpronounsintheliterature . . . . . . . . . . . 14 2.2 PronounsinStandardandSwissGerman . . . . . . . . . . . . . . . . . . . . 17 2.3 StandardGerman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.3.1 KindsofPronouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.3.2 TheLocationofPronouns . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.3.3 Pronoun(re-)orderingandscrambling . . . . . . . . . . . . . . . . . . 24 2.3.4 Diagnosingstrongpronouns . . . . . . . . . . . . . . . . . . . . . . . 30 2.4 SwissGermanPronouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 2.4.1 PronounsinSwissGerman . . . . . . . . . . . . . . . . . . . . . . . . 44 2.4.2 PronounOrder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 2.4.3 Thelocationofreducedpronouns . . . . . . . . . . . . . . . . . . . . 49 2.4.4 Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.4.5 ModificationandtheStructureofPronouns . . . . . . . . . . . . . . 57 2.5 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 2.5.1 Pesetsky&Torrego(2007) . . . . . . . . . . . . . . . . . . . . . . . . . 61 2.5.2 ThestructureofPronouns . . . . . . . . . . . . . . . . . . . . . . . . . 62 2.5.3 ThestructureofDPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 3 PreviousLiterature 71 3.1 Anagnostopoulou(2003,2008,2015) . . . . . . . . . . . . . . . . . . . . . . . . 71 1 CONTENTS 2 3.2 Walkow(2013) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 3.3 Stegovec(2017) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 3.4 PanchevaandZubizarreta(2017) . . . . . . . . . . . . . . . . . . . . . . . . . 89 3.5 CliticLogophoricRestriction . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 3.5.1 Charnavel&Mateu(2015) . . . . . . . . . . . . . . . . . . . . . . . . . 96 3.5.2 TheCLRinPanchevaandZubizarreta(2017) . . . . . . . . . . . . . . 99 4 SwissGermanPCC 101 4.1 Thefulldataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 4.1.1 The“PCC”inSwissGerman . . . . . . . . . . . . . . . . . . . . . . . 101 4.1.2 Onthebanof3DAT>3ACC . . . . . . . . . . . . . . . . . . . . . . . . 103 4.1.3 LocatingthestructuraloriginsofthePCC . . . . . . . . . . . . . . . . 106 4.2 TheAnalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4.2.1 DerivingtheOptionality . . . . . . . . . . . . . . . . . . . . . . . . . . 111 4.3 ThepurposeofPer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 4.3.1 TheSwissGermanData . . . . . . . . . . . . . . . . . . . . . . . . . . 113 4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 5 ThePCCinStandardGerman 116 5.1 TheCLR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 5.2 ThePCCinStandardGerman . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 5.3 ExperimentalData . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 5.3.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 5.3.2 PersonCaseConstraint-Experiment1 . . . . . . . . . . . . . . . . . . 120 5.3.3 LCR-Experiment2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 5.3.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 5.4 AnalysisforStandardGerman . . . . . . . . . . . . . . . . . . . . . . . . . . 132 6 DerivingtheVariationinthe PCC 136 6.1 DescribedVariation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 6.1.1 StrongPCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 6.1.2 Super-StrongPCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 6.1.3 Ultra-strongPCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 6.1.4 Me-firstPCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 6.2 PredictedVariation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 6.2.1 No-PCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 6.2.2 StrongPCCversion2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 CONTENTS 3 6.2.3 OnlyMePCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 6.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 7 Conclusion 144 Acknowledgements I want to give special thanks to my advisor Maria Luisa Zubizarreta. She spent years, teaching me in classes, guiding me in meetings, commenting on my drafts and pushing me towards and over the finish line. She pointed me toward the topic that eventually became this thesis and helped me keeping on track so that I could write this thesis. I also want to thank Roumyana Pancheva for being on virtually every committee that I ever formed at USC. She gave me insightful feedback on every one of them. I was lucky enough to sit in many of her classes also one on the PCC in which I formed the basis of the analysis presented here. She taught me enumerable things and made sure that I had the background to start on the quest that became this dissertation. I was extremely luckythatStefanKeinejoinedtheUSClinguisticsdepartmentwhenIwasworkingonthis thesis. His insightful questions and understanding of the matter shines through in the better parts of the present work. I am very grateful to Andrew Simpson who always had encouragementandaninsightfulperspectiveonmostthings,aperspectivethatwasmore than needed at times. Gabriel Uzquiano not only introduced me to the beauty of modal logicbutalsotookthetimetositthroughhoursofmypresentationsandlistentothemas anexternalcommitteememberandIammostgratefultohimforthis. Ialsowanttogive thankstoallofthefacultyintheUSClinguisticsdepartmentinparticulartoBarrySchein formanyagooddiscussion,toHajimeHojiandAudreyLiformanyanencouragingword andinterestingtalksandCananIpekforteachingmehowtodealwithbiggerdatasets. IalsowanttogivethankstoallofmyfellowgradstudentsatUSCwhosharedtheups and downs of graduate life with me. It is hard to imagine my life at USC without this fantastic crowed. I want to particularly mention my cohort mates Binh Ngo and Monica Do for the needed encouragement and uplifting words. I am very happy to have been a partoftheUSCSyntaxplusgroupwileatUSCandIamverygratefultotheitsmembers forallthegreatMondayafternoonstalkingaboutoursharedinterestinSyntaxandallthe feedbackandinspirationtheyprovided. 4 CONTENTS 5 This dissertation would not have been possible without Ulrike Steindl. She made suretoremovemostoftheobstaclessometimesbeforeIrealizedthattheymightbecomea problemandallowedmetorideinherdraft. Ilearnednearlyallofmytechnicalskillsfrom her, she functioned as a sanity check for most grammaticality judgments and proposals and helped me putting things in perspective when I was losing it. She gave me most valuablecommentsonnearlyeverypartofmywork. Iwouldnothavestartedwritingthis thesis let alone every finishing it if it would not have been for her. The contributions that she made to my work big and small are too numerous to name them all. I am extremely grateful for all that she has done and consider myself so very lucky that she was willing to spend the last decade with me and hope she will still want to spend time with me the future. I am also thankful to Lacey Schauwecker who regularly checked in to make sure that I am still alive on days otherwise spend in solitude and Sarah Signor, Brad Foley, Eliso Melero and Amanda Wagenbach who were always ready to spend an evening talking when I could not spend any more time writing and looking at a screen. I am very happy to know Sarah Ouwayda with whom I had the chance to spend many Sunday brunches talking about the joy of syntactic analysis and Zoe Wu with whom I had the chance to spendmanyaSaturdayafternoonhikingthroughthespectacularbeautythataretheSan Gabrielmountains. Chapter1 Introduction Various languages have restrictions on the co-occurrence of pronominal arguments. This isillustratedforFrenchin(1). Wecanseethata1stor2ndpersonobjectcliticcanco-occur withanaccusative3rdpersonclitic(1a)whileitcannotco-occurwitha3rdpersondative clitic (1b). Since the restrictions are dependent on the person but also case configuration oftheinternalarguments,Bonet(1994)labeleditthePersonCaseConstraint(PCC). (1) French a. Elle she te/me 2sg/1sg le 3masc.acc présentera. will.introduce ‘Shewillintroducehimtoyou/me.’ b. *Elle she te/me 2sg/1sg lui 3masc.dat présentera. will.introduce ‘Shewillintroduceyou/metohim.’ (Pancheva&Zubizarreta2017:ex(1)) Bonet (1991) realized that the there was some variability in judgments. While all the Catalan speakers accepted the contrast in (2), judgments varied for (3). We see in (2) a similar contrast for Catalan that we saw in (1) for French. A 3rd person accusative clitic canco-occurwitha1stor2ndpersondativeclitic(2a),thereverseisnottrue. A1stor2nd person accusative clitic cannot co-occur with a 3rd person dative clitic (2b). (3) illustrates that some Catalan speakers allow for both, the dative and the accusative clitic to be 1st and2ndpersonclitics,whileothersdonotallowthiscombination. (2) Catalan a. En the Josep, Josep, me 1sg.dat ’l 3sg.acc varecomanar recommended.3sg la the Mireia Mireia ‘She(Mireia)recommendedhim(Josep)tome.’ 6 CHAPTER1. INTRODUCTION 7 b. *A to en the Josep, Josep, me 1sg.acc li 3sg.dat varecomanar recommended.3sg la the Mireia Mireia ‘She(Mireia)recommendedmetohim(Josep).’ (Bonet1991:178) Speakerswhoaccept(3)asgrammaticalwerelabeledspeakersofaweakPCCvarietywhile speakerswhorejectsentenceslike(3)werelabeledspeakersofastrongPCCvariety1 . (3) %Te 2sg me 1sg recomendaron. recommended.3.pl Catalan a. ‘Theyrecommendedmetoyou.’ b. ‘Theyrecommendedyoutome.’ (Bonet1991:179) Later research into the PCC revealed that there is a much finer grained distinction in the PCC.Iwilldiscussthispointfurtherinsection1.1below. 1.1 VariationinthePCC Bonet (1991) points out that the PCC comes in two different variations, namely the weak and the strong PCC. The weak PCC stipulates that if a sentence contains a 3rd person clitic, this clitic has to be an accusative clitic. The strong PCC requires that the accusative clitic has to be 3rd person. Nevins (2007) extends this list of PCC variations by two further kinds of PCCs, the me-first and the ultra-strong PCC. To Nevins (2007), the me first–PCC is a restriction that requires any first person dative argument to precede any accusativeargument. Thismeansthatitonlyrestrictscombinationscontaining1stperson arguments. Such first-person arguments have to be dative arguments. The ultra-strong PCC is the combination of the weak PCC and the me-first constraint. In other words, the Ultra-strong PCC bans any combinations that are banned by the weak or the me-first constraint. This means that all the combinations that are either banned by the weak PCC or the me-first PCC are also banned by the ultra-strong PCC. Haspelmath (2004) also introduces the super-strong PCC. The super-strong PCC is like the strong PCC but also bansthecombinationstwo3rdpersons. Table1.1belowsummarizesthisdata. 1(3)hastwopossibletranslations. Thisisbecausetheorderofthedativeandaccusativecliticisnotfixed withrespecttoeachother. Theorderofclitic pronounswith respectto each otheris notpurely syntacticin naturebuthasaphonologicalcomponent across languages (cf. Anagnostopoulou 2015). CHAPTER1. INTRODUCTION 8 Table1.1: TableofknownPCCvariationfromPancheva(2016) DAT ACC weakPCC me-first Ultra-strongPCC strongPCC super-strong 1 3 2 3 3 3 x 1 2 x x 3 2 x x x x 2 1 x x x x 3 1 x x x x x We can see in the table above that only the super-strong PCC bans 3rd-3rd combi- nations. This does not capture the typological variation found in natural language, but ratherreflectstheanalysisprovidedfortheindividualphenomena. Nevins(2007)argues thatrestrictionson3rd-3rdcombinationsarenottrulyreflectionsofPCCrestrictions,but the result of independent language specific morphological restrictions on two internal arguments with the same person specification. A point I will further discuss in chapter 4. This is likely also the reason for why Nevins (2007) does not try to specifically account for Haspelmath’s (2004) super-strong PCC. However, it is important to note that classical Arabic, which Nevins (2007) uses as an example for the ultra-strong PCC, does not allow for combinations of two 3rd person internal argument clitics (Ouwayda 2016). Walkow (2013:248), who also gives an analysis of the ultra-strong PCC using central Catalan and classicalArabicdata,definestheultra-strongPCCasin(4). (4) TheUltra-StrongPerson-CaseConstraint: Twointernalargumentscliticizeif a. Thehigheronehasamorelocalpersonspecificationthanthelowerone, b. where: 1 > local local2local > local 3 This means that the variation in table 1.1 might well underrepresent the variation in the PCCbutanyanalysisofthePCChastobeabletocaptureatleastthevariationintable1.1. 1.2 SwissGermanPCC-The Problem Bonet(1991)describedthePCCinanumberoflanguagesandwasalsothefirsttodescribe it in Swiss German. Swiss German is a particular case in the discussion for two reasons. The PCC in Swiss German is dependent on the order of the pronominal elements. It is CHAPTER1. INTRODUCTION 9 alsodifferenttoanumberofotherdescriptionsofthePCCinthatiteffectsweakpronouns andnotonlyclitics. (5)illustratesthatpronounsinSwissGermancancomeintwoorders. Intheacc>datorderasin(5a)and thedat>acc orderasin(5b) (5) a. D’Maria theMaria zeigt shows en him mir to-me (ACC>DAT) b. D’Maria theMaria zeigt shows mir to-me en him (DAT>ACC) ‘Mariashowshimtome’ (Bonet1991:188) Thisfreeorderingofinternalargumentsisonlypossibleiftheaccusativeargumentis3rd person and the dative argument is 1st or 2nd person. Note that this is a configuration where we do not see PCC restrictions cross linguistically as is illustrated by the check marks in the top two rows of table 1.1. If the dative argument is 3rd person and the accusative argument is 1st person, the order is stricter. In (6) we see that ACC>DAT is possible (6a) but that DAT>ACC is only possible if the accusative object is a strong pronounasin(6b)butnotifitisa weakpronounsasin(6c). (6) a. D’Maria theMaria zeigt shows mi me em to-him (ACC>DAT)/(clitic) b. D’Maria theMaria zeigt shows em to-him miich me (DAT>ACC)/(strong) c. *D’Maria theMaria zeigt shows em to-him mich me (DAT>ACC)/(weak) ‘Mariashowsmetohim (Bonet1991:188) The small data set above shows that Swiss German clearly exhibits a kind of PCC. The restriction is dependent on the case and person of the pronoun combination. We are looking at a restriction against a 3rd person dative pronoun and a first person accusative pronoun to co-occur. This restriction has been observed in many other languages (e.g. Greek, French, Spanish, Arabic). The fact that strong pronouns do not participate in the constraint, as is illustrated in (6b) is not surprising in this context. Strong pronouns are not affected by the PCC cross linguistically (cf. Bonet 2008). I will discuss the difference between weak and strong pronouns in chapter 2 below and will give an analysis which accountsforthedifferentbehavior. Basedonthedatasetabove,SwissGermanisgenerallyviewedasalanguagethathas the PCC (c.f. Béjar & Rezac 2003). The data set has also been used to argue for or against particular points. Albizu (1997) argued based on the ordering restrictions on the PCC CHAPTER1. INTRODUCTION 10 in Swiss German, that the PCC has to be a syntactic phenomenon and cannot be purely morphological in nature. Anagnostopoulou (2003) argues that the influence of the order of the arguments in Swiss German can be viewed as a more general fact, namely that the PCC can only occur when the dative argument c-commands the accusative argument in therelevantlevelofrepresentation2. There are some questions raised through the data set above. Contrary to many other languages discussed in the PCC literature, the PCC phenomena in Swiss German is not restricted to clitics but also involves weak pronouns. This is demonstrated in (6) where we see the accusative pronoun in three different forms. This is important since some researcherstiedthePCCstronglytocliticisation(e.g.Walkow2013). InSwissGermanwe are dealing with two distinct pronominal forms, at least phonologically: weak pronouns, and clitic pronouns. Both forms trigger the PCC. I will demonstrate in section 2.4 that bothformsbehaveinfactlikeXPsandthattheso-calledcliticsarephonologicalcliticsbut notsyntacticclitics,sincetheycannotattachtoanotherhead. Iwillthereforerefertoboth forms collectively as reduced pronouns or weak pronouns. In my examples the so-called clitic forms will predominantly occur since they are more clearly marked as a reduced form. The fact that Swiss German does have the PCC with weak pronouns makes the language particularlyinterestingalsoforanotherreason. Anagnostopoulou (2015) argues that the order of clitics is not necessarily governed by syntax but might be the result of phonological constraints. She argues that the order of weak pronouns, on the other hand, is the result of syntactic operations. The order of pronounsisthereforemoreinformativeifwelookatweakpronounsandnotclitics. The other aspect why Swiss German is of particular interest is that the order of the pronounscorrelateswiththePCC.IwillarguethatSwissGermancangiveusanunusual insightintothenatureofthePCC becauseofthis. 1.3 Sketchingouttheanalysis I will follow authors like Anagnostopoulou (2003) and Walkow (2013) in arguing that the PCC is the result of an intervention effect and like Pancheva & Zubizarreta (2017) I will argue that the PCC is triggered by interpretable person features. Diverging from previous authors, I will show that these interpretable person features are not bound to a particularheadbutthatpartoftheobservablevariationinthePCCacrosslanguagesand withinlanguagesisduetotheparticularlocationoftheinterpretablepersonfeatures. The 2Bonet(2008)respondsthatsurfacewordordercannotbetheonlyrelevantfactoracrosslanguagessince theaccusativecliticprecedesthedativeclitic in Catalan, a language which still has the PCC. CHAPTER1. INTRODUCTION 11 personfeaturesprobeintotheirdomainandiftheprobefindsmultiplepotentialgoalsin itsc-commanddomainitwillAgreestrictlywiththeminimallyclosestgoalbutwillsearch forthegoalthatsatisfiesthemostfeatures. Thisisbasedonaconditionfrom(VanUrk& Richards2015:132),whichtheycallMultitasking,definedin(7)below. (7) Multitasking: Ateverystepinaderivation,ifaprobecantriggertwooperationsAandB,andthe features checked by A are a superset of those checked by B, the grammar prefers A. The PCC is therefore more generally the result of a universal Agree mechanism that considersmultiplepotentialgoalsandlanguagespecificconsiderationsonheadscarrying unvaluedpersonfeatures. Iwilldiscusstheproposalindetailinchapter4anddemonstrate how it can capture the ultra-strong PCC and order sensitivity in Swiss German. I will thenturntothePCCinStandardGermananddemonstratehowtheaccountcapturesthe weak PCC in Standard German chapter 5, before illustrating how the account can derive therange ofvariationintable1.1. 1.4 StructureoftheWork ThisworkwilldealpredominantlywithdatafromStandardGermanandSwissGerman. In chapter 2 I will discuss in a first step how to distinguish strong pronouns from weak pronounsinbothlanguagesandwhatthistellsusaboutthedifferencesbetweenthesetwo kindsofpronouns. Thisisparticularlyimportantsinceonlyweakpronounsareaffectedby thePCCandunderstandingthedifferencesbetweenthetwokindsofpronounsallowsus tounderstandwhystrongpronounsare“immune”againstthePCC.Iwillthenintroduce theliteraturethatIwillrelyonmostforthediscussionandanalysisofthePCCinchapter 3. Chapter 4 will be divided in two parts. I will introduce in a first step the relevant data demonstratingthekindofPCCthatSwissGermanhasandthenturntotheanalysisofthe PCCinSwissGerman. IwillthenprovideexperimentalevidenceshowingthatStandard German also has the PCC and illustrate how my account can capture the difference between Swiss German and Standard German. In chapter 6 I will then demonstrate how my analysisfor SwissGerman andStandard Germancan account forthe variationfound inthePCCacrosslanguages. Chapter2 WeakandStrongPronouns The PCC across languages depend on the nature of the pronouns. Many languages have multiple parallel pronouns systems and the PCC effects only one of them. We can see in (8) bellow the PCC in Catalan which are Bonet’s (2008) ex. (2) and (1) respectively. The 3 rd person accusative clitic in (8a) can co-occur with a 1 st person dative clitic. However, a 3 rd person dative clitic cannot co-occur with a 1 st person accusative clitic as is illustrated in(8b). (8) Catalan a. El the director, director, me medat l’ him.acc ha has recomanat recommended la the Mireia. Mireia ‘Asforthedirector,Mariahasrecommendedhimtome.’ b. *Al the director, director, me meacc le him.dat ha has recomanat recommended la the Mareia. Mireia ‘Asforthedirector,Mireiahasrecommendedhimtome’ The PCC is not dependent on reference. This is illustrated in (9) which is Bonet’s (2008) (6). We see that co-occurrence of two clitics in (9a) is ungrammatical, but if the dative is expressed as a proper name (9b) or a strong pronoun (9c), the sentence is grammatical. Notethatthedativeargumentinallthreeexamplescanhavethesamereferent. (9) Catalan a. *Me me.acc le 3sg.dat recomendó. recommmended.3sg ‘Herecommendedmetohim’ b. Me me.acc recomendó recommmended.3sg a to Pedro. Pedro ‘HerecommendedmetoPedro’ 12 CHAPTER2. WEAKANDSTRONGPRONOUNS 13 c. Me me.acc recomendó recommmended.3sg a to él. him ‘Herecommendedmetohim’ Bonet (1991) argues that the PCC is universal with weak pronouns and clitics. Across languages,lexicalDPs/NPsaswellasstrongpronounsseemtobeunaffected(cf.Stegovec 2017). Wehaveseeninsection1.2thatthisisalsotrueforSwissGerman. Iwillprovidea formalanalysisofthedifferencebetweenstrongandweakpronounsinStandardGerman andSwissGerman. Iwillproposefollowing,Cardinaletti&Starke(1994),thatthedifferencebetweenweak and strong pronouns is a difference in structure. I will more concretely argue that strong pronouns are DPs with an expended structure while weak pronouns are pronouns that haveaminimalstructure. IwillproposethatbothcontainanelidedNPwhichisconsistent withElbourne’s(2001)analysisofdonkeypronouns. (10) strongpersonalpronouns DP ... nP NP n ... D (11) weakpersonalpronouns nP NP n WeakandstrongpronounshavethesamefeaturesandthesamepotentialnP.Aweak pronoun becomes a strong pronoun by the additional structure. I make the assumption that as soon as nP merges with other elements within the nominal domain that such a structure will require to be embedded in a full DP, a requirement that does not exist for the structure of a weak pronoun in (11). The different behavior of weak and strong pronouns is than, that strong pronouns can satisfy all the feature checking requirements of within the DP while weak pronouns need to check these features within vP and TP. The advantage of this analysis is that it captures the intuition that strong and weak pronouns are fundamentally similar, while weak pronouns lack something that lexical DPsandstrongpronounshave. ItfurthermoreaccountsforthedistributionoflexicalDPs andstrongpronounsincontrasttoweakpronouns. In the next section, I will introduce Cardinaletti & Starke’s (1996) proposal that the difference between strong and weak pronouns, before turning to Standard German and CHAPTER2. WEAKANDSTRONGPRONOUNS 14 Swiss German. The difference between the two varieties is that weak pronouns in Swiss German are less restricted in their surface location and better phonologically marked than weak pronouns in Standard German. I will use these sections also to introduce the pronominalsystemofbothlanguages,whichwillberelevantinthediscussionofthePCC inlaterchapters. 2.1 Thestatusofstrongandweakpronounsintheliterature Cardinaletti & Starke (1994) point out that personal pronouns do not behave homoge- neously. Strongandweakpronounssurfaceindifferentformsinsomelanguages. (Cardi- naletti & Starke 1994:41) give the Italian example in (12) which is their (1) to demonstrate thedifference. essein(12a)canhaveahumanaswellasanon-humanreferencebutcannot becoordinated. loroin(12b)canonlyhaveahumanreferentandcanbecoordinated. (12) Italian a. Esse 3.pl.fem.nom (*e and quelle those accanto) besides sono are troppo too alte. tall/high +human/ -human b. Loro 3.pl.fem.nom (e and quelle those accanto) besides sono are troppo too alte. tall/high +human/*-human (Cardinaletti&Starke1994:41) As the gloss above indicates, esse and loro ‘they’ in (12) seem to fulfill the same role. TheybothtranslatetothesamepronouninEnglish. Buttheyarenotonlyphonologically distinct,theyalsobehavedifferentlyfromeachothersyntactically. Whilelorocanhaveonly human reference, esse can have human or non-human reference. (12) also demonstrates that loro can be coordinated with a lexical DP while esse cannot. Cardinaletti & Starke (1994) call the esse type pronouns deficient, since they lack certain abilities in contrast to the loro type pronouns, which they call strong. Cardinaletti (1999) also call deficient pronouns weak pronouns, both terms are used in the literature. Some authors call the samesetofpronounsalsow-pronouns(e.g.Anagnostopoulou2015). Iwillusetheseterms somewhatinterchangeablethedissertation,thechoiceofwordwilldependsomewhaton theliteraturewhichIwillrelyonforaparticularsection. While Cardinaletti & Starke (1994, 1996, 1999a) argue that strong pronouns need to have human reference, I will demonstrate that this is not true for German, as is also pointedoutinBayer(1999). CHAPTER2. WEAKANDSTRONGPRONOUNS 15 Cardinaletti & Starke (1994) demonstrate that across languages, strong and weak/de- ficient pronouns are not always phonologically distinct. Strong and deficient pronouns behave distinct from each other even in languages where both classes are phonologically identical. Thisisimportantforthis worksinceGermanisamongtheselanguages. Cardinaletti & Starke (1994, 1996, 1999b) propose that there is at least a three-way distinctionbetweenpronominalelements(13). The basic idea is that clitic pronouns are deficient heads, while weak pronouns are deficient XPs. Only strong pronouns are not deficient in this view. Deficient referential elements are missing properties that regular heads or XPs have. The idea of deficient heads in (13) will not be applicable to this work since neither German nor Swiss German has clitics in the same way as many Slavic or Romance languages do. I will come back to this point in the discussion of Swiss German pronouns in section 2.4. The notions in Cardinaletti&Starke(1996)aresummarizedin(13). (13) a. cliticelementsaredeficientX 0 . b. weakelementsaredeficientXPs. c. strongelementsarenon-deficientXPs. (Cardinaletti&Starke1996:36) Cardinaletti & Starke (1994, 1996) argue that the difference between strong and weak pronouns is based in the structure. The structures that they give are given in (14).The subscripted L stands for a lexical head. The idea is that the I in (14c) is not a full lexical element, whereas the I is a non-defective element in (14b) and in (14a). The difference betweenthestrongpronounandtheweakpronounsistheexistenceofC L in(14a)whereas weakpronounsonlyhavefunctionalprojections(P). (14) a. StrongPronouns CP L P L IP 1 LP I L 0 1 0 C L 0 b. WeakPronouns P L IP 1 LP I L 0 0 c. CliticPronouns IP LP I 0 I will adopt a proposal in the general vein of Cardinaletti & Starke (1996). I will CHAPTER2. WEAKANDSTRONGPRONOUNS 16 proposethatdifferentkindsofpronounsareembeddedindifferentsizestructures. Strong pronounshavemorestructurewhiledefectivepronounshavelessstructure 1. Weak pronouns behave differently from strong pronouns and lexical DPs because they are not DPs. Cardinaletti & Starke (1994) give a range of ways in which strong and defective pronouns behave differently from each other. I will use these characteristics as diagnosticsforwhetherpronounsarestrongordefective. Cardinaletti&Starke(1994)use thesecriteria: • onlystrongpronounscanbecoordinated. • onlystrongpronounscanbetopicalized. • onlystrongpronounscanbemodified. • strongpronounsanddefectivepronounsdonotsurfaceinthesameposition. All of the characteristics above point towards structural, syntactic difference between strongandweakpronounsandcanbecapturedbytheideathatstrongandweakpronouns differininternalstructure. Iwillelaborateonhowtheanalysiscapturesthesedifferences in section 2.5. I will use these four characteristics to diagnose first standard German pronounsinsection2.3andlaterSwissGermanpronounsinsection2.4. I will show that the distinction between strong and defective pronouns captures an importantcontrastinStandardGermanandSwissGerman. Oneofthedifferencesbetween the two variates is that Swiss German marks deficient pronouns more consistently in its phonologicalshapewhichmakesidentifyingthemeasier. Whileweakpronounsaremorerestrictedintheirdistribution,theyaremorefrequent. Cardinaletti&Starke(1996:48)proposetherulein(15). (15) Choiceofapronoun Choosethemostdeficientpossibleform 1Müller (2002) has an in-depth discussion of German pronouns and has a finer, four-way distinction of pronouns in Standard German. He aligns these four kinds of pronouns (stressed, human reference, non-humanreferenceand3 rd personneuter pronoun) on the scale in table 2.1 Table2.1: PersonalPronounScalefromMüller(2002) Pron strong > Pron unstressed > Pron weak > Pron reduced IHN[+stress] ihn[+anim] ihn[-anim] es I believe that Müller (2002) makes many valuable observations. However, his finer grained distinction is not relevant for the PCC and can be broken down to a two way distinction between strong and weak pronouns. CHAPTER2. WEAKANDSTRONGPRONOUNS 17 The rule can be characterized as stating that deficient pronouns are the elsewhere case. Strong pronouns only surface in situations where deficient pronouns or clitics are un- available (cf. López 2009). This seems to be true cross-linguistically even though I am nearlyexclusivelyusingexamplesfromEuropeanlanguages. Onecanseeverynicelyhow deficient pronouns are the elsewhere case in Newmann’s (2000) description of Hausa, a west Chadic language. Newmann (2000:477-478) gives what might be an exhaustive list ofcaseswherestrongpronounsareused,andcliticscannotbeused. • incoordinations • focusedortopicalizedelements • whennotfollowingtheverb (i.e.thepositionofclitizisedpronouns) • withinprepositions • determinersofanimatenounphrases. • sentenceswithoutverbs(e.g.copulaconstructions) We will see that strong pronouns in (Swiss) German appear in the same position and instances as strong pronouns in Hausa - with the exception of verb less sentences which donotexistinStandardGermanorSwissGerman. 2.2 PronounsinStandard and Swiss German Itisclearthatthedistinctioninpronounsisnotjustabinaryonebutmuchfinergrained. Cardinaletti (1999) points out that not all Romance clitics behave alike and that weak pronouns in standard German behave like Romance clitics in some respects. Müller (2002) establishes five categories of pronouns for Standard German, which have different phonologicalorsyntacticproperties. Cooper(1994)distinguishesfourkindsofpronouns inSwissGermanwhichhavedifferentphonologicalproperties. The distinction between strong pronouns and the rest, between pronouns that are not affected by the PCC and the pronouns that are affected by the PCC is important for our concerns. I will introduce here the finer grained distinction in the pronoun system of the two languages before collapsing this finer grained distinction into a binary one which is relevantforthePCCinbothlanguages. Across all languages that I looked at, the PCC does not affect independent/strong pronounsandr-expressionsbutonlytheweak/boundforms. CHAPTER2. WEAKANDSTRONGPRONOUNS 18 2.3 StandardGerman 2.3.1 KindsofPronouns TherearetwosetsofpronounsinstandardGerman. Thefirstsetofpronounsare‘regular’ personal pronouns that authors like Müller (2002) or Cardinaletti & Starke (1996) discuss. Thissetof pronounsislistedintable2.22. . Table2.2: ‘Personalpronouns’inGerman sg nom dat acc 1 st ich mir mich 2 nd du dir dich 3 rd masc er ihm ihn fem sie ihr sie neut es ihm es pl nom dat acc 1 st wir uns uns 2 nd ihr euch euch 3 rd sie ihnen sie The other set of pronouns, listed in table 2.3, are clearly derived from definitearticles. I will refer to them as d-pronouns following Wiltschko (1998)3 D-pronouns have the same form as the articles in German but have pronominal function. Abraham (2007) observes that d-pronouns only occur with focus emphasis. Some occurrences of the d-pronoun mighthaveademonstrativefunction. ThroughoutmydiscussionIwillonlyconsidercases whicharefelicitousalsoincaseswherethereferentisnotpresentatthetimeofutteranceto ensurethatthatwearenotdealingwithapronounthatfulfillsademonstrativefunction. Table2.3: ‘D-pronouns’ sg nom dat acc 3 rd masc der dem den fem die der die neut das dem das pl nom dat acc 3 rd die denen die 2Idonotaddthegenitivepronouns. Theformofgenitivepronounsdependsonthegenderandthecase oftheheadnoun. Iamnotgoingtobetalkingaboutgenitivepronounsinparticularandaddingthemhere wouldintroduceunnecessaryconfusion 3Abraham (2007) calls the same set of pronouns article pronouns, while Patel-Grosz & Grosz (2017) call thissetofpronouns demonstrativepronouns. CHAPTER2. WEAKANDSTRONGPRONOUNS 19 Iwilldemonstratethatthedifferencebetweenthesetwosetsofpronounsisthatwhile personal pronouns can be strong or weak/deficient pronouns, d-pronouns will always behavelikestrongpronouns. 2.3.2 TheLocationofPronouns There are three things to consider when it comes to the location of pronouns. One is the location of pronouns in contrast to the location of lexical DPs. It is widely accepted that weak pronouns and clitic do not surface in the base position of their lexical DP counterparts (cf. Abraham 1996). It has also been observed that strong pronouns do not have to undergo the same kind of movement e.g. (Müller 2002). I will first illustrate in section2.3.2thatweakpronounsandstrongpronounsdonotappearinthesameposition before turning to the question where weak pronouns move to in section 2.3.2. In section 2.3.3 I will address the question of weak pronoun order in Standard German and how it relatesto orderoflexicalDPs. I will start this section by laying out some assumptions about relevant diagnostics in StandardGermanwhicharealsoapplicabletoSwissGerman. Diagnosingpronounlocation It is commonly assumed that modal particles (MP) and adverbs are stable and do not undergo movement in German. This means that word order differences between ad- verbs/MPs on the one hand and other elements such as DPs are due to movement of the DPs,andnotoftheAdverbs/MPs. ThisfactisusedinJäger(2001)(buildingonDiesing’s (1992) work) who argues that certain readings in German are present or absent in certain DPmovements. Whilelowsubjects,wherethesubjectremainsinspecvP,suchasin(16a) and (17a), only have an existential reading, movement of the subject to spec TP triggers a generic reading as in (16b) and (17b). Predicates that allow only for generic but not for existential readings are therefore ungrammatical with low subjects as is demonstrated in (17a). (16) a. (weil) (since) angeblich allegedly Feuerwehrmänner firemen verfügbar available sind. are ‘Allegedly,firemenareavailable’(onlyexistentialreading) b. (weil) (since) Feuerwehrmänner firemen angeblich allegedly verfügbar available sind. are ‘Allegedly,firemenareavailable’(onlygeneric) (17) a. *(weil) (since) angeblich allegedly Feuerwehrmänner firemen selbstlos altruistic sind. are CHAPTER2. WEAKANDSTRONGPRONOUNS 20 b. (weil) (since) Feuerwehrmänner firemen angeblich allegedly selbstlos altruistic sind. are ‘Allegedly,firemanarealtruistic.’ (Jäger2001:86) The whole observation is based on the idea that adverbs cannot move and are therefore a good indicator for the movement of other constituents. In other words, if the subject precedes an adverb or MP it is due to movement of the subject. If the adverb or MP precedes the subject it is because the subject did not move to spec TP and remains in spec vP. I assume here that the same is true for Swiss German varieties and that the positionofmodalparticlesandadverbsrelativeargumentsisagoodindicatorofargument movement4. Wackernagelmovement Cardinaletti & Starke (1994) point out that strong and deficient pronouns do not appear in the same position across language and neither do they appear in the same position in German. Thisisdemonstratedinthecontrastin(19)whichisadoptedfrom(López2009:68)for German. Thedifferencebetweentheweakpronounihn‘him’in(19a)and(19b)incontrast to the strong pronoun in (19c) is that ‘him’ is not contrasted with another element while sie ‘she’ in (19c) is. Sie ‘her’ in (19c) is in contrast with some other salient referent. (19) demonstrates that not all pronouns appear in the same position. I will later demonstrate that the lower position in (19b), (19c) is reserved for strong pronouns while defective pronounshavetomovetowhatMüller(2002)referstoastheWackernagelposition. (19) Context: John and Mary go together to Mass on a daily basis. Did you see John yesterday? a. Ja, Yes, ich I.nom habe have ihn 1 him.acc gestern yesterday t 1 gesehen. seen. b. #Ja, Yes, ich I.nom habe have gestern yesterday ihn him.acc gesehen seen. ‘YesIhaveseenhimyesterday.’ 4In contrast to Jäger (2001) I don’t want to use subordinate clauses with weil - ‘since’ if avoidable since it isoneoftherarecaseswhereacomplementizer does not force a V final structure as (18) demonstrates. (18) Weil since angeblich allegedly sind are Feuerwehrmänner firemen verfügbar. available ‘Allegedly,firemenareavailable’ CHAPTER2. WEAKANDSTRONGPRONOUNS 21 c. Nein, No, ich I.nom habe have gestern yesterday (nur) (only) sie her.acc gesehen seen. ‘NoIhaveseen(only)heryesterday.’ I will demonstrate in the next sections that ihm ‘him’ in (19a) is a deficient pronoun while sie ‘her’ in (19c) is a strong pronoun. The deficient pronoun ha to move, (19b) is infelicitouswhilethestrongpronoundidnot(19c). Crucially(19b)isalsoungrammatical because ihn ‘him’ has to be a reduced pronoun and cannot be a strong pronoun since the sentence could be expressed with a reduced pronoun , therefore it has to be expressed with a deficient pronoun (cf. (15)). Cardinaletti & Starke (1994, 1996) demonstrate that most regular pronouns have a strong and deficient counterpart, but es ‘it’ only occurs as a deficient pronoun. I will furthermore show that derived pronouns do not pattern with deficientpronounsbutbehavelikestrongpronouns. Müller(2002)arguesthatunstressedpronounsinstandardGermanhavetomovetoa positioninbetweenC 0 andT 0 apositionhecallsWackernagelposition5. This is demonstrated in (20) - (21) which are based on Müller (2002) (7)-(8). Pronouns that do not carry phonological stress cannot remain in a lower position may they be IO as in (20) or DO as in (21). They can surface in the base position when the pronouns are stressedasin(20c)and(21c). (20) indirectobjectpronoun a. dass that ihr 1 her.dat der the Fritz Fritz.nom gestern yesterday t 1 ein a Buch book.acc geschenkt given hat has b. *dass that der the Fritz Fritz.nom gestern yesterday ihr 1 her.dat ein a Buch book.acc geschenkt given hat has c. dass that der the Fritz Fritz.nom gestern yesterday IHR 1 her.dat ein a Buch book.acc geschenkt given hat has (21) directobjectpronoun a. dass that sie 1 she.acc der the Fritz Fritz.nom gestern yesterday der the Maria Maria.dat t 1 geschenkt given hat has b. *dass that der the Fritz Fritz.nom gestern yesterday der the Maria sie 1 Maria.dat geschenkt she.acc hat givenhas c. dass that der the Fritz Fritz.nom gestern yesterday der the Maria Maria.dat SIE 1 she.acc geschenkt given hat has (Müller2002:209) 5I will address the question of stress later. I will not want to say that a strong pronoun is strong due to phonological stress and that phonological stress can make a weak pronoun into a strong pronoun, but the two tend to correlate since a contrastive pronoun has to be strong and will carry phonological stress. I will addressthisquestionmorein-depthinsection 2.5 CHAPTER2. WEAKANDSTRONGPRONOUNS 22 We can account for the variation in the examples above by Cardinaletti & Starke’s (1994) analysis of sie/ihn ‘heracc/dat’ as being homophones in that it can be the phonological realization of the strong or the defective pronoun in German. Cardinaletti & Starke (1994) also observe that es ‘it.neut’ can only functions as a defective pronoun but not as a strong pronoun. This is illustrated in (22). The pronoun ‘es’ has to move (22a) vs(22b). Phonological stress on the pronoun does not license a low position of ‘es’ (22c). This is in contrast to stress on her in (20c) and (21c) above. This is the first indication that stress is not responsible for turning a weak pronoun into a strong pronoun. The contrastive interpretation that a strong pronoun yields can be achieved by using the d-pronoun das ‘it.neut’. (22) a. dass that es 1 she.acc der the Fritz Fritz.nom gestern yesterday der the Maria Maria.dat t 1 geschenkt given hat has b. *dass that der the Fritz Fritz.nom gestern yesterday der the Maria Maria.dat es 1 she.acc geschenkt given hat has c. ??dass that der the Fritz Fritz.nom gestern yesterday der the Maria Maria.dat ES 1 she.acc geschenkt given hat has d. dass that der the Fritz Fritz.nom gestern yesterday der the Maria Maria.dat das 1 she.acc geschenkt given hat has (Müller2002:209) One might think that es ‘it’ is defective due to some semantic reason but this assumption isnotsupported. Thepronounesanddascanhavethesamereference. Whileesisalways a defective pronoun, das is never defective. I will elaborate on this point throughout this section, demonstrating that es behaves differently from other pronouns in standard Germanandthatthespecialstatus ofes doesnotholdforthed-pronoundas. ThelocationoftheWackernagelposition I will take a closer look at the location of the Wackernagel position before diving further in the syntactic properties of strong and defective pronouns. I have mentioned above that Müller (2002) locates the Wackernagel position initially between C 0 and T 0 . This is consistent with where second position clitics in Slavic languages are often presumed to reside(cf. Fuchs(2015)) Examples(20)to(22)demonstratethatsomepronounshavetoberealizedinaposition other than the base position of normal arguments - a position which I will call, following Müller(2002)theWackernagelposition. TheinterpretationthattheWackernagelposition CHAPTER2. WEAKANDSTRONGPRONOUNS 23 hastobelocatedbetweenC 0 andT 0 islessobviousinGerman. Thewholeargumentrests onthereasonableassumptionthatthedifferencebetween(23a)and(23b)isthatthesubject moved out from a vP internal position and that the temporal adverb gestern “yesterday” isbasegeneratedoutsidethevPshell6. ThisisverymuchinlinewithWurmbrand(2006) whodemonstratesthatnominativeargumentsinGermancanremainintheirbaseposition notonlyatPFbutalsoatLFanddonothavetoberealizedinspecTP. (23) a. dass that gestern yesterday der the Fritz Fritz.nom der the Maria Maria.dat ein a Buch book.acc geschenkt given hat. has b. dass that der the Fritz Fritz.nom gestern yesterday t 1 der the Maria Maria.dat ein a Buch book.acc geschenkt given hat. has ‘thatFritzgaveMariaabookyesterday.’ The relevant question for us is: where is the subject located in a sentence like (23b)? If subjects can only appear low in their base position or high in spec TP then everything preceding the higher subject has to be outside of Spec TP. However there seem to be multiplepositionsforsubjectstomoveinto. We see in the contrast in (24) that subjects can be realized after the moved pronoun (24a)andbeforethemovedpronoun(24b). (24) a. dass that es it.acc der the Fritz Fritz.nom gestern yesterday t 1 the der Maria.dat Maria given geschenkt has hat. b. dass that der the Fritz Fritz.nom es it.acc gestern yesterday t 1 the der Maria.dat Maria given geschenkt has hat. (Müller2002:209) Therearetwopossibleanalysesforthedataabove. Oneoptionisthatthesubjectcanmove toapositionprecedingorfollowingtheWackernagelpositionwhilethepronounremains inthesameposition. Theotheroptionisthatthesubjectremainsinthesameposition,but theobjectpronouncanmove. ThislaterviewisputforwardbyAnagnostopoulou(2015). I consider the view in which subject move more freely more parsimonious due to data such as (25). We can see in the example below that the nominative argument can surface inmanydifferentlocations. ItisclearthatsubjectsinGermandonotonlysurfaceintheir basepositionorSpecTPbutalsoinpositionsin-between. Itisthusreasonabletoassume that the contrast between (24a) and (24b) is also due to movement of the subject and not movementoftheobjectpronoun. 6TemporaladverbsareassumedtobebasegeneratedoutsideofthevPshellbecausetheytakescopeover the preposition and taking scope over and c-commanding are generally viewed to be one and the same (cf. Barker2012) CHAPTER2. WEAKANDSTRONGPRONOUNS 24 (25) a. dass that (der the Fritz) Fritznom es it.acc (derFritz) sicherlich certainly (derFritz) gestern yesterday (derFritz) der the Maria Maria.dat vorbeigebracht bring hat has thatFritzhascertainly broughtittoMariayesterday’ T in German is head final. It is therefore not clear where the specifier position is located. We know subjects can move to spec TP, the question is which of the many positions that the subject can surface in is spec TP? In other words; which of the many positions is an A position and not an A’ position in which the subject scrambled to? The contrast in (26) sheds some light on this question. We can see in (26a) that an accusative object can move tothepositiondirectlyfollowingthepronoun. (26b)illustratesthattheaccusativemarked objectcannotmovetothepositionprecedingthepronoun. (26) a. dass that es it.acc der the Maria 1 Maria.dat gestern yesterday der the Fritz Fritz.nom t 1 geschenkt give hat has b. *dass that der the Maria Maria.dat es it.acc gestern yesterday der the Fritz Fritz.nom geschenkt give hat has The data in (26) is in line with an analysis that Spec TP actually precedes the pronoun position and that the position following the pronoun is an A’ position which is also available to non-subject DPs. This explains why the position preceding the pronoun is available only to nominative DPs - which can A move to the spec TP position but not to internal arguments - which cannot A move to the spec TP position in the presents of an externalargument. This means that the Wackernagel position in German is likely lower but close to T. I will propose that weak pronouns move to the Wackernagel position in order to get an anaphoric interpretation which they can only get in the left sentential periphery while strong pronouns do not have to move there because they can get the anaphoric interpretationwithintheDP. 2.3.3 Pronoun(re-)orderingandscrambling Deficient/weakpronounsinStandardGermancomeinanS<DO<IOorder. Incontrastto r-expressions, weak pronouns cannot be scrambled with respect to each other. The order ofweakpronounsisfixed(cf.Haider2010). Thisisdemonstratedbytheexamplesin(27). Notethatallthreeexamplesin(27)areungrammaticalbecauseweakpronounshavetobe realizedintheS<DO<IOorder. (27) a. *,weil because ich I.nom ihm him.dat dich you.acc gestern yesterday gezeigt shown habe. have. CHAPTER2. WEAKANDSTRONGPRONOUNS 25 b. *,weil because ihm him.dat dich you.acc ich I.nom gestern yesterday gezeigt shown habe. have. c. *,weil because dich you.acc ihm he.dat ich I.nom gestern yesterday gezeigt shown habe. have. Only weak pronouns have this order restriction. Strong pronouns like dich ‘you.acc’ in (28a)orich ‘I.nom’in(28b)donotfollowthesameorderrestrictions. (28) a. ,weil because ich I.nom ihm him.dat gestern yesterday nur only dich you.acc gezeigt shown habe. have ‘becauseIshowedonlyyoutohimyesterday.’ b. ,weil because dich you.acc ihm him.dat gestern yesterday nur only ich I.nom gezeigt showed habe. have ‘becauseonlyIshowed youtohimyesterday.’ I will discuss next the word base order of arguments in German and demonstrate lexical DPs. The base word order of lexical DPs tends to bedat>acc with only very view verbs whichhaveargumentsinaacc>datorder. HoweverasHaider(2010)pointsout,theorder ofweakpronounsisnotaffectedbytheargumentstructureoflexicalDPs. WordorderofLexicalDPs Swiss German and Standard German allow for scrambling. Determining the base word orderoftwoelementsinrelationtoeachother,canbetricky(c.f.Sternefeld&Featherston (2003)). ForstandardGerman,Lechner&Yatsushiro(2001)andSternefeld&Featherston (2003) have proposed diagnostics which demonstrate that the IO is generally base gener- ated higher than the DO. The diagnostics range from binding to weak and strong cross overcases. However,tome,theclearestandstrongestargumentsarescope/reconstruction caseswhichIwillgooverintofollowing. IwilloutlinetheargumentputforwardinLech- ner & Yatsushiro (2001) for Standard German for high datives, i.e. datives that are base generatehigherthantheaccusativeargument. Iwillthanturntosomecasesoflowdative, i.e. dative arguments that are base generated lower than the accusative argument before illustrating that the word order of Lexical DPs has no influence on the word order of pronominal arguments. This is an important fact that any analysis of the Person Case Constraintwillhavetotakeintoaccount. Germanisalanguagewithaquiteflexiblewordorderbutquiterigidscope. Thismeans thatscopealwayscorrelateswithovertsyntaxinneutralintonation(c.f.Krifka1998). Both inneutralandverumfocusintonation,asentencelike(29a)hasonlyoneinterpretation,one in which the subject takes scope over the object. However, (29b) is ambiguous between a surface scope interpretation where the object quantifier takes scope over the subject and CHAPTER2. WEAKANDSTRONGPRONOUNS 26 an inverse scope interpretation where the subject quantifier takes scope over the object quantifier. The second reading is the result of a reconstruction (c.f. von Fintel & Heim 2011). I will refer to the availability of an inverse scope reading as a reconstruction effect. Note that the subject in (29a) moves from a position that c-commands the object DP to anotherpositionthatc-commandstheobjectDP.Reconstructingthesubjectwilltherefore not make a difference in the scope interpretation of the sentence. However, in (29b) the objectmovesoverthesubject. AtPFtheobjectisthereforeinapositionwhichc-commands thesubjectwhileit’slowercopyisbeingc-commandedbythesubject. (29) a. [Genau exactly ein one. nom Freund] nom friend hat has t nom [jeden each. acc Roman] acc novel gelesen. read ‘Exactlyonefriendhas readeachnovel. 9 1 >8= 8 >9 1 b. [Genau exactly einen one. acc Roman] acc novel hat has [jederFreund] nom each. nom t acc friend gelesen. read ‘Eachfriendhasreadexactlyonenovel. 9 1 >8=8>9 1 The object in (29b) moves from a position which is c-commanded by the subject to the spec CP position where it c-commands the subject DP. In the higher position, the object thereforetakesscopeoverthesubject,whilereconstructionwillgenerateaninterpretation wherethesubjectquantifiertakesscopeovertheobjectquantifier. (29) demonstrates that an ambiguity arises if one quantifier is moved over another quantifierasin(29b). However,inaconstructionwheretheorderbetweenthequantifiers ispreservedthroughthederivation,asin(29a),noscopalambiguityarises. This is important because it predicts that different argument orderings between the dative and the accusative argument trigger differences in the availability of scopal am- biguities. When the dative argument precedes the accusative argument, only one in- terpretation is possible, namely one in which the dative argument takes scope over the accusative argument. This is demonstrated in (30a). If the accusative argument linearly precedes the dative argument, the sentence allows for two readings, one in which the accusative argument takes scope over the dative argument, but also a reading where the dativeargumenttakesscopeovertheaccusativeargument. Thisisshownin(30b). (30) a. weil since sie she [genaueinem exactlyone.dat Freund] dat friend [jeden each.acc Gast] acc guest vorgestellt introduced hat. has ‘sincesheintroducedeveryguesttoexactlyonefriend.’ 9 1 >8= 8>9 1 b. weil since sie she [genaueinen exactlyone.acc Gast] acc friend [jedem each.dat Freund] dat guest vorgestellt introduced hat. has. ‘sincesheintroducedexactlyonefriendtoeveryguest.’ 9 1 >8=8>9 1 (Lechner&Yatsushiro2001:9) CHAPTER2. WEAKANDSTRONGPRONOUNS 27 In(31),wecanseethesameistrueformainclauses. Weonlygetscopalambiguitieswhen theDOc-commandstheIOatPFbutnotwhentheIOc-commandstheDO. (31) a. [Genaueinem exactlyone.dat Freund] dat Friend hat has sie she [jeden each.acc Gast] acc guest vorgestellt. introduced ‘sheintroducedexactly onefriendtoeveryguest.’ 9 1 >8= 8>9 1 ’ b. [Genaueinen exactlyone.acc Gast] acc guest hat has sie she [jedem each.dat Freund] dat friend vorgestellt. introduced ‘sheintroducedexactlyonefriendtoeveryguest.’ 9 1 >8=8>9 1 ’ (Lechner&Yatsushiro2001:9) TheseresultsareconsistentwithananalysisthatassumesthattheIOfortheverbvorstellen ‘introduce’ is base generated in a position c-commanding the DO. A DO c-commanding a IO at PF has then moved over the IO somewhere in the derivation and can either be interpretedwithsurfacescope,i.e.widescope,orcanbereconstructedallowingtheIOto takewide scopeovertheDO. It has been argued that the so-called kriegen passive constructions can function as diagnostics for high and low dative constructions in that they are only grammatical with highdativesbutnotlowdatives(cf.McIntyre(2006)). We have seen above that the verb vorstellen ‘introduce’ takes two internal arguments whereas the IO is base generated in a higher position than the DO. A sentence like (32a) canbepassivizedintwoways. Theregularwerden‘become’passivewillcreateastructure where the DO is promoted and will get nominative case assigned (32b). In the case of kriegen ‘got’passivestheDOispromotedandwillgetnominativecaseassigned(32c). (32) a. Die The Maria Maria.nom stellt introduce mir me.dat den the Max Max.acc vor. to ‘MariaintroducesMaxtome’ b. Mir me.dat wird become der the Max Max.nom vorgestellt. introduced DO withnom ‘Maxisintroducedtome’ c. Ich I.nom kriege got den the Max Max.acc vorgestellt. introduced IOwithnom ‘IgotMaxintroducedtome.’ Neither passive construction is contingent on ditransitive verbs. Helfen takes an external argumentandadativemarkedinternalargument,asisillustratedin(33). Butonlyinthe case of the kriegen passive in (33c) is the internal argument Dative marked. In the case of the becomepassivenoargumentismarkednominative(33b). CHAPTER2. WEAKANDSTRONGPRONOUNS 28 (33) a. Ich I.nom helfe help dir. you.dat ‘Ihelpyou.’ b. Dir you.dat wird become geholfen. helped nonom ‘Youarebeinghelped.’ c. Du you.nom kriegst get hilfe. helped IOwithnom ‘Yougethelp.’ Most dative arguments are base generated higher than the DO in German, but some verbs are said to subcategorize for low datives that are c-commanded by the accusative argument in their base position. Examples for such words are aussetzen ‘to expose’, unterziehen ‘to subject’, zuführen ‘to bring’, entgegenhalten ‘to hold towards/against’ (cf. Georgala 2011:98). Sentences with low dative behave differently with respect to the tests above. Frey & Pittner (1998) propose a topic/focus test and a C-command test, which leads them to conclude that vorziehen ‘prefer’ subcategorizes for a low dative, i.e. an IO whichisgeneratedlowerthantheaccusativeargument. Wecanseein(34)thatvorziehen‘prefer’doesdoesallowforpassivizationwithwerden ‘become’(34b)butnotwithkriegen‘get’(34c) (34) a. Die The Maria Maria.nom zieht prefers den the Peter Peter.acc mir me.dat vor. pre. ‘MariaprefersPeteroverme’ b. Der The Peter Peter.nom wird becomes mir me.dat vorgezogen. preferred ‘Peterispreferredoverme’ c. *Ich I.nom krieg got den the Peter Peter.acc vorgezogen. preferred If we look at the scope, (35a) as well as (35b) allow for the inverse reading. This is in con- trast to the cases where dative arguments are base generated higher than the accusative argument, such as in the verb vorstellen ‘introduce’. There, only the acc>dat order is allowed for an inverse reading. Georgala (2011) argues that in German dative arguments that are base generated low have to move to Spec ApplP to be case licensed. This move- mentmovesthedativeargumentovertheaccusativeargument,whichcouldexplainwhy sentences with low dative arguments are always ambiguous, while sentences with high dative arguments are only ambiguous when the accusative argument moved over the dativeargumentinthesurfacestructure. (35) StandardGerman CHAPTER2. WEAKANDSTRONGPRONOUNS 29 a. weil because ich I jedem each.dat Gast guest genau exactly einen one.acc Freund friend vorziehe. prefer 8>9 1 =9 1 >8 ‘becauseIpreferexactlyonefriendovereveryguest.’ b. weil because ich I jeden each.acc Freund friend genau exactly einem one.dat Gast guest vorziehe. prefer 8>9 1 =9 1 >8 ‘becauseIprefereachFriendoverexactlyoneguest.’ Wesawthatverbsthattakelowdativesandverbsthattakehighdativesbehavedifferently withrespecttokriegen‘get’passivesandscope. While, depending on the verb, datives might be generated lower or higher than the accusative argument, the weak pronoun order remains the nom > acc > dat for both. This is illustrated in (36) and (37) below. vorstellen ‘introduce’ is a high dative verb as cancan be seen by the grammaticality of the kriegen passive (36a) while vorziehen is a low dative verb, which can be seen by the ungrammaticality of the kriegen passive formation in (37a). Importantly, the word order for weak pronouns remains the same for the verb takinghighdatives(36b)orlowdatives(37b). (36) a. Noch still gestern yesterday hat has er he.nom dich you.acc vorgestellt introduce gekriegt. got ‘Hegotintroducedtoyoualreadyyesterday.’ b. Noch still gestern yesterday habe have ich I.nom (dich) you.acc ihm him.dat (*dich) you.acc vorgestellt. introduced ‘Iintroducedyoutohimalreadyyesterday.’ (37) a. *Noch still gestern yesterday hat has er he.nom dich you.acc vorgezogen referred gekriegt. got ‘Untilyesterday,hewaspreferredoveryou.’ b. Noch still gestern yestday habe have ich I.nom (dich) you.acc ihm him.dat (*dich) you.acc vorgezogen. referred ‘Ireferredyouoverhimuntilyesterday.’ WecanconcludefromtheexamplesanddiscussionabovethatweakpronounsinStandard German come in a fixed order which is independent from regular argument order. We see that we have reordering of pronominal arguments with respect to the base word order of lexical DP arguments with high dative verbs such as vorstellen ‘introduce’. This reordering of pronominal arguments is however not a feature of weak pronouns per se. Weak pronouns with low dative verbs such as vorziehen ‘prefer’ are not reordered with respect to the base word order of lexical DP arguments. In other words; the base word orderofargumentsdoesnothaveaninfluenceonthewordorderofweakpronouns. Weak pronouns have a fixed acc>dat order independently of whether the dative argument is basegeneratedinaloworahighposition. CHAPTER2. WEAKANDSTRONGPRONOUNS 30 Thefactthattheorderofweakpronounsisfixedtellsusfurthermoreaboutscrambling of weak pronouns. Lexical DP arguments and strong pronouns can occur in theacc>dat and dat>acc order. The assumption is that this flexibility with respect to word order is the result of scrambling in the German middle field. The fact that weak pronouns have a constantorderwithrespecttoeachotherisastrongindicationthatweakpronounscannot bescrambled. I will now turn to Cardinaletti & Starke’s (1994) diagnostics for weak and strong pronouns. 2.3.4 Diagnosingstrongpronouns Coordination Cardinaletti & Starke (1994, 1996, 1999b) begin their discussion about different kinds of pronouns with an observe that Italian makes a distinction between classes of pronouns. This is illustrated in (12) repeated as (38) below. The pronoun esse in (38a) can have humanornon-humanreferencebutcannotbecoordinated. Thepronounloroin(38b)can becoordinatedbutcanonlyhavehumanreference. (38) Italian a. Esse 3.pl.fem.nom (*e and quelle those accanto) besides sono are troppo too alte. tall/high +human/ -human b. Loro 3.pl.fem.nom (e and quelle those accanto) besides sono are troppo too alte. tall/high +human/ *-human (Cardinaletti&Starke1994:41) Cardinaletti & Starke (1994, 1996, 1999b) propose that the connection between obligatory humanreferenceandcoordinabilityisauniversalacrosslanguages. Theyarguethatsome languagessuchasGermandonotmakeaphonologicaldistinctionbetweenthetwokinds ofpronounsbutstillhavethesamedistinction. TherestrictioninGermanisdemonstrated in(39)whichis(4)inCardinaletti&Starke(1994). Whilesie’they’canrefertoahumanor anon-humanentityin(39a),itcan onlyrefertoanon-humanentityin(39b). (39) a. Sie they sind are groß. tall/bigg +human/ -human b. Sie they und and die those daneben besides sind are groß. tall/bigg +human/*-human (Cardinaletti&Starke1994:43) CHAPTER2. WEAKANDSTRONGPRONOUNS 31 They conclude from this that only strong pronouns can be coordinated and that strong pronouns have to have a [+human] reference. Diesing (1999) points out that demonstra- tives in English can be coordinated without having human reference and Bayer (1999) raises doubts that the restriction on human reference is true for d-pronouns in German. I want to follow Cardinaletti & Starke (1999a) and assume that demonstrative pronouns might behave different than personal pronouns and d-pronouns, but the problem raised in Bayer (1999) persists and is even bigger than hinted at by him. We can quite freely createcoordinatedconstructionsoftheformpronounandXPwhicharegrammaticalorat leastdramaticallybetterthan(39b)wouldsuggest. Theseexamplesarein(40)–(42) (40)demonstratesthatapronouncanrefertoaninanimateobjectlikeanIkeachair. (40) Dieser This Ikea Ikea Stuhl 1 chair is is voll full scheiße. crappy Er 1 it.masc.masc und and alle all anderen other Ikea Ikea Stühle chairs gehören belong öffentlich publicly verbrannt. burned ‘ThisIkeachairiscrappy. ItandallotherIkeachairsoughttobepubliclyburned.’ (41) illustrates that a pronoun in a coordinated structure can also have an animate but non-humanreferentsuchasadog. (41) Ich I hasse hate deinen your Hund 1 . dog Er 1 it.masc.masc und and alle all anderen other Schäferhunde shepherddogs sollten ought eigentlich actually aus out Wien Vienna verbannt exiled werden. become Ihateyourdog 1 . It 1 andallothershepherddogsoughttobebandfromVienna. Er ‘it.masc’ is clearly a personal pronoun. The referent does not have to be present for thesesentencestobefelicitous. The coordinated pronoun can also be in object position, as (42) demonstrates. Note that there is some flexibility here. A speaker might choose a personal pronoun (42a), a d-pronoun(42b)oralexicalnoun(42c)torefertoBeethoven’s4thsymphony. (42) Context: DidtheViennaPhilharmonicOrchestraplayBeethoven’s4thsymphony yesterday? a. Ja yes sie they.nom haben have sie it.fem.acc und and Mahlers Mahler’s Unvollendete unfinished aufgeführt. preformed b. Ja yes sie they.nom haben have die it.fem.artpro.acc und and Mahlers Mahler’s Unvollendete unfinished aufgeführt. preformed ‘Yes,theyperformeditandMahler’s c. Ja yes sie they.nom haben have Beethovens Beethoven’s 4. 4th und and Mahlers Mahler’s Unvollendete unfinished aufgeführt. preformed CHAPTER2. WEAKANDSTRONGPRONOUNS 32 I think that Cardinaletti & Starke; Cardinaletti & Starke; Cardinaletti’s (1994; 1996; 1999) distinction between strong and weak/deficient pronouns is fundamentally correct and that only strong pronouns can be coordinated. However I have demonstrated above that humanreferenceisnotrequiredforcoordinationandforstrongpronouns. Iwillargueand demonstratethroughoutthissectionthatd-pronounsarestrongpronounsirrespectiveof their referent. I will furthermore demonstrate that es ‘it.neut’ is always a weak pronoun, again irrespective or its referent. The human/ non-human distinction in (39) is therefore notduetoafundamentalconnectionbetweenstrongpronounsandhumanreferencebut due to a tendency to interpret strong pronouns as having human referent. I will get back tothispointintheanalysisofstrongpronounsandwillproposeacauseforthistendency. Cardinaletti&Starke(1994)pointoutthates‘it’isasomewhatspecialcaseinGerman. They argue that es only exist as a deficient pronoun and cannot be coordinated. This is relevant because the 3 rd person neuter pronoun can refer to girls, children and babies in Germanandcanthereforehaveahumanreference. When talking about the neighbors crying baby it would be appropriate to say some- thinglike(43). (43) Es it schreit screams nochimmer. still ‘He/sheistsillscreaming.’ WhenIwanttoexpressthatIdroppedoffthesamecryingbabyandastrolleratdaycare I will have to use a masculine or feminine pronoun like in (44a), es ‘it’ is ungrammatical asisdemonstratedin(44b). (44) a. Ich I habe have sie/ihn her/him.acc und and den the.acc Kinderwagen stroller in in der the Kita day-care abgegeben. brought ‘Ihavelefther/himandthestrollerintheday-care.’ center’ b. *Ich I habe have es her/him.acc und and den the.acc Kinderwagen stroller in in der the Kita day-care abgegeben. brought ‘Ihavelefther/himandthestrollerintheday-care’ Usingphonologicalstressonesasin(45a)doesnotmakethesentencegrammatical. Using thed-pronoundasinthissituationisinfelicitousas(45b)illustrates. Incontrasttoeswhich canhavehumanandnon-humanreference,dascannothavehumanreference. (45) a. *Ich I habe have ES it.acc und and den the.acc Kinderwagen stroller in in der the Kita day-care abgegeben. brought b. #Ich I habe have das it.acc und and den the.acc Kinderwagen stroller in in der the Kita day-care abgegeben. brought CHAPTER2. WEAKANDSTRONGPRONOUNS 33 es cannot be coordinated, even if it has a human reference. es with non-human reference can also not be coordinated as is illustrated in (47a). However, one can use either the d-pronouns dasoranountorefertothereferentindiscussion(47b)(47c). Context: I am having a party but don’t have enough cutlery. I have invited my friend Susiandherboyfriendandaskhertobringsomecutleryalong. (46) Susi: OK OK ich I werde will Besteck cutlery.(neut) mitbringen, bring, soll should ich I sonst else noch still was something mitnehmen? bring? Susi: OKIwillbringcutlery. ShouldIbringsomethingelse? (47) a. *Bring bring.2.sg.imp nur only ES it.3sg.neut.acc und and deinen your Freund friend.acc mit. with ‘Justbringthecutleryandyourboyfriendalong.’ b. Bring bring.2.sg.imp nur only das it.3sg.neut.dec und and deinen your Freund friend.acc mit. with c. Bring bring.2.sg.imp nur only das the Besteck cutlery und and deinen your Freund friend.acc mit. with ‘Justbringthecutleryandyourboyfriendalong.’ I illustrated above that the contrast between pronouns that have human reference and pronouns that do not have human reference is not clear as proposed by Cardinaletti & Starke (1994, 1996, 1999b). The 3.sg,neut personal pronoun es can never be coordinated. WewillseethroughthissectionthatCardinaletti&Starke(1996)arecorrectinsayingthat es is always a deficient pronoun. Other personal pronouns can be coordinated. I agree with Cardinaletti & Starke (1996) that these pronouns have a strong and a deficient form which are homophones but have different syntactic behavior. I disagree with them in theirassertionthatstrongpronounscannothaveanon-humaninterpretation. d-pronouns are always strong pronouns. It is tempting to think that d-pronouns are strong pronoun counterparts to personal pronouns, but we have seen in the examples abovethatdasisnotjustthestrongversionofthedeficientpronounes. escanhavehuman or non-human reference but das can only have non-human reference. It is also important to say that stress is not the morpheme that causes or marks strong pronouns. Strong pronouns do not need to be phonologically stressed and phonological stress does not forceastrongpronounparsing. Topicalization Cardinaletti & Starke (1994, 1996) argue that topicalized pronouns require human refer- ence. Topicalized pronouns behave similar to the coordinated pronouns. We can see in CHAPTER2. WEAKANDSTRONGPRONOUNS 34 (48) that embedded object can move to the position preceding the verb in second posi- tion, a position commonly referred to as vorfeld. The difference between (48a) and (48b) is that the pronoun ihm ‘him’ has a human reference in the former example but has to haveanon-humanreferenceinthelater. Thetwoexamplesalsohaveastarkdifferencein judgments. Whiletheearlierisgrammaticalthelatterisnot7. (48) a. Ihn him.acc hat has Maria Maria gesagt, said dass that Hans Hans schon already eingeladen invited hat. has ‘MariasaidthatHansalreadyinvitedhim’ b. *Ihn him.acc hat has Maria Maria gesagt, said dass that Hans Hans schon already gekauft bought hat. has ‘MariasaidthatHansalreadyboughtit’ (Cardinaletti&Starke1996:29) (49) demonstrate the same point for extractions out of TP complements. We see the same facts surfacing below where the embedder subject is topicalized. Er ‘he’ in (49a) has a human reference, er in (49b) and es ‘it’ in (49c) cannot have a human reference and is ill formed. (49) a. Er He.masc.nom glaube believes.1sg ich I.nom ist is.3.sg sehr very nett. nice. ‘Ibelieveheisverynice.’ b. #Er It.masc.nom glaube believe.1sg ich I.nom ist is sehr very teuer. expensive. ‘Ibelieveitisveryexpensive’ c. *Es It.masc.nom glaube believe.1sg ich I.nom ist is sehr very teuer. expensive. ‘Ibelieveitisveryexpensive’ (Cardinaletti&Starke1996:29) Up to this point topicalization mirrors what we have seen with respect to coordination above. The weak pronoun es is ungrammatical with topicalization and the topicalization withpersonalpronounsthatdonothavehumanreferenceisinfelicitous. Thed-pronoun der in (50a) can have the same referent as er in (49b) and das in (50b) can have the same referentasesin(49c). Importantly,topicalizationwithd-pronounsasin(50)isgrammatical evenifthepronounsdohaveanon-humanreference. (50) a. Der it.3sg.masc.der glaube believe.1sg ich is.3sg ist very sehr expensive. teuer. ‘Ibelieveitisveryexpensive.’ 7NotallGermanspeakersallowforlongdistantextraction,tothembothsentenceswillbeungrammatical. CHAPTER2. WEAKANDSTRONGPRONOUNS 35 b. Das it.3sg.neut.der glaube believe.1sg ich is.3sg ist very sehr expensive. teuer. ‘Ibelieveitisveryexpensive.’ We can see from this examples two things: strong pronouns behave differently from deficient pronouns and d-pronouns tend to pattern with other strong pronouns and humanreferenceisnotacrucialdistinguishingfactorforstrongpronouns. WeakpronounsintheVorfeld Wehaveseeninthesectionabovethatdeficientpronouns cannot be topicalized. It follows therefore that es ‘it’ can never be topicalized. This does notmeanthatdeficientsubjectorobjectpronounscannotoccupytheVorfeld,sincenotall elements occupying this position are topics. If no element moves to the sentence initial position for other purposes, the highest element will move to to spec CP to fulfill verb secondrequirement. Thedistributionofnon-topicalizedelementsintheVorfeldgivesus a good indication about the structure since it indicates that the element in the vorfeld is the highest element after all features on C were checked. Weak pronouns in the vorfeld positionarethereforeadiagnosticforwhatelementsprecedecliticsinthederivation. Subject pronouns can occur in the vorfeld position. This is not only true for strong pronouns but also for defective pronouns as is illustrated in (51) to (53). The subject pronoun sie ‘she’ in (51) can be in the vorfeld whether it has a human referent (51a) or a non-humanreferent(51b). (51) a. Sie She.nom hat has den the Schraubenzieher screwdriver dann then doch still noch still gefunden. found (sie (she = = Maria) Maria) ‘Shefoundthestrewdriverafterall.’ b. Sie She.nom quietscht squeeks jedes every Mal, time, wenn when es it kalt cold wird. become (sie (she = = die the Türe) door) ‘Itsqueekseverytimeitgetscold.’ (52)demonstratesthatthedefectivesubjectpronounes‘it’canberealizedinthevorfeld position,whetheres hashumanreferenceasin(52b)orinanimatereferenceasin(52a). (52) a. Es It.neut.def liegt lies auf on dem the Tisch table (es (it = = das the Brot) bread) ‘Itisstilllyingonthetable’ b. Es It.neut.def ist is noch still im in.the Kindergarten kindergarten (es (it = = das the Kind) child) ‘He/sheisstillinkindergarten.’ (53)demonstratesthatinthepresenceofmultipledefectivepronouns,thefirstpronoun will always be interpreted as the subject. We see that Stress will not have an affect on CHAPTER2. WEAKANDSTRONGPRONOUNS 36 the interpretation as is illustrated in (53b). The object can only scramble over the subject pronounifitisastrongpronoun(53c)afullDPasin(53d) (53) Context: One does not have to pretend that the unicorn is not just a horse (Pferd.neut) with a strapped on horn. The child already knows that it is just a horse. a. Es it.neut.nom/*acc hat has es it.neut.acc/*nom schon already einmal once ohne without Horn horn gesehen. seen. b. ES it.neut.nom/*acc hat has es it.neut.acc/*nom schon already einmal once ohne without Horn horn gesehen. seen. c. Das it.neut.artpro.#nom/acc hat has es it.neut.nom/#acc schon already einmal once ohne without Horn horn gesehen. seen d. Das The Pferd horse hat has es it.neut.nom/#acc schon already einmal once ohne without Horn horn gesehen. seen ‘S/hehasseenthehorsewithoutthehornalready. Iwanttoarguethatdefectivepronounsdonotmovetothevorfeldtochecktopicorother interpretive features in C but only to check an EPP feature located on the C head. Frey (2006)callsthisoperationFormalMovement. Thisassumptionmakesveryclearpredictions. We have seen at the beginning of this chapter, that defective subject pronouns have to precede defective object pronouns. This means that defective object pronouns cannot move to the vorfeld if a defective subject pronoun is present since this would violate minimality. This explains why the pronoun in the vorfeld in (53b) can only be interpreted as the subject pronoun. I would also predict that defective object pronouns can move to the vorfeld position if the subject remains low, e.g. in spec vP or a low focus position (cf. Wurmbrand 2006). If the subject remains low, the subject does not function as an intervener and formal movement can target weak object pronouns in the Wackernagel position. This prediction is born out. Contrary to a claim in Cardinaletti & Starke (1996) Meinunger (2007) shows that the deficient object pronoun es and not only the deficient subject pronouns can surface in the vorfeld. The object pronoun es with human or non- humanreferenceinthevorfeldiswellformedunderparticularcircumstances. Meinunger (2007) points out that the grammaticality of defective object pronouns in the vorfeld depend on the subject. Only subjects that can remain low are possible. Wurmbrand(2006)showsthatindefinitesubjectsdonothavetomovetospecTP,notonly at PF but also at LF. The idea is that (54a) is grammatical because jemand ‘somebody’ can CHAPTER2. WEAKANDSTRONGPRONOUNS 37 remainlowanddoesnothavetomovetospecTP.Theweakaccusativepronouncanthan move from the Wackernagel position to spec CP. (54b) demonstrates that the indefinite subjectdoesnothavetoremainlow,andcanmovetothevorfeldinwhichcaseesremains intheWackernagelposition. (54) Context: Thethiefhidthepieceofjeweleryunderaloosefloorboard. Theythought nobodywouldfinditthere. a. Es i it.neut.def hat has e i then dann still doch somebody.nom jemand discovered enteckt. b. Jemand somebody.nom hat has es it.neut.def dann then doch still enteckt. discovered ‘Somebodydiscovereditafterall.’ (55)illustratesthatthesubjectisrelevant. Definitesubjectspreventthedefectivepronoun frombeingrealizedinthevorfeld. (55) Context: ThethiefhidthepieceofjeweleryunderaloosefloorboardsothatPeter willnotfindit. Theythoughtnobodywouldfinditthere. a. *Es it.neut.der hat has der the Peter Peternom dann than doch still enteckt. found b. Der the Peter Peter.nom hat has es it.neut.der dann than doch still enteckt. found ‘Peterfounditafterall.’ This tells us something important about defective pronouns. Word order in Standard Germanisquitefreeandscramblingiscommonplacebutdefectivepronounsdonottake part in it. Their relative order is stable, they move to the Wackernagel position and can only appear higher than the Wackernagel position to check EPP features if they are the minimally closest XP which can fulfill the V2 requirement. This means that if we see re- orderingofweakpronouns,thisreorderingisnotinformationstructuredrivenscrambling. Becauseifthiswouldbepossiblewewouldseeweakobjectpronounsappearingsimilarly unrestrictedasotherXPs. Modificationandthestructureofpronouns Strong pronouns can be modified by what Cardinaletti & Starke (1994) call c-modifiers, adverbs that modify the whole DP. Siewierska (2004:10) points out that pronouns can be modified by numbers. I will illustrate that modification is only possible with strong pronouns. CHAPTER2. WEAKANDSTRONGPRONOUNS 38 c-modification This is demonstrated by (56a) where nur ‘only’ is able to modify the strong pronoun sie ‘her’. Note that Mary does not need to be explicitly mention for (56a) to be felicitous. It is sufficient that it is part of the common ground that John and Mary usually go to Mass together. The d-pronoun in (56b) is infelicitous in a context where Maryisnotmentionedinthecontext. (56) Context: JohnandMarygotogethertoMassonadailybasis. DidyouseeJohn? a. Nein, No, ich I.nom habe have nur yesterday sie (only) gesehen. her.acc seen. ‘NoIhaveseen(only)heryesterday.’ b. #Nein, No, ich I.nom habe have nur yesterday die (only) gesehen. her.acc.der seen. ‘NoIhaveseen(only)heryesterday.’ Ihaveintroducedin(15)Cardinaletti&Starke’s(1996)ideathatthemostdeficientpossible formhastobechosen. (57a)isinfelicitousinthiscontextbecauseaweakpronounwould be felicitous here and the most deficient pronoun has to be used. A strong pronoun can refer to John as in (57b), if it refers to a member of a focus set. Note that (57b) and (57c) do not only say that the speaker saw John, it also makes clear that the speaker did not see Mary. I want to point to the contrast between (56b) and (57c). The contrast seems to indicate that d-pronouns require a clear discourse antecedent which might be more local thanotherstrongpronouns. (57) Context: JohnandMarygotogethertoMassonadailybasis. DidyouseeJohn? a. #Ja, Yes, ich I.nom habe have nur only ihn him.acc gesehen. seen ‘Yes,Ihaveonlyseenhim.’ b. Ja, Yes, aber but ich I.nom habe have nur only ihn him.acc gesehen. seen. ‘YesbutIhaveonlyseenhim.’ c. Ja, Yes, aber but ich I.nom habe have nur only den him.artpro.acc gesehen. seen. ‘YesbutIhaveonlyseenhim.’ The examples above demonstrate that strong pronouns can be modified by c-modifiers. Weakpronouns,ontheotherhand,cannotbemodifybyc-modifiers. Thisisdemonstrated in(58). Neither(58a)nor(58b)canhaveanon-humanreferent. (58) Adverbmodification a. *Nur Only es it.neu.nom war was teuer. expensive CHAPTER2. WEAKANDSTRONGPRONOUNS 39 b. #Nur Only er it.masc.nom war was teuer. expensive Thisisagaindifferentford-pronouns. Thecounterpartto(58a)withd-pronounsin(59a) isgrammaticalandthecounterpartto(58b)in(59b)isslightlymarkedbutbetter. (59) Adverbmodification a. Nur Only das it.neu.nom war is teuer. expensive b. Nur Only der it.masc.nom war is teuer. expensive Itseemsthatc-modificationofsomed-pronounsrequireaparticularcontext. Introducing aclearantecedentsuchasin(60)doesnothelpweakpronounsbutd-pronouns. (60) Context: Iboughtquiteanumberofthings. a. *Der the Tisch 1 table.nom war was teur; expensive; eigentlich infact war was nur only er 1 it.masc teuer. expensive b. Der the Tisch 1 table.nom war was teur; expensive; eigentlich infact war was nur only der 1 it.masc teuer. expensive Interestingly, we cannot replicate an example such as (56) with inanimate reference as in (61). It is not surprsing by now that the defective pronoun in (61a) is ill-formed. We see again that the d-pronoun in (61b) can also not refer to the baldachin like the d-pronoun in(56b)isunabletoreferbacktoMary. Onlyusinganr-expressionasin(61c)isfelicitous inthiscontext. (61) Every year, for the Corpus Christi celebration the priest carries the monstrance (Monstrantz.fem)throughthevillageandfourmencarryabaldachin(Baldachin.masc)above thepriest. Didyouseemonstranceattheprocession? a. #Nein No, ich I habe have nur only ihn him.acc gesehen. seen b. #Nein No, ich I habe have nur only den him.der.acc gesehen. seen c. Nein No, ich I habe have nur only den the Baldachin Baldachin.acc gesehen. seen ‘No,Ihaveonlyseenthebaldachin.’ C-modification is only possible with strong pronouns. I want to generalize the pattern above to say that weak pronoun do not allow for focus alternatives. C-modification like nur ‘only’ requires a set of focus alternatives since it crucially picks one element from a CHAPTER2. WEAKANDSTRONGPRONOUNS 40 set. C-modificationofweakpronounsisinfelicitousbecausec-modificationofasingleton setissuperfluousandviolatespragmaticmaxims. Thisisclearlyinsufficienttoexplainthefulldiversitythatwesee. Cardinaletti&Starke (1994) argue that the differences are in fact between human vs non-human reference. I havealreadygivenanumberofexamplesthatdemonstratethatthisisnotnotthedefining feature. Wehaveseensomeexampleswherestrongpronounsareassociatedwithhuman reference. IdontthinkthathumanreferenceisparticularmarkedinGermanthedifference between (56) and (61) point towerds a different in gender marking between human and non-humanpronouns. Whilethegenderofthestrongpronounin(56)issufficienttopick outthecontextuallantecedent,genderseemstobenotsufficentin(61). The difference between Mary and John in (56) and the monstance and the baldachine in(61)isthatgenderintheformerisinterpretable,whileitisnotinterpretableinthelater. I am drawing here on a distinction from De Vincenzi & di Domenico (1999) exemplified forItalianintable2.4from(Alexiadouetal. 2007:242)below. Table 2.4: Explaining the difference between the different kinds of pronouns based on DeVincenzi&diDomenico(1999) Noun Interpretability Variabilty ragazza ’girl’ + + donna ’woman’ + - sedia ‘chair’ - - Theintuitionhereisthatgenderrepresentssomethingdifferentforahumanthanfora chair. ItisnotaccidentalthatwomanandsisterhavethesamegenderinGerman,however it is accidental that die Sonne ‘the sun’ and die Datei ‘the file’ are feminin too. This is clear justbylookingatFrenchwerelafammeandlasœrarefeminintoobutlesoleilandlefichier are not. So why is there a difference between human and non-human reference and not just between animate and inanimate? I would argue that this is also a effect of gender. The gender of animals is in fact often underspecified in German. Animals have a default gender which does not have to match the actual sex of the animal. I can use the default genderevenifIknowtherealsexoftheanimal. TheSentencesin(62)arefineevenifmy dogisfemaleormycateismalejustbecausethedefaultgenderfordogsismasculineand forcatsis female. (62) a. Ich I mag like unseren our.masc Hund dog viel much lieber rather wie than andere other Hunde. dogs Ilikeourdogmuchmorethanotherdogs. CHAPTER2. WEAKANDSTRONGPRONOUNS 41 b. Ich I habe have einen a.masc Hund dog und and eine a.fem Katze. cat ‘Ihaveadogandacat.’ It is clear that we do not want to say that gender with animals is not always interpretable but it ought to be in cases where the term is clearly reflective of the sex of the animal. das Pferd ‘the horse’ is marked neuter and does not tell us anything about the sex of the animalbutthetermsHengst‘stallion’,Stute‘mare’oraWallach‘gelding’areunambiguous masculine,feminineormasculine. We cannot use a pronoun to refer to a contextually salient horse if we introduce just a horsewithoutmakethegenderclear(63). Thepronounin(63d)cannotrefertothehours but could only be interpreted as referring to Mary.The only continuation that is possible insuchacontextis(63e)wherewe usealexicalDPtorefertothehorse. (63) Context: Marygoesforlongridesonherhorse(neut)everymorning. Everybody isworriedtoday. Whathappened? a. *Heute Today früh morning ist is nur only es he.nom nach to Hause home gekommen. came Onlyhecamehomethismorning. b. #Heute Today früh morning ist is nur only das she.nom nach to Hause home gekommen. came das=horse Onlyhecamehomethismorning. c. #Heute Today früh morning ist is nur only er she.nom nach to Hause home gekommen. came er=horse Onlyhecamehomethismorning. d. #Heute Today früh morning ist is nur only sie she.nom nach to Hause home gekommen. came sie=horse Onlyhecamehomethismorning. e. Heute Today früh morning ist is nur only das the Pferd house.nom nach to Hause home gekommen. came Onlyhecamehome thismorning. Howeverifthegenderofthehorseisspecifiedasin(64),eithercontinuation(64a)or(64b) ispossible. ItisclearthatweareworriedaboutMaryin(64a)sincewedonotknowwhere sheisandweareworriedabouta riderlessstallionin(64b). (64) Context: Marygoesforlongridesonherstallion(masc)everymorning. Everybody isworriedtoday. Whathappened? a. Heute Today früh morning ist is nur only er he.nom nach to Hause home gekommen. came Onlyhecamehomethismorning. CHAPTER2. WEAKANDSTRONGPRONOUNS 42 b. Heute Today früh morning ist is nur only sie she.nom nach to Hause home gekommen. came Onlyhecamehomethismorning. The apparent bias for human or animate reference in German seems to be a bias for elements that have an interpretable gender feature. I assume that there are only two genders in German, feminine and masculine. Neuter is not marked by features but is the absence of gender features. This would mean that neuter pronouns such as es and dascannotbeusedtodifferentiatebetweenantecedentsolelybasedongenderfeaturesin contrasttopronounssuchaser and sie asin(64). Thispredictionseemstohold. I do not have an ultimate answer here but it seems the answer has to lay in the connection between gender and sex. It is clear that gender in inanimate objects are necessarily purely grammatical, telling us only something about the lexicon. The gender for animate entities can be informative, with human extensions gender is necessarily informative,atleastin(Swiss)German. Numeral modification Siewierska (2004:10) points out that pronouns can be modified bylownumerals. ShegivetheEnglishexamplein(65)todemonstratethis. (65) ustwo/wefour StrongpronounscanbemodifiedbylownumeralsinStandardGermanas(66)illustrates. (66) Context: MaryandJohnaretwoexceptionallybrightstudentsinaclassof12. a. Nur They sie two zwei will werden it es sometime mal to zu something was make bringen. b. (Nur) They.artpro die two zwei will werden it es sometime mal to zu something was make bringen. ‘Thosetwowillmakesomethingoutoftheirlives.’ Strong pronouns referring to speech act participants can be modified by numerals as (67) demonstrates. This is true independent of case marking. (67a) shows this for ac- cusativemarkedpronounsand(67b)illustratesthepropertiesfornominativecasemarked pronouns. (67) a. Euch You.acc drei three kann can man one nie never alleine alone lassen. let ‘Onecanneverleavethethreeofyoualone.’ b. Ihr You.nom vier four lebt live ja MP wie like die the Maden maggots im in.the Speck. bacon ‘Thefourofyouarelivingthegoodlife.’ CHAPTER2. WEAKANDSTRONGPRONOUNS 43 Human reference is not required but weak pronouns are infelicitous with numeral mod- ification(68a). Note that the strong pronoun/d-pronoun in (68b) does not have a human referentbutisgrammatical. (68) a. #Sie They zwei two waren were teuer. expensive b. Die they.der zwei two waren were teuer. expensive Howevernumeralmodificationisrestrictedtosmallernumbers. Thisisdemonstrated with the contrast in (70). While a small number can specify the number of addressees in (70a),theminimalpairwithahighernumberin(70b)isungrammatical8. (70) a. Euch you.pl.acc zwei two brauch need ich I.nom noch still später. later ‘Iwillneedthetwoofyoulater.’ b. *Euch you.pl.acc 15 15 brauch need ich I.nom noch still später. later ‘Iwillneedthe15ofyoulater.’ Adjective modification Siewierska (2004) argues that pronouns with some restrictions can be modified by adjectives. She points out that accusative pronouns in English can be modified by a limited list of adjectives which precede the pronoun. poor me/lucky you are finewhile*poorI/*luckyhearenot. The English observation does not seem to be reproducible in German. Pronouns modifiedbyadjectivesarebadoutoftheblue. (71) adjectivemodification-NPElipse? a. Ihr you.pl.nom werdet will mit with (*schönen) us.dat uns beautiful ( ?? schönen) talk reden want wollen. b. (*schönen) beautiful wir we.nom ( ?? schönen) beautiful werde will mit with euch you.pl.dat reden talk wollen. want 8There seems to be one exception to the ban on large number modification which is illustrated in (69). While the personal pronouns sie ‘they’ can only occur with small numbers (69a) (69b), the d-pronoun die ‘they’ does not seem to have the same restriction (69c). I assume that this is not the pronominal use of die butthatwearedealingwithNPellipsishere. I will elaborate on this point further in the next section. (69) a. Siezweibrauchichnochspäter. b. *Sie15brauchichnochspäter. c. Die15(Schrauben)brauchich noch später. CHAPTER2. WEAKANDSTRONGPRONOUNS 44 Expletives Cardinaletti & Starke (1994) observe that only deficient pronouns can function as exple- tives. This is also true for German. We can see in (72) and (73) that only the deficient pronoun es can be used. Using a strong pronoun as in (72b) or (73b) make the sentence ungrammatical. (72) a. Es it.nom regnet. rains ‘Itisraining.’ b. *Das it.nom.der regnet. rains (73) a. Es it.nom ist is ein an Unglück accident geschehen. happen ‘Anaccidentoccurred.’ b. *Das it.nom.der ist is ein an Unglück accident passiert happen 2.4 SwissGermanPronouns TheSwissGermanpronounsystemhasanadditionallayerincomparisontoit’sstandard German counterpart. I will first introduce the Zurich German pronoun system as dis- cussed in Cooper (1994, 1999) since a number of analysis in the literature focus on this dialect. (Cooper1999)giveanindepthanalysisofZurichGermanclitcsanddemonstrates thatthesecliticsbehaveverydifferenttoromanceclitics. Shemakesabroaderdistinction betweenobjectandsubjectcliticsobservingthatsubjectcliticssurfaceinasomewhatmore restrictive environment than object clitics. In general clitics in Zurich German are seem lessrestrictedinthepositionthey canappearincomparisontoclassicalromanceclitics. 2.4.1 PronounsinSwissGerman I will first introduce the pronoun system in Zurich German, an East High Alemannic dialect, as described in Cooper (1994). I mentioned above that Werner (1999) notes that shecouldnotfindthiscliticinherresearch. Cooper(1994,1999)andWerner(1999)arethe mostextensivestudiesonSwissGermanPronounssofar. BothdealwithZurichGerman in particular, it seams therefore appropriate to introduce it here. The difference between Cooper’s(1994)strongandweakpronounsisinstressonly. Thedifferencebetweenclitics andtheindependentformsisthatthecliticformdoesnotcarrywordstress,hasonlyschwa vowels(@/1)plussomeconsonantreduction. Thissystematicdifferencebetweenfulland CHAPTER2. WEAKANDSTRONGPRONOUNS 45 Table2.6: PluralpronounsinZurichGermanbasedonCooper(1994:75) strong weak proclitic enclitic 1pl NOM miir mir mer- -mer ACC Ois ois DAT Ois ois 2pl NOM iir ir er- -er ACC Oi oi DAT Oi oi 3pl NOM sii si s- -s ACC sii si -s DAT Ine ine -ene,-ne clitic pronouns in Zurich German is not a property shared by all Swiss German dialects. Nübling (1992) reports about Bernese German, an Estern High Alemannic dialect, that cliticpronounsinthatlanguagearenotjustaphonologicalreductionoffullpronouns. Table2.5: SingularPronounsinZürichGermanbasedonCooper(1994:75) strong weak proclitic enclitic 1sg NOM iich ich ch- -i/; ACC miich mich -mi DAT miir mir -mer 2sg NOM dU du ;- -;/-d ACC diich dich -di DAT diir dir -der 3masc NOM eer er r- -r ACC iin in -en DAT iim im -em 3fem NOM sii si si- -si ACC sii si -si DAT ire ire -ere,-re 3neut NOM ees es s- -s ACC iins ins -s DAT iim im -em Thecarefulreadermighthavenoticedthattheparadigmhastwogaps. Thereareno cliticcounterpartsfor1 st and2 nd personpluralpronounsintheaccusativeandthedative form. Thisseemsnottobeduetoanyparticularreason. ThisgapdoesnotexistinBernese German (cf. Nübling 1992). The other gap is more important: We see that there are no CHAPTER2. WEAKANDSTRONGPRONOUNS 46 proclitic forms for accusative and dative clitics. Cooper (1994, 1999) only classifies clitics inthevorfeldasproclisisandwehaveseeninthediscussiononGermanabovethatweak object pronouns can appear in the vorfeld position only in very restricted circumstances. I will argue that object clitics can appear in the vorfeld position in these very particular cases. My discussion will mostly center on the Alemanic varieties spoken in Liechtenstein and just across the boarder in Switzerland and Vorarlberg. All my informants grew up lessthan20kmfromeachotherandwenttothesamesecondaryschool. Iwillrefertothe variety somewhat loosely as Lichtenstein German glossing over some lexical difference betweensubjects. ThevarietyunderdiscussionbelongstotheWestHighAllemanicdialectgroupandis thereforeverycloselyrelatedtoZurichGerman. ThePronounsystemdiffersfromZurich German mostly in the loss of the fricatives in the full forms (Ich! I, mich! mi, dich !di). Table2.7: PronounSysteminLichtensteinGerman full reduced 1sg NOM I -i ACC mi -mi DAT mir -mer 2sg NOM du -d/-; ACC di -di DAT dir -der 3masc NOM er -r ACC in -(e)n DAT im -(e)m 3fem NOM si -si ACC si -si DAT ire -(e)re 3neut NOM es -s9 ACC es -s DAT im -em full reduced 1pl NOM wi(r) -mer ACC üs DAT üs 2pl NOM ir -er ACC oi DAT oi 3pl NOM si -s ACC si -s DAT ine -ene/-ne Only full pronouns can receive stress, while clitics are never bearer of phonological stress. Clitics in Swiss German do not attach to words belonging to a particular syntactic category. This leads Cooper (1994, 1999) to argue that that clitics in Swiss German are CHAPTER2. WEAKANDSTRONGPRONOUNS 47 phonologicalbutnotsyntacticinnatureanddonotinvolveheadmovementaviewwhich I follow throughout this work. With respect to pronouns, Swiss German differs from Standard Germaninfourmajorrespects. i) weakpronounscanappearintheacc>datandthedat>accorder. ii) weak pronouns can be realized not only in the wackernagel position but also in a lowerposition,whichwillbeexploredblow. iii) SwissGermanhasphonologicalcliticswhicharealwaysweak/defectivepronouns. iv) SwissGermanhaspro-dropinrestrictedcases. SwissGermanprodrop SwissGerman1 st and2 nd personsingularsubjectpronounsmaybedropped. Pluralor3 rd person subject pronouns cannot be dropped. This is illustrated in the declarative in (74) (modified from Werner (1999:62)). We can see that in these declaratives the embedded subjectcanbedropped,whetherthisisa1 st person(74a)ora2 nd person(74b)pronoun. (74) a. I I bi am sicher, sure, das that ;en him.acc gnoo took ha. have1sg ‘IamsurethatItookit.’ b. I I bi am sicher, sure, das that ;en him.acc gnoo took häsch. have2sg ‘Iamsurethatyoutookit.’ Pro drop is not contingent on the kind of matrix subject or even on being embedded. We can see in (75) that the matrix subject can be dropped and that an embedded 1 st person singularsubjectpronoun(75a)orasecondpersonsingularsubjectpronounin(75b)does nothavetobeovertlyrealized. (75) a. Bisch are.2sg ;sicher, sure das that ; en him.acc gno took ha? have1sg ‘AreyousurethatItookit?’ b. Bisch are.2sg sicher,das sure ; that engno him.acc häsch? took have2sg ‘Areyousurethatyoutookit?’ Pro drop in Swiss German is not only restricted in that it can only affect the first and secondpersonsingularsubjectpronouns,butitalsohasphonologicalandmaybefurther structural restrictions. (76a) is a case where the subject pronoun has to be realized. In CHAPTER2. WEAKANDSTRONGPRONOUNS 48 the absence of the embedded subject pronoun the sentence cannot be interpreted as a declarativebutonlyasaninterrogativeasillustratedin(76b). (76) a. I I wöt want wüsse, know, wo where *(d) you wonsch. live.2sg ‘Iwanttoknowwhereyoulive.’ b. I I wöt want wüsse: know: Wo Where wonsch live.2sg ;? ‘Iwanttoknow: Wheredoyoulive?’ The exact restrictions on pro drop are not of particular relevant for this work. I will thereforonlypointtotheextensivediscussioninCooper(1994)andWerner(1999)forthe interestedreader. I will explore in the below order and location of weak pronouns in Swiss German before turning the diagnostics for weak pronouns. I will use the same diagnostics for weak pronouns as I did for Standard German above the only thing that this section will add, is to demonstrate that clitics/reduced pronouns are always weak and that the diagnostics,whicharealldiagnosticsforstrongpronounspreventreducedpronounsfrom occurring. 2.4.2 PronounOrder Reduced subject pronouns in Swiss German behave like weak subject pronouns in Stan- dard German and have to precede reduced object pronouns. This is illustrated in (77a). The full subject pronoun can also appear in other positions such as after or between the reducedobjectpronouns(77b). (77) a. ...,das that (d) you.nom mer me.dat (*d)en him.acc (*d)halt mp morn tomorrow bringsch. bring.2sg b. ...,das that (du) you.nom mer me.dat (du)en him.acc (du)halt mp morn tomorrow bringsch. bring.2sg ‘...thatyouwillbringhimtometomorrow.’ (Werner1999:81) However, reduced object pronouns can appear in both orders relative to each other in Swiss German. This is illustrated in (78). The dative-accusative order in (78a) as well as theaccusativedativeorderin(78b)aregrammatical. (78) a. De the Max Max hät has mer me.dat en him.acc voorgstellt. introduced CHAPTER2. WEAKANDSTRONGPRONOUNS 49 b. De the Max Max hät has en him.acc mer me.dat voorgstellt. introduced ‘Maxintroducedhimtome.’ Mostofmysubjectspreferredthe(78a)over(78b)inanoutofthebluecontextbutaccepted both. The dative–accusative order is ungrammatical in PCC violation cases (79a)10. In this caseonly theaccusative–dativeorderispossible(79b). (79) a. De the Max Max hät has mi me.acc em him.dat voorgstellt. introduced b. *De the Max Max hät has em him.dat mi me.acc voorgstellt. introduced ‘Maxintroducedme tohim.’ (Werner1999:81) 2.4.3 Thelocationofreducedpronouns IhavedemonstratedinthesectiononStandardGerman,thatweakpronounhavetomove to the Wackernagel position. The situation in Swiss German is different at least when we look at the PF linearization. In this section I will demonstrate that weak accusative pronoun cannot surface in the same position as the lexical DP counterpart. I will further moreillustratethatthereisalowpositionwithinvP,wheredativeandaccusativereduced pronounscansurfacewhilesubjectpronounshavetoberealizedoutsideofvP. We can see in (80) that reduced object pronouns, in contrast to subject pronouns, can surfaceinmultiplelocations. Wecanseein(80a)thatbothinternalargumentscanappear high directly following the subject pronouns. Both pronouns can be realized following a modal particle (i.e. halt) and a temporal adverb withinvP (80c), however the two object pronounsdonothavetoappearin thesamepositionas(80b)illustrates. (80) a. ...,das that d you.nom mer me.dat en him.acc halt mp morn tomorrow [ vP bringsch]. bring.2sg b. ...,das that d you.nom mer me.dat halt mp morn tomorrow [ vP en him.acc bringsch]. bring.2sg c. ...,das that d you.nom halt mp morn tomorrow [ vP mer me.dat en him.acc bringsch]. bring.2sg ‘...thatyouwillbringhimtometomorrow.’ 10Werner (1999) reports, that her informants, all speaker of Zurich German, had trouble interpreting the sentence in the dative–accusative order. Many of her informants rescued the ungrammatical sentence by using the full form mich instead of the reduced form mi. My subjects, who do no not make a clear phonological contrast between the full and reduced from, both are mi, corrected this sentence by using the accusative–dativeorder. CHAPTER2. WEAKANDSTRONGPRONOUNS 50 (Werner1999:82) Werner (1999) reports that (80a) is the least marked of the three. However there are also caseswherethelowerpositionisnotmoremarked. Thelowerpositionin(81)isnotmore markedthanthehigherposition. (81) ...,das that (mer (me.dat en) him.acc) den then halt mp de the Max Max (meren)zaigt. shows ‘...thatMaxshowshimtome.’ It seems that the crucial aspect about (81) is the low position of the subject which carries phonologicalstressandisinterpretedasacontrastivetopic. Speakersweregenerallymuch morewillingtoacceptlowrealizationsoftheobjectpronouniftheprecedingconstituent carriescontrastivestress. In contrast to reduced object pronoun, reduced subject pronouns can only occur in a high position to the left of modal particles. This is illustrated in (82a) note that this restriction is particular to reduced pronoun, strong pronouns can occur in all of these position(82b). (82) a. ...,das that (d) you.nom halt mp (*d) me.dat morn tomorrow (*d) him.acc mer bring.2sg enbringsch. b. ...,das that (du) you.nom halt mp (du) me.dat morn tomorrow (du) him.acc mer bring.2sg enbringsch. ‘...thatyouwillbringittometomorrow.’ The generalization that can be drawn form the examples above is that the subject has to move to a high position following C, presumably the spec TP position, while object pronouns do not. Cooper (1994) and Werner (1999) argue that reduced object pronouns can occur in the same position as lexical DPs however the while the data does support thisviewfordativearguments(83),reducedaccusativepronounshavetomove(84). The examples in (83) and (84), from (Werner 1999:82) illustrate this. We can see in (83a) that the reduced dative pronoun em ‘him’ does have to appear in front of a the low lexicalDPmarkedwithaccusativecase. Thisisnotalinearrestriction,theweakpronoun can appear linearly after the accusative argument if the lexical DP has moved to a higher position (83b). According to the test for high and low dative discussed in the previous section, bringe ‘bring’ is a high dative and the results of (83) fit with a view in which the reduced dative pronouns in the low position is realized in the same position as their lexicalcounterpart. (83) a. ...,das that mer we.nom (em) him.dat halt MP morn tomorrow (em)s’ the Gschenk gift.acc (*em)bringend. bring.pl CHAPTER2. WEAKANDSTRONGPRONOUNS 51 b. ...,das that mer we.nom (em) him.dat s’ the Gschenk gift.acc halt MP morn tomorrow (em)bringend. bring.pl ‘...thatwearegoingtobringhimthegifttomorrow.’ (Werner1999:82) Reduced accusative pronouns on the other hand have to move. (84) demonstrates this point. Since bringe ‘bring’ subcategorizes for a high dative argument, the accusative ar- gument is base generated lower than the dative and follows the dative argument in its base position. (84a) illustrates that the reduced accusative argument can surface in many positionsbutitcancruciallynotfollowthedativeargumentifthedativeargumentisina low position. Note that this cannot be due to a constraint on linearizion or some phono- logicalconstraint. Thedativeargumentcanprecedethereducedaccusativepronounifthe dativeargumentclearlymovedoutovP(84b). Itcanalsooccurinthesamephonological environmentfollowingalexicalDPaslongasthisDPisthesubject(nominativemarked) andnottheIO(dativemarked)(85). (84) a. ...,das that mer we.nom (en) it.acc halt MP morn tomorrow (en)am the Max Max.dat (*en)bringend. bring.pl b. ...,das that mer we.nom (en) him.acc am the Max Max.dat halt MP morn tomorrow (en)bringend. bring.pl ‘...thatwearegoingtobringittoMaxtomorrow.’ (Werner1999:82) (85) ...,dass that mer me.dat den then halt mp de the Max Max en it.acc bringt. bring.3sg ‘...thatMaxwillbringittome.’ ThedataaboveisrelevantsinceitcanonlybeexplainedbyassumingthatthereducedDO cannotberealizedinthesamepositioninwhichlexicalDOarebasegenerated. Thelowest positionwherereducedaccusativepronounscanberealizedisabovethebasepositionof high dative arguments, and likely lower than the base position of subjects. Is is therefore located between spec vP and Appl, somewhere between the location where subjects and IOarebasegenerated. Wehavetoassumethatmovementofthereducedsubjectpronoun and movement of the reduced accusative object pronoun is required, while the IO object mightremaininsitu. Thisassumptionsareillustratedinthethreein(86). (86) Movementofweakpronouns CHAPTER2. WEAKANDSTRONGPRONOUNS 52 TP TP T vP vP v AgrOP AgrOP AgrO ApplP ApplP Appl VP V acc dat acc nom nom The generalization in (86) is that reduced pronouns have to move to the specifier of the case assigning head. The dative argument does not have to move, since dative case is controlled by the applicative head, the nominative pronoun has to move to spec TP since TassignsnominativecaseandtheaccusativepronounhastomovetospecAgrOPbecause AgrOassignsaccusativecase. I stipulate here that the reduced accusative argument moves to a low case assigning head,belowvPwhichIlabeledasanobjectagreementheadAgrOinthelineofHaegeman (1993)11. Iwilldiscussinchapter4theadvantageofassumingsuchalowposition. When the pronoun agrees with the case assigning head, it can check features on this case assigning head. Importantly the agree relation is triggered by case and is not sensitive to intervention of other arguments. In the case of the accusative reduced pronouns the dativeargumentisnotanintervenerandwhenitmoves,itallowstheonntocheckany featuresonAgrO.Iwillexplore thispointmoreinsection2.5. I will assume that deficient pronouns, may they be full or reduced pronouns, move cyclically from a low position to a the Wackernagel position. This movement is tied to 11It is unusual to assume that the case assigning head is notv. By doing so, I lose the way to account for Burzio’sgeneralization,butitseemsnecessary to account for the Swiss German data. CHAPTER2. WEAKANDSTRONGPRONOUNS 53 the interpretation of the reduced pronoun as an anaphora (cf. López 2009). I want to suggestthatthereasonforthevariouslocationsofreducedpronounsistheresultofcyclic movementofweakpronounstotheWackernagelpositionsimilartotheSportiche’s(1988) analysisoffloatingquantifiers. ThisanalysisissimilartoFranks&ŽeljkoBošković’s(2001) analysis of clitics in some Slavic languages. They argue for multiple spell-out locations forcliticsinBulgarianandMacedonian. We have seen that reduced pronoun in Swiss German, is not restricted to a particu- lar position and can linearly follow elements of most lexical categories. Cooper (1999) concludes from this that reduced pronouns in Swiss German are not phonological not syntactic clitics. This means that movement of reduced pronoun in Swiss German is not the result of head movement but phrasal movement. They are in that different from pronouncliticsinRomanceorSlaviclanguages. 2.4.4 Diagnostics IwilldemonstrateherethatreducedpronounsinSwissGermanareweakpronouns. This allows us to easily identify weak pronouns, since reduced pronouns are often phonolog- ically marked. I use the same test for Swiss German as for Standard German with very similarresults. Coordination Coordinate pronouns in Swiss German mirror Standard German. (87) demonstrates this point. Thesubjectpronounin(87b)canhaveahumanoranon-humanreferent,whilethe subjectpronounin(87b)isrestrictedtohavingahumanreferent. (87) a. Er he isch is uhura very gross. big +human/ human ‘He/itisverybig.’ b. Er He und and de the danebe next.to isch is uhura very gross. big. +human/ human ‘Heandtheonenexttohimisverybig.’ Swiss German makes a phonological distinction between reduced pronouns that cannot have stress and full pronouns that can carry phonological stress. Reduced pronouns cannotpartakeincoordination(88a),onlyfullpronounscan(88b). (88) a. *I i ha have em him.dat und and de the Maria Maria a a gschenk gift koft. bought ‘IboughtMariaand himagift.’ CHAPTER2. WEAKANDSTRONGPRONOUNS 54 b. I i ha have im him.dat und and de the Maria Maria a a gschenk gift koft. bought ‘IboughMariaandhimagift.’ As with Standard German, es ‘it.neut’ is somewhat a special case. Cooper (1994) reportsthatstressedesinZurichGerman,canonlyhaveanimatereferenceIwouldagree withthat. Thestrongd-pronoundas‘it’in(90)canonlyhavenon-humanreference. The generalization seems to be that in contrast to other personal pronouns, es has to be clitizised independently of the phonological make up of the host. The only exception isinthevorfeldwhereitstilloftengetsrealizedasaproclitic. (89) I I ha have ’s/*es it.acc id in.the Spielgruppe daycare brocht. brought ‘Ibroughthim/her/ittodaycare.’ (90) I I ha have des it.acc.der id in.the Spielgruppe daycare brocht. brought ‘Ibroughtittothedaycarecenter.’ We have seen that es ‘it’ in Standard German cannot be coordinated because it functions alwaysasaweakpronoun. Thereduced’s‘it’canalsonotbecoordinatedinSwissGerman (91) (91) *I I.nom ha’s have-it.acc und and de the Kinderwage stroller id in.the Spielgruppe day-care brocht. brought ‘Ibroughthim/her/itandthestrollertoday-care.’ Topicalization Swiss German mirrors the data that we have seen for Standard German. The important extension is again the presence of phonological reduced pronouns which always behave asdefectivepronouns. Cooper(1999)arguesthatreducedobjectpronounscannotsurface in sentence initial position. Her argument is based on sentences like (92). The reduced pronoun cannot move to the vorfeld by itself (92a), only the full form, or the d-form can occurinthisposition(92b)(92c) (92) a. *En him.acc holt pickup sicher certainly de the Max Max.nom ab. from b. In him.acc holt pickup sicher certainly de the Max Max.nom ab. from c. De him.d.acc holt pickup sicher certainly de the Max Max.nom ab. from ‘Maxiscertaintopickhimup’ CHAPTER2. WEAKANDSTRONGPRONOUNS 55 However this generalization is not quite true. The reduced form can occur in the vorfeld position if its movement is piggybacked on some other movement. The examples in (93) are based on Werner’s (1999) (4:13). The reduced pronoun en ‘him’ can occur in the Wackernagel position (93a) or within vP (93b). Importantly, it can also move, together with the main verb, into the vorfeld position (93c). The contrast between (93c) and (92a) is important. Note that in the former the whole vP, including the reduced pronoun, is topicalized (which is grammatical) while in the later only the reduced pronoun is topicalized,whichisungrammatical. (93) a. De the Max Max hät has en him.acc scho already geschtert yesterday abgholt. pickedup b. De the Max Max hät has scho already gestert yesterday en him.acc abgholt. pickedup c. ?[En him.acc abgholt] pickedup hät has de the Max Max scho already geschtert. yesterday ‘Maxhaspickedhimupalreadyyesterday’ The same is also true for reduced dative pronouns. My informants considered (94) grammatical, particular with a context were helping was set in a contrast with other activitieslikelistening. Werner(1999:98)reportsalsothatsuchstructureswerejudgedas grammaticalbyherinformants. (94) [Em him.dat zhelfe] to-help han have i I.nom gar even nöed not probiert. treid ‘Ihavenoteventriedtohelphim.’ We see that like in Standard German, that topicalization of reduced object pronouns is ungrammatical. This is not due to some phonological restriction that bars them from movingthere. Wecanseein(93c)and(94)thatreducedobjectpronounscanoccurinthe sentence initial position, but such movement is only possible in a bigger constituent that alsocontainsthereducedpronounorincaseswherethesubjectdoesnothavetomoveto specTP.(95)isconstructedinthesamewayasMeinunger’s(2007)examplesforstandard German. Similarly to standard German, the object pronoun can move into the vorfeld position in the context of a indefinite subject which does not have to (covertly) move to spec TP. If the subject remains low, the reduced object pronoun will be promoted to spec CPtochecktheEPPfeatureonCgeneratingthestructurein(95) (95) De The Peter Peter.nom hät has s’ the Geld money.acc irgendwo somewhere im in.the Wald forest vergrabe. buried. S’ It.acc.wk findet finds sicher certainly niemert nobody me. more ‘Peterburiedthemoneyin theforest. Nobodyisgoingtofinditanymore.’ CHAPTER2. WEAKANDSTRONGPRONOUNS 56 WeakpronounscannotscrambleoverotherXPstomovespecCP.(96a)isungrammatical since the object pronoun cannot scramble over the subject pronoun i ‘I’ to move to spec CP. Strong pronouns (96b) and d-pronouns (96c) can move to spec CP independently of Frey(2006)styleFormalMovementi.e.movementtofulfillaV2requirement. (96) a. *En i him.acc.wk han have.1sg i I.nom e i no still nie never gse. seen. b. In him.acc.st han have.1sg i I.nom no still nie never gse. seen. c. De him.acc.d han have.1sg i I.nom no still nie never gse. seen. ‘Istillhaven’tseenhim The examples in (97) demonstrate the same for embedded subjects. The reduced subject pronoun cannot move to the topic position a movement which is possible for the strong pronounin(97b)andthed-pronounin(97c). (97) a. *r- he.nom-wk glob belive.1sg I I.nom isch is.3sg no still nie never do here gsi. been b. Er he.nom-st glob belive.1sg I I.nom isch is.3sg no still nie never do here gsi. been c. De he.nom-d glob believe.1sg I I.nom isch is.3sg no still nie never do here gsi. been ‘Ibelievehestillhasn’tbeenhere.’ Thedataabovedemonstratesthatweakpronounscannotscrambleoverotherintervening elements such as other weak pronouns. This is particularly important later on for the discussion about the availability of multiple word orders for weak pronouns in Swiss German. Reordering of reduced pronouns cannot be simply due to scrambling since, as hasbeendemonstratedhere,weakpronounscannotpartakeinscrambling. Null pronouns do not function as interveners. This is illustrated bellow. (98) from Cooper (1994:93) demonstrates what we already have seen above, which is that only full object pronouns can scramble over subject pronouns (98a) but reduced object pronouns cannot (98b). With a reduced subject and reduced object pronoun, only the reduced subjectpronounnotthereducedobjectpronouncanberealizedinSpecCP.Howeverpro asin(99b)doesnotfunctionasanintervener,inthissituationtheweakpronounhastobe promoted (99c)12. 12This example demonstrates that we are dealing with pro drop and not with Standard German topic drop,whichisdiagnosablebyverbinitialstructures. Note that this is out here. CHAPTER2. WEAKANDSTRONGPRONOUNS 57 (98) a. De the Peter Peter meint thinks [ cp im him hetted had mer we müese must aalüüte.] phone ‘Peterthinksweshouldhavephonedhim.’ b. *De the Peter Peter meint thinks [ cp em him hetted had mer we müese must aalüüte.] phone (Cooper1994:93) (99) a. De the Peter Peter meint thinks [ cp mer we hetted had em him müese must aalüüte.] phone ‘Peterthinksweshouldhavephonedhim.’ b. De the Peter Peter meint thinks [ cp em him hetsch had.2sg ;müese must aalüüte.] phone c. *De the Peter Peter meint thinks [ cp ;hetsch had.2sg em him müese must aalüüte.] phone ‘Peterthinksyoushouldhavephonedhim.’ 2.4.5 ModificationandtheStructureofPronouns c-modification We have seen above that c- modification cannot modify defective pronouns. We can see in (100a) that full pronouns can be modified by adverbs. Reduced pronouns, as in (100b) are ungrammatical. As in standard German, using the derived pronoun is infelicitous in suchacontextbecauseaweakerversionofthepronounisavailable. (100) Context: JohnandMarygotogethertoMass. DidyouseeMary? a. Nai, no i I.nom ha have.1sg nur only in him.acc.st gseh. seen ‘No,Ihaveonlyseenhim.’ b. *Nai, no i I.nom ha have.1sg nur only en him.acc.wk gseh. seen c. #Nai, no i I.nom ha have.1sg nur only de him.acc.d gseh. seen Thecontextin(100)isnottheonlyreasonwhyreducedpronounsarebad. (101)illustrates thatusingreducedpronounswithadverbialmodificationisungrammatical. (101) I I.nom ka can.1sg gar even nöd not globe, believe, dass that du you.nom nur only *en/in him.acc.wk/st gseh seen häsch. have.2sg ‘Ireallycannotbelievethatyouhaveseenonlyhim.’ Full forms cannot have an inanimate reference as is illustrated in (102a) however deter- minerpronounscanhaveaninanimatereferenteveniftheyaremodified(102b) CHAPTER2. WEAKANDSTRONGPRONOUNS 58 (102) Context: Iboughtquiteanumberofthings. a. *De the Tisch 1 table isch is tür expensive gsi, been, eigentlich actually isch is nur only er 1 it.masc.st tür expensive gsi. been. b. De the Tisch 1 table isch is tür expensive gsi, been, eigentlich actually isch is nur only de 1 it.masc.det tür expensive gsi. been. ‘Thetablewasexpensive,actuallyonlyitwasexpensive’. Thepicturethatwearegettinghereandthatwillbesupportedfurtherinthissectionis as follows: Swiss German full pronouns and determiner pronouns behave like pronouns anddeterminerpronounsinStandardGerman. Thedifferencebetweenthetwovarietiesis thatSwissGermanhasreducedformsandthisreducedformsalwaysbehavelikedefective pronouns. TheremarkablethingaboutSwissGermandefectivepronounsisthattheycan occurverylow. In contrast to human references, gender is not sufficient to disambiguate between inanimatereferenceswithpronouns. (103) Context: Every year, for the Corpus Christi celebration, the priest carries the monstrance (Monstrance.fem through the village and four men carry a baldachin (Baldachin.masc abovethepriest. Didyouseethemonstranceattheprocession? a. #Nai, No, I I ha have nur only in it.masc.acc.st gseh. seen b. #Nai, No, I I ha have nur only de it.masc.acc.st gseh. seen c. Nai, No, I I ha have nur only de the Baldachim baldachin gseh. seen ‘No,Ihaveonlyseenthebaldachin.’ The data in (103) again mirrors what we have seen for Standard German. The gender of a pronoun can only be used in the case of natural gender but not if gender is purly grammatical. In other words, gender has to be interpretable in the sense of De Vincenzi &diDomenico(1999). Numeralmodification Numeral modification with reduced pronouns is not possible (104a). Numeral modifica- tionrequiresthefullform(104b)orthed-form(104c). CHAPTER2. WEAKANDSTRONGPRONOUNS 59 (104) a. *I I.nom ha have ’s them.acc.wk zwei two gseh. seen b. I I.nom ha have si them.acc.st zwei two gseh. seen c. I I.nom ha have die them.acc.d zwei two gseh. seen ‘Ihaveseenthosethree’ The same is true for participants as illustrated in (105). (105a) illustrates that the reduced pronoun mer cannot be modified by a numeral but the full pronoun wir can (105b). Numeralmodificationisnotlimitedtonominativepronounsas(105c)illustrates. (105) a. *Gestert yesterday hend have -mer we.nom.wk drü three si her/them.acc gseh. seen b. Gestert yesterday hend have wir we.nom.wk drü three si her/them.acc gseh. seen ‘Thethreeofushaveseenher/themyesterday’ c. Gestert yesterday hend have -s they.nom.wk üs us.acc drü three gseh. seen ‘Theyhaveseenthethreeofusyesterday.’ Adjective Modification Adjective modification is not freely available with pronouns. Only some pronouns can be modified by pronouns (cf. Siewierska 2004). 3 rd person reduced or full pronouns are ungrammaticalwithadjectivemodification(106a)(106b). Only3 rd persondeterminersare possiblewithadjectives. (106c)is aclassicalcaseofNPellipsis. (106) a. *Denn then händ have d’ the Engländer English en him erliche honest eifach simply abgwähtlt. votedout b. *Denn then händ have d’ the Engländer English in him erliche honest eifach simply abgwähtlt. votedout c. Denn then händ have d’ the Engländer English de the.mas erliche honest eifach simply abgwähtlt. votedout ‘ThentheEnglishjustvotedthehonestoneout.’ Adjective modification is possible with 2 nd person pronouns as is illustrated in (107). In this case only full pronouns can be modified by adjectives, not reduced pronouns. (107a) and(107b)differinthatyouin(107a)functionsasthesubjectofacopulaclausewhileyou in (107b) functions as a determiner in a DP. Note that the you in (107a) can surface as a fullpronoun,areducedpronounorprowhileonlythefullformispossiblein(107b). The same is true for (107c) where the full pronoun only modifies an adjective. The difference CHAPTER2. WEAKANDSTRONGPRONOUNS 60 between regular NP ellipsis and cases such as (107c) is that NP ellipsis typically requires thattheelidedelementislicensedbebeingsalientinthecontexthowevernoappropriate nounneedstobesalientinthecontextforasentencesuchas(107b)tobefelicitous. (107) a. Du/d’/; you.nom bisch are.2sg en an erliche honest (ma) man und and gibsch give.2sg ’s it mer me.dat sicher certainly zruk back ‘Youareanhonestmanandwillsurlygiveitbacktomeagain.’ b. Du/*d’/*; you.nom erliche honest ma man gibsch give.2sg ’s it mer me.dat sicher certainly zruk back ‘Youareanhonestmanandwillsurlygiveitbacktomeagain.’ c. Du/*d’/*; you.nom erliche honest gibsch give.2sg ’s it mer me.dat sicher certainly zruk back 2.5 Analysis Strong pronouns behave like lexical DPs. This is clear for the PCC since full DPs also do not partake in the PCC. However, they also appear in the same environments as full (definite) DPs. For Standard as well as Swiss German this means that they can be coordinated, they are quite free with respect to scrambling; they can appear freely in the Vor-feld, can scramble in relation to each other in the middle-feld and they can be stressed. They can be modified by focus sensitive adverbs and can contain numeral modification. Weak pronouns are much more restricted in their behavior. They are restricted to certain positions, cannot be coordinated, cannot be modified and cannot be topicalized. This differentbehaviorissummarizedintable(2.8). Table2.8: Propertiesofweakandstrongpronouns Pronouns movement modification coordination topicalization strong weak obligatory X X X WeakandStrongpronounsclearlybehavedifferentlyinmanyrespects. Iwillpurpose here that the difference between the two kinds of pronouns is a structural one. I follow Cardinaletti & Starke (1994, 1996, 1999b), Déchaine & Wiltschko (2002) Stegovec (2017), in saying that the difference is a structural one and agree will play a role. I will utilize Pesetsky&Torrego’s(2007)system,asystemthatIwanttointroducebeforeturningtoan accountforthedifferentbehaviorof strongandweakpronouns. CHAPTER2. WEAKANDSTRONGPRONOUNS 61 2.5.1 Pesetsky&Torrego(2007) Pesetsky & Torrego’s (2007) system of Agree makes two important modifications to the standard agree model Chomsky (2000, 2001). One is the idea of features sharing the other isgthatvaluationandinterpretabilityofafeaturearenotlinkedbyabi-conditional. Featuressharingisdefinedasin(108)below. (108) Agree(Featuresharingversion) (i) AnunvaluedfeatureF(aprobe)onaheadHatsyntacticlocation (F )scans itsc-commanddomainforanotherinstanceofF(agoal)atlocation (F )with whichtoagree. (ii) ReplaceF withF ,so thatthesamefeatureispresentinbothlocations. (Pesetsky&Torrego2007:268) Pesetsky&Torrego(2007)adoptthenotationin(109). (109)denotesasinglevaluedfeature Fbeingsharedacrossfourlocations. (109) Notationforfeaturesharing F[73]...F[73]...Fval[73]...F[73] Pesetsky & Torrego (2007) second modification is to separate interpretability from valuation. In Chomsky (2001) interpretability is tied to valuation in that interpretable features are interpretable iff they are valued. This means that a feature could be stored in the lexicon in one of two ways, interpretable or uninterpretable. Pesetsky & Torrego (2007)arguethatthisisnotthecasebutthatafeaturecanbeinterpretablebutunvaluedin the lexicon. In that case the feature will be a probe and be valuated via Agree. A feature mightalsobevaluedbutuninterpretable,in thatcasethefeaturewillhavetoAgreewith aheadthathasaninterpretablebutunvaluedcounterpart. Thepossiblekindsoffeatures arein(110)whichisPesetsky&Torrego’s(2007)(9). (110) Typesoffeatures(boldface=disallowedinChomsky(2000,2001)) uF uF val [ ] uninterpretable, uninterpretable, valued unvalued iF iF val [ ] interpretable, interpretable, valued unvalued I do not want to go over Pesetsky & Torrego’s (2007) full motivation and argumentation for separating interpretability from feature valuation. The intuitive appeal of Pesetsky & Torrego (2007) comes from the idea that words that are marked for certain features, like wh-words for a question feature Q, are valued but have to agree with another head because they are uninterpretable in their base position. Functional heads that are clearly CHAPTER2. WEAKANDSTRONGPRONOUNS 62 associatedwithsuchwordscarrytheinterpretablecounterpartbutarenotvaluedforthis feature. WhatIwanttosayinthecaseofpronounsisthatpronounsarealwaysbasegenerated with valued features. The restrictions on weak pronouns originates in the lack of interpretabilityinthisbaseposition. Weakpronounsthereforehavetoagreewithahead thatcarriessuchinterpretable features. 2.5.2 ThestructureofPronouns I have illustrated above that Swiss German as well as standard German require a way to distinguish between pronouns. Both languages have a class of deficient pronouns and a class of strong pronouns. Cardinaletti & Starke’s (1996) strong pronouns can be split betweend-pronounsandstrongpersonalpronouns. Inthatstrongpersonalpronounsdo havethesamerequirementsasd-pronounsonpossibleantecedents. D-pronounsrequire asingleclearantecedentinthecontext,whichstrongpersonalpronounsdonot. Weak pronouns are nPs lacking further functional projections (112), while strong pronounsaren heads,with features(111)embeddedinafullDP13. 13This proposal is not in line with the DP-hypothesis and with the observation that nearly all lexical DPs inSwissGermanaswellasmanyvarieties of Standard German have overt determiners with lexicalNPs. CHAPTER2. WEAKANDSTRONGPRONOUNS 63 (111) strongpersonalpronouns DP ... nP NP n ... D (112) weakpersonalpronouns nP NP n Adjective and number modification is possible with strong pronouns but not with weakpronounsbecausenumberphrasesandadjectivephrasesneedtobeembeddedina DP. Thefeaturesinnarevaluedinthecasesdiscussedsofar. Notallpronounscomewith valued phi features. Kratzer (2009) points out that pronouns can be defective in that they come without valued features and gives examples along the line of (113a) as empirical evidence. Theimportanceoftheexamplebelowisthatthesentenceisambiguousbetween a bound variable reading where nobody else is picking up their son and a strict reading where nobody else is picking up our son but might be picking up their own child. The ambiguitydisappearsin(113b)wherethepossessivepronoundoesnotmatchthematrix subjectanymore. (113) a. Wir We sind are die the einzigen only die who unseren our Sohn son abholen. pickup.pl ‘Wearetheonlyoneswhopickupourson.’ b. Wir We sind are die the einzigen only die who deinen your Sohn son abholen. pickup.pl ‘Wearetheonlyoneswhopickupourson.’ Kratzer (2009) argues that the difference between the bound reading and the referential reading in (113b) is that the pronoun which gives raise to the bound reading enter the derivation withoutvaluedphifeatureswhilethereferentialpronounsdo. Déchaine & Wiltschko (2002) and Wiltschko (1998) argue that d-pronouns cannot function as bound pronouns. This is illustrated in (114) below. er ‘him’ in (114a) can be a weak pronoun and can have a bound reading. er in (114b) is c-modified and can only be a strong pronoun, the strong pronoun can have a bound interpretation, even though the speakers preferred the weak pronoun (114a) over the strong pronoun in (114b). The d-pronounin(114c)cannothaveaboundinterpretation. CHAPTER2. WEAKANDSTRONGPRONOUNS 64 (114) a. Niemand i Nobody i glaubt believes zu to halluzinieren, hallucinate, selbst even wenn if er i he i der the einzige only ist is der who die the Oase oasis sehen see kann. can. ‘Nobody believes to be hallucinating even if he is the only one who can see theOasis.’ b. ? Niemand i Nobody i glaubt believes zu to halluzinieren, hallucinate selbst even wenn if nur only er i he i die the Oase oasis sehen see kann. can Nobodybelievestobehallucinatingevenifonlyhecanseetheoasis c. #Niemand Nobody i glaubt believes zu to halluzinieren, hallucinate selbst even wenn if nur only der he i die the Oase oasis sehen see kann. can I want to suggest that this is due to the fact that D functions as a phase head. This meansthatunvaluedfeatureshavetomovetospecDPinordertogetvaluated. Iwantto stipulatethatstrongpersonalpronounsdojustthatinStandardGerman. Thiscanexplain why strong personal pronouns appear before modifying numerals or adjectives. n does not move to the left edge with d-pronouns, which prevents d-pronouns from getting a bound reading. The prediction is that a language that has bound readings with strong pronounwillalsohaveboundreadingwithweakpronounsbutnotnecessarilyviceversa. Déchaine&Wiltschko(2002)arguethatpronounsembeddedwithinDcannevergeta bound reading and that strong pronouns which can get a bound reading cannot be a DP. ItisuncleartomewhyhavingDprojectionswouldnecessarilyexcludedboundreadings. This should only be true for a language that prevents pronouns from moving to spec DP or to head move to D itself. I would argue that there are counter arguments against Déchaine & Wiltschko’s (2002) view. German does not only have d-pronouns but also indefinite pronouns. Indefinite pronouns are derived from indefinite determiners which are also base generated in D. We can see in (115) that an indefinite pronoun can get a bound reading. I will explore the case of indefinite pronouns a bit more in section 2.5.3 below. (115) Niemand i Nobody will wants dass that seine i =j his Mutter mother einen i one.masc.acc beim at.the Rauchen smoking erwischt. cought ‘Nobodywantstobecaughtbyhismothersmoking.’ Thedifferenceinbehaviorbetweenweakandstrongpronounsstemsfromtheinternal structureofDPs. IassumeinlinewithworkbyKoopman(2000)andAlexiadouetal. (2007) thattheDPnotonlycontainsanumberbutalsoapersonandagenderprojectionandthat CHAPTER2. WEAKANDSTRONGPRONOUNS 65 D like C makes finer grained distinctions. I will simplify the structural representation by unifyingtheseheadsunderasingle headinthetreesthroughout. 2.5.3 ThestructureofDPs Alexiadouetal. (2007)arguethattheDPhastwofunctionalfieldsdominatingthelexical field,ahigherfunctionalfieldwithdiscourse-orientedfunctionsandaloweroneencoding agreement properties and specificity. They argue that the discourse-oriented aspects of D can be seen with demonstratives. They propose that demonstrative pronouns move to spec DP from a lower projection for reasons of deixis. This movement would explain the occurrence of low demonstrative in Romanian (116) and high demonstratives, preceding thedefinitearticleinGreekasin(117). (116) tabloul painting acesta this rotund round al of său his Romanian (Alexiadouetal. 2007:111) (117) Dhen not perimena expected-1sg afti this tin the antidharsi reaction Greek (Alexiadouetal. 2007:108) The difference between the two languages is whether the demonstratives moves overtly or covertly into the D domain to get its deictic interpretation. In the case of Romanian, demonstrativesdonotovertlymoveintotheDdomainwhereinGreektheydo. Alexiadou et al. (2007) furthermore connect the D domain to the C domain in that both of these domainsconnecttotheutterancetothediscourse. I will not examine these higher functional projections. Alexiadou et al. (2007) and Simpson & Syed (2016) argue for a focus projection in the right DP periphery. There is reason to assume that some strong pronouns can occur in the deictic domain but not weak pronouns. Déchaine & Wiltschko (2002) argue for the special statues of we/us and you/you in contrast to they/them since only first and second person pronouns can fill the determinerroleinconstructionssuchasusstudentsoryouprofessorswhereastheylinguists isungrammatical14 ThecorrespondingstructuresrequirestrongpronounsinSwissGerman. Thereduced formisungrammaticalin(118) (118) SwissGerman a. Wir we.st / / *mer wk Schüler students gwünend win jeds every johr year gegen against d’ the lerer. teachers ‘Westudentswineveryyearagainsttheteachers.’ 14Déchaine & Wiltschko (2002) note that them linguists is grammatical for some speakers of American English. CHAPTER2. WEAKANDSTRONGPRONOUNS 66 b. Jeds Each johr year gwünend win ir you.st / / *er wk Schüler students gegen against d’ the lerer. teachers ‘Youstudentswineveryyearagainsttheteachers.’ c. *Jeds Each johr year gwünend win si they.st / / s’ wk Schüler students gegen against d’ the lerer. teachers ‘Thestudentswineveryyearagainstthestudents.’ I assume that the ungrammaticality of singular pronouns and third person pronouns with structures such as in (118) might be due to the fact that neither of these structures would further specify the referent of the DP. The referent of a first-person pronoun is always specified, in the presents of a lexical noun, 3 rd persons pronouns are redundant and second person singular pronouns are not that bad with nouns if the referent is not immediately clear. ‘Could you, Lacy, give me the cheese.’ is a good out of the blue utterance inacrowdedkitchen. I want to suggest that anaphoric elements have to move in order to get an anaphoric interpretation. López (2009) proposes that anaphoric elements have an [a] feature which gets evaluated at a phase head. Elements that have the [a] feature therefore move to phase heads. An alternative implementation of the same idea is that there are functional projections in the left sentential periphery and within the left periphery within the DP that license an anaphoric interpretation. Strong pronouns move to the left periphery of theDPinordertobeinterpretedanaphorically. Weakpronounsarenotembeddedwithin aDPandhavetomovetoleftsententialperipheryinordertolicensethe[a]feature. This accounts for why strong pronouns do not have to move while weak pronouns surface in theWackernagelposition. Iwanttosuggestthatthisisalsotheoriginoftherestrictionon coordination. Thecoordinatestructureconstraint(CSC)(Ross1967)excludesasymmetric movement out of coordinations. This means that a weak pronoun/clitic which has to move would violate the CSC. We know that across the board movement does not cause a CSC violation which is demonstrated for English in (119). Reduced pronouns in Swiss Germancanalsomoveacrossthe boardwhichisillustratedin(120). (119) WhatdidIbuyfromPeter andgivetoJohn. (120) I I ha have.1sg ’s i it.acc vom from.the Peter Peter.dat e i koft bought und and em the Max Max.dat e i geh. given ‘IboughtitfromPeterandgaveittoMax.’ The movement requirement of weak pronouns and the coordination restriction of weak pronounsarecausallyrelated. The last observation is that only strong pronouns can be topicalized, not weak pro- nouns. I want to suggest that this is connect to the possibility of c-modification. Only CHAPTER2. WEAKANDSTRONGPRONOUNS 67 strong pronouns can be modified by focus sensitive operators and only strong pronouns can be topicalized. López (2009) formalizes this insight with the introduction of a [c] (contrastive) feature. There is nothing in my approach that forces this. Elements can be contrastivewithoutbeingembeddedwithinaDP15. The whole analysis of weak pronouns raises an important question. Why is it that if wehavemorethannthatwearecommittedtothebuildingofthewholeDP?Whyshould syntaxmakeastarkdifferencebetweenoneelement(weakpronouns)andmanyelements (strongpronouns). IwanttoproposethatthismightbeduetothefunctionofDasaphase head. The purpose of phases is to chunk elements together in order to reduce working memory (cf. Richards 2011). In the case of weak pronouns, the load on memory cannot be decreased anymore since weak pronouns only have one element. However, a phase headwillreducememoryonastructurethatcontainsmorethanoneelementwhichgives us the D requirement on all nominal phrases but weak pronouns in Swiss German and Standard German. I will argue in chapter 4 that the reason the PCC is that weak pronoun need to check valued but uninterpretable features in the derivation and that strong pronouns do not have this restriction. The reason for why strong pronouns do not have this restriction is that strong pronouns are embedded within a DP and that all these features are checked against unvalued but interpretable features within the DP. Authors like Koopman (2006) and Alexiadou et al. (2007) argue for the existence of person number and gender heads within the DP. This is in line with work by Sigurðsson & Holmberg (2008) and others who argue that person and number are independent probes for subject-verb agreement. My approach is compatible with such a view, but I will treat for simplicity reasons all these probes as a bundle at a single head and assume that D enters the derivation with a interpretable but unvalued features. Assuming that D has unvalued features is warranted since determiners do agree with nouns in number and gender in German, as wellasSwissGerman. TheunvaluedfeaturesonDprobeforvaluedfeaturewhichitwillfindinthefeatures inn. Theestablishmentoftheagreementrelationwillvaluatetheinterpretablefeatures 15Multiple authors have argued for focus projections with in DPs to account for information structural driven reordering in the nominal domain (e.g. Simpson & Syed (2016)). It is tempting to assume that the lack of DP internal focus projections might be the origin for the lack of contrastive weak pronouns but elements like adverbs, which are not base generated within a DP, can also get a contrastive interpretation andbetopicalizedas(121)demonstrates. (121) Nur only langsam slowly ist is er he gelaufen. run ‘Heranonlyslowly.’ CHAPTER2. WEAKANDSTRONGPRONOUNS 68 onDmakingsurethatstrongpronounsarealwaysinterpretable. This is not necessarily the case for weak pronouns. There is nothing in the system which demands that weak pronouns enter the derivation with uninterpretable features but if they do, they will need to agree with a head with corresponding interpretable features. This is able to capture Béjar & Rezac’s (2003) Person-Licensing Condition (PLC) in (122) without stipulating that the PLC only applies to clitic/weak pronouns and not strongpronouns. (122) Person-LicensingCondition(PLC) "Aninterpretable1 st =2 nd personfeaturemustbelicensedbyenteringintoanAgree relationwithafunctionalcategory." (Béjar&Rezac2003:53) D-pronounsandindefinitepronouns Ihaveusedd-pronounsasanothercaseofstrongpronounshintingthatthereareadditional restrictions on d-pronouns. D-pronouns are what Schwarz (2009) calls strong definite articles. Iwouldarguethatonedifferentbetweenstrongpronounsandthesed-Pronouns lies in that d-pronouns have additional material in D (cf. Patel-Grosz & Grosz 2017) howeverthematerialinnisneverseparatelypronouncedfromtheelementsinD.Murphy (2017)proposesthatindefinitepronounsinStandardGermanallowustoseparateDfrom n. Some indefinite pronouns differ in form from their indefinite article counterpart. This is true for neuter pronouns which is illustrated in (123). eines ‘a.neut is only grammatical inthepronominalformbutnotinthedeterminerfunction. (123) a. Er he hat has ein(*-es) a Auto. car.acc.neut ‘Hehasacar.’ b. Er He hat has ein*(-es) a.acc.neut Auto. car ‘Hehasone.’ Inthepresenceofanadjective isalwaysexpressedontheadjective. (124) a. Er He hat has ein a sauber*(-es) clean.acc.neut Auto. car b. Er He hat has ein a sauber*(-es) clean.acc.neut Auto. car c. #Er He hat has ein-es a.acc.neut sauber clean Auto. car CHAPTER2. WEAKANDSTRONGPRONOUNS 69 Murphy (2017) argues that -es is the phonological realization of agreement withn that is expressedonDin(123b)andontheadjectiveifitispresent(124). Idonotwanttogointo to much details of Murphy’s (2017) account but point out that we can observe the same forSwissGerman. (125) a. I I.nom ha have e a.fem gros-i tall.fem frau seen gse. ‘Ihaveseenatallwomen.’ b. I I.nom ha have ai*(-ni) a.fem-fem gse. seen ‘Ihaveseenone.’ I would predict that indefinite pronouns behave like strong pronouns and this seems to be born out. Indefinite pronouns follow weak pronouns in Standard German (126a) and Swiss German (126b). Note that weak pronouns would have to precede these dative pronounsinbothlanguages. (126) a. Ich I habe have (einen) one ihm him.dat (einen) given gegeben. StandardGerman ‘Igavehimone.’ b. Iha (aini)em(aini)ge. Ihaveonehim.datgiven SwissGerman ‘Igavehimone.’ 2.6 Conclusion I have illustrated in this chapter how the difference between strong and weak pronouns can be viewed as a difference in structure. Weak and strong pronouns start of with the same set of structure and features asnPs with elided NPs. The difference is that strong pronouns are embedded within a DP while weak pronouns are not. Strong pronouns behave like lexical DPs because they are DPs. Weak pronouns on the other hand are not DPs and do therefore not behave like DPs. Weak pronouns are not DPs and are therefore arguments that need to have their valued but uninterpretable features checked by a functional projection that has the corresponding interpretable but unvalued feature set. They are, because of this susceptible to intervention effect, something that I will talk aboutinsubsequentchapters. Theotherdifferencebetweenstrongandweakpronounsis thatweakpronounsonlyhavefeatures. Weakpronounsareonlyn’swhichonlycarry features. Strong pronouns on the other hand are composed of multiple functional heads CHAPTER2. WEAKANDSTRONGPRONOUNS 70 andcanthereforealsohavemorethanjustthefeaturesfromn. Wecanseethisinthefact that strong pronouns come in multiple forms, such as d-pronouns, indefinite pronouns or strong personal pronouns. This means that strong pronouns can appear in and move topositionsthatareunavailableto weakpronouns. Chapter3 PreviousLiterature Before diving into the analysis put forward in this paper, I want to discuss the analyses proposed for the ultra-strong PCC and point out why these analyses cannot capture the Swiss German data presented in this work. Since Anagnostopoulou (2008) mentions the Swiss German explicitly I will discuss her proposal in more depth, before turning to Walkow (2013) who proposes an analysis which can capture the super-strong PCC in a collectionoflanguagesnotincludingSwissGerman. 3.1 Anagnostopoulou(2003,2008,2015) The main idea in Anagnostopoulou (2003, 2008, 2015) is that the PCC is the result of a single probe establishing an agree relation with multiple goals. The configuration is illustratedin(127)whichisAnagnostopoulou’s(2015)(7) (127) G2 G1 H Anagnostopoulou (2003) derives the PCC by stipulating that 3rd person accusative pro- nounsdonothaveapersonfeaturebut3rdpersondativefeaturesdo. 71 CHAPTER3. PREVIOUSLITERATURE 72 StrongPCC Anagnostopoulou (2003, 2008, 2015) has certain assumptions about the nature of the featuresofpronouns. Both1stand2ndpersonpronounsaremarked+personandarenot distinguishable from each other in this respect. However, they are distinguishable from 3rdpersonpronouns,whichdonothaveapersonfeatureiftheyareaccusativepronouns and a –person feature if they are a dative pronoun. The features are summarized in table 3.1. Accusativepronounsaredistinctfromdativepronounsinthattheirnumberfeatures areaccessiblewhilethenumberfeaturesondativepronounsandcliticsarenotaccessible. Thismeansthatdativepronounsandcliticscannotvaluenumberfeaturesonanyprobe. Table3.1: featuredistributionin(Anagnostopoulou2008:4) acc dat 1,2 +person,number +person,(inaccessiblenumber) 3 number,(noperson) –person,(inaccessiblenumber) With this feature schema, the strong PCC is accounted for via cyclic agree, where person and number features are being probed one after the other, cyclically. The base orderisshownin(128). InthePCCcases,DP1isthedativeargumentorIO,whileDP2is theaccusativeorDO. (128) Base: FP XP X’ ...ZP DP2 X DP1 {P,N} In a first step the person feature will now probe and find DP1. Minimality will prevent probingDP2. DP1cannowmovetoaspecifierpositionofFPasillustratedin(129a)Note thatDP1,i.e. thedativeargument,doeshaveaccessiblepersonfeatures,butnotaccessible number features. DP1 is therefore not an appropriate goal for the number feature. How- ever, DP2, that is the accusative pronoun, is an appropriate goal for the number probe. By establishing the agree relation, the DO may move as well. Anagnostopoulou (2003) CHAPTER3. PREVIOUSLITERATURE 73 assumesthattheDOmaymovetoapositionlowerthantheIOinaprocessreferredtoas tucking-in. Thisisillustratedin(129b). (129) a. StepI:PersonAgreewithhighargu- ment FP FP XP X’ ...ZP DP2 X e {0,N} DP1 b. Step II: Number Agree with lower argument FP FP FP XP X’ ...ZP e X e {0,0} DP2 DP1 Anagnostopoulou (2008) makes the prediction that in the presence of a dative pro- noun, only 3rd person accusative pronouns can be licensed, since they do not have a person feature which requires to be checked via agree. Following Chomsky (2000, 2001), (Anagnostopoulou 2003:274) argues that completephi-feature checking of DPs for struc- turalcase checkingmusttakeplace. In other words, for Anagnostopoulou (2003, 2008), the PCC is a result of structural case checking. If we rely on cyclical agree with an intervening dative pronoun, the DO can only get its number feature but not its person feature checked. Since 3rd person accusative pronouns do not have person features but only number features, they are the only pronouns that can get case licensed. The person feature on a first or second person accusative pronoun cannot be checked if a dative pronoun intervenes, preventing case assignmentandrenderingthestructureungrammatical. Note that 3rd dative clitics preceding 1/2 accusative clitics are ruled out by Anagnos- topoulou(2003,2008)becausethe3rdpersondativeclitichasa[-person]feature. Anagnostopoulou (2015) proposes that word order of weak pronouns functions as an indicator whether movement is triggered by a single head or by multiple heads. She assumestuckinginwillhappenifmovementistriggeredviamultipleagree/cyclicalagree andorderreversalifthemovementistriggeredbymultipleheads. CHAPTER3. PREVIOUSLITERATURE 74 WeakPCCinAnagnostopoulou(2003) Anagnostopoulou(2003)accountsfortheweakPCCinasimilarfashionasforthestrong PCC. She extends the framework to account for the weak PCC. According to Anagnos- topoulou (2003), the main difference between the strong PCC and the weak PCC is that whilethestrongPCCreliesoncyclicalagree,languageswithweakPCCallowformultiple agreetooccur. Thismeansthataprobecannotonlyagreewithonegoalbutwithmultiple goals one after another. To restrict the applicability of multiple agree, Anagnostopoulou (2003)introducestheconditionin(130). (130) AConditiononMultipleAgree Multiple Agree can take place only under non-conflicting feature specifications of theagreeingelements. (Anagnostopoulou2008:6) (130) is used in particular to prevent dative 3rd person reduced pronouns and clitics from co-occurring with 1/2 person reduced pronouns. Dative 3rd person reduced pro- nounshavea[-Person]featurewhile1/2personhavea[+Person]feature. Thismeansthat they have a conflicting feature specification preventing multiple Agree from taking place. However, having a 1/2 person dative argument and a 1/2 person accusative argument does not prevent multiple agree, since both arguments are [+Person]. 3rd person ac- cusative arguments do not have a person feature and do therefore not conflict with 3rd persondativeargumentsor1/2persondativearguments. ThisgeneratestheweakPCCrestrictionwhichonlybans3rdpersondativearguments fromco-occurringwith1/2personaccusativearguments. AccountingfortheSwissGermanData Anagnostopoulou(2003,2008)bringupBonet’s(1991)SwissGermandatasetintroduced in(131)and(132)below. Whatthesedatashow,isthatthereissomesortofPCCinSwiss German,whichonlyoccurswhendat>acc butnotwhenacc>dat. (131) a. D’Maria theMaria zeigt shows en him mir to-me b. D’Maria theMaria zeigt shows mir to-me en him ‘Mariashowshimtome’ (Bonet1991:188) (132) a. D’Maria theMaria zeigt shows mi me em to-him clitic CHAPTER3. PREVIOUSLITERATURE 75 b. D’Maria theMaria zeigt shows em to-him miich me strong c. *D’Maria theMaria zeigt shows em to-him mich me ‘Mariashowsmetohim weak (Bonet1991:188) WithoutmakingitclearwhichkindofPCCwearelookingatAnagnostopoulou(2008) accounts for the data in the small data set above in her system. To her, the difference between the two orders of reduced pronouns in Swiss German is the result of two-word ordersinthebaseposition. Shearguesthatthetwodifferentordersforreducedpronouns coincidewithtwodifferentpositionsforwherethedativeargumentinSwissGermancan bebasegenerated. (133) a. vP vP vP’ vP v’ VP v t i vAPPL t j v-TR ACC i DAT j b. vP vP vP’ VP V’ t i v t j v-TR DAT i ACC j Onlythestructurein(133a),wherethedativeargumentisbasegeneratedhigherthan the accusative argument, does it function as an intervener, restricting the occurrence of the PCC to structureslike (133a). If thedative argument is base generated lowerthan the accusativeargument,asschematizedin(133b),thedativeargumentdoesnotfunctionasan intervenerbetweentheaccusativeargumentandtheprobe. Thefeatureontheaccusative argument will therefore be checked by the probe and allow for case to be assigned. The dative argument is not assigned structural case and does therefore not require that all of itsfeaturesarecheckedforcasereasons. Tuckingingeneratesthesameargumentorderat PFastheoriginalbaseorder. CHAPTER3. PREVIOUSLITERATURE 76 ProblemswithAnagnostopoulou(2003,2008,2015) TheaccountspresentedinAnagnostopoulou(2003,2008)effectsrunintotwomajorprob- lems when accounting for the Swiss PCC effects. On the one hand, they are not able to account for the ultra-strong PCC effects in Swiss German. On the other hand, they have tostipulatetwobasepositionsfordativeargumentsinSwissGerman. AsfarasIcansee, wecanonlyfindargumentsagainstthisassumption,butnoargumentsinfavor. Accounting for the ultra-strong PCC I have demonstrated above that the PCC inSwiss German is neither a case of the strong nor the weak PCC but a case of the ultra-strong PCC. However, the mechanism proposed in Anagnostopoulou (2003) cannot capture the ultra-strong PCC, since it cannot make a distinction between the first and second person. Toaccountfortheultra-strongPCC,herproposedfeaturesystemhastobeextended. First,wewanttoaccountforthefactthatfirstpersondativescanco-occurwithsecond person accusative reduced pronouns but second person datives cannot cooccur with first person accusatives. Therefore, we are forced to assume that we are dealing with a probe that allows for multiple agree, since this is the only way that we can account for having thedativeaswellastheaccusativeargumentbeingaspeechactparticipant. However,we also must rule out that a second person dative argument can co-occur with a first person accusative argument. To do that, we have to make sure that first person and second person are distinguishable in their feature configuration. Consequently, I introduce the featureauthor,whichappearsas[+author]forfirstpersonandas[-author]with2ndperson dativereducedpronounsandisabsentin3rdpersonreducedpronounsandin2ndperson accusative pronouns. By introducing this feature set, we are able cover the same facts as Anagnostopoulou (2003): the weak PCC prohibits 3rd person dative arguments from co-occurring with first or second person arguments but not with 3rd person accusative arguments. Table3.2: ModificationofFeaturedistributiontocaptureultra-strongPCC acc dat 1 +person,+author,number +person,+author(inaccessiblenumber) 2 +person,number +person,-author(inaccessiblenumber) 3 number,(noperson) -person,(inaccessiblenumber) While the introduction of a [+author] feature for first person is not a bold step, the introductionofa[-person]featurefordativeargumentsisanovelty. Asaconsequence,we CHAPTER3. PREVIOUSLITERATURE 77 areleftwithafeaturedistributionwhereaccusativepronounsonlyhavepositivefeatures while dative features have no accessible number features. We can increase symmetry in our feature system by stipulating that 3rd person dative pronouns also come with a [-author]featureasillustratedintable3.3. Table3.3: ModificationofFeaturedistributiontoenhancesymetry acc dat 1 +person,+author,number +person,+author,(inaccessiblenumber) 2 +person,number(noauthor) +person,-author,(inaccessiblenumber) 3 number,(noperson,noauthor) -person,-author,(inaccessiblenumber) I am proposing a paradigm that does not account for why the ultra-strong PCC in SwissGermanalsobansco-occurrenceofa3rdpersondativeanda3rdpersonaccusative reduced pronouns. I have argued above that this restriction is part of the Swiss German PCC.Idon’tseeawayofnaturallyincorporatingthisrestrictionintotheaccountproposed byAnagnostopoulou(2003). The feature system as in Anagnostopoulou (2003) as well as my extension in table 3.3 exhibit a particularity. Anagnostopoulou (2003) employs ternary features, since a feature can be either [+], [–] or absent. However, these features are not freely distributed, accusativepronounsneverhavenegativefeatures,dativepronounsneverhaveanyabsence offeatures. Inotherwords,bothdativeandaccusativefeaturespecificationsareinternally binary. Thisisaconsequenceofthespecificphenomenathatwearecapturing,butthereis nothinginthesystemthatpreventsotherconfigurations. Unsurprisingly,werunintothe problem of over generating possible PCCs that to my knowledge do not exist1. Note that overgeneralization is not a problem that Anagnostopoulou (2003) is faced with, since she only accounts for strong and weak PCCs and does not propose the extension of features asintable3.3buttheextensionisnecessarytocapturetheultra-strongPCCinhersystem. There is no obvious way to extend Anagnostopoulou (2003) account for the weak and strong PCC to capture the ultra-strong PCC without creating a system that ends up dramaticallyovergeneratingpossiblevariantsofthePCC.Furthermore,therestrictionon 1There is no reason in this system for why the PCC ought not also occur reversed. One can imagine a weakPCCthatrequires3rdpersonargumentstobedativeinsteadofaccusative. Thiswouldbearchivedby changing accusative 3rd person pronouns to have a [–person] feature and the dative argument not having a person feature. We could also image a language that prevents 3rd person arguments to combine with anything but other 3rd person pronouns by having accusative and dative 3rd person reduced pronouns havinga[–person]feature. Thisisclearly undesirable since such languages do not seem to exist. CHAPTER3. PREVIOUSLITERATURE 78 3rd person accusative and 3rd person dative pronouns from co-occurring in in the PCC order in Swiss German cannot be captured and would require additional stipulation. In addition, there is also another problem with her account of the Swiss German PCC data, concerningherapproachtothequestionofordering,whichIwilldiscussinthefollowing paragraphs. Alternative Word orders Anagnostopoulou’s (2003) analysis hinges on the idea that Swiss German allows for dative arguments to be base generated in one of two positions, onelowerthantheaccusativeargumentandonehigherthantheaccusativeargument. In her account, the two possible locations for base generation of dative arguments result in thetwoordersofreducedpronouns. Shedoesnotgiveanyempiricalreasonsforassuming such a move which is not grounded in her proposal and while having two possible sites for dative arguments works well in the system as scotched by Anagnostopoulou (2003, 2008),suchanassumptioniscontradictorytothedatathatwecanobserve. Idemonstratedinsection(2.4)thatweseereconstructioneffectsinSwissGermanwhen theaccusativeargumentprecedesthedativeargumentbutnottheotherwayaround. This indicates that the dative argument has to be base generated higher than the accusative argument. IfbothorderswherepossibleasstipulatedbyAnagnostopoulou(2003,2008)we wouldnotexpectthat. Wewouldmuchratherpredict,contrarytofact,thatconstructions with two internal arguments are always scopally ambiguous in Swiss German since the dativeargumentcouldbeinterpretedasbeingbasegeneratedloworhigh. There is also at least one other argument against the view that Swiss German allows for dative arguments to be base generated optionally in a lower and a higher position. And that this optional base word order is the cause for two different surface word orders and the existence of the PCC in only one of them. The argument comes from applicative constructions. As is demonstrated in (134) Swiss German has applicatives which are dative marked. Applicatives in Swiss German have a benefactive meaning. khoft ‘bought’ in (134a) is a transitive verb. It does not force us to assume any particular benefactor of the speaker’s action. Asentencelike(134b)hasanexplicitbenefactorofthespeaker’saction. Notethat this does not necessarily involve any sort of transfer of possession as the continuation in (134b)or(134c)demonstrates2. (134) a. I I ha have an a wie wine khoft. bought 2(134c) requires a bit of a context but is perfectly fine in a situation where Maria for example is lying in thehospital. CHAPTER3. PREVIOUSLITERATURE 79 ‘Iboughtabottleofwine’ b. I I ha have der you.dat an a wie wine khoft bought (ka can der you.dat en 3sg.masc.acc jetzt now aber but no yet noed not geh.) give ‘IboughtabottleofwineforyoubutIcannotgiveittoyoujustyet.’ c. I I ha have der you.dat bluame flowers khoft bought (und and ha have si it i in dim your name name scho already de the Maria Maria ge.) given ‘IboughtflowersandhavealreadygivenittoMariainyourname.’ (135) demonstrates that we can have weak pronominal referring to the internal ar- gument as well as to the applicative argument. Importantly the ordering of reduced pronouns is not restricted in this case as is illustrated in the (135a)(135b) pair. The dative pronouncanbeprecedingorfollowingtheaccusativepronoun. Importantly,theordering of the two arguments does not make a semantic difference. The dative argument in both sentences isnecessarilyabenefactiveandnotnecessarilytherecipient. (135) a. I I ha have -n 3sg.masc.acc der you.dat khoft. bought ‘Ihaveboughthimforyou’ b. Ihaderenkhoft. Ihaveyou.dat3sg.masc.accbought ‘Ihaveboughthimforyou’ ThisisrelevantsinceweknowfromPylkkänen(2008)thattheheightoftheapplicative is related to its interpretation. She shows that benefactive applicative constructions cross linguistically are merged in high position such that the applicative head takes scope over the predicate, in (134) this is the buying of the wine or flowers3. She argues that a benefactive interpretation requires a high applicative structure since the benefactive is in fact not a relation between the internal argument and the applied argument. In other words, the hearer is not the benefactor of the wine in (134b) but the benefactor of the buying of the wine. This is quite clear in (134c) where the hear does is not ever in possession of the flowers or does much with them. The hearer however is the benefactor ofthespeakerspurchaseoftheflowers. 3For Pylkkänen (2008), high applicatives are of the formx.e. Benefactive(e,x) which means that the specifier of the applicative head is interpreted as the benefactor of the predicate via event identification. Theapplicativeheadthereforeestablishes a relation between an event and an argument of type e. CHAPTER3. PREVIOUSLITERATURE 80 The crucial bit here is that the benefactive interpretation requires a high applicative, for our case this means that the dative argument is merged higher than the accusative argument. But even though we have a benefactive interpretation of the dative argument in (135) both word orders are possible. The surface word order can therefore not be contingentonthebasewordorder. Thebasewordordermustbethesameindependently ofthesurfaceordering. 3.2 Walkow(2013) Walkow(2013)triestoaddressthequestionofvariationwithrespecttothePCCandwith respect to resolution strategies employed. He mainly looks for this endeavor at standard ArabicandCatalan. AsothersyntacticanalysisofthePCCheusesthePLCasstatedinBéjar&Rezac(2003) copiedin(136). (136) Person Licensing Condition (PLC): An interpretable 1st/2nd person feature must belicensedbyenteringintoanagreerelationwithafunctionalcategory. In contrast to most other literature, Walkow (2013) argues that the PCC is the result of a probeagreeingwiththeDObeforeagreeingwiththeIO.Thedifferencebetweentheultra- strongPCCandthestrongPCCistohimadifferenceinprobespecificity. Indifferenceto the previous analysis he is able to capture the ban on two 3rd person internal arguments asaresultofthePCC. I will sketch his analysis in the following however there is on main problem for ex- plaining the Swiss German data. To Walkow (2013) the reason for the PCC is that the accusativeargumentisblockingaprobefromagreeingwithadativeargument. Crucially the accusative argument has to intervene between the probe and the dative argument in somesenseforthistowork. However,IwilldemonstratethatthePCCdisappearsexactly inthesecaseswhereWalkow(2013)expectsthePCCtoarise. Theoriginsofclitics Walkow(2013)proposesthatcliticsarethemorphologicalrealiza- tion of an agree operation between a head and an internal argument. He adopts the idea thatpersoncategoriesarebundlesofprimitivefeaturesasinlistedintable3.4. CHAPTER3. PREVIOUSLITERATURE 81 Table3.4: Personcategoriesasbundlesofprimitivefeatures(Béjar&Rezac2009:42)from (Walkow2013) Person: 3 rd 2 nd 1 st Feature [] [] [] Specification: [part(icipant)] [part(icipant)] [spea(ker)] ThePCCisnowtheresultthataheadwithunvaluedfeaturesprobesforitsfeaturesbut notasabundlebuteachfeatureafteranother. Theorderoftheprobesisnotimportant. DerivingtheSuperStrongPCCwithDOprecedingIO Walkow(2013)arguesthatthe accusative object in Catalan precedes the dative argument. The probe, which he locates onv probesdownandwillfirstfindtheDOandthentheIOasisdemonstratedinFigure 1 from (Walkow 2013). In the case of the super strong PCC the v enters the derivation with an unvalued , an unvalued [part] and unvalued [spea] feature. [u] will probe downandwillfindanappropriatetargetintheDO.Minimalitywillforceagreementwith DO and prevent the probe from finding the [] in IO. However, if [upart will not find an appropriate target in DO since 3rd person does not have an [part] and will probe further downuntilitfindsanappropriatetargetinIOasisillustratedinFigure2. If clitics require to be licensed via agree, it is easy to see that the mechanism will only generate convergent derivations if the IO is more specified than the DO. Which is the generalization of the super strong PCC. If the DO is more or as specific as the IO, the DO will block an agree relationship betweenv and IO. However, PCC in cases where the IO precedetheDOaremorecommonintheliterature. Deriving the Super Strong PCC with IO preceding DO We have seen in the above cases that if the DO precedes the IO Walkow (2013) assumes that the v introduces the CHAPTER3. PREVIOUSLITERATURE 82 relevantunvaluedfeatures. ForlanguageswhereIOprecedestheDOheassumesthatthe head introducing the unvalued features is the same head that introduces the IO itself. In most theories, this is the applicative head. Walkow (2013) labels in his Arabic example the relevant head as a v shell which introduces the IO while presumably the external argumentisintroducedbyanadditionalv head. He derives the PCC now by first merging the DO with the probing head. The head checks its unvalued features first with the DO object since it is the only argument in the derivationatthispoint. Afterthis first roundofagreetheIOismerged. Tolabelthe new structure v projects its label and with it the unvalued features. From this position the unvalued features agree with the IO. Again, the IO needs to be licensed via agree which canonlyoccuriftheDOislessspecifiedthantheIOderivingtheultra-strongPCC. Walkow’s(2013)accountseemstofollowwithouttoomanynewassumptionsincases where the DO is in a position where it c-commands the IO and is itself c-commanded by the relevant probing head. However, in cases where the IO c-commands the DO this is not the case. Walkow (2013) needs to make some rather unusual assumptions about the naturestructurebuildingoperations,oraboutthenatureofagree. It looks like he needs to assume either that downward agree is not only an operation that probes the c-command domain of an element but also all elements that it dominates orthatagreetakesplacebeforeastructureislabeled. Some authors, e.g. Hornstein (2009), who argue that merge is composed into two processes, concatenate and label. However, to my knowledge nobody proposed that a c- commandrelationcanbeestablishedbetweentwoconcatenatedelementswithoutlabeling thestructurefirst. Infact,Larson(2013)explicitlywantstoexcludethispossibility. The other option is to assume that agree establishes relations in one of two ways. One is dominance and the other is c-command. This is what Walkow’s (2013) tree above suggests. Note that we need agree via dominance only to check undervalued features on labels,aproblemcreatedbyassumingthatunvaluedfeaturescanbepartofalabel. Walkow(2013)arguesthatthatcliticsarebasicallythemorphologicalrealizationofan agree relation. If no agree relation can be established the DP that was not agreed with is pronouncedregularly. Thecliticisthereforethepronunciationofthecheckedfeatureson CHAPTER3. PREVIOUSLITERATURE 83 theprobe. However,itistomyknowledgearatherunusualassumptionthatthechecking of this features causes the DP (the interpretable features) not to be pronounced at PF. It seems we would expect clitic doubling to be the norm for languages with PCC. This is to myknowledgenotthecaseandWalkow(2013)doesnotexplainwhyorhowtheDPisnot beingpronounced. Besides these technical issues with the hypothesis there are also empirical problems. Note that the Swiss German PCC only exist when the dative pronoun precedes the ac- cusative pronoun, however to Walkow (2013) the PCC is the result of a licensing probe agreeingfirstwiththeaccusativeargument. This idea runs quickly into different troubles with the data set discussed here. For simplicity reasons, I want to look at the question of apparent repair strategies for PCC violations. Walkow(2013)arguesthattheapparentrepairstrategiesthatlanguagesdeploy are in fact the just the PF realization in the absence of agree. The problem that the Swiss Germancasesposeisthattherepairstrategydoesnotinvolvedifferentpronounsbutonly differentlinearorder. IfthereasonforthePCCinSwissGermanisthatnotbothreduced pronouns can be agreed with it is hard to capture that this only affects one of the two orders,i.e.IO>DObutnottheotherorder. 3.3 Stegovec(2017) Stegovec (2017) discusses an order sensitive version of the PCC in Slovenian. (137) and (138) are his (6) and (8) respectively. (137a) illustrates the classical PCC. In the dat>acc order, the dative may be any person as long as the accusative argument is 3rd person (137a). If the dative argument is a 3rd person pronoun, the accusative argument cannot bea1stor2ndpersonpronoun(137b). (137) Slovenian a. Mama mom mi/ti/mu 1/2/3.m.dat ga 3.m.acc bo will predstavila. introduce 1/2/3.io>3.do ‘Momwillintroducehimtome/you/him.’ b. *Mama mom mu 3.m.dat me/te 1/2/m.acc bo will predstavila. introduce 3.io>1/2.do ‘Momwillintroduce me/you/himtohim.’ Iftheaccusativepronounprecedesthedativepronoun,weseethereverseparadigm(138). If the reduced accusative pronoun precedes the reduced dative pronoun, the accusative pronouncanbeanypersonifthedativepronounis3rdperson(138a)butiftheaccusative pronounisa3rdpersonpronoun,thedativepronouncannotbea1stor2ndperson(138b). CHAPTER3. PREVIOUSLITERATURE 84 (138) Slovenian a. Mama mom me/te/ga 1/2/3.m.acc mu 3.m.dat bo will predstavila. introduce 1/2/3.do>3.io ‘Momwillintroduceme/you/himtohim.’ b. *Mama mom ga 3.m.acc mi/ti 1/2/m.dat bo will predstavila. introduce 3.do>1/2.io ‘Momwillintroduce himtome/you.’ If the both arguments are 3rd person, both orders are possible, however Stegovec (2017) reports that dat>acc order is the unmarked order in that case. There seems to be inter- speakervariationwithrespecttowhetherSlovenianhasthestrongorweakPCC.Stegovec (2017) records that some speakers do not allow for the co-occurrence of two internal participant arguments while others do (139), his example (9) and example (10). Speakers who consider (139a) and (139b) ungrammatical are speakers of the strong PCC variety and speakers who consider these sentences grammatical are speakers of the weak PCC variety. (139) Slovenian a. %Mama mom mi/ti 1/2.dat te/me 1/2.acc bo will predstavila. introduce ‘Momwillintroduce you/metome/you’ b. %Mama mom te 2.acc mi 1.dat bo will predstavila. introduce ‘Momwillintroduceyou/metome/you’ The Slovenian case is interesting since the PCC in Slovenian is not actually tied to case but to the order of internal arguments. In that, it is reminiscent of the PCC in Swiss German, a connection which is made by Stegovec (2017) himself. I will discuss his proposal more in depth before turning to why it is unable to capture the Swiss German data. Stegovec’s (2017) analysis is based on Pesetsky & Torrego’s (2007) notion of agree and the idea that interpretable features are not necessarily valued features. This idea is summarized in more detail in section 2.5.1. Stegovec (2017) proposes that deficient pronouns (weak pronouns or clitics) do not come with valued person features but only interpretable ones. The only valued features deficient pronouns (pro DF ) come with are number and gender features. For him, first and second person feature valuation for internalargumentsoriginatesinvwhilethirdpersonisthedefaultspell-outifvdoesnot valuatethepersonfeaturesonthe deficientpronoun. Stegovec (2017) derives the strong PCC assuming thatv comes with valued but unin- terpretable features and unvalued number and gender features (u:__). The unvalued CHAPTER3. PREVIOUSLITERATURE 85 features probe for valued and find them on the IO (141). The unvalued features on the IO now get valuated by the uninterpretable but valued features onv via Maximize Agreeasdefinedin(140). (140) Maximize Agree: If Agree holds between heads X and Y for any feature, then all the unvalued features on X and Y must be valuated by any matching features on theotherelementregardlessofthedirectionofvaluation. (Stegovec2017:18) Importantly since all the features onv are checked at this point,v does not function as a probe anymore and does not establish an agree relation with the DO (142). This means thattheinterpretable featurescannotbevaluedviaAgree. (Stegovec2017:14)proposes that featurescangetadefault3Pvalueiffvaluationviaagreeisnotpossible(143). (141) DerivingthestrongPCCpart1 vP ApplP ApplP VP DO i:val i:__ v Appl IO i:val i:__ v u:__ u:val (142) DerivingthestrongPCCpart2 vP ApplP ApplP VP DO i:val i:__ )3P v Appl IO i:val[2] i:1/2/3P v u:[2] u:val (143) a. Apro DF hasunvalued[i]featuresthatmustbevaluedbeforespell-outbecause anunvalued[i]isillicitatLF(Pesetsky&Torrego2007); b. An unvalued [] can get a value either (i) via Agree with a valued [], or (ii) bygettingadefault3p valueiffvaluationviaAgreeisimpossible. The reverse order is derived via movement of the DO object to spec ApplP. Unvalued u probe finds valued on DO and Maximize Agree will check the features onv IO with the featuresonv (144). (144) DerivingthereversedstrongPCC CHAPTER3. PREVIOUSLITERATURE 86 vP ApplP ApplP ApplP VP DO v Appl IO i:val i:__ )3P DO i:val[2] i:1/2/3P v u:[2] u:val In Stegovec’s (2017) approach, the strong PCC is the result of both the person features values originating onv and the fact that only the higher of the two arguments enters in an Agree relation with v. Like this, the lower argument cannot valuate its unvalued featureswhichleadthesefeaturestobespelledoutat3P.TheapproachhingesonMaximize Agree which mimics upward agree but requires that a downward Agree relation can be established. Stegovec’s (2017) account is innovative and very interesting but there are some con- ceptual issues. The interpretation of the pronouns is due to the valuation onv but there is nothing in the account that controls the movement of internal argument belowv. This isimportantsincethismovementultimatelydeterminesthevaluationoftheinterpretable features. Theapproachwillonlygenerategrammaticalsentencesbuttheinterpretation oftheinternalargumentswouldendupbeingrandomsinceitiscontrolledbythemove- ment or not movement of the IO. This equates to a significant lookahead problem. Note that the interpretation of the pronoun depends on features introduced on v, not on the pronounitself. Reorderingofthepronounsoccursbeforehandbutcannotbemotivatedby reference to the specifications of the pronouns themselves because they are for all intents andpurposesequivalentatthispoint. DerivingtheWeakPCC Stegovec (2017) also proposes a way to derive the weak PCC. He assumes that the main difference between the weak and the strong PCC is the specification of the features inv. Whilev has an unvalued feature in case of the strong PCC, the feature is valued in case of the weak PCC. This means thatv has only valued features and does not function CHAPTER3. PREVIOUSLITERATURE 87 asaprobepreventing featureswithinApplPfromgettingvaluatedviaAgree. Stegovec (2017) proposes that this inability of feature valuation triggers movement of defective pronouns to a position c-commanding the head containing the valued features (145a). Stegovec (2017) argues that this movement does not lead to reordering of the arguments duetotuckingin(cf.Anagnostopoulou2003). Themotivationforthiswillbecomeapparent shortly. Importantly, thanks to the afore mentioned movement, both pronouns have now an appropriate goal in their c-command domain which functions as an appropriate goal fortheirunvalued feature(145b). (145) a. DerivingtheweakPCC XP X’ X’ YP Y’ t 2 Y t 1 X u:val pro DF i:__ pro DF i:__ b. DerivingtheweakPCC XP X’ X’ t 1 ...t 2 ... X u:val pro DF i:1/2P pro DF i:1/2P Withoutmodification,(145b)wouldseemtorequirethatbothinternalargumentshave the same features since the features are valuated from the same goal. Stegovec (2017) makesacoupleofassumptionstoderivetheempiricalfactsoftheweakPCCinSlovenian. He first assumes that features are privative along the lines of Béjar & Rezac (2009), as illustratedintable3.4,repeatedintable3.5below. CHAPTER3. PREVIOUSLITERATURE 88 Table3.5: Personcategoriesasbundlesofprimitivefeatures(Béjar&Rezac2009:42)from (Walkow2013) Person: 3 rd 2 nd 1 st Feature [] [] [] Specification: [part(icipant)] [part(icipant)] [spea(ker)] Hearguesthatsincehavinga[spea]featurerequiresaheadtoalsohavea[part]feature, any head that receives [spea] feature will first have agreed with a head that has a [part] feature. 3rd person is again default agreement that results from not agreeing withv due totheabsenceofmovementorbecausetherelevantfeaturesonvhavebeendeleted. Since thehigherargumentmovesfirstandthereforeagreesfirst,thehigherpronounwillnever be 3rd person when the 2nd pronoun is not. The problem arises in the cases where both pronouns are first or second person pronouns. Remember that Slovenian allows for the 1P>2P as well as the 2P>1P order. Stegovec (2017) derives this by assuming that if V has valued participant and valued speaker features, both arguments pronouns can move intoapositionc-commandingv (145b)bothagree(inturn)withv tofirstgetaparticipant feature and then either one of them can agree with a potential speaker feature onv. This view can generate the weak PCC. The mechanism needed for the weak PCC seem to be inconsistentwithStegovec’s(2017)viewofMaximizeAgree,whichisimportantforderiving the strong PCC and requires that if an agree relation is established between two heads, "all the unvalued features [...] must be valuated by any matching features on the other element". ItisunclearwhythisshouldnotapplyinthecaseoftheweakPCC.Interestingly enough,strictadherencetoMaximizeAgreewouldgenerateacaseindependentultra-strong PCCinthesenseofNevins(2007),whichmeansthattheprecedingelementneedstobeas prominent or more prominent than the following element. It is not clear how Stegovec’s (2017) account would be capable of capturing the me-first PCC or restrictions on the co- occurrenceoftwo3rdpersonobjectpronouns. Conclusion Stegovec(2017)proposesaninnovativeaccountofthePCCthatisabletocaptureordering restrictions on the PCC. His account would run into problems with the Swiss German databecausethereisnoclearwaythatitcancapturetherestrictiononthePCCaspartof theorderingrestriction. CHAPTER3. PREVIOUSLITERATURE 89 Table3.6: PersonHierarchyfromPancheva&Zubizarreta(2017)ex(2) 1P/2P > 3P 1P > 2P 3PProximate > 3PObviative 3.4 PanchevaandZubizarreta (2017) Pancheva & Zubizarreta (2017) connect the PCC to interpretation. They see it as being more than just a purely formal syntactic effect but driven by an interpretable feature. Descriptively they argue that the variation in the PCC can be described by the Person Hierarchy in table 3.6. Any analysis of the PCC should therefore be able to capture the PersonHierarchywithoutbeingmotivatedbyit. Pancheva&Zubizarreta(2017)maketwoassumptionswhichwehavealreadyseenin similar form in Walkow (2013) above. Pancheva & Zubizarreta (2017) assume that clitics present the establishment of an Agree relation. In contrast to Walkow (2013), they are agnosticaboutwhetherthecliticis anargumentorthespelloutofagreementitself. Pancheva & Zubizarreta (2017) also assume a feature specification similar to the one that we have seen in Walkow (2013) in table 3.5. The difference between the two is that Pancheva & Zubizarreta (2017) assume that 3rd person features sometimes are further specifiedforproximate. Theyalsoassumethatfeaturesarebinaryandnotprivative4. Table3.7: PersonFeaturesinPancheva&Zubizarreta(2017) 1P [+proximate] [+participant] [+author] 2P [+proximate] [+participant] [-author] 3Pproximate [+proximate] [-participant] [-author] 3Pobviative [-proximate] [-participant] [-author] Pancheva & Zubizarreta (2017) assume that the difference between proximate and obviative only plays a role in the presence of a 3P argument or if the difference between obviativeandproximateismorphologicallymarkedinthelanguage. Pancheva & Zubizarreta (2017) propose that the PCC is the result of the encoding of perspective by a person feature (p-feature). In PCC languages the p-feature is located on the applicative head and "mediates the mapping of a DP at the phase edge onto a logophoricentity"(Pancheva&Zubizarreta2017:9). 4SeeWatanabe(2013)foranargumentforbinaryandagainstprivativefeaturesbutthedifferencebetween privativeandbinaryfeatureswillnotmakeadifferenceintheimplementationofPancheva&Zubizarreta’s (2017)proposal. CHAPTER3. PREVIOUSLITERATURE 90 The second role of the person feature is being the host of the P-Constraint as stated in (146)TheP-Constrainthasfourcomponents,eachsubjecttopotentialparametricvariation. The Domain of application regulates the head carrying the p-feature. Differences in the domain of application can generate the difference between a direct/inverse (p-feature on vorInfl)systemandthePCC(p-featureonAppl)(cf.Zubizarreta&Pancheva2017). The otherconstraintsareP-Prominence,P-UniquenessandP-Primacy. Iexplainwhatthesethree latterconstraintsdobelow. (146) P-Constraintonphases headedbyaninterpretablep(erson)-feature a. The interpretable person feature is present on all heads of a certain functional category(default),unlessrestricted(Domainofapplication) b. There must be an n-valued D located at the edge of that enters into an agreementrelationwiththen-valuedinterpretablepersonfeatureonthehead of . n is [+proximate] (default) or restricted to [+participant] or [+author]. (P-Prominence) c. There can be at most one DP in eligible to agree with the interpretable P-featureontheheadof. (P-Uniqueness) d. IfthereismorethanoneDPthatcanagreewiththeinterpretablep-featureon theheadof,theDPmarked[+author]istheonethatagrees. (P-Primacy) (Pancheva&Zubizarreta2017:10) P-Prominence requires that the DP in spec ApplP is [+proximate]. This insures that the IO is a potential perspective center. The value for P-Prominence can also be set to [+participant] which would ensure that the IO is always 2nd or 1st person or [+author] whichensuresthattheIOisalways1stperson. P-Uniqueness Pancheva & Zubizarreta (2017) propose that the p-feature checks the features of both internal arguments. The P-Uniqueness constraint states that the person featuresonApplmustbedistinct. P-Primacy is conditional on P-Uniqueness. In case there are multiple potential can- didates, P-Primacy can resolve the conflict by setting the argument with the [+author] featureas theperspectivecenterimposingthe1P>2PinthePersonhierarchy. The P-Uniqueness constraint requires that that both the person feature of the IO as wellastheDOareconsidered. Pancheva&Zubizarreta(2017)suggestthatthisisderived locally by having two person features on Appl, one interpretable and valued [iP:n] and theotheruninterpretableandunvalued[uP:_]. [uP:_]probesintoitsc-commanddomain, agreeingwiththeDO,whilethe[iP:n]agreeswiththeIO.Thisisillustratedin(147)below. CHAPTER3. PREVIOUSLITERATURE 91 (147) featurecheckingonApplPPancheva&Zubizarreta(2017)ex.(13) ApplP VP DO [iP:q] v Appl [iP:n][uP:_] IO [iP:n] The P-Uniqueness constraint requires that the value the two P features (n and q in (147)) isdistinct. Thismeansthatifoneis[+proximate],theotherhastobe[-proximate]. InPancheva&Zubizarreta’s(2017)system,itisnotnecessaryfornontheApplheadto bethemostspecificpersonfeatureontheIO.TheinterpretablepersonfeatureonApplis independently valuated from the person feature on IO. This means that [iP:+participant] onAppliscompatiblewithafirstorsecondpersonpronouninIO.[uP:_]onApplonthe otherhandvaluatesviaagreewith DOandsharesallthefeaturesofDO. DerivingthePCCvariationinPancheva&Zubizarreta(2017) Pancheva & Zubizarreta (2017) have a way to derive the variation in the PCC discussed in the literature by manipulating the variables in the P-Constraint. They argue that the strong PCC is the most stable version of the PCC across languages and that this is due to the fact that it is the default setting of the P-Constraint. The P-Constraint as given in (148) derives the strong PCC, it also requires that that the DO is not only 3rd Person but [-proximate],apointwhichIwillturntoinsection3.5. (148) P-ConstraintStrongPCCPancheva&Zubizarreta(2017)ex: (15) a. TheinterpretablepersonfeatureispresentonallApplheads. (Domainof application) b. Theinterpretablep-featureonApplisvalued[+proximate] (P-Prominence) c. There can be at most one DP in eligible to agree with the interpretable p-featureontheheadof (P-Uniqueness) Pancheva & Zubizarreta (2017) derive the ultra-strong PCC by adding P-Primacy to the configuration of the strong PCC. This will allow for the co-occurrence of a 1st and 2nd person argument but only if the IO is first person. The configuration as illustrated here CHAPTER3. PREVIOUSLITERATURE 92 willgeneratetheultra-strongPCCthatwillallowfortheco-occurrenceoftwo3rdperson internalargument. (149) P-ConstraintUltra-strongPCCPancheva&Zubizarreta(2017)ex: (19) a. TheinterpretablepersonfeatureispresentonallApplheads. (Domainof application) b. Theinterpretablep-featureonApplisvalued[+proximate] (P-Prominence) c. There can be at most one DP in eligible to agree with the interpretable p-featureontheheadof (P-Uniqueness) d. IfthereismorethanoneDPthatcanagreewiththeinterpretablep-featureon theheadof,theDPmarked[+author]istheonethatagrees. (P-Primacy) I will demonstrate in chapter 4 that Swiss German has the ultra-strong variety that bans the co-occurrence of two 3rd person internal arguments. The Swiss German style ultra- strongPCCcanalsobederivedwiththeP-Constraintsetas(150). Theonlydifferenceisin P-Prominence. Theinterpretablep-featureonApplhastobevalued[+participant]which willban3rdpersonIOs. (150) P-ConstraintforSwissGermanstyleUltra-strongPCC a. TheinterpretablepersonfeatureispresentonallApplheads. (Domainof application) b. Theinterpretablep-featureonApplisvalued [+participant] (P-Prominence) c. There can be at most one DP in eligible to agree with the interpretable p-featureontheheadof (P-Uniqueness) d. IfthereismorethanoneDPthatcanagreewiththeinterpretablep-featureon theheadof,theDPmarked[+author]istheonethatagrees. (P-Primacy) The P-Constraint setting in (149) and (150) makes another interesting prediction. The setting in (149) bans the co-occurrence of [+proximate] DOs with 3rd and 2nd person IOs because they violate P-Uniqueness without triggering P-Primacy. The Swiss German P-Constraint setting in (150) on the other hand does not ban the co-occurrence of a 2nd personIO witha[+proximate]DO. The weak PCC is derived with the P-Constraint configuration in (151). The important aspect of the configuration is the absence of the P-Uniqueness constraint. This allows the co-occurrence of 1st and 2nd person arguments as well as two 3rd person arguments as longastheIOargumenthasthefeature[+proximate]. (151) P-ConstraintweakPCCPancheva&Zubizarreta(2017)ex:(22) CHAPTER3. PREVIOUSLITERATURE 93 a. TheinterpretablepersonfeatureispresentonallApplheads. (Domainof application) b. Theinterpretablep-featureonApplisvalued[+proximate] (P-Prominence) As it stands, the P-Constraint configuration in (151) does not ban a 3rd person IO from co-occurring with a 1st or 2nd person DO. Pancheva & Zubizarreta (2017) prevent such a system from occurring by the assumption that languages do not morphologically mark pronouns [proximate] in the presence of another 3rd person pronoun. This means that a language that has the PCC would only allow for a 3rd person IO object from co- occurring with a [+participant] DO if 3rd person pronouns are morphologically marked for[+proximate]. ThisisnotthecasefortheEuropeanlanguagesthathavebeendescribed ashavingtheweakPCC. Pancheva & Zubizarreta (2017) derive the Super-strong PCC by slightly modifying the P-Constraint for the strong PCC by setting the value for the p-feature on Appl as [+participant]. ThispreventstheIOfrombeing3rdPerson. (152) P-ConstraintSuper-strongPCCPancheva&Zubizarreta(2017)ex: (26) a. TheinterpretablepersonfeatureispresentonallApplheads. (Domainof application) b. Theinterpretablep-featureonApplisvalued [+participant] (P-Prominence) c. There can be at most one DP in eligible to agree with the interpretable p-featureontheheadof (P-Uniqueness) The last of the attested PCCs is the me-first PCC. Pancheva & Zubizarreta (2017) derive the me-first PCC by setting the domain of application only to Appl heads that have a [+author]featureandhavingtheinterpretablePfeaturesvaluesetto[+author]. (153) P-ConstraintMe-firstPCCPancheva&Zubizarreta(2017)ex: (28) a. TheinterpretablepersonfeatureispresentonlyonheadsofApplphraseswith atleastoneDPwitha[+author]feature. (Domainofapplication) b. Theinterpretablep-featureonApplisvalued[+author] (P-Prominence) c. There can be at most one DP in eligible to agree with the interpretable p-featureontheheadof (P-Uniqueness) Thissettingofthedomainofapplicationderivesthefactthattheme-firstPCConlyeffects combinations that have an argument that is [+author], and the setting of P-Prominence ensurestheinsuchacasethe1stpersonargumentistheIO. CHAPTER3. PREVIOUSLITERATURE 94 PredictionofUnattestedPCC-Configurations DifferentpossiblesettingsoftheP-Constraintcreate36(4*3*3)possiblegrammars. Going through all of them here would exceed the scope of this work. Some of these settings willgeneratethesameoutput,otherswillgenerateverysmalldifferencepertainingtothe availabilityoftwo[-proximate]internalarguments,sothenumberofavailablegrammars is de facto less than 36. The fact that some differences are not attested might be due to the factthatonewouldhavetoactivelylookfortheabsenceoftheminabigenoughsampleof speakers. Pancheva&Zubizarreta’s(2017)accountofthePCCalsopredictstheexistence of unattested PCC configurations that might be more noticeable but are predicted to be lesslikelysincetheywouldbehighlymarkedintheirsystem. Iwanttoillustratethiswith onesettingwhichPancheva&Zubizarreta(2017)call‘possiblegrammar3’or‘PG3’. (154) PG 3basedonPancheva&Zubizarreta(2017)ex:(33) a. Domainofapplication: AllApplheads b. P-Prominance: [iP:+author] c. P-Uniqueness: active PG 3 is a language where the IO always has to be 1st person. There are two reasons for why this is an unlikely setting. This setting is marked since P-Prominence does not have thedefaultsetting. TheotherreasonisthatalearnerthatspeaksanyotherPCClanguage willgetmanycounterexamplestosuchasettingintheprimarylinguisticdata. TheideaofamarkedP-Constraintsettingalsopredictsthatsomeinterspeakervariation ismorelikelythanother. Inthepresentaccount,theweakPCCandtheultra-strongPCC are similarly marked since both require one deviation from the unmarked strong PCC setting. Thedifferencebetweenthetwoisalsoinquitesmallintheprimarylinguisticdata sincetheyonlydifferinwhether2IO>1DOisacceptableornot. Pancheva&Zubizarreta’s (2017)accountpredictsthatthereisconsiderableinter-speakervariationbetweentheweak PCC and ultra-strong PCC in languages. This seems to be sported in the variation found inanumberofEuropeanlanguages. ProblemsaccountingfortheSwissGermanData Across languages, PCC restriction seems to be a restriction on Datives. We never see a reverse me-first PCC that requires that only the accusative argument can be first person. Pancheva&Zubizarreta(2017)ensuresthatthisobservationisnotjustaccidentalbytying the PCC to a restriction on the specifier of the ApplP. However, this also means that the PCC should not be affected by word order. In their analysis PCC is caused by a local CHAPTER3. PREVIOUSLITERATURE 95 restrictiononfeaturevaluationonApplandtakesplacewhenApplandtheIOobjectare merged. Any later reordering of arguments cannot have an effect on it. The reordering facts in Swiss German can therefore not be captured in the analysis of the PCC as it is in Pancheva&Zubizarreta(2017). 3.5 CliticLogophoricRestriction Ormazabal & Romero (2007) based on work by Roca (1992) observe that 3rd person accusative clitics in Spanish cannot be co-referent with particular matrix subjects in the presence of 3rd person dative clitics. This is demonstrated in (155) below. The accusative clitic lo can refer to the matrix subject in (155a) while it cannot do so in (155b). The important difference between the two is the presence of the dative clitic se in (155b), which blocks the relevant interpretation. Note that it is not the mere presence of a dativeargumentthatrestrictstheinterpretationoftheaccusativeclitic,(155a)hasadative argument,itistheexistenceofthedativecliticistherelevantfactor. (155) Spanish a. Mateo i Mate piensa think que that lo i 3acc entregaste handed.2sg.subj a to la the policà police ‘Mateothinksthatyouhandedhimovertothepolice.’ b. Mateo i Mate piensa think que that se 3dat lo i 3acc entregaste handed.2sg.subj a to la the policà police ‘Mateothinksthatyouhandedhimovertothepolice.’ (Ormazabal&Romero2007:327) Ormazabal & Romero (2007) note that the kind of matrix subject is relevant and that the effect in question only occurs with animate matrix subjects but not with inanimate ones. This is illustrated in (156) which is (32) in (Ormazabal & Romero 2007:328). Coreference with the inanimate matrix subject is possible irrespective of whether the dative clitic is present,asin(156b)ornot,like(156a). (156) Spanish a. El The paquete i package especifica specifies que that lo i 3acc entregues hand.subj en at la the porterà frontdoor ‘Thepackagespecifiesthatyoushouldhanditoveratthefrontdoor.’ b. El The paquete i package especifica specifies que that se 3dat lo i 3acc entregues hand.subj en at la the porterà frontdoor ‘Thepackagespecifiesthatyoushouldhanditoveratthefrontdoor.’ CHAPTER3. PREVIOUSLITERATURE 96 Charnavel&Mateu(2015)argue,contrarytoBhatt&Šimík(2009),thattherelevantfactor for this restriction is not c-command between the two co-referent elements but that the matrix subject can function as a logophoric center. This means that the matrix subject is the attitude holder in their view, which in turn entails that it has to be animate. (156) is therefore felicitous because a package, as an inanimate object, cannot be a logophoric center. (157) demonstrates that c-command is not a requirement for the restriction on coreference. prisionero ‘prisoner’ in the example below does not c-command the clitic lo, coreferenceisneverthelessimpossible. (157) *La the carta letter [del of.the prisionero] i prisoner explica explains que that se k 3.dat lo i 3.acc entregaron hand.pst.3pl [al of.the juez] k judge sin without pruebas. evidence ‘The prisoner’s letter explains that they handed him over to him, the judge, withoutevidence.’ Charnavel & Mateu (2015) capture the restriction under the label Clitic Logophoric Restric- tion (CLR)asdefinedin(158). (158) CliticLogophoricRestriction(CLR) When a third person dative clitic and an accusative clitic co-occur in a cluster, the accusativecliticcannotco-referwithalogophoriccenter. (Charnavel&Mateu 2015:11) They test the CLR in French and Spanish using full lexical datives as a baseline. Subjects wereaskedtojudgesentencesonlineonacontinuousscale. Theyconsideredaparticular condition to be ungrammatical iff the condition was judged significantly lower than the baseline. They report that the CLR holds in both languages, and that logophoricity and notbindingistherelevantfactor. Ormazabal&Romero(2007),Bhatt&Šimík(2009)aswellasCharnavel&Mateu(2015) directly connect the CLR to the PCC. However, the connection is a priori not very strong sinceallofthelanguagesthatthevariousauthorslookedatwerelanguageswiththePCC as well as the CLR. Pancheva & Zubizarreta (2017) argue that the CLR and the PCC are connected but not due to the same syntactic constraints. I will present the analysis of Charnavel & Mateu (2015) first, then introduce Pancheva & Zubizarreta’s (2017) ideas on theconnectionbetweenthePCCandtheCLR. 3.5.1 Charnavel&Mateu(2015) Charnavel & Mateu (2015) conducted an experimental study with speakers of French CHAPTER3. PREVIOUSLITERATURE 97 and Spanish to support their argument that the CLR is not a restriction on binding, but a restrictiononcoreferencewiththelogophoriccenterofanutterance. Thisisdemonstrated in the French examples in (159) which are Charnavel & Mateu’s (2015) (6a) and (23a) respectively. Thematrixsubjectin(159a)isthelogophoriccenter,andtheaccusativeclitic cannotrefertothislogophoriccenterinthepresenceofa3rdpersondativeclitic. In(159b) thematrixsubjectisinanimateandcannotbethelogophoriccenter,whichmeansthatthe accusative cliticcanrefertothematrixsubject. (159) French a. *Pierre i Peter dit says qu’ that on s.o le i acc.3sg lui k dat.3sg a has présenté, introdcued á to [la the Reine] k . Queen ‘Peter i saysthattheyintroducedhim i toher k -[theQueeen] k .’ b. [Le the paquet] i package spécifie indicates qu’ that il it faut must le i acc.3msg lui k dat.3sg remettre, give [au to.the concierge] k doorman ‘[The package] i specifies that they should hand it i over to him k - [the doorman] k Charnavel & Mateu (2015) point out that this anti-logophoricity effect can only occur in 3rd person clitic clusters in French and Spanish. This is demonstrated in (160) which is Charnavel&Mateu’s(2015)ex. (25a). (160) [La the petite little fille] i girl espère hopes qe’ that on s.o va will te dat.2sg la i acc.3fsg confier. entrust French ‘[Thelittlegirl] i hopesthattheywillentrusther i toyou. TheygiveasemanticmotivationfortheCLRandthePCCandarguethatbotharisefroma perspective conflict when two relevant logophoric elements co-occur. To them the dative argument encodes a point of view and have to be inherently logophoric. Introducing a logophoricaccusativeargumentcreatesagrammaticalconflict5 5 While Charnavel & Mateu (2015) have the experimental data from French and Spanish to back up their claim, Bhatt & Šimík (2009) consider crosslinguistic data, which seems to indicate that there may be some variation between languages, and that the restriction on coreference might not be as closely related to logophoricity in some languages, as Bhatt & Šimík’s (2009) example in (161) from Serbo-Croatian demon- strates. (161) Na On pacetu i package piše writes da that mu him:dat ga i=j it:acc treba needs poslati. send Serbo-Croatian ‘Thepackage i specifiesthatyoushould send it i=j to him.’ (Bhatt&Šimík2009:3) NotethatCharnavel&Mateu(2015)arguethatinanimateargumentsdonotfunctionaslogophoriccenters. They have given (159b) as an example in support of their proposal for the CLR. However, (161) seems to CHAPTER3. PREVIOUSLITERATURE 98 TheAnalysisinCharnavel&Mateu(2015) Charnavel&Mateu(2015)assumethattherearedifferenttypesoflogophoriccentersand notallinteractionsbetweenlogophoriccenterscreateaconflict. Logophoriccentersare: i) discourseparticipantswhicharespeakers(1stperson)orhearers(2ndperson), ii) empathy locus which is the event participant with whom the speaker identifies, such as3rdpersondativeclitics iii) attitudeholderswhichinvolvesindirectintegrationofperspectivesuchasa3rdperson accusativeclitic. These logophoric centers are ordered on a hierarchically as in (162). An first person is necessarily an empathy locus and a discourse participant, while a 3rd person dative is necessarilyaempathylocusandanattitudeholder. (162) discourseparticipant> empathylocus> attitudeholder (Charnavel&Mateu2015:20) Fortheconcreteimplementation,Charnavel&Mateu(2015)introducetheideaofalo- gophoric operator which carries uninterpretable features and must probe its c-command domain. Multiple elements carrying the same logophoric feature(s) give rise to an inter- ventioneffect. Therelevantfeaturesdistributionisgivenin(163). (163) a. Discourseparticipants: [A,B] b. Empathylocus: [B,C] c. Attitudeholder: [C] (Charnavel&Mateu2015:23) Charnavel&Mateu’s(2015)viewaccountsforthe(super-)strongPCCandtheCLReffects in a Anagnostopoulou (2003) or Walkow (2013) style framework. Charnavel & Mateu’s (2015) assume that there is at most one logophoric operator which probes its c-command domain. All the logophoric elements need to get their features checked by OP log , which ensures that only combinations are licensed that do not violate the PCC or the CLR as outlinedintable3.8. indicatethatSerbo-CroatianmightnotfollowthesamegeneralizationasFrenchandSpanishwhenitcomes totheCLR. Iwillnotresolvethedataquestionswhichpreviousliteratureonthetopichasraised. Whatisclearfrom Bhatt&Šimík(2009)aswellasCharnavel&Mateu(2015)isthatmanylanguageswhichhavethePCCalso showrestrictionsoncoreferencebetweenmatrix subjects and accusative clitic. CHAPTER3. PREVIOUSLITERATURE 99 Table3.8: DerivingthePCCandtheCLRinCharnavel&Mateu(2015) *OPlog [A;B;C] [la C lui [B;C] ] (CLR) *OPlog [A;B;C] [me [A;B] lui [B;C] ] (PCC) OPlog [A;B;C] [me [A;B] le [C] ] TheproblemwithCharnavel&Mateu’s(2015)accountisthatitonlycapturestheCLR and PCC for languages that have the strong PCC, but not for languages that have the weak,theultra-strongortheme-firstPCC. 3.5.2 TheCLRinPanchevaandZubizarreta(2017) Pancheva & Zubizarreta (2017) propose that the interpretable person feature on Appl encodes the fact that the IO clitic is a point of view center. This requirement is imposed with the P-Prominence constraint. They connect the point of view to the concept of pivot inSells(1987). Sells(1987)arguesthatthereisnounifiednotionoflogophoricitybutthat it consists of three primitive “roles”, the pivot, self and source. These concepts are defined in (164), which are from Sells (1987:455), the bracketed terms are terms sometimes used inPancheva&Zubizarreta(2017). (164) a. Thepivotrepresentstheonefromwhosepointofviewthereportismade(point ofview) b. The self representsthe onewhose"mind"isbeingreported(attitudeholder) c. The source istheonewhomakesthereport(speaker) (Sells1987:455) Pancheva & Zubizarreta (2017) point out that there is a mapping between the Sells’s (1987) roles and the Person features in that they impose a minimum requirement on interpretable P-features. If a pronoun is [+proximate] it has to be at least a pivot. If P is valued [+participant], it has to function at least as a self and if P is valued [+author] the referenthastobeatleastasource. Thisiscapturedintable3.9. Table3.9: [iP:n]requiresthatthereferentisapivot [+proximate] thereferentmustbeatleastapivot [+participant] thereferentmustbeatleastaself [+author] thereferentmust beatleastasource Logophoric roles might overlap. A speaker is a source by definition but can also be a CHAPTER3. PREVIOUSLITERATURE 100 self as an attitude holder, an agent marked [+proximate] can also be the attitude holder andthesource,e.g.thesubjectofsay. Pancheva & Zubizarreta (2017) propose that the CLR and the PCC are different phe- nomenathatshareacommoncore. ThePCCiscausedbyasyntacticprocess,whereasthe CLR is a semantic restriction, which requires compliance with the Point-of-view Principle in(165). (165) Point-of-viewPrinciple: Withinalogophoricdomainmarkingpointofview,ifthere areattitudeholdersamongtheeventparticipants,oneofthemhastobethepoint- of-viewcenter. (Pancheva&Zubizarreta2017:ex(48)) (165) requires that if there is an attitude holder, an attitude holder has to be the point- of-view center. The CLR is then the result of the interaction between the P-Prominence constraint and the Point-of-view Principle. The P-Prominence constraint assigns point of view to the dative argument in the syntax. If the dative argument is the speaker or the hearer, the Point of view Principle will always be fulfilled since the syntax marked an attitudeholderasthepointofviewwhichtcanbepoint-of-viewcenteraccordingto(165). A problem can only arise if the dative argument is not an attitude holder, possible only if it is 3rd person, and the accusative is an attitude holder. In such a case the Point-of-view Principle andtheP-ProminenceconstraintarenotreconcilablecausingtheCLR. Chapter4 SwissGermanPCC This chapter will discuss the PCC data in Swiss German and propose an analysis for it. I will introduce the relevant data in section 4.1, showing that Swiss German allows for reduced internal arguments to surface in acc>dat and dat>acc order, and we can see the ultra-strong PCC in the dat>acc order. Importantly the ultra-strong PCC in Swiss German also bans the co-occurrence of two 3rd person reduced pronouns. I will discuss the analysis that I propose for Swiss German in section 4.2 before turning to restrictions onbound pronounsinthePCCconfigurationinsection4.3. 4.1 Thefulldataset Previousliteraturedoesnottakethefullparadigmintoaccount. Anagnostopoulou(2003, 2015)orStegovec(2017)forexample,analyzethePersonCaserestrictioninSwissGerman as another case of strong or weak PCC and give it the same treatment as their analysis of the weak and the strong PCC. However, upon a closer look, this generalization does nothold. Infact,itturnsoutthatSwissGermanhasavariantoftheultra-strongPCC(cf. Nevins2007). 4.1.1 The“PCC”inSwissGerman ThefirstpieceofevidencetoconsideristhatthereisnoPCCwhentheaccusativeargument linearlyprecedesthedativeargumentastheexamplesin(166)demonstrate. Thisiswhat Bonet (1991) already noted. Note that the combination of two 3rd person weak internal pronounsin(166g)isgrammaticalaswell. 101 CHAPTER4. SWISSGERMANPCC 102 (166) a. D’Maria theMaria hät has mi me der 2sg.dat zeigt. shown 1> 2 ‘Mariahasshownmetoyou’ b. D’Maria theMaria hät has mi me em 3m.dat zeigt. shown 1> 3 c. D’Maria theMaria hät has di you em 3m.dat zeigt. shown 2> 3 d. D’Maria theMaria hät has di you mer 1sg.dat zeigt. shown 2> 1 e. D’Maria theMaria hät has en him mer 1sg.dat zeigt. shown 3> 1 f. D’Maria theMaria hät has en him der 1sg.dat zeigt. shown 3> 2 g. D’Maria theMaria hät has en him em 3m.dat zeigt. shown 3> 3 However,ifwelookatthecaseswherethedativeargumentsprecedetheaccusativeargu- mentsin(167)weseethatcertaincombinationsofreducedpronounsareungrammatical. (167a)to(167c)areexampleswherethedativeargumentismorelocalthantheaccusative argumentalongthedefinitionin(4). However,alltheexampleswherethedativeargument isnotmorelocalthantheaccusativeargument,i.e. (167d)to(167g),areungrammatical. (167) a. D’Maria theMaria hät has mer 1sg.dat di you zeigt. shown 1> 2 ‘Mariahasshownyoutome’ b. D’Maria theMaria hät has mer 1sg.dat en him zeigt. shown 1> 3 c. D’Maria theMaria hät has der 2sg.dat en him zeigt. shown 2> 3 d. *D’Maria theMaria hät has der 2sg.dat mi me zeigt. shown 2> 1 e. *D’Maria theMaria hät has em 3m.dat mi me zeigt. shown 3> 1 f. *D’Maria theMaria hät has em 3m.dat di you zeigt. shown 3> 2 g. *D’Maria theMaria hät has em 3m.dat en him zeigt. shown 3> 3 Atthispoint,acommentonthebanon3rd-3rdin(167g)isinorder. Ihavementionedin section1.1abovethatNevins(2007)doesnotconsiderbanson3rdpersondativearguments with 3rdperson accusativearguments tobe partof thePCC. The questionis whetherthe CHAPTER4. SWISSGERMANPCC 103 ungrammaticalityof(167g)arisesfromthesamecausesastheungrammaticalityin(167d) to (167f)? Furthermore, are there any good arguments for separating the constraint on two3rdpersoninternalargumentfromtheotherrestrictions,orissuchamovepurelyad hoc? I will show in the next section that the data indicates that the ban on 3rd>3rd is in factpartofthesamePCCparadigmandthatNevins’s(2007)argumentsforexcludingthe banonco-occurrenceoftwo3rdpersonpronounsdoesnotholdinSwissGerman. 4.1.2 Onthebanof3DAT>3ACC Cooper (1999) hypothesizes that the restriction exemplified by (167g) might be a phono- logical restriction. However, this hypothesis appears incorrect. Note that the restriction is unaffected by the phonological realization of the reduced pronouns. Replacing the masculine with the feminine counterpart in (168) does not make the sentence any more grammatical, although the feminine pronouns ere ‘her’ (dat) and si ‘her’ (acc) are phono- logicallyquitedifferentfromthemasculinepronounsem ‘him’(dat)anden ‘him’(acc). (168) a. *D’Maria theMaria hät has ere 3f.dat (e)n him.acc zeigt. shown 3> 3 b. *D’Maria theMaria hät has em 3m.dat si her.acc zeigt. shown 3> 3 c. *D’Maria theMaria hät has ere 3f.dat si her.acc zeigt. shown 3> 3 If the accusative w-pronoun precedes the dative argument, the sentences in (168) are grammaticalasisillustratedin(169) (169) a. D’Maria theMaria hät has (e)n him.acc ere 3f.dat zeigt. shown 3> 3 b. D’Maria theMaria hät has si her.acc em 3m.dat zeigt. shown 3> 3 c. D’Maria theMaria hät has si her.acc ere 3f.dat zeigt. shown 3> 3 Variations in number also do not save the construction. (170) demonstrates that plural pronounswillnotmaketheconstructiongrammatical. (170) a. *D’Maria theMaria hät has ene 3pl.dat si them/her zeigt. shown 3> 3 b. *D’Maria theMaria hät has ene 3pl.dat en him zeigt. shown 3> 3 CHAPTER4. SWISSGERMANPCC 104 Again, the sentences are grammatical if the accusative w-pronoun precedes the dative pronoun(171). (171) a. D’Maria theMaria hät has si them/her.acc ene 3pl.dat zeigt. shown 3> 3 b. D’Maria theMaria hät has en him.acc ene 3pl.dat zeigt. shown 3> 3 Astheexamplesin(168)and(170)demonstrate,therestrictionisona3rdpersonaccusative reduced pronoun, preceding a 3rd person dative reduced pronoun. The restriction is independentofthephonologicalrealization,thegenderorthenumberofthepronounin question. Thismeansthatwearenotdealingwithaphonologicalrestriction. Nevins (2007) argues that the le lo constraint in Spanish is due to a morphological restriction. I will show in the following that his argument for Spanish does not apply for SwissGerman. Iwillpresentdatashowingthattherestrictionontwo3rdpersoninternal pronouns must be of the same nature as the other PCC violations. Therefore, an account ofthePCCinSwissGermanhastoaccountfortherestrictionontheco-occurrenceoftwo 3rdpersonpronouns. Nevins (2007) gives a thorough discussion of the le lo restriction in Spanish. The restriction is demonstrated in (172) below. Spanish bans a 3rd person dative clitic from preceding a 3rd person accusative clitic as is demonstrated in (172a). The repair for such a violation can be seen in (172b). Importantly, the repair is not a regular PCC repair in Spanish. Spanish PCC violations are otherwise repaired by using a strong pronoun (cf. Bonet2008) (172) Spanish a. *A to Pedro, Pedro, le the premio, prize, le 3dat lo 3acc dieron gave-pl ayer. yesterday. ‘ToPedro,theprize, theygaveittohimyesterday. b. A to Pedro, Pedro, le the premio, prize, se se lo 3acc dieron gave-pl ayer. yesterday. ‘ToPedro,theprize,theygaveittohimyesterday. Nevins (2007) argues that the origins of the constraint on combinations of two 3rd person internal clitics is not a PCC phenomenon but the same as the restriction on two firstortwosecondpersoncliticswhicharealsoungrammaticalinSpanish. However,forSwissGermantherestrictionseemstobefundamentallydifferent. Also, in Swiss German it is true that combinations of two first or second person internal argu- mentsareungrammatical,asisdemonstratedin(173). CHAPTER4. SWISSGERMANPCC 105 (173) a. *De the Peter Peter hät has mer 1sg.dat mi me zuateilt. assigned dat>acc b. *De the Peter Peter hät has der 2sg.dat di you zuateilt. assigned dat>acc There is a fundamental difference between the cases of two first or second person arguments in (173) and the case of two 3rd person internal arguments. While having two 3rd person internal arguments is grammatical if the accusative argument precedes the dative argument, having two first or two second person internal arguments is always ungrammatical irrespective of the order of the two arguments. This is shown in (174) whichdemonstratesthathavingtwo3rdpersoninternalarguments(174a)ortwosecond personinternalarguments(174b)isungrammatical,alsointheacc>datorderofreduced pronouns. (174) a. *De the Peter Peter hät has mi me mer 1sg.dat zuateilt. assigned acc>dat b. *De the Peter Peter hät has di you der 2sg.dat zuateilt. assigned acc>dat Thisisnotthecasefortwo3rdpersonarguments. Whilethedat–accorderisungram- matical, as is demonstrated in (175b), repeated from (167g), the reversed order in (175a), repeated from (166g), is grammatical. In contrast to the examples containing two first or twosecondpersonreducedpronouns,two3rdpersonpronounsaresensitivetothesame orderingrestrictionsastherestoftheultra-strongPCCinSwissGerman. (175) a. D’Maria theMaria hät has en him em 3m.dat zeigt. shown acc>dat b. *D’Maria theMaria hät has em 3m.dat en him zeigt. shown dat>acc There is therefore no reason to assume that the ban of 3rd dat before 3rd acc is not partofthePCCinSwissGerman1. 1It is also interesting to note that other German dialects do not have the same restrictions on the co- occurrenceoftwo3rdpersonpronounsasonecanseeforViennese,aBavariandialectbelowwherewesee that both orders are fine. The co-occurrence of two first person pronouns is therefore clearly not a problem crosslinguistically. (176) Viennese(BavarianGerman) a. ea he hot has eam-an 3m.dat ge:m him (DAT gave ACC) ‘Hegavehimtohim b. ea he hot has -n him eam 3m.dat ge:m gave (ACCDAT) (Abraham 1996:4) CHAPTER4. SWISSGERMANPCC 106 4.1.3 LocatingthestructuraloriginsofthePCC The PCC is not contingent on reduced pronouns, which appear in the Wackernagel posi- tion. The PCC also arises when reduced pronouns are realized within vP. We can see in (177) and (178) that all of the hallmark patterns of the PCC in Swiss German are present. Theacc>datorderisalwaysgrammatical,asisillustratedin(178a)and(177a),whilethe dat>acc orderissensitivetothePCC,asshownby(178b)and(177b). (177) a. Waisch know.you eh, MP, dass that warschinli probably [ vP a a Frau woman mi me em 3m.dat vorgstellt] introduced hät. has ‘Youknow,itwasprobablyawomanthatintroducedmetohim’ b. *Waisch know.you eh, MP, dass that warschinli probably [ vP a a Frau woman em 3m.dat mi me vorgstellt] introduced hät. has (178) a. Waisch know.you eh, MP, dass that warschinli probably [ vP a a Frau woman en him mer 1sg.dat vorgstelt] introduced hät. has ‘Youknow,itwasprobablyawomanthatintroducedmetohim’ b. Waisch know.you eh, MP, dass that warschinli probably [ vP a a Frau woman mer 1sg.dat en him vorgstelt] introduced hät. has Wehaveseeninsection2.4.3abovethattheweakpronounsdonothavetoappearadjacent to each other. The PCC is not affected by the surface position of the weak pronouns. We seethePCCdistributioninthedat>accorderalsoifthew-pronounsdonotoccuradjacent to each other as is illustrated in (179). A first person dative pronoun can precede an 3rd person accusative pronoun in (179a) but a 3rd person dative pronoun cannot precede a 1st person accusative pronoun (179b). (179c) demonstrates that the same is true for 3rd persondativepronounspreceding3rdpersonaccusativepronouns. (179) a. Waisch know.you eh, MP, dass that mer me.dat warschinli probably [ vP a a Frau woman en 3m.acc vorgstellt] introduced hät. has ‘Youknow,itwasprobablyawomanthatintroducedhimtome’ b. *Waisch know.you eh, MP, dass that em 3m.dat warschinli probably [ vP a a Frau woman mi me vorgstellt] introduced hät. has ‘Youknow,itwasprobablyawomanthatintroducedmetohim’ c. *Waisch know.you eh, MP, dass that em 3m.dat warschinli probably [ vP a a Frau woman en him.acc vorgstellt] introduced hät. has ‘Youknow,itwasprobablyawomanthatintroducedhimtohim’ The examples in (179) demonstrate that the PCC in Swiss German cannot be the result of aphonologicalorotherrestrictionthatrequiresapplylocallyonsurfacestructure. CHAPTER4. SWISSGERMANPCC 107 4.2 TheAnalysis Ihavearguedinsection2.4.2thattheaccusativereducedpronounhastomovetoaposition outside of the applicative phrase. I have labeled this head AgrO and argued that it is associatedwithaccusativecaseassignmentandchecksvaluedbutuninterpretablefeatures on the weak pronoun. I have demonstrated above that reduced nominative pronouns do have to move to spec TP in Swiss German where they can get their uninterpretable features checked. Only strong subject pronouns can remain in a position lower than T. Since T is associated with nominative case assignment, it is reasonable to assume that the reduced subject pronoun has to move to the case assigning head where it its uninterpretable features can be interpreted. Movement of reduced accusative pronouns to the specifier of their case assigning head aligns the movement of the DO with the movement of reduced nominative pronouns. Since dative arguments are case licensed in their base position, reduced dative pronouns do not have to move. I assume that Swiss German has interpretable but unvalued features which means that the features uninterpretable features on the weak IO pronoun can be interpreted in spec ApplP and are licensed. It is important here that the reduced accusative pronoun is licensed by moving to AgrO, i.e. that no further agreement is needed to license weak pronouns and that that weak dative pronoun is interpretable in Swiss German by virtue of being base generated in the specifier of a head carrying interpretable but unvalued features. This is illustratedin(180) (180) DOmovement vP AgrOP AgrOP ApplP ApplP VP V acc Appl dat AgrO acc v CHAPTER4. SWISSGERMANPCC 108 There is no reason to assume that the transitive structure looks much different from the ditransitivestructurein(180). Theonlydifferencewouldbetheabsenceoftheapplicative phrase. However,itwouldbeimpossibletoseethemovementoftheaccusativeargument sinceAgrOdoesnothaveaphonologicalrealization. This means that the structure in (180), and the fact that both reduced object pronouns arelicensedintheirrespectivepositions,showwhytheacc>datorderdoesnothaveany personrestrictions 2. SohowdoesthePCCforthedat>accorderarise? Iwilldemonstrate thatthepersonrestrictionscanbeanalyzedasaresultofamovementoperation. Ipropose thatSwissGermanhasaPersonhead(Per)situatedbetweenvandAgrO.Thisheadcomes with unvalued author and participant features (181). These features will now probe in theirc-commanddomainforanappropriategoalandfinditeitherwiththeIOortheDO. (181) LocationofthePerhead vP PerP AgrOP AgrOP ApplP ApplP VP V acc Appl dat AgrO acc part[] author[] v Some remarks about the mechanism used here are in order. I propose that this head is obligatory in Swiss German and will derive the optionality of movement by proposing an optional pro. The optionality could also be derived by having an optional Per head or optionalovertmovement(cf.Stegovec2017) 2I speculate that the dative argument is not an intervener for the probe on AgrO for the same reason that the dative arguments do not seem to intervene for nominative case assignment in languages with odd subjects,e.g.Icelandic. Ihavearguedmoreextensivelyforthisinchapter2.4.3thatdativeargumentsarein factnotintervenerforcaseassignmentand agreement CHAPTER4. SWISSGERMANPCC 109 Iwillassumeinthefollowingthatthecheckingofunvaluedfeatureshastobeattempted in syntax, but a failure to check these features will not lead the derivation to crash (cf. Preminger2011) 3 ThelastassumptionthatIwillmakeisthattheprobesearchesforthegoalthatsatisfies themostfeatures. Thisisbasedonaconditionfrom(VanUrk&Richards2015:132),which theycallMultitasking,definedin(182). (182) Multitasking: Ateverystepinaderivation,ifaprobecantriggertwooperationsAandB,andthe features checked by A are a superset of those checked by B, the grammar prefers A. Thisassumptionisnotacrucialone. ItseemstomethatAgreeisafallibleoperationwhich will,atpoints,forceusintosomesortofMultitaskingassumption. However,onewouldbe able to reach the same results via multiple agreement operation and ordering them that authorandthenparticipanthastobeprobeforfirst. I chose these assumptions and mechanics over alternatives for two reasons. Many of thesemechanicshavefoundbroaderuseintheliterature,anditsimplifiesthederivations by reducing the number of lexical variations and the number of agree relations that are being established. The derivations and trees will therefore be somewhat simpler. Armed with these assumptions we can generate the grammatical and ungrammatical sentences underconsiderationandcapturethevariationthatispossibleinSwissGerman. Remember that the generalization is as follows: dative reduced pronouns can only precede reduced accusative pronouns if the dative argument is higher on the person hierarchy than the accusative argument. That means a 2 nd person dative argument can precedea3 rd personaccusativeargument,butnota1 st personaccusativeargument. This follows from the structure proposed here. The undervalued features in Per probe within theirc-commanddomainforappropriategoals. Iftheaccusativeobjectdoesnothaveany relevantfeatures,orifitisonlyabletovalueasubsetofthefeatures,theprobewillsearch further for a “better” goal. If it finds a more appropriate goal in the dative object, the Per head will agree with the dative object. The argument that the Per head agrees with can now move to the specifier of the Per head. If the Per head agrees with the accusative reducedpronoun,thewordorderwillremainthesameandthemovementoperationwill remain without detectable consequences. The operation will only be detectable if the reduced dative pronoun is more local than the reduced accusative pronoun. This two 3Thisisnotacrucialassumption,sincethesamecouldbeachievedviahomophonesinthelexicongoverned by a conditional enumeration rule ensuring to select the Per head which has only and all the features valuedontheweakpronounsselected. CHAPTER4. SWISSGERMANPCC 110 steps are illustrated in (183). If the dative is more local the dative will agree with PerP andmovetochecktheEPPfeature(183a),iftheaccusativeargumentismorelocal,itwill agreewithPerPandmovetochecktheEPPfeature(183b). (183) a. AgreelicensesmovementofIO(DAT>ACC PerP PerP AgrOP AgrOP ApplP ApplP VP V acc Appl dat AgrO acc part[] author[] EPP[ ] dat agree move b. AgreelicensesmovementofDO PerP PerP AgrOP AgrOP ApplP ApplP VP V acc Appl dat AgrO acc part[] author[] EPP[] acc agree PerPhastobeseparateandlowerthanvP.Ifthepersonfeatureswerelocatedinthevhead wewouldnotexpectmovementtooccurinthepresentsofanexternalargumentbecause theexternalargumentcanchecktheEPPfeaturebyvirtueofbeingmergedthere4. IfPerP 4This assumption is based on my use of the EPP and of merge over move (cf. Hornstein et al. (2005)). If CHAPTER4. SWISSGERMANPCC 111 is merged after v we would expect that a first-person subject would be an appropriate intervener, which prevents the movement of dative objects. This is not the case as (185) demonstrates. Bothordersarepossibleinthiscase 5 . (185) a. D’ The Susi Susi wais, knows dass that I I -n him der 2sg.dat vorgstellt intrduced ha. have. b. D’ The Susi Susi wais, knows dass that I I der 2sg.dat -n him vorgstellt intrduced ha. have. ‘SusiknowsthatIhaveintroducedhimtoyou.’ (186) a. Gestert yesterday han have -I -I der 2sg.dat -n him vorgstellt. introduced b. Gestert yesterday han have -I -I -n -him der 2sg.dat vorgstellt. introduced ‘Ihaveintroducedhimtoyouyesterday. 4.2.1 DerivingtheOptionality IhavestatedintheprevioussectionthatIassumethatPerisobligatory. Thiswouldpredict that reordering must be obligatory if possible. This is not the case. Theacc>dat order is always grammatical. I propose here, that this is possible by introducing a silent pronoun movement is triggered by strong features, there would be no problem with v being the head which carries thefeatures 5Stegovec (2017) cites (184a) from Winter (2001) which seems to indicate that the acc>dat order might sometimes be band in Siwss German and that this is reducible to the PCC. Stegovec (2017) assumes that the Swiss German PCC might involve obligatory reordering for some speakers creating (184a) just like in Slovenian, while other speakers have freer reordering. Subjects that allow for optional reordering require reordering in (184a). The reordering requirement seems however to be quite restricted and does not exist if the second person subject is either doped or a strong pronoun (184b) or if it is 1st person instead of 2nd (184c). The reduced second person pronoun is also bad if you use bring as a transitive verb (184d). It seems that this apparent exaction to the optionality of reordering is in fact not due to a syntactic but to a phonologicalrestriction. (184) a. das that d’ you.nom (*en)mer me.dat (en) him.acc halt MP morn tomorrow bringsch. bring.2sg ‘thatyouwilljusthavetogiveit to me tomorrow.’ b. das that du/; you.nom (en)mer me.dat (en) him.acc halt MP morn tomorrow bringsch. bring.2sg ‘thatyouwilljusthavetogiveit to me tomorrow.’ c. das that I I.nom (en)der you.dat (en) him.acc halt MP morn tomorrow bring. bring.2sg ‘thatyouwilljusthavetogiveit to me tomorrow.’ d. das that (*d’) you.nom en him.acc halt MP morn tomorrow bringsch. bring.2sg ‘thatyouwilljusthavetogiveit to me tomorrow.’ CHAPTER4. SWISSGERMANPCC 112 that does not contain interpretable or valued features and probes its c-command domain for an appropriator head to valuate. The closest head it will find with valued features is than the DO in spec AgrOP. This allows an DO in Swiss German to be associated with Per even though the DO is not the more local pronoun. Note that this process is not available to IO in Swiss German. The merge over move preference will ensure that if, this silentpronounisintheenumeration,reorderingwillnottakeplace. The question arises, what the purpose of Per is and why speakers would ever choose not to reorder, or in other words why they would ever chose this silent dummy pronoun topreventreordering. 4.3 ThepurposeofPer I have introduced Charnavel & Mateu’s (2015) Clitic Logophoric Restriction CLR in section 3.5 above and mentioned that they make a direct connection between the CLR and the PCC.Iwilldemonstrateinsection5.3.3thatthereisaCLRinStandardGermanandthat, counter to what the name would indicate, is not restricted to clitics but is also pertains to weak pronouns. I will furthermore illustrate that Standard German is another language wheretheCLRandthePCCgohandinhand. Iwilldiscussheredatathatpointstoward Pancheva & Zubizarreta (2017) view on the CLR, i.e. that the CLR and the PCC are connected butnotreducibletoasinglephenomenon. Remember that Pancheva & Zubizarreta (2017) argue that the PCC is triggered by an interpretablepersonfeatureandthatthispersonfeatureconnectstologophoricityrollsvia table 3.9 repeated in table 4.1 below. If a head has an interpretable [+proximate] feature, thespecifierofthatheadhastointerpretedasapivot/pointofview. Table4.1: [iP:n]requiresthatthereferentisapivot [+proximate] thereferentmust beatleastapivot [+participant] thereferentmustbeatleastaself [+author] thereferentmust beatleastasource In my analysis of Swiss German Per has an interpretable [participant] and [speaker] feature which means that the argument in spec PerP will be interpreted as the attitude holderandthesource. There is one configuration where Spec PerP might remain empty, that is if there is no potential internal argument to fill it. In this case I do not predict any influence on the CHAPTER4. SWISSGERMANPCC 113 interpretationofthearguments. Thisisthecasewhenbothargumentsare3rdperson. In allothercasesSpecPerPwillbefilledbybyoneoftwoways: i) viaagreeandmove-themore localargumentwillbepromotedtospecPerP ii) via feature sharing - feature sharing between empty pronoun in spe PerP and struc- turallyhigherargument. IfspecPerPisfilled,weexpecttohavesomelogophoriceffectsifPancheva&Zubizarreta (2017)isontherighttrack. Iwillshowinthefollowing,thatthisseemstobetrue. 4.3.1 TheSwissGermanData The contrast that I am going to present here tend to be on the level of preferences. Some preferences are clearer and shared by all the subjects and other are less clear. I will mark the degree of preference by using ?S ?? will mark a strong preference while ? will mark a small preference. Note that this is different to the PCC violations where subjects have clearjudgments. Swiss German does not have the classical CLR effect which is demonstrates in (187). Coreference between the matrix subject and the reduced accusative pronoun is possible intheacc>datorder. (187) De the Thomas i Thomas wais knows dass that I I -n i him ere 3f.dat vorstelle intrdocue werd. will ‘Thomas i knowsthatIwillintroducehim i toher’ SinceSwissGermandoesnotallowtwothirdpersonpronounstoco-occurinthedat>acc order,wecannotseelogophoricityeffectswithtwo3rdpersonpronouns. Thisisexpected since this it the one case where Spec PerP does not have to be filled. The much more interesting case is when the dative argument is more local than the accusative argument since the word order is more flexible in that instance. This case is demonstrated in (188). Note that coreference with the matrix subject is possible in the acc>dat order in (188a), butstronglydispreferredinthedat>accorderin(188b). (188) a. De the Thomas i Thomas sait, says, dass that -n i=j him d’ 2sg.dat Maria the der Maria vorgstellt introduced hät has . ‘Thomassays,thatMariahasintroducedhim i=j toyou’ b. De the Thomas i Thomas sait, says, dass that der the en?? i=j Maria d’ 2sg.dat Maria him vorgstellt introduced hät. has CHAPTER4. SWISSGERMANPCC 114 It is crucial in (188) that Thomas is an attitude holder and the source. As in Spanish andFrench,thecontrastdisappearswhenthematrixsubjectisnotanattitudeholderand cannot be a logophoric center. This is demonstrated in (189) where the criminals are not attitudeholders. (189) a. De the fabrecher i criminal isch is abkaut escaped befor before de the Max Max en i=j him der 2sg.dat zeige shown hät has könne. can b. De the fabrecher i criminal isch is abkaut escaped befor before de the Max Max der 2sg.dat en i=j him zeige shown hät has könne. can ‘ThecriminalescapedbeforeMaxcouldshowhimtoyou.’ The table in (190) can be coreferenced with a reduced pronoun independent of the order ofpronouns. (190) a. Uf On em the.dat tisch i package stoht, writes dass that de the Peter Peter n i=j it der 2sg.dat schicke sent soet. should b. Uf On em the.dat tisch i package stoht, writes dass that de the Peter Peter der 2sg.dat n i=j it schicke send soet. should ‘ItsaysonthepackagethatPetershouldsendittoyou.’ Whilethedataaboveisbasedonalimitednumberofparticipants,thedataseemtopoint towards anti-logophoricity in Swiss German, as described by Charnavel & Mateu (2015) forSpanishandFrench. Weseetheseanti-logophoricityeffectsinthesamesituation that we see the PCC in Swiss German, i.e. in the dat>acc word order. The major difference is that the Swiss German CLR occurs with 2nd and 1st person dative reduced pronouns. Thecaseof3rdpersondativereducedpronounsisruledoutbythePCCandcantherefore notbetested. Informants in general prefer sentences where 1st person datives linearly precede 2nd personaccusativesasin(191). (191) a. De the Peter Peter hät has mer me.dat di you.acc scho already gestert yesterday vorgstellt. introduced b. ?? De the Peter Peter hät has di you.acc mer me.dat scho already gestert yesterday vorgstellt. introduced ‘Peterhasintroducedyoutomealreadyyesterday.’ However this clear preference goes away, if the 2nd person can be interpreted as the source. (192). (192) a. Waish you.know eh, MP, dass that sie she ( ? di) you.acc mer me.dat (di)scho already gestert yesterday zeigt shown hät? has ‘Areyouaware,thatshehasalreadyshownyoutomeyesterday?’ CHAPTER4. SWISSGERMANPCC 115 b. Du you wish know eh, MP, dass that sie she (di) you.acc mer me.dat ( ? di)scho already gestert yesterday zeigt shown hät. has ‘Youknow,thatshehasshowedyoutomealreadyyesterday.’ Speakers of Swiss German seem to have a general preference for thedat>acc order if possible. It is interesting that this preference is reversed if the accusative argument is the source and not the dative argument, which is the case if the dative argument is only the empathy locus (second person) and no source. Speaker differ in their order preference if thedativepronounis1stpersonandtheaccusativeargumentisinterpretedasalogophoric center. Notethatthisisthecasewherebothargumentsarepotentiallogophoriccenters. 4.4 Conclusion IproposeananalysisfortheSwissGermanPCCthatisinfluencedbyWalkow(2013)and Pancheva&Zubizarreta(2017). IadoptWalkow’s(2013)ideathatthePCCcanbeviewed asanaccusativeargumentinterveningbetweenahigherprobeandadativeargumentthat needstoagreewiththisprobe. ItakefromPancheva&Zubizarreta(2017)theideathatthis probe is an interpretable P-feature that can have different specifications (P-Prominence) and that the P-feature sets interpretative requirements for the argument in its specifier position. However, I separate the interpretable P-feature from the Appl head in order to account for the influence of argument order on the PCC in Swiss German. Adopting the idea of an interpretable [author] on Per makes interesting and correct predictions about theinterpretationoftheargumentinSpecPer. IwillturnnexttothePCCinStandardGerman,whichcanbeclassifiedasaninstanti- ationoftheweakPCC. Chapter5 The PCCinStandardGerman The question of the PCC in Standard German has been raised in the literature. Haspel- math (2004) argues that Standard German does not have PCC restrictions while Anag- nostopoulou (2008, 2015) provides data that suggests that Standard German is a weak PCC language. I will present data in this chapter that shows that Standard German has the PCC and the CLR before turning to analyzing the weak PCC in Standard German alongthelinesoftheanalysisproposedforSwissGerman. Iwillgiveashortrecapofthe relevant information on the CLR even though I have introduced it in chapter 3, focusing onhowCharnavel&Mateu(2015)collectedtheirdatainsection5.1. Iwillalsointroduce Anagnostopoulou’s(2008)StandardGermandatawhichwillfunctionasthebasisforthe hypothesis tested here. I will present the results from two experiments here, one dealing withthePCCinStandardGermanandonedealingwiththeCLR. 5.1 TheCLR Ormazabal & Romero (2007), based on work by Roca (1992), observe that 3rd person accusative clitics in Spanish cannot be coreferent with particular matrix subjects in the presentsofa3rdpersondativeclitic. Thisisdemonstratedin(193)below. Theaccusative clitic lo can refer to the matrix subject in (193a) while it cannot do so in (193b). The important difference between the two is the presence of the dative clitic se in (193b), whosepresenceblockstherelevantinterpretation. Notethatitisnotthemerepresenceof a dative argument that restricts the interpretation of an accusative clitic, since (193a) has adativeargument,theexistenceof thedativecliticistherelevantfactor. 116 CHAPTER5. THEPCCINSTANDARDGERMAN 117 (193) Spanish a. Mateo i Mate piensa think que that lo i 3acc entregaste handed.2sg.subj a to la the policà police ‘Mateothinksthatyouhandedhimovertothepolice’ b. Mateo i Mate piensa think que that se 3dat lo i 3acc entregaste handed.2sg.subj a to la the policà police ‘Mateothinksthatyouhandedhimovertothepolice’ (Ormazabal&Romero2007:327) Ormazabal & Romero (2007) note that the kind of matrix subject is relevant and that the effect in question only occurs with animate matrix subjects but not with inanimate ones. This is illustrated in (194) which is (32) in (Ormazabal & Romero 2007:328). Note coreferencewiththeinanimatematrixsubjectispossibleirrespectiveofwhetherthedative cliticispresent(194b)ornot(194a). (194) Spanish a. El The paquete i package especifica specifies que that lo i 3acc entregues hand.subj en at la the porterà frontdoor ‘Thepackagespecifiesthatyoushouldhanditoveratthefrontdoor.’ b. El The paquete i package especifica specifies que that se 3dat lo i 3acc entregues hand.subj en at la the porterà frontdoor ‘Thepackagespecifiesthatyoushouldhanditoveratthefrontdoor.’ Charnavel&Mateu(2015)argue,contrarytoBhatt&Šimík(2009),thattherelevantfactor forthisrestrictionisnotc-commandbetweenthetwocoreferentelements,butthatmatrix subject can function as a logophoric center. This means in their view that the matrix subject has to be the attitude holder, which means it has to be at least animate. (194) is therefore felicitous because a package as an inanimate object cannot be a logophoric center. (195) demonstrates that c-command is not a requirement for the restriction on coreference. prisionero ‘prisoner’ in the example below does not c-command the clitic lo, coreferenceisneverthelessimpossible. (195) *La the carta letter [del of.the prisionero] i prisoner explica explains que that se k 3.dat lo i 3.acc entregaron hand.pst.3pl [al of.the juez] k judge sin without pruebas. evidence ‘The prisoner’s letter explains that they handed him over to him, the judge, withoutevidence.’ Charnavel & Mateu (2015) capture the restriction under the label Clitic Logophoric Restric- tion (CLR)asdefinedin(196). CHAPTER5. THEPCCINSTANDARDGERMAN 118 (196) CliticLogophoricRestriction(CLR) When a third person dative clitic and an accusative clitic co-occur in a cluster, the accusativecliticcannotcoreferwithalogophoriccenter. (Charnavel&Mateu 2015:11) They test the CLR in French and Spanish using full lexical datives as a baseline. Subjects wereaskedtojudgesentencesonlineonacontinuousscale. Theyconsideredaparticular condition to be ungrammatical iff the condition was judged significantly lower than the baseline. They report that the CLR holds in both languages, and that logophoricity not binding is the relevant factor. I will also consider ungrammaticality to be marked by a significantdropinacceptability,similartoCharnavel&Mateu(2015). IwillintroduceAnagnostopoulou’s(2008)datainthenextsection,showingthatStan- dard German might be a PCC language, before turning to the experiment testing the connectionbetweenCLRandPCC inStandardGerman. 5.2 ThePCCinStandard German The section on Standard German relies heavily on the work by Anagnostopoulou (2008, 2015) who first observes the PCC in Standard German. Anagnostopoulou (2008) argues thatsomespeakersofStandardGermanshowtheweakPCCinaparticularsituation. She arguesthatthePCConlyaffectsweakpronounsifthepronounsprecedethesubject. Wehaveseeninsection2.3.2thatlexicalsubjectsinGermancanberealizedpreceding orfollowingweakobjectpronouns. Thisisdemonstratedin(197)below. (197) StandardGerman a. weil because irgendwer somebody.nom ihn him.acc ihr her..dat vorgestellt introduced hat has b. weil because ihn him.acc dir her..dat irgendwer somebody.nom vorgestellt introduced hat has ‘becausesomebodyintroducedhimtoher’ Anagnostopoulou(2008,2015)arguesthatthepersonconfigurationmatterswhenthe subject follows the pronoun cluster as in (198). Note that the dative argument is third person while the accusative argument is a participant in (198a) and in (198b). Both are PCCviolations,andbothareill-formed. (198) StandardGerman a. *weil because dich you-acc ihm him-dat irgendwer someone-nom vorgestellt introduced hat has PCC ‘becausesomeonehasintroducedyoutohim CHAPTER5. THEPCCINSTANDARDGERMAN 119 b. ??weil because mich me-acc ihr her-dat irgendwer someone-nom vorgestellt introduced hat has PCC ‘becausesomeonehasintroducedmetoher (Anagnostopoulou2008:26) Theconstructioniswell–formedifneitherofthew-pronounsisthirdperson,orifbothof thew-pronounsarethirdpersonpronounsasillustratedin(199)below. (199) StandardGerman a. weil because dich you-acc mir me-dat irgendwer someone-nom vorgestellt introduced hat has ‘becausesomeonehasintroducedyoutome’ b. weil because mich me-acc dir you-dat irgendwer someone-nom vorgestellt introduced hat has ‘becausesomeonehasintroducedmetoyou’ c. weil because ihn him-acc ihr her-dat irgendwer someone-nom vorgestellt introduced hat has ‘becausesomeonehasintroducedhimtoher’ (Anagnostopoulou2008:26) Based on this data, Anagnostopoulou (2008, 2015) concludes that Standard German has theweakPCC(cf.Bonet1991). Anagnostopoulou (2008, 2015) argues that there is no PCC if the subject precedes the weakobjectcluster,asin(200). (200) StandardGerman a. weil because sie she.nom dich you.acc ihm him.dat vorgestellt introduced hat has ‘becausesheintroducedyoutohim’ b. weil because die the Maria Maria.nom mich me.acc ihr her.dat vorgestellt introduced hat has ‘becauseMaryintroducedmetohim’ (Anagnostopoulou2008:26) In a first step, I want to verify Anagnostopoulou’s (2008) claim that Standard German hasthePCCandthatthisisonlythecaseifthesubjectfollowsthepronouncluster. Iwant to test in a second step whether Standard German has the CLR in the same cases that it hasthePCC. CHAPTER5. THEPCCINSTANDARDGERMAN 120 5.3 ExperimentalData I conducted an experiment on the CLR and one on the PCC in German. Both experi- ments were presented together in order to collect data for both conditions from the same participants. Anagnostopoulou (2015) points out that the PCC in German is restricted to participants who judge order reversal of pronouns, i.e. the dat>acc, as ungrammati- cal. She argues that the PCC in languages with weak pronouns are not as robust as for speakers of clitic languages and that speakers who allow for order reversal do not judge PCC violations as ungrammatical. To screen for subjects that allow for order reversal, all subjects where confronted with the sentence in (201). Subjects that did not give (201) a valueof 3 ona7pointLikertscalewereexcludedfromthestudy. (201) Jedem Everyone muss must bewusst aware sein, be, dass that ihm him.dat mich me.acc niemand nobody geben give kann. can ‘Everyonehastoknowthat nobodycangivemetohim.’ 5.3.1 Participants Data from forty native speakers of Standard German was included in the final analysis. Datafrom8participantswasexcludedfromtheanalysisbecausetheparticipantsdidnot judge the sentence in (201) sufficiently bad. There is no clear correspondence between a numericalLikertvalueandgrammaticalityjudgments,butalowvaluelikelyindicatesthat the subject rejects the sentence. 31 of the 40 participants self–identified as having grown up in Austria. One person each self–identified as having grown up in Liechtenstein and Southern Tirol, respectively, and 7 participants as having grown up in Germany. The German group was predominantly from the south (4 participant from Bavaria, 1 participantfromBaden-Württenberg)andtwoparticipantsfromNorth-Rhine-Westphalia. Alltheparticipantswherenaïvetothepurposeofthestudy. 5.3.2 PersonCaseConstraint-Experiment1 The first experiment tested the connection between linear order of the subject in relation to the object pronoun cluster and the PCC. Participants were presented with a context introducing a referent and a target sentence. Subject were then asked to judge whether thetargetsentenceisapossibleGermansentenceandrateitonaseven-pointLikertscale. The order between the pronoun cluster and the subject as well as whether the pronoun clusterwasexpectedtogenerateaPCCviolationornotwasmanipulated. Inotherwords, theexperimenthadtwoindependentbinaryvariables. CHAPTER5. THEPCCINSTANDARDGERMAN 121 MaterialandDesign Eight sets of context and target sentences were constructed. The subjects were simulta- neously presented with the context and the target sentence. The target sentence had a differentbackgroundcolorinordertovisuallyseparatethetwofromeachother1. Figure5.1: Illustrationofquestionaspresentedtothesubjects. The experiment had a 2x2 design. The experiment had two independent variables, whether the subject preceded the pronoun cluster or followed the pronoun cluster and whether the object pronouns violated the PCC or not. Keeping the context constant, the wordorderbetweensubjectandobjectclusterwasmanipulated,sothatonegroupwould seethetargetinanIODOSorderandanothergroupwouldseethetargetsentenceinthe SIODOorder. All the PCC configurations described in the literature block the co-occurrence of a 1st personaccusativeclitic/weakpronounwitha3rdpersondativeclitic/weakpronoun. The PCC condition used the cluster mich ihr ‘me.acc’ ‘her.dat’ or mich ihm ‘me.acc’ ‘him.dat’, thenonPCCconditionusedtheclusterihn/siemir‘him/her.acc’me.dat. Theuseofgender was evenly split between items; half the items had a masculine 3rd person reference and theotherhalfhadafeminine3rdpersonreference. The paradigm is illustrated in (202) and (203). (202) is the PCC condition. (202a) is the target sentence with the subject preceding the object pronouns. (202b) is the target sentencewiththeobjectpronounsprecedingthesubject. (202) Ich I.nom habe have nun now schon already zwei two Monaten months bei at meiner my Mutter mother gewohnt. lived. Ich I.nom werde will 1TheGlossforthisexampleisgivenin (202) below. CHAPTER5. THEPCCINSTANDARDGERMAN 122 aber but morgen tomorrow zu to meiner my Freundin girlfriend ziehen. move. ‘I have already lived at my mother’s place for two months. But I will move to my girlfriendtomorrow.’ a. Meine My Mutter mother freut happy sich self super, greatly, dass that jemand somebody mich me.acc ihr her.dat abnimmt. takeof ‘Mymotherisveryhappythatsomebodyistakingmeofherhands.’ b. Meine My Mutter mother freut happy sich self super, greatly, dass that mich me.acc ihr her.dat jemand somebody abnimmt. takeof ‘Mymotherisveryhappythatsomebodyistakingmeofherhands.’ The context for the non-PCC condition in (203) had to be slightly modified in order to allow for the reversal of the internal arguments in relation to the PCC condition. The target sentences vary minimally in relation to each other with the subject preceding the pronounclusterin(203a)andfollowingtheobjectclusterin(203b) (203) Meine my Freundin girlfriend hat has nun now schon already zwei two Monate months lang long bei with mir me.dat gewohnt. lived. Sie She wird will aber but morgen tomorrow für for einen a Monat month zu to ihren her Eltern parents ziehen. move ‘Mygirlfriendhaslivedfortwomonthswithmebuttomorrowshewillbemoving toherparentsforamonth.’ a. Ich I freu happy mich myself super, greatly, dass that jemand sombody sie her.acc mir me.dat abnimmt. takeof b. Ich I freu happy mich myself super, greatly, dass that sie sombody mir her.acc jemand me.dat abnimmt. takeof ‘Iamveryhappythatsomebodyistakingherofme.’ Participantswerepresentedwith24context-sentencepairsintotal. Thisincluded2train- ing questions, 8 from the PCC-Experiment (Experiment-1), 8 from the LCR-Experiment (Experiment-2), which will be introduced in section 5.3.3 and 6 fillers. The items were pseudo-randomized and kept constant across groups. This was done in order to control potentiallearningeffects. Predictions The difference between Haspelmath’s (2004) and Anagnostopoulou; Anag- nostopoulou’s (2008; 2015) is an empirical question which can be addressed experimen- tally. IfHaspelmath(2004)iscorrect,wewouldexpecttherenottobeadifferencebetween the+PCCandthe-PCCconditionindependentlyofthewordorder. CHAPTER5. THEPCCINSTANDARDGERMAN 123 Table5.1: ExperimentpredictionifnoPCC(Haspelmath2004) +PCC -PCC Sub> DO> IO DO> IO> Sub IfAnagnostopoulou(2008,2015)iscorrect,weexpectthatthereisadifferencebetween +PCC and -PCC and that there is an interaction between word order and the +PCC con- dition, this means that subjects will judge +PCC/OS condition worse than the -PCC/OS conditionandthe+PCC/SOandthePCC/SOconditionnotsignificantlydifferent. Table5.2: PCConlyinSub>DO>IOorder(Anagnostopoulou2008) +PCC -PCC Sub> DO> IO DO> IO> Sub * The last option that I want to consider is that Standard German has the PCC, at least insubordinateclauses. Ifthisistruewepredictthatthe+PCCconditionismarkedworse thanthe-PCCconditionandthatthereisnointeractionbetweenwordorderandthePCC condition. Table5.3: ExperimentpredictionifPCCindependentofwordorder +PCC -PCC Sub> DO> IO * DO> IO> Sub * Procedure Agrammaticaljudgmenttaskwasused. Allitemswerepresentedinwritingonacomputer screenusingGooglesheets. Participantsaccessedthequestionnaireonlineviaalink. The context, the target sentence and the scale were displayed simultaneously for each item. The participant had to judge the sentence by clicking on the appropriate rating from 1-7. Therangewaslabeledwithonekannmansichernichtsagen‘onecancertainlynotsaythat’ to 7 kann man sicher sagen ‘one can certainly say that’. After the question was answered, participants had to press ‘next’ to go to the next question. Participants could evaluate theirprogresswithaprogressbardisplayedinthelowerrightcornerofthewindow. CHAPTER5. THEPCCINSTANDARDGERMAN 124 Results The results for all conditions are illustrated in figure 5.2 below as histograms. We can see that the PCC condition, in the top row, tended to be judged worse than the non-PCC condition, inthebottomrow. Figure5.2: HistogramofresponsestoPCC-Experiment1. The Box plot graph in figure 5.3 indicates the distribution of answers given for each condition, and the median answers, indicated by the line in the box. It shows that the mediananswerforthePCCconditionwerelowerthanthemediananswerforthenon-PCC condition. CHAPTER5. THEPCCINSTANDARDGERMAN 125 Figure5.3: BoxplotgraphforPCC-Experiment1. It is important to stress that subjects judged sentences on a Likert scale and although subjects were given labels for the ends, it would be wrong to treat the data as an inter- val/ratio data. A subject choosing to answer 1 on the first sentence and 2 on the second doesnotexpressthatthesecondsentenceistwiceasgoodasthefirstone. Itwouldalsobe wrongtosimplyassumethateverythingabove4indicatesgrammaticalityandeverything below 4 indicates ungrammaticality. Mangiafico (2016) points out that the median is the primarymeasureoflocation,notmeans. Figure 5.4 illustrates the spread of median Likert score with the bars indicating the first and third quartile to indicate the spread of data from each median.We can see that theanswersforthePCCconditionontherightisconsiderablylowerthantheanswersfor thenon-PCCconditionontheleft. CHAPTER5. THEPCCINSTANDARDGERMAN 126 Figure5.4: MedianLikertScorePCC-Experiment1. IpreformedarepeatedordinalregressionanalysisonthedatautilizingtheCumulative Link Mixed Models function of the R ordinal library (Christensen 2015; R Core Team 2013) with word order (subject preceding, or subject following the pronoun cluster) and PCC (=3 rd :acc=1 st :dat or 1 st :acc=3 rd :dat as independent variables, and participant as a random/blockingvariable. The effect of the PCC condition was significant (p<0.001), while neither the effect of wordordernortheinteractionbetweenwordorderandPCCweresignificant. Participants judged sentences of the PCC condition worse (-1.147) than sentences with the non-PCC condition while they judged sentences in the SO word order not significantly better than theintheOSwordorder(0.203/p>.4). Itestedthemodelassumptionswithnominal_test scale_testoftheordinalpack- age(followingMangiafico2016)andconcludednoviolationsofthemodelassumptions. Discussion The experiment supports the existence of a PCC effect in subordinate clauses in German. ThisPCCeffectisnotdependentonthewordorderofthesubjectwithrespecttotheweak CHAPTER5. THEPCCINSTANDARDGERMAN 127 object pronouns, contrary to observations made in Anagnostopoulou (2008, 2015). The experimentreliedonindefinitesubjectquantifiers. Thiswasdonetoallowforsubjectsto more naturally occur in a lower position (cf. Wurmbrand 2004). There is no significant effectofwordorderandnosignificantinteractionbetweenwordorderandPCC. ItispossiblethatthePCCeffectseemsstrongerwithdefinitesubjectswhenconducting simplegrammaticalityjudgmenttestswithsubjectsinthepostobjectposition,sincethisis anon-canonicalpositionfordefinitesubjects. Inthatcasetheremightbeacompounding effectofPCCandapreferredsurfacepositionofsubjects. 5.3.3 LCR-Experiment2 StandardGermandoesnothavecliticsanditisdifferentfromSpanishandFrench,theLCR as formulated in Charnavel & Mateu (2015) therefor does not capture Standard German. We have seen in the results from Experiment 1 that Standard German has the PCC, at least in subordinate clauses. Experiment 2 tested whether we see the same effects in Standard German as Charnavel & Mateu (2015) captured in Spanish and French and I will demonstrate that weak accusative pronouns in Standard German have coreference restrictionswhichweakdativepronounsdonothave. MaterialandDesign Eight sets of context and target sentences were constructed. As in for experiment I, participants were simultaneously presented with the context and the target sentence. Visually, experiment 2 mirrored experiment 1. The context introduced two potential human references by name, one male and one female. Both names were accompanied by definite articles, this was done for two reasons; participants were mostly from areas wherenamesexclusivelyoccurredwithdefinitearticles. Definitearticlesalsoadditionally marked the gender of the referent, making sure that subjects could easily identify the intended antecedent of pronouns. The experiment was set up in a 2 by 2 design with word order and reference as independent variables. Subject would see a context as in (204) followed by a target sentence. The target sentence would have an embedded clause with either subject> DO> IO word order as in (204a) (204c) or DO> IO> subject as in (204b) (204d). The other independent variable was which pronoun was coreferent with the matrix subject. The accusative object pronoun could corefer with the matrix subject (204a)(204b) (+LCR/ACC condition) or the dative object pronoun could corefer with the matrixsubject(204c)(204d)(-LCR/DATcondition). CHAPTER5. THEPCCINSTANDARDGERMAN 128 (204) Die the.fem Susi Susi will wants den the.mas Fritz Fritz kennenlernen. gettoknow ‘SusiwanttogettoknowFritz.’ a. Die the.fem Susi Susi möchte wants daher, therefore, dass that jemand somebody.nom sie her.acc ihm him.dat vorstellt. introduces b. Die the.fem Susi Susi möchte wants daher, therefore, dass that sie her.acc ihm him.dat jemand somebody.nom vorstellt. introduces. ‘Susithuswantssomebodytointroducehertohim.’ c. Die the.fem Susi Susi möchte wants daher, therefore, dass that jemand somebody.nom ihn him.acc ihr her.dat vorstellt. introduces. d. Die the.fem Susi Susi möchte wants daher, therefore, dass that ihn her.acc ihr him.dat jemand somebody.nom vorstellt. introduce. ‘Susithuswantssomebodytointroducehimtoher.’ We do not expect differences between the conditions if the LCR is in fact only a restriction on clitics. If the LCR is dependent on the PCC and not restricted to clitics, we expect the same distribution for the LCR as for the PCC. This means that we expect the +LCR condition to be judged significantly lower than the -LCR condition, and no interaction between word order and LCR. This prediction is illustrated in table 5.4 We expect Table5.4: ExperimentpredictionofCLR +LCR/ref. withacc -LCR/ref. withdat Sub>DO>IO * Sub>DO>IO * Thispredictionininfactborneout. Wewillseeinthenextsectionthatcoreferencebe- tweenthematrixsubjectandtheembeddedaccusativepronounwerejudgedsignificantly lowerthancoreferencebetweenthematrixsubjectandtheembeddeddativepronoun. Results The results for the four conditions are illustrated in the graph in figure (5.5) below. The top row corresponds to the results for -CLR cases where the matrix subject corefers with the dative argument of the embedded clause, the bottom row are the +CLR cases where thesubjectmatrixwascoreferentwiththeembeddedaccusativepronoun. Thegraphson theleftrepresentthetwoconditions,wherethesubjectfollowedtheobjectpronounsand therightcolumnrepresenttheresultsfortwoconditions,wherethesubjectprecededthe objectpronouns. CHAPTER5. THEPCCINSTANDARDGERMAN 129 Figure5.5: ResultsCLR-Experiment2. We can see in the figure above that the frequency of subjects responded with either 6 or7(‘onecansaythat’)ismuchhigherinthecaseswherethedativeandnottheaccusative pronounreferredtothematrixsubject. Figure 5.6 below gives the box plot for the results from the CLR experiment. I give this more for completeness’ sake, means are not particularly informative for Likert items. However, one can see that the averages for the SO ordered targets are higher than for OS ordereditems. CHAPTER5. THEPCCINSTANDARDGERMAN 130 Figure5.6: BoxplotCLR-Experiment2. Mangiafico (2016) points out that median-Likert plots are more suited to graphically represent results on Likert scales. Figure 5.7 is the median Likert plot for the CLR data. The points indicate the median score for the condition while the bars indicate the first and third quartile. We can see that median scores for dative reference is higher than for accusative. We can also see that the SO order (blue) tended to be judged better than the correspondingOSorderedsentence(red). CHAPTER5. THEPCCINSTANDARDGERMAN 131 Figure5.7: Median-LikartplotCLR-Experiment2. Boththedifferencesbetweendativeandaccusativereferenceaswellasthewordorder differencesweresignificant. Iperformedarepeatedordinalregressionanalysisonthedata utilizing the Cumulative Link Mixed Models function of R ordinal library (Christensen 2015) with word order (subject preceding, or subject following the pronoun cluster) and CLR (accusative pronoun or dative pronoun referring to the matrix subject) as indepen- dent variables, and participant as a random/blocking variable. Sentences in the +CLR conditionwerejudged1.71pointsworsethansentencesinthe-CLRcondition. Thisdif- ference was significant (p<0.0000001).The OS word order was judged.83 points worse than sentences in the SO configuration. This difference was also significant (p0.007). Importantly, there was no significant interaction between CLR and word order (p>0.5). The modal assumptions where tested with the nominal_test and the scale_test of the ordinal package (following Mangiafico 2016). No violations of the model assumptions werefound. CHAPTER5. THEPCCINSTANDARDGERMAN 132 Discussion The experiment demonstrates that there is a CLR effect in Standard German. Subjects judged sentences where accusative pronouns refer to matrix subjects significantly worse thansentenceswhereadativepronounhadthesamereferenceasthematrixsubject. This supports the view that weak pronouns in Standard German are subject to the CLR. This effectdoesnotdependonthepositionoftheembeddedsubject. Incontrasttoexperiment 1, word order differences had a significant effect in this experiment. I propose that this difference is due to an unintentional garden pathing of the participants in the OS word orderbutnotintheSOorder. Thereisa3sg.fem.nom/3sg.fem.accsyncretisminGerman. sie‘she/her’doesnotdistinguishbetweenthenominativesubjectandtheaccusativeobject form. This means that the string dass sie ihm ‘that she/her him’ is ambiguous between a reading where the first pronoun is the subject of the sentence, and one where the firs pronounisadirectobjectpronoun. Kempen & Harbusch (2004) have a corpus study of subordinate clauses that shows that the combination subject pronoun-IO pronoun is much more frequent than the order DO pronoun-IO pronoun. It is reasonable to assume that subjects reading the sentence willhavetoreinterpretsieasanobjectafterinitiallyparsingitasasubjectpronounwhich couldlead suchsentencestoberatedworse. 5.3.4 Conclusion This experimental data supports the hypothesis that Standard German has the PCC in- dependently of word order, at least in subordinate clauses. This contradicts Haspelmath (2004), who claims that Standard German is a language that does not have the PCC. The experimental data presented here also suggest that the CLR has to be expanded and is in fact not a restriction on clitics but on clitics as well as weak pronouns. The CLR therefore affects the same class of words (weak pronouns and clitics) that are also affected by the PCC. 5.4 AnalysisforStandard German There are two differences between Standard German and Swiss German that need to be accountedfor. ThePCCinSwissGermanisaninstantiationoftheultra-strongPCC,while the Standard German PCC is an instantiation of the weak PCC. The second aspect is that the PCC in Swiss German is restricted to the dat<acc order, the acc<dat order does not CHAPTER5. THEPCCINSTANDARDGERMAN 133 have a PCC restriction. Standard German on the other hand does not have a PCC in the acc<datorderandbansthedat<acc orderforweakpronouns2 I will derive the PCC in Standard German by modifying two aspects of the Swiss German analysis, one is the location of Per, the other is the feature specification on Per. I have argued that Per is located above AgrOP in Swiss German. The Standard German Per head is located between AgrO and Appl and has each an interpretable but unvalued [prox]andaninterpretablebutunvalued[part]feature. Thisisillustratedin(205). (205) LocationofPerinStandardGerman AgrOP PerP ApplP ApplP VP V acc Appl dat Per [] [prox] [part] AgrO The IO is in a privileged position because no relevant argument intervenes between Per and the IO. This means that if the IO is first or second person, Per will agree with the IO. The only way that Per could agree with the DO is if the DO is a 3rd person pronoun (without[part])andtheDOisa1stor2ndpersonpronoun(with[part])(207). 2Anagnostopoulou(2008)pointsoutthatthepronounses‘it’doesnotseemtobefollowingthisrestriction andcaninfactsurfaceineitherorder. ItisnoticeablethattheSwissGermanequivalent’sisalsounaffected fromorderingrestrictionswhichapplytootheraccusativeweakpronouns. Iwillnotdealwiththisparticular issue here, but I want to suggest that it might be connected to the fact the es often refers to events and not toelementsoftype<e>. CHAPTER5. THEPCCINSTANDARDGERMAN 134 (206) a. Deriving the Weak PCC - Agree withIO PerP ApplP ApplP VP V acc Appl dat u[] u[prox] (u[part]) Per [] [prox] [part] b. Deriving the Weak PCC - Move the IO PerP PerP ApplP ApplP VP V acc Appl dat Per [] [prox] [part] dat u[] u[prox] (u[part]) 3 (207) DerivingtheWeakPCC-failedagreewiththeIO AgrOP PerP ApplP ApplP VP V acc u[] u[prox] u[part] Appl dat u[] u[prox] Per i[] i[prox] i[part] AgrO x 3 The weak PCC in Standard German results in this analysis from a low Per head. The wordorderdifferencebetweenSwissGerman(flexiblewordorder)andStandardGerman (restrictedwordorder)alsoresultsfromthelowPerhead,inparticularformthePerhead being in the c-command domain of AgrO. After Per is merged and triggers movement of the dative argument, AgrO is merged and triggers movement of the weak DO. This is illustratedin(208)below. CHAPTER5. THEPCCINSTANDARDGERMAN 135 (208) DerivingtheDO>IOorder AgroOP AgrOP PerP PerP ApplP ApplP VP V acc Appl dat Per dat AgrO acc ThedifferencebetweenthePCCofSwissGermanandStandardGermancanbeexplained by thelocation ofthe Per head,the features onPer and whetherthe weakIO has to move tospecPerP. Chapter6 DerivingtheVariationinthePCC The variation in the PCC can be derived by varying the location of the Per head and the feature modification on Per. While there are many different settings, 70 to be precise, manyofthesemodificationswillgeneratethesameoutput. I will demonstrate how this approach derives the PCC variation described in the literaturebeforeturningtothequestionwhichotherPCCsthisapproachpredicts. 6.1 DescribedVariation I have introduced in table 1.1 the described variation in the PCC. The table is repeated in table6.1below. Table6.1: TableofknownPCCvariationformPancheva(2016) DAT ACC weakPCC me-first Ultra-strongPCC strongPCC super-strong 1 3 2 3 3 3 x 1 2 x x 3 2 x x x x 2 1 x x x x 3 1 x x x x x An analysis of the PCC should be able to account for the observed variation. I will demonstrateinthefollowinghowmyanalysisofthePCCinSwissGermanandStandard GermancanaccountfortheknownvariationinthePCC. 136 CHAPTER6. DERIVINGTHEVARIATIONINTHEPCC 137 6.1.1 StrongPCC The strong PCC results from having the Per head preceding AgrO. This means that the accusative argument can function as an intervener, having an interpretable [part] and [prox]featureonPer. ThereisnothingthatpreventsPerfromalsohavinga feature,but itwillnotaffectthederivation. (209) StrongPCC PerP PerP AgrOP AgrOP ApplP ApplP VP V acc Appl dat AgrO acc prox[] part[] dat agree move The accusative argument functions as an intervener if it has a participant feature, which derives the basic facts of the strong PCC, namely that the accusative argument has to be 3rd person. The second thing that this structure predicts is that the accusative argument cannot be marked proximate if the dative argument is third person. This is Charnavel & Mateu’s (2015) CLR. It furthermore predicts that the dative argument has to have at least a[+prox] featurewhichmeansithastobeapointofview. 6.1.2 Super-StrongPCC Thedifferencebetweenthestrongandthesuper-strongPCCisthatthesuper-strongPCC bans the co-occurrence of two 3rd person reduced pronouns. The ultra-strong PCC is derived by changing the feature specification on Per. Per in (210) can only agree with the dative argument if only the dative argument is a participant. Since Per is not probing for features that 3rd person pronouns have, Per can never agree with a 3rd person dative argumentifanaccusativeargumentintervenes. CHAPTER6. DERIVINGTHEVARIATIONINTHEPCC 138 (210) StrongPCC PerP PerP AgrOP AgrOP ApplP ApplP VP V acc Appl dat part[val] AgrO acc part[] dat agree move 6.1.3 Ultra-strongPCC IhaveshownwiththeSwissGermanexamplehowtoderivetheultra-strongvarietythat blocks the co-occurrence of two 3rd person pronouns. Nevins (2007) argues for an ultra- strongPCCconfigurationthatdoesnotbantheco-occurrenceoftwo3rdpersonreduced pronouns. This can be derived in two ways. One is similar to the strong PCC, with an addedauthorfeature. Thisisillustratedin(211)below. (211) Nevins’s(2007)ultra-strongPCC CHAPTER6. DERIVINGTHEVARIATIONINTHEPCC 139 PerP PerP AgrOP AgrOP ApplP ApplP VP V acc Appl dat AgrO acc prox[] part[] author[] dat agree move Thisconfigurationrequiresthatadativeargumenthastobeatleastthepointofviewholder and that the accusative argument cannot have prox[val] feature if the dative argument is 3rdperson,whichforcestheCLR. Nevins’s (2007) ultra-strong PCC can also be derived if the PerP is lower than AgrO. The only difference to the one above is that the syntax does block the co-occurrence two internal arguments that have the same feature specification and that the CLR has to be derivedbyaseparatemechanism,whichseemstoexistindependentlyanyway. 6.1.4 Me-firstPCC Theme-firstPCCisderivedbyhavingPermergedbelowAgrOandhavinganauthorand aproximate/ featureasisillustratedin(212). ThiswillensurethatthattheIOcanagree withPeraslongastheAccusativeargumentisnot1stperson. (212) Me-FirstPCC CHAPTER6. DERIVINGTHEVARIATIONINTHEPCC 140 AgrOP PerP PerP ApplP ApplP VP V acc Appl dat prox[] author[] dat AgrO agree move 6.2 PredictedVariation 6.2.1 No-PCC Thisapproachalsopredictssomevariationthatisnotcapturedintable6.1above. Thisac- countwouldpredictthattherearelanguagesthathavenorestrictionsontheco-occurrence of two internal pronouns or block the co-occurrence of two internal reduced pronouns. Both of these variations would be caused by having Per with only a feature. If Per is lower than AgrO, as in (213), the agree relation can never be blocked, which means that there will be no intervener. If Per is higher than AgrO, every DO will block agreement between Per and the IO (214). Doliana (2014) calls these two versions the Zero and the Giga PCC. He gives German as a Zero-PCC case and Hausa as a Giga-PCC case. Work by Anagnostopoulou (2015) as well as the study presented here raises doubt whether Germanis infactacaseofZero-PCC. (213) ZeroPCC CHAPTER6. DERIVINGTHEVARIATIONINTHEPCC 141 AgrOP PerP PerP ApplP ApplP VP V acc Appl dat prox[] author[] dat AgrO agree move (214) GigaPCC PerP AgrOP AgrOP ApplP ApplP VP V acc Appl dat AgrO acc x Hausahasasetofstrongandreducedpronouns. Theaccusativereducedpronounisonly grammatical in the absence of a dative pronoun (215a) (215c)1. The only way to express the accusative pronoun in the presence of a dative pronoun is by using a strong pronoun (215d). (215) a. Audù Audu yaa he kaawoo brought tà her/it ‘Audobroughtit.’ b. Audù Audu yaa he dafàa’ cooked makà for you.dat ‘Audocookedforyou.’ 1Note that the tone on the weak accusative pronoun is a polar tone depending on the preceding tone. It isthereforelowin(215a)butwouldbehigh in (215c) CHAPTER6. DERIVINGTHEVARIATIONINTHEPCC 142 c. *Audù Audu yaa he kaawoo brought makà her/it ta foryou.dat ‘Audobroughtitforyou.’ d. Audù Audu yaa he kaawoo brought makà her/it ita foryou.dat ‘Audobroughtitforyou.’ (Doliana2014:5) (216) demonstrates that we are not dealing with a restriction on local pronouns (216a) or somePCCthatbanstheco-occurrenceof3rdpersonpronouns(216b). (216) a. *Audù Audu yaa he kaawoo brought masàa her/it ta foryou.dat ‘Audobroughtitfor him.’ b. *Audù Audu yaa he kaawoo brought minìi you ka forme.dat ‘Audobroughtitfor you.’ (Doliana2014:4) 6.2.2 StrongPCCversion2 My account also predicts the occurrence of a PCC that is very similar to the Strong PCC which bans the co-occurrence of 1st or 2nd person datives with reduce accusatives pronouns which are marked proximate. Ormazabal & Romero (2007) report that some Spanish leista speakers ban the co-occurrence of 1st and 2nd person dative with animate accusative clitics. Thisisillustratedin(217). (217) Someleistadialect a. Te 2dat lo 3acc.inanimate di ‘Igaveittoyou.’ b. *Te 2dat le 3acc.animate di ‘Igaveittoyou.’ Assumingthatanimatepronounsaretherealizationofaproximatefeature,thisrestriction can be derived by changing the Per head for the strong PCC in (209) from having a participantandaproximatefeaturetoonlyhavingaproximatefeature. CHAPTER6. DERIVINGTHEVARIATIONINTHEPCC 143 6.2.3 OnlyMePCC The last possible PCC that I want to discuss here is a PCC that requires that the dative argumentonlybea1stpersonpronoun. (218) OnlyMePCC PerP AgrOP AgrOP ApplP ApplP VP V acc Appl dat AgrO acc [author] Agree Thishas,tomyknowledge,notbeenobservedinnaturallanguage. Theonlywaythatmy account can account for why the only-me PCC is not more common in the observed lan- guagesisthatlearnersofalanguagewheretheyencountersomevariationwithrespectto thePCCwillmostlikelyhavepositiveevidencethattheirtargetlanguageisnotame-only language, since every other type of PCC allows combinations which are ungrammatical intheme-onlyPCC. 6.3 Conclusion The PCC variation can be reduced to a generalized syntactic agree mechanism, Maximize Agree,andlanguagespecificspecificationsonafunctionalheadthatcarriesonlyunvalued person features. My approach reduced the variation in the PCC to a lexical difference which varies between languages and speakers. The reasons why we see inter–speaker variation is that language learners have to make inferences about the lexical specification ofthePerhead. Chapter7 Conclusion Thecontributionsofthisdissertationaretwo-fold,theyarebothempiricalandtheoretical. Ontheempiricalside,anewdatasetispresentedwhichisrelevantandsofar,remains unexplainedinthePCCliterature,namelythecaseofSwissGermanorderingrestrictions on reduced pronouns. This data set poses specific challenges because it effects elements that are clearly syntactic and the influence of structure on the PCC. An important part of this dissertation focused on the solicitation of this data and on providing a quantitative analysisforthePCCinSwissGermanaswellasStandardGerman. The PCC is a restriction on weak pronouns but not on DP arguments and strong pronouns. Iprovidedanaccountforweakandstrongpronounsinchapter2thataccount forthispatternandderivethePCCasarestrictiononargumentsthatlackaDPlayer. On the theoretical side, I propose an analysis of the PCC in general, which provides a naturalanalysisfortheSwissandStandardGermandatapresented. Thepresentanalysis only relies on two assumptions: the principle of Maximize Agree (Van Urk & Richards 2015) as well as the assumption of a probe goal structure of agreement. In addition to these assumptions, no specialized mechanisms are required. Not only does this analysis accountfor thedata inthe twoGerman varietiesthat aremainly discussedin thepresent work, it is also capable of deriving all other PCC types that have been discussed in the literaturesofar. As such, this proposal allows for easy comparison and clear conclusion of PCC phe- nomena across languages. It makes the clear statement that PCC is a syntactic phe- nomenon,anditisnotduetoaphonologicalprocess. Inaddition,itfollowsfromassump- tions about the nature of Agree. The data from Standard German demonstrates that the CLRproposedbyCharnavel&Mateu(2015)isnotrestrictedtocliticsbutalsoeffectsweak pronouns and the data in Swiss German suggest that the CLR might have an effect that goes beyond 3rd person arguments. Speaker preferences on ordering in Swiss German 144 CHAPTER7. CONCLUSION 145 supportPancheva&Zubizarreta’s(2017)assumptionsabouttheconnectionbetweenCLR and PCC, in that they are not due to the same phenomena but liked via the interpretive requirementsofinterpretablepersonfeaturesinthesententialspine. Bibliography Abraham, Werner. 1996. German standard pronouns and non-standard pronominal clitics: Typologicalcorollaries.Pages1–14of: Lippi-Green,RosinaL,&Salmons,JosephC (eds), Germaniclinguistics: syntacticanddiachronic. JohnBenjaminsPublishing. Abraham, Werner. 2007. Discourse binding - DP and pronouns in German, Dutch, and English. Pages 21–47 of: Stark, Elisabeth, Leiss, Elisabeth, & Abraham, Werner (eds), Nominal Determination: Typology, context constraints, and historical emergence. John Benjamins. Albizu, Pablo. 1997. Generalized Person-Case Constraint: A case for a syntax-driven inflectional morphology. Pages 1–33 of: Mendikoetxea, Amaya, & Uribe-Etxebarria, Myriam (eds), Theoretical issues on the morphology-syntax interface. ASJU XL. Donostia: GipuzkoakoForuAldundiaandEuskalHerrikoUnibertsitatea. Alexiadou, Artemis, Haegeman, Liliane, & Stavrou, Melita. 2007. Noun Phrase in the GenerativePerspective. deGruyter. Anagnostopoulou, Elena. 2003. The Syntax of Ditransitives, Evidence from Clitics. Berlin: MoutondeGruyter. Anagnostopoulou,Elena. 2008. Notes on the Person Case Constraint in Germanic (with special reference to German). Page 15 of: D’Alessandro, Roberta, Fischer, Susann, & Hrafnbjargarson, Gunnar Hrafn (eds), Agreement restrictions, vol. 15. Walter de Gruyter. Anagnostopoulou,Elena.2015(September). OrderingPatternsandtheSyntaxofthePerson CaseConstraint(PCC). 12thConferenceonGreekLinguistics. FreieUniversiätBerlin. Barker, Chris. 2012. Quantificational binding does not require c-command. Linguistic inquiry,43(4),614–633. 146 BIBLIOGRAPHY 147 Bayer, Josef. 1999. Comment on Cardinaletti and Starke “The typology of structural deficiency”. Pages 235–241 of: vanRiemsdijk,Henk (ed), Clitics in the Langues of Europe, vol.1. Berlin: MoutondeGruyter. Béjar,Susana,&Rezac,Milan. 2003. Person licensing and the derivation of PCC effects. Pages 49–62 of: Pérez-Leroux, Ana Teresa, & Roberge, Yves (eds), Romance Linguistics: TheoryandAcquisition. Amsterdam;Philadelphia: JohnBenjamins. Béjar,Susana,&Rezac,Milan.2009. CyclicAgree. LinguisticInquiry, 40(1),35–73. Bhatt,Rajesh,&Šimík,Radek.2009(October). Variablebindingandthepersoncaseconstraint. 25th Annual Meeting of the Israel Association for Theoretical Linguistics. Ben Gurion UniversityoftheNegev. Bonet,Eulalia.1991. Morphologyaftersyntax: PronominalcliticsinRomance. MITPress. Bonet,Eulalia.1994.ThePerson-CaseConstraint: Amorphologicalapproach.Pages33–52 of: Phillips, H. Harley & C. (ed), MIT working papers in linguistics 22: The morphology- syntaxconnection. Cambridge,MA:MITWPL. Bonet,Eulàlia. 2008. The Person-Case constraint and repair strategies. Pages 103–128 of: D’Alessandro, Roberta, Fischer, Susann, & Hrafnbjargarson, Gunnar Hrafn (eds), Agreementrestrictions. Berlin-NewYork: MoutondeGruyter. Cardinaletti,Anna. 1999. in Germanic and Romance Languages : An overview. In: van Riemsdijk,Henk(ed), CliticsintheLanguagesofEurope. Berlin: MoutondeGruyter. Cardinaletti, Anna, & Starke, Michael. 1999a. Responses and Demonstratives. Pages 273–290 of: van Riemsdijk, Henk (ed), Clitics in the Langues of Europe, vol. 1. Berlin: MoutondeGruyter. Cardinaletti, Anna, & Starke, Michal. 1994. The typology of structural deficiency. On thethreegrammaticalclasses. Workingpapersinlinguistics,4(2),41–109. Cardinaletti,Anna,&Starke,Michal.1996.Deficientpronouns: AviewfromGermanic. Pages 21–65 of: van Riemsdijk, Henk (ed), Clitics in the Languages of Europe. Mouton de Gruyter. Cardinaletti, Anna, & Starke, Michal. 1999b. The typology of structural deficiency: A casestudyofthethreeclassesofpronouns. Pages145–234of: vanRiemsdijk,Henk(ed), CliticsintheLanguesofEurope,vol.1. Berlin: MoutondeGruyter. BIBLIOGRAPHY 148 Charnavel, Isabelle, & Mateu, Victoria. 2015. The clitic binding restriction revisited: Evidenceforantilogophoricity. TheLinguisticReview, 32(4),671–701. Chomsky, Noam. 2000. Minimalist Inquaries: The framework. Pages 89–155 of: Martin, Roger, Michaels, David, & Uriagereka, Juan (eds), Step by step: Essays on minimalist syntaxinhonorofHowardLasnik. Cambridge,Mass.: MITPress. Chomsky,Noam.2001. Beyondexplanatoryadequacy. MIT. Christensen,R.H.B.2015. ordinal—RegressionModelsforOrdinalData. Rpackageversion 2015.6-28.http://www.cran.r-project.org/package=ordinal/. Cooper,Katharin.1994. TopicsinZurichGermansyntax. Ph.D.thesis,Univ.Edinburgh. Cooper, Kathrin. 1999. On the nature and distribution of Zurich German pronominal clitics. Pages 711–730 of: van Riemsdijk, Henk (ed), Clitics in the Languages of Europe. MoutondeGruyter. DeVincenzi,Marica,&diDomenico,Elisa. 1999. A distinction among phi-features: The roleofGenderandNumberintheretrievalofpronounantecedents. Genderandnumber in normal and impaired language processing, Special Issue of Rivista di Linguistica, 11(1Oana Ciucivara),21–74. Déchaine,Rose-Marie,&Wiltschko,Martina. 2002. Decomposing pronouns. Linguistic inquiry,33(3),409–442. Diesing,Molly.1992. Indefinites. Cambridge(Mass.): MITPress. Diesing, Molly. 1999. Comment on Cardinaletti and Starke “The typology of structural deficiency”. Pages 242–247 of: vanRiemsdijk,Henk (ed), Clitics in the Langues of Europe, vol.1. Berlin: MoutondeGruyter. Doliana,Aaron.2014(Sept). HardandSoftPerson–CaseConstraints. CGSW29. University ofYork. Elbourne,Paul. 2001. E-type anaphora as NP-deletion. Natural Language Semantics, 9(3), 241–288. Franks,Steven,&ŽeljkoBo²kovi¢.2001. AnArgumentforMultipleSpell-Out. Linguistic Inquiry, 32(1),174–183. BIBLIOGRAPHY 149 Frey, Werner. 2006. How to get an object-es into the German prefield. Pages 159–185 of: Fuss, Eric, & Brandt, Patrick (eds), A festschrift presented to Günther Grewendorf on the occasionofhis60 th birthday. Berlin: AkademieVerlag. Frey, Werner, & Pittner, Karin. 1998. Zur Positionierung der Adverbiale im deutschen Mittelfeld. LinguistischeBerichte,176(176),489–534. Fuchs, Zuzanna. 2015. Second Position and “Floating” Clitics in Wakhi. Pages 133– 153 of: Jurgensen,AnnaE.,Sande,Hannah,Lamoureux,Spencer,&andAlisonZerbe, KennyBaclawski(eds),ProceedingsoftheAnnualMeetingoftheBerkeleyLinguisticsSociety, vol.41. Georgala,Effi. 2011. Why German Is Not an Exception to the Universal<IO, DO> Base Order of Double Object Constructions. In: Washburn, Mary Byram, McKinney-Bock, Katherine,Varis,Erika,Sawyer,Ann,&Tomaszewicz,Barbara(eds),Proceedingsofthe 28thWestCoastConferenceonFormalLinguistics. Somerville,MA:CascadillaProceedings Project. Haegeman, Liliane. 1993. The Morphology and Distribution of Object Clitics in West Flemish. StudiaLinguistica, 47(1),57–94. Haider,Hubert.2010.TheSyntaxofGerman.Cambridge,NewYork: CambridgeUniversity Press. Haspelmath, Martin. 2004. Explaining the ditransitive person-role constraint: A usage- basedapproach. Constructions, 2(2004),1–71. Hornstein,Norbert. 2009. A theory of Syntax: Minimal Operations and Universal Grammar. Cambridge: CambridgeUniversityPress. Hornstein,Norbert,Nunes,Jairo,&Grohmann,KleanthesK. 2005. Understanding mini- malism. Cambridge: CambridgeUniversityPress. Jäger, Gerhard. 2001. Topic-Comment Structure and the Contrast Between Stage Level andIndividualLevelPredicates. JournalofSemantics, 18(2),83–126. Kempen,Gerard,&Harbusch,Karin. 2004. A corpus study into word order variation in Germansubordinateclauses: Animacyaffectslinearizationindependentlyofgrammat- ical function assignment. Pages 173–181 of: Pechmann,Thomas,&Habel,Christopher (eds), Multidisciplinaryapproachestolanguageproduction. Berlin: MoutondeGruyter. BIBLIOGRAPHY 150 Koopman, Hilda. 2000. The internal and external distribution of pronominal DPs. Pages 77–118of: TheSyntaxofSpecifiersandHeads-CollectedEssaysofHildaJKoopman.Routledge. Koopman, Hilda. 2006. Agreement: in defense of the “Spec head configuration”. Pages 159–199of: Boeckx,Cedric (ed), Agreementsystems. Benjamins. Kratzer,Angelika. 2009. Making a pronoun: Fake indexicals as windows into the prop- ertiesofpronouns. LinguisticInquiry, 40(2),187–237. Krifka,Manfred. 1998. Scope inversion under the rise-fall contour in German. Linguistic Inquiry, 29(1),75–112. Larson,Bradley. 2013. Arabic Conjunct-Sensitive Agreement and Primitive Operations. LinguisticInquiry,44(4),611–631. Lechner,Winfried,&Yatsushiro,Kazuko.2001. Projektantrag: BindingandScopeinDouble ObjectConstructions. UniversityofTübingen. López,Luis.2009.Aderivationalsyntaxforinformationstructure.Oxfordstudiesintheoretical linguistics,vol.23. NewYork,NY:OxfordUniversityPress. Mangiafico, Salvatore S. 2016. Summary and Analysis of Extension Program Eval- uation in R, version 1.13.0. rcompanion.org/handbook/. (Pdf version: rcompan- ion.org/documents/RHandbookProgramEvaluation.pdf.). McIntyre,Andrew. 2006. The interpretation of German datives and English have. Pages 185–211 of: Hole,Daniel,Meinunger,André,&Abraham,Wernder (eds), Datives and othercases. JohnBenjaminsAmsterdam. Meinunger,Andeé.2007. AboutObjectesintheGermanVorfeld. LinguisticInquiry,38(3), 553–563. Müller,Gereon. 2002. Harmonic Alignment and the Hierarchy of Pronouns in German. Pages 205–232 of: Simon, H., & Wiese, H. (eds), Pronouns–Grammar and Representation. Amsterdam: Benjamins. Murphy,Andrew.2017(Oct.). PronominalinflectionandNPellipsisinGerman. MS.Leipzig University. Nevins, Andrew. 2007. The representation of third person and its consequences for person-caseeffects. NaturalLanguage&LinguisticTheory, 25(2),273–313. BIBLIOGRAPHY 151 Newmann, Paul. 2000. The Hausa Langauge: An Encyclopedic Reference Grammar. New Haven&London: YealUniversityPress. Nübling,Damaris.1992. KlitikaimDeutschen: Schriftsprache,Umgangssprache,alemannische Dialekte. GunterNarr. Ormazabal, Javier, & Romero, Juan. 2007. The Object Agreement Constraint. Natural Language&LinguisticTheory, 25(2),315–347. Ouwayda,Sarah.2016. personalcommunication. Pancheva,Roumyana.2016. PersonCaseConstraint. USCSyntax-SeminarHandout. Pancheva, Roumyana, & Zubizarreta, Maria Luisa. 2017. The Person Case Constraint: TheSyntacticEncodingofPerspective. NaturalLanguage&LinguisticTheory. Patel-Grosz,Pritty,&Grosz,PatrickG.2017. RevisitingPronominalTypology. Linguistic Inquiry, 48(2),259–297. Pesetsky,David,&Torrego,Esther.2007. Thesyntaxofvaluationandtheinterpretability of features. Pages 262–294 of: Wilkins,Wendy,&Karimi,Simin (eds), Phrasal and clausal architecture: Syntacticderivationandinterpretation. JohnBenjamins. Preminger, Omar. 2011. Agreement as a fallible operation. Ph.D. thesis, Massachusetts Institute ofTechnology. Pylkkänen,Liina.2008. Introducingarguments. Ph.D.thesis,MIT. RCoreTeam.2013. R:ALanguageandEnvironmentforStatisticalComputing. RFoundation forStatisticalComputing,Vienna,Austria. Richards,MarcD.2011. DerivingtheEdge: What’sinaPhase? Syntax, 14(1),74–95. Roca,Francesc. 1992. On the licensing of pronominal clitics: The properties of object clitics in SpanishandCatalan. M.Phil.thesis,UniversitatAutnomadeBarcelona. Ross,JohnRobert.1967. ConstraintsonVariablesinSyntax. Ph.D.thesis,MIT. Schwarz,Florian. 2009. Two types of definites in natural language. Ph.D. thesis, University ofMassachusetts,Amherst. Sells,Peter.1987. Aspectsoflogophoricity. Linguisticinquiry, 18(3),445–479. BIBLIOGRAPHY 152 Siewierska, Anna. 2004. Person. Cambridge textbooks in linguistics. Cambridge, New York: CambridgeUniversityPress. Sigurðsson,HalldórÁrmann,&Holmberg,Anders.2008. Icelandicdativeintervention: Personandnumberareseparateprobes. Pages251–280of: Agreementrestrictions. Berlin: MoutondeGruyter. Simpson,Andrew,&Syed,Saurov. 2016. Blocking Effects of Higher Numrerals in Bangla: APhase-BasedAnalysis. LinguisticInquiry, 47(4),754–763. Sportiche, Dominique. 1988. A theory of floating quantifiers and its corollaries for con- stituentstructure. LinguisticInquiry,19(3),425–449. Stegovec,Adrian.2017(October). TakingCaseoutofthePerson-CaseConstraint. UConn. Sternefeld, Wolfgang, & Featherston, Sam. 2003. The German reciprocal einander in double object constructions. Pages 239–265 of: v.Gunkel,L.,Müller,G.,&Zifonun,G. (eds), ArbeitenzurReflexivierung. Tübingen: MaxNiemeyerVerlag. VanUrk,Coppe,&Richards,Norvin.2015. Twocomponentsoflong-distanceextraction: SuccessivecyclicityinDinka. LinguisticInquiry, 46(1),113–155. vonFintel,Kai,&Heim,Irene.2011. IntensionalSemantics. MIT. Walkow,Martin.2013. Locatingvariationinpersonrestrictions. UniversityofPennsylvania WorkingPapersinLinguistics,19(1),28–. Watanabe,Akira.2013. Person-numberinteraction: impoverishmentandnaturalclasses. Linguisticinquiry, 44(3),469–492. Werner,Ingegerd.1999. DiePersonalpronomenimZürichdeutschen. Lundergermanistische Forschungen,vol.63. Stokholm: Almqvist&WiksellInternational. Wiltschko, Martina. 1998. On the Syntax and Semantics of (Relative) Pronouns and Determiners. JournalofComparativeGermanicLinguistics, 2(2),143–181. Winter,Yoad. 2001. Flexibility Principles in Boolean Semantics: The Interpretation of Coordi- nation,Plurality,andScopeinNaturalLangauge. CambridgeMA:TheMITPress. Wurmbrand,Susi.2004.NoTP-frontingmeetsnearlyheadlessnick.UniversityofConnecticut. Wurmbrand,Susi.2006. LicensingCase. JournalofGermanicLinguistics, 18(3),175–236. BIBLIOGRAPHY 153 Zubizarreta, Maria Luisa, & Pancheva, Roumyana. 2017. A formal characterization of Person-basedAlignment: ThecaseofParaguayanGuaraní.NaturalLanguage&Linguistic Theory,Feb. Appendix Experiment1–PPCexperiment: Items (219) a. Ich vertraute den Vorschlägen meiner Internet-Datingseite nicht immer voll undganz. Abersohab ichmeinenPartnerkennengelernt. i. Ichwaretwasüberrascht,dasseinAlgorithmusihnmirvorgeschlagenhat. ii. Ichwaretwasüberrascht,dassihnmireinAlgorithmusvorgeschlagenhat. b. MeinPartnervertrautedenVorschlägenvonInternet-Datingseitenichtimmer vollundganz. Abersohatermichkennengelernt. i. Er war etwas überrascht, dass ein Algorithmus mich ihm vorgeschlagen hat. ii. Er war etwas überrascht, dass mich ihm ein Algorithmus vorgeschlagen hat. (220) a. AlsTemp-ArbeiterkannsichdieMariaihrenArbeitgebernichtaussuchen. i. Heutearbeitetsiefürmich,weileinAnzugträgersiemirgeliehenhat. ii. Heutearbeitetsiefürmich,weilsiemireinAnzugträgergeliehenhat. b. AlsTemp-ArbeiterkannichmirmeinenArbeitgebernichtaussuchen. i. Heute arbeite ich für die Maria, weil ein Anzugträger mich ihr geliehen hat. ii. Heute arbeite ich für die Maria, weil mich ihr ein Anzugträger geliehen hat. (221) a. Meine Freundin hat nun schon zwei Monate lang bei mir gewohnt. Sie wird abermorgenfüreinen MonatzuihrenElternziehen. i. Ichfreumichsuper,dassjemandsiemirabnimmt. ii. Ichfreumichsuper,dasssiemirjemandabnimmt. b. IchhabenunschonzweiMonatenbeimeinerMuttergewohnt. Ichwerdeaber morgenzumeinerFreundinziehen. 154 BIBLIOGRAPHY 155 i. MeineMutterfreutsichsuper,dassjemandmichihrabnimmt. ii. MeineMutterfreutsichsuper,dassmichihrjemandabnimmt. (222) a. IchbineinTollpatschundwaresauch,alsmeinkleineSchwestereinBabywar. i. MeineMutterhatmirerzählt,dasssiemirniemandentgegengehaltenhat. ii. MeineMutterhatmirerzählt,dassniemandsiemirentgegengehaltenhat. b. MeineTanteisteinTollpatschundwaresauch,alsicheinBabywar. i. Meine Mutter hat mir erzählt, dass mich ihr niemand entgegengehalten hat. ii. Meine Mutter hat mir erzählt, dass niemand mich ihm entgegengehalten hat. (223) a. DieWirtschaftboomt,eralsArbeitsloserbrauchteinenJobundichalsArbeit- geberbraucheArbeitskräfte. i. Eswirdbeidenbessergehen,wennjemandihnmirzuführenkann. ii. Eswirdbeidenbessergehen,wennihnmirjemandzuführenkann. b. Die Wirtschaft brummt, er als Arbeitsloser braucht einen Job und ich als Ar- beitgeberbraucheArbeitskräfte. i. Eswirdbeidenbessergehenwennjemandmichihmzuführenkann. ii. Eswirdbeidenbessergehenwennmichihmjemandzuführenkann. (224) a. IchunddieJohannasindzweigrundverschiedenePersonen. i. Ichkannnichtglauben,dassirgendwersiemirgleichsetzt. ii. Ichkannnichtglauben,dasssiemirirgendwergleichsetzt. b. IchunddieJohannasindzweigrundverschiedenePersonen. i. Ichkannnichtglauben,dassirgendwermichihrgleichsetzt. ii. Ichkannnichtglauben,dassmichihrirgendwergleichsetzt. (225) a. IchglaubefestandieLiebefürsLebenundhabjetztauchZeitsiezufinden. i. Ich muss mir meinen Lebenspartner schon selber suchen, weil niemand ihnmirsuchenwird. ii. Ich muss mir meinen Lebenspartner schon selber suchen, weil ihn mir niemandsuchenwird. b. IchglaubefestandieLiebefürsLebenhababerkeineZeitsiezufinden. i. MeinLebenspartnerwirdmichschonselbersuchenmüssen,weilmichihm niemandsuchenwird. BIBLIOGRAPHY 156 ii. MeinLebenspartnerwirdmichschonselbersuchenmüssen,weilniemand michihmsuchenwird. (226) a. IchkennedieDanielanunschonseiteinemJahr. IchhattevoreinemJahreinen schlimmenFahrradunfall,undsiewardieNotärztindiegleichzurStelle war. i. WasfüreinGlück, dassjemandsiemirgerufenhat. ii. WasfüreinGlück,dasssiemirjemandgerufenhat. b. IchkennedieDanielanunschonseiteinemJahr. SiehattevoreinemJahreinen schlimmenFahrradunfallundichwardieNotärztin,diegleichzurStelle war. i. WasfüreinGlück, dassjemandmichihrgerufenhat. ii. WasfüreinGlück,dassmichihrjemandgerufenhat. Experiment2–CLRexperiment: Items (227) a. DieSusiwilldenFritzkennenlernen. i. DieSusimöchtedaher,dassjemandsieihmvorstellt. ii. DieSusimöchtedaher,dasssieihmjemandvorstellt. b. DieSusiwilldenFritzkennenlernen. i. DieSusimöchtedaher,dassjemandihnihrvorstellt. ii. DieSusimöchtedaher,dassihnihrjemandvorstellt. (228) a. Die Eva trifft den Max nicht häufig. Der Max redet immer viel und die Eva hasstdasganzeGeschwafel. i. ZumGlückkannsichdieEvasichersein,dasskeinFreundsieihmausset- zenwürde. ii. ZumGlückkannsichdieEvasichersein,dasssieihmkeinFreundausset- zenwürde. b. Die Eva trifft den Max nicht häufig. Der Max redet immer viel und die Eva zeigtimmer,wiesieseineRedenhasst. i. Zum Glück kann sich der Max sicher sein, dass sie ihm kein Freund aus- setzenwürde. ii. Zum Glück kann sich der Max sicher sein, dass kein Freund sie ihm aus- setzenwürde. (229) a. DieKanzlerinmusssichumdasGemeinwohlkümmernundistkeinWirtschafts- bosswieMarkZuckerberg. BIBLIOGRAPHY 157 i. Merkelmusswissen,dassniemandsieihmgleichsetzenkann. ii. Merkelmusswissen,dassniemandsieihmgleichsetzenkann. b. DieKanzlerinmusssichumdasGemeinwohlkümmernundistkeinWirtschafts- bosswieMarkZuckerberg. i. FacebooksMarkZuckerbergmusswissen,dassniemandsieihmgleichset- zenkann. ii. FacebooksMarkZuckerbergmusswissen,dasssieihmniemandgleichset- zenkann. (230) a. Ichbinmirsicher,dassdieStefaniedenTomkennt. i. DieStefanieerrinertsichabernichtmehr,dassjemandsieihmschoneinmal gezeigthat. ii. DieStefanieerrinertsichabernichtmehr,dasssieihmjemandschoneinmal gezeigthat. b. Ichbinmirsicher,dassdieStefaniedenTomkennt. i. Der Tom errinert sich aber nicht mehr, dass jemand sie ihm schon einmal gezeigthat. ii. Der Tom errinert sich aber nicht mehr, dass sie ihm jemand schon einmal gezeigthat. (231) a. Teamworksollfrühgelerntwerden. Schülersuchensichaberimmerihrebesten Freunde für Gruppenarbeiten aus und arbeiten nie mit Schülern des anderen Geschlechts zusammen. Für die heutige Gruppenarbeit will die Lehrerin ein Experiment wagen. Die Schüler sollen nicht ihre eigenen Partner aussuchen sondern dies für einander tun. Jeder Schülerin soll ein Schüler ausgesucht werden. DerSebastianmagvondenMädchenabernurdieNathalie. i. DerSebastianhofft,dassjemandihnihraussuchenwird. ii. DerSebastianhofft, dassihnihrjemandaussuchenwird. b. Teamworksollfrühgelerntwerden. Schülersuchensichaberimmerihrebesten Freunde für Gruppenarbeiten aus und arbeiten nie mit Schülern des anderen Geschlechts zusammen. Für die heutige Gruppenarbeit will die Lehrerin ein Experiment wagen. Die Schüler sollen nicht ihre eigenen Partner aussuchen sondern dies für einander tun. Jeder Schülerin soll ein Schüler ausgesucht werden. DieNathaliemagvondenBubenabernurdenSebastian. i. DieNathaliehofft,dassjemandihnihraussuchenwird. ii. DieNathaliehofft, dassihnihrjemandaussuchenwird. BIBLIOGRAPHY 158 (232) a. Der Martin ist ein ganz ausgezeichneter Koch auf der Suche nach Arbeit und die Paula eine Braut auf der verzweifelten Suche nach einem Koch für ihre Hochzeit. i. DerMartinwarsehrzufrieden,dassjemandihnihrvorgeschlagenhat. ii. DerMartinwarsehrzufrieden,dassihnihrjemandvorgeschlagenhat. b. Der Martin ist ein ganz ausgezeichneter Koch auf der Suche nach Arbeit und die Paula eine Braut auf der verzweifelten Suche nach einem Koch für ihre Hochzeit. i. DiePaulawarsehrzufrieden,dassjemandihnihrvorgeschlagenhat. ii. DiePaulawarsehrzufrieden,dassihnihrjemandvorgeschlagenhat. (233) a. Für die gestrige Schicht hab ich dem Max die Petra zugeteilt. Die Petra war sichihresGlücksnochnichtbewusst. i. DerMaxahnteallerdingsschon,dassjemandihnihrzugeteilthat. ii. DerMaxahnteallerdingsschon,dassihnihrjemandzugeteilthat. b. Für die gestrige Schicht hab ich dem Max die Petra zugeteilt. Die Petra war sichihresGlücksnochnichtbewusst. i. DiePetraahnteallerdingsschon,dassjemandihnihrzugeteilthat. ii. DiePetraahnteallerdingsschon,dassihnihrjemandzugeteilthat. (234) a. Die Stefanie ist auf der Suche nach einer Tanzpartnerin. Der Patrick ist eine großartiger Tänzer hat aber schon eine Tanzpartnerin. Er wäre bereit mit der Stefaniezutanzen. i. DerPatrickglaubtaber,dassniemandihnihrüberlassenwird. ii. DerPatrickglaubtaber,dassihnihrniemandüberlassenwird. b. Die Stefanie ist auf der Suche nach einer Tanzpartnerin. Der Patrick ist eine großartiger Tänzer hat aber schon eine Tanzpartnerin. Er wäre bereit mit der Stefaniezutanzen. i. DieStefanieglaubtaber,dassniemandihnihrüberlassenwird. ii. DieStefanieglaubtaber,dassihnihrniemandüberlassenwird. Fillers (235) GesternbeimMarathonlaufhabichkeinenApfelabgekriegt. a. Mirwarnichtbewusstdasskeinermirihnentgegenhaltenwird. BIBLIOGRAPHY 159 (236) Ich bineinIndividuum. a. Jedemmussbewusstseindassihmmichniemandgebenkann. (237) Der Peter ist als besonders redsam bekannt und niemand möchte deswegen auf FeiernmitihmZeitverbringen. a. Eshatmichdaherüberrascht,dassmirihnjemandeineFreundausgesetzthat. (238) DerMaxwillnichtdassinseinemFirmenwagengerauchtwird. a. Ichhabegehört,dassihmeseinVorgesetzterverbotenhat. (239) DiePetrabrauchtdringensteinenHut. a. DieserHutschreitförmlich,dassjemandihnihraufsetzensoll. b. DieserHutschreitförmlich,dassjemandihnihraufsetzensoll. (240) EinereicheSammlerinistgekommenumeinenantikenTeppichzukaufen. a. DerTeppichistabernochnichtimvorführraum,obwohlihnihrjemandinfünf Minutenvorführensoll. b. DerTeppichistabernochnichtimVorführraum,obwohljemandihnihrjemand infünfMinutenvorführensoll.
Abstract (if available)
Abstract
Various languages have restrictions on the co-occurrence of pronominal arguments referred to in the linguistic literature as the Person Case Constraint (PCC). Swiss German as well as Standard German have been mentioned in the literature as special cases in the literature on the PCC. ❧ On the empirical side, a new data set is presented which is relevant and so far, remains unexplained in the PCC literature, namely the case of Swiss German ordering restrictions on reduced pronouns. This data set poses specific challenges because it effects elements that are clearly syntactic and the influence of structure on the PCC. ❧ On the theoretical side, I propose an analysis of the PCC in general, which provides a natural analysis for the Swiss and Standard German data presented.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Perspective in Turkish complementation
PDF
Superlative ambiguities: a comparative perspective
PDF
The morphosyntax of states: deriving aspect and event roles from argument structure
PDF
Building adjectival meaning without adjectives
PDF
Functional categories: the syntax of DP and DegP
PDF
Silence in answers: a study of ellipsis in Hindi
PDF
Towards the unity of movement: implications from verb movement in Cantonese
PDF
Copy theory of movement and PF conditions on spell-out
PDF
A reduplicative analysis of sentence modal adverbs in Spanish
PDF
Syntax-prosody interactions in the clausal domain: head movement and coalescence
PDF
The grammar of correction
PDF
Narrowing the focus: experimental studies on exhaustivity and contrast
PDF
Building phrase structure from items and contexts
PDF
The grammar of individuation, number and measurement
PDF
Subjectivity, commitments and degrees: on Mandarin hen
PDF
Reference time in the dynamics of temporal dependency in Korean
PDF
Decomposing Slavic aspect: the role of aspectual morphology in Polish and other Slavic languages
PDF
When things are left unsaid: existential and anaphoric implicit objects in discourse
PDF
The theory of empty noun in Chinese: with special reference to the right node raising construction
PDF
Syntactic and non-syntactic factors in reflexive pronoun resolution in Mandarin Chinese
Asset Metadata
Creator
Borer, Thomas Michael
(author)
Core Title
The case of a person: The person case constraint in German
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Linguistics
Publication Date
11/19/2018
Defense Date
05/08/2018
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
agreement,Alemannic,German,OAI-PMH Harvest,PCC,pronouns,Swiss German,syntax
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Zubizarreta, Maria Luisa (
committee chair
), Keine, Stefan (
committee member
), Pancheva, Roumyana (
committee member
), Simpson, Andrew (
committee member
), Uzquiano-Cruz, Gabriel (
committee member
)
Creator Email
thomasborer@gmx.ch
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-107839
Unique identifier
UC11675796
Identifier
etd-BorerThoma-6973.pdf (filename),usctheses-c89-107839 (legacy record id)
Legacy Identifier
etd-BorerThoma-6973.pdf
Dmrecord
107839
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Borer, Thomas Michael
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
agreement
Alemannic
PCC
pronouns
Swiss German
syntax