UNIVERSITY OF NOVA GORICA GRADUATE SCHOOL ON THE OPTIONALITY OF WH-FRONTING IN A MULTIPLE WH-FRONTING LANGUAGE DISSERTATION Petra Mišmaš Mentor: Assoc. Prof. Franc Marušič Nova Gorica, 2015 Abstract This thesis explores the fact that in Slovenian multiple wh-questions not all whphrases have to front. This suggests that multiple wh-movement in Slovenian is optional. The majority of the existing literature on multiple wh-fronting focuses on questions in which all wh-phrases have to move to clause initial positions, I, on the other hand, focus on optionality in multiple wh-questions. I show movement in Slovenian is not avoided because of phonological, syntactic or semantic restrictions that influence other languages (cf. Bošković 2002), and that the Principle of Distinctness (Richards 2010) does not account for all cases of optional multiple whfronting in Slovenian. Three types of multiple wh-questions in Slovenian are determined and analyzed: (i) questions in which all wh-phrases move to clause initial positions (i.e. questions with multiple wh-fronting), (ii) questions in which one wh-phrase has to be moved to a clause initial position and the rest undergo movement to a clause internal position (multiple wh-questions with short movement), (iii) questions in which at least one wh-phrase has to be moved to a clause initial position and the rest stay in situ (multiple wh-questions with wh-in-situ). Crucially, in all three types at least one wh-phrase has to move to a clause initial position for a question to receive a true question reading. I assume the Cartographic approach and propose an account of multiple whfronting in Slovenian in which one wh-phrase has to move to an Interrogative Projection (the clause initial position) in the Left Periphery while the remaining whphrases move to Wh-Projections in the Left Periphery, questions in (i), or the Low Periphery, questions in (ii). I propose that wh-phrases with a wh-feature undergo whmovement, which means that wh-movement is in fact obligatory in Slovenian. In questions of type (iii), wh-phrases that do not undergo movement are in fact bare whpronouns, which one also finds in polarity contexts, that are licensed by the interpretable Q+wh-feature located in the Interrogative Projection. Because the bare wh-pronouns do not come with a wh-feature, they do not have to move. I conclude that wh-movement in Slovenian only appears to be optional. Key words: multiple wh-fronting, short movement, optionality, Interrogative Projection, Left Periphery, Low Periphery, bare wh-pronouns, wh-in-situ i Povzetek: O neobveznosti večkratnega k-premika v jeziku z večkratnim k-premikom V disertaciji je kot primer jezika z večkratnim k-premikom, v katerem k-zveze niso nujno skladenjsko premaknjene, obravnavana slovenščina. To nakazuje, da je kpremik v slovenščini neobvezen. Ker so v preteklosti raziskovalci večino pozornosti namenili večkratnim k-vprašanjem, v katerih se premaknejo vse k-zveze, se ta disertacija osredotoča na neobveznost v večkratnih k-vprašanjih. Pri tem so kot mogoče pojasnilo za odsotnost premika izključene fonološke, skladenjske in semantične omejitve, ki pojasnijo odsotnost k-premika v drugih jezikih (gl. Bošković 2002). Hkrati je pokazano, da ima na k-premik v slovenščini vpliv načelo razlikovanja (Richards 2010), a da to ne pojasni vseh primerov neobveznega kpremika v slovenščini. Določeni in analizirani so trije tipi večkratnih k-vprašanj v slovenščini: (i) vprašanja, v katerih se vse k-zveze premaknejo na začetek stavka (tj. vprašanja z večkratnim k-premikom), (ii) vprašanja, v katerih se vsaj ena k-zveza premakne na začetek stavka, medtem ko se ostale premaknejo na pozicijo znotraj stavka (tj. večkratna k-vprašanja s kratkim premikom), (iii) vprašanja, v katerih se vsaj ena kzveza premakne na začetek stavka, medtem ko ostale k-zveze ostanejo na mestu (tj. večkratna k-vprašanja s k-zvezo na mestu). V vseh tipih k-vprašanj se vsaj ena kzveza premakne na prvo mesto v stavku. V nalogi privzemam kartografski pristop. Na podlagi tega je predlagana analiza večkratnega k-premika v slovenščini, v kateri se ena sama k-zveza premakne v vprašalno projekcijo (prvo mesto v stavku, InterP) v levi periferiji, medtem ko se preostale k-zveze premaknejo v k-projekcije (Wh-P), ki so bodisi v levi periferiji, vprašanja v (i), ali v nizki periferiji stavka, vprašanja v (ii). Predlagano je, da se tako premaknejo vprašalne k-zveze, ki imajo k-oznake, kar pomeni, da je k-premik v slovenščini dejansko obvezen. Na mestu ostanejo zgolj goli k-zaimki, ki jih lahko zasledimo tudi v pogojnih stavkih ali odločevalnih vprašanjih. Tovrstni zaimki so v vprašanjih v (iii) dovoljeni zaradi interpretabilne Q+k-oznake, ki se nahaja v vprašalni projekciji. Ker goli k-zaimki nimajo k-oznake, se ne premaknejo, kar ponovno kaže na to, da k-premik v slovenščini dejansko ni neobvezen. Ključne besede: večkratni k-premik, kratek premik, neobveznost, vprašalna projekcija, leva periferija, nizka periferija, goli k-zaimki, k-zveze na mestu ii Acknowledgements First and foremost I want to thank my advisor, Franc Marušič - Lanko. Hvala za vso pomoč, potrpljenje, komentarje, kritike, ideje in priložnosti, ki sem jih bila deležna v zadnjih letih. Hvala, hvala, hvala. For their helpful comments and suggestions I am sincerely grateful to the members of my committee: Lisa Lai-Shen Cheng, Iliana Krapova, and Athur Stepanov. I have spent many years at the University of Nova Gorica and there are a lot of people who I am thankful to. In the first group is everybody I had the pleasure of working with. Velik hvala Roku Žaucerju za vso pomoč in mnogo koristnih nasvetov. Za mnogo priložnosti hvala tudi Katji Mihurko Poniž. Velika zahvala tudi celotni stalni jezikoslovni zasedbi na Univerzi v Novi Gorici, ki jo danes sestavljajo: Franc Marušič - Lanko, Rok Žaucer, Arthur Stepanov, Penka Stateva in Sara Andreetta. Delo s temi ljudmi (mislim, da lahko prav vsakega opišem z mentsh) mi je prineslo vsaj toliko znanja kot sam študij, vsak dan odprlo kup vprašanj in vedno zbudilo tisto trmo, zaradi katere se vedno trudim delati več in bolje. Za vso podporo in ogromno prijaznih besed se zahvaljujem tudi ostalim sodelavcem na Fakulteti za humanistiko. Hvala, hvala, hvala. In the second group is everybody who was there during my years as a student at the University of Nova Gorica. In addition to Franc Marušič - Lanko, Penka Stateva, Arthur Stepanov, and Rok Žaucer, I want to also thank Slavica Kochovska for her many words of encouragement, Andrew Nevins for all the great ideas he shared with me and everybody else who was a part of my studies at UNG. Really, I am so happy that I had the chance to work with so many amazing people. I honestly do not know if this would be possible anywhere else. Thank you, thank you, thank you. Dolg je tudi seznam vseh, s katerimi je bil vsak dan v Rožni Dolini zanimiv in pozitiven. Še posebej hvala podstrešnikom (naj bo vaša senca dolga) in Ireni. Pa tudi študentom, ki so mi vedno dali kopico idej in vprašanj. Hvala, hvala, hvala. I am grateful to everybody who helped me with the data. Big thanks to everybody who helped me out with the surveys. (Velik hvala vsem, ki ste sodelovali v anketah.) I want to especially thank Adrian Stegovec (who helped me when I had doubts about Slovenian, hvala), Neda Todorović and Marija Runić for their Serboiii Croatin judgments, and Jitka Bartošová and Hana Strachoňová who helped me with the Czech data. So many thanks go to Calum Riach who helped me with my English. Also, many thanks to all of you for being a great group of people to discuss ideas, problems, issues (and True Detective, blankets, gossips, lemons, and all the remaining important stuff). I owe you all a lot of gratitude and beer. In all honesty, there are so many other people who should be included in the group of linguist buddies and who I should add to the list… But this would turn into a novel fast. So, dear you, I hope you know that you are one of them and that you are just great. Thank you, thank you, thank you. Hvala nejezikoslovnim prijateljem. Moniki (na še mnoge skupne obede in jake intelektualne razprave, cheerio), Ani (kuki, Islandija sigurno, zen in karma bosta pošlihtana, Godot pravi, da bo tam) in Poloni (ki mi ob pisanju tega poroča o dogodkih preteklega dneva, pri čemer ji odgovarjam z vsemi 18 črkami najine abecede). Mislim, da sem vam dolžna več, kot sploh veste – zagotovo pa kakšen par bobničev. Hvala Saši in Janji (kakšna polnočna tura s plavanjem po lužah, volnenimi nogavicami in tortelini bi se počasi prav prilegla). In spet, v to skupino spada še gora ljudi. In spet, dragi moji, upam, da veste, kdo ste, da sem vam hvaležna in da ste fajn. Hvala, hvala, hvala. In za konec še res velike reči. Hvala mojim staršem. Mislim, da je malo ljudi, ki jim je doma dano toliko podpore, kot sem jo od ranih nog dobivala jaz (še posebej jo cenim, ker vem, da je bila nudena kljub mojim občasno povsem neutemeljenim, nerazumljivim odločitvam). Hvala, ker sta mi pomagala na tej poti – upam, da vesta koliko mi vajina podpora pomeni. Hvala tudi sestri in njeni družini, ki so mi (poleg kopice drugih reči) družno dali enega izmed mojih najljubših nazivov – teta. No, pred petimi leti, sem v diplomski nalogi napisala tole: »Posebna zahvala še vsej moji družini. Tisti, ki je z mano vedno in povsod. Tisti, ki je na žalost že odšla. Tisti, ki je vmes prišla.« Vsaj to se med tem časom ni spremenilo. Hvala, hvala, hvala. (If you are reading this in English, you should know that my family is amazing and that I am really happy to have them.) In hvala Martinu. Ker si poslušal in ravno prav komentiral in nisi pobegnil in si me znal pomiriti bolj od česar koli na svetu in ker si tu, ko te rabim … in za vse tisto, kar je z besedami težko povedati, ker pomeni preveč. Hvala, hvala, hvala. (You should know that Martin is ah-maaaaa-zing.) iv I have spent many sleepless nights thinking about the thesis and (lately) about whom I need to thank. And I know that this list does not include everybody it should – blame it on my nonexistent memory. I guess thinking about questions made me forget about the facts. Still, one last thank you to everybody seems to be in place. Hvala. v Contents 1 Introduction .............................................................................................................. 1 1.1 Framework ........................................................................................................... 2 1.2 The data ............................................................................................................... 5 1.3 Organization of the thesis .................................................................................... 6 2 Optionality and obligatoriness of multiple wh-fronting .......................................... 8 2.1 What is optionality? ............................................................................................. 9 2.2 Multiple wh-fronting is not obligatory .............................................................. 13 2.2.1 Serbo-Croatian ............................................................................................. 23 2.2.2 Excluding Malagasy .................................................................................... 28 2.3 Slovenian - the data ........................................................................................... 35 2.3.1 Multiple wh-questions with two wh-phrases ............................................... 36 2.3.2 Questions with more than two wh-phrases .................................................. 41 2.3.3 Optionality in multiple wh-questions in Slovenian...................................... 44 3 Multiple wh-questions with multiple wh-fronting in Slovenian ............................ 46 3.1 Multiple wh-fronting - the [+/–MFS] analysis (Rudin 1988) ............................ 46 3.1.1 The [+/– MFS] analysis and Slovenian (Golden 1996a, b, 1997a).............. 53 3.1.2 Problems with the [+/–MFS] analysis and Slovenian .................................. 59 3.2 Multiple wh-fronting – ‘real’ and non-wh-movement (Bošković 1997a and after) .................................................................................................................... 64 3.2.1 Superiority as a diagnostic for Slovenian .................................................... 74 3.2.2 Problems with the focus analysis of multiple wh-fronting in Slovenian and beyond .............................................................................................................. 77 3.3 Cartography of multiple wh-questions............................................................... 87 3.3.1 Cartography and the Left Periphery in multiple wh-fronting languages ..... 87 3.3.2 Bulgarian ...................................................................................................... 89 3.3.3 Russian ......................................................................................................... 93 3.4 Word order in Slovenian multiple wh-questions ............................................. 100 3.4.1 The order of wh-objects and wh-subjects .................................................. 102 3.4.2 The order of wh-adjuncts with respect to wh-subjects ............................... 108 3.4.3 The order of wh-adjuncts with respect to wh-objects ................................ 115 3.4.4 Position of wh-phrases with respect to focused phrases and topics ........... 118 vi 3.4.5 Deriving the free word order of wh-phrases ............................................. 120 3.5 Obligatoriness of the clause initial wh-phrase ................................................ 132 3.6 Wh-movement ................................................................................................. 135 3.7 Summary ......................................................................................................... 139 4 Restrictions on multiple wh-fronting: when a wh-phrase must stay in situ ........ 140 4.1 Phonological restrictions ................................................................................. 140 4.2 Phonological restrictions in Slovenian............................................................ 143 4.3 Semantic restrictions ....................................................................................... 144 4.4 Semantic restrictions in Slovenian .................................................................. 146 4.5 Syntactic restrictions ....................................................................................... 148 4.6 Non-wh-islands as restriction on Slovenian .................................................... 149 4.7 The Principle of Distinctness as a restriction on multiple wh-movement....... 150 4.8 The Principle of Distinctness as a restriction on Slovenian ............................ 155 4.9 Summary ......................................................................................................... 162 5 Multiple wh-questions with a wh-phrase in situ in Slovenian............................. 163 5.1 Wh-in-situ as a consequence of the properties of wh-phrases ......................... 169 5.1.1 Wh-pronouns in Slovenian and other Slavic languages ............................ 170 5.1.2 Two readings of bare wh-pronouns ........................................................... 177 5.1.3 Assigning interrogative reading to wh-phrases in situ in Slovenian ......... 183 5.2 Further issues .................................................................................................. 189 5.2.1 Cases with a wh-subject in a clause final position .................................... 189 5.2.2 Two types of ‘what’ and other dialectical variation.................................. 191 5.2.3 Koliko ‘how much/many’ and zakaj ‘why’ ............................................... 196 5.2.4 Data from Serbo-Croatian ......................................................................... 199 5.3 Summary ......................................................................................................... 201 6 Multiple wh-questions with short movement ...................................................... 202 6.1 Short movement in Slovenian: Scrambling vs. wh-movement ....................... 209 6.2 Short movement as wh-movement to the Low Periphery ............................... 218 6.3 Clause initial wh-phrase and short movement ................................................ 224 7 How optional is multiple wh-movement in Slovenian ........................................ 227 7.1 Interpretation of multiple wh-questions in Slovenian ..................................... 227 7.1.1 The interpretation of multiple wh-questions with wh-in-situ .................... 228 7.1.2 Interpretation of wh-questions with short movement................................ 234 vii 7.1.3 Excluding cases with all wh-phrases in situ............................................... 236 7.2 Obligatory and non-obligatory wh-movement in Slovenian ........................... 238 7.2.1 Optionality and wh-in-situ ......................................................................... 239 7.2.2 Optionality and movement to the peripheries ............................................ 242 8 Conclusion ........................................................................................................... 246 9 References ............................................................................................................ 250 viii 1 Introduction Ever since the influential work on multiple wh-fronting by Bošković (1997c et seq.), many researchers have considered multiple wh-fronting to be obligatory (e.g. Zavitnevich-Beaulac 2005). And while some authors have observed non-uniform behavior of multiple wh-fronting in a variety of languages (e.g. Rudin 1988, Citko 2010 for Polish, Šimík 2010 for Czech), the phenomenon of optional multiple whfronting has been, for the most part, pushed aside. Most attention has been paid to ‘well-behaved’ multiple wh-fronting in which all wh-phrases move to the clause initial position and which has been an extremely popular topic since Rudin’s (1988) seminal paper. The goal of this thesis is twofold: First, this thesis aims to be a systematic overview of multiple wh-questions in Slovenian. Slovenian in general did not appear in the literature on multiple wh-fronting (with the exception of Golden 1996a, 1997a) and it therefore seems reasonable to explore the phenomenon and propose an account for it. I take Slovenian as the focal point because the initial work on this language has shown it to be a multiple wh-fronting language, but it was at the same time observed that multiple wh-fronting in Slovenian is not obligatory, in that not all wh-phrases have to move to the clause initial position – a property that was then explored no further. The second goal is related to optionality. I explore different varieties of multiple wh-questions in Slovenian which point toward the fact that Slovenian exhibits optional multiple wh-fronting. And since multiple wh-fronting has been considered to be obligatory, Slovenian multiple wh-questions present an interesting area of exploration that might help us understand the phenomena of multiple whfronting better. I will show that in multiple wh-questions in Slovenian three positions are possible for wh-phrases: a clause initial position (the Left Periphery of the clause), a clause internal position (the Low Periphery), or the wh-phrase stays in situ. But despite the availability of three positions, a common property of all multiple whquestions in Slovenian is that at least one wh-phrase needs to move to the clause initial position in order for the question to be interpreted as a true wh-question. This means that three types of multiple wh-questions can be found in Slovenian: (i) multiple wh-questions with multiple wh-fronting to the clause initial positions, (ii) 1 multiple wh-questions with (partial) short movement (to the Low Periphery), and (iii) multiple wh-questions with (partial) wh-in-situ. Furthermore, I will show that multiple wh-questions (ii) and (iii) are not a result of phonological, semantic, and syntactic restrictions, nor are they the result of Principle of Distinctness. Taken this into account, I will be considering three instances of multiple wh-questions within one language and I will be looking at each pattern of questions separately. In the thesis, I show that wh-fronting in Slovenian is obligatory, contrary to what the data seem to show at first. I will propose an account of multiple wh-fronting in Slovenian in which one wh-phrase has to move to an Interrogative Projection (the clause initial position) in the Left Periphery, which hosts the interpretable Q+whfeature (see Soare 2007), while the remaining wh-phrases move to Wh-Projections, which host the wh-features, and can be located inside the Left Periphery or the Low Periphery. Both also come with an EPP subfeature which means that movement of a wh-phrase with a wh-feature is obligatory. However, I additionally argue that in addition to interrogative wh-phrases with a wh-feature another type of wh-phrases exist. These wh-phrases can stay in situ in Slovenian and are in fact bare whpronouns that lack inherent quantificational force and are underspecified for the whfeature. Because they are underspecified, they do not undergo wh-movement. Still, they are interpreted as interrogative wh-phrases in multiple wh-questions in which one wh-phrase moves to the clause initial position. Because the underspecified whphrase is in the scope of the complex Q+wh-feature, the wh-phrase in situ is interpreted as being an interrogative phrase. 1.1 Framework While this thesis is based mainly on the research done in the Minimalist approach to syntax (Chomsky 1993 et seq.), I additionally adopt the Cartographic approach (for an overview see for example Belletti 2004: Introduction). Neither is considered a framework, but rather research programs based on the Principles and Parameters framework. And while tensions between the two research topics exist, the two are in fact not exclusive, as it was already recognized in the past (see Cinque & Rizzi 2008, 2 Shlonsky 2010). Even more, as Cinque & Rizzi (2008) note, the Cartographic approach is based on the Minimalist guidelines. Furthermore, each of the topics can be understood as having separate tasks and the two approaches can be used together to account for structure generation (a task for Minimalism) and outlining of the details of the structure (Cartography). In this sense these are “two research topics which can be pursued in parallel in a fully consistent manner, and along lines which can fruitfully interact […]” (Cinque & Rizzi 2008: 60). Such a division of labor and consequent interaction will be evident in the majority of thesis. I will not go into details of each topic at this point, but rather give the background when necessary. However, since wh-movement is in the center of this thesis, I will here briefly show how wh-movement is assumed to behave in the Minimalist program. In the Minimalist program, Move is taken to be a combination of Agree and Merge1 and is taken to be a last resort2 operation, that is, Move is chosen when nothing else is possible. An example of single wh-movement (taken to be movement of wh-phrases to SpecCP) is shown below. Example (1) shows the Agree relation between the head C0 and the wh-phrase. Example (2) shows movement of the whphrase to SpecCP. 1 In Chomsky (2004), Move is taken to be Internal Merge. In External Merge, the two elements that are merging are separate objects (for example α and β merge to form a new object γ = {α β}, while in Internal Merge one item is part of the other (as for example, α is a part of γ and the two can merge and form a new object δ). It is also stipulated that the element that undergoes movement leaves a copy in its place of origin (Chomsky 2000). The theory of movement is therefore known as the Copy Theory of movement. 2 In the Minimalist Program (Chomsky 1995), Last Resort is a principle that forces movement to happen only when necessary: (i) Move F raises F to target K only if F enters a checking relation with a sublabel of K. (Chomsky 1995: 280, (51)) 3 (1) C' wo C [uQ, WH, EPP] TP ei she T' 3 did eat vP 3 what [Q, uWH] (2) CP wo what C' [Q, u:WH:WH] wo C TP [uQ:Q, WH, EPP] 3 she T' 3 did vP 3 eat what Following Chomsky (2000), the wh-phrase has an uninterpretable wh-feature (that is: the feature’s value is unspecified) and an interpretable feature Q. These features match the features on the probe C that has an uninterpretable Q feature and an interpretable wh-feature. The probe C and the goal (wh-phrase) can agree because the wh-phrase is active, i.e. it has an uninterpretable feature. The wh-phrase is active until the uninterpretable wh-feature is matched (matching, i.e. agreement, happens when features are identical) and deleted. All this can be taken as the Agree part of movement, but we also need the second part – the Merge. In order for a wh-phrase to move, the head C0 also needs an EPP feature, which is, following Chomsky (2000), taken as an uninterpretable (nonsemantic) feature that influences the interpretation through the configuration it establishes. The EPP feature on C0 triggers necessary movement. I will return to the EPP and its role in optionality in section 2.1. In the Cartographic approach movement is understood in a similar way, or as Cinque & Rizzi (2008) put it for A’-movement in general: “A’-movement conforms to the general fact that movement is formally triggered by the featural constitution of a c-commanding head” (Cinque & Rizzi 2008: 62). This means that just as in 4 Minimalism, movement is feature driven. There is, however, a difference in the quantity of heads that act as probes. Specifically, the Cartographic approach assumes that there is a split of what Minimalism takes to be a single projection (for example the CP) into several projections, see for example Rizzi (1997) for the split CP approach. It is crucial for these functional projections that the head of each projection comes with its own morphosyntactic feature, which attracts a matching feature to the head’s specifier. Furthermore, in Cartography, all the features on the heads that are contained in the syntactic structure are interpretable. Also, just as in Minimalism, it holds that all syntactic representations should by the end contain only interpretable information which is visible at the interfaces, as Belletti (2004) notes for Cartography based on Chomsky (1995).3 This means that in Cartography “peripheral functional heads can be seen as overt “flags” carrying very transparent instructions to the interface systems on how their immediate dependents are to be interpreted” (Cinque & Rizzi 2008: 63). To summarize, while the two approaches have separate tasks, it holds in both that wh-movement is feature driven. I will assume this throughout the thesis, while other assumptions and background will be given when needed. 1.2 The data The thesis focuses on multiple wh-questions in Slovenian, and the relevant data is largely based on my own judgements and those of a few other informants. In addition to this I collected data from Slovenian speakers in two separate experiments. In both experiments, speakers were asked to fill out a questionnaire, but the two questionnaires were different because they were developed to check different facts and were based on different studies. In section 3.4 I report on the experiment with which I tested preferences in the word order of wh-phrases in multiple wh-questions 3 In Minimalism a computation converges at the interface level only if it is legible at the interface level (Chomsky 1995, 2000). This means that the structure only includes interpretable features (and no uninterpretable features, which are not legible): (i) The Interpretability Condition (Chomsky 2000: 113, (30b)) [Lexical Items] have no features other than those interpreted at the interface, properties of sound and meaning. 5 and in section 4.8 I report on an experiment in which I tested the influence of the Principle of Distinctness (Richards 2010) on Slovenian multiple wh-questions (this was also a part of a larger study that has tested the effects of the Principle of Distinctness on Sluicing and double accusative constructions in Slovenian, see Mišmaš (2013)). I give the details of each experiment in each of the two sections of the thesis. In addition to Slovenian data, I also report on data from other languages. I largely rely on data that has been reported in other studies but add judgements for additional data from Czech and Serbo-Croatian. The judgements were provided by two speakers of Czech and two speakers of Serbo-Croatian. 1.3 Organization of the thesis In the next chapter, chapter 2, I describe optionality in multiple wh-fronting languages based on the data from Slovenian as instances of multiple wh-questions in multiple wh-fronting languages in which not all wh-phrases have to move to the clause initial position but rather at least one wh-phrase has to move to the clause initial position while the rest can undergo short movement or stay in situ. I show that this phenomenon can also be found languages other than Slovenian (e.g. SerboCroatian) and show that some languages which were considered to have optional multiple wh-fronting (e.g. Malagasy in Sabel 2003) are, in fact, not optional multiple wh-fronting languages. The central issue in chapter 3 is multiple wh-fronting in Slovenian in which all wh-phrases move to the clause initial position. I argue against the analyses of multiple wh-fronting as proposed by Rudin (1988) and Bošković (1997a et seq.) and instead argue for an analysis of multiple wh-fronting in Slovenian in which whmovement proceeds to the extended Left Periphery of the sentence (the CP) and in which the word order of wh-phrases in the CP is free, but with some preferences (zakaj ‘why’ and kako ‘how’ precede other wh-phrases and [+human] subjects precede [+human] objects). A structure of the Left Periphery and an analysis of multiple wh-fronting are proposed. 6 Chapter 4 shows that phonological, semantic, and syntactic restrictions that apply to other multiple wh-fronting languages do not apply in Slovenian and therefore do not account for instances of optional multiple wh-fronting in Slovenian. In addition, it is shown that although the Principle of Distinctness (Richards 2010) does restrict multiple wh-fronting in Slovenian, it does not account for all instances of optional multiple wh-movement. In chapter 5, I look at multiple wh-questions with a wh-phrase in situ. In these at least one wh-phrase has to move to the clause initial position. Based on this and the fact that in Slovenian bare wh-pronouns can also be used in polarity contexts in which they are interpreted as indefinite pronouns, while they are always interpreted as interrogative in questions with a clause initial wh-phrase, I propose that instances of wh-in-situ in Slovenian multiple wh-questions are instances of underspecified bare wh-pronouns. In chapter 6, I show that multiple wh-questions with short movement are a phenomenon separate from multiple wh-questions with multiple wh-fronting to the clause initial position. Because of the parallel behavior of wh-phrases in questions with short movement and in multiple wh-fronting in Slovenian and Polish, I propose that short wh-movement proceeds to Wh-Projections in the Low Periphery. Chapter 7 shows that all types of multiple wh-questions in Slovenian receive the same interpretation. In addition, I discuss optionality of wh-in-situ and short movement in Slovenian and show that wh-in-situ in Slovenian is not an instance of optional wh-movement, because of the deficiency of the relevant wh-phrases. What seems to be optional is whether a wh-phrase undergoes wh-movement to the WhProjection in the High or Low Periphery. Still, wh-fronting to the clause initial Interrogative Projection in the Left Periphery in Slovenian is obligatory in true multiple wh-questions. Chapter 8 gives the conclusions of the thesis. 7 2 Optionality and obligatoriness of multiple wh-fronting Different languages exploit different mechanisms to produce wh-questions. For example, English moves at most one wh-phrase, while Japanese leaves all whphrases in situ. But there exists a group of languages in which wh-phrases can either move or stay in situ when forming wh-questions. Among these are, following Sabel (2006, but see also references cited therein), French, Bellunese, Greek, Spanish, Brazilian Portuguese, the Athabaskan languages Ancash Quechua, BabineWitsuwit’en, Navajo, West-Apache, Slave and Austronesian languages such as Malagasy, Malay, Tagalog and also Bantu languages like Duala, Kikuyu, Kinyarwanda, Tuki and Zulu, and Afro-Asian languages like Iraqi Arabic, Hausa and Coptic. Note, however, that this list includes both languages in which wh-ex-situ is achieved via wh-movement, as in Babine-Witsuwit’en (Denham 1997, 2000), or via a wh-cleft construction, for example Zulu (Sabel & Zeller 2006). With the exception of Malagasy, which some researchers claim to be a multiple wh-fronting language that exhibits optional multiple wh-fronting (cf. Sabel 2003) and to which I will return in section 2.2.2, multiple wh-fronting languages are missing from this list. I focus on multiple wh-fronting in this thesis, using Slovenian as an example of a multiple whfronting language which seems to exhibit optional movement. Slovenian is a multiple wh-fronting language (Golden 1997a). In Slovenian, all wh-phrases can move to the clause initial position. But this is not the only option for forming multiple wh-questions in this language. As an alternative to moving to the clause initial position, wh-phrases in a multiple wh-question can move to some clause internal position or stay in situ when at least one wh-phrase appears clause initially. Crucially, all three types of questions, shown below, can be used in the same context and have the same interpretation (I return to this in 7.1). (1) a. Koga je čemu znanstvenik izpostavil? who.ACC AUX what.DAT scientist ‘Who did the scientist expose to what?’ b. Koga je znanstvenik čemu izpostavil? c. Koga je znanstvenik izpostavil čemu? d. Čemu je koga znanstvenik izpostavil? 8 expose e. Čemu je znanstvenik koga izpostavil? f. Čemu je znanstvenik izpostavil koga? These examples suggest that multiple wh-fronting in Slovenian is not obligatory, but rather optional. I take optional multiple wh-fronting to mean that in multiple wh-questions not all wh-phrases have to move to the clause initial position. In fact, a more precise overview of some Slovenian data in section 2.3 shows that it is obligatory for at least one wh-phrase to move to the clause initial position. In this chapter, I focus on optionality in general and how it is exhibited in multiple wh-questions. I start by discussing the notion of optionality in theoretical terms and then turn to optionality in multiple wh-fronting languages to show that optionality of multiple wh-fronting is a wider phenomenon in that it can be found in other multiple wh-fronting languages as well. 2.1 What is optionality? The notion ‘optionality’ can be very intuitively described as: “a situation in which different ways of saying what seems to be the same thing show a clear correspondence in form” (Müller 1999: 3). With respect to movement we could, again intuitively, describe optionality as an ‘either/or’ situation – being optional then means ‘either you move or you do not’. But while the term itself is not problematic on an intuitive level, optionality is problematic from the Minimalist perspective since such semantically vacuous alternative operations are precluded from the grammar, as Roussou and Vlachos (2011) note based on Chomsky (1995). If they are indeed ruled out, then we should simply not find any cases of optional operations. But this is not the case, as I will show in this chapter. In fact, it turns out that two kinds of optionality exist: the first is the ‘minimalist’ optionality (‘minimalist’ in that it is predicted by the Minimalist program) and the second type, which has been termed ‘true optionality’ and is not predicted by the Minimalist program. According to the Minimalist program (see section 1.1) the operation Move is taken to be a combination of Agree and Merge and is taken to be a last resort operation (i.e. it only happens when nothing else is possible). Crucially for wh9 movement, in order for a wh-phrase to move to CP the head C0 needs an EPP feature, see section 1.1. This means that movement is necessarily triggered when the head C contains the EPP feature. The EPP feature is, following Chomsky (2000), taken as an uninterpretable (nonsemantic) feature which has an influence on the interpretation through the configuration it establishes4; it is a selectional category which seeks a phrase to merge with the category it heads. And while it is (perhaps) universally present on T, its presence for the phase heads v/P varies parametrically among languages and it is optional whether a language has it or not (Chomsky 2000).5 When on v or C, the EPP feature is a property of the phase: the head H of a phase PH may be assigned an EPP- and P-feature (where a P-feature is a feature of the peripheral system, like the Q-feature is for interrogative C) (Chomsky 2000: (24)). In this sense the EPP is an edge feature. In Minimalism, obligatoriness of movement, in our case wh-movement, is then related to the EPP feature – when the phase head has the EPP feature, movement is obligatory, as I show in example (2) in section 1.1. But an edge position required for an Internal Merge (i.e. Move) is optional. This optionality is determined in the lexicon. The edge position is made available for a head H by an EPP (which is also known as the OCC, i.e. occurrence feature) feature. Whether H has such a feature or not is determined in the lexicon (Chomsky 2004). Crucially: “Optimally, OCC should be available only when necessary, that is, when it contributes to an outcome at SEM that is not otherwise expressible - the basic Fox-Reinhart intuition about optionality. Hence H has OCC only if that yields new scopal or discourse related properties” (Chomsky 2004: 112).6 This means that in the Minimalist system optionality is pushed into the lexicon and is necessarily related to different interpretations, since the EPP provides new interpretations (new outcomes), while it, by default, provides different outcomes at PF. 4 See Yeo (2010) and the references cited therein on why the semantic contribution of the EPP is problematic. 5 I am not concerned with this type of optionality here, as this concerns variations between languages (that is, does a language have wh-movement or not). I am rather concerned about optionality within a single language. 6 For example Fox (1995): “[The] SSO [Scope Shifting Operation] can apply only if it yields a semantic interpretation which would be impossible without its application. In other words, SSOs are allowed only when necessary (as last resort) for achieving a designated interpretation” (Fox 1995: 284). 10 We can then imagine a situation where there are two distinct heads H in the lexicon – one with an EPP feature and one without. These two heads are then related to two different interpretations. Head H1 with an EPP will trigger movement and result in the interpretation A1, head H2 without EPP will not trigger movement and receive the interpretation B2 (see example (2) below). But the understanding of optionality in which optional movement always results in different interpretations is problematic, since there are examples of languages which display optionality, but in which optional movement is not related to different interpretations. Biberauer and Richards (2006) call this semantically vacuous optionality ‘true optionality’, see (3) below. The German examples in (2) show an instance of optionality in which interpretation of the sentence is affected by the position of oft ‘often’. Example (2a), in which ein Buch ‘a book’ follows oft, receives an interpretation in which ‘one often reads some book (but not necessarily the same book)’. Example (2b), in which oft is located after ein Buch ‘a book’, gets an interpretation in which a specific book is often read. The Afrikaans examples in (3), on the other hand, both receive the same interpretation, regardless of the position of het ‘has’ (i.e. in (3a) the auxiliary het ‘has’ is sentence final, in (3b) the auxiliary is in the second position/V2). This is problematic from the minimalist standpoint, since movement does not lead to a new interpretation. (2) German (Biberauer and Richars 2006: (1)) a. Er hat oft ein he has often a Buch gelesen book read ‘He often read a (non-specific) book.’ [weak reading; cf. Diesing 1992] b. Er hat ein he has a Buch oft gelesen book often read ‘There’s a book that he often read.’ [strong reading; cf. Diesing 1992] (3) Afrikaans (Biberauer and Richars 2006: (2)) a. Ek weet dat sy I know that she dikwels Chopin gespeel het. often Chopin played has ‘I know that she has often played Chopin.’ 11 b. Ek weet dat sy I know that she het dikwels Chopin gespeel. has often Chopin played ‘I know that she has often played Chopin.’ The question is then how to account for such semantically vacuous movement. Biberauer and Richards (2006) base their proposal on the distinction between optional and obligatory EPP features. As shown above, optional EPP features (the ones that are optionally assigned to phase heads) lead to different interpretations, but obligatory EPP features (the ones that are inherently assigned to a functional head, such as the EPP feature on T) must be present in the structure and do not influence LF. The ‘optionality’ which is exhibited with obligatory EPP features arises because of how the obligatory EPP feature is satisfied. An example of different ways to satisfy a feature is shown below for Russian wh-movement, where either the entire DP moves to satisfy the [+wh] feature on C, (4a), or only č’ju ‘whose’ does so, (4b). (4) Russian (Biberauer and Richards 2006: (28)) a. b. Č’ju knigu ty čital? whose book you read Č’ju ty whose you read book čital knigu? ‘Whose book did you read?’ The authors assume that in Russian the interrogative C head has an obligatory EPP feature and elements of the type č’ju ‘whose’ (and also kakoj ‘what’, ‘which’, kotoryj ‘which’, ‘what’) occupy the SpecDP position. Because of this they have the status of a maximal projection, so they can either move to CP on their own, as in (4b), or pied-pipe the entire DP to check the [+wh] feature, (4a). That is: the [+wh] feature needs to be checked and can be checked in two different ways and because this is an obligatory EPP feature, it does not matter in which way it is checked (see Biberauer & Richards 2006 for more on true optionality). Different types of movement in different languages exhibit optionality (scrambling, quantifier shift, object shift), but I will here focus only on optional 12 multiple wh-movement. In the next section I show initial data that indicate that multiple wh-fronting can be optional. 2.2 Multiple wh-fronting is not obligatory A review of the existing literature on multiple wh-fronting shows there are several different claims about optionality in multiple wh-fronting languages. I will take obligatory multiple wh-fronting to be instances of multiple wh-fronting in which all wh-phrases must move to the clause initial position under a true question reading.7 Early work on multiple wh-fronting takes the phenomena to be obligatory – except when the question appears in a special context. Wachowicz (1974), for example, gives data from Polish and Russian in which all wh-phrases appear in the clause initial position. Example (5) shows this for Polish: (5) Polish (Wachowicz 1974: (9)) Kto kogo who.NOM who.ACC budzi? wakes up ‘Who wakes up who(m)?’ On the other hand, Wachowicz (1974) also claims that in Polish a wh-phrase can stay in situ in a clarifying question. For example, (6) could be uttered in a situation in which the speaker knows that several people are doing several tasks and there have been proposals about the pairings of people and tasks – the speaker, however, wants a fixed plan, so he asks the question in (6). In addition, in Polish question in which not all wh-phrases move to the clause initial position, such as (7), is ungrammatical under a true question reading but can be interpreted as a yes/noquestion which contains an indefinite pronoun (Wachowicz 1974). 7 By true question reading I mean a non-echo-interpretation of a multiple wh-question (as in Bošković 2002). 13 (6) Polish (Wachowicz 1974: (15)) W końcu kto robi co? finally, who does what ‘Finally, who’s doing what?’ (7) Polish (Wachowicz 1974: (23)) * Co Monika what.ACC Monika.NOM komu dała? who.DAT gave # ‘What did Monika give to whom?’ ‘Did Monica give anything to anybody?’ Given these examples we could conclude that multiple wh-fronting to a clause initial position in Polish is obligatory in order to get a true question reading, since wh-in-situ and wh-movement8 to a non-initial position are acceptable only in special contexts or when they receive a special interpetation. Such a conclusion was adopted in Pesetsky (1987). Based on the examples from Wachowicz (1974), Pesestsky argues that only D-linked wh-phrases can stay in situ in Polish (as in (6), which he takes to be an example of D-linking), while non-D-linked wh-phrases move to an A’position. While he does not specify what the A’-position is, Pesetsky only gives examples in which all wh-phrases appear in a clause initial position. Again, the conclusion is that under a true question reading, all wh-phrases obligatorily front. Rudin (1988) shows that all wh-phrases obligatorily front in Bulgarian. For example: (8a), in which both wh-phrases are moved, is acceptable, but (8b) is ungrammatical because kŭde ‘where’ is not moved to a clause initial position (the same situation obtains if kŭde ‘where’ moves and koj ‘who’ is left in situ or if any of the wh-phrases just move within the embedded clause, cf. Rudin (1988: (6c–e)) also shown in section 3.1, example (8)). (8) Bulgarian (Rudin 1988: (6)) a. Koj kŭde misliš [če e who where think.2SG that has otišŭl _ _]? gone ‘Who do you think (that) went where?’ 8 I take wh-fronting and wh-movement to be the same phenomenon (cf. Dyakonova 2009). 14 b. * Koj misliš [če e otišŭl _kŭde]? Rudin (1988) concludes that all wh-phrases have to move to the clause initial position for a question to be grammatical in Bulgarian. On the other hand, more recently Pesetsky (2000) shows that in Bulgarian wh-questions with more than two wh-phrases, one wh-phrase can stay in situ: (9) Bulgarian (Pesetsky 2000: (40)) a. Koj na kogo kakvo s kakvo who to whom what with what napisa? wrote ‘Who wrote what to whom with what?’ b. ? Koj na kogo kakvo napisa s kakvo? [3 out of 4 move] c. ??? Koj na kogo napisa kakvo s kakvo? [2 out of 4 move] d. ** Koj napisa kakvo na kogo s kakvo? [1 out of 4 moves] He accounts for the facts by assuming that there are two different C heads in Bulgarian: a C1-Spec which only requires one wh-specifer and a Cm-Spec which requires more than one wh-specifier. Multiple wh-questions in Bulgarian are introduced with the latter and are additionally regulated by a general preference that all wh-phrases in a multiple wh-question move. When this preference is maximally satisfied, as in (9a), the question is completely acceptable. When no wh-phrases move, the question is unacceptable, and when two or three wh-phrases move, the sentences are “indistinguishably unnatural to a mild degree” (Pesetsky 2000: 21). Returning to Rudin (1988), she claims that wh-phrases can be left in situ in Serbo-Croatian9 and that questions of this type still get a non-echo reading, as in (10), but Polish and Czech can leave wh-words in situ only in echo questions, as shown in example (11) below for Czech. According to Rudin, (11a) is a ‘real’ multiple wh-question and (11b) is ungrammatical under the true question reading but is acceptable as an echo question: 9 In this thesis I use the term Serbo-Croatian as a cover term for Bosnian, Croatian, and Serbian, which are, as a group, also referred to as BCS (see Brown and Alt 2004, Section 0). I use SerboCroatian because this is the term typically used in the literature. When I cite examples and statements about Serbo-Croatian or any of the dialects belonging to the group, I will use the name that is used by the author I am citing. 15 (10) Serbo-Croatian (Rudin 1988: (106)) a. Ko je who has koga video? whom seen ‘Who saw whom?’ b. Ko je video koga? (11) Czech (Rudin 1988: (105)) a. Kdo koho viděl? who whom saw ‘Who saw whom?’ b. * Kdo viděl koho? Based on this we could conclude that multiple wh-fronting is only optional in Serbo-Croatian, while it is obligatory in Bulgarian, Czech and Polish. But this conclusion is challenged by Bošković (1997c), who claims that wh-phrases cannot remain in situ in either true or echo questions in Serbo-Croatian. This obligatoriness of wh-movement (regardless of whether it is motivated by a focus or a wh-feature, cf. section 3.2) holds for wh-adjuncts and wh-arguments in Serbo-Croatian. The only case where one wh-phrase does not move is when movement of all wh-phrases would lead to a violation of a restriction (which will be discussed in chapter 4). In addition, Bošković (2002) states that all wh-phrases must move in multiple wh-fronting languages and that this holds for real wh-fronting (as Bulgarian) and the non-whfronting languages (such as Russian, see Bošković (2002)), i.e. focus movement of wh-phrases is obligatory too.10 This obligatoriness is shown in examples below – 10 Based on the data available in the literature, in multiple wh-fronting languages, wh-phrases cannot stay in situ in single wh-questions either. (i) Serbo-Croatian (Bošković 1997c: (24)) ?* Jovan je kupio šta? Jovan.NOM AUX bought what.ACC ‘John bought what?’ (ii) Bulgarian (Bošković 2002: (16b)) ?* Ivan e kupil kakvo? Ivan.NOM AUX bought what ‘Ivan bought what?’ 16 cases in which a wh-phrase stays in situ are ungrammatical.11 Example (12) shows that wh-adjuncts cannot stay in situ, (13) shows this for wh-arguments, and (14) shows that multiple wh-questions in Serbo-Croatian are equally unacceptable regardless of the number of wh-phrases that do not move (cf. Bulgarian above). Bošković also claims that wh-fronting is obligatory in Polish, (15), Russian, (16), and Bulgarian, (17).12 (12) Serbo-Croatian (Bošković 1997c: (21)) a. Ko je gdje zaspao? who AUX where fallen-asleep ‘Who fell asleep where?’ b. * Ko je zaspao gdje? (13) Serbo-Croatian (Bošković 1997c: (11a)) a. Ko šta kupuje? who what buys ‘Who buys what?’ b. * Ko kupuje šta? (14) Serbo-Croatian (Bošković 1997a: (21)) a. Ko šta gdje who what where kupuje? buys ‘Who buys where what?’ b. * Ko kupuje šta gdje? c. * Ko šta kupuje gdje? 11 Bošković (1997a) takes this to be evidence for the claim that wh-movement (in his analysis, focus movement) is motivated by the strong focus feature in the wh-phrase that undergoes movement (if the strong feature is in the Agr, only one wh-phrase would need to move). This topic is discussed further in section 3.2. 12 Bošković (2002) does not mention Czech when discussing obligatoriness of multiple wh-fronting. Checking the acceptability of multiple wh-questions has shown that speakers find examples with multiple wh-fronting completely acceptable, but do not find multiple wh-questions in which one whphrase stays in situ completely acceptable or completely unacceptable. (i) a. b. c. Kdo co přinesl? who what brought ‘Who brought what?’ ? Kdo přinesl co? ? Co přinesl kdo? 17 d. * Ko gdje kupuje šta? (15) Polish (Citko 1998: (1)) a. Co Jan położył? gdzie what where Jan put ‘What did Jan put where?’ b. * Co Jan położył gdzie? c. * Jan położył co gdzie (16) Russian (Bošković 2002: (15)) a. Kto čto kupil? who what bought ‘Who bought what?' b. * Kto kupil čto? (17) Bulgarian (Bošković 2002: (14)) a. Koj kakvo e kupil? who what is bought ‘Who bought what?’ b. * Koj e kupil kakvo? Before I turn to other claims relating to obligatoriness of multiple wh-fronting that exist in the literature, I should note what has been said about obligatoriness of focus fronting. In section 3.2 I will show that instances of multiple wh-movement in which Superiority is violated have been analyzed as focus movement by Bošković (1997a et seq.). Bošković (2002) notes that speakers prefer focused non-wh-phrases to be moved and claims that we can ignore cases in which speakers do not front focused non-wh-phrases “since [they are] clearly dispreferred, perhaps for all speakers” (Bošković 2002: 356, fn. 7). He offers two possible accounts for this pattern, which he calls Variety I.13 The first account proposes that focused non-wh13 For example, as Dyakonova (2009) shows, colloquial Russian has the option of leaving a focused non-wh-phrase in situ. In the example below, the proper name Ljudmila is focused and can appear insitu, as in (i.c), clause initially as in (i.b) or after the subject, as in (i.a). This poses a problem for an analysis of wh-fronting as focus fronting in Russian, since it is not clear why focused non-wh-phrases can stay in situ while focused wh-phrases cannot. (i) Russian (Dyakonova 2009: 188, (11)) (Context: Ivan invited Olga for the dinner tonight, right?) 18 phrases can have either a strong or a weak focus feature – if they have the former they must move, if they have the latter, they stay in situ. Wh-phrases, on the other hand, always come with a strong feature (and therefore always move). The possibility of leaving non-wh-phrases in situ is then defined in the lexicon. The second account is based on the behavior of wh-phrases in Malay. In Malay, argument wh-phrases stay in situ (with no operator movement at all) or move, but wh-adjuncts must move, because they are uninterpretable in situ. A parallel can then be established between Malay and Variety I: wh-phrases in Variety I are like whadjuncts in Malay and are therefore uninterpretable when they are in situ. Focused phrases in Variety I, on the other hand, are like wh-phrases in Malay and can either move or not. Still, Bošković (2002) only proposes these accounts for focused nonwh-phrases and says nothing about the option of leaving a wh-phrase in situ in multiple wh-fronting languages, since, as we have seen above, Bošković takes whfronting to be obligatory. In fact, Bošković has often claimed that multiple whfronting is obligatory, for example: “[…] as we have seen above, wh-phrases in Slavic obligatorily undergo fronting independently of wh-movement” (Bošković 1998: 17), see also Bošković (1997c, 1999, 2002). Since he takes wh-fronting in certain contexts of Serbo-Croatian (see section 3.2) to be an instance of focus fronting, the obligatoriness to move wh-phrases is unexpected. The claim that multiple wh-fronting is obligatory (regardless of the motivation) has been assumed by many authors, for example Cheng (1991, based on Wachowicz 1974 and Rudin 1988), Stepanov (1998), Grohmann (2003), Stjepanović (2003), Meyer (2004), and Zavitnevich-Beaulac (2005). However, some authors have noticed that not all wh-phrases in multiple wh-fronting languages have to move to the clause initial position. This is obvious when one takes a closer look at languages such as Polish. For example, Dornisch (1998) shows that in Polish wh-phrases can be moved to the clause initial position. Alternatively, a wh-phrase can undergo ‘short a. b. c. Net, on LJUDMILU pozval. no he.NOM Ludmila.ACC invite.PST.MASC ‘No, he invited LUDMILA.’ Začem! LJUDMILU on pozval! no.way Ludmila.ACC he.NOM invite.PST.MASC Net, on pozval LJUDMILU. no he.NOM invite.PST.MASC Ludmila.ACC 19 wh-movement’, which Dornisch describes as follows: “wh-phrases other than the first one must indeed move but potentially only as far as the immediately preverbal position” (Dornisch 2000: 47, quote taken from Błaszczak & Fischer 2001: 72). Short movement in Polish has also been explored by Citko (2010), who shows that as in typical multiple wh-questions in Polish, there are also no Superiority effects in these cases. This is shown in (18) for a wh-argument and a wh-adjunct, while (19a) shows short wh-movement in Polish for arguments and (19b) shows movement to the clause initial position. In Citko’s analysis the lower wh-phrase moves to a position between TP and vP. As for leaving a wh-phrase in situ, according to Błaszczak and Fischer (2001), this can only happen in certain circumstances, that is with D-linking and in echo questions, (19c), in which the wh-phrase in situ is emphasized. (18) Polish (Citko 2010: (12)) a. Kiedy Ewa kogo odwidziła? when Eve.NOM who.ACC visited ‘When did Eve visit whom?’ b. Kogo Ewa kiedy odwidziła? (19) Polish (Błaszczak and Fischer 2001: (127)) a. Co by what.ACC COND-AUX Anna komu poleciła? Anna whom.DAT recommend ‘What would Anna recommend to whom?’ b. Co komu by Anna poleciła? c. Co by Anna poleciła KOMU? Based on Polish, Błaszczak and Fischer (2001) conclude that it is one of the ‘hard facts’ that in multiple wh-fronting languages all wh-phrases have to move, but that wh-phrases need not necessarily move to the clause initial position but rather at least to some preverbal position.14 As Błaszczak and Fischer note, the question is why the non-initial wh-phrases need to move at all and why this movement can be short. I will return to this question in chapter 6. 14 As I will show, in Slovenian multiple wh-questions not all wh-phrases have to move, i.e. all but one wh-phrase can stay in situ. 20 Short movement can also be found in Russian, as Grebenyova (2006b) shows. She calls this wh-fronting ‘partial wh-fronting’, but it in fact needs to be teased apart from what is typically known as partial wh-movement, see Fanselow (2006) for an overview of the phenomenon.15 I take (20b, c) to be instances of short wh-movement. (20) Russian (Grebenyova 2006b: (8)) a. Ivan podaril? Komu čto who.DAT what.ACC Ivan gave ‘Who did Ivan give what (as a present)?’ b. Komu Ivan čto podaril? c. Kogo Ivan gde uvidel? who.ACC Ivan where saw ‘Who did Ivan see where?’ A similar observation about multiple wh-questions in Russian has been made by Dyakonova (2009), who also claims that in multiple wh-questions the non-inital wh-phrase can either move to a clause initial position or it can only move to some preverbal position, as (21) shows. At the same time, a wh-phrase cannot stay in situ, as (22) shows. A similar observation about the availability of short movement has also been made by Liakin (2005), who notes that wh-phrases can be split by a topic and based on this proposes that wh-phrases can occupy several positions in Russian. (21) Russian (Dyakonova 2009: (6)) Čto otec komu what.ACC father.NOM who.DAT kupil? buy ‘What did the father buy for whom?’ 15 In typical ‘partial question’ constructions, Russian uses kak ‘how’ as a scope marker (and not what, as in many other languages). In addition, cases of partial wh-fronting are typically biclausal. For more on partial questions in Russian see, for example, Stepanov (2000). (i) Russian (Fanselow 2005: (17a)) Kak vy dumaete kogo ona how you think who she ‘Who do you think she loves?’ ljubit? loves 21 (22) Russian (Zavitnevich 2001: (6)) * Kto ty who you dumaesh pridet kogda? think arrives when Multiple wh-fronting does not behave uniformly in Czech either – examples from Veselovská (1993) show that non-initial wh-phrases can appear either before or after the subject (similarly to short movement in Polish) but that one wh-phrase needs to precede the clitic cluster. While Veselovská claims that a wh-phrase cannot stay in situ in Czech, Šimík (2010) shows that wh-in-situ is available in Czech in some contexts, (24).16 This again means that wh-fronting to a clause initial position is not obligatory in Czech. (23) Czech (Veselovská 1993: (7)) a. Co jsem já what am I komu kdy udělala? who.DAT when dome ‘What have I ever done to anybody?’ 16 Also, Czech speakers find multiple wh-questions with short movement or wh-in-situ acceptable. The latter goes against the claim that wh-phrases cannot stay in situ in Czech. (i) a. b. c. d. Koho komu Jan who.ACC who.DAT Jan.NOM ‘Who introduced Jan to whom?’ Koho Jan komu představil? Koho Jan představil komu? Komu Jan představil koho? představil? introduce This is also true for multiple wh-questions with more than two wh-phrases. Example (ii) shows that wh-phrases can stay in situ when one wh-phrase moves to the clause initial position (in this case kdo ‘who’). In these cases the order of wh-phrases matters. A wh-argument and a wh-adjunct cannot be placed in clause final position, (iii). (ii) (iii) a. Kdo představil komu koho? who.NOM introduce who.DAT who.ACC ‘Who introduced who to whom?’ b. ? Kdo představil koho komu? a. Co jesm já komu kdy udělala? what.ACC AUX me.NOM who.DAT when do ‘What have I when done to who?’ b. *? Co jesm ja udělala komu kdy? c. *? Co jesm ja udělala kdy komu? 22 b. Co jsem kdy komu já udělala? c. * Co jsem já udělala kdy komu? (24) Czech (Šimík 2010: (7)) a. Kdo co KOUPIL? who what bought ‘Who bought what?’ b. Kdo koupil CO? While there are several inconsistencies and contradicting claims in the literature about the obligatoriness of multiple wh-fronting in multiple wh-fronting languages, it seems that multiple wh-fronting is not universally obligatory. For languages such as Polish, Russian and Czech it is not the case that all wh-phrases must move to the clause initial position, however, it does seem that in Bulgarian and Romanian multiple wh-fronting is obligatory, since, to best of my knowledge, no claims about optional wh-fronting have been made about these languages. One language that has not been discussed as an optional multiple wh-fronting language in the literature thus far is Serbo-Croatian. Despite the claims about obligatoriness of multiple wh-fronting in this language, speakers’ judgements show that multiple whfronting is not obligatory in Serbo-Croatian either. 2.2.1 Serbo-Croatian To my knowledge, there have been no explicit claims about the availability of short wh-movement or the existence of optional multiple wh-fronting in Serbo-Croatian. However, wh-in-situ has been reported for cases when wh-movement would violate some kind of a restriction, see chapter 4. Still, there are attested examples which show that fronted wh-phrases can be divided by a focused phrase, (25),17 and that a wh-phrase can stay in situ, (26). 17 Given the analysis of multiple wh-fronting as focus fronting in Serbo-Croatian matrix questions (e.g. Bošković 2002, section 3.2 herein), this example could simply be interpreted as one in which both wh-phrases and the focus phrases front because of the [focus]-feature and no Superiority effects arise. 23 (25) Serbo-Croatian (Halupka-Rešetar 2013: (23a–c)) Znam ko je šta kupio PETRU. A ... ‘I know who bought what for Peter. But...’ a. ko je šta kupio who.NOM AUX what.ACC bought MARIJI? Mary.DAT ‘who bought what for Mary?’ b. ko je šta MARIJI kupio? c. ko je MARIJI šta kupio? (26) Serbo-Croatian (Zlatić, online) Ko je video šta? who AUX saw what? ‘Who saw what?’ Given examples such as these, we can predict that multiple wh-fronting is not obligatory for all speakers of Serbo-Croatian. As we will see below, this prediction is confirmed. Starting with multiple wh-questions with two wh-phrases in Serbo-Croatian, my two informants judge as completely acceptable examples in which only one whphrase moves, while the other stays in situ, such as (27), though one speaker noted that she prefers questions in which all wh-phrases move. That šta ‘what’ in (27a) is in fact left in situ can be concluded based on the ungrammaticality of (27b) in which ko ‘who’ appears in the clause final position. The availability of wh-in-situ contradicts the judgements given by Bošković (1998), according to whom all whphrases must move. In addition, Serbo-Croatian speakers allow short movement of the wh-phrase (i.e. movement to a position below the subject), (28). And while both word orders in (28) were judged as acceptable, one speaker preferred the word order in (28a). (27) a. Ko vidi šta? who sees what ‘Who sees what?’ 24 b. * Šta vidi ko?18 (28) a. Šta je Ivan kome dao? what.ACC AUX Ivan who.DAT give ‘What did Ivan give to whom?’ b. Kome je Ivan šta dao? Speakers give similar judgements with D-linked questions, as in (29). There are some preferences regarding the word order of different wh-phrases: for example, one speaker prefers the example in which the accusative wh-phrase stays in situ, but does not judge any of the three options given in (29) as ungrammatical. It has to be noted that movement of D-linked phrases is subject to restrictions on movement which will be discussed in section 4.3. (29) a. Koji dečko je nazvao koju devojku? which boy.NOM AUX called which girl.ACC ‘Which boy called which girl?’ b. Koju devojku je nazvao koji dečko? c. Koji dečko je koju devojku nazvao? These examples clearly show that multiple wh-fronting is not obligatory in Serbo-Croatian. This is further confirmed by questions with three wh-phrases. In these examples we also have two options. All wh-phrases can move to the clause initial position, as in (30), in any word order. This is expected if we assume multiple wh-fronting to be obligatory. (30) a. Ko je kome šta napravio? who.NOM AUX who.DAT what.ACC done ‘Who did what to whom?’ 18 When asked about this question, speakers suggest examples in (i) as alternatives. I take this as an indication that speakers prefer questions with multiple wh-fronting in Serbo-Croatian. (i) a. b. Šta ko who.NOM what.ACC ‘Who sees what?’ Ko šta vidi? vidi? see 25 b. Ko je šta kome napravio? c. Šta je ko kome/kome ko napravio? d. Kome je šta ko/ko šta napravio? But multiple wh-fronting is not obligatory here either. In these questions one wh-phrase can stay in situ, while two move to a clause initial position – the word order of fronted wh-phrases being free. Serbo-Croatian speakers also allow for questions in which two wh-phrases stay in situ. But for such questions speakers note that the order of wh-phrases is important and the direct object wh-phrase precedes the wh-phrase for the indirect object. (31) a. Šta je ko napravio what.ACC AUX who.NOM did kome? who.DAT ‘Who did what to whom?’ b. Ko je šta napravio kome? c. Ko je kome napravio šta? d. Kome je ko napravio šta? (32) a. Ko je napravio šta who.NOM AUX did what.ACC who.DAT kome? ‘Who did what to whom?’ b. ? Ko je napravio kome šta? In questions with three wh-phrases, short movement is available for one or two wh-phrases, (33). Also possible is a combination of short movement and wh-in-situ, (34). There is also an option of moving more than one wh-phrase to the clause initial position while one wh-phrase undergoes short movement, (35). Here, all word orders of wh-phrases are acceptable, but I am only giving some of the possible word orders.19 19 I use questions with a non-wh-subject to show the position of the wh-phrase that undergoes short movement. Since wh-adjuncts, such as kada ‘when’ are able to stay in situ or undergo short movement, I expect wh-adjuncts to behave just like wh-arguments. 26 (33) a. Kome je who.DAT AUX dao?20 Ivan kada šta Ivan when what.ACC give ‘What did Ivan give when to whom?’ b. (34) a. Kome je Ivan šta kada dao? Kome je who.DAT AUX Ivan kada dao šta?21 Ivan when give what.ACC ‘What did Ivan give when to whom?’ b. Kome je Ivan šta dao kada? (35) Kome who.DAT je šta Ivan kada dao? AUX what.ACC Ivan when give ‘What did Ivan give when to whom?’ And finally, a question such as (36) cannot get a true question reading, but rather only an echo question reading or a rhetoric question reading. (36) Ivan je kome Ivan AUX who.DAT šta kada dao? what.ACC when give # ‘When did Ivan give what to whom?’ ‘Ivan gave what, to whom, when?’ (echo) ‘When did Ivan ever give what to whom?’ Based on the examples above, we can conclude that Serbo-Croatian exhibits optionality in multiple wh-questions, a fact that has previously gone unnoticed. This is another confirmation of the existence of optional multiple wh-fronting in multiple wh-fronting languages. In the previous section we saw some observations about optional multiple whfronting in Polish and Czech, while this section has shown that multiple wh-fronting is, contrary to standard assumptions, not obligatory in Serbo-Croatian. In section 2.3 I will show this also holds for Slovenian. But before I explore the different types of multiple wh-questions in Slovenian, Malagasy needs to be discussed since this 20 This question can also get a rhetoric question reading (as in: When did Ivan ever give anything to anybody?!). 21 One speaker noticed that the clause final wh-phrase needs to be emphasized. 27 language has been described as having optional multiple wh-fronting (Sabel 2003) and this might help us account for optionality multiple wh-fronting in Slovenian. 2.2.2 Excluding Malagasy Malagasy has been described as having optional multiple wh-fronting (Sabel 2003). However, as will be shown, wh-phrases in Malagasy do not in fact undergo whmovement, but rather pseudo-clefting (Potsdam 2006). Because of this, I will not be considering Malagasy as an optional multiple wh-fronting language. Following Sabel (2003), Malagasy is an Austronesian VOS language in which wh-phrases can be moved or left in situ. In single wh-questions, only the adjunct or the argument phrase in the sentence final position (SpecIP) can move, as shown in (37), where the sentence final phrase is underlined. As we can see, either iza ‘who’ or inona ‘what’ can move, but only when they move from a sentence final position, hence the ungrammaticality of (37b), in which the argument moves from a sentence internal position. This is, following Sabel (2003), known as the ‘subjects only’ restriction. This restriction does not hold for adjuncts, as they can move, despite not being sentence final before movement, as shown in (38). (37) Malagasy (Sabel 2003: (3a), (3b), (4a)) a. Iza no manasa ny lamba amin’ ny savony who PRT wash the clothes with the soap t? ‘Who washes the clothes with soap?’ b. * Inona What no manasa t amin ny savony ny reny? PRT wash with the soap the mother amin’ ny savony ‘What does the mother wash with soap?’ c. Inona no sasan’ ny reny What PRT wash the mother with the soap ‘What does the mother wash with the soap?’ 28 t? (38) Malagasy (Sabel 2003: (6a)) Aiza no manasa ny lamba amin’ny savony ny reny? Where PRT wash the clothes with the soap the mother ‘Where does the mother wash the clothes with the soap?’ If we consider the wh-movement approach, wh-phrases can undergo partial, (40), or full wh-movement, (39), in embedded contexts. According to Sabel (2003), in all cases wh-movement is triggered by [+focus] feature and the difference between in situ and wh-movement is in the strength of the [+focus] feature. The feature is weak in in-situ constructions and strong in cases with full and partial wh-fronting. (39) Malagasy (Sabel 2003: (7)) [CP1 Inona no heverin-dRabe [CP2 what FOC believe-Rabe fa novidin-dRakoto t]]? that buy-Rakoto ‘What does Rabe believe that Rakoto has bought?’ (40) Malagasy (Sabel 2003: (8)) [CP1 Heverin-dRabe [CP2 believe-Rabe fa inona that what no novidin-dRakoto t]]? FOC buy-Rakoto ‘What does Rabe believe that Rakoto has bought?’ Non-subject wh-arguments stay in situ in Malagasy, as shown in (41). In addition, Sabel (2003) shows three conditions under which a wh-phrase in Malagasy has to move. The first is the presence of the focus particle no. When there is no focus particle, the wh-phrase must stay in situ, (42a), but when the focus particle is present, the wh-phrase has to move, (42b). (41) Malagasy (Sabel 2003: (8)) [CP1 Heverin-dRabe [CP2 PRES-TT-believe-Rabe fa nividy inona Rakoto t]]? that PAST-AT-buy what Rakoto any amin’ ny magazay Rabe? there in 29 shop Rabe ‘What does Rabe believe that Rakoto has bought?’ (42) Malagasy (Sabel 2003: (10a, b)) a. Mividy (*no) inona buy (* PRT what the ‘What does Rabe buy in the shop?’ b. Inona what *(no) vidin-dRabe PRT buy-Rabe any amin’ ny magazay? there in the shop ‘What does Rabe buy in the shop?’ The second condition applies to wh-words in SpecIP which cannot stay in situ, as shown in (43). This, following Sabel (2003), is due to the fact that wh-phrases are non-specific and in Malagasy no non-specific NPs can stay in SpecIP (regardless of whether the non-specific phrase is a wh-phrase or not, see also Potsdam (2006a)). Wh-phrases must therefore move from SpecIP. The last restriction holds for nonreferential adjuncts (how and why), which must always occur sentence initially, (44), while referential adjuncts (where and when) can appear either in situ or sentence initially. (43) Malagasy (Sabel 2003: (11)) a. * Novidin-dRabe (ny) inona? buy-Rabe (the) what b. Inona no novidin-dRabe? what PRT buy-Rabe ‘What has Rabe bought?’ (44) Malagasy (Sabel 2003: (15a), (17b)) a. * Nanasa wash ny lamba amin’ ny savony nahoana the clothes with the soap why ny reny? the mother ‘Why did the mother wash the clothes with soap?’ b. Nahoana no nanasa ny lamba why FOC wash amin’ the clothes with ny savony ny reny? the soap the mother ‘Why did the mother wash the clothes with soap?’ The examples above only show single wh-movement, but Malagasy also has multiple questions in which wh-phrases can appear in situ (cf. Sabel 2003) or display optional fronting, which is allowed for adjuncts, such as aiza ‘where’, and arguments. In multiple wh-fronting, all the conditions mentioned above hold, so multiple fronting with a subject and an adjunct is permitted, (45a), but movement of 30 both a subject and another argument is excluded (while the subject must move, another argument, which is not in SpecIP, cannot), as shown in (46). Also, as (45) shows, Superiority effects must be observed in Malagasy. (45) Malagasy (Sabel 2003: (48)) a. Aiza iza no mividy ny vary? where who PRT buy the rice ‘Where does who buy the rice?’ b. * Iza aiza no mividy ny vary? (46) Malagasy (Sabel 2003: (35)) a. * Iza inona who what b. * Inona no mividy? PRT buy iza no mividy? Sabel (2003) accounts for the data by proposing that wh-phrases form a cluster that subsequently moves.22 As already mentioned, wh-fronting in Malagasy is the result of a strong [+focus] feature. According to Sabel, some feature of the wh-word is attracted to the same feature in the head C, but first the feature on the wh-word attracts a similar feature on the other wh-word. This can be seen in (45a), in which the strong [+focus] feature of iza ‘who’ attracts the wh-phrase aiza ‘where’. They form a cluster, as shown in (47), and this cluster is attracted to the head C due to the [+focus] feature that is on that head. Additionally, Sabel assumes that only the C0 head has the strong [+focus] feature in cases in which only one wh-phrase in a multiple wh-question is fronted in Malagasy. In examples where no wh-phrases 22 This multiple wh-fronting is not the result of scrambling, as Malagasy does not allow scrambling, nor is it a result of movement to multiple specifiers, as, following Sabel (2003), this would predict the availability of long movement across short movement, which is not possible, (i.a). In these cases, the wh-phrase that would undergo long movement has to stay in situ, (i.b). (i) (Sabel 2003: (38a, c)) a. * Oviana no mihevitra i Piera [CP fa when FOC believe ART Piera that ‘When does Piera believe that who did return?’ b. Mihevitra i Piera [CP fa iza no niverina believe ART Piera that who PART return ‘When does Piera believe that who did return?’ 31 iza no niverina tt]? who FOC return oviana t] when move, neither the head C0 nor the wh-phrase has the [+focus] feature. I am leaving other details aside here, but see Sabel (2003) for more. (47) CP 3 C’ 3 C IP 3 I’ NP1 2 2 …t2… XP2 NP1 Sabel (2003) takes Malagasy to be similar to Bulgarian and Romanian since in all three languages multiple wh-fronting is available under similar conditions: the order of moved wh-phrases is fixed, nothing can intervene between the moved whphrases, and long distance wh-fronting is available. The difference between Malagasy and Bulgarian, according to Sabel (2003), is that the latter does not exhibit optional wh-fronting while Malagasy does so. Still, it needs to be emphasized that while it is true that more than one wh-phrase can appear in a sentence initial position in Malagasy, this only concerns examples in which the subject (i.e. the wh-phrase that originates in the SpecIP position) and a wh-adjunct ‘where’ or ‘when’ move. This means that multiple wh-fronting is much more restricted in Malagasy than in Bulgarian, where all wh-phrases move, regardless of being a subject, object, or adverb, which one should consider when classifying a language as a multiple whfronting language. In addition, Sabel’s claim that wh-movement in Malagasy is ‘optional multiple wh-fronting’ also seems to be too strong if we take optional movement to be an either/or operation. It is however true that wh-movement in Malagasy is subject to many restrictions based on which an item either moves or does not. What seems to be optional in Malagasy is whether the adjuncts oviana ‘when’ and aiza ‘where’ move. Because of this, I do not take multiple wh-fronting in 32 Malagasy to be comparable to wh-fronting in, for example, Bulgarian, nor do I consider the optional wh-movement in Malagasy comparable to that in Slovenian.23 In fact, as Potsdam (2004, 2006) shows, the properties of Malagasy multiple wh-fronting and the differences between Malagasy and Bulgarian can be accounted for if we adopt a pseudo-cleft analysis of optional wh-questions in Malagasy. Under the pseudo-cleft approach, single wh-questions in which a wh-phrase appears in a clause initial position are “covert pseudo-clefts in which the wh-phrase is a predicate and the remaining material is a headless relative clause in subject position” (Potsdam 2006: 2156). The structure for a simple matrix question with a wh-phrase in the clause initial position is shown in (48).24 (48) Malagasy (Potsdam 2006: (5b), (7)) a. Izza no nihomehy? who laugh PRT ‘Who laugh?’ b. [IP [predicate izza] [DP/headless rel. no Opi who nihomehy ti]] laugh lit. ‘The one that laughed is who?’ ‘Who laughed?’ Potsdam (2006) also argues against a multiple wh-fronting analysis of Malagasy multiple wh-questions. One reason for this is the observed anti-superiority effects. That is, in Bulgarian the wh-subject precedes the wh-adjunct (see section 3.1), but in Malagasy the wh-adjunct must precede the wh-subject, (45). The second argument comes from what can be fronted in Malagasy. While in Bulgarian any combination of wh-phrases can be fronted, in Malagasy multiple wh-fronting is restricted to wh-subjects and wh-adjuncts. Moreover, even multiple wh-questions with multiple fronted wh-adjuncts are ungrammatical in Malagasy: 23 For example, in Slovenian, optional wh-movement can concern arguments as well as adjuncts. The full range of optional multiple wh-fronting is shown in 2.3. 24 I will not go through the entire argumentation for the pseudo-cleft analysis here, but see Potsdam (2006 and the references cited therein) for the reasoning against a wh-movement analysis. 33 (49) Malagasy (Potsdam 2004: (28a, b)) a. * Taiza oviana where when no nividy mofo Rasoa? PRT buy.ACT bread Rasoa ‘When did Raosa buy the bread where?’ b. * Oviana taiza no nividy mofo Rasoa? Under the pseudo-cleft approach, multiple wh-questions in Malagasy can be accounted for if we assume Malagasy has the so called ‘bodyguard construction’, see Potsdam (2004; and the references cited therein). The term ‘bodyguard’ refers to the subject which can optionally appear between the fronted wh-adjunct (or a focused phrase) and the focus particle. The two possible positions of the subject (i.e. sentence final or sentence internal) in a wh-question are shown below in (50). Following Potsdam (2004), this ‘bodyguard’ appears in the SpecDP of the headless relative that appears in the subject position (50) Malagasy (Potsdam 2004: (22)) a. Aiza no mividy mofo Rasoa? where PRT buy.ACT bread Rasoa ‘Where does Rasoa buy bread?’ b. Aiza Rasoa no mividy mofo? The notion of a ‘bodyguard’ also helps us to account for instances of multiple wh-fronting. In these cases the ‘bodyguard’ is a wh-phrase. Example (51) shows the structure of example (45). This also accounts for the word order of fronted whphrases in Malagasy. (51) IP 3 I’ DP 3 3 I PredP DPk D’ 55 3 aiza iza D CP ‘where’ ‘who’ no 6 Opi nividy vary tk tk ‘buy rice’ 34 Based on the very restricted nature of multiple wh-fronting in Malagasy and the fact that optional movement in Malagasy only concerns wh-subjects and referential adjuncts, I do not consider Malagasy a language with optional multiple wh-fronting. I will assume Malagasy wh-phrases are moved in pseudo-cleft constructions, as proposed by Potsdam (2004, 2006). Because of this, I leave Malagasy aside from now on. In the next section I turn to the Slovenian data and show why Slovenian can be taken as an example of optional multiple wh-fronting language. 2.3 Slovenian - the data One of the central goals of this thesis is to establish that optional wh-fronting is a valid pattern in multiple wh-fronting. In order to achieve this, I will mainly focus on Slovenian. Because of this, I will first establish how optionality of multiple whfronting is displayed in Slovenian. In this section I show the typical patterns of multiple wh-questions in Slovenian. In establishing these patterns I look at D-linked and non-D-linked wh-phrases – the latter including both argument and adjunct whphrases. The possible patterns will be shown for matrix and embedded questions. I will start with questions with two wh-phrases, but will also look at wh-questions with three wh-phrases. 25 25 Optional multiple wh-fronting also has to be separated from partial wh-movement in Slovenian. I will not go into detail about partial wh-movement, but a couple of differences have to be mentioned. Given the typology of partial wh-movement (Fanselow 2006), Slovenian is a language with partial movement with scope marking by the most unmarked wh-word. In this respect Slovenian is similar to German, see example (i), Frisian, Serbo-Croatian, Czech, Romani, Hungarian, Finnish, and Warlpiri (see Fanselow 2006 and the references therein). (i) German (Fanselow 2006: (12a)) Was glaubst du wen what believe you who.ACC ‘Who do you believe that Irina loves?’ Irina Irina t liebt? loves In all these languages, scope is marked by the most unmarked wh-phrase, which is (in these languages) the word what. On the other hand, some languages (such as Polish) use how in the scope marking position. Slovenian, like German, uses the word kaj ‘what’, as (ii) shows. This unmarked whphrase does not move from the embedded clause. In this group of languages, the scope marking position in the matrix clause is filled by the most unmarked wh-word, while the wh-phrase moves only within the embedded clause. 35 2.3.1 Multiple wh-questions with two wh-phrases In wh-questions with two wh-phrases, three possible types of questions that receive a true question reading emerge: Both wh-phrases can move to the clause initial position (i.e. multiple wh-fronting), one wh-phrase moves to the clause initial position and one moves to some clause internal position (i.e. short movement), or else one wh-phrase moves to the clause initial position and one wh-phrase stays in situ. Multiple wh-fronting will be explored in chapter 3, but some properties need to be noted at this point: In Slovenian, all wh-phrases can move to the clause initial position of a question and in these cases there are no Superiority effects. This pattern holds both in matrix, (52), and in embedded questions, (53). Example (54) shows multiple wh-fronting for adjunct wh-phrases in matrix and (55) in embedded questions. Examples (56) and (57) illustrate that multiple wh-fronting is available also for D-linked wh-phrases in matrix and embedded questions, respectively. The latter is surprising, given that multiple wh-fronting of D-linked wh-phrases is said not to be possible in certain other multiple wh-fronting languages, such as SerboCroatian, and D-linking can in fact be viewed as a restriction on movement (cf. section 4.3). Note that all of the questions below can receive either a single pair or a pair list answer. Furthermore, multiple wh-movement out of embedded questions is not acceptable, as (58) shows, cf. section 3.1.2. (ii) Kaj misliš, koga ima Janez v mislih? what thinks who have Janez in thoughts ‘Who do you think Janez is thinking of?’ Two things stand out immediately: optionality can be displayed in both matrix and embedded questions in Slovenian (which I will show in this section) and when it is displayed in embedded questions, any wh-phrase can move to the clause initial position of the matrix clause, not only kaj ‘what’. The availability of optional multiple wh-fronting in matrix questions and the fact that any whphrase can be optionally fronted shows that partial wh-movement and optional wh-movement need to be looked at as two distinct phenomena and that we cannot account for the data from optional multiple wh-questions with a partial wh-movement analysis. 36 (52) a. Kdo je kaj kupil? who.NOM AUX what.ACC buys ‘Who buys what?’ b. (53) a. Kaj je kdo kupil? Janez sprašuje, kdo je kaj kupil. Janez asks who.NOM AUX what.ACC buy ‘Janez is asking who bought what.’ b. (54) a. Janez sprašuje, kaj je kdo kupil. Kje je kdaj razprodaja oblek? where AUX when sale clothes ‘Where are the clothes on sale when?’ b. (55) a. Kdaj je kje razprodaja oblek? Janez sprašuje, kje je kdaj razprodaja oblek. Janez asks AUX when sale where clothes ‘Janez is asking where the clothes are on sale when.’ b. (56) a. Janez sprašuje, kdaj je kje razprodaja oblek. Katera punca je katerega fanta povabila na ples? which girl.NOM AUX which boy.ACC invite to dance ‘Which girl invited which boy to the dance?’ b. (57) a. Katerega fanta je katera punca povabila na ples? Janez sprašuje, katera punca je katerega fanta Janez asks which girl which boy povabila na ples. invite to dance AUX ‘Janez is asking which girl invited which boy to the dance.’ b. Janez sprašuje, katerega fanta je katera punca povabila na ples. (58) a. * Kdo koga who.NOM who.ACC povabil plesati? invite dance.INF Miha sprašuje Miha.NOM asks Majo, Maja.ACC if b. * Koga kdo Miha sprašuje Majo, če je povabil plesati? 37 če je AUX As examples above show, multiple wh-fronting is a valid option for multiple whquestions in Slovenian.26 Questions with short movement, similar to the ones in Polish (cf. Citko 2010) can also be found in Slovenian. In this pattern, the lower wh-phrase is located after the subject (I only consider questions with non-wh-subjects here, but see chapter 6 for other examples). As in the previous pattern, this holds for argument wh-phrases in matrix, (59), and embedded questions, (60), and adjunct wh-phrases in matrix, (61), and embedded questions, (62). (59) a. Kaj je Janez kupil? what.ACC AUX Janez.NOM who.DAT komu buy ‘What did Janez buy for whom?’ b. (60) a. Komu je Janez kaj kupil? Tone sprašuje, kaj je Janez kupil. Tone asks what.ACC AUX Janez.NOM who.DAT komu buy ‘Tone is asking what Janez bought for whom.’ b. (61) a. Tone sprašuje, komu je Janez kaj kupil. Kdaj je Janez kam šel na počitnice? when AUX Janez.NOM where go on holiday ‘When did Janez go where on holiday?’ b. (62) a. Kam je Janez kdaj šel na počitnice? Tone sprašuje, kdaj je Janez kam šel na počitnice. Tone asks when AUX Janez.NOM where go on holiday ‘Tone is asking where Janez went on holiday where?’ b. Tone sprašuje, kam je Janez kdaj šel na počitnice? 26 Slovenian clitics are second position clitics and always follow the first syntactic constituent of the clause (e.g. Golden and Sheppard 2000; Marušič 2009): (i) a. b. c. Vesel mlad pek je spekel torto. happy young baker AUX bake cake ‘A happy young baker baked a cake.’ * Vesel je mlad pek spekel torto. * Vesel mlad je pek spekel torto. 38 Again, the pattern can also be found with D-linked phrases. Their word order is free. (63) a. Kateri učiteljici je Maja katero darilo prinesla? which teacher.DAT AUX Maja which present.ACC bring ‘To which teacher did Maja bring which present?’ b. (64) a. Katero darilo je Maja kateri učiteljici prinesla? Ravnatelj sprašuje, kateri učiteljici je Maja headmaster asks which teacher.DAT AUX Maja.NOM which darilo katero prinesla. present.ACC bring ‘The headmaster is asking to which teacher did Maja bring which present?’ b. Ravnatelj sprašuje, katero darilo je Maja kateri učiteljici prinesla. Questions with short movement will be explored in chapter 6. The final form of multiple wh-questions with two wh-phrases which receive a true question reading in Slovenian are questions in which one wh-phrase moves to the clause initial position and one is left in situ. This is again grammatical in matrix and embedded questions, as (65) and (66) show for argument wh-phrases and (67) and (68) show for adjunct wh-phrases. This pattern can also be found with D-linked wh-phrases. (65) a. Kaj je Miha kupil komu? what.ACC AUX Miha buy who.DAT ‘What did Miha buy for whom?’ b. (66) a. Komu je Miha kupil kaj? Janez sprašuje, komu je Ana kupila kaj. Janez asks who.NOM AUX Ana buy what.ACC ‘Janez is asking what Ana bought for whom.’ b. Janez sprašuje, kaj je Ana kupila komu. 39 (67) a. Kdaj je Janez šel kam? when AUX Janez went where ‘When did Janez go where?’ b. (68) a. Kam je Janez kdaj šel? Zanima me, kdaj je Janez šel wonder I.DAT when AUX Janez went where kam. ‘I wonder when Janez went where.’ b. (69) a. Zanima me, kam je Janez šel kdaj. je Kateremu sorodniku which relative.DAT Maja predstavila katerega fanta? AUX Maja introduce which boy.ACC ‘To which relative did Maja introduce which boy?’ b. (70) a. Katerega fanta je Maja predstavila kateremu sorodniku? Zanima me, kateremu sorodniku wonder I.DAT which katerega fanta. which boy.ACC je Maja predstavila relative.DAT AUX Maja introduce ‘I wonder, to which relative Maja introduced which boy.’ b. Zanima me, katerega fanta je Maja predstavila kateremu sorodniku. This last group of questions, (65)–(70), shows that it is not the case that wh-phrases need to move to a preverbal position, which Błaszczak and Fischer (2001) took to be one of the crucial properties of multiple wh-fronting. However, if we compare the last group of questions, which I will return to in chapter 5, to questions with multiple wh-phrases which all stay in situ, we can see that it is necessary for one wh-phrase to move to a clause initial position (while the rest can stay in situ or undergo some type of movement) in order to get a true question reading. Questions in which all wh-phrases stay in situ are not acceptable under a true question reading. They are, however, acceptable as echo questions or yes/noquestions. When komu ‘who’ and kaj ‘why’ are pronounced without an emphasis and a rising intonation, as in (71), the example gets a yes/no-question interpretation. The yes/no-question interpretation is available as all interrogative wh-phrases are homophonous with indefinite pronouns. If the wh-phrases are emphasized, the question is interpreted as an echo question. 40 (71) Miha je kupil Miha AUX bought KOMU KAJ? who.DAT what.ACC #‘What did Miha buy for whom?’ ‘Did Miha buy something for someone?’/’Miha bought WHAT FOR WHOM?’A common property of multiple wh-questions in Slovenian is that at least one whphrase has to move. Crucially, all three word orders in (72) receive the same true question interpretation in Slovenian. (72) a. Kaj je komu Janez kupil? what.ACC AUX who.DAT Janez buy ‘What did Janez buy for whom?’ b. Kaj je Janez komu kupil? c. Kaj je Janez kupil komu? Before accounting for all multiple wh-questions with a true question reading, I turn to questions with more than two wh-phrases to get a clearer picture of the multiple wh-questions in Slovenian. 2.3.2 Questions with more than two wh-phrases In order to get a complete idea about the behavior of multiple wh-questions in Slovenian, we need to include questions with more than two wh-phrases. The first option that is available in this type of questions is that all wh-phrases move to a clause initial position. Again, there are no Superiority effects in these questions. Example (73) shows this possibility for matrix questions with argument and adjunct wh-phrases and (74) shows multiple wh-fronting of wh-arguments in an embedded question. (73) a. Kdo je koga kam povabil? who.NOM AUX who.ACC where invited ‘Who invited whom where?’ 41 b. Kdo je kam koga povabil? c. Koga je kdo kam/kam kdo povabil? d. Kam je kdo koga/koga kdo povabil? (74) a. Rok sprašuje, Rok.NOM asks kdo je kaj who.NOM AUX what.ACC who.DAT komu kupil. buy ‘Rok is asking who bought what for whom.’ b. Rok sprašuje, kdo je komu kaj kupil. c. Rok sprašuje, kaj je kdo komu/komu kdo kupil. d. Rok sprašuje, komu je kdo kaj/kaj kdo kupil. This means that multiple wh-fronting is available with three (and even more) whphrases.27 Questions with short movement are again available. As example (75) shows, one wh-phrase moves to a clause initial position, while the rest of the wh-phrases move only to a lower position, beneath the subject. Different combinations are also possible: two wh-phrases can move to the clause initial position and only the remaining one undergoes short movement, (76). There are no Superiority effects in these examples. While I am only giving an example of matrix questions with non-Dlinked wh-phrases, such short movement is also possible in embedded questions and with D-linked phrases. (75) a. Kaj je Maja what.ACC AUX Maja.NOM when who.DAT kdaj komu povedala? tell ‘When did Maja tell what to whom?’ b. Kaj je Maja komu kdaj povedala? c. Komu je Maja kaj kdaj/kdaj kaj povedala? d. Kdaj je Maja kaj komu/komu kaj povedala? 27 Multiple wh-fronting can also be found with three D-linked wh-phrases. Questions such as (i) are, however, interpreted as clarifying questions (which is not surprising, since the wh-phrases in it refer to entities that are given in the discourse). While there is no Superiority in these questions, I only give one word order below. (i) Katera punca je kateremu fantu katero darilo which girl.NOM AUX which boy.DAT which gift.ACC ‘Which girl brought which gift for which boy?’ 42 prinesla? bring (76) a. Kaj je komu Maja kdaj povedala? what.ACC AUX who.DAT Maja.NOM when tell? ‘When did Maja tell what to whom? b. Komu je kaj Maja kdaj povedala? c. Kaj je kdaj Maja komu povedala? d. Kdaj je kaj Maja komu povedala? e. Komu je kdaj Maja kaj povedala? f. Kdaj je komu Maja kaj povedala? As expected, with more than three wh-phrases, any combination of the two patterns is possible and one could potentially ask any of the following questions in (77). The word order of the wh-phrases is again free. (77) a. Koga je kdaj Maja komu kje predstavila? who AUX when Maja whom where introduce ‘Who did Maja introduce to whom, when and where?’ b. Koga je Maja komu kdaj kje predstavila? c. Koga je kdaj komu Maja kje predstavila? In addition to multiple wh-fronting and short movement, wh-in-situ is also possible in questions with three wh-phrases (or more). In questions with more than two wh-phrases, one, (78a), or more, (78b), wh-phrases can stay in situ. In these questions, certain restrictions seem to hold. Examples with two wh-phrases in situ are more acceptable when both wh-phrases are arguments. In these cases word order plays a role. I take the verb izpostaviti ‘expose’ in which we are questioning both the accusative and the dative object. With izpostaviti ‘expose’ the objects always appear in a fixed order in which the accusative precedes the dative (Marvin and Stegovec 2012). This order is also preferred with wh-phrases. (78) a. Kdo je čemu izpostavil koga? who.NOM AUX who.DAT expose ‘Who exposed whom to what?’ b. Kdo je izpostavil koga čemu? 43 who.ACC c. ?? Kdo je izpostavil čemu koga? If a wh-adjunct is clause final, the wh-question is less grammatical. The question seems less acceptable when more wh-phrases appear in the clause final positions. (79) a. ?? Kaj what.ACC je Ana kupila komu kdaj? AUX Ana buy who.DAT when Intended: ‘What did Ana buy for whom and when did she do it?’ b. ?? Kaj je Ana kupila kdaj komu? c. ?? Komu je Ana kupila kaj kdaj/kdaj kaj? d. ?? Kdaj je Ana kupila komu kaj/kaj komu? e. ? Kaj je komu Ana kupil kdaj? This means that again all three patterns of multiple wh-questions can be found in questions with three or more wh-phrases. It should come as no surprise that the combination of ‘move at least one’, ‘move behind the subject’ and ‘clause final’ can be combined in a single question, as shown below: (80) Kdaj je when AUX Miha kaj kupil komu? Miha what.ACC buy who.DAT ‘When did Miha buy what for whom?’ This question receives a true question reading. It can be used in the same contexts as a multiple wh-fronting question. This indicates that multiple wh-fronting is optional in Slovenian. 2.3.3 Optionality in multiple wh-questions in Slovenian In Slovenian multiple wh-questions, a wh-phrase can appear in any of three distinct positions: clause initial, clause internal (beneath the subject) and clause final. Crucially, it holds that at least one wh-phrase has to move to a clause initial position 44 for a question to receive a true question reading. This gives us three types of questions: (i) multiple wh-questions with multiple wh-fronting to the clause initial positions, (ii) multiple wh-questions with (partial) short movement (to the Low Periphery), and (iii) multiple wh-questions with (partial) wh-in-situ. Based on the Slovenian data, I define optionality in multiple wh-fronting languages as instances of multiple wh-questions in multiple wh-fronting languages in which not all wh-phrases have to move to the clause initial position but rather at least one wh-phrase has to be moved to the clause initial position while the rest have the option to undergo movement to the clause initial position, to undergo short movement, or to stay in situ. However, this means that at least one wh-phrase is obligatory moved in Slovenian. In what follows I will explore each type of questions separately while focusing mainly on the data from Slovenian. In the next chapter I will explore questions with multiple wh-fronting in Slovenian. 45 3 Multiple wh-questions with multiple wh-fronting in Slovenian Multiple wh-fronting has been a topic of exploration since the 1980’s. In the last decades several accounts of the phenomena have emerged. In the first part of this chapter, I go through two of the more influential theories of multiple wh-fronting. The focus of section 3.1 is Rudin’s (1988) analysis of multiple wh-fronting, and in 3.2 I focus on the analysis in Bošković (1997a et seq.)28. I then show how these analyses can be applied to Slovenian. At the same time I point out the most important properties of multiple wh-fronting languages and concepts important for the presented research. In section 3.2.2 I discuss why the analysis of Slovenian as a language without real wh-movement is problematic and show why mainstream views on Slovenian multiple wh-fronting should be reanalysed. Finally, I turn to the Cartographic approach to multiple wh-fronting and show how it can be applied to Slovenian. 3.1 Multiple wh-fronting - the [+/–MFS] analysis (Rudin 1988) With respect to wh-movement, languages can be divided into three groups. Some languages do not move wh-phrases but rather leave them in their original positions. These languages constitute the first group of languages, the so called wh-in-situ languages, such as Chinese and Japanese, (1). The second group of languages move only one wh-phrase, such as English, (2), and the third group is comprised of languages that can front more than one wh-phrase – the so called multiple whfronting29 languages, (3). Among the multiple wh-fronting languages we find different Slavic languages, all of which have in common the fact that they can front all the wh-phrases in a sentence. This work focuses on the last group of languages. 28 The issue of multiple wh-fronting and related phenomena in Serbo-Croatian and other languages is discussed in Bošković (1994, 1997a, 1997b, 1997c, 1998, 1999, 2000, 2002, 2007). 29 I will be using the term multiple wh-fronting for all the cases in which more than one wh-phrase moves (regardless of the motivation for movement). 46 (1) Japanese (Richards 1997: (2)) Taroo-ga dare-ni nani-o ageta no? Taroo.NOM who.DAT what.ACC gave Q ‘Who did Taroo give what?’ (2) English a. Who bought what? b. * Who what bought? (3) Serbo-Croatian a. Ko koga vidi? who.NOM whom.ACC sees ‘Who sees whom?’ b. Koga ko vidi? Multiple wh-fronting has been a widely researched topic since the 1980’s (with some initial work done in the 1970’s), with much of the work appearing after Rudin’s (1988) paper on multiple questions.30 This paper is to this day one of the most influential papers on multiple wh-fronting and is still used by many as the basis for analyses of phenomena related to multiple wh-fronting. Rudin (1988) shows that, despite superficial similarities (i.e. fronting all whphrases to the beginning of a clause), multiple wh-fronting languages can be divided into two groups that differ with respect to the order of fronted wh-phrases (Superiority effects), possibility of multiple extraction of wh-phrases out of embedded tensed clauses, wh-island effects, and whether fronted wh-phrases form a constituent. Examples (4) and (5) show different behavior with respect to Superiority. Example (4) shows a strict word order of fronted arguments in Bulgarian in which the subject must precede the object. This is an indicator that Bulgarian exhibits Superiority effects (this is similar to English, where the subject wh-phrase must be fronted in sentences with subject and object wh-phrases). On the other hand, Serbo-Croatian multiple questions do not show Superiority effects, (5), and allow all possible word orders of fronted wh-phrases (in this case, all arguments). 30 Prior to Rudin (1988), there was also work on multiple wh-fronting by Wachowicz (1974), Toman (1981), and earlier work by Rudin (1982 et seq.). 47 (4) Bulgarian (Rudin 1988: (54)) a. Koj vižda? kogo who.NOM whom.ACC sees ‘Who sees whom?’ b. * Kogo koj vižda? (5) Serbo-Croatian (Rudin 1988: (56)) a. Ko je što dao? who.NOM AUX what.ACC who.DAT kome given ‘Who gave what to whom?’ b. Ko je kome što dao? c. Što je ko kome dao?/ kome ko dao? d. Kome je to što ko dao? / ko što dao? In Bulgarian, wh-phrases cannot be separated by a clitic, parenthetical, or an adverb, as shown below in (6), where wh-phrases are ungrammatically separated by the adverb prŭv ‘first’. This means that, in Bulgarian, wh-phrases form a constituent. In Serbo-Croatian, wh-phrases do not form a constituent. This is shown by the fact that a clitic, parenthetical or an adverb, as in example (7), can appear between them. (6) Bulgarian (Rudin 1988: (42)) a. Zavisi ot tova, koj depends on this who.NOM who.ACC kogo prŭv e udaril. first has hit ‘It depends on who hit whom first.’ b. * Zavisi ot tova, koj prŭv kogo e udaril. (7) Serbo-Croatian (Rudin 1988: (43)) a. Ko je koga prvi udario? who.NOM AUX whom.ACC first hit ‘Who hit whom first?’ b. Ko je prvi koga udario? The examples in (8) and (9) show different behavior with respect to multiple wh-extraction out of an embedded clause: In Bulgarian (and Romanian) multiple whextraction out of an embedded clause is obligatory, hence the grammaticality of (8a). 48 If a wh-phrase stays in situ, (8b, c), or moves only to the SpecCP of the embedded clause (8d, e), the sentence is ungrammatical. According to Rudin (1988), multiple long distance extraction is prohibited in Serbo-Croatian (but see fn. 31) and only one wh-phrase can move out of an embedded clause while the other stays in situ, as in (9a, b). If all wh-phrases move out of the embedded clause (9c, d), or if only one whphrase moves out of the embedded clause and one moves to SpecCP of the embedded clause, (9d, e), the sentence is ungrammatical. (8) Bulgarian (Rudin 1988: (6)) a. Koj kŭde misliš [če e otišŭl _ _]? who where think.2SG that has gone ‘Who do you think (that) went where?’ b. * Koj misliš [če e otišŭl _ kŭde]? c. * Kŭde misliš [če koj e otišŭl _ _]? d. * Koj misliš [kŭde (če) e otišŭl _ _]? e. * Kŭde misliš [koj (če) e otišŭl _ _]? (9) Serbo-Croatian (Rudin 1988: (11)) a. Ko želite who want.2PL [da vam šta kupi _]? that you.DAT what buy.3SG ‘Who do you want to buy you what?’ b. Šta želite [da vam ko kupi _ ]? what want.2PL that you.DAT who buy.3SG ‘What do you want who to buy you?’ c. * Ko šta želite da vam kupi _ _]?31 d. * Šta ko želite [da vam kupi _ _]? e. * Ko želite [šta da vam kupi _ _]? f. * Šta želite [ko da vam kupi _ _]? 31 Bošković (1997c), on the other hand, claims that there are some Serbo-Croatian speakers who find multiple long distance fronting, as in (i), acceptable: (i) Serbo-Croatian (Bošković 1997c: (8)) a. Ko si koga tvrdio, who.NOM AUX who.ACC claimed ‘Who did you claim beat whom?’ b. * Koga si ko tvrdio, da je istukao? 49 da that je AUX istukao? beaten Rudin (1988) explains the grammaticality of (8a) with the availability of a multiply filled specifier of CP in Bulgarian. In order for movement not to violate Subjacency, the wh-phrases must move out of the embedded clause through SpecCP. Bulgarian has a multiply filled SpecCP through which multiple wh-phrases can move, but Serbo-Croatian has only a SpecCP position through which only one whphrase can move. This, following Rudin (1988), is also related to wh-islands, which can be violated in Bulgarian, as shown in (10), but not in Serbo-Croatian, (11). (10) Bulgarian (Rudin 1988: (19)) Vidjah edna kniga, kojatoi se čudja saw1SG a wonder1SG who knows who sells book which [koj znae [ koj prodava _i]] ‘I saw a book which I wonder who knows who sells (it).’ (11) Serbo-Croatian (Rudin 1988: (24a)) * Šta si what have.2SG me pitao ko može da me asked who can to uradi? do ‘What did you ask me who can do?’ Rudin (1988) attributes these differences to the ability of a language to multiply fill the SpecCP position, which she states in The multiply filled SpecCP hypothesis: (12) The multiply filled SpecCP hypothesis (Rudin 1988: (70)) SpecCP can contain multiple wh-words at S-structure in Bulgarian and Romanian ([+MFS] languages]), but only one in Serbo-Croatian, Polish, and Czech ([–MFS] languages). According to Rudin, the two groups then have different structures for multiple wh-questions. The structure for [+MFS] languages is shown in (13) and the structure for [–MFS] languages is shown in (14). 50 (13) SpecCP in [+MFS] languages (Rudin 1988: (73b)) CP 3 SpecCP IP 3 SpecCP WH 3 SpecCP WH g WH (14) SpecCP and IP in [–MFS] Language (Rudin 1988: (82)) CP 3 SpecCP C’ g 3 WH Comp IP 3 WH IP 3 WH IP In [+MFS] languages, the moved wh-phrases are right adjoined to SpecCP as in (13). In [−MFS] languages, only the first wh-phrase is in SpecCP at the point of Spell-Out and the rest are adjoined to IP. At LF, the wh-phrases, which are in [–MFS] languages adjoined to IP, move to SpecCP.32 The two structures also account for the different behavior of the two groups of languages with respect to Superiority. Superiority effects have already been shown, in examples (4) and (5). To explain the difference in word order, Rudin (1988) uses the split version of the Empty Category Principle (i.e. empty categories must be properly governed), according to which an empty element must be governed by a lexical head at PF (head 32 Rudin (1988) assumes that wh-raising is universal and that the difference between languages is in the timing of movement to SpecCP. She accounts for the difference using the Condition on SpecCP adjunction (CSA) (based on Adams 1984): (i) Condition on SpecCP adjunction (CSA) (Rudin 1988: (92)) *[SpecCP α SpecCP] (nothing may be adjoined to SpecCP) Languages differ with respect to when CSA applies. In the [+MFS] the CSA never holds (it does not hold at PF, LF or S-structure), in [–MFS] languages it holds at S-structure (but movement to SpecCP can proceed at LF and PF, that is: these languages have multiple questions, but do not violate islands). 51 government) and an A’-anaphor must be A’-bound in its domain at LF (The local binding condition, Aoun, Hornstein, Lightfoot and Weinberg (1987)).33 Rudin (1988) proposes that in Bulgarian wh-fronting proceeds as follows: “the single wh- which is substituted to SpecCP becomes the head of SpecCP, and its index percolates to SpecCP” (Rudin 1988: 482).34 From SpecCP the index is passed to its Comp – the head of CP which is visible at PF (via Spec-Head agreement). The whphrase and Comp are therefore coindexed. When the subject wh-moves first it is coindexed with Comp (a visible head at PF) and the Comp consequently governs the empty element (i.e. the subject trace). In this case the object trace is head-governed by V. If it was the object wh-phrase that moved first, Comp would be coindexed with the object. The subject trace would be left ungoverned (violating head government) and the sentence would be ungrammatical. As for the lack of Superiority effects in the [–MFS] group: When the subject wh-phrase moves first to SpecCP, Comp is coindexed with it and Comp governs the subject trace. The object trace is head governed by V (as in [+MFS] languages). When the object wh-phrase moves to SpecCP, Comp is coindexed with the object wh-phrase and governs the object trace. The subject trace is governed by INFL (subject wh-phrase moves to SpecIP and INFL is coindexed with it). To conclude, Rudin (1988) shows that there are two different types of multiple wh-fronting. The first is movement of all wh-phrases to CP (as in [+MFS] languages) and the second is movement of only one wh-phrase to CP (as in [–MFS] languages). But there are problems with Rudin’s account. These and subsequent proposals for a 33 Government and domain are defined as follows: (i) (ii) “A (X0) GOVERNS B iff all maximal projections dominating A also dominate B and A is dominated either by all maximal projections dominating B or by all maximal projections dominating the maximal projection of B” (Rudin 1988: (66)). “The DOMAIN for an expression A is the first clause (IP or CP) or NP which contains an accessible SUBJECT for A, where SUBJECT = AGR, [NP, IP], or [NP, NP] (Chomsky 1981), and where B is accessible to A iff A is in the c-command domain of B and assigning the index of B to A would violate neither the i-within-i condition nor BINDING CONDITION C (the condition that R-expressions must be A-free)” (Rudin 1988: (68)). 34 Substitution is, besides adjunction, one of the types of movement assumed in Chomsky (1986). In substitution, a maximal projection moves to the specifier position. In this system, wh-movement to SpecCP is one of the major cases of substitution. 52 new analysis will be described in section 3.2, but I will first show how Rudin’s account was applied to Slovenian. 3.1.1 The [+/– MFS] analysis and Slovenian (Golden 1996a, b, 1997a) The first analyses of multiple wh-fronting in Slovenian were made by Golden (1996a, 1996b, 1997a). Golden (1996a) compares different properties of Slovenian multiple wh-fronting with English wh-movement. She observes that Slovenian allows multiple wh-phrases at the beginning of a multiple questions, (15), but notes that there are restrictions on which wh-phrases can be fronted, i.e. while two argument wh-phrases or an adjunct wh-phrase and an argument wh-phrase can front easily, (17), zakaj ‘why’ and kako ‘how’ cannot, (16a), except if the two are coordinated, (16b):35 (15) Kdo komu who whom kaj kupuje? what buys ‘Who is buying what for whom?’ (16) (Golden 1997a: (10a)) a. * Zakaj why kako Peter pretepa Toneta? how Peter beats Tone 35 The restriction only seems to apply to zakaj ‘why’ and kako ‘how’. Adjuncts such as kje ‘where’ and kdaj ‘when’ can both front in Slovenian, as shown in (i). This does not seem to be the case for Serbo-Croatian, as noted by Citko and Gračanin-Yuksek (2013), who show that multiple wh-questions with two fronted adjuncts are not acceptable in Croatian, but when only one wh-adjunct is fronted, the question is acceptable. (i) (ii) Slovenian Kdaj kam potuješ? when where-to travel ‘When and where are you traveling?’ Croatian (Citko and Gračanin-Yuksek 2013: (38a), fn. 31, (i)) a. * Gdje kada Ivan nastupa? where when Ivan performs ‘Where does Ivan perform when?’ b. Gdje Ivan kada nastupa? 53 b. Zakaj in kako je when and how AUX ženska umrla? woman die? ‘Why and how did the woman die?’ (17) (Golden 1996a: (8b)) Kako koga Peter pretepa? why who Peter beats ‘How does Peter beat who?’ Golden claims that questions with multiple adjuncts sound better when the two wh-adjuncts are separated by intervening material, such as clitics in (18), or when one wh-adjunct is not fronted, as in (19) and (20).36 36 Slovenian is not the only example of a language with such a constraint against multiple fronted ‘true’ wh-adjuncts, nor is it the only language where such cases can be improved. As Golden notes, Bošković (1994, also 1998) discusses a similar constraint in Serbo-Croatian. As shown in (i), two (true) adjuncts cannot be wh-fronted in Serbo-Croatian. (ii) shows that an adjunct and an argument can be fronted in Serbo-Croatian: (i) (ii) Serbo-Croatian (Boškovič 1994: (7a)) * Zašto je kako istukao Petra? why AUX how beat Peter? ‘Why and how did he beat Petar?’ (Boškovič 1994: (8a)) Kako je koga Petar istukao? how AUX who.ACC Petar beat ‘How did Petar beat whom?’ There are two mechanisms of improving these examples in Serbo-Croatian. As in Slovenian, the two adjunct wh-phrases can be coordinated in Serbo-Croatian, (iii), or another non-adjunct whphrase can be added, (iv), Bošković (1998): (iii) (Bošković: 1998 (53)) Zašto i kako je istukao Petra? why and how AUX beaten Petar (vi) ? Zašto je koga kako istukao? why AUX whom how beaten ‘Why did he beat whom and why did he beat him?’ An analysis of the constraint was proposed in Bošković (1994), who claims this is an ECP type of effect that arises because of optional coindexing of C with the moved element (Spec-Head Agreement). In this analysis, the C, which is coindexed with the moved element, is the antecedent governor for the relevant trace and the C can only govern one trace (either at SS or LF) – if more than one trace needs to be antecedent governed, as in the case of fronted ‘how’ and ‘why’, the sentence will be ungrammatical as one of the traces will be left ungoverned, thus violating the ECP (cf. Bošković 1994 and the references cited therein). 54 (18) (Golden 1997a: (11a)) ?* Zakaj jih CL.ACC why je kako Peter opravljal? AUX how Peter performed ‘Why and how was Peter taking them?’ Possible answer: Peter was taking the exams orally because he broke his arm. (19) (Golden 1997a: (11b)) ?? Zakaj why je Peter kako opravljal izpit iz matematike? AUX Peter how performed exam from math ‘Why and how was Peter taking the math exam?’ (20) (Golden 1997a: (11c)) Zakaj so ga why him where how-much time before AUX kje koliko časa pred kom kako zasliševali? who how interrogate ‘Why did they interrogate him where, for how long, in front of whom, and how?’ The order of fronted wh-phrases is free, which means Slovenian does not display Superiority effects, as shown in examples (21) and (22). (22) also shows that fronted wh-phrases do not form a constituent, since the two wh-phrases are divided by an auxiliary clitic. The non-constituency is again shown in (23), in which fronted wh-phrases are separated by a parenthetical po tvoje ‘in your opinion’. Golden (1997a) notes that with respect to these properties Slovenian behaves like SerboCroatian, since these are the properties of languages without multiply filled Specifiers, in Rudin’s (1988) typology. (21) (Golden 1997a: (3)) a. Kdo komu who.NOM who.DAT dolguje? kaj what.ACC owes ‘Who owes what to whom?’ b. Kdo kaj komu dolguje? c. Komu kdo kaj / kaj kdo dolguje? d. Kaj kdo komu / komu kdo dolguje? 55 (22) a. Kdo je koga udaril? who.NOM AUX whom.ACC hit ‘Who hit whom?’ b. (23) a. Koga je kdo udaril? po Kdo tvoje kdaj kuha kosilo? who.NOM after you when cooks lunch ‘Who in your opinion cooks lunch when?’ b. Kdaj po tvoje kdo kuha kosilo? Golden (1996a) also notes that a wh-phrase can occur with an overt complementizer, (24), in embedded clauses and that when a wh-phrase (an argument or an adjunct) is extracted out of an embedded clause, a complementizer must be used, (25): (24) (Golden 1996a: (13a, c, d)) a. Rad bi vedel, koga (da) je Peter videl? like would know who (that) is Peter saw ‘I would like to know who Peter saw.’ b. Sprašujem se, koga ali wonder1SG REFL who whether Špela ljubi?37 Špela loves ‘I wonder whether Špela loves who?’ c. Nisem ga vprašal, AUX-NEG-1S him ask da zameri. komu kaj whom what that resent ‘I did not ask him who he resents for what.’ (25) Kaj je Špela what AUX Špela mislila, think *(da) that je Peter kupil. AUX Peter bought ‘What did Špela think that Peter bought?’ 37 Golden (1996a) presents this as an acceptable sentence in Slovenian, but speakers I have contacted do not find (24b) (completely) acceptable. The reason for this might be just the choice of the complementizer ali ‘if’. Changing ali to če ‘if’, (i), improves the grammaticality of the sentence for the speakers I have contacted. (i) Sprašuješ, kdo če pride. wonder.2SG who if comes ‘You are wondering who is coming.’ 56 As shown by the examples above, Slovenian allows extraction of a single whphrase out of an embedded clause. Golden (1996a, 1997a) also claims that Slovenian allows multiple extraction out of embedded clauses, as shown in example (26a) below, and is thus comparable to Bulgarian (see above). The judgements reported here are taken from Golden (1996a, 1997a). They are not always shared by me or by the speakers who I have consulted. I will give a more detailed discussion of the judgements and the availability of multiple extraction in section 3.1.2, but see also Marušič (2008). (26b) shows that multiple extraction out of an embedded clause can be optional in Slovenian, but note that the same does not hold in Bulgarian (Golden 1997a). (26) (Golden 1997a: (19a, b) a. Sprašujem te, Ask1SG b. you who.ACC Sprašujem te, Ask1SG koga koga da ogovarjam.38 komu Peter trdi, who.DAT Peter claims that disparage Peter trdi, da ogovarjam komu. you who.ACC Peter claims that disparage who.DAT ‘I am asking you who Peter claims that I am disparaging to whom.’ Based on examples with multiple extraction out of an embedded clause, such as (26a), Golden (1996a, 1997a) observes similarities between Slovenian and Bulgarian. Remember that in Rudin’s analysis of languages with multiple whfronting (1988) Bulgarian is said to allow multiple wh-extractions from a clause, while Serbo-Croatian only allows single wh-extraction. Rudin explains this with multiple SpecCP positions: because Bulgarian has multiple Specifiers, and wh- 38 Many speakers do not find this example completely acceptable or get a different interpretation. The example also includes the verb ogovarjati ‘to disparage’, which is not very common. When it is replaced with a more common verb opravljati ‘to gossip’ judgements improve, but still the majority of speakers get a different reading in which we are not asking about Peter’s claims about who and to whom I am gossiping, but rather to whom Peter is claiming this. In this case only one wh-word would have been extracted from the embedded clause. I return to this in section 3.1.2. (i) Sprašujem te, koga komu Peter trdi, da opravljam. ask.1SG you who.ACC who.DAT Peter claims that gossip.1SG ‘I am asking you to whom is Peter claiming that I am gossiping about whom.’ 57 phrases can move through them. The availability of multiple extraction then, according to Golden, suggests that Slovenian too allows multiple SpecCP positions. From the availability of multiple extraction, Rudin predicts that Multiple Specifier Languages do not obey wh-island constraints, which is the case for Bulgarian (see Rudin 1988). As shown above, Golden (1997a) claims that Slovenian allows multiple extraction out of an embedded clause and she also shows that Slovenian displays no wh-island effects (as predicted by the presence of multiple whextraction), (27a). She notes that wh-extraction of adjuncts from wh-islands is not completely acceptable, but that it is still grammatical, (27b). Again, the judgements presented here are from Golden (1997a) and more will be said about these examples in section 3.1.2. (27) (Golden 1997a: (23a), (24)) a. Komu mi povej, kaj Špela who.DAT me.DAT tell what Špela zavida? envies ‘Tell me whom Špela envies what.’ b. ? Zakaj why si prepričan, katere knjige ne bodo prevedene? AUX convinced which books not will-be translated ‘Why are you convinced which books will not be translated?’ Both multiple wh-extraction out of embedded clauses and the lack of wh-island effects, are, following Rudin (1988), properties of [+MFS] languages, but (as shown above) Golden (1997a) notes, that Slovenian does not behave as other [+MFS] languages with respect to Superiority effects and constituency of fronted wh-phrases. Note however, that Golden (1997a) does not discuss Superiority effects in cases with multiple wh-extraction out of embedded clauses. Based on the non-uniform behavior Slovenian displays with respect to Rudin’s diagnostics, Golden suggests that Slovenian requires two distinct structures for multiple wh-questions: The [–MFS] structure in direct multiple wh-questions (one wh-phrase in SpecCP) and the [+MFS] structure for questions with multiple whextraction out of embedded clauses and wh-islands. According to Golden (1997a), this also shows that not all multiple wh-fronting languages will behave uniformly with respect to Rudin’s typology of multiple wh-fronting languages. 58 3.1.2 Problems with the [+/–MFS] analysis and Slovenian While the [–MFS] analysis of matrix questions in Slovenian seems to be unproblematic, there are in fact two problems with Golden’s analysis of Slovenian as a language which allows the [+MFS] structure for instances of multiple extraction out of embedded clauses and for wh-island violations. The first is related to the data presented in Golden (1997a), the second is a problem with the theory which relates the possibility of violation of wh-islands to multiple Specifiers of CP. Following Golden, the first argument for the availability of the [+MFS] structure comes from the availability of multiple extraction out of embedded clauses. But as already noted above, speakers do not agree with this. For example, Marušič (2008) shows that Slovenian allows only single long distance movement out of embedded finite clauses. He also notes that Slovenian allows multiple wh-movement out of non-finite clauses, as shown in (29), but that this movement is optional. (28) (Marušič 2008: (15)) Kaj je (*komu) rekel Janez, (*komu) da je Peter what is who.DAT said Janez, whom.DAT that is Peter dal *(komu)? give whom ‘What did Janez say that Peter gave whom?’ (29) (Marušič 2008: (16a)) Komu si kaj pozabil dati? Who.DAT AUX.2SG what.ACC forgot give.INF ‘Whom did you forget to give what?’ As for examples from Golden (1997a), which I repeat below with a different verb (see fn. 38), speakers find example (30b) completely grammatical, which means that single wh-movement out of an embedded clause is acceptable (given the similarities with [–MFS] languages and example (28) this is not surprising). Speakers, however, either do not accept example (30a) or try to assign it an interpretation under which we are asking ‘to whom is Peter claiming that I am gossiping and who am I gossiping about?’. But this interpretation indicates that speakers analyse the sentence as an instance of single wh-movement out of the 59 embedded clause and wh-movement out of the matrix clause and not as multiple whmovement out of an embedded clause. (30) (Golden 1997a: (19a, b)) a. Sprašujem te, ask.1SG b. you who.ACC Sprašujem te, ask.1SG koga koga you who.ACC komu Peter trdi, da opravljam. who.DAT Peter claims that gossip Peter trdi, da opravljam komu. Peter claims that gossip who.DAT ‘I am asking you, who Peter claims that I am gossiping to whom?’ When speakers get examples in which such an interpretation is not possible, as in (31), they do not find it acceptable. (31) *Pišem, kdo write.1SG who.NOM je koga Miha pravil Tonetu, da je udaril. AUX who.DAT Miha said Tone.DAT that is hit Intended: ‘I am writing about who Miha said to Tone hit whom.’ This indicates that multiple wh-movement out of embedded clauses in Slovenian is not acceptable and that the [+MFS] structure cannot be applied in these examples since Slovenian behaves like a [–MFS] language. The second argument for the availability of multiple SpecCP (i.e. the [+MFS] structure of multiple wh-questions) provided by Golden (1997a) is the possibility of violating wh-islands in Slovenian. According to Golden, examples in which a whphrase moves from a wh-island are completely acceptable. This was shown in (27), which I repeat below as (32). Example (32a) shows movement of an argument and example (32b) of an adjunct wh-phrase. (32) (Golden 1997a: (23a), (24)) a. Komu mi povej, kaj who.DAT me.DAT tell what Špela ‘Tell me whom Špela envies what.’ 60 Špela zavida? envies b. ? Zakaj why si prepričan, katere knjige ne bodo prevedene? AUX convinced which books not will-be translated ‘Why are you convinced which books will not be translated?’ But these two examples, based on which Golden concludes that Slovenian can violate wh-islands, are problematic. Example (32b) is acceptable, but only when it receives the irrelevant meaning in which zakaj ‘why’ is understood as originating in the matrix clause. If the wh-phrase ‘why’ was moved from the island, the paraphrase of the question (32b) would be ‘why the books will not be translated’. In fact, speakers understand (32b) as ‘why is one convinced that the books will not be translated?’. This implies that zakaj ‘why’ is not moved out of the wh-island but rather merged in the higher clause. Note also that adjunct extraction out of whislands is also unacceptable in Bulgarian (which Rudin described as a [+MFS] language) in both questions and relativization, as observed in Bošković (2003). This is shown below in the unacceptable example (33a). The fact that it is extraction from islands that is unacceptable is confirmed by example (33b), in which an adjunct is extracted out of an embedded clause and which is acceptable. (33) Bulgarian (Bošković 2003: (16b, c)) a. * Zašto /poradi kakva pričinai znae why for which reason knows whether Boris e zaminal ti]? is left [dali Boris ‘Why/for which reason does he know whether Boris left?’ b. Zašto /poradi kakva pričinai misliš [če Boris e why for which reason think.2SG that Boris is zaminal ti]? left ‘Why/for which reason do you think that Boris left?’ Golden’s example shown in (32b) therefore cannot be taken as an example of extraction out of a wh-island in Slovenian or as an argument for the availability of the [+MFS] structure. 61 In addition, (32a) is surprising given what we know about Bulgarian. In Bulgarian, wh-island violations are only possible with D-linked questions (Rudin 1988, Comorovski 1996) (also, see below for Swedish). It is therefore unexpected that wh-island violations in Slovenian are available with non-D-linked questions. However, it is also not clear whether Slovenian even allows wh-movement out of wh-islands. This is apparent if we compare example (32a), which is marked as acceptable in Golden (1997a) to (34a), which speakers find unacceptable. (34b) shows that this is not because of word order. In addition, example (35) shows that Dlinking does not improve grammaticality of wh-island violations. Example (36) shows that single extraction from an embedded clause is grammatical in Slovenian. This indicates that it is the extraction out of the wh-island that makes (34) ungrammatical. (34) a. * Kdo who.NOM je Miha vprašal Majo, AUX Miha ask Maja.ACC, what watches kaj gleda na TV? on TV Intended: ‘Miha asked Maja who watches what on the TV.’ b. * Kaj je Miha vprašal Majo, kdo gleda na TV? (35) *Katera which punca je Rok vprašal Majo, girl Rok ask AUX kaj gleda Maja.ACC what watches na TV? on TV Intended: ‘Miha asked Maja which girl watches what on the TV.’ (36) Kdo Miha trdi who Miha claims Maji, da gleda Maja that watches TV? TV ‘Miha is claiming that who watches TV?’ In addition, the speakers that judged (34) unacceptable also find (32a) unacceptable or try to interpret the sentence as komu ‘who.DAT’ moving from the main clause, despite the fact that this reading is impossible because of the dative clitic mi ‘me.DAT’. This indicates that wh-island violations in Slovenian are not acceptable. Another argument against the multiple SpecCP analysis of Slovenian for examples in which Golden claims that Slovenian allows wh-island violations comes from Bošković (1999, 2003). He provides an argument against Rudin’s analysis of wh-islands on which Golden (1997a) bases her arguments for movement to SpecCP. According to Bošković, Bulgarian can violate wh-islands with argument wh-phrases 62 but Serbo-Croatian cannot (see examples (10) and (11) in section 3.1 above), but it is not the case that a language must necessarily have a multiple SpecCP available in order to violate wh-islands. He shows that wh-island violations are possible in Swedish, a language that does not allow multiple wh-fronting, as shown in (37). Extraction out of wh-islands is possible with relativization, (38a) and D-linked questions, (38b). (37) Swedish (Bošković 2003: (17a), based on Maling 1978) *Vad frågade Jan vem som skrev? what asked John who that wrote ‘What did John ask who wrote?’ (38) Swedish (Bošković 2003: (17b, c), based on Maling 1978 and Engdahl 1986) a. Det är melodin, som Jan frågade vem som skrev. this is song.DEF that John asked who that wrote ‘This is the song that John asked who wrote.’ b. Vilken film var det which film was it du gärna you gladly ville veta vem wanted know.INF who som hade regisserat? that had directed ‘Which film did you want to know who had directed?’ Bošković (2003) also shows that Serbo-Croatian does not allow wh-island violations even in environments in which wh-phrases are fronted to SpecCP (see section 3.2 for more on movement to SpecCP in Serbo-Croatian), such as questions with li. (39) Serbo-Croatian (Bošković 2003: (26a)) * Koju od which of tih knjigai these books se pitaš ko li REFL wonder2SG who Q prodaje ti? sells ‘Which of these books do you wonder who on earth sells?’ Based on Swedish and Serbo-Croatian data, Bošković (2003) concludes that wh-island effects cannot be accounted for with the possibility of having more than 63 one wh-phrase in SpecCP at S-Structure. This conclusion indicates two things. First, the potential availability of wh-island violations in Slovenian would not necessarily indicate a [+MFS] structure in these examples (but, again, Slovenian speakers do not necessarily accept wh-island violations). Second, this shows us that Rudin’s (1988) account of island violations does not account for all the possible data and that it needs to be revised (see Bošković 2003). 3.2 Multiple wh-fronting – ‘real’ and non-wh-movement (Bošković 1997a and after) One of the central concepts in the research of wh-fronting is Superiority. Superiority was first defined in Kuno and Robinson (1972) as the Wh Crossing Constraint: “a wh word cannot be preposed crossing over another wh” (Kuno and Robinson 1972: 474), while Chomsky (1973) proposed the Superiority Condition, which is stated in (40). (40) The Superiority Condition (Chomsky 1973) a. No rule can involve X, Y in the structure … X … [ … Z … WYV …] … where the rule applies ambiguously to Z and Y, and Z is superior to Y. b. the category A is ‘superior’ to category B if every major category dominating A dominates B as well but not conversely. Later, in the Minimalist program, Superiority is defined as in (41) and is termed as ‘Attract Closest’, where Attract F is understood as (42) which incorporates both Last resort and the Minimal Link Condition: (41) α can raise to target K only if there is no legitimate operation Move β targeting K, where β is closer to K. (Chomsky 1995: 296, (82)) (42) K attracts F if F is the closest feature that can enter into a checking relation with a sublabel of K. (Chomsky 1995: 297, (84)) 64 Returning to Rudin (1988), Rudin claims that while Bulgarian and other [+MFS] languages are subject to the Superiority Condition, Serbo-Croatian and other [–MFS] languages are not, see above. The question is then why Serbo-Croatian behaves differently, or as Bošković (1997c) puts it: the lack of Superiority in SerboCroatian is conceptually problematic, since the Superiority Condition follows from the Principles of Economy, which are one of the fundamental properties of languages, thus it would be surprising for it to vary in different languages. In addition to this observation, Bošković (1997b, 1997c, 1999, 2002) shows that Rudin’s analysis does not provide an account for all the available data from multiple wh-fronting languages. In order to expand the data in question, Bošković (1997c) explores different wh-questions in Serbo-Croatian (in addition to short distance matrix questions, which had already been discussed in Rudin (1988)) and shows that it is not the case that Serbo-Croatian simply does not obey the Superiority Condition, as it exhibits Superiority in some contexts (see below). In addition, he shows that Superiority effects can be used as a diagnostic tool for determining the motivation and location of wh-movement. Bošković shows that Serbo-Croatian has examples without Superiority effects, such as the short distance matrix question in (43), but he also shows that one can find Superiority effects in other contexts. Bošković (1997c) shows that Serbo-Croatian exhibits rigid word order in long distance questions, (44), indirect questions, (45), embedded questions with correlative constructions, (46), and overt C questions, (47). (43) Serbo-Croatian a. Ko koga who.NOM who.ACC vidi? sees ‘Who sees whom?’ b. Koga ko vidi? 65 (44) Serbo-Croatian (Bošković 1997c: (8)) a. Ko si koga tvrdio, da je istukao?39 who.NOM AUX who.ACC claimed that AUX beaten je koga istukao. AUX who.ACC beaten. o njemu i govori. about him even talks ‘Who did you claim beat whom?’ b. * Koga si ko tvrdio, da je istukao? (45) Serbo-Croatian (Bošković 1997c: (12)) a. Jovan i Markone znaju ko Jovan and Marko not know who.NOM ‘Jovan and Marko do not know who beat whom.’ b. ?* Jovan i Marko ne znaju koga je ko istukao. (46) Serbo-Croatian (Bošković 1997c: (16)) a. [Ko voli], taj koga who.NOM whom.ACC loves that-one ‘Everyone talks about the person they love.’ b. ?* [Koga ko voli], taj o njemu/o njemu taj i govori. (47) Serbo-Croatian (Bošković 2002: (10)) a. Ko li who.NOM Q koga voli? who.ACC loves ‘Who on earth loves whom?’ b. * Koga li ko voli? Bošković (1997c) shows that the environments in which Serbo-Croatian exhibits Superiority effects are the same as the environments in which French exhibits obligatory wh-movement. The contrast between availability of in situ whquestions and obligatory movement can be seen by comparing (48) and (49). The former is a short-distance matrix question in which a wh-phrase can stay in situ and the latter is an embedded finite clause out of which a wh-phrase (obligatorily) moves. 39 Notice that it is also possible to move only one wh-phrase from an embedded clause with two whwords – in this case it is only the higher one that can be fronted: (i) (Bošković 1997c: (10)) a. Ko tvrdiš da koga who.NOM claim that who.ACC ‘Who do you claim that loves who?’ b. * Koga tvrdiš da ko voli? voli? loves 66 (48) French (Bošković 1997c: (18)) Tu as vendu you have sold quoi? what ‘What did you sell?’ (49) French (Bošković 1997c: (19)) a. * Jean Jean et Marie croient que Pierre a and Marie believe embrassé qui? that Peter has kisses who ‘Who do John and Mary believe that Peter has kissed?’ b. Qui Jean et Marie croient-ils que Pierre a embrassé? In addition, Bošković (2002) observes a parallelism with languages which always exhibit Superiority effects (i.e. Bulgarian and Romanian) and languages that always have overt wh-movement (English) on the one hand, and languages that never exhibit Superiority effects (such as Russian, see below) and languages that never exhibit wh-movement (e.g. in situ languages, such as Japanese). This leads Bošković to posit a correlation between Superiority and wh-movement – languages only exhibit Superiority when wh-movement takes place. This means that Serbo-Croatian short-distance wh-questions do not have wh-movement and, contrary to what Rudin (1988) claims, there is no movement to SpecCP in examples without Superiority. On the other hand, ‘real’ wh-movement (movement to check the [+wh] feature on C) occurs in Serbo-Croatian long distance, overt C, and embedded questions (i.e. contexts with Superiority). Here, the first wh-phrase moves to SpecCP. Following Bošković (2002), this essentially means that in multiple wh-fronting languages there are two types of movement: real wh-movement and non-wh-fronting. The former movement exists in English, which is a single wh-fronting language, Bulgarian, Romanian, both of which are multiple wh-fronting languages, and in some contexts in French and in the same environments Serbo-Croatian. The latter movement is not taken as ‘real’ wh-movement and exhibits no Superiority effects, as in some environments in Serbo-Croatian and in Russian. In addition, because Russian never has ‘real’ wh-movement, it is similar to Japanese, which also never has whmovement. Based on this Russian can be taken as an in situ language, as in Stepanov (1998). 67 A couple of questions remain: What triggers wh-fronting in environments without Superiority effects? Why does does it not result in Superiority? And, since movement of only one wh-phrase is enough to check the [+wh] feature on C (which is the case in English), why does more than one wh-phrase move in multiple whfronting languages? Bošković (1997a, 1997b, 1997c, 2002) bases his analysis of non-whmovement on the proposal by Stjepanović (1995, 1999), who shows that contrastively focused40 phrases and wh-phrases exhibit parallel behavior in SerboCroatian, in that focused phrases and wh-phrases occupy the same position. Stjepanović (2003) shows that focused phrases must be fronted in Serbo-Croatian, as shown in example (50), and that the relevant positions for the focused phrases are either above TP or between VP and TP, which she claims based on examples (51) and (52). (50) Serbo-Croatian (Stjepanović 2003: (13)) PETRA Marija voli. Petar Marija loves ‘It is Petar that Marija loves.’ (51) Serbo-Croatian (Stjepanović 2003: (17a, b)) a. Oni mudro JOVANA savjetuju. they wisely Jovan.ACC advise ‘It is wise of them to advise Jovan.’ ‘They advise Jovan in a wise manner.’ b. Oni JOVANA mudro they Jovan.ACC wisely savjetuju. advise *‘It is wise of them to advise Jovan.’ ‘They advise Jovan in a wise manner.’ 40 Stjepanović (2003) assumes that contrastive focus is the focus which is subsumed in the notion of identificational focus, as stated in Kiss (1998), which means that it represents a subset of the set of contextually or situationally given elements for which the predicate phrase can potentially hold; it is identified as the exhaustive subset of this set for which the predicate phrase actually holds (Kiss 1998: 245). 68 (52) Serbo-Croatian (Stjepanović 2003: (18)) JOVANA oni mudro Jovan.ACC they wisely savjetuju. advise ‘It is wise of them to advise Jovan.’ ‘They advise Jovan in a wise manner.’ In examples (51) and (52) Stjepanović uses an adverb mudro ‘wisely’ to identify the position of focused phrases. Mudro ‘wisely’ can receive either the sentential reading or the manner reading. When the adverb receives the manner reading, it is placed in the VP. On the sentential reading (that is, the speaker oriented reading), the adverb is adjoined to the TP. This is slightly modified by Bošković (1997a) who proposes that the adverb is adjoined to AgrOP when it has the manner reading and to AgrSP when it has the sentential reading. Stjepanović interprets the unavailability of the sentential reading in (51b) and the availability of both readings in (51a) and in (52) with two possible positions of focused phrases. Where the sentential reading is available, the focused phrase is in the TP, where the manner reading is available, (51), the focused phrase is in the VP. This, according to Stjepanović (2003), means that a focused phrase is at least as high as the VP adjoined position, but that it can also be located in TP, as in (52) where the focused phrase is sentence initial, that is, following Stjepanović (2003), the AgrSP position. What is crucial for the analysis of multiple wh-fronting: is that wh-phrases in Serbo-Croatian move to the same positions as focused phrases, as shown below: (53) Serbo-Croatian (Stjepanović 2003: (19)) a. Ko mudro who wisely koga savjetuje? whom advises ?? ‘Who is it wise of to advise whom?’ b. Ko koga who whom mudro savjetuje? wisely advises * ‘Who is it wise of to advise whom?’ ‘Who advises whom in a wise manner?’ 69 c. Koga ko mudro whom who wisely savjetuje? advises ?? ‘Who is it wise of to advise whom?’ ‘Who advises whom in a wise manner? Stjepanović claims that in these examples the judgements about the availability of readings the adverbs receive are the same as for examples with non-wh-focused phrases. Based on this she argues that wh-phrases and contrastively focused phrases occupy the same position. Builiding upon this assumption, Bošković (1997a, 2000, 2002) shows that when fronting of wh-phrases does not display Superiority effects (and therefore does not involve overt movement to check the [+wh] feature), it happens because of the [+focus] feature. Returning to the issue of Superiority, the question is why there is Superiority in ‘real’ wh-movement and no Superiority in non-wh-fronting. Starting with single whfronting in languages such as English: In English the strong feature that motivates movement is in the target ([+wh] feature in C), as in (54). (54) (Bošković 1998: (38)) F wh-phrase1 wh-phrase2 wh-phrase3 +wh +wh +wh +wh strong weak weak weak Because the strong feature is in the target, it is enough for only one wh-phrase to move. After movement, Spec-Head agreement is triggered and results in checking of the strong [+wh] feature. Crucially, this movement is subject to the Superiority condition, as stated in (40), and must proceed in the most economical way. Given this, the closest wh-phrase has to check the strong feature on C. Because of this, (55) is grammatical – it is the higher wh-phrase that moves. Example (56) is ungrammatical because the lower wh-phrase moves to the clause initial position, which violates the Superiority Condition, see (40): (55) Who wrote what? 70 (56) *What did who write? Returning to multiple wh-fronting and Superiority: Languages with Superiority and multiple fronting (i.e. Bulgarian and certain contexts in Serbo-Croatian) are different from English in that all wh-phrases move in multiple wh-questions. In this case, movement occurs because of the strong [+wh] feature on C and focus features on the lower wh-phrases. Because C has a strong [+wh] feature it is the closest whphrase that has to check it, but all other wh-phrases move freely because the strong focus feature is in the element undergoing movement.41 Turning to focus movement, Bošković (1998, 1999, 2002) claims that with focus fronting (or non-wh-movement) the strong feature that is the driving force behind movement is in the moved element. The operation is called Attract and is shown in (57). Because the strong features are in the elements that undergo movement, it is irrelevant in which order they move, as they must cross the same nodes – hence no Superiority.42 (57) (Bošković 1998: (39)) F wh-phrase1 wh-phrase2 wh-phrase3 +focus +focus +focus +focus weak strong strong strong How is it possible to have both contexts without Superiority (short distance questions) and with Superiority (long distance questions, indirect questions, (45) embedded questions with correlative constructions, and overt C questions) in SerboCroatian? The suggested analysis for the two types of multiple wh-movement is 41 The word order of wh-fronting in Bulgarian confirms this – Superiority only holds for the first moved wh-phrase (that is the wh-phrase that under this analysis moves to check the [+wh] feature on C, and the word order of the remaining wh-phrases is free: (i) a. (Bošković 1998: (32)) Koj kogo kak e tselunal? who.NOM who.ACC how AUX kissed ‘Who kissed whom how?’ b. Koj kak kogo e tselunal? 42 Bošković (1998) notes that the difference between wh-fronting and focus-fronting is also in the attracting head. The C head has an Attract-1 property, while the F head has an Attract-All property. 71 based on the different options for the Merge of C. In accordance with the Minimalist Program (Chomsky 1995), Bošković (1997c) states that an element with its features can be Merged before Spell-Out or after it (either at PF or LF).43 The features can either be weak or strong – and it is the strong features that need to be checked as soon as possible. What Bošković then proposes is that in French and Serbo-Croatian environments with overt movement and Superiority, respectively, the C head with the strong [+wh] feature is merged in overt syntax together with a Q feature. In short distance questions in Serbo-Croatian, on the other hand, a phonologically null C with a strong Q feature can be merged at LF. In both cases movement must proceed as soon as possible to check the strong feature – in short distance questions this happens at LF, but in long distance questions this happens in the overt syntax. To summarize this for Serbo-Croatian: the difference between examples with and without Superiority effects in Serbo-Croatian is in the timing of the insertion of the interrogative C (with a [+wh] feature). In examples without Superiority effects (such as matrix questions in Serbo-Croatian), wh-phrases move overtly to check the focus feature on the wh-phrase, but the interrogative C is inserted at LF – a wh-phrase undergoes covert wh-movement to check the [+wh] feature on C and no overt whmovement is needed. In Serbo-Croatian environments with Superiority effects, interrogative C is inserted overtly. This C is phonologically null in embedded and long distance questions, but in questions with the particle li the C is overt. In any case, wh-movement proceeds overtly to check the [+wh] feature. This proposal was slightly revised in Bošković (1999) in which it is proposed that the motivation for focus movement is not in the strong [+focus] feature in the element undergoing movement, but rather that the target head has an Attract-all-F specification. In ‘real’ wh-fronting on the other hand, the target head has an Attract1-wh specification. The result, however, is the same as in the previous proposal, which is: there are no Superiority effects in focus fronting. As for the location of focus fronting, Bošković (1997a) again follows Stjepanović’s analysis, in which focus fronting proceeds to an IP-adjoined or a VPadjoined position. Bošković (1997a) modifies it in accordance with the Split INFL hypothesis, according to which there are two AgrP positions - AgrsP above TP and 43 After Spell-Out, an element can be inserted in PF if it has no semantic features or at LF if the element has no phonological features. 72 AgroP below TP. AgrP positions are, following Bošković (1997a), the positions where focus gets checked in Serbo-Croatian.44 While I have until this point shown examples of a language that always exhibits Superiority effects (i.e. Bulgarian) and a language that in some cases exhibits Superiority effects (i.e. Serbo-Croatian), the paradigm can be completed with a language that never exhibits Superiority effects. This is Russian, as shown by Stepanov (1998). The lack of Superiority effects is shown below for matrix clauses, (58), and embedded questions, (59). Based on the analysis by Bošković (1997a, 1997b, 1998c, 2000), Stepanov shows that the lack of Superiority effects in Russian indicates that wh-phrases in Russian never move to check the [+wh] feature, but rather to check the [+focus] feature. Still, contrary to Bošković, Stepanov does not adopt the Split-INFL version of the analysis, but assumes the complement of C is I and the fronted wh-phrases are adjoined to IP (and not the Agr projection). Crucially, wh-phrases in Russian do not overtly front to SpecCP. (58) Russian (Stepanov 1998: (11)) a. Kto kogo who.NOM who.ACC videl? saw ‘Who saw whom?’ b. Kogo kto videl? (59) Russian (Stepanov1998: (18)) a. Ivan i Petr ne Ivan and Peter not pomnjat kto kogo remember who.NOM who.ACC pobil. beat ‘Ivan and Peter do not remember who beat whom.’ b. Ivan i Petr ne pomnjat kogo kto pobil. Based on this we can conclude that there are two different motivations for wh-fronting (following Bošković 1998, 1999, 2002; Stepanov 1998). The first is the 44 Another argument for different locations of multiple wh-fronting comes from the interpretation of multiple wh-questions. Bošković (2001a, 2002) shows that wh-movement to SpecCP forces a pair list reading and that when no wh-element is overtly moved to SpecCP both single pair and pair list answers are available. Bulgarian and Romanian have only pair list answers, while Serbo-Croatian, Russian and Polish have pair list and single pair answers, which means they move wh-phrases below CP (Bošković 2002). 73 checking of the [+wh] feature, which results in Superiority effects, and the second is the checking of the [+focus] feature, which does not result in Superiority effects. Languages then fall into three different groups: languages that always exhibit Superiority effects, such as Bulgarian, which are the ‘wh-fronting languages’ (based on the similarities with English), languages that never exhibit Superiority effects, such as Russian, these are then the ‘in situ-languages’ (based on similarities with Japanese), and the mixed type, such as Serbo-Croatian (this group is then similar to French). In the next section I show how Superiority can be used as a diagnostic to analyze Slovenian multiple wh-fronting and show that Slovenian falls into the ‘in situ’ group of languages, in that it never exhibits Superiority. 3.2.1 Superiority as a diagnostic for Slovenian As already mentioned, Golden (1997a) has established that Slovenian exhibits no Superiority effects in matrix questions. As shown in (60), there is no strict word order between subject and direct object wh-phrases. (61) shows that the same holds for the subject and an adjunct wh-word. (60) a. Kdo je koga poljubil? who.NOM AUX who.ACC kisses ‘Who kissed whom?’ b. (61) a. Koga je kdo poljubil? Kdo je kdaj poljubil Toneta? who.NOM AUX when kissed Tone ‘Who kissed Tone when?’ b. Kdaj je kdo poljubil Toneta? But as argued by Bošković, the word order of wh-phrases in matrix questions is not enough to establish where and why wh-phrases move. In order to do this, we must also look at different contexts such as embedded or long distance questions. 74 Starting with embedded questions: (62) shows that the word order of whphrases is free in Slovenian embedded questions (for argument and non-argument wh-phrases). (62) a. Miha razmišlja, kdo je koga poljubil. Miha thinks AUX who.ACC kissed who.NOM ‘Miha is thinking about who kissed whom.’ b. (63) a. Miha razmišlja, koga je kdo poljubil. Miha razmišlja, koga je kdaj poljubil. Miha thinks AUX when kissed who.ACC ‘Miha is thinking when he kissed whom’ b. Miha razmišlja, kdaj je koga poljubil. Embedded questions in Slovenian differ from embedded questions in SerboCroatian, which exhibit Superiority effects in this environment. Russian, on the other hand, behaves as does Slovenian and does not have a rigid word order of fronted whphrases in embedded questions, as shown in example (59) in the previous section. There are also no Superiority effects in correlative sentences. These show that Slovenian behaves like Russian. As in Russian, wh-words have an indefinite meaning in correlative sentences and the examples below are not questions (see Stepanov 1998). Still, word order can vary in correlative sentences. (64) a. Če kdo if who.NOM who.ACC koga vidi, bo ta tega see.3SG will this.NOM that.ACC spoznal. recognize ‘If somebody sees someone, he will recognize him.’ b. (65) a. Če koga kdo vidi, bo ta tega spoznal. Če se bi if REFL would who.NOM when kdo takrat uspelo. that-time succeed kdaj 75 potrudil, bi temu tries will this ‘If somebody would put an effort in something, he would succeed at that time.’ b. Če se bi kdaj kdo potrudil, bi temu takrat uspelo. There are also no Superiority effects in questions where a constituent is topicalized in front of wh-phrases. This is, following Stepanov (1998), an environment in which the topicalized phrase is adjoined to CP and the interrogative Q of C is projected overtly. As in Russian (see Stepanov 1998 for Russian examples), the word order of fronted argument and non-argument wh-phrases is free in Slovenian, as shown below. (66) a. V tej šoli, in this school kdo koga who.NOM who.ACC uči? teaches ‘In this school, who teaches whom?’ b. (67) a. V tej šoli, koga kdo uči? V tej šoli, in this school kdaj uči? kdo who.NOM when teaches ‘In this school, who teaches when?’ b. V tej šoli, kdaj kdo uči? Multiple long distance questions cannot be used to test for Superiority in Slovenian, as multiple fronting from embedded clauses is not acceptable in Slovenian (see section 3.1.2), (68a). What we can observe is that examples with multiple long distance fronting are equally unacceptable regardless of the word order or fronted wh-phrases.45 45 We also cannot take single wh-movement out of embedded question as a test for Superiority as embedded wh-questions are syntactic islands in Slovenian which means that a wh-phrase cannot move from them: (i) a. b. * Kdo je Miha vprašal, komu who.NOM AUX Miha ask who.DAT Intended: ‘Miha asked who bought a gift for whom.’ * Komu je Miha vprašal, kdo je kupil darilo? 76 je AUX kupil darilo? buy gift (68) a. * Kdo je who.NOM is dal komu Miha trdil Maji, da who.DAT Miha claimed Maja.DAT that je AUX darilo? give gift ‘Who is Miha claiming to Maja that has a gift to whom?’ b. * Komu je kdo Miha trdil Maji, da je dal darilo? Examples from (60) to (67) show that word order of fronted wh-phrases in Slovenian is not rigid in any of the environments that can be used to determine the motivation for fronting of wh-phrases in a language. The data above indicates that multiple wh-fronting in Slovenian is similar to Russian multiple wh-fronting in that it never exhibits Superiority. Using Superiority as a diagnostic tool, as described in section 3.2, we can conclude (following Bošković 1998, 1999, 2002) that wh-phrases in Slovenian are never moved to check the [+wh] feature, but rather to check the [+focus] feature which means that wh-phrases undergo focus movement in Slovenian (and not wh-fronting). In addition, this would also mean that the wh-phrases do not move to CP, but to IP. But this analysis is problematic, as I show below. 3.2.2 Problems with the focus analysis of multiple wh-fronting in Slovenian and beyond While the ideas presented so far have been accepted by many researchers who have looked at multiple wh-questions in their languages, they are not unproblematic. In this section, I show some of the problems with the analysis proposed by Bošković (1997a et seq.) – some of which have been raised before. I will first start with Slovenian counterexamples as the first problem for the focus analysis of multiple whfronting in languages with no Superiority. One of the supporting facts for the focus analysis of multiple wh-fronting is, according to Bošković (1998), the fact that all wh-phrases in multiple wh-questions must move. If we assume the focus analysis of multiple wh-fronting, then we assume that the strong focus features that motivate movement are in the wh-phrases that undergo movement and not in the C head (Bošković 1998: 14): “If the relevant 77 strong feature were to reside in the target it would suffice to front only one of SC whphrases in multiple questions […].” In Bošković (1999), this is restated in terms of an Attract-all-F head. It is because of this Attract-all-F head that all wh-phrases need to move. We can then predict that in multiple wh-fronting languages in which there are no Superiority effects, all wh-phrases will always have to move, as Bošković claims to be the case in Serbo-Croatian. (69) Serbo-Croatian (Bošković 1998: (37)) a. Ko šta gdje kupuje? who.NOM what.ACC where buys ‘Who buys what where?’ b. * Ko kupuje šta gdje? c. * Ko šta kupuje gdje? d. * Ko gdje kupuje šta? Slovenian, however, is a clear counterexample to this prediction. As shown in section 3.2.1, Slovenian exhibits no Superiority effects, but as shown by (70), whfronting in Slovenian is optional (see also section 2.3). (70) a. Komu je Jože kupil kaj? who.DAT AUX Jože buy what ‘For whom did Jože buy what?’ b. Kaj je Jože kupil komu? What is more, many Serbo-Croatian speakers do not agree with the judgements in Bošković (1998) and claim that questions in which only one wh-phrase moves are just as acceptable as questions with multiple fronted wh-phrases: (71) Serbo-Croatian a. Ko šta vidi? who.NOM what.ACC see ‘Who sees what?’ b. Šta ko vidi? 78 c. Ko vidi šta? The question is then how to account for this unexpected data from languages that exhibit no Superiority. The second problem for the focus analysis of multiple wh-fronting languages with no Superiority, in this case Slovenian, is the existence of questions with multiple fronted D-linked phrases. In Serbo-Croatian D-linked phrases can stay in situ, as (72) shows, but they can move in Slovenian. Bošković (2002) takes the behavior exhibited by D-linked phrases in Serbo-Croatian to be a confirmation of a focus analysis of multiple wh-fronting in matrix clauses. (72) Serbo-Croatian (Bošković 2002: (26a), (27)) a. Ko je who.NOM is kupio koju knjigu? bought which book.ACC ‘Who bought which book?’ b. ? Ko je koju knjigu kupio? Following Bošković, D-linked phrases do not move because they are already part of the discourse. Accordingly, D-linked phrases are not inherently focused and therefore cannot undergo focus fronting (see also section 4.3). Bošković notes, however, that they can optionally be scrambled. In Slovenian D-linked phrases can be fronted, as shown in (73), but see also sections 4.4 and 4.3. (73) a. Kateri slikar je katero sliko naslikal za svojo ženo? which painter AUX which painting paint for self wife ‘Which painter painted which painting for his wife?’ b. Katero sliko je kateri slikar naslikal za svojo ženo? Such examples cannot be accounted for under the focus analysis, since both fronted wh-phrases are D-linked and therefore given in the discourse. Following Bošković, these wh-phrases are moved because of scrambling. But the question is then, why, in questions with two D-linked wh-phrases, both phrases cannot stay in 79 situ (if scrambling is optional), as (74) shows. Note, that (74) is possible as an echo question, but is unacceptable under a true question reading. (74) ?* Maja je posodila Maja AUX lend kateri film kateremu fantu? which movie which boy ‘To which boy did Maja lend which movie?’ One reason for the ungrammaticality of (74) might be Clause Typing – all sentences must be typed (Cheng 1991) and because no wh-phrase moves in this example, the sentence is not typed as a wh-question and is therefore ungrammatical. But this would mean that scrambling is responsible for Clause Typing and is in this case obligatory, which goes against the understanding of scrambling as a semantically vacuous, optional operation. Another option is that Slovenian is an Attract-all-wh language. Following Bošković (2007), Hungarian may behave as an Attract-all-wh language (both D-linked and non-D-linked wh-phrases in multiple whquestions need to move because of an Attract-all-wh property – again, if a language is an Attract-all-focus language, D-linked phrases do not need to move). A similar situation holds in Basque (Bošković 2007). But if this is the case, we can expect that all D-linked phrases need to obligatorily front. This, however, is not the case, as example (75) shows (nor is it the case that all wh-phrases have to front in Slovenian). (75) a. Kateri film je Maja posodila kateremu fantu? which movie AUX Maja lend which boy ‘To which boy did Maja lend which movie?’ b. Kateremu fantu je Maja posodila kateri film? It is therefore not clear how to account for movement of D-linked wh-phrases in Slovenian under the analysis proposed by Bošković (2007). Another problem comes from sluicing. Slovenian, as a multiple wh-fronting language, allows for multiple sluicing, i.e. sluicing with multiple remnants. Example (76) is an instance of single sluicing and example (77) of multiple sluicing. 80 (76) a. Tone misli, da je nekoga izpostavil virusu. Tone thinks that AUX someone.ACC expose virus.DAT ‘Tone thinks that he exposed someone to a virus.’ b. Koga? who.ACC ‘Who?’ (77) a. Tone pravi, da je Tone says that AUX nekoga someone.NOM nečemu izpostavil. someone.DAT expose ‘Tone is saying that he exposed somebody to something.’ b. Koga čemu? who.ACC what.DAT ‘Who to what?’ c. ?* Čemu koga? While (77c) seems at first glance to be ungrammatical because of Superiority, that is not the case. This can be shown with (78) in which the reversed word order of wh-remnants can be observed in the sluice. Note that I am using the verb izpostaviti ‘expose’ with which the objects come in a fixed order, i.e. the accusative object precedes the dative prior to movement, but the order can be reversed after movement. (78) a. Tone pravi, da je nečemu nekoga izpostavil. Tone says that AUX something.DAT someone.ACC expose ‘Tone is saying that he exposed somebody to something.’ b. Čemu koga? c. ?* Koga čemu? This example shows that sluicing in Slovenian does not exhibit Superiority effects, but rather a parallelism between the antecedent sentence and the sluice. This can be further confirmed with examples in which the sluice does not correspond to an overt indefinite in the antecedent:46 46 While some authors have claimed in the past that ditransitives in Slovenian have two different structures (Marvin and Stegovec 2012), I will assume that the examples in (78) have the same structure prior to movement, since the dative wh-phrase in both only gets the low applicative reading. 81 (79) a. Do svojega until his devetdesetega rojstnega dneva je Miha ninetieth birth AUX Miha prodal vse premoženje. sold all belongings day ‘By his ninetieth birthday Miha had sold all his belongings.’ b. Res? Povej, kaj komu. really tell what.ACC who.DAT ‘Really? Tell me what to whom.’ c. Res? Povej, komu kaj. Given the standard understanding of Sluicing in the sense of Merchant (2001), Sluicing is taken to be the deletion of a TP complement of a C head, which has the +Q and the +wh features, at PF. This however is not compatible with an IP analysis of Slovenian multiple wh-fronting. This problem has also been observed in Grebenyova (2006a) for Russian and Stjepanović (2003) for Serbo-Croatian. Stjepanović shows that Serbo-Croatian exhibits Superiority effects in sluicing. Given this, Stjepanović proposed that in sluicing, wh-phrases move to check the wh-feature in the CP and the TP is deleted (in accordance with Merchant (2001)). Grebenyova (2006a), on the other hand, proposes that in Russian the Focus head can license TP deletion. Because the deletion is licensed by the Focus head, a focused R-expression may be a remnant of sluicing in Russian, as shown in (80), in which a wh-phrase and the R-expressions are remnants in a sluice. Focused R-expressions, however, cannot be remnants in Sluicing in Slovenian, as (81) shows. (80) Russian (Grebenyova 2006a: (13)) a. Ty ne you not pomniš kogda Ivan remember when vstretil Ivan.NOM met Mašu? Mašak.ACC ‘You don’t remember when Ivan met Maša?’ b. Net. Ja ne pomnju no. not remember why I POČEMU SERGEJ Sergej.NOM ‘No. I don’t remember WHY SERGEJ (met) LENA.’ 82 LENU. Lena.ACC (81) a. Ne spomniš se, kdaj je Ivan spoznal not remember2GS self when AUX Ivan.NOM met Majo? Maja.ACC ‘You don’t remember when Ivan met Maša?’ b. * Ne. Ne no not spomnim se, kdaj JANEZ TINO. remember self when Janez Tina Intended: ‘No. I don’t remember when JANEZ (met) TINA.’47 The fact that multiple sluicing in Slovenian does not exhibit Superiority effects, assuming the analysis by Bošković, would imply that wh-movement in Slovenian does not proceed to CP, yet the Focus head does not license TP deletion. It is then not clear how to account for data from Slovenian multiple wh-sluicing under the focus analysis of multiple wh-fronting in Slovenian.48 Arguments against the focus analysis of wh-fronting come from other languages as well. As shown in section 2.2.1, Serbo-Croatian speakers do not agree that Serbo-Croatian has only obligatory multiple wh-fronting. But optionality is not the only property about which one can find disagreement about the Serbo-Croatian data from Bošković (1997a et seq.). Moreover, speakers’ disagreement about data is a constant in theories about multiple wh-fronting. Granted, Bošković (2007) notes that not all speakers of a single wh-fronting language have the same judgements and the distinctions between judgements are subtle. In addition, different speakers of multiple wh-fronting languages report that their language behaves as a typologically different language (for example, their judgements for Serbo-Croatian correspond to the behavior of Bulgarian, i.e. their Serbo-Croatian always has Superiority effects). But Bošković claims that this does not change the fact that multiple wh-fronting 47 Examples like (i) are possible in Slovenian. But since Janez is already discourse given in the antecedent sentence, I do not consider it to be focused, but rather a topic. (i) a. Peter je rekel, da je udaril Roka in Janeza. Peter AUX said that AUX hit Rok.ACC and Janez.ACC ‘Peter said that he hit Rok and Janez.’ b. Ne vem, zakaj Janeza. not know why Janez.ACC ‘I don’t know why he hit Janez.’ 48 Example (81) above shows that focused expressions cannot be the only remnants of sluicing, but Marušič et al. (2014) show that in Sluicing in Slovenian discourse particles can appear in sluices with wh-phrases. Assuming that discourse particles are located in the extended CP in the sense of Rizzi (1997), as Marušič et al. (2014) do, this again indicates that wh-phrases move to the CP and not the IP and that the larger portion of CP can survive sluicing in Slovenian. 83 languages can be divided into three distinct groups and states that speakers’ judgements will fall into one of the three patterns. However, I believe that disagreements about the data within one language weaken Bošković’s proposal. For example, Mihaliček (2010) reports on a different state of affairs for environments which, according to Bošković (2002), exhibit Superiority. She and her informants (speakers of Serbian, Croatian and Bosnian) disagree with Bošković’s ungrammatical judgements, a conclusion Mihaliček reaches on the basis of novel, constructed data, and naturally occurring data (and examples from the literature). For example, out of the three examples below, the first two are cases of embedded questions. The first is from the literature (Godjevac 2000), and the second is an example of naturally occurring data, in this case from a Bosnian dialect (she provides examples from Serbian and Croatian too) and the third is a case of a novel, constructed long distance extraction. (82) Serbo-Croatian (Godjevac 2000: (137c)) Pitam se šta ko sada govori. ask.1SG REFL what.ACC who.NOM who.DAT kome now says ‘I wonder who is saying what to whom now.’ (83) Bosnian (Mihaliček 2010: (14c)) Sarajevo tačno zna gdje je Sarajevo exactly knows where AUX who.NOM what.ACC ko radio, ko je who.NOM done who.NOM AUX odakle ko, i je šta AUX gdje živi. ko from-where and who.NOM where lives ‘Sarajevo knows exactly who is where, who has been doing what, who is from where, and who lives where.’ (84) Serbo-Croatian (Mihaliček 2010: (16b)) Šta ko želite da what.ACC who.NOM want C vam pokloni za rođendan? you give for birthday ‘Who do you want to give you what for your birthday?’ She provides further examples for the lack of Superiority in Serbo-Croatian for correlatives, single long distance extraction out of an embedded clause, sluicing, and questions with the overt complementizer li (see Mihaliček 2010). 84 There is also disagreement about the data from other languages that were taken as evidence in Bošković’s analysis. For example, Bošković (2002) argues for an insitu-group of multiple wh-fronting languages based on the data presented in Stepanov (1998) who shows that there are no Superiority effects in Russian.49 Rojina (2011) (and her informants), on the other hand, disagrees with these judgements and claims that there are restrictions on the order of fronted wh-phrases in Russian, which will be shown in section 3.3.3. For example, she notes that she and her informants can get only multiple pair answers to questions such as (85), see footnote 44, and claims that the unacceptability of (85b) is due to Superiority effects: (85) Russian (Rojina 2011: (62)) a. Kto čto kupil? who what bought ‘Who bought what?’ b. *Čto kto kupil? The question is then how one can deal with this data, if we assume Bošković’s theory. One could simply adopt the line of reasoning in Bošković (2007) and claim that the data Mihaliček (2010) provides corresponds to speakers who fall in the ‘Russian’ category of multiple wh-fronting (i.e. these speakers never exhibit superiority). But, as Mihaliček (2010) writes, the additional data from SerboCroatian indicates, that we do not necessarily need two mechanisms for multiple whquestions within one language. However, it has to be noted that the existence of this data does not exclude the existence of speakers who agree with Bošković’s judgements, but as Mihaliček (2010) notes, the lowered acceptability of examples with violations of Superiority could be accounted with extragrammatical processing factor. But still, such disagreements about data weaken the motivation for Bošković’s analysis, specifically, for the need for two types of movement within one language. In addition, Dyakonova (2009: 188) claims that focus and wh-fronting in Russian exhibit non-uniform behavior in that focused phrases do not have to appear 49 In addition, Krapova and Cinque (2005) show that in Bulgarian strict word order can be found between all wh-phrases. This goes against a prediction by Bošković that Superiority only applies to the first wh-phrase that is wh-moved in Bulgarian (cf. fn. 15 in Krapova and Cinque 2005). 85 in a preverbal position and that Russian allows only one focus per clause. Based on this, it is problematic to take wh-fronting in Russian as an instance of focus fronting. There are also theoretical problems with the focus analysis of multiple whfronting, specifically with the C insertion in non-wh-fronting, as Cheng and Rooryck (2000) point out. As it was mentioned above, one of the central arguments which Bošković (1997c) provides as support for late insertion of C and consequent movement of wh-phrases to IP in cases where there are no Superiority effects, comes from his analysis of French. Bošković claims that environments in which SerboCroatian exhibits Superiority effects are same as the environments in which French exhibits obligatory wh-movement. For example, in French a wh-phrase can stay in situ in matrix questions but not in embedded questions, see examples (48) and (49) in 3.2. Remember that in matrix questions Serbo-Croatian exhibits no Superiority effects and the word order of wh-phrases is rigid in embedded questions. Based on this parallel between Serbo-Croatian and French (and similarities between Japanese and Russian (languages with no wh-movement) vs. English and Bulgarian (languages with wh-movement)), Bošković claims that when C is inserted overtly, wh-phrases front in a rigid word order. Under his analysis, French allows for the insertion of C overtly and the insertion of C together with the [+wh] feature in the LF. The latter option is shown to be problematic. As Cheng and Rooryck (2000) note, the covert insertion of C is problematic since it does not account for the intonation of French in situ questions – if C is only inserted covertly and wh-phrase undergoes feature movement, how does one account for the intonation of in situ question (the same as in yes/no-questions) which is different from the intonation in questions with wh-movement. Assuming the Ymodel of language, one cannot do so. This means that the initial correlation between French and Serbo-Croatian, on which Bošković builds, is problematic from the start. In addition, again as Cheng and Rooryck (2000) note, there are other issues with Bošković’s proposal for feature movement in LF (that is, when C is inserted covertly, only the features of one wh-phrase move to check the C head). Feature movement in LF is blocked by A’-heads (consequently there are no in situ whphrases in embedded clauses – no features can move from an embedded clause), but this does not account for the blocking effects of quantifiers and modals. It is also not clear how Bošković’s analysis would account for presuppositional effects if the 86 difference between in situ questions in French (these require a strongly presupposed context and cannot get a negative answer) and wh-movement is in pied piping. Because of these issues, I will not be assuming the theory of multiple whfronting as proposed by Bošković (1997a et seq.). And while there are other Minimalist accounts of multiple wh-fronting, for example Citko (1998), Grewendorf (2001), Richards (2001), etc., I propose an account of multiple wh-fronting in Slovenian which is based on the Cartographic approach. 3.3 Cartography of multiple wh-questions In sections 3.1 and 3.2, I have shown two approaches to multiple wh-fronting (Rudin 1988, Bošković 1997 and after) and how they can be applied to Slovenian. But, as I have shown, there are several issues with these approaches. Because of this, I turn to a different approach, to Cartography, to propose a different analysis of multiple whfronting in Slovenian. I start the section with an overview of the Cartographic approach and how it has been applied to multiple wh-fronting in two different languages – Bulgarian and Russian. I chose these examples because the two languages behave completely differently according to the approaches shown in the previous section – Bulgarian is said to always exhibit Superiority effects, while Russian never does. However, as I will show below, both languages in fact seem to exhibit Superiority effects (i.e. strict word order of wh-phrases in the clause initial positions) if we look into the order of multiple wh-questions in detail. This leads to the idea that one needs to look at Slovenian examples more closely in order to conclude whether Slovenian does in fact exhibit strict word order of wh-phrases. 3.3.1 Cartography and the Left Periphery in multiple wh-fronting languages Cartography is an approach to syntax that tries to draw precise maps of syntactic configurations, and assumes that languages share both principles of phrase and clause composition and the same functional make-up of phrases and clauses (Cinque 87 and Rizzi 2008). Here I will be concerned with the internal make-up of the CP projection (the Left Periphery), as first described in Rizzi (1997).50 Rizzi (1997) states that the CP acts as an interface between the superordinate structure (e.g. the higher clause) and the content expressed in the IP. This means that the CP expresses information facing the outside (Force) and information facing inside (Finiteness). In addition, the CP can also express topic (as many topics as needed) and focus (one focus per clause in Italian). This Topic-Focus system is only present in the structure when needed and is positioned within the Force-Finiteness system, as I show below: (86) (Rizzi 1997: (8)) … Force …. (Topic) … (Focus) … Fin IP The complementizer system is also the location of wh-phrases. These are, for example, compatible with topics in Italian matrix questions (wh-phrases have to follow topics, cf. Rizzi (1997)), but incompatible with focus, (87). Because of the incompatibility with focus, Rizzi assumes that in Italian matrix questions the whphrase then moves to SpecFoc position of the Left Periphery. In embedded questions, on the other hand, wh-phrases are marginally compatible with embedded focalized elements, (88). The wh-phrase in embedded questions is then located in an independent position (not SpecFoc) (Rizzi 1997: fn.18). (87) Italian (Rizzi 1997: (25)) a. * A to GIANNI che cosa hai detto (, non a Piero)? Gianni what thing did say not to Piero ‘TO GIANNI what did you tell (, not to Piero)?’ b. * Che cosa A GIANNI hai detto (, non a Piero)? (88) (Rizzi 1997: 330, (ii)) ? Mi domando, I wonder A GIANNI che cosa abbiano to Gianni what thing have detto (, non a Piero) say not to Piero ‘I wonder TO GIANNI what they say (, not to Piero).’ 50 See Cinque and Rizzi (2008) for references concerning other syntactic configurations. 88 This approach to movement to the Left Periphery is regulated by Relativized Minimality, a condition which ensures that movement is local. I will show one implementation of this condition and say more about it when discussing the Bulgarian data in the following section and in section 3.4.5. I will also use Bulgarian and Russian to further show how mapping of the CP is achieved and how the Cartographic approach deals with multiple wh-fronting languages. 3.3.2 Bulgarian In section 3.1 I showed, following Rudin (1988), that the word order of wh-phrases in Bulgarian is strict. Rudin’s work was expanded upon by Krapova and Cinque (2005), who compared the acceptability of multiple wh-questions with different word orders of wh-phrases and showed the detailed word order of fronted wh-phrases in Bulgarian. For example, starting with adjuncts, they show that koga ‘when’ necessarily precedes kâde ‘where’, and kâde ‘where’ necessarily precedes kak ‘how’. As expected based on transitivity, they show that koga ‘when’ also precedes kak ‘how’. (89) Bulgarian (Krapova and Cinque 2005: (8)) a. Koga kâde šte hodiš tova ljato? when where will go-you this summer ‘When will you go where, this summer?’ b. * Kâde koga šte hodiš tova ljato? (90) Bulgarian (Krapova and Cinque 2005: (9)) a. Kâde kak si se dâržal? where how are-you behaved ‘Where did you behave how?’ b. * Kak kâde si se dâržal? (91) Bulgarian (Krapova and Cinque 2005: (10)) a. Koga kak si se dâržal? when how are-you behaved 89 ‘When did you behave how?’ b. * Kak koga si se dâržal? The word order of wh-adjuncts proves to be crucial for determining the order of other wh-phrases, as Krapova and Cinque (2005) show. Arguments are placed in different positions with respect to adjuncts and these positions depend on the internal makeup of the arguments. For example, na kogo ‘to whom’ must precede all adjuncts, but na kolko N ‘to how many N’ is placed between koga ‘when’/kâde ‘where’ and kak ‘how’, as shown below: (92) Bulgarian (Krapova & Cinque 2005: (12)) a. Na kogo kâde si daval podarâci? to whom where are-you given presents ‘To whom did you give presents where?’ b. ??? Kâde na kogo si daval podarâci? (93) Bulgarian (Krapova & Cinque 2005: (11)) a. Na kogo kak šte prepodadeš tozi urok? to whom how will teach-you this lesson ‘To whom will you teach this lesson how?’ b. * Kak na kogo šte prepodadeš uroka? (94) Bulgarian (Krapova & Cinque 2005: (14)) a. Koga/kâde na kolko xora si pomagal? when/where to how many people are-you helped ‘How many people did you help when/where?’ b. *? Na kolko xora koga/kâde si pomagal? (95) Bulgarian (Krapova & Cinque 2005: (15)) a. Na kolko xora to how many people kak možeš da pomogneš? how can-you to help ‘How many people can you help how?’ b. * Kak na kolko xora možeš da pomogneš? In addition, they show that arguments can be ordered among themselves. For example, the subject precedes the object wh-phrase in Bulgarian, as already noted by 90 Rudin (1988), see example (4) in section 3.1, but it also holds that the direct object precedes the indirect object, as shown in (96). (96) Bulgarian (Krapova and Cinque 2005: (17)) a. Kogo na kogo šte predstaviš? whom to whom will introduce-you ‘Whom will you introduce to whom?’ b. * Na kogo kogo šte predstaviš? I will not go through all the examples with which Krapova and Cinque (2005) establish the hierarchy of wh-phrases in Bulgarian, but a comparison of acceptability of different word orders of wh-phrases gives results that are summarized in Table 1. D-linked wh- Non-D-linked wh-phrases 51 koj/koja/koe/koi (N) (kogo) (na kogo) (marked)kakvoSub/Obj (marked) kâde/koga kogo (whom) na kogo (to whom) koga (when) kâde (where) kakvoSubj (what) kolkoSubjN (how many) kakvoObj (what) na kolkoObjN (how many) kak (how) Table 1: Word order of wh-phrases in Bulgarian (Krapova and Cinque 2005: Table 3)) The table shows that D-linked phrases precede all non-D-linked phrases (but note that koj ‘who’ precedes even D-linked phrases), and wh-objects (non-D-linked) that are specified as [+human] (kogo ‘whom’ and na kogo ‘to whom’) precede adjunct wh-phrases for time and place. The latter are then followed by non-D-linked wh-subjects and these precede wh-objects, which are both specified as [–human] (kakvo ‘what’) or are underspecified for [human] (kolko ‘how many’). All whphrases are followed by kak ‘how’. What this table does not show is an additional layer of D-linked wh-phrases, which come with resumptive clitics. These must precede all other wh-phrases and can occur before a parenthetical, as in (97) (for more see Krapova and Cinque 2005). 51 Koj/koja/koe/koi (N) are ‘which’ phrases and kogo ‘whom’ and na kogo ‘to whom’ in this column refer to D-linked uses of these wh-phrases. 91 (97) (Krapova and Cinque 2005: (42a)) Koja kartina, spored tebe, kâde *(ja) e risuval tozi xuždonik? which painting, according to you, where it is painted this artist ‘According to you, which painting did this artist paint where?’ According to Krapova and Cinque (2005), this word order reflects the word order of wh-phrases before movement to the Left Periphery. The underlying word order is preserved due to a requirement that only the chain as a whole counts as an intervener, (98), where ‘chain’ is defined as in (99). This requirement on chains is formulated based on Relativized Minimality as proposed by Rizzi (2001b), (100). (98) (Krapova and Cinque 2005: (58)) Only a whole chain, not just a link of a chain, counts as an ‘intervener’. (99) (A1, …, An) is a chain iff, for 1 ≤ i < n (i) Ai = Ai+1 (ii) Ai c-commands Ai+1 (iii) Ai+1 is in a Minimal Configuration with Ai (100) Y is in a Minimal Configuration with X iff there is no Z such that (i) Z is of the same structural type as X, and (ii) Z intervenes between X and Y With (98) they essentially subsume the notion Superiority under the notion of Relativized Minimality. Superiority violations, as in (101b) are accounted for, since the chain as a whole intervenes between the trace of kak ‘how’ and the target kak ‘how’. Example (101a), on the other hand, does not violate the condition, since there are only links of the chain intervening between the trace of kak and the target kak – which is not a problem under (98).52 52 Where one finds free word order among wh-phrases in Bulgarian, as with D-linked phrases, which I am leaving aside here, Krapova and Cinque (2005) argue that movement of the phrase is triggered by a special feature which the intervening element does not have. Hence no violation of Relativized Minimality arises, because the feature of the intervener is distinct from the target. 92 (101) Bulgarian (Krapova and Cinque 2005: (25)) a. Kakvo kak šte napraviš? what how will do-you ‘What will you do how?’ b. * Kak kakvo šte napraviš? (102) a. [CP kakvo [+wh] b. * [CP [CP kak [IP šte napraviš [+wh] kak [CP kakvo [+wh] [+wh] t t [+wh] [IP šte napraviš t t [+wh] To summarize: The word order of wh-phrases prior to movement is crucial for word order of wh-phrases in the Left Periphery in Bulgarian. When word order is preserved, the question is acceptable and when it is violated, the question is unacceptable. Krapova and Cinque (2005) account for this with a condition on chains according to which only the whole chain acts as an ‘intervener’, (98). 3.3.3 Russian While it has been claimed that Russian does not exhibit Superiority effects (e.g. Stepanov 1998), a closer look at the orderings of different wh-phrases gives different results. This was shown by Rojina (2011), who gathered judgements of twenty Russian speakers in two separate experiments – in the first she tested matrix questions and in the second embedded questions, in order to examine the word order of wh-phrases. While she notes that speakers claim that with the order of adjuncts and objects the order depends on which information is more prominent, Rojina also shows that some Superiority effects arise.53 For example, in matrix questions the 53 Granted, there is a problem with how Rojina (2011) reports the data, since she at times speak about preferences and at times just reports on grammaticality and does not give quantitative results of her study. However, since she marks examples as either acceptable or not, I will also take them as such. 93 subject must always precede the object (regardless whether the subject is [+human] or not, (103) and (85) in 3.2.2, or D-linked or not, (104)): (103) Russian (Rojina 2011: (117)) a. Kto uvidel? kogo who.NOM who.ACC saw ‘Who saw whom?’ b. * Kogo kto uvidel? (104) Russian (Rojina 2011: (138)) a. Kakoj malčik uvidel? kogo which boy.NOM who.ACC saw ‘Which boy saw who?’ b. * Kogo kakoj malčik uvidel? On the other hand, while a [+human] object must proceed kak ‘how’, (105), there is no such strict ordering for a [+human] object with respect to wh-adjuncts, as shown below for gde ‘where’, (106), in matrix clauses. The latter case, however, is subject to preference – there is a preference of placing object before adjunct, but speakers also report that the ordering depends on which piece of information is more prominent. A similar situation can also be found with [–human] objects (see Rojina 2001: 60, 61). (105) Russian (Rojina 2011: (106)) a. * Kak komu how who.DAT ty pomog? you helped ‘Whom did you help and how?’ b. Komu kak ty pomog? (106) (Rojina 2011: (107)) a. ? Gde where kogo ty poslednij who.ACC you last ‘Who did you see last time and where?’ b. Kogo gde ty poslednij raz videl? 94 raz videl? time saw Rojina also shows that there are Superiority effects with D-linked phrases in matrix questions in Russian. For example kto ‘who’ must be higher than D-linked elements, (107), and D-linked subjects precede [+human] non-D-linked wh-objects, (104) above. (107) Russian (Rojina 2011: (134)) a. Kto kakuju studentku who.NOM which student budet ekzamenovat? will examine ‘Who will examine which student?’ b. * Kakuju studentku kto budet ekzamenovat? Comparing different word orders, Rojina (2011) suggests that wh-phrases are ordered as shown in Table 2. The subject must be placed before the [+human] object or the D-linked object, which is placed before the [–human] object and adjuncts of place or time. And everything precedes the manner object kak ‘how’. Subject [+human] Object D-linked Object [-human] Object Adjunct (place, time) Adjunct (manner) kto (who) čto (what) kakoj/kakaja N (which N) komu (whom) kogo (whom) kakoj/kakaja N (which N) kogda (when) gde (where) kak (how) čto (what) skol‘ko N (how much N) PP object kak (how) Table 2: Russian - matrix questions (Rojina 2011: 65, Table 6) In embedded questions the word order of wh-phrases is stricter. For example, in embedded questions, wh-phrases for direct [+human] objects must precede [+human] wh-phrases for indirect objects, (108). A strict word order is also established between čto ‘what’ and wh-adjuncts, I show this in (109) for čto ‘what’ and gde ‘where’, and with D-linked and non-D-linked phrases. 95 (108) Russian (Rojina 2011: (151)) a. Ona sprosila, kogo s she who.ACC with whom asked kem ja sputal. I mixed ‘She asked who I confused with whom.’ b. * Ona sprosila, s kem kogo ja sputal. (109) Russian (Rojina 2011: (154)) a. * Ona sprosila, she asked, ty videl. gde čto where what.ACC you saw ‘She asked what you saw and where.’ b. Ona sprosila, čto gde ty videl. By comparing different word orders of wh-phrases in embedded wh-questions, Rojina suggests the following word order: Subject D-linked object Direct obj [+hum] Indir. obj [+hum] Object [–hum] PP object Adjunct (place/ time) Adjunct (manner) kto (who) čto (what) kakoj N (whichN) kakoj N (which N) kogo (who) komu (to who) skolko N (how much N) čto (what) o čëm (about what) s kem (with whom) gde (where) kogda (when) kak (how) Table 3: Embedded wh-questions in Russian (Rojina 2011: 74, Table 9) These results are, to some extent, similar to what was discussed in Meyer (2004) who shows that speakers of Russian, Polish and Czech (which were previously assumed to be languages without Superiority effects) prefer some word orders over others. Meyer conducted a magnitude estimation study that provided evidence that Russian speakers prefer the word order in which the wh-subject precedes the wh-object. This is slightly more noticeable when the wh-object is [+animate] but D-linking also has an influence: the wh-subject precedes the whobject more easily if the wh-subject is not D-linked, regardless of the object. Meyer also finds a weak preference for the word order wh-subject > kak ‘how’, but contrary to Rojina, Meyer finds no influence of embedding. However, as Rojina (2011) also 96 points out, Meyer did not test all the combinations that Rojina did, so different conclusions about the influence of embedding might arise because of different materials.54 Before I return to the derivation of the word order of wh-phrases in multiple wh-questions in Russian, a couple of more things have to be noted. First, Rojina does not adopt the analysis according to which wh-phrases move for focus reasons and in which they move to the IP, see section 3.2. Rojina rather proposes that wh-movement proceeds to check strong wh-features and claims that wh-phrases move to the Left Periphery of the sentence, which is the CP. In the Left Periphery, wh-phrases in Russian can co-occur with focus phrases, as in (110) and it is because of this that Rojina proposes that focused and wh-phrases do not compete for the same positions. In addition she shows that focused phrases can either precede or follow wh-phrases, (110). The topic, on the other hand, can never precede all wh-phrases, but it can appear between wh-phrases or follow all of them in multiple wh-questions, (111).55 (110) (Rojina 2011: (189), (190)) (Context: two people talking about presents for children. One of them is asking: ‘What did you buy for the kids, especially for Masha?’ The other replies that he got a car for Peter, a doll-house for Kate etc., not mentioning Masha first. The first speaker is impatient and is asking:) a. b. Da MAŠE čto to MASHA what you bought, ty kupil, (MAŠE) čto (MAŠE) ty MASHA.DAT what MASHA.DAT you a ne drugim? but not to others kupil, a ne drugim? bought, but not to others (111) (Rojina 2011: (191)) a. Kogda Maša komu zvonila? when Masha whom called 54 What is especially interesting for us is that Meyer also finds that in Czech and Polish speakers prefer the ‘subject > object’ word order, but only when both wh-phrases are animate. I return to this when discussing Slovenian. 55 Example (111b) shows that ‘Maša’ in (111a) is the Topic – following Rizzi (1997) Topics cannot be quantificational elements. 97 b. * Kogda kazhdyj rebenok komu zvonil? when every child whom called c. * Maša kogda komu zvonila? when whom called kogda zvonila? Masha when called Kogda Maša zvonila? Masha d. ?? Maša e. Based on this, Rojina proposes that the Left Periphery of multiple wh-questions in Russian has the structure shown in (112): (112) Force . . . (Wh-P) . . . (Topic) . . . (Wh-P) . . . (Focus). . . (Wh-P) . . . Fin IP The order of wh-phrases in the Left Periphery is strict, as I have shown above, and can be summarized as in (113) for embedded questions and as in (114) for matrix questions. (113) (Rojina 2011: 88) [ForceP [FocP [SubjP [d.obj +h [ind.obj+h[obj –h [PP obj [Adjunct-adv [Manner adv]]]]]]]]]]] (114) (Rojina 2011: (195)) [ForceP [FocP [SubjP [[+hum]ObjP [ObjP, AdjunctP[ Manner AdjunctP]]]]]] Based on Krapova and Cinque (2005), Rojina proposes that word order of whphrases prior to movement is preserved after movement (but at the same time, it is also regulated by prominence and internal make up, for example [+human] precedes [–human]). Rojina (2011) describes Russian as a free word order language, but the default word order of arguments is the one shown in (115) – the subject precedes the direct object and the direct object precedes the PP object. There is no preference for an adjunct of place over time which is shown in (116) (cf. Table 2 and Table 3, the two are marked as Adjunct-adv and AdjunctP in (113) and (114), respectively) and the manner adjunct must precede all other adjuncts. However, with respect to manner adjuncts, such as gromko ‘loudly’ in (117), Rojina adopts an analysis in Nilsen 98 (2000) and Cinque (2002) according to which a manner adjunct is merged lower in the structure and is then moved to a higher position. (115) Russian (Rojina 2011: (177)) Ivan pročёl stat‘ju o beţencax. Ivan read article about refugees ‘Ivan read an article about refugees.’ (116) Russian (Rojina 2011: (181)) Ja ego videl (včera) v universitete I him saw yesterday in university (včera). (yesterday) ‘I saw him yesterday at the university.’ (117) Russian (Rojina 2011: (182)) On gromko razgovarival (gromko) v sosednej komnate (*gromko). he loudly spoke (loudly) in next room (loudly) ‘He spoke loudly in the next room.’ The word order of wh-phrases prior to movement is preserved. The preservation in the Left Periphery is achieved with a requirement for chain crossing. An example of a derivation is shown in (119) for the embedded question in (118): (118) Russian (Rojina 2011: (192)) Ona interesovalas, komu Pavel kak pomog. She wonder who.DAT Pavel how help object [+h] manner adjunct ‘She wondered whom Pavel how helped.’ (119) [ForceP [FocP [WhP komuk [TopP Pavel [WhP kaki [… [IP[… [ObjP tk [MannerP ti … If multiple wh-questions in which chains are crossed are acceptable, we can in turn also expect that when the chains are not crossed, the multiple wh-question is ungrammatical. This is borne out, as we can observe if we compare the ungrammatical example (120) in which the order of wh-phrases is kak ‘how’ followed by komu ‘whom’, to (118). 99 (120) Russian (Rojina 2011: (142a)) * Ona interesovalas‘, she wondered kak komu ty pomog. how whom you helped ‘She was wondering whom I helped and how.’ What we can take away from the Bulgarian and the Russian data and the analysis is that both languages seem to preserve the order of phrases prior to movement in the Left Periphery. Given this, we can expect other multiple whfronting languages to behave similarly. I will explore whether this is the case in Slovenian in the next chapter. 3.4 Word order in Slovenian multiple wh-questions To determine the order of wh-phrases in Slovenian I have collected data from 16 Slovenian speakers with a survey. The survey included 60 items. Out of these 60 items, 32 were fillers (examples with sluicing) and 28 were multiple wh-questions which are of interest here. The stimuli (multiple wh-questions) were composed based on Meyer (2004) who hypothesizes that a language either exhibits a preference for the subject > object or object > subject order or else exhibits no preference in the order of wh-phrases. If a language exhibits such a preference, it is further influenced by animacy, main clause/embedded clause distinction and D-linking. Because of this, word order preference between subjects and objects in matrix and embedded questions when the two arguments have the same or different value for [human]56 and when one of them is D-linked or neither is D-linked were tested. In addition, the word order or adjunct wh-phrases with respect to argument wh-phrases were tested (based on the findings presented in Rojina (2011) and Meyer (2004)). It was also 56 The wh-word kdo ‘who’ is taken as [+human], rather than [+animate] in Slovenian and kaj ‘what’ to [–human] rather than [–animate] since they are taken to question human and nonhuman entities, respectively (cf. Toporišič 2001: 276). However, the declension of kdo ‘who’ coincides with the declension pattern of [+animate] nouns and the declension of kaj ‘what’ with the declension pattern of [–animate] nouns. This means that effects of animacy of wh-phrases and effects of the [human] feature of wh-phrases are comparable. 100 taken into consideration whether receiving a single pair or pair list answer has an influence on the word order of wh-phrases. Based on the data presented in section 3.2.1, a lack of strict word orders of wh-phrases in Slovenian was predicted. The survey confirmed the prediction as it showed that there is no strict word order between wh-phrases in multiple wh-questions in Slovenian. Still, there are some preferences which are described in the following sections. In the survey, speakers were asked to choose an acceptable sentence for a specific context they were given. They could choose between two possible orders of wh-phrases and two additional possibilities: ‘both options are equally acceptable’ and ‘neither option is acceptable’. In addition, speakers could suggest an acceptable sentence. An example is shown below: (121) Context: Shranjene imam vse podatke o Evroviziji. ‘I have all the information about the Eurovision Song Contest saved.’ Responses: a. Zanima me, kdo je kdaj zastopal wonder I.DAT, who when represented Slovenia AUX Slovenijo? ‘I wonder who represented Slovenia when.’ b. c. Zanima me, kdaj je kdo zastopal Slovenijo? wonder I.DAT, when AUX who represented Slovenia Obe možnosti sta nesprejemljivi. ‘Both options are unacceptable.’ d. Obe možnosti sta enako sprejemljivi ‘Both options are equally acceptable.’ e. Other: _________________________ In the provided context there was no explicit mentioning of the adjuncts/arguments that later appeared as a wh-phrase in the questions in order not to influence the word order in the possible responses. In addition to the word order between the two wh-phrases, I was also looking at how these wh-phrases were placed with respect to other elements typically found in 101 the Left Periphery. The data that explores these placements was not tested in the experiment, which means the judgements are my own. 3.4.1 The order of wh-objects and wh-subjects Based on section 3.2.1, no Superiority effects (i.e. no strict word order) between subjects and objects were expected in Slovenian. However, I found non-uniform behavior of argument wh-phrases. Based on this, I divided the results into several categories. I start with matrix questions in which both subject and object were [+human]. The results for these questions are shown in the following table: subject > object object > subject both neither Single-pair 13 1 2 0 Pair-list 13 0 3 0 26 1 5 0 81,2% 3,1% 15,6%, Sum 57 Total: 32 Table 4: Word order of [+human] subjects and objects in matrix questions When both the subject and the object wh-phrases were [+human], speakers preferred the subject to precede the object, (122). The same results held both in environments that evoked pair list and single pair answers. Still, I mark the ‘object > subject’ word order with ‘??’ because there were speakers who claimed that both word orders were equally acceptable. (122) V soboto je bil prvi krog državnega tekmovanja v šahu, po katerem je izpadla polovica tekmovalcev. ‘The first round of the national chess competition took place on Saturday, after which half of the contestants were eliminated.’ 57 In the Sum line I give the sum of all the answers, in the Total line I give the results in percents and exclude the number of speakers who do not accept any of the offered word orders. 102 a. Kdo je koga premagal? who.NOM AUX who.ACC beat ‘Who beat whom?’ b. ?? Koga je kdo premagal? Superiority effects were not detected when the subject wh-phrase was [+human] and the object was not, as the Table 5 shows. One such example is in (123). subject [+h] > object [–h] object [–h] > subject [+h] both neither Single-pair 4 0 12 0 Pair-list 4 4 5 1 Sum 8 4 17 1 Total: 29 27,6% 13,8% 58,6% Table 5: Word order of [+human] subjects and [–human] objects in matrix questions (123) V soboto bo praznovanje slovenskega kulturnega praznika, na katerem bo več nastopov. ‘The celebration for the Slovenian cultural holiday will happen on Saturday. There will be several performances.’ a. Kdo bo kaj recitiral? who.NOM AUX what.ACC recite ‘Who will recite what?’ b. Kaj bo kdo recitiral? These results are similar to the results for Polish and Czech as described in Meyer (2004), who calls the preference of the ‘subject > object’ word order with animate arguments Selective Superiority. In Polish and Czech this preference only occurs with animate subjects and objects, but not with inanimate objects (the same goes for Slovenian). In addition, Meyer describes the phenomenon as a ‘reverse animacy effect’ since it goes against expectations that subjects precede objects and animates precede inanimates. Meyer offers no true explanation and attributes the effect to an independent factor. The phenomenon has also been observed in German, 103 in which the object wh-phrase can move over the subject wh-phrase but only when the animacy of the two is different (Fanselow 2010, Fanselow et al. 2013): (124) German (Fanselow 2010: (5)) a. Wer hat who.NOM has wen gesehen? who.ACC seen ‘Who has seen who?’ b. # Wen hat wer gesehen? Fanselow (2010) and Fanselow et al. (2013) argue against a grammatical constraint which would rule out this type of movement. Rather, they propose that the ‘subject > object’ word order is preferred due to a processing difficulty that occurs with movement of similar phrases (see Fanselow 2010, Fanselow et al. 2013 for more). Based on Meyer (2004), it is not clear how Polish and Czech behave with respect to Superiority in embedded questions, but it seems that they exhibit Selective Superiority in embedded contexts as well (Meyer writes for Polish: “Embedding had no influence on the Superiority-like effect, which shows that Polish wh-questions are typologically distinct from Serbo-Croatian ones, as analyzed in Bošković (1998)” (Meyer 2004: 52) and makes a similar observation for Czech (cf. Meyer 2004: 53)). In Slovenian embedded questions, the influence of the same value for [human] was weaker, yet ‘subject > object’ word order was still preferred. The results of testing examples in which both wh-arguments of an embedded wh-questions are [+human], (125), and examples in which only the wh-subject is [+human], (126), are summarized in Table 6 and Table 7, respectively. (125) V soboto bo družabni večer s plesom. Zanimajo nas udeleženci tega večera. ‘There will be a dance on Saturday. We want to know who will attend it.’ a. Janez vpraša, kdo bo koga povabil. Janez asks who.NOM AUX who.ACC invite ‘Janez is asking, who will invite whom.’ b. Janez vpraša, koga bo kdo povabil. 104 (126) Včeraj se je zgodila velika eksplozija. Ljudi zanima vse o dogodku. ‘Yesterday there was a big explosion. People want to know everything about the event.’ a. Novinarji sprašujejo, kdo je kaj journalists ask AUX what.ACC blow-up who.NOM razstrelil. ‘The journalists were asking who blew up what.’ b. Novinarji sprašujejo, kaj je kdo razstrelil. subject [+h] > object [+h] object [+h] > subject [+h] both neither Single-pair, [+h] 10 0 6 0 Pair-list [+h] 9 0 6 1 Sum 19 0 12 1 Total: 31 61,2% 0% 38,7% Table 6: Word order of [+human] subjects and [+human] objects in embedded questions subject [+h.] > object [–h.] object [–h.]> subject [+h.] both neither Single-pair 4 5 6 1 Pair-list 4 3 9 0 Sum 8 8 15 2 Total: 31 25,8% 25,8% 48,4% Table 7: Word order of [+human] subjects and [–human] objects in embedded questions We can observe that many speakers still prefer the ‘subject > object’ word order in cases where both arguments are human, but there are a lot of speakers that find both orders (‘subject > object’ and ‘object > subject’) equally acceptable. We can also take this as an indicator that the ‘subject > object’ preference with two [+human] arguments is not due to a grammatical constraint (assuming that a grammatical constraint would not disappear in an embedded clause) but due to an independent factor (see above). The results more balanced in cases with a [+human] subject and a [–human] object. Here the majority of speakers found both ‘subject > object’ and ‘object > subject’ word order equally acceptable. Crucially, because we find both patterns in embedded questions, I conclude that there is no fixed word order between subject and object wh-phrases in embedded multiple wh-questions. 105 Also tested were the preferences in word order of D-linked arguments with respect to non-D-linked arguments. The results are given in Table 8. subject > object object > subject both neither wh-subject D-linked 6 2 4 3 wh-object D-linked 9 0 6 158 Sum 15 2 10 4 Total: 27 55,6% 7,4% 37,0% Table 8: Word order of D-linked arguments w.r.t. non-D-linked arguments in matrix questions When the subject was D-linked and the object was not, (127), the preferred word order was ‘subject > object’ in matrix questions but speakers also found both ‘subject > object’ and ‘object > subject’ word orders equally acceptable. There was no preference for D-linked wh-arguments to be placed before non-D-linked wharguments, given that no speakers chose ‘D-linked object > non-D-linked subject’ word order. Given this, I conclude that there is no strict word order between D-linked and non-D-linked arguments in matrix clauses but that the ‘subject > object’ order is preferred. (127) Pripravljamo jedilnik za živali v Živalskem vrtu Ljubljana. ‘We are preparing the menu for the animals at Ljubljana Zoo.’ a. Katera žival bo kaj jedla? which animal.NOM AUX what.ACC eat ‘Which animal will eat what?’ b. 58 Kaj bo katera žival jedla? One speaker offered a different word order (i.c): (i) Prejšnji teden smo organizirali večerjo za slovenske diplomate v tujini. ‘Last week we organized a dinner for Slovenian diplomats abroad.’ a. Kdo je katerega diplomata povabil? who.NOM AUX which diplomat.ACC invited ‘Who invited which diplomat?’ b. Katerega diplomata je kdo povabil? c. Kdo je povabil katerega diplomata? 106 In embedded questions with D-linked and non-D-linked wh-phrases, as (128) in which the object is D-linked, no clear preference could be detected: subject > object object > subject both neither wh-subject - D-linked 5 2 8 059 wh-object - D-linked 2 8 3 260 Sum 7 10 11 1 Total: 28 25,0% 35,7% 39,3% Table 9: Word order of D-linked arguments w.r.t. non-D-linked arguments in embedded questions (128) Lani sem organizirala pesniški nastop za izbrane pesnike. Komisija je prejela nekaj kontroverznih kandidatov. ‘I have organized a performance for selected poets. The judges accepted some controversial candidates.’ a. Sprašujem se, wonder kdo self who.NOM je katerega pesnika izbral. AUX which poet.ACC select ‘I wonder who selected which poet.’ b. Sprašujem se, katerega pesnika je kdo izbral. c. Sprašujem se, kdo je izbral katerega pesnika. With D-linked phrases in embedded questions, speakers also proposed their own answers, leaving either the D-linked or the non-D-linked phrase in situ, as I show in (128c), which was a response to (128), and in (129c) – while both of these two examples are embedded, the same also happened in a matrix question (see fn. 58). (129) Romane je pisalo veliko slovenskih pisateljev. ‘A lot of Slovenian writers wrote novels.’ a. Zanima me, kateri pisatelj je wonder I.DAT which writer.NOM AUX what.NOM wrote kaj ‘I wonder which writer wrote what.’ 59 60 One speaker proposed another word order which is shown in (128c). Two speakers proposed a different word order which is shown in (129c). 107 napisal. b. Zanima me, kaj je kateri pisatelj napisal. c. Zanima me, kateri pisatelj je napisal kaj. I also tested whether the type of expected answer (i.e. pair list or single pair) has an effect on the word order of the two wh-phrases in Slovenian matrix or embedded wh-questions. The type of answer was controlled with specific contexts. For example, in (126) above, which is an embedded question, the felicitous answer is a single pair answer, and, as we have seen, no Superiority was found. On the other hand, the felicitous answer to (123), which is a matrix question, is a pair list answer. Again, the type of answer does not lead to Superiority effects. To summarize: In Slovenian we can observe selective Superiority, which holds in matrix wh-questions when subject and object wh-phrases are both human. In this condition the [+human] subject precedes the [+human] object. In embedded questions this preference is also observed, but to a lesser extent. When subject and object have different values for human, no Superiority effects are found. When either subject or object is D-linked, no Superiority effects can be found (but the ‘subject > object’ order is preferred). In embedded questions, again no Superiority effects can be found. The word order of wh-phrases is also unaffected by the type of answers (single pair vs. pair list). Based on the presented results, I conclude that there are no ‘real’ Superiority effects between wh-subjects and wh-objects in Slovenian. 3.4.2 The order of wh-adjuncts with respect to wh-subjects I divide the results of testing word order of a wh-adjunct with respect to the whsubject into two groups: the order of zakaj ‘why’, kako ‘how’, kje ‘when’, and kdaj ‘where’ with respect to the subject in matrix questions and the order of zakaj ‘why’, kako ‘how’, kje ‘when’, and kdaj ‘where’ with respect to the subject in embedded questions. Table 10 shows the results of the questionnaire exploring the word order of a wh-adjunct with respect to the wh-subject in matrix questions. 108 subject > adjunct adjunct > subject both neither wh-subject - why 2 9 2 3 wh-subject- how 3 10 2 1 wh-subject - when 2 10 4 0 wh-subject- where 5 7 4 0 Sum 12 36 12 4 Total: 60 20,0% 60,0% 20,0% Table 10: The word order of wh-adjuncts w.r.t. the wh-subject in matrix questions Checking the influence of word order of the wh-subject with respect to whadjuncts has again shown that we cannot find real Superiority effects (i.e. strict word order of wh-phrases) in Slovenian, but rather just preferences. For example in matrix questions, the majority of speakers preferred for zakaj ‘why’, (130), kako ‘how’, (131), and kdaj ‘when’, as in (132), to precede the subject wh-phrase. The preference for ‘adjunct > subject’ word order was slightly weaker with kje ‘where’, (133). (130) Minister že cel dan bere pritožbe o novem davku. ‘The minister has been reading complaints about the new tax all day long.’ a. Zakaj je kdo poslal pritožbo? why AUX who send complaint ‘Why did who send the complaint?’ b. ? Kdo je zakaj poslal pritožbo? (131) Že cel teden zbiramo rešitve matematične domače naloge. ‘We have been collecting the solutions for the math homework all week long.’ a. Kako je kdo rešil nalogo? how AUX who solved task ‘How did who solve the task?’ b. ? Kdo je kako rešil nalogo? (132) Zbiramo podatke o vseh tekmovanjih za roman leta. ‘We are collecting the information about every contest for the novel of the year.’ 109 a. Kdaj je kdo zmagal? when AUX who won ‘Who won when?’ b. ? Kdo je kdaj zmagal? (133) Prinesli smo gobe iz različnih slovenskih gozdov. ‘We brought mushrooms from different Slovenian forests.’ a. Kje je kdo nabral gobe? where AUX who pick mushrooms ‘Who picked the mushrooms where?’ b. Kdo je kje nabral gobe? The overall preference of placing a wh-adjunct before the wh-subject can also be observed in embedded questions. This is shown Table 11. subject > adjunct adjunct > subject both neither wh-subject - why 1 11 3 1 wh-subject - how 3 10 2 1 wh-subject - where 2 6 7 1 wh-subject- when 3 4 9 0 Sum 9 31 21 3 Total: 61 14,8% 58,8% 34,4% Table 11: The word order of wh-adjuncts w.r.t. the wh-subject in embedded questions In embedded questions, most speakers preferred placing the wh-adjuncts zakaj ‘why’, (134), and kako ‘how’, (135), before the subject. On the other hand, there is no such preference when kje ‘where’, (136), and kdaj ‘when’, (137), precede the whsubject in embedded questions. In this case most speakers found the two possible word orders equally acceptable. (134) Ravnatelj že cel dan piše obvestila o ukorih. Na teh bo utemeljil vse kazni. ‘The headmaster has been writing notifications about reprimands the whole day long. He is going to substantiate all the punishments.’ 110 a. Starše zanima, zakaj je kdo dobil kazen. parents wonder why AUX who get punishment ‘Parents wonder who got punished why.’ b. ?? Starše zanima, kdo je zakaj dobil ukor. (135) Že kakšno leto vemo vse o krajah gesel na spletni strani. Lahko vam pomagamo pri vaši preiskavi. ‘We have found out everything about the password thefts on the web page. We can help you with the investigation.’ a. Zanima me, kako je kdo kradel podatke. wonder I.DAT how who steal information AUX ‘I wonder who has stolen the information how.’ b. ? Zanima me, kdo je kako kradel podatke. (136) Napisali smo seznam slavnih bitk. ‘We have written down a list of famous battles.’ a. Ugotovite, kdo je kje premagal nasprotnike. guess where beat who AUX enemy ‘Guess who beat the enemy where.’ b. Ugotovite, kje je kdo premagal nasprotnike. (137) Shranjene imam vse podatke o Evroviziji. ‘I have all the information about the Eurovision Song Contest saved.’ a. Zanima me, kdo je kdaj zastopal wonder I.DAT, who when represented Slovenia AUX Slovenijo. ‘I wonder who represented Slovenia when.’ b. Zanima me, kdaj je kdo zastopal Slovenijo. To summarize the results: In matrix and embedded questions there is a general preference for wh-adjuncts to precede the wh-subject. In embedded questions there is no such preference with kdaj ‘when’ and kje ‘where’. Still, despite these preferences, we cannot claim that a strict word order of wh-phrases in multiple wh-questions in Slovenian exists, since speakers also accept different word orders. This means that word order of wh-phrases is free, but there are some preferences. Assuming transitivity holds, we can also expect a preference for the ‘adjunct > object’ word 111 order, since we observed a slight preference of ‘subject > object’ word order for arguments. Still, something needs to be said about the word order in multiple wh-questions with zakaj ‘why’ and kako ‘how’, since there was a preference to place these whadjuncts before the wh-subject in both matrix and embedded questions in Slovenian. I will not go into detail about these two adjunct wh-phrases, but before proceeding, two things need to be noted about the data and the results of the survey. First I need to emphasize that the results might seem strongly in favor of the word order zakaj ‘why’/kako ‘how’ > ‘all other wh-s’, but this is still a preference, as there were speakers who chose both word orders or the reverse order to be possible. In addition, we can find evidence for the reverse order in naturally occurring data. I give examples from Gigafida (a written corpus of the modern Slovenian language which contains texts from different genres that amount to 1,2 billion words, see Gigafida for more) in which the wh-subject precedes zakaj ‘why’ and kako ‘how’ below (but again, word order ‘why/how > subject’ is more common in the corpus too): (138) (Gigafida, both examples taken from 24ur.com, 2010) a. Kdo je zakaj komunist who why communist AUX omejeval in nadzoroval limit and control - pripadnik gnilega režima, ki follower rotten regime, that AUX o je življenja in njih odločal? lives and about them decide ‘Who is a communist and why – a follower of a rotten regime which has limited and controlled lives and decided about them?’ b. Pa pol da se vidi kdo je and then that self see who AUX kako nastopil ... who perform ‘And then we will see who preformed and how…’ Therefore we are again speaking about a preference and not a single grammatical option regarding word order. In addition, the cases that were used in the survey were all examples in which zakaj ‘why’ can primarily be interpreted as a ‘reason why’. While the purpose interpretation of why is possible (the exception being (134)), it is less obvious. This 112 might have influenced the results, as different behavior between the two types of why has been noticed in the past. That is, in Slovenian, zakaj ‘why’ can be interpreted either as the reason why or the purpose why (to which I will be from now on referring as whyR and whyP, respectively). Because of this, a question such as (139a) can receive either (139b) or (139c) as a potential answer.61 On the other hand, in some languages, such as Russian, the two different types of ‘why’ are expressed with two different wh-words, as the examples below show. (139) a. Zakaj si šel v kino? why go in cinema AUX ‘Why did you go to the cinema?’ b. Ker je doma pregorela varovalka. because AUX home burn fuse ‘Because the fuse has blown at home.’ c. [Reason] Da bi pogledal novo slovensko komedijo. that AUX look new Slovenian comedy ‘In order to see the new Slovenian comedy.’ [Purpose] (140) Russian (Stepanov and Tsai 2008: (52)) a. Začem P why Ivan sjuda prišel? Ivan here came ‘For what purpose did Ivan come here?’ b. Čtoby kupit’ pivo. in-order-to buy beer ‘In order to buy beer.’ (141) Russian (Stepanov and Tsai 2008: (53)) a. Počemu Ivan sjuda prišel? whyR Ivan here come ‘Why did Ivan come here?’ b. Potomu čto emu bylo skučno. because him was boring 61 The two can potentially be substituted with s katerim namenom ‘with which purpose’ and s katerim razlogom ‘with which reason’. I will be leaving the two aside. The Slovenian equivalent to the SerboCroatian zbog čega ‘whyP’, that is zaradi česa ‘why’, can be used as a reason or purpose why. 113 ‘Because he was bored.’ Stepanov and Tsai (2008) notice that in Russian the reason and purpose why behave differently. For example, whyR is not able to appear in multiple wh-questions, while whyP is able to do just that.62 In Serbo-Croatian, on the other hand, both types of why can appear in multiple wh-questions. (142) Russian (Stepanov and Tsai 2008: (58), (64)) a. Kto začem prišel? who whyP came b. * Kto počemu who whyR c. ?* Počemu R why prišel? came kto prišel? who came Because we have not accounted for the two different types of zakaj ‘why’ in the survey, it is hard to draw any conclusions about their behavior in Slovenian. We have also not accounted for the two different readings of kako ‘how’ – the instrumental and the manner reading, so again it is hard to make any conclusions about the position or movement for the two adjuncts. The same also holds for the results of the survey presented in the next section. Because the understanding of zakaj ‘why’ and kako ‘how’ is not crucial for the larger goal of this thesis and because there is insufficient data, I will leave the two aside from now on and focusing on multiple wh-questions with the object wh-phrases and wh-adjuncts.63 62 Rojina (2011) again makes a different observation and claims that začem ‘whyP’ can appear in any multiple wh-question (except with kak ‘how’) while počemu ‘whyR’ can only do so with arguments. 63 However, Stepanov and Tsai (2008, see also and the references cited therein) have examined why and how and have proposed that whyR is base generated in the CP and whyP is base generated in the vP, below the NegP. This is confirmed, for example, by the fact that in Russian whyR can be used in negative questions while whyP cannot. Rizzi (2001a), on the other hand, argues that in Italian perché ‘why’ in matrix questions is base generated high, in the specifier of Interrogative Projection (IntP, beneath Force, above Focus in the Left Periphery) or moves to it from an embedded clause. Shlonsky and Soare (2011) argue that it is merged in ReasonP, which is above NegP, adverbs and possibly the subject, and then moves to the Specifier of the Interrogative Projection. 114 3.4.3 The order of wh-adjuncts with respect to wh-objects I again divide the results of testing word order between a wh-adjunct and the whobject into two groups: the order of zakaj ‘why’, kako ‘how’, kje ‘when’, and kdaj ‘where’ with respect to the wh-object in matrix questions and the order of zakaj ‘why’, kako ‘how’, kje ‘when’, and kdaj ‘where’ with respect to the wh-object in embedded questions. I show the results for each condition separately. As expected based on the preference of the word order of wh-adjuncts with respect to the wh-subject, we again find a preference of placing wh-adjuncts before the object wh-phrase, but speakers again allow both word orders. The results are shown below: object > adjunct adjunct > object both neither wh-object - why 1 8 5 2 wh-object- how 3 8 3 2 wh-object - when 1 8 5 2 wh-object- where 4 5 5 2 Sum 9 29 18 8 Total: 56 16,1% 51,8% 32,1% Table 12: Word order of wh-adjuncts w.r.t. the wh-object in matrix questions Below, I give examples of the word orders I have tested. The preference to place the wh-adjunct before the wh-object can be detected with all the adjuncts, but it is the least strong with kje ‘where’. Still, in this condition, no strict word order can be detected, as speakers also take both word orders to be equally acceptable or accept the questions in which the wh-adjunct precedes the wh-object. (143) Tone cel dan govori o podobah na svojih fotografijah. ‘Tone has been speaking about his photographs the entire day.’ a. Zakaj je kaj slikal? why AUX what take-picture-of ‘Why did he take the picture of what?’ b. ?Kaj je zakaj slikal? 115 (144) Na predavanju so razložili mehanizem različnih vremenskih pojavov. ‘In the lecture they explained mechanisms of different weather phenomena.’ a. Kako so kaj how what explain AUX razložili? ‘How did they explain what?’ b. * Kaj so kako razložili? (145) Tone piše življenjepis znanega slovenskega pisatelja. ‘Tone is writing the biography of a well-known Slovenian writer.’ a. Kaj je kdaj počel? when AUX what do ‘When did he do what?’ b. Kdaj je kaj počel? In embedded questions, speakers again preferred the wh-adjuncts to precede the wh-object in general, as the table below shows. object > adjunct adjunct > object both neither wh-object - why 2 13 0 1 wh-object- how 2 11 3 0 wh-object - when 5 6 5 0 wh-object- where 2 4 9 0 Sum 11 34 17 1 Total: 62 17,7% 54,8% 27,4% Table 13: Word order of wh-adjuncts w.r.t. the wh-object in embedded questions Most speakers preferred zakaj ‘why’, (146), and kako ‘how’, (147), to be placed before the object wh-phrase in embedded questions. (146) Profesor je predaval o zgodovinskih požigih umetniških del. ‘The professor gavea lecture about burnings of art works.’ a. Jože je vprašal, zakaj so kaj Jože is aks what burn why AUX sežgali. ‘Jože asked why they burned what.’ 116 b. * Jože je vprašal, kaj so zakaj sežgali. (147) Razlagali so, da so celo leto iskali pohištvo za muzejsko zbirko. ‘They explained that they were looking for furniture for the museum collection for a year.’ a. Povej mi, kako so kaj tell what find me how AUX našli. ‘Tell me how they found what.’ b. * Povej mi, kaj so kako našli. Speakers had no strong preference to place kje ‘where’ or kdaj ‘when’ before the whobject, and the majority of speakers claimed both word orders are equally acceptable. (148) Menda so Banksyjevi grafiti po celem svetu. Pojdiva si jih ogledat. 64 ‘Supposedly there are Banksy’s graffiti all over the world. Let’s see them.’ a. Zapiši si, write self what kaj je kje narisal. AUX where paint ‘Write down where he painted what.’ b. Zapiši si, kje je kaj narisal. To summarize: In matrix and embedded questions there is a general preference to place wh-adjuncts before the wh-object. This preference is especially strong with zakaj ‘why’ and kako ‘how’ in embedded questions. Again, we do not find strict word order in placing wh-adjuncts with respect to wh-objects in Slovenian since speakers accept a variety of word orders. 64 One speaker opted for a different word order with a wh-phrase in situ and proposed the matrix question in (i): (i) Kje je narisal kaj? where AUX draw what ‘Where did he draw what?’ 117 3.4.4 Position of wh-phrases with respect to focused phrases and topics Before we continue to the derivation of word order of wh-phrases in multiple whquestions, we need to look at the word order of wh-phrases with respect to other elements typically found in the Left Periphery. In Slovenian, a wh-phrase can cooccur with a focused phrase, as examples below show. If a question only contains one wh-phrase and a focused phrase, the wh-phrase needs to precede the focused whphrase to get a true question reading. At least one wh-phrase also has to precede the focused phrase in a multiple wh-question. (149) a. peljal Kdo je MAJO who.NOM AUX Maja.ACC take na večerjo? on dinner ‘Who took MAJA to dinner?’ b. # MAJO je kdo peljal na večero? ‘Did anybody take MAJA to dinner?’ (150) a. Kdo je MAJO who.NOM AUX Maja.ACC where kam peljal? take ‘Who took MAJA where?’ b. Kdo je kam MAJO peljal? c. # MAJO je kdo kam peljal? Because wh-phrases can co-occur with the focused phrase, I propose that whphrases move to their own projection, which I will call a Wh-P (Wh-Projection), which can be recursive in Slovenian, since multiple wh-fronting is possible.65 In addition, another argument for a Wh-P, which is separate from FocusP, is that Dlinked phases behave exactly as non-D-linked phrases in Slovenian. Given that D65 It also has to be noted that in Slovenian a sentence can have more than one focus per clause, while Rizzi (1997) claims that in Italian only one focus per clause is available. For example, the Slovenian sentence in (i.a) has two instances of focus: (i) Miha je peljal Majo na večerjo. ‘Mija took Maja to dinner.’ a. Ne, TINO je peljal NA no, Tina is take on ‘No, he took Tina to lunch.’ KOSILO. lunch 118 linked phrases are given in the context, they do not convey new information, i.e. they are not focused and therefore do not appear in the FocusP. Since they do front, just as non-D-linked wh-phrases front, I assume they front to Wh-P. In multiple wh-questions a topic phrase can appear either after one, (151a), or all wh-phrases, (151b). Crucially, a topic cannot appear clause initially in multiple wh-questions, (151c). And while (151c) is in fact a possible question in Slovenian, it can only receive a yes/no-interpretation (see section 3.5). I take temu fantu ‘this boy’ to be the topic of the examples because it is definite (as it was previously mentioned in the context sentence as Janez).66 (151) Janez vsako leto dobi goro daril. ‘Janez gets a bunch of presents every year.’ a. Kdaj je temu fantu when AUX this boy.DAT Maja kupila za kaj what.ACC Maja buy for rojstni dan? birthday ‘When did Maja buy what for this boy?’ b. Kdaj je kaj temu fantu Maja kupila za rojstni dan? c. ?* Temu fantu je kdaj kaj Maja kupila za rojstni dan? With respect to placement of focus and topic phrases in the Left Periphery of Slovenian multiple wh-questions, Slovenian is similar to Russian (Rojina 2011). Assuming the split CP structure proposed in Rizzi (1997) and the Slovenian data, the Slovenian CP can be represented as shown in (152). The starred projections are in the CP of wh-questions only when needed and the projections in brackets are interchangeable when in the CP. That is: in a wh-question, one wh-phrase has to precede all other elements. This is why a Wh-Projection is located above the Topic, 66 A sentence such as (i) is acceptable in the context in (151). While one could take this as an indicator that the wh-phrase does not have to be sentence initial in a multiple wh-question, this is not the case. I take temu fantu ‘this boy’ this example to be an instance of a ‘hanging’ topic (see Krapova 2004). I do not take the hanging topic to be a part of the wh-question. One argument for this is the position of the clitic – Slovenian displays the second clitic phenomenon (the clitics appear after the first syntactic constituent), but in this case the clitic is located after the second constituent. Given this and a slight pause after the topic temu fantu ‘this boy’, I take the hanging topic to be a part of a separate clause and consequently not a part of the same Left Periphery as the two wh-phrases. (i) Temu fantu, kdo je kaj this boy who.NOM AUX what.ACC ‘For this boy, who bought what?’ kupil? buy 119 Focus and a Wh-Projection (which refers to all non initial wh-phrases), all of which are interchangeable among themselves. I will return to the clause initial WhProjection in section 3.5 and revise the structure of the Left Periphery in Slovenian. (152) Force … Wh-P …(Topic*) …(Focus*) ... (Wh-P*) …Fin IP In this structure we also need to include all the different wh-phrases and how they are ordered. We saw from the survey that two preferences emerged: when both the subject and the object wh-phrase are [+human], the subject precedes the object and when one of the wh-phrases is either zakaj ‘why’ or kako ‘how’ this wh-phrase precedes the remaining wh-phrases. The former can be explained with processing difficulties that arise with such combinations (see Fanselow 2010, Fanselow et al. 2013 for more on this type of Superiority). The latter can be accounted for by the different behavior of ‘real’ wh-adjuncts (i.e. why and how), which I am leaving aside. In the next section I show the derivation of the free word order of wh-phrases in multiple wh-questions in Slovenian. 3.4.5 Deriving the free word order of wh-phrases Even though some preferences were observed in Slovenian, I will assume the word order of wh-phrases in the Left Periphery in Slovenian multiple wh-questions is free. This appears to be true for both wh-questions with two wh-phrases, which were tested in the survey, and for questions with three wh-phrases, as (153) shows (see also examples in 2.3.2). (153) a. Kdo je komu kaj kupil za who.NOM AUX who.DAT what.ACC buy for ‘Who bought what for whom for his birthday?’ b. Kdo je kaj komu kupil za rojstni dan? c. Komu je kdo kaj kupil za rojstni dan? d. Komu je kaj kdo kupil za rojstni dan? e. Kaj je kdo komu kupil za rojstni dan? 120 rojstni dan? birthday f. Kaj je komu kdo kupil za rojstni dan? It does however seem that the word order can be regulated by other (nonsyntactic) factors in the sense of Kuno’s (1982) sorting key, originally stated in the Sorting Key Hypothesis (but also circumstances and pragmatic knowledge, see Kuno 1993). (154) Sorting Key Hypothesis (Kuno 1993: (102)) In a multiple wh-question the leftmost wh-word represents the key for sorting relevant pieces of information in the answer. This means that the questions in (153) would receive different answers. These answers would reflect the word order of wh-phrases in multiple wh-questions.67 The question is then how to derive the free word order of wh-phrases. One common property of Bulgarian and Russian multiple wh-fronting that was observed in Krapova and Cinque (2005) and Rojina (2011) is that wh-phrases in the Left Periphery preserve the word order they have prior to wh-movement.68 Rojina proposes that in Russian, the CP (the Vorfeld) preserves the word order from the Mittelfeld.69 This is based on Krapova and Cinque’s proposal that in Bulgarian the order from the IP is preserved. In addition, in Russian matrix questions, the word order of wh-phrases is, according to Rojina, ruled by the internal makeup of whelements, prominence of information and function of the elements, but crucially the word order from the Mittelfeld is preserved due to Relativized Minimality in the 67 For example, a felicitous answer to (153a) is in (ii.a) and a felicitous answer to (153c) is in (ii.b). In the two answers, the word order is different and corresponds to the word order of wh-phrases in the multiple wh-question. (ii) a. Miha je Maji avto kupil, Tone Jožetu kolo ... Miha.NOM AUX Maja.DAT car.ACC buy, Tone.NOM Jože.DAT bike.ACC ‘Miha bought Maja a car, Tone bought Jože a bike…’ b. Maji je Miha avto kupil, Jožetu Tone kolo ... 68 See also Laenzlinger and Soare (2005) for Romanian. 69 Rojina follows Laenzlinger and Soare (2005) who suggest that wh-elements move from the VP to the Mittelfeld, the IP, in order to check A-features (case and phi-features). They do so in positions labeled as [Spec, SubjP] for wh-subjects and [Spec, ObjP] for wh-objects. 121 sense of Krapova and Cinque (2005) – that is, no chain can intervene between the trace and the target but a part of a chain can intervene. It was shown in section 3.3.2 how strict word order can be achieved under the assumption that word order is preserved if we also assume the requirement on chains, so the question is how to account for word order variations. What Rojina proposes for multiple wh-questions in Russian in which two orders of wh-phrases are possible is that two word orders are possible in the Mittelfeld. Below is an example of such ambiguity. The two derivations of (155) in which gde ‘where’ can appear before or after kogda ‘when’, are shown in (156) and (157) below. (155) Russian (Rojina 2011: (193), (194)) a. Ona interesovalas‘, gde she kogda was wondering where when ja ego videl. I him saw ‘She was wondering when I saw him where.’ b. Ona interesovalas‘, (156) [ForceP kogda gde ja ego videl. [FocP [WhP gdei [WhP kogdak …[IP [LocP ti [TimeP tk]]]]]]] (157) [ForceP [FocP [WhP kogdai [WhP gdek …[IP [TimeP ti [LocP tk]]]]]]] Still, we need a way of accounting for the free word order of all wh-phrases in the Left Periphery in Slovenian. There seem to be two possible ways of doing this. First, we could potentially continue with the line of reasoning proposed in Krapova and Cinque (2005) or Rojina (2011) in that the order of wh-phrases in the Left Periphery is determined prior to movement to the CP – that is, the order of whphrases is determined in the IP/Mittelfeld. The second way is to assume that requirement on chains as proposed in Krapova and Cinque (2005) can be violated in Slovenian. Starting with the first possibility, we need first to establish what the Mittelfeld is. The Mittelfeld is taken to be the area between the CP (the Vorfeld) and the VP. The assumed Mittelfeld then looks like (158), as in (Laenzlinger and Soare 2004: (3)). Note that I follow Laenzlinger and Soare (2004) in the structure below, and use Sub(ject) and Obj(ect) to label grammatical projections in the tree. However, the 122 authors point out that SubjP corresponds to AgrsP and ObjP to AgroP, see Laenzlinger and Soare (2004) for more on the given structure of the Mittelfeld. (158) … MoodP 3 SubjP 3 S Infl(P) 3 V ObjP 3 O ModeP 3 SubjP 3 S Infl(P) 3 V ObjP 3 O AspP … Located in the Mittelfeld, following Laenzlinger and Soare (2005), are recursive Apositions and adverbs, which are, following Cinque (1999), located in different semantic projections. Crucially, Laenzlinger and Soare (2005) assume the Full VP Evacuation Principle according to which “all arguments must leave vP domain in order to have their A-features (i.e. Case and phi-features) and I-features (i.e. informational features such as topic, focus) checked/matched/assigned a value in the overt syntax” (Laenzlinger and Soare 2005: 107, (2)). Slovenian is an SVO language but exhibits free word order. Still, in (159) the first option is the most unmarked while the remaining orders get a marked reading (i.e. the first constituent is understood as focused). The marked readings are less strong in embedded sentences, as in (160) below which shows the different word orders of the subject and the dative and accusative objects. I assume that the material between the complementizer da ‘that’ and the negation is the Mittelfeld. (159) a. Ivan piše knjigo Ivan writes book o kavi. about coffee ‘Ivan is writing a book about coffee.’ b. Knjigo o kavi piše Ivan. 123 c. Knjigo o kavi Ivan piše. d. Ivan knjigo o kavi piše. (160) a. Jože pravi, da Miha Tonetu ne Jože says that Miha.NOM Tone.DAT not piše pesmi. write songs.GEN ‘Jože is saying that Miha is not writing songs for Tone.’ b. Jože pravi, da Tonetu Miha ne piše pesmi. c. Jože pravi, da pesmi Miha Tonetu ne piše. d. Jože pravi, da pesmi Tonetu Miha ne piše. If we then assume that Slovenian is in fact a language with free word order and we continue to assume movement to the Mittelfeld, we can observe how the order of wh-arguments is preserved. Because the word order is free in the IP, as (159) and (160) show, it is also free in the CP. The derivation would then be similar to the Russian example (155) schematized in (156) and (157). In this case the two adverbs are located in different projections. In Slovenian the object and subject can also be located in different positions in the Mittelfeld, but this is problematic, as I show below. Still, different word order in the Mittelfeld means that two word orders can arise in multiple wh-questions with an object and subject wh-phrase. (161) a. Zanima ga, kdo komu wonder he.ACC who.NOM who.DAT piše pesmi. writes songs ‘He wonders who writes songs for whom.’ b. Zanima ga, komu kdo piše pesmi. [WhP kdoi [WhP komuk …[IP [SubjP ti [ObjP (162) [ForceP (163) [ForceP [WhP komui [WhP kdok …[IP [ObjP tk]]]]]] ti [SubjP tk]]]]]] As we can see, in (162) and in (163) the chain as a whole never intervenes, so there is no violation of the requirement on chains. The word order from the Mittelfeld is also preserved in both cases but the word order differs from one case to another. 124 Based on this, however, the first problem emerges: If we assume the word order from the Mittelfeld is preserved due to a requirement on chains, as proposed by Krapova and Cinque (2005), this will account for the word order in the Vorfeld and the data in Slovenian if we assume the word order in the Mittelfeld is free. But if the word order in the Mittelfeld is free, this would indicate the requirement on chains can be violated at least when movement to the Mittelfeld occurs. Multiple wh-questions in which the subject wh-phrase precedes the object wh-phrase, as in (161a) above, are not problematic. The subject precedes the object in the Mittelfeld and the requirement of chains is not violated by movement to the CP or by the A-movement to the Mittelfeld (if we assume the subject is merged in the vP, which is above VP, which is where the object is merged). Movement from vP to the Mittelfeld (IP) is shown in (164a). The problem occurs in a wh-question in which the object precedes the subject. In order to get the ‘required’ word order in the Mittelfeld, we need the object to move across the subject when movement from the vP proceeds. In this case, the chain that is created by moving the subject intervenes between the object and its trace, which should lead to ungrammaticality. Yet, as (161b) above shows, a question in which the wh-object precedes the subject wh-phrase is completely acceptable. This problematic movement from the vP to the Mittelfeld is shown in (164b). (164) a. … [IP [SubjP kdoi b. … [IP [ObjP [ObjP kajk [… [vP ti …[VP tk …]]]]]] kajk [SubjP kdoi [… [vP ti …[VP tk …]]]]]] The problem is then how to account for the fact that a requirement on chains should hold during wh-movement (that is movement from the Mittelfeld to the CP) and not during A-movement from the VP to the Mittelfeld. The availability of the order preservation option can be tested using elements that have a fixed position within the Slovenian Mittelfeld. As we have seen above, the Mittelfeld consists of recursive A-positions and adverbs. These adverbs are located in different semantic projections which have a fixed hierarchy (Cinque 1999). We will use adverbs which are located in the Mittelfeld to test if the order of wh-phrases is preserved during wh-movement. Such wh-movement of adverbs is 125 possible, since, as Cinque (1999) notes, “some Adverb Phrases can undergo whmovement, while other, “higher” adverbs might resist this option. When forming such questions, we are typically asking for a degree relating to the adverb. But whether an adverb can be fronted is not related to the fact that an adverb can be modified by a degree” (Cinque 1999, fn. 45). For example, Cinque (1999) notes that courageously in English cannot be wh-fronted, while we can modify it with a degree. In Slovenian, on the other hand, a wh-question can be formed with pogumno ‘courageously’, but not with an adverb such as the speaker oriented iskreno ‘honestly’, despite the fact that we can modify the degree of both adverbs. (165) a. He very courageously fought the aliens. b. * How courageously has he fought the aliens? (166) a. Zelo pogumno se very courageously self je AUX bojeval z nezemljani. fought with aliens ‘He very courageously fought the aliens.’ b. Kako pogumno se je bojeval z nezemljani? how courageously self AUX fought with aliens ‘How courageously has he fought the aliens?’ (167) a. Povsem iskreno Miha piše knjigo. completely honestly Miha writes book ‘Completely honestly, Miha is writing a book.’ b. Kako iskreno Miha piše knjigo? how honestly Miha writes book # ‘How honest are you in saying that Miha is writing a book?’ ‘How honest is Miha when writing his book?’ If we want to test whether the Mittelfeld word order is preserved, we need to use questions with more than one wh-phrase. By now we have assumed a Mittelfeld in which an argument can take a position between any of the semantic-functional projections in which the adverbs are located. The position of these adverbs is fixed, but the arguments can appear in different positions with respect to these adverbs. We can then test the prediction that the order that wh-phrases have prior to movement to the CP is preserved with questions in which both wh-phrases relate to the adverbs in 126 the Mittelfeld. If the word order from the Mittelfeld is indeed preserved, then we can expect that wh-phrases that refer to the adverbs in the Mittelfeld will have a fixed word order (if the chains cannot intervene in Slovenian). For example, (168) shows that adverbs verjetno ‘probably’ and pogosto ‘often’ have a fixed order with respect to each other. Example (169) shows that this word order is less fixed with wh-phrases which refer to these adverbs. (168) a. Miha je verjetno Miha AUX probably pogosto obiskoval svojo mamo. often visited self mother ‘Miha probably often visited his mother.’ b. * Miha je pogosto verjetno obiskoval svojo mamo. (169) a. ? Kako verjetno how probably je kako pogosto Miha obiskoval svojo mamo? AUX how often Miha visited self mother ‘How probable is it that Miha how often visited his mother?’ b. ?? Kako pogosto je kako verjetno Miha obiskoval svojo mamo? Based on these examples, the first observation is that a multiple wh-question with two wh-phrases referring to the degree of an adverb is less grammatical than multiple wh-questions with two argument wh-phrases. It is also clear that the order of wh-phrases is less strict after wh-movement than it is before movement (based on my judgements, (169b) is not completely unacceptable, despite the reverse word order). This indicates that in Slovenian the order of wh-phrases in the Mittelfeld is not necessarily preserved in the CP. Because of this I will argue that it is not the case that word order in the Mittelfeld has to be preserved. I will rather argue for the second option, that is, Slovenian allows for violations of the requirement on wh-chains that Krapova and Cinque (2005) propose for Bulgarian, repeated below in (170). In fact, the requirement on chains also does not regulate focus or topic movement – I will return to this below. (170) (Krapova and Cinque 2005: (58)) Only a whole chain, not just a link of a chain, counts as an ‘intervener’. 127 This requirement is formulated based on Relativized Minimality (as in Rizzi 2001a) which is defined as in (171).70 Relativized Minimality explains a variety of phenomena, including wh-islands. Given that we cannot find wh-island violations in Slovenian, as we have already seen in section 3.1.2, we can see that Relativized Minimality holds in Slovenian. I illustrate this below in (172): (171) Relativized Minimality in the configuration: …X…Z…Y… X and Y cannot be connected by movement (or other local relations) if Z intervenes between them, and Z is of the same structural type as X.71 (172) a. * Komu sprašuješ, kdo kupuje t X Z Y : z_________________=_________________m who asks who buys darilo? gift Intended: *‘For whom are you asking is who buying a gift.’ b. * Kdo sprašuješ komu kupuje darilo? If Relativized Minimality did not hold in Slovenian, then we would expect island violations, but as (172) shows this is not the case. Because the wh-phrase kdo ‘who’ intervenes between the moved wh-phrase komu ‘who.DAT’ and its trace, the 70 Relativized Minimality is not the only constraint on locality and similar conditions were postulated within the Minimalist approach also. For example, in Chomsky (1995) the Minimal Link Condition (MLC), (i), is proposed. In (i), ‘closer’ is defined as (ii), in which τ refers to the target of raising. (i) (ii) Minimal Link Condition (Chomsky 1995: 311, (110)) K attracts α only if there is no β, β closer to K than α, such that K attracts β. β is closer to K than α unless β is in the same minimal domain as (a) τ or (b) α. As Rizzi (2004) notes, MLC differs from Relativized Minimality (RM) in that (i) it refers to a specific syntactic operation Attract, (ii) it operates in the course of derivation (while RM operates on representations), and (iii) it puts an emphasis on featural make up – in that the featural make up of the intervener must be identical to the one of the target. The latter point was considered in the revised proposal for RM, which is termed refers as Feature-Based Relativized Minimality in Rizzi (2011), see fn. 72. 71 Structural types are: argumental (person, number, gender, case), quantificational (Wh, Neg, measure, focus), modifier (evaluative, epistemic, Neg, frequentative, celerative, measure, manner) and topic (Rizzi 1990). 128 sentence is ungrammatical.72 Relativized Minimality therefore holds in Slovenian, but I suggest the extension of the condition on chains does not. Pursuing this option, we do not have to assume the Full VP Evacuation Principle (all arguments move to the Mittelfeld) which we needed in order to achieve the correct word order with the ‘order preservation option’, but note that the outcome does not change if we do assume it.73 The crucial point is that the wh-phrases can then simply move in any word order. For example, izpostaviti ‘expose’ is a verb that selects only the accusative dative order in Slovenian (Marvin and Stegovec 2012), (173). Following Marvin and Stegovec (2012), I assume that both wh-phrases are merged in the VP, where the dative object is introduced by a non-overt P (see (175) below). Crucially, in a multiple wh-question with the verb izpostaviti ‘expose’ in which we are questioning both the accusative and the dative object, both word orders of fronted wh-phrases are possible: (173) a. Zdravnik je izpostavil pacienta virusu. doctor AUX expose virus.DAT patient.ACC ‘The doctor exposed the patient to the virus.’ b. ?? Zdravnik (174) a. je izpostavil virusu Čemu je koga what.DAT AUX who.ACC pacienta. zdravnik izpostavil? doctor expose 72 Weak islands (i.e. islands out of which some elements can be extracted), on the other hand, can be violated in Slovenian. This is not surprising given a later version of Relativized Minimality – the Feature-Based Relativized Minimality, see Rizzi (2004, 2013). (i) Kaj si rekel kako je Maja zapela? what AUX say how is Maja sing ‘What did you say how she sang?’ Following Rizzi (2004), in Feature-Based Relativized Minimality only specifiers count as interveners and specifiers are licensed by the features on their heads. Under this approach features fall in the same groups as ‘structural types’ in fn. 71 and the notion ‘same structural type’ is replaced by “Spec licensed by features of same class” (Rizzi 2004: (62)). In the grammatical example (i), we could argue that the adverb kako ‘how’ does not count as an intervener as it does not licensed by the same features as the moved wh-phrase. I will not discuss such cases here any further, but see Rizzi (2004, 2011) for more examples. 73 The issue of phases emerges, but nothing changes if we consider vP and CP to be phases – whphrases can move through them in any possible word order. For now, I will ignore vP as a phase for a simpler representation. 129 ‘To what did the doctor expose whom?’ b. Koga je čemu zdravnik izpostavil? Assuming the underlying word order in (173) and that the direct and indirect objects do not move to the Mittelfeld prior to wh-movement, two derivations are possible.74 The derivation for (174a) is shown in (175c) and the derivation for the word order of the wh-phrases in (174b) is shown in (175b): (175) a. [WhP [whP ...[vP [v] [VP [V izpostavil] [[DP pacienta][PP <P> [DP virusu]]]]]]] b. [WhP kogai [whP čemuk ...[vP [v] [VP [V izpostavil] [[DP ti][PP <P> [DP tk]]]]]]] c. [WhP čemuk [whP kogai ...[vP [v] [VP [V izpostavil] [[DP ti][PP <P> [DP tk]]]]]]] Because (175c) is grammatical despite a configuration in which a chain intervenes between the target wh-phrase and the trace of the larger chain, we can assume that the requirement on chains does not hold in Slovenian. But this also means that we can expect that a requirement on chains in the sense of Krapova and Cinque (2005) does not necessarily hold in all languages. A conclusion that a language either does or does not obey the condition on chains leads to a further prediction that a language which obeys the condition in wh-fronting will exhibit strict word order in other types of movement, but a language that does not obey the condition on chains in multiple wh-fronting will also exhibit free word order in other phenomena. This prediction is borne out if we look at Topicalization in Bulgarian. Lambova (2003) shows that Bulgarian allows multiple instances of Topicalization. In these cases, all topics must move to the beginning of the clause. Moved topics exhibit superiority effects. This means the order of moved topics is strict, as we have predicted based on multiple wh-fronting in Bulgarian. 74 Again, nothing changes if we assume movement to Mittelfeld – in this case wh-phrases move to and from the Mittelfeld in any word order possible. 130 (176) Bulgarian (Lambova 2003: (35)) a. Mama decata mom.TOP kids.TOP šte vodi na will take to cirk. circus ‘Mom, the kids will take to the circus.’ b. * Decata mama šte void na cirk. c. * Mama šte vodi decata na cirk. In addition, having the condition on chains would also mean that Bulgarian would not allow for scrambling, given that scrambling leads to the word order variation. This is confirmed, since Bošković (2004) notes that Bulgarian does not have scrambling.75 Slovenian, on the other hand, allows for Scrambling, as example (177) shows. While it is hard to disambiguate between Scrambling and Focalization, I use the example below as an example of Scrambling because it can be used in an all new context in which, for example, somebody walks in a room and asks ‘What is going on?’. This means that nothing is given (and therefore not a topic) and nothing needs to be emphasized (as a focused phrase) for the sentence below to be an acceptable answer. Because of this I will assume that arguments are scrambled in the embedded clause. This means that again the requirement on chains does not hold in Slovenian.76 75 A different proposal was presented in Müller (2002) who claims that Bulgarian has scrambling, based on examples such as (i). However, as Bošković (2008) notes, Bulgarian does not have Japanese style scrambling, i.e. long distance scrambling from finite clauses (see Bošković 2008: fn. 5). Example (i) would then not be relevant for his observation. (i) Bulgarian (Müller 2002: (15))) vratata2 a. Ivan1 otvori Ivan opened door the b. Ivan1 vratata2 otvori t2 c. Vratata2 Ivan1 otvori t2 76 Boškovič (2008) claims that “it is well-known that multiple top/foc is disallowed, while multiple scrambling is allowed” (Bošković 2008: fn.16) in Slovenian. I am not aware of any such prohibition. But given the similarities between topicalization and scrambling, it is impossible to say how both the objects are moved to the clause initial position in (i). What is crucial for us is again, that they can be moved in any possible order (cf. Bulgarian multiple topicalization in (176)). (i) a. Paciente bo virusu zdravnik izpostavil. Patients.ACC will virus.DAT doctor expose ‘The doctor will expose the patients to the virus.’ 131 (177) a. Janez misli, da Janez.NOM thinks that je Ivan napisal pismo Maji AUX Ivan.NOM wrote letter.ACC Maja.DAT ‘Janez thinks that Ivan wrote a letter to Maja.’ b. Janez misli, da je Ivan pismo napisal Maji. c. Janez misli, da je Ivan Maji napisal pismo. d. Janez misli, da je Maji Ivan napisal pismo. Because of the free word order in multiple wh-fronting and Scrambling, I assume that wh-fronting in Slovenian is not regulated by the condition on chains (that is, chains are not interveners for wh-movement). This means that wh-movement to the CP can occur in any order possible. We have so far only accounted for the free word order of wh-phrases in the Left Periphery, but the mechanism behind movement still needs to be explained. Before this, an account has to also be given for the fact that a wh-phrase needs to precede other elements in the Left Periphery (i.e. topics and foci), which was shown in 3.4.4. 3.5 Obligatoriness of the clause initial wh-phrase In section 3.4.4 we have already seen that a wh-phrase also needs to appear in a clause initial position in questions in which all wh-phrases appear in the Left Periphery of the clause for the question to be interpreted as a true multiple whquestion.77 We need to clarify why this is the case. I will argue that the reason for this is Clause Typing which is defined as: b. Virusu bo paciente zdravnik izpostavil. Similarly, it holds that at least one wh-phrase has to appear in the clause initial position in order for the question to be interpreted as a true multiple wh-question and not, for example, an echo or a yes/noquestion. For instance, the question (i.e) would be interpreted as an echo question if the wh-phrases were pronounced with an emphasis (this is true regardless of the word order of wh-phrases) or a yes/no-question if the question would be pronounced with a rising intonation, which is generally typical of Slovenian yes/no-questions. Examples (i.a) to (i.d), on the other hand, receive a true question reading. 77 (i) a. Kaj what.ACC je is Maja kdaj komu Maja when who.DAT 132 povedala? tell (178) Clausal Typing Hypothesis (Cheng 1991: 29 (9)) Every clause needs to be typed. In the case of typing a wh-question, either a wh-particle in C0 is used or else fronting of a wh-word to the Spec of C0 is used, thereby typing a clause through C0 by spec head agreement. While clauses are typically subject to Clause Typing, I am here only concerned with multiple wh-questions in Slovenian. Note that Cheng (1991) states that in multiple wh-fronting languages, which Slovenian is, not all wh-phrases are fronted for Clause Typing, but that fronting is related to the interrogative reading of whwords (I will return to this in chapter 5). While I agree that it is not the case that all wh-phrases in Slovenian move to the beginning of the clause for Clause Typing, I will assume that the movement of the wh-phrase that ends up in the clause initial position is closely related to Clause Typing. If movement to any (or all) WhProjections were related to Clause Typing, then we would expect that a wh-phrase could simply be moved to a Wh-Projection below, for example, a Topic or a Focus Projection but the data we have examined shows that this is not the case. Rather Clause Typing occurs in the highest projection that hosts wh-phrases. I will refer to this projection as Interrogative Projection (Inter-P). Based on this we can modify the schematics of the Left Periphery in (152) from section 3.4.4 as (179) below:78 (179) ForceP … InterP … (TopicP*) … (FocusP*) … (Wh-P*) Such an Interrogative Projection was already proposed in Rizzi (2001a) for Italian and Spanish, adopted also by Aboh and Pfau (2011) for Gungbe and Dyakonova (2009) for Russian. But while Rizzi argues that only a small class of whphrases in Italian move to the Interrogative Projection (such as perche ‘why’), I b. c. d. e. ‘When did Maja tell what to whom?’ Kaj je Maja komu kdaj povedala? Komu je Maja kaj kdaj/kdaj kaj povedala? Kdaj je Maja kaj komu/komu kaj povedala? # Maja je kdaj kaj komu povedala? 78 The starred projections are in the CP only when needed and the projections in brackets are interchangeable when in the CP. 133 argue that there is no restriction on which wh-phrases can move to this projection in Slovenian. But crucially, in a multiple wh-question, one wh-phrase has to appear in the Interrogative Projection (i.e. this wh-phrase is clause initial). Based on the evidence in section 3.4.5, I claim that Inter-P precedes all other projections in the Left Periphery of the clause – with the exception of Force.79 I take the Force Projection and Interrogative Projection to be two different projections based on the assumption that Force conveys information relevant to the higher clause and it is subject to the ‘higher selector’ (Rizzi 1997, 2001a).80 Assuming this and based on the fact that Slovenian exhibits multiple wh-questions which can be embedded under a declarative complementizer da ‘that’, which is required by the matrix verb, I argue that Inter-P is separate from the ForceP. Examples such as (180) are interpreted as embedded questions. Crucially, embedded questions are not true wh-questions. For more arguments for separating Force from Interrogative Projection see Rizzi (2001a). (180) Maja je Maja AUX rekla, da kdo koga said that who.NOM who.ACC tepe. hits Literally: ‘Maja said that who hits whom?’ Before continuing, note that I am not claiming that the wh-phrase in the Interrogative Projection is responsible for Clause Typing. Rather I assume that Clause Typing is achieved by the complex interpretable Q+wh-feature (Soare 2007) which is located in the Interrogative Head. Because this complex feature also comes with an EPP subfeature, a wh-phrase needs to be moved to SpecInterP in order to check the features on the Interrogative Head, hence the close relation between Clause Typing and wh-movement to the clause initial position. Note also, that I am not the first to propose that overt wh-movement is not necessarily related to Clause Typing, see also Aboh and Pfau (2011). More about the feature make up and motivation for 79 A similar proposal was made for Russian by Dyakonova (2009) who also places the Interrogative Projection above the Topic Projection in Russian (but places a Frame Projection above InterP, see Dyakonova 2009). 80 Rizzi (1997) claims that Clause Typing is done by the Force head, but see Rizzi (2001a) for arguments why ForceP and InterP should be treated as two different heads. 134 movement in multiple wh-questions in Slovenian will be said in the following section. 3.6 Wh-movement In what follows, I assume, following Rizzi (1997, 2004), that a head of a projection in the Left Periphery agrees with its specifier in that the two share the same feature (and interpretation).81 I also assume that features that are located in the heads in the Left Periphery are interpretable (see for example Cinque and Rizzi 2008) and therefore visible at LF. And finally, I follow Soare (2007), and assume an interpretable Q-feature (similar to the Q-operator) and a wh-feature. In this approach there is a split between these two different features (see Soare (2007) for the motivation for this split), but in some languages, such as English, Romanian, or Slovenian, the two form a syncretic Q+wh-feature. In addition, in Soare’s (2007) approach the Q- or the wh-feature can also have an EPP subfeature – when this subfeature is present, the movement occurs and when there is no subfeature, no movement happens. The idea that all ‘real’ questions, both wh- and yes/no-, have a C projection with an abstract question operator Q is not new, see also Katz and Postal (1964), Baker (1970), Pesetsky (1987), Hagstrom (1998), den Dikken and Giannakidou (2002), Cable (2010), etc. Typically, in these approaches it is this Q-operator that turns a proposition into a question and it can be phonetically realized, as in Japanese or Chinese, or not. Similarly, in Soare (2007) the Q-feature can be realized as an overt Q-particle or not. Still, even when the Q-feature is not realized as a Q-particle, it is responsible for typing the sentence as interrogative (i.e. Clause Typing does not happen because of a moved wh-phrase, but because of the Q-feature). As for the whfeature, in Soare (2007) an interpretable unvalued wh-feature is also universally 81 Rizzi (2004) defines the Spec-Head relation that is crucial for feature checking as: (i) Rizzi (2004: (15)) Feature K is licensed (checked, valued…) on (H, XP) only if (a) XP is in a minimal configuration with H, and (b) c-command holds. 135 merged in the Focus head (together with the Q-feature). And finally, in this approach, (interrogative) wh-phrases enter the derivation with an uninterpretable valued wh-feature. Based on this, Soare proposes a typology of languages that divides languages on the basis of how languages form wh-questions. This typology also includes languages in which the wh-feature and the Q-feature form a complex feature that appears on a syncretic Focus head. Following Soare, this means that the two features are fused and both appear on the same head (I will refer to this complex feature as Q+wh-feature). Again, in these languages an EPP subfeature can be present (or not). According to Soare, the EPP-subfeature of the Q+wh-feature is available in languages with single wh-fronting (such as English or French cases with whmovement) or multiple wh-movement, such as Romanian or Bulgarian. Instances of wh-in-situ in French are, following Soare, cases of complex Q+wh-feature with no EPP-feature. I propose that Slovenian is a language similar to Romanian and Bulgarian in Soare’s typology, in that it forms a complex Q+wh-feature. This complex feature appears on the Interrogative Head (and not on the Focus0 as proposed by Soare (2007)) and comes with an EPP subfeature. Because of the EPP-subfeature, movement of one wh-phrase, which comes with an uninterpretable wh-feature, to the Specifier of the Interrogative Projection is obligatory. The complex Q+wh-feature on the Interrogative Projection is also responsible for Clause Typing. This means that the clause initial wh-phrase is simply a consequence of the clause typing Q+whfeature and is not itself responsible for the interpretation of a question as a whquestion but that Clause Typing and wh-movement are still closely related in Slovenian. In addition, Soare (2007) adopts the analysis of multiple wh-fronting in Romanian presented in Laenzinger and Soare (2005) according to which Romanian has wh-chunk movement. Wh-chunk is a structure consisting of only wh-phrases and this chunk moves to a single projection. Specifically, under this approach whmovement in Romanian proceeds to SpecFocus where it enters into a multiple agree relation with Foc0. While Soare (2007) makes no explicit claims about the nature of multiple wh-movement in Bulgarian, I assume she is adopting the view from Laenzinger and Soare (2005), according to which Bulgarian does not have wh-chunk 136 movement, but rather multiple movement to multiple Focus Projections, cf. Grewendorf (2001) who proposes wh-chunk movement for Bulgarian. I propose that Slovenian is also different Romanian in that it does not form wh-chunks, but rather move each wh-phrase individually. This means that the remaining wh-phrases move to the Wh-Projections. Wh-heads of Wh-Projections come with interpretable wh-feature (and not a complex Q+wh-feature) and an EPP-subfeature. Wh-phrases, which have an uninterpretable valued wh-feature, agree with the Wh-heads, and move to SpecWhP because of the EPP-feature. There are again no Superiority effects, which is not problematic if we assume that chains do not have to be crossed in Slovenian, see section 3.4.5.82 Below I show how the movement of wh-phrases proceeds (the structure omits everything but the wh-movement). The reverse word order of both wh-phrases is also available. (181) a. Koga je čemu zdravnik who.ACC AUX what.DAT doctor izpostavil? expose ‘Who did the doctor expose to what?’ b. Čemu je koga izpostavil zdravnik? 82 Citko’s (2010) phase based approach to locality with which she accounts for the free word order of wh-phrases in Polish short wh-movement also accounts for Slovenian. Under this approach locality is evaluated at phase edges. Movement to the phase edge goes on in a tucking in fashion. Once on the edge, wh-phrases can move in any order possible to the operator position, as operator movement frees them from locality violations. This does, however, require further stipulation, that is: the Left Periphery is located on top of a phase edge (a periphery is not equal to phase edge since wh-movement to the Left Periphery is an instance of operator movement which means that wh-phrases in Clause Internal or Left Periphery are not available for further movement due to operator freezing effect). 137 (182) a. ForceP 2 Force’ 2 Force0 InterP 2 Inter’ 2 Inter0 Wh-P [Q+wh] 2 [EPP] Wh’ 2 Wh0 FinP [wh-] 2 [EPP] Fin’ 2 Fin vP Agree 6 kogai… čemuj [wh] [wh] b. ForceP 2 C’ 2 InterP Force0 2 koga Inter’ 2 Inter0 Wh-P [iQ+wh] 2 [EPP] čemu Wh’ 2 Move Wh0 FinP [iwh-] 2 [EPP] Fin0 2 Fin vP 6 kogai… čemuj [wh] [wh] Wh-fronting to the clause initial position therefore proceeds as one wh-phrase (with an uninterpretable valued wh-feature) moves to check the interpretable unvalued complex Q+wh-feature on the Interrogative Head which in addition hosts an EPP-subfeature. The complex Q+wh-feature is also responsible for Clause Typing in Slovenian. Wh-phrases in addition move to Wh-Projections in the Low or High Periphery in which the Wh-Head has an interpretable wh-feature and the EPPsubfeature. 138 3.7 Summary This section has argued against the analysis of multiple wh-fronting in Slovenian based on the proposals by Rudin (1988) and Bošković (1997a et seq.) and argued for an analysis of multiple wh-fronting in Slovenian in which wh-movement proceeds to the extended Left Periphery of the sentence (the CP). I formulate this proposal based on the analysis of Bulgarian and Russian by Krapova and Cinque (2005) and Rojina (2011), respectively. It was shown that the word order of wh-phrases in the CP is free, but that there are some preferences: zakaj ‘why’ and kako ‘how’ precede other wh-phrases and [+human] subjects proceeds [+human] objects. I have argued that the word order of wh-phrases in multiple wh-questions is free because the requirement on chains which was proposed to regulate Bulgarian by Krapova and Cinque (2005) does not hold in wh-fronting in Slovenian. I proposed that wh-phrases in Slovenian move to two types of projections: the Interrogative and the Wh-Projection. The Interrogative Projection hosts the interpretable syncretic Q+wh-feature (also responsible for Clause Typing) and the Wh-Projection the interpretable wh-feature. Both come with an EPP subfeature which means that movement of wh-phrases to these projections is obligatory. The central issue in this chapter was multiple wh-fronting in Slovenian in which all wh-phrases move to the clause initial position. But as was shown in the previous chapter, this is not the only type of multiple wh-question that one finds in Slovenian, as wh-movement seems to be optional. Before turning to optional multiple wh-fronting in Slovenian, I will first show that optional multiple wh-fronting is not a result of restrictions on movement that have previously been used to account for whin-situ in multiple wh-fronting languages (e.g. Bošković 2002). I explore these restrictions in the next chapter. 139 4 Restrictions on multiple wh-fronting: when a wh-phrase must stay in situ While multiple wh-fronting is said to be obligatory in multiple wh-fronting languages, there are some exceptions to the rule. Multiple wh-movement is avoided and a wh-phrase can stay in situ when movement would violate a certain restriction. In this chapter I go over the phonological, semantic, and syntactic restrictions that account for instances of wh-phrases in situ in multiple wh-fronting languages (based on Bošković 2002). I show that these do not apply to Slovenian and therefore cannot be responsible for the seemingly optional multiple wh-fronting in Slovenian. I end the chapter with a restriction that does hold in Slovenian (and Serbo-Croatian) – the Principle of Distinctness (Richards 2010). 4.1 Phonological restrictions Bošković (1997b, 2001, 2002) observes that in Serbo-Croatian, wh-movement is typically obligatory, but that wh-phrases do not have to move when they are phonologically identical or similar (this observation is credited to Wayles Browne). This contrast can be observed in (1), (2) and (3) below: (1b) is unacceptable, since one wh-phrase stays in situ and this wh-phrase is not identical or similar to the moved wh-phrase (if both move, the question is acceptable, (1a)). Example (2), following Bošković (1997c), is not completely acceptable (he assesses it as reasonably acceptable), because the two wh-phrases are phonologically similar and koga ‘who’ is not fronted to the beginning of the sentence. Crucially, example (3a) is completely acceptable – because the two wh-phrases are homophonous, one must be left in situ. If both homophonous wh-words are fronted and form a sequence, the question is ungrammatical (or marginally acceptable if the second wh-phrase is heavily stressed, see Bošković 2001), as in (3b). It has to be noted that not all speakers of SerboCroatian share these judgements and see examples like (3b) as completely acceptable (it seems that this also holds for cases where the second wh-phrase is not heavily stressed). 140 (1) Serbo-Croatian (Bošković 1997c: (11a)) a. Ko šta kupuje? who what buys ‘Who buys what?’ b. (2) *Ko kupuje šta? Serbo-Croatian (Bošković 1997b: (23a)) ?Ko je who AUX ubio koga? killed who ‘Who killed who?’ (3) Serbo-Croatian (Bošković 2002: (37), Bošković 2001b: 102, fn. 5) a. Šta uslovljava šta? what conditions what ‘What conditions what?’ b. *Šta šta uslovljava? According to Bošković (2002) the same restriction against sequences of homophonous wh-phrases also holds in Bulgarian and Russian, as shown below, and also Romanian. (4) Bulgarian (Bošković 2002: (39a, b)) a. Kakvo obuslavlja kakvo? what conditions what b. * Kakvo kakvo obuslavlja? (5) Russian (Bošković 2002: (39c, d)) a. Čto obuslovilo čto? what conditioned what b. * Čto čto obuslovilo? Examples (3), (4), and (5) show that sequences of homophonous wh-phrases need to be avoided by leaving one wh-phrase in situ for the question to be grammatical. When a such sequence is not formed and the homophonous wh-phrases are divided by additional material, such as an adverb in example (6), both wh-phrases 141 must move (Bošković 2001b, 2002). Russian avoids sequences of identical whphrases in question (7) in a similar way: In Russian the copula is usually phonologically null, but it has to be pronounced between homophonous wh-phrases (Bošković 2001b: 103, fn. 6). (6) Serbo-Croatian (Bošković 2002: (38a, b)) a. Šta neprestano šta uslovljava? what constantly what conditions ‘What constantly conditions what?’ b. ?* Šta (7) neprestano uslovljava šta? Russian (Bošković 2001: 103, (ii)) a. Kto *(est’) kto? who who is ‘Who is who?’ b. Kto (*est’) professor? who is professor ‘Who is the professor?’ Bošković (2001b, 2002) accounts for the data with Franks’ (1998) proposal for the deletion of copies in the PF. According to this proposal, the deletion of a tail of a non-trivial chain in the PF is a preference. But when the pronunciation of the head would lead to a PF violation, the lower copy is pronounced and the higher one is deleted. Under this approach, both wh-phrases move because of their [+focus] feature (see section 3.2), as shown in (8), but because the pronunciation of the higher copy of the second šta would lead to a PF violation, the lower copy is pronounced, as shown in (9). As in Bošković (2001b), I am also ignoring the lower copy of the first šta ‘what’.83 83 A possible alternative analysis is presented in Bošković (2002): In accordance with Chomsky’s (1995) Move F analysis, the second wh-phase only moves features, but leaves phonological features behind (and the second wh-phrase is pronounced in situ). The first wh-phrase moves via phrasal movement. (i) [FP Šta FF(štai) [uslovljava štai]] 142 (8) [FP Šta Štai [uslovljava štai]] (9) [FP Šta Štai [uslovljava štai]] Richards (2010), on the other hand, claims that the ungrammaticality of (3) is a consequence of the Principle of Distinctness (see section 4.8). The two wh-phrases have a syncretic form, which is visible when Distinctness applies, possibly because of Impoverishment which deletes parts of the feature bundle (for example: in Greek, the case feature is deleted when the feature neuter is present, see Richards 2010).84 Still, as Richards notes, not all languages are sensitive to syncretism (i.e. German) possibly because in these languages syncretism is induced after lexical insertion. Phonological restrictions limit movement in examples in which wh-phrases are homophonous. To avoid sequences of homophonous wh-phrases, movement is suppressed and one of the identical wh-phrases has a lower copy pronounced. The question is then if such a restriction also holds for Slovenian. 4.2 Phonological restrictions in Slovenian Restrictions against sequences of homophonous wh-words do not hold in Slovenian. As (10a) shows, when both instances of kaj ‘what’ move, the question is grammatical, but the second wh-phrase can stay in situ in these examples, (10b). The same holds for examples in which the homophonous wh-words are divided by an adverb (i.e. examples where both wh-phrases must move in Serbo-Croatian) – in Slovenian the second wh-phrase can either move or stay in situ, (11). (10) a. Kaj kaj povzroča? what what causes ‘What causes what?’ 84 According to Richards (2010), difficulties, which lead to Distinctness effects, arise for linearization when two functional heads are indistinguishable. In Greek, such difficulties occur with nominative and accusatives for neuter – regardless of whether the two DPs have the same phonological form or not, see Richards (2010) for examples. To account for Greek data with nominative and accusative neuter, Richards proposes Impoverishment, which in this case deletes the case feature, which makes the two DPs distinguishable. 143 b. (11) a. Kaj povzroča kaj? Kaj vedno kaj what always what brings prinese? ‘What always brings what?’ b. Kaj vedno prinese kaj? Examples (10) and (11) show that optionality of Slovenian wh-fronting cannot be explained with phonological restrictions. In addition, there are many examples of questions with non-homophonous wh-phrases in which one of the phrases can stay in situ, such as (12). Wh-in-situ in these examples cannot be accounted for with a phonological restriction as described above. (12) a. Kdo občuduje who.NOM admires koga? who.ACC ‘Who admires whom?’ b. Kdo koga občuduje? 4.3 Semantic restrictions Following Bošković (2002), Serbo-Croatian (but also Russian and Bulgarian) is limited by semantic restrictions which only apply to D(iscourse)-linked phrases. As (13a) shows, a D-linked phrase stays in situ and the sentence is grammatical. Example (13b) shows that the D-linked wh-phrase can (optionally) be fronted (which again also holds in Russian, Polish, etc.). Even more, Bošković (2002) notes that most Bulgarian speakers allow optional fronting of the D-linked wh-phrase, which is shown in (14) – example (14a) shows the D-linked phrase in situ and example (14b) the optionally fronted D-linked phrase. (13) Serbo-Croatian (Bošković 2002: (26a), (27)) a. Ko je kupio koju knjigu? who.NOM AUX bought which book.ACC ‘Who bought which book?’ 144 b. ? Ko je koju knjigu kupio? (14) Bulgarian (Bošković 2002: (26b), (28)) a. Koj e kupil koja kniga? who.NOM AUX bought which book.ACC ‘Who bought which book?’ b. Koj koja kniga e kupil? Bošković (2002) accounts for Serbo-Croatian data and the contrast between Serbo-Croatian and Bulgarian (the contrast being that most Bulgarian speakers allow fronting of D-linked phrases and that only some Serbo-Croatian speakers allow optional fronting of D-linked phrases) with the motivation for wh-fronting in these languages. Bošković assumes that in Serbo-Croatian, as was shown in section 3.2, wh-movement is essentially focus movement and wh-phrases are focused. On the other hand, D-linked phrases are present in the discourse and an answer to a D-linked phrase is limited to familiar objects. Because they are already given, they are not focused and therefore cannot undergo focus movement. When they are fronted, as in (13b), this is the result of optional scrambling.85 In Bulgarian, on the other hand, wh85 This proposal would predict that D-linked phrases in single questions in Serbo-Croatian do not have to move. But Bošković (2002) states that it is not completely clear if D-linked phrases in single questions move, as example (i) is degraded, but not fully unacceptable: (i) Serbo-Croatian (Bošković 2002: (32)) ?? On je kupio koju knjigu? he is bought which book ‘He bought which book?’ Bošković suggests that such examples are degraded because they are not clause typed as interrogative sentences (in the sense of Cheng 1997) – Clause Typing in Serbo-Croatian is done via fronting of the wh-phrase (Bošković notes that the wh-phrase needs to be pronounced in the highest phonological realized projection in overt syntax) and this can be achieved through scrambling or wh-movement, the latter because of the parallels between French and Serbo-Croatian that Bošković claims to exist (see section 3.2). While I do not plan to devote much attention to single questions in Slovenian, it needs to be noted that single questions with D-linked phrases in situ in Slovenian receive an echo-interpretation or a yes/no-interpretation: (ii) a. b. Katero knjigo je which book AUX ‘Which book did he buy?’ # Kupil je katero knjigo? kupil? buy.3SG 145 movement is not focus movement, but rather movement to check the [+wh] feature. Since none of the wh-phrases move for focus reasons, it is not surprising that a Dlinked phrase in Bulgarian can move (however, it is surprising that this movement is only an option). Optionality of scrambling also explains why some speakers find (15) below grammatical (but marked): (15) Koju which knjigu je ko kupio? book AUX who bought ‘Who bought which book?’ To summarize: Semantic restrictions described in Bošković (2002) account for non-fronting of D-linked phrases in languages in which wh-phrases undergo focus fronting. The question is, whether these restrictions also apply in Slovenian. 4.4 Semantic restrictions in Slovenian D-linked phrases in Slovenian can stay in situ, (16a), or move, (16b). In addition, as (16c) shows, the order of fronted wh-phrases is free. (16) a. Kdo je poljubil katero punco? who.NOM AUX kissed which girl.ACC ‘Who kissed which girl?’ b. Kdo je katero punco poljubil? c. Katero punco je kdo poljubil? If we were to adopt the proposal for multiple wh-fronting for languages that do not exhibit Superiority (like Slovenian) given by Bošković (see section 3.2), this would lead us to the conclusion that Slovenian wh-fronting is an instance of focus fronting. Given the fact that D-linked phrases are not focused, we would expect that D-linked phrases do not move in Slovenian. This expectation is not fulfilled, as Dlinked phrases can move in Slovenian. This means that the semantic restrictions 146 described in Bošković (2002) cannot account for optionality of Slovenian multiple wh-fronting.86 However, it was shown in section 3.2.2 that there are several problems with this account. Because of this a different approach was assumed, under which whfronting in Slovenian is not an instance of focus fronting. I have argued that multiple wh-fronting in Slovenian is an instance of wh-movement. Assuming this, neither whin-situ nor wh-fronting with D-linked wh-phrases can be considered a result of a restriction on movement, but rather a consequence of some other factor (see chapter 5). In addition, non-D-linked phrases can also stay in situ in Slovenian, (17a). Since these wh-phrases are not D-linked, such examples cannot be accounted for with semantic restrictions on movement, which were described above. (17) a. Kdo je poljubil koga? who.NOM AUX kissed who.ACC ‘Who kissed whom?’ b. Kdo je koga poljubil? From these examples we can conclude that the semantic restrictions described in Bošković (2002) cannot account for optionality of Slovenian multiple wh-fronting. 86 Note that Slovenian also behaves differently with respect to non-wh-focus phrases. While these have to move in Serbo-Croatian, (i), they do not have to move to a clause initial position in Slovenian. (i) (ii) Serbo-Croatian (Bošković 2002: (17)) a. JOVANA savjetuje. Jovan.ACC advises ‘(S)he advises Jovan.’ b. ?* Savjetuje JOVANA. Slovenian a. JOŽETU svetuje. Jože.ACC advises ‘(S)he advises Jože.’ b. Svetuje JOŽETU. 147 4.5 Syntactic restrictions According to Bošković (2002), syntactic restrictions are restrictions which apply to non-Relativized Minimality islands (i.e. non-wh-islands) in languages such as Romanian. Following Chomorovski (1996), in Romanian, echo wh-phrases must always front – this also holds for contexts in which non-echo wh-phrases cannot move. The contrast is shown below – a non-D-linked wh-phrase cannot be extracted out of a whisland in Romanian – this is the reason why (18a) is ungrammatical. If the whquestion receives an echo intonation, the question is acceptable, (18b): (18) Romanian (Comorovski 1996: 60, (10)) a. *[Cei [nu what not știi [cinej [ej you-know who a spus ei ]] ] ] ? has said a spus ei ]] ] ] ? has said (standard question intonation) b. [Cei [nu știi [cinej [ej what not you-know who (echo-question intonation) Following Comorovski (1996), Bošković shows that syntactic restrictions apply to echo wh-phrases within non–Relativized Minimality islands (i.e. non-whislands) in Romanian. So while echo wh-phrases must move in Romanian (at least in some dialects), as (19) shows for matrix and embedded questions and as (18b) shows for wh-islands, they stay in situ in non-wh-islands. This is shown in (21) for a complex NP island. Overt wh-movement of non-echo wh-phrases from an island is not allowed in either overt or true questions, (20). (19) Romanian (Bošković 2002: (53), p. 376, fn. 44: (i)) a. * Ion Ion b. a adus CE? has brought what *Ion crede Ion believes că Petru a that Petru has cumpărat CE? bought what 148 (20) (Comorovski 1996: (12a)) * Cei vă intrigă what you intrigues zvonul că ar fi cumpărat ei? rumour-the that he-may be bought (21) (Comorovski 1996: (13a)) Vă intrigă zvonul you intrigues că ar rumor-the that he-may fi cumpărat CE? be bought what ‘You are intrigued by the rumor that he may have bought what?’ Bošković (2002) assumes that islandhood is syntactic in nature and based on this claims that the grammaticality of (21) can be accounted for under the Move F analysis (feature movement is not subject to non–Relativized Minimality islands but phrasal movement is): Full phrasal movement of the echo wh-phrase is not possible, (20), but the wh-phrase can undergo feature movement, as in (21) (for more see Bošković 2002). 4.6 Non-wh-islands as restriction on Slovenian In Slovenian, echo wh-phrases can stay in situ or move in matrix questions: (22) a. KOGA je Hilda podkupila? who AUX Hilda bribe ‘Hilda bribed WHO?’ b. Hilda je podkupila KOGA? Furthermore, Slovenian does not allow wh-fronting from syntactic islands for either wh-phrases or echo wh-phrases. This means that syntactic restrictions described in Bošković (2002) always apply in Slovenian: (23) a. ? Janez Janez je slišal AUX heard govorico, da je Petru rumor Peter.DAT bought what that AUX ‘Janez heard the rumor that Peter has bought what?’ b. Janez je slišal govorico, da je Petru kupil KAJ? 149 kupil kaj? (24) a. * Kaj je Janez What AUX Janez slišal govorico, da heard rumor that je Petru kupil? AUX Peter.DAT bought b. * KAJ je Janez slišal govorico, da je Petru kupil? Since echo wh-phrases do not obligatorily front in Slovenian, it is not an exception that they can stay in situ in complex NP islands. The kind of syntactic restrictions that hold for Romanian do not apply in Slovenian. In addition, as the examples in previous sections have shown, a non-echo wh-phrase can stay in situ in questions without islands, so this type of restriction cannot account for all the available Slovenian data. 4.7 The Principle of Distinctness as a restriction on multiple wh-movement Languages display differences in the morphological dissimilation effects that they use to avoid accidental repetition of elements, because of the general prohibition against elements of the same type appearing too close together. These constraints “operate on both form and content of morphemes, banning adjacent identity within a circumscribed domain” (Nevins 2012: 84). Such restrictions are known as repetition avoidance, morphological dissimilation, haplology, anti-homophony or the morphological OCP (Nevins 2012). One constraint of this kind is the Principle of Distinctness (Richards 2010). Richards (2010) bases his theory on the idea that some properties of syntax follow from the interaction between syntax and phonology, specifically from the ban which prohibits elements of the same type to be close together. In order to explain this ban, Richards assumes (following Chomsky 2000, 2001) that Spell-Out occurs cyclically during the derivation after the completion of strong phases, where strong phases include vP, CP, PP, and KP87 (i.e. Kase Phrase, the highest projection of the noun phrase). After each strong phase the material is sent to PF (Spell-Out domain) through Spell-Out. 87 When talking about English, Richards (2010) focuses on DPs. Here I will also be looking at DPs (and not KPs). 150 Richards (2010) also assumes that the head of a phase and its specifier (i.e. the edge) are a part of the following phase for the purposes of the calculation of distinctness, since they are linearized with the material of the higher phase (Chomsky 2000). Following Chomsky (2000), he also assumes that trees generated by syntax do not contain information about linear order and that linearization of nodes is performed by the operation of Spell-Out. It is also assumed that linearization is achieved by a version of the Linear Correspondence Axiom (LCA) as proposed by Kayne (1994), by which a total order of terminals within a Spell-Out domain (phase) is established. Based on these assumptions, Richards (2010) formulates the Principle of Distinctness: (25) Principle of Distinctness (Richards 2010: (5)) If a linearization statement <α,α> is generated, the derivation crashes. According to (25), a linearization statement is only interpretable when the nodes in the linearization statement are distinct from each other. Two nodes of the same type (i.e. <α, α>) in an asymmetric c-command relation cannot be linearized in the same Spell-Out domain. For example: the linearization statement <DP, DP>, which comes from the linearization of the structure in (26) in which XP does not stand for any of the strong phases (i.e. vP, CP, PP, or KP)88, cannot be linearized and causes a crash: (26) XP 3 DP X’ John 3 X DP Mary Richards (2010) also states that Distinctness only affects functional heads which enter the derivation as feature bundles and that Distinctness effects arise before vocabulary insertion. Functional heads are therefore represented only as 88 When X stands for one of the strong phases the linearization is not problematic: Since the head of a phase and its specifier are linearized with the material of the higher phase, the two DPs are not linearized in the same phase and the derivation is successful. 151 feature bundles and, according to Richards (2010), languages are different with respect to the richness of feature bundles and can be divided into two groups. The first group includes languages such as English, in which being of the same type means having nodes with the same label (e.g. <DP, DP>). Other languages are, according to Richards (2010), sensitive to the value of features which DPs have, for example case and gender. An example of a violation of the Principle of Distinctness in English is shown in (27), which shows sentences with exceptives: (27) (Richards 2010: (6a), (7a)) a. Every man danced with every woman, except John with Mary. b. * Every man danced with every woman, except John Mary. In (27) both remnants are DPs, both are in the same Spell-Out domain and form a linearization statement <DP, DP>. The ungrammaticality of (27b) follows from the Principle of Distinctness – two nodes in one Spell-Out domain are of the same type, which means that they cannot be distinguished from each other. They thus cannot be linearized and the linearization statement causes a crash. Example (27a) however is acceptable – the Principle of Distinctness is not violated since the two remnants are not of the same type – since PP is also a phase, linearization only sees a DP and a PP, which gives the linearization statement <DP, PP>. Richards (2010) notes that there are languages which allow multiple DPs in a single phase of the derivation. Among these are languages such as Japanese, German, Dutch, and Greek. For example, in Japanese case features make two DPs distinct, as shown below. (28) (Richards 2010: (89a, b)) a. [Sensei-o hihansita] gakusei-ga koko-ni oozei iru teacher.ACC criticized student.NOM here.DAT many be kedo, dare-ga dare-o ka oboeteinai Q remember.NEG but who.NOM who.ACC ‘There are lots of students here who criticized teachers, but I don’t remember who criticized who.’ 152 b. * [Sensei-ga teacher.NOM suki na] gakusei-ga koko-ni oozei iru like student.NOM here.DAT many be kedo, dare-ga but who.NOM who.NOM Q dare-ga ka oboeteinai remember.NOM ‘There are lots of students here who like teachers, but I don’t remember who likes who.’ The sentence in (28a) is an example of multiple sluicing in Japanese and is completely acceptable, despite the fact that there are two DP remnants. Richards claims that the acceptability can be accounted for with the Principle of Distinctness: since the two DPs have different case features (nominative and accusative), example (28a) is acceptable. When the two DPs have the same case feature, as in (28b), the sentence is unacceptable – following Richards, the two DPs are not sufficiently distinct to appear in the same phase of the derivation. Principle of Distinctness also works on multiple wh-fronting languages, as shown by Richards (2010). Richards (2010) discusses the influence of Distinctness on Serbo-Croatian, but claims that Serbian and Croatian display differences with respect to which features in the two languages make fronted DPs distinct (because of this I will treat the two separately in this section). In Serbian, case makes DPs distinct, but gender does not: (29) Serbian (Richards 2010: (106)) a. * Kojem je čovjeku which.DAT AUX man.DAT kojem dječaku mrsko which boy.DAT boring pomogati? help.INF ‘Which man doesn’t feel like helping which boy?’ b. * Kojem je čovjeku which.DAT AUX man.DAT kojoj ženi mrsko which woman.DAT boring pomogati? help.INF ‘Which man doesn’t feel like helping which woman?’ c. Kojem je čovjeku which.DAT AUX man.DAT kojeg dječaka mrsko pozdraviti? which boy.GEN boring greet.INF ‘Which man doesn’t feel like greeting which boy?’ 153 In (29a) both fronted DPs are dative and both have the same gender feature (masculine). The sentence is, according to Richards (2010), ungrammatical and it seems that this is not because both gender features are the same, but rather because of the same case feature. We can conclude this by comparing (29a) to (29b) in which the DPs have the same case feature (dative), but different gender feature (masculine and feminine), which has no effect on the grammaticality (i.e. (29b) is ungrammatical). In (29c), on the other hand, the two DPs are both masculine, but have different case feature (dative and genitive) – the sentence is, according to Richards (2010), completely acceptable, which suggests that being distinguishable (or distinct) by case, but not gender, is more important in Serbian. Croatian acts slightly differently from Serbian and the judgements for examples which are comparable to the ones in Serbian are different. (30) Croatian (Richards 2010: (107a), (109), (108)) a. ?? Kojem which.DAT je čovjeku kojem dječaku pomoči? AUX man.DAT which boy.DAT help.INF ‘Which man is to help which boy??’ b. ? Kojem which.DAT je čovjeku kojoj ženi pomoči AUX man.DAT which woman.DAT help.INF ‘Which man is to help which woman?’ c. Kojem je čovjeku which.DAT AUX man.DAT kojeg dječaka pozdraviti? which boy.GEN greet.INF ‘Which man is to greet which boy?’ In (30a), the two DPs have the same case and gender features (dative, masculine) and the sentence is almost unacceptable (the differences in the acceptability are very subtle). But as shown in (30b), the sentence improves if the two DPs differ at least in gender. Case seems to play a crucial role here too, since (30c) in which the two DPs have different case but the same gender is completely acceptable. Since (30b) is less acceptable than (30c), we can conclude that it is more important to have different case than to only differ in gender in Croatian. The crucial observation based on the data presented in Richards (2010) is that in multiple wh-fronting languages two fronted DPs must have distinct features for the 154 sentence to be completely acceptable. When the two DPs have similar features, the sentence is less grammatical (as in (30a) for Croatian). But, as Richards notes, there are mechanisms used to avoid Distinctness violations. These mechanisms include adding and deleting structure and suppressing movement. According to Richards (2010), suppressing movement is the mechanism used in multiple wh-fronting languages, as shown for Croatian: (31) Croatian (Richards 2010: (107b)) Kojem je čovjeku which.DAT AUX man.DAT pomoči kojem dječaku? help.INF which.DAT boy.DAT ‘Which man is to help which boy?’ If we compare (30a) and (31), we can see that (30a), in which both wh-phrases have the same case and gender and both move to a clause initial position and are therefore in the same phase of the derivation, is less acceptable than (31), in which movement is suppressed. Avoiding movement of the lower wh-DP divides the two DPs in different phases of the derivation, which means they do not form a linearization statement and Distinctness is not violated (Richards 2010). The Principle of Distinctness can then be taken as a restriction on multiple whmovement that is similar to semantic or phonological restrictions described in Bošković (2002). As with phonological and semantic violations, a violation of the Principle of Distinctness can also be avoided by not moving all wh-phrases to a clause initial position, i.e. leaving one wh-phrase in situ. But contrary to phonological and semantic restrictions, the Principle of Distinctness also holds in Slovenian, as I will show in the next section. 4.8 The Principle of Distinctness as a restriction on Slovenian In this section, I will show how Principle of Distinctness restricts multiple whmovement in Slovenian. In what follows I will be assuming that all wh-phrases move to the Left Periphery of the clause and are located in the same phase of the derivation. Assuming this and the Principle of Distinctness as defined in Richards 155 (2010), we can predict that the Principle of Distinctness restricts wh-movement in Slovenian. I tested this prediction with an experiment and the results show that it is borne out. Slovenian acceptability judgements were gathered from fifteen speakers in a small experiment. The list of examples consisted of multiple wh-fronting examples that had different case, gender and animacy combinations, but also multiple whsluicing examples and fillers. The experiment included 36 examples of multiple whfronting, 9 examples of multiple sluicing which I am leaving aside (but see Mišmaš 2013), and 34 grammatical and ungrammatical fillers. The fifteen speakers were asked to judge the sentences using a 1 to 5 scale, 5 being completely acceptable and 1 being completely unacceptable. Examples were recorded and the speakers assessed the sentences after hearing them. They also had the sentences written down on a sheet of paper on which they wrote down their judgements. As was reported for Croatian examples in Richards (2010), the differences in the acceptability are subtle but noticeable. In order to determine whether adjacency in linearization (being in the same phase) is one of the crucial factors for acceptability of Slovenian multiple whquestions, the tested examples had either both wh-phrases fronted or just one of the wh-DPs fronted (only examples with two wh-phrases were tested). In addition we tested how different grammatical features influence the acceptability of multiple whquestions. Starting with the influence of features: Given that case appears to be the most important feature in Serbian and Croatian (according to Richards 2010), we can first compare examples with the same or different case features on the fronted whphrases. Looking at the results, the first observation is that sentences with two whDPs which have the same case feature (all other features being the same) are in general less acceptable than sentences with DPs which have different case features. For example: questions with double datives got grades from 2.6 up to 3.3 (where the number is the average grade an example was given) while sentences with one dative DP and one genitive DP got responses from 3.2 up to 3.7.89 89 We are interested in the contrast in the acceptability of examples with different values of features, so the marks (i.e. whether an example is graded by a star or a question mark) are assigned relative to each other (and not based on a certain scale). 156 (32) ??Kateri punci [which girl]DAT.F kateri tetki ni zoprno pomagati? [which aunt]DAT.F not-be annoying help.INF ‘Which girl is not annoyed to help which aunt?’ (33) ?Kateri punčki [which girl]DAT.F katere starke ni zoprno narisati? [which old-lady]GEN.F not-be annoying draw.INF ‘Which girl is not annoyed to draw which old-lady?’ Example (33) got an average grade of 3.5 and is thus more acceptable than (32), which got an average grade of 3. From the point of view of the Principle of Distinctness this is not surprising assuming that case is relevant in distinguishing DPs in Slovenian. The linearization statement of (32) is <[DP, DAT, F] [DP, DAT, F]>, while for (33) it is <[DP, DAT, F] [DP, GEN, F]>. In (33) the case feature is responsible for making the two DPs distinguishable from each other.90 After observing the general difference between examples with DPs with different case features and DPs which have the same case feature, we can turn our attention to examples with the same case features. Looking at the gender feature, we can observe that examples with the same case and different gender features seem to be slightly more acceptable than the ones with the same case and same gender DPs: (34) ??Kateremu fantu [which boy]DAT.M je kateremu dedku zoprno pomagati? AUX [which grandfather]DAT.M annoying help.INF ‘Which boy is annoyed by helping which grandfather?’ (35) ?Kateremu starčku [which old-man]DAT.M je kateri teti zoprno pomagati? AUX [which aunt]DAT.F annoying help.INF ‘Which boy is annoyed by helping which aunt?’ 90 The data from the acceptability judgement test is also an argument against the structure of Slovenian matrix questions with multiple wh-fronting which was proposed by Golden (1997a), which was shown in section 3.1.1. Since we have assumed the Principle of Distinctness as proposed by Richards (2010), the graded acceptability of Slovenian multiple wh-questions can only be accounted for if both wh-phrases are in a single phase of the derivation, which is not the case for the structure proposed by Golden (1997a) – in this structure one wh-phrase would be in the CP and therefore on the edge of the phase which means it would be linearized with material of the higher phase. Movement of all wh-phrases to one phase, however, can account for the data – all the DPs are in a single phase of the derivation and the Principle of Distinctness can apply, which it does. 157 A sentence comparable to (35) got an average response of 3, while (34) got an average response of 2.6. If we compare the linearization statements of the two examples, we can see that in (34) the linearization statement is <[DP, DAT, M] [DP, DAT, M]>, while in (35) it is <[DP, DAT, M] [DP, DAT, F]>. This means that in (35) the gender feature plays a part in the computation of distinctness since the statements with different gender features are more acceptable than the ones with the same gender features. Number also has an influence on the acceptability of examples in which DPs have the same case and gender features – such as (34). The influence of number was not tested in the experiment, but the intuition of a few speakers is that sentences (36) and (37) are more acceptable than (34): (36) ?Kateremu fantu [which boy]DAT.M.SG je katerim dedkom zoprno pomagati? AUX [which grandfather]DAT.M.PL annoying help.INF ‘Which boy is annoyed by helping which grandfathers?’ (37) ?Katerim fantom [which boy]DAT.M.PL je kateremu dedku zoprno pomagati? AUX [which grandfather]DAT.M.SG annoying help.INF ‘Which boys is annoyed by helping which grandfather?’ In (36) the linearization statement is <[DP, DAT, M, SG] [DP, DAT, M, PL]> and in (37) <[DP, DAT, M, PL] [DP, DAT, M, SG]>. These examples are more acceptable than (34) which suggest that number is a feature that can make DPs distinct, but since the sentences are less acceptable than examples with different case feature this suggests that number has a lesser influence on acceptability. The influence of grammatical animacy was tested with double accusatives, as grammatical animacy can only be observed in Slovenian on masculine DPs in accusative singular (Toporišič 2004), but we cannot make any conclusions about the influence of grammatical animacy. Double accusatives in which both wh-DPs have an inanimate grammatical feature (regardless of whether the two DPs were close or in different phases) were judged the worst with an average of 1.8: 158 (38) *Kateri računalnik [which computer].ACC.M.-AN uči programer kateri teach programmer.NOM.M [which teorem? theorem].ACC.M.-AN ‘Which computer is being taught which theorem by the programmer?’ (39) *Kateri računalnik [which computer].ACC.M.-AN kateri teorem uči [which theorem].ACC.M.-AN teaches programer? programmer.NOM.M ‘Which computer is being taught which theorem by the programmer?’ As both (39) and (38)91 were given comparably low grades (example (39) got on average 1.8), (39) seems to be unacceptable for reasons outside of the Principle of Distinctness. I claim that these sentences are not ungrammatical because of the Principle of Distinctness because of example (38) in which the two DPs are not in one phase of the derivation and the sentence is still unacceptable. One possible explanation for the ungrammaticality of examples (38) and (39) is the verb used in these examples. That is, double accusatives in Slovenian occur with the verb učiti ‘to teach’ and it seems that sentences where both the ‘what is being taught’ and ‘who is being taught’ are inanimate objects are lexically not preferred. An argument that sentences such as (38) and (39) are ungrammatical for reasons outside of the Principle of Distinctness (assuming the principle holds) is also that changing the number feature has no effect on the acceptability of these sentences (grammatical number was not tested in the experiment): (40) *Kateri računalnik uči programer [which computer].ACC.M.-AN.SG teach programmer.NOM.M katere teoreme? [which theorem].ACC.M.-AN.PL ‘Which computer is being taught which theorem by the programmer?’ 91 Example (38) was not tested with speakers, but a comparable example was. Based on the tested example, which got an average judgement of 1.9, a mark was assigned to (38). 159 (41) *Kateri računalnik katere teoreme uči [which computer].ACC.M.-AN.SG [which theorem].ACC.M.-AN.PL teaches programer? programmer.NOM.M ‘Which computer is being taught which theorem by the programmer?’ Sentences with DPs with different number features are still unacceptable and if we assume that having at least one different feature on DPs is enough for a sentence to be acceptable this would mean that example (41), with two inanimate accusative DPs, is unacceptable for reasons not related to the Principle of Distinctness (in (41) the linearization statement is <[DP, ACC, M, -AN, SG] [DP, ACC, M, -AN, PL]>). Grammatical animacy feature still has some influence, and the sentence in (42) is more acceptable than examples (38) and (39), with an average acceptability of 3.3: (42) ?Katerega učenca kateri teorem uči profesor? [which student].ACC.M.+AN [which theorem].ACC.M.-AN teaches professor ‘Which girl is the professor teaching which subject?’ This sentence is more acceptable, but, again, it is not clear whether this is because of lexical reasons related to the verb učiti ‘to teach’ (one DP is naming the person who is being taught, the other the subject which is taught) or is it because of the distinctness – since the two DPs have different linearization statements: <[DP, ACC, M, +AN] [DP, ACC, M, -AN]>. Looking at the Slovenian data from a very general perspective, one thing is clear: The Principle of Distinctness has an influence on acceptability of Slovenian multiple wh-fronting. This is clear, as examples with fronted DPs with the same features are less acceptable than examples with DPs with distinct features. It can also be observed that different features have a different influence on acceptability, for example: if two DPs have different cases and the same gender, these examples are generally more acceptable than examples in which two DPs have the same case but different gender features. This suggests a hierarchy of features (see Mišmaš 2013), but crucially for us: The Principle of Distinctness affects Slovenian multiple wh160 fronting. Also, as Richards (2010) shows for Croatian, a violation of the Principle of Distinctness can be avoided by suppressing movement of one of the wh-phrases. We also tested with speakers whether sentences with two wh-DPs that have indistinguishable features are more acceptable if not all wh-phrases are fronted. According to Richards (2010), suppressing movement is one of the mechanisms used to avoid Distinctness violations, as shown above for Croatian. This is possible since avoiding movement of the lower wh-phrase divides the two DPs in different phases of the derivation, which means they do not form a linearization statement and Distinctness is not violated (Richards 2010). Example (43) shows that this holds for Slovenian. If we compare example (43), in which the two DPs have the same case and gender feature and only one wh-phrase is fronted to clause initial position, to example (32), repeated below in (44), in which the two DPs have the same case and gender feature but both wh-phrases are fronted to a clause initial position, we can see that speakers fully accept (43) but not (44). Since the only difference between the two examples is whether both or only one whphrase moves, I take this to be the reason for different acceptability. (43) Kateri punci ni which girl]DAT.F not-be zoprno pomagati kateri tetki? annoying help.INF [which aunt]DAT.F ‘Which girl is not annoyed to help which aunt? (44) ?? Kateri punci kateri tetki ni [which girl]DAT.F [which aunt]DAT.F not-be zoprno pomagati? annoying help.INF ‘Which girl is not annoyed to help which aunt?’ This suggests that avoiding movement can improve the grammaticality of Slovenian multiple wh-questions with wh-phrases that have indistinguishable features. It also means that the Principle of Distinctness is a restriction that affects Slovenian multiple wh-fronting – violations of the Principle of Distinctness are avoided by leaving a wh-phrase in situ (i.e. in a separate phase of the derivation). But while we can see that the Principle of Distinctness is a restriction on multiple wh-movement in Slovenian, it does not account for all cases of optionality 161 in Slovenian. For example, in the multiple wh-question (45), the two wh-phrases have distinct features but one wh-phrase still does not have to move.92 (45) a. Katera punca riše katerega fanta? [which girl].NOM.F draws [which boy].ACC.M ‘Which girl is drawing which boy?’ b. Katera punca katerega fanta riše? This question is completely acceptable when both wh-phrases move, (45b), but still a wh-phrase can be left in situ, (45a). This suppression of movement cannot be accounted for with the Principle of Distinctness since movement would not lead to ungrammaticality as the two DPs are distinct. The question is, then, how we can account for examples such as (45a). 4.9 Summary This chapter has shown that phonological, semantic, and syntactic restrictions that apply to other multiple wh-fronting languages do not account for instances of wh-insitu in Slovenian. In addition, we have seen that the Principle of Distinctness does restrict multiple wh-fronting in Slovenian, but it does not account for optionality in Slovenian multiple wh-questions in general, given that not all wh-phrases front even when they are distinct. Since we cannot account for optional multiple wh-fronting with restrictions, a new account is needed. 92 Moreover, the two wh-phrases can even have different labels and a wh-phrase can still stay in situ. (i) a. b. Kdaj je Maja kupila kaj? when AUX Maja buy what.ACC ‘When did Maja buy what?’ Kdaj je kaj Maja kupila? 162 5 Multiple wh-questions with a wh-phrase in situ in Slovenian As we have seen above, there are cases of multiple wh-questions in Slovenian in which at least one wh-phrase moves to the clause initial position and one wh-phrase (or more) stays in situ. This pattern is grammatical in matrix and embedded questions, (1) and (2). This pattern can also be found with D-linked wh-phrases, as I show in (3) for a question with one D-linked phrase, but see also section 2.3. In example (4), two wh-phrases are left in situ. Izpostaviti ‘to expose’ is a verb that occurs with the accusative object preceding the dative object. If we assume that the wh-phrases appear in situ in these cases, it is then not surprising that a wh-question in which the accusative wh-phrase precedes the dative, (4a), is much more acceptable that dative preceding the accusative, (4b). Crucially, in cases such as (1) to (4) all wh-phrases receive an interrogative reading. (1) a. Kaj je Miha kupil komu? what.ACC AUX Miha buy who.DAT ‘What did Miha buy for whom?’ (2) b. Komu je Miha kupil kaj? a. Janez sprašuje, komu je Ana kupila kaj. Janez asks who.DAT AUX Ana buy what.ACC ‘Janez is asking what Ana bought for whom.’ (3) b. Janez sprašuje, kaj je Ana kupila komu. a. Kaj je Miha kupil kateri punci? what.ACC AUX Miha buy which girl.DAT ‘What did Miha buy for which girl?’ (4) b. Kateri punci je Miha kupil kaj? a. Kdo je izpostavil koga čemu? Who.NOM AUX expose what.DAT who.ACC ‘Who did what to whom?’ b. ?? Kdo je izpostavil čemu koga? 163 While I will focus mostly on multiple wh-questions with an object wh-phrase in situ in this section, questions with a clause final wh-subject are also possible. I will consider these in section 5.2.1. Questions in which at least one wh-phrase moves to the clause initial position and a wh-phrase stays in situ are interpreted as true multiple wh-questions with multiple wh-fronting (that is, questions such as (1)–(4) receive the same interpretation as questions with multiple wh-fronting, they receive either a pair list or a single pair answer).93 I will say more about the interpretation of these questions in section 7.1.1; at this point I only want to emphasize that these questions are different from ones in which no wh-phrases move to the clause initial position (i.e. all whphrases stay in situ). Questions in which all wh-phrases stay in situ are ungrammatical under a true wh-question reading and are only acceptable under two, different readings: if the wh-phrases are pronounced with an emphasis, the question is interpreted as an echo question (either as a surprise or a request-for-repetition question) and if the sentences are pronounced with a yes/no-question (rising) intonation or if they have a yes/no-question marker in the clause initial position, they will be interpreted as yes/no-questions and the in situ wh-phrases will be interpreted as indefinite pronouns. I will leave the echo and yes/no-questions questions aside and 93 The situation is then to some extent similar to Polish, since, as Dornisch (1998) notes, only one whphrase needs to obligatorily move to the clause initial position (in her analysis SpecCP) in Polish hence the ungrammaticality of (i). But Dornisch also notes that a wh-pronoun can stay in situ but only if it carries heavy, focal stress as in (ii), which is not the case for Slovenian. (i) Polish (Dornisch 1998: 123, (1’); 124, (10)) * Pozbawil-by kogo jakxch przywilejów? deprive-cond whom what privileges Intended: ‘Whom would he deprive of what privileges?’ (ii) Co by Anna poleciła KOMU? what cond. Anna reccomend to-whom ‘What would Anna recommend to whom?’ Dornisch suggests that the difference between the wh-phrases that must move and wh-phrases that remain in situ is in the deficiency pronouns exhibit. Following Cardinaletti and Starke (1994) she assumes that deficient pronouns cannot occupy Theta positions and must move to a derived position (as in (i)). On the other hand, the stressed KOMU ‘to whom’ in (ii) is a strong pronoun and can, because of this, stay in situ. In addition, according to Dornisch, the weak pronouns carry a strong whfeature that motivates movement, while the strong pronouns carry a weak wh-feature. The movement to the clause initial position is motivated by the strong wh-feature on C. The weak feature on the strong pronoun, however, does not have to be checked before LF and can stay in situ because of this. 164 focus on true wh-questions, but the behavior of wh-phrases in yes/no-questions will prove to be crucial for our analysis of multiple wh-questions with a wh-phrase in situ. (5) Zdravnik doctor je AUX izpostavil KOGA ČEMU? expose who.ACC what.DAT ‘The doctor exposed WHO to WHAT?’ #‘Who did the doctor expose to what?’ (6) (A) je zdravnik QY/N AUX doctor izpostavil koga čemu? expose what.DAT who.ACC ‘Did the doctor expose somebody to something?’ #‘Who did the doctor expose to what?’ Because wh-phrases can stay in situ only when at least one wh-phrase moves to the clause initial position, I will not be arguing that Slovenian is a wh-in-situ language, as for example Strahov (2001) did for Russian.94 The central question of this section is then: How can we account for a multiple wh-question in which a whphrase is left in situ? Building on previous studies, there seem to be at least two ways of proceeding: The explanation is in the landing site of the wh-phrase or in the whphrases themselves. 94 Specifically, Strahov (2001) claims Russian is a wh-in-situ language in which wh-phrases are scrambled. The claim is based on Cheng’s (1991) generalization according to which languages that have a special marking in yes/no-questions are in situ languages and also have a wh-typing (overt or covert) particle (cf. Bruening 2007). While Strahov (2001) shows that that Russian particle li is a yes/no-clausal typing particle and that, consequently, Russian is a wh-in-situ language the same does not hold for Slovenian. In Slovenian the particle ali/a appears in yes/no-matrix questions, such as (6) above, but a/ali can also appear in embedded yes/no-questions, (i), in disjunction, (ii), and in infinitivals, (iii). Since Slovenian a/ali can be used in these three contexts, it can be treated in the same way as English whether or Polish czy which are, following Cheng (1991), not particles, but rather wh-phrases. (i) (ii) (iii) Miha je vprašal, a poznaš Miha AUX ask QY/N know ‘Miha asked if you know Maja.’ Miha pleše ali poje. Miha dance Q sing ‘Miha dances or sings.’ Miha ne ve, ali plesati ali Miha not knows Q dance.INF or ‘Miha doesn’t know whether to dance or not.’ Majo. Maja.ACC ne. not 165 Starting with the first option, the difference in the landing site has been shown to be responsible for optionality of wh-movement in different languages, e.g. Cheng and Rooryck (2000) for French, Pires and Taylor (2007) for Brazilian Portuguese, Denham (1997, 2000) for Babine-Witsuwit’en, etc. For example, Denham (1997, 2000) shows that Babine-Witsuwit’en, an SOV Athabaskan language spoken in northern British Columbia, allows only one wh-phrase per sentence. The examples below show matrix questions with optional fronting – ndu ‘what’ can appear in situ as in (7a) or it can move to the beginning of the clause, as in (7b) (for more examples see Denham 2000).95 The in-situ option and questions with movement are possible with arguments, as in (7), and with adjuncts, (8), and the same facts that hold in matrix clauses also hold in complex questions (for both the embedded and matrix clause of a complex question, cf. Denham (2000)). Optional fronting only applies to wh-phrases, as the ungrammatical example (9b) shows. (7) Babine-Witsuwit’en (Denham 2000: (5)) a. Lillian ndu yunkët? Lillian what 3s.bought.3s ‘What did Lillian buy?’ b. (8) Ndu Lillian yunkët? Babine-Witsuwit’en (Denham 2000: (14)) a. Sharon book nts’ën’a yik’iyetalhdic? Sharon book how 3s.will read.3s ‘How will Sharon read the book?’ b. (9) Nts’ën’a Sharon book yik’iyetalhdic? Babine-Witsuwit’en (Denham 2000: (7)) a. Lillian dus yunkët. Lillian cat 3s.bought.3s7 ‘Lillian bought a cat.’ 95 Given that there is no difference in the interpretation of questions in which a wh-phrase moves or not, Babine-Witsuwit’en is then an example of a language which exhibits ‘true’ optionality (i.e. semantically vacuous optionality, see Section 2) which is not predicted by the Minimalist program. Denham avoids the problem by proposing an independent projection, which is responsible for interpretation. What is optional in this proposal is the selection of C from the lexicon. This is reminiscent of optional selection of the EPP feature, and accounts for the either/or nature of whmovement in Babine-Witsuwit’en. 166 b. * Dus Lillian yunkët. Denham (2000) accounts for the data by proposing optional selection of C from the lexicon. When C is selected from the lexicon wh-fronting proceeds, and when C is not selected the wh-phrases stay in situ. Following Denham, this interrogative C does not influence the interpretation, it only motivates movement, which is achieved with a –interpretable feature on C and a +interpretable feature on the wh-phrases. Because the +interpretable feature is present in the structure, the interpretation is unaffected by movement. Additionally, Clause Typing and scope are ‘pushed’ to a typing projection TyP which in wh-questions has a +interpretable Q and a Scope feature. Because TyP has a +interpretable feature, the feature does not trigger movement and is interpreted at LF (for more see Denham (2000)). If we continue with such a line of reasoning and try to account for optional multiple wh-fronting in Slovenian with the nature of the C head (while maintaining that wh-phrases only come in one variety) and in addition pursuing a Minimalist account with no extended Left Periphery, this would lead us to few different C heads. To account for Slovenian data, the C head would come in several different varieties – attract all or attract one. But since there are cases in which more than one wh-phrase can move (while one wh-phrase stays in situ, i.e. there is no clear either/or situation in Slovenian multiple wh-fronting) this would not cover all the data. Potentially, a third C head could be added – attract some.96 Given the data, this type of reasoning seems to lead to at least three different Cs in the lexicon, and these Cs are, as far as I can see, unmotivated by anything else in Slovenian. Also, if stipulated two (or more) different Cs are related to wh-movement then given the Minimalist understanding of optionality (see section 2.1), we would also expect two different interpretations for examples in which all wh-phrases move or only one wh-phrase moves. As it will be shown in section 7.1.1, this is not the case. We can therefore conclude that assuming more than one C that motivates multiple wh-movement in Slovenian does not seem to be a sufficient way of accounting for optional wh-movement in which a wh-phrase stays in situ. Therefore, I will not be arguing for such an analysis of Slovenian. 96 Or an additional generalization should be put in place, see Pesetsky (2000) for Bulgarian, cf. Section 2.2. 167 These issues also remain unresolved if we adopt a Cartographic approach for the structure of Slovenian wh-questions, as it was done in the previous sections. Under the assumed approach, the wh-question is typed as such in the Interrogative Projection. Clause Typing always occurs before Spell Out and is related to a whphrase in Slovenian since one wh-phrase needs to appear clause initially. Given the Slovenian data explored here, there is no optionality with respect to movement to the Interrogative Projection. That is, in all multiple wh-questions, at least one wh-phrase appears in the clause initial position. When this does not happen and all wh-phrases stay in situ, the sentence does not receive a true wh-question reading. However, as we have seen, wh-phrases also move to Wh-Projections in Slovenian. If we still pursue the analysis by Denham (2000), we can then simply claim that these WhProjections are not selected from the lexicon and therefore the wh-phrases do not move (or differently, these Wh-Projections come with an interpretable wh-feature but no EPP-subfeature and again, no movement occurs).97 But if we take movement to be feature driven and assume that wh-phrases enter the derivation with an uninterpretable valued wh-feature which is checked because of the interpretable unvalued feature on the Wh-head (see section 3.6), how would the feature on the whphrase be checked (i.e., why do we need Agree and Move for the wh-phrase which moves to the Interrogative Projection, but only Agree is enough with WhProjections)? The problem also remains if we assume a criterion approach to movement in the sense of Rizzi (1996), where the wh-criterion is defined as (10) below. According to the criterion, if a wh-operator enters the derivation it needs to move in the appropriate level of representation (Rizzi 1996). The wh-operator is defined as in (11): (10) The wh-criterion (Rizzi 1996: (6)) a. A wh-operator must be in Spec-head configuration with X0 [+wh] b. An X0 [+wh] must be in a Spec-Head configuration with a wh-operator (11) Wh-operator = a wh-phrase in an A-bar position 97 If we take the EPP-subfeature to be responsible for optionality, the question remains why it is the case that a Wh-Projection can have optional EPP, but the Interrogative Projection has an obligatory EPP. 168 This means that only wh-operators move, but wh-phrases that are not operators can stay in situ. For example, in English matrix questions such as ‘Who hit whom?’ only one argument wh-phrase moves to the clause initial position while the other whargument stays in situ. This is not problematic as the wh-in-situ argument is in an argument position (i.e. in the position in which it receives its theta role) and is therefore not an operator, cf. (11). These cases are subject to static agreement, but an additional option exists for languages with optional wh-movement. For example, cases of optional wh-movement as in French are subject to dynamic agreement in which agreement can apply either in the syntax or in the LF (Rizzi 1996). Following Rizzi, we could say that Slovenian optional wh-movement to Wh-Projections is subject to dynamic agreement which would mean that wh-phrases either move in syntax or at LF. I will not be pursuing this line of reasoning here as it is not clear why dynamic agreement would then only apply when wh-phrases move to WhProjections, while movement to the Interrogative Projection is subject to static agreement (and is therefore obligatory). Because of the many issues that come with the proposal that optional whfronting is a result of the landing sites of wh-phrases that are responsible for multiple wh-fronting, I will rather be turning to wh-phrases themselves, as there is independent evidence for different varieties of wh-phrases in Slovenian. 5.1 Wh-in-situ as a consequence of the properties of wh-phrases Two initial observations can be made, based on which we can pursue the second possibility outlined above, i.e. wh-in-situ in multiple wh-questions in Slovenian is a consequence of there being different wh-phrases. First, in questions with multiple wh-phrases in Slovenian a wh-phrase can stay in situ only in questions in which at least one wh-phrase moves to the clause initial position. Second, we have independent evidence that wh-phrases come in two varieties, since wh-phrases in yes/no-questions are not interpreted as interrogative wh-phrases, as we have already seen. In these the wh-phrase is interpreted as an indefinite pronoun and typically 169 appears in situ. Because of this, it seems reasonable to investigate the relation between interrogative wh-phrases and indefinite pronouns further. 5.1.1 Wh-pronouns in Slovenian and other Slavic languages The relation between wh-pronouns and indefinite pronouns has been discussed in the literature before (cf. Cheng 1991, Citko 1998). I start with the system of wh-, indefinite, and negative pronouns (the latter two are built from wh-stems) in the languages that have multiple wh-fronting questions and have previously been described as having some variation with respect to movement in multiple whquestions (Polish, Serbo-Croatian, and Czech are taken from Citko (1998), but Russian (based on Yanovich 2005) and Slovenian have been added to the list). Crucially, wh-pronouns (leftmost column in the table below) can have two meanings in Slovenian (but also Czech, Serbo-Croatian, Russian, and Polish): they can receive an interrogative or an indefinite interpretation. I will refer to these wh-pronouns as bare wh-pronouns due to the lack of prefixes or suffixes that one finds in indefinite pronouns (or negative pronouns, etc.). The pronouns are shown in Table 14. The contrast between negative pronouns and wh-pronouns in Slovenian may seem larger than the respective contrast in Polish, Czech, and Serbo-Croatian. But, following Ilc (2008), negative pronouns in Slovenian are derived from wh-pronouns to which a prefix ni- and a suffix -r are added. In the nominative form this process triggeres the use of a substitution form. In addition, Ilc (2008) writes that older Slovenian grammars also noted nikdo ‘nobody’ as a negative pronoun that is clearly related to the wh-phrase kdo ‘who’. This means that despite superficial differences, Slovenian negative pronouns are derived from wh-pronouns. 170 Wh-pronouns Indefinite pronouns Negative pronouns Polish kto ‘who’ kto-ś ‘somebody’ ni-kt ‘nobody’ co ‘what’ co-ś ‘something’ ni-c ‘nothing’ gdzie ‘where’ gdzie-ś ‘somewhere’ ni-gdzie ‘nowhere’ kiedy ‘when’ kiedy-ś ‘some time’ ni-gdy ‘never’ Czech kdo ‘who’ kdo-si ‘somebody’ ni-kdo ‘nobody’ co ‘what’ co-si ‘something’ ni-c ‘nothing’ kde ‘where’ kde-si ‘somewhere’ ni-kde ‘nowhere’ kdy ‘when’ kdy-si ‘some time’ ni-kdy ‘never’ Serbo-Croatian (t)ko ‘who’ ne-(t)ko ‘somebody’ ni-(t)ko ‘nobody’ što/šta ‘what’ ne-što ‘something’ ni-šta ‘nothing’ gd(j)e ‘where’ ne-gd(j)e ‘somewhere’ ni-gd(j)e ‘nowhere’ kada ‘when’ ne-kada ‘some time’ ni-kada ‘never’ Russian98 kto ‘who’ kto-to’ ‘someone’ ni-kto ‘nobody’ čto ‘what’ čto-to’ ‘something’ ni-čto ‘nothing’ gde ‘where’ gde-to’ ‘somewhere’ ni-gde ‘nowhere’ kogda ‘when’ kogda-to’ ‘sometime’ ni-kogda ‘never’ Slovenian kdo ‘who’ ne-kdo ‘someone’ nihče ‘nobody’ (ni-kdo) kaj ‘what’ ne-kaj ‘something’ nič ‘noting’ kje ‘where’ ne-kje ‘somewhere’ ni-kje-r ‘nowhere’ kdaj ‘when’ ne-kdaj ‘some time’ ni-kda-r ‘never’ (ni-koli) Table 14: List of wh-, indefinite, and negative pronouns 98 According to Yanovich (2005) there are several series of indefinite pronouns, he lists four: the series with the morpheme -to, series with -nibud’, with -libo and with ni. I give only the -to series in the table, as these can be used in declarative sentences, but see Yanovich (2005) for more. 171 Starting with indefinite pronouns, Table 14 shows that indefinite pronouns are a combination of a wh-phrase with a suffix (e.g. Polish) or a prefix (e.g. Slovenian). Such indefinite pronouns can be used in, for example, declarative sentences in Slovenian: (12) Miha je nekoga Miha AUX somebody.ACC predstavil svoji sestri. introduce self sister.DAT ‘Miha introduced somebody to his sister.’ But in addition to indefinite pronouns, bare wh-pronouns can also be used as indefinite pronouns in some environments, such as yes/no-questions or conditional sentences. As Citko (1998) notes, this is possible in Serbo-Croatian, (13), and Polish, (13), and Dyakonova (2009) shows this for Russian bare wh-pronouns, (15).99 Finally, (16) shows the use of a bare wh-phrase as an indefinite in a yes/no-question for Slovenian.100 It also needs to be noted that while I here concentrate on examples 99 Note that Dyakonova claims that question such as (15) can get either a yes/no-interpretation if a rising accent is on the final verb or even a wh-question reading if a raising accent is on the wh-word and the verb that follows the wh-phrase is completely de-accented, see Dyakonova (2009) for more. 100 Citko (1998) shows that in Polish bare wh-pronouns, which are interpreted as indefinite pronouns, and negative pronouns appear in the clause initial position. Progovac (2005) observes parallel behavior of negative, polarity and (interrogative) wh-words in Serbo-Croatian. She shows that unmarked order of these phrases in Serbian is preverbal, which means that (i.a) is unmarked while (i.b) is emphatic. In fact, Progovac proposes that all polarity items in Serbian check in this projection. Since wh-words also appear in a preverbal position, Progovac takes this to show that wh-phrases come with a polarity feature ([-neg] in this proposal) which, according to Progovarc, is additionally confirmed by the fact that wh-words can be used as indefinites in Serbian. In Slovenian, these indefinite wh-phrases can appear in a preverbal position too. But this is not wh-fronting and it also happens with prefixed indefinites, (iii), and also negative pronouns, (iv). (i) (ii) (iii) Serbian (Progovac 2005: (28)) a. On nikoga ne voli. he nobody not like ‘He doesn’t like anybody.’ b. ? On ne voli nikoga. Serbian (Progovac 2005: (34)) Da li je on koga uvredio? that Q AUX he whom insulted ‘Has he insulted someone?’ A je France kaj / nekaj QY/N AUX France what.INDEF something ‘Did France write anything/something?’ 172 napisal? write with simple wh-pronouns, the same behavior can also be observed with D-linked whphrases, (17). (13) Serbo-Croatian Ima šta has da what to kupi? buy ‘There is something for him to buy.’ (14) Polish (Cheng 1991: (82)) Czy Janek tam kogo zobacył? whether Janek there who.ACC saw ‘Did Janek see anyone?’ (15) Russian (Dyakonova 2009: 200, (39)) My v vyhodnye kuda pojdem? we.NOM on weekends where go.FUT.1PL ‘Are we going anywhere on the weekends?’ (16) Slovenian A si QY/N did jedel kaj? eat what.ACC ‘Did you eat anything?’ (17) A si že QY/N AUX already prebral katero knjigo? read which book.ACC ‘Have you read any books yet?’ As already indicated, bare wh-pronouns can be used in conditionals as well. For example, wh-phrases in Russian can be bound by the non-interrogative operator esli ‘if’ (Strahov 2001). Cheng (1991) shows that a bare wh-form can also be used in conditionals in Polish, (19). Wh-words can also be used in conditionals in Slovenian, as I show in (20). (iv) A ni France QY/N AUX.NEG France ‘Didn’t France write anything?’ nič nothing napisal? write 173 (18) Russian (Strahov 2001: (6b, c)) a. Esli čto if slučit'sja, to what.NOM happens zvoni mne nemedlenno. then call me.DAT immediately ‘If anything happens, call me immediately.’ b. Esli kogda budeš’ v našom rajone, to zaxodi if when will in our area then come by ne stesnjajsja. not be-shy ‘If sometime you are in our area, drop in, don’t be shy.’ (19) Polish (Cheng 1991: (83)) Jeżeli kto tu if zapali who here light papierosa, to cigarette ja then I się wścieknę. refl will get mad ‘If anyone smokes here, then I will get mad.’ (20) Slovenian a. Če bo kdo prišel, if AUX who.NOM come me pokliči. me.DAT call ‘If anybody/somebody comes, call me.’ b. Če bo Miha koga If AUX Miha.NOM who.ACC poklical, bomo imeli težave. call AUX have trouble ‘If Miha calls anybody/somebody, we will have problems.’ Cheng (1991) calls this kind of reading of wh-pronouns in Polish a ‘polarity reading’, since wh-pronouns receive this reading in polarity/affective contexts, that is, yes/no-questions and conditionals. According to Cheng, in these cases in Polish the wh-pronoun is interpreted as ‘anyone’ or ‘anyhow’. Citko (1998) takes these bare wh-pronouns to be indefinite pronouns. In Russian wh-pronouns receive either an interrogative, indefinite or a universal interpretation, see Zavitnevich (2001). For example kto ‘lit. ‘who’ can be interpreted as ‘who’, ‘someone’, ‘somebody’, ‘anyone’, ‘anybody’, or ‘everyone’ and čto lit. ‘what’ is interpreted as ‘what’, ‘something’, or ‘anything’. Similarly, Slovenian bare wh-pronouns, such as kdo lit. ‘who’ or kaj lit. ‘what’, receive the interpretation ‘anyone’ or ‘someone’ and ‘anything’ or ‘something’ in polarity contexts (e.g. conditionals and yes/noquestions). In addition, in some languages, such as Mandarin Chinese, indefinite 174 pronouns can also be licensed under negative markers, as Cheng (1991) notes. In Polish and Slovenian on the other hand, bare wh-pronouns cannot appear under negation. Cheng (1991) shows that in Polish a negative affix needs to be attached to a bare wh-pronoun under negation, (21), and proposes that this might be because of negative concord. Example (22) shows that the negative pronouns need to be used under negation in Slovenian too. (21) Polish (Citko 1998: (8a)) Nikt nic nie widzał. nobody nothing NEG saw ‘Nobody saw anything.’ (22) Slovenian Nihče ni nič videl. nobody NEG.AUX nothing see ‘Nobody saw anything.’ And finally, wh-questions need to be discussed. In Slovenian, a bare wh-phrase in situ in a wh-question with a fronted wh-phrase cannot be interpreted as an indefinite pronoun, as shown in (23). Only wh-pronouns combined with the indefinite prefix ne- are interpreted as indefinite in wh-questions in Slovenian. A question with one fronted wh-phrase and an indefinite pronoun (with the prefix ne-) will of course be interpreted as a single wh-question. This is shown below, in examples (23) in which (23a) is interpreted as a multiple wh-question and (23b) as a single question, as shown with the answers that are felicitous for such questions. The observation that wh-in-situ in multiple wh-questions is interpreted as an interrogative is in line with Šimík’s (2010) observation that only answers in which we assign values to both wh-phrases are felicitous responses to questions with wh-in-situ in Czech (see example (25) and section 7.1.1). (23) a. Kdo je kupil kaj? who buy what AUX ‘Who bought what?’ (Mary bought beer./ #Mary did.) 175 b. Kdo je kupil nekaj? who buy something AUX ‘Who bought something?’ (#Mary bought beer. / Mary did) There seems to be some variation between speakers of multiple wh-fronting languages with respect to the interpretation of wh-questions with non-fronted whphrases. For example, Wachowicz (1974) states that Polish examples such as (24) are interpreted as a yes/no-question (i.e., example (24) is interpreted as ‘Did Monika give anything to anybody?’) in which both wh-phrases are in fact indefinite pronouns. According to Citko (2010), on the other hand, these cases are interpreted as multiple wh-questions, see chapter 6. (24) Polish (Wachowicz 1974: (23)) * Co Monika what.ACC Monika komu dała? who.DAT gave ‘What did Monika give to whom?’ Šimík (2010) observes multiple wh-questions with a wh-phrase in situ in Czech, such as the one in (25) in which the pitch accent is on the clause final constituent, to which I will also return in section 7.1.1, and argues against the analysis of bare wh-pronouns in situ as indefinites in Czech. Šimík notes that a value needs to be assigned to both wh-phrases when answering such a question in Czech. This suggests that the wh-phrase in situ is truly interrogative. (25) Czech (Šimík 2010: (7)) a. Kdo co KOUPIL? who.NOM what.ACC bought ‘Who bought what?’ b. Kdo koupil CO? Based on what we have seen so far, we can conclude that in Slovenian, whpronouns that appear in situ can be interpreted in two different ways: they either 176 receive an interrogative reading when they appear in a question in which a wh-phrase appears in the clause initial position or they receive an indefinite reading when they appear in a polarity context, such as a conditional or a yes/no-question. Because of this, I propose that a wh-phrase in situ in a multiple wh-question is an underspecified wh-pronoun, which is licensed by another element in the clause. Note that I am not arguing that the wh-in-situ is an indefinite pronoun in Slovenian. Rather I am arguing that the wh-in-situ is a wh-phrase containing a bare wh-pronoun that receives an interrogative interpretation because of the features in the Left Periphery. More about the proposal follows in the next section. 5.1.2 Two readings of bare wh-pronouns The following facts are crucial for the proposal I am making here for multiple whquestions with a wh-phrase in situ in Slovenian. First, a bare wh-pronoun can stay in situ only in multiple wh-questions in which a wh-phrase appears in the clause initial position, i.e. the Interrogative Projection of the Left Periphery. There is empirical support for the fact that a wh-phrase in the clause initial position, i.e. Interrogative Projection, is crucial for wh-in-situ – examples in which a wh-phrase is moved to a non-initial position, such as the Wh-Projection preceded by a focused phrase, are not felicitous under a wh-question reading (they can get a yes/no- or an echo-question reading): (26) MIHA je komu Miha AUX who.DAT kupil kaj? buy what.ACC #‘What did MIHA buy for whom?’ ‘Did MIHA buy something for somebody?’ Second, in a multiple wh-question in which one wh-phrase appears in the clause initial position, a wh-phrase in situ always gets an interrogative reading. In such cases, a wh-phrase in situ never gets an indefinite interpretation, (27), and a value needs to be assigned to both wh-phrases. In these environments the only pronoun that will get an indefinite reading is a fully pronounced indefinite pronoun, (28). In turn, with a yes/no-question marker a wh-phrase in situ never gets an 177 interrogative reading, (29).101 The asymmetry in interpretation also holds if the whphrase in situ is a D-linked phrase, as (30) shows for a wh- and (31) for a yes/noquestion. (27) Kdo je who AUX jedel kaj? eat what ‘Who ate what?’ (John ate cake./# Mary.) (28) Kdo je who AUX jedel nekaj? eat something ‘Who ate something?’ (John. / # John ate cake.) (29) A je jedel kaj? QY/N AUX eat what.ACC ‘Did he eat something?’ (Yes./# (He ate) cakes, cookies, ice cream.) (30) Kdo je who AUX kupil katero knjigo? buy which book ‘Who bought which book?’ (Tom bought The Hobbit. /# Yes.) (31) A je kateri QY/N AUX which kos tvoja kreacija? piece your creation ‘Are any of the pieces your creation?’ (#This statue./Yes.) 101 There is another option for this type of questions: They can be assigned a yes/no-intonation (i.e. rising intonation, regardless if they have an interrogative complementizer a/ali).In such cases whphrases are again interpreted as indefinite pronouns. (i) Je jedel kaj? [Rising intonation] eat what ‘Did he eat anything? (Answer: Yes/No./ # John ate cake.) AUX 178 Because the examples so far indicate that the interpretation of a wh-pronoun in situ is connected to what appears in the clause initial position, I propose that whphrases in situ in Slovenian are bare wh-pronouns that need to be licensed by another element in the clause. The reason for this is a deficiency of one class of pronouns – because of this deficiency the bare wh-pronouns in situ behave differently from whphrases which undergo movement. This means that the situation with Slovenian whin-situ is opposite to that which Cheng and Rooryck (2000) propose for French whin-situ – in French, wh-in-situ is a result of an underspecified Q-morpheme on the C0 head and not on the wh-phrase that stays in situ (as I propose for Slovenian). In French a wh-phrase can either move or stay in situ. (32) French (Cheng and Rooryck 2000: (4a), (6), (5b)) a. Jean a acheté quoi? Jean has bought what ‘What did Jean buy?’ b. Jean a acheté un livre? Jean has bought a book ‘Jean has bought a book?’ [rising intonation in French and English] c. Quel livre est-ce que Jean a acheté? which book EST-CE QUE Jean has bought ‘Which book did Jean buy?’ Cheng and Rooryck (2000) propose that in examples with wh-in-situ, such as (32a) above, the head C comes with a underspecified Q-morpheme, represented as [Q: ]. This underspecified morpheme can either be valued as [Q: wh] or [Q: y/n]. In yes/noquestions, such as (32b), the value of the underspecified Q-morpheme is set to [Q: y/n] by a default operation and the question is interpreted as a yes/no-question. In cases with wh-in-situ the Q-morpheme licenses the wh-phrase in situ. At LF the underspecified Q-feature is specified via feature movement of the wh-feature of the wh-phrase. If this did not happen, the Q-feature would be specified with the default yes/no-value, which would yield an illegitimate interpretation in cases with a whphrase. 179 The proposal by Cheng and Rooryck (2000) is not applicable to Slovenian, since at least one wh-phrase needs to move to the clause initial position, which indicates that wh-fronting is, at least to some extent, obligatory. However, the idea of featural deficiency can be applied to Slovenian data – I propose that the difference between interrogative wh-phrases that move and those that remain in situ is in the wh-phrases themselves. Wh-phrases enter the derivation with an uninterpretable valued wh-feature that is checked because of movement to Interrogative Projection or the Wh-Projections, as I have, based on Soare (2007), claimed in section 3.6. On the other hand, bare wh-pronouns in situ are underspecified for a wh-feature and therefore do not need to move to the Interrogative Projection or the Wh-Projections. But because of their deficiency, these wh-phrases need to be licensed by another element in the sentence. We can summarize the wh-pronouns that are of interest here102 as: (33) Pronouns with interrogative and indefinite reading (i) Interrogative wh-pronouns Wh-pronouns specified with a wh-feature: these wh-pronouns receive an interrogative reading and move to check the wh-feature in the Left Periphery. (ii) Indefinite pronouns Wh-pronouns specified with an indef-feature which is, in Slovenian, specified on the phonologically overt prefix ne-. These are not dependent on any special contexts. (iii) Underspecified bare wh-pronouns These stay in situ and need to be licensed. Based on what licenses them, they get one of the readings that the wh-pronouns in (i) or (ii) get.103 102 I am here only mentioning wh-pronouns prefixed by ne-, which creates indefinite pronouns like ‘someone’ etc., but other prefixes exist in Slovenian. These are, for example: marsi- ‘quite a few’ as in marsikdo ‘quite a few people’, malo- ‘few’ as in malokdaj ‘a few times’. For an overview see Greenberg (2008). 103 While I here argue that only a subset of wh-pronouns is underspecified in Slovenian, a different proposal was made for Russian by Zavitnevich-Beaulac (2005) who argues that wh-phrases in single wh-movement, in situ, and multiple wh-fronting languages are universally underspecified in the lexicon and that wh-phrases in wh-movement languages are wh-proforms combined with a question operator Q. For more see Zavitnevich-Beaulac (2005). 180 Cheng (1991) proposes a similar division of wh-phrases in multiple wh-fronting languages, such as Polish, and argues that bare wh-words in these languages can get an interrogative or a polarity reading (Hungarian on the other hand only allows the interrogative reading). Cheng assumes that wh-words are made up from a core and a determiner104 and that the core itself does not have/inherit quantificational force. Under this approach there are three options for wh-phrases in multiple wh-fronting languages: a wh-core can be combined with an affix (the determiner), which contributes existential quantificational force, and thereby gets an indefinite reading, the wh-core can be combined with a wh-determiner and be interpreted as an interrogative wh-phrase, or the wh-core appears in a polarity context and is interpreted as an indefinite pronoun. This is shown for the Hungarian pronoun valaki ‘someone’. In (34), the core, NP ki, has no quantificational force, and it is the D (vala) that contributes existential quantificational force. The Hungarian valaki ‘someone’ is then similar to Slovenian pronouns described in (33, ii). (34) (Cheng 1991: (66)) DP 3 D’ 3 D NP vala ki Applying such reasoning to Slovenian leads to the conclusion that wh-pronouns are specified with a prefix ne- and do not need to be licensed by an operator in the context. This is already done by the prefix, which carries an indef-feature. This accounts for the fact that indefinite pronouns can appear in declarative sentences, as for example nekoga ‘somebody.ACC’ in example (12) in section 5.1.1, or in an interrogative context, such as in (28) with nekaj ‘something’, and always receive an indefinite reading (i.e. they have existential force). 104 Such an analysis is problematic under the assumption that some Slavic languages (like SerboCroatian) are article less and do not project the DP layer (cf. Bošković 2008), but note that some researchers propose a DP layer despite the lack of articles (cf. Pereltsvaig 2007). 181 For wh-phrases, such as ki ‘what’ in Hungarian, (35), Cheng (1991) proposes that the core is again without interrogative force and the interrogative force is contributed by the null determiner, which Cheng defines as [D 0 [+wh]].105 (35) Cheng 1991: (67) DP 3 D’ 3 D NP 0 [+wh] ki Again, a similar analysis can be applied to Slovenian: interrogative wh-pronouns, as described in (33, i) are specified with a wh-feature and move to the Left Periphery in order to check the wh-feature on the Interrogative Projection or the Wh-Projections. This movement was described in section 3.6. For bare wh-pronouns in polarity contexts, Cheng proposes that these pronouns are interpreted as having an existential reading. Cheng proposes that the bare whform is interpreted as an indefinite when the wh-word has no determiner, but just as a core. In these cases the core is a polarity item and therefore needs a polarity trigger (for example, a yes/no-question marker). The existential force is achieved with the rule of Existential Closure, as in Heim (1982), which introduces a non-overt existential quantifier to a sentence and then binds an indefinite (allowing it to be interpreted existentially). In Cheng (1991), Existential Closure introduces an (phonologically null) existential quantifier which binds the core and gives it existential force.106 105 In Cheng (1991) this null determiner needs to be licensed in multiple wh-fronting languages and it is licensed trough Spec-Head agreement with the head C which carries a [+wh] feature. In multiple wh-languages, this holds for all wh-phrases and because all wh-phrases need to be in a local relation with the C head, all wh-phrases have to move. For details of the account see Cheng (1991). 106 Yanovich (2005) observes that bare indefinites, (i.b), are existentially closed in the scope of the question and differ from interrogative kto ‘who’, (i.a) in that bare indefinites need a licensor in Russian but interrogative pronouns do not. (i) Russian (Yanovich 2005: (29)) a. Kto prišel? who came ‘Who came?’ 182 What Cheng (1991) does not predict is that bare wh-pronouns also get an interrogative reading, which I propose for Slovenian.107 I focus on how wh-in-situ is interpreted as interrogative in Slovenian in the next section. 5.1.3 Assigning interrogative reading to wh-phrases in situ in Slovenian The previous section showed that underspecified bare wh-pronouns receive an indefinite reading in polarity contexts and an interrogative reading when the clause has a clause initial interrogative wh-phrase. The underspecified wh-phrases in Slovenian are in this respect similar to wh-phrases in wh-in-situ languages in that whin-situ in both these cases needs to be licensed by some other element. Cheng (1991) observes that in languages, such as Japanese, Korean, and Mandarin, wh-words are lexically ambiguous. For example, in Mandarin Chinese a wh-word can be interpreted as either an interrogative wh-word, (36), or as an existential quantifier when in the scope of negation, in yes/no-questions, (37), and in conditionals. This is reminiscent of bare wh-pronouns in Slavic languages, as was shown above. (36) Mandarin Chinese (Cheng 1991: (28), (3)) Shei mai-le sheme (ne) who buy.ASP what QWH ‘Who bought what?’ (37) qiaofong mai-le Qiaofong buy.ASP sheme ma what QY/N ‘Did Qiaofong buy anything?’ b. Razve kto prišel? is-it-true-that who came ‘Is it true that someone came?’ 107 Cheng’s (1991) proposal has already been extended in the past. For example, Zavitnevich (2001) assumes Cheng (1991), and proposes that in Russian all interrogative wh-words are polarity items with a [+Q] feature (and an [F] feature for focus). They receive this feature at S-structure and are therefore not bound by the same operator, which means that they all have to move to the clause initial position in order to check their [+Q] feature in Russian. In English wh-phrases all have a [+Q] feature in the lexicon which means that all of them can be bound by the same operator and that only one whword needs to move and the rest stay in situ. For more see Zavitnevich (2001). 183 As Cheng notes, a similar observation was made in, for example, Nishigauchi (1990) about Japanese. In Japanese, the wh-word also gets interpreted in relation to the particle with which it occurs. That is, when it appears with a particle -ka it is interpreted as an existential, when it is in the scope of the particle -mo as a universal, (38b), and when in the scope of wh-particle -ka it is interpreted as an interrogative wh-phrase, (38a). (38) Japanese (Cheng 1991: (20a, b)) a. Dare-ga ki-masu-ka who N come-Q ‘Who is coming?’ b. Dare-ga ki-te mo boku-wa aw-a-nai who-N come Q I-T meet-not ‘For all x, if x comes, I would not meet (x).’ Cheng accounts for the behavior of wh-phrases in Mandarin Chinese by building on Heim’s (1982) proposal for indefinites and Nishigauchi’s (1990) proposal for Japanese wh-/indefinite alternation. Nishigauchi (1990) claims that, in Japanese, wh-words are like indefinites in that neither have inherent quantificational force. This has previously been shown for indefinites in Heim (1982). Because indefinites do not have their own quantificational force they need to receive it from other elements which have inherent quantificational force (e.g., from adverbs of quantification such as always108) or an interpretative rule. In Heim (1982), this rule is Existential Closure, which introduces a non-overt existential quantifier to a sentence and then binds an indefinite (allowing it to be interpreted existentially). Cheng builds on Nishigauchi’s proposal that takes wh-words and indefinite pronouns in Japanese as lacking inherent quantificational force. The quantificational force is only determined by the particle. That is, only when a wh-word appears with an interrogative particle -ka in Japanese, it is interpreted as an interrogative wh- 108 In example (i) the indefinite gets a universal reading because it is bound by always: (i) If a man owns a donkey, he always beats it. 184 phrase, see example (38) above. In Mandarin Chinese the quantificational force is determined in a similar way, with a wh- or a yes/no-particle, a negative marker or a universal marker. The particle is a licensor that licenses wh-words as polarity items, but they also need a binder. The binder in yes/no- and negative contexts is the Existential Closure, which determines the quantificational force. As for the interrogative reading, only association with a wh-particle will lead to an interrogative interpretation of a wh-phrase. Wh-words that receive an interrogative reading are, however, slightly different from polarity items in Mandarin Chinese in that the whparticle is both the licensor and the unselective binder of the wh-word. This means that the particle contributes the interrogative force to the wh-word in Mandarin Chinese (potentially more than one wh-word at a time). I propose that cases of wh-phrases in situ in Slovenian multiple wh-questions are similar to Japanese in that both languages possess a wh-phrase that needs to be licensed by some other element in the sentence in order to get an interrogative reading. This means that, just as in the case with wh-words in Japanese, bare whpronouns in Slovenian lack quantificational force. The difference between receiving the interrogative reading in the two languages is in what licenses the wh-phrase – in Japanese the interrogative reading is licensed by an overt interrogative particle, but in Slovenian there is no overt interrogative particle and the licensing is done by the complex Q+wh-feature. As already shown in section 3.6, I have argued that in Slovenian wh-questions an interpretable complex Q+wh-feature is located in the Interrogative Projection in the Left Periphery. The Q+wh-feature comes with an EPP-subfeature and one whphrase with an uninterpretable wh-feature moves in order to check the EPP feature on the Interrogative head. Checking of the Q+wh-feature is responsible for Clause Typing, but the interpretable Q+wh-feature is also responsible for licensing wh-insitu in Slovenian. Crucially, the Q+wh-feature on the Interrogative Head is interpretable, and because of this, visible at LF (see for example Cinque and Rizzi 2008). We have seen above that bare wh-pronouns in Slovenian are similar to the ones in Russian. I will take two properties of indefinite pronouns to hold in Slovenian. First, the wh-phrase in situ is a bare wh-pronoun that has no wh-feature and hence does not agree with the Q+wh-feature on Inter0 and does not move (nor can it agree 185 with the interpretable wh-feature on the Wh-head). Second, because it is a bare indefinite pronoun it needs to be licensed. In multiple wh-questions the licensing operator is the interpretable Q+wh-feature on the Inter0 head. If the underspecified wh-phrase is in the scope of the Q+wh-feature the wh-phrase in situ is interpreted as an interrogative wh-phrase. This means that it is not the moved wh-phrase that assigns a reading to the wh-phrase in situ. Rather, one wh-phrase which has the whfeature and which agrees with the complex Q+wh-feature moves because of the EPPsubfeature on the complex Q+wh-feature, but it is the interpretable complex feature in the Interrogative head that is responsible for licensing the interrogative reading of the wh-phrase in situ. The situation is then slightly different in yes/no-questions. These do not come with a complex Q+wh-feature, but rather just a Q (to which I will refer as QY/N). This QY/N can be checked with an overt yes/no-complementizer a/ali ‘or’ or with a phonologically null complementizer which leads to a yes/no-, i.e. rising, intonation. This QY/N also cannot be checked by a wh-phrase (since there would be no agreement). Because the bare wh-pronoun in situ in yes/no-questions is not in the scope of the complex Q+wh-feature it does not receive an interrogative reading. I will assume that in these cases, the QY/N is the licensor and the bare wh-pronouns are interpreted as indefinite pronouns because they are existentially closed, see Yanovich (2005) for Russian.109 For example, we can account for the readings of examples such as (25) in such a way. In this case the Interrogative head does not host a complex Q+wh-feature, because of which no wh-phrase moves to the clause initial position. Because the bare wh-pronoun is not in the scope of this Q+wh-feature it 109 A question which has been pointed out is why an existential quantifier can intervene between the QY/N and the bare wh-pronoun but a situation in which an existential quantifier intervenes between a Q+wh and the wh-phrase in situ is not allowed. A possible solution was suggested by the anonymous reviewer for Mišmaš (2015), that is, a violable constraint which requires licensor and binder to be the same. So when a QWH, such as in Q+wh, can license and bind a wh-pronoun it has to do so, but the QY/N is not able to bind and a quantifier must be introduced, as in yes/no-questions with bare whpronoun. I agree with such a proposal – assuming that the two Qs are different in that QY/N does not have inherent quantificational force, while QWH in Q+wh has it, introducing an existential quantifier in a wh-question with a wh-in situ would be superfluous since the QWH can both bind and license the bare wh-pronoun in situ. In fact, a similar proposal has been made by Cheng (1991: 127, 128) who proposes that the derivation in which one particle can both license and bind is less costly than a derivation in which an existential quantifier is introduced by Existential Closure (Cheng is assuming the Principle of economy of derivation as in Chomsky (1989)). Because of this, Existential Closure only applies when there is no other binder available. 186 cannot get an interrogative reading (but does get an existential reading because it is in the scope of a QY/N). The wh-item in situ is therefore interpreted based on which operator has scope over it.110 While licensing wh-in-situ might resemble wh-scope marking questions in Slovenian, the two are in fact different phenomena. If we compare the partial question (i.e. a wh-scope marking question, see Stepanov (2000)) in (39) to the multiple wh-question with a wh-phrase in situ, (40), we can see that we can answer ‘Miha invited Tone’ to both. This means that in answering (39) we only provide referents for the wh-phrases in the embedded clause. In questions such as (40) we need to provide referents for all wh-phrases. 110 We could potentially test if the licensing relation between the licensor and the bare wh-pronoun is crucial for determining the interrogative reading of the bare wh-pronoun with intervention effects in the sense of Beck (2006). I take (i) to describe the generalization regarding intervention effects and intervention effects can be schematized as in (ii). An example of intervention effects is shown in (iii). The wh-scope marking question in German is ungrammatical when niemand ‘nobody’ (the quantificational element) intervenes between the wh-phrase in the scope position of the wh-scope marking question. (i) Beck 2006: (9) A quantificational or focusing element may not intervene between a wh-phrase and its licensing complementizer. (ii) Beck (2006: (11)) * [Qi [… [ intervener [… wh-phrasei… ]]]] (iii) German (Beck 2006: (8a, b)) * Was glaubt Luise/ *niemand wen Karl gesehen hat? what believes Luise nobody whom Karl seen has ‘Who does nobody/Luise believe that Karl saw?’ Questions in (iv) are grammatical in Slovenian. All three can receive either a single pair or a pair list answer, but example (iv.a) can also be answered by only referring to the moved interrogative whphrase kaj ‘what.ACC’ (komu ‘who.DAT’ receives an indefinite reading). This indicates that a quantified element intervenes between the bare wh-pronoun and the Interrogative Projection that licenses it. In examples (iv.b, c) such single answer is not possible. I suggest this is the case because komu ‘who.DAT’ does not need licensing by a licensing operator – since it is moved, I assume komu ‘who.DAT’ in these cases is an interrogative wh-pronoun. Because intervention effects are something that needs to be further explored in Slovenian and deserve more attention, I will not be considering them any further. (iv) a. b. c. Kaj je samo Miha kupil what.ACC AUX only Miha buy ‘What did only Mihy buy for whom?’ Kaj je samo Miha komu kupil? Kaj je komu samo Miha kupil? komu? who.DAT 187 (39) Kaj misliš, what think kdo je who.NOM AUX koga povabil na večerjo? who.ACC invite on dinner ‘What do you think, who invited who to dinner?’ (40) Kdo who.NOM je na večerjo povabil koga? AUX on dinner introduce who.ACC ‘Who introduced who to dinner?’ The examples so far have shown that the features of the Interrogative Projection are crucial for licensing wh-in-situ in for matrix clauses, but the same also holds for wh-in-situ in embedded clauses. In fact, these confirm the influence of the features of the Interrogative Projection. According to Yanovich (2005), bare indefinites in Russian need to associate with the nearest licensor. This also holds for bare wh-pronouns in Slovenian. As example (41) shows, the wh-phrase in situ gets interpreted as an indefinite pronoun when the embedded clause is headed by če ‘if’ – in this case the embedded clause is a yes/no-question. In such a question, the whphrase in situ can also be pronounced with an emphasis, and in this case the embedded clause is interpreted as an echo-question. But crucially, when the embedded clause is headed by a wh-phrase, as in (42), and has a wh-phrase in situ, the wh-in-situ is interpreted as an interrogative pronoun.111 (41) Koga who.ACC je spraševal, AUX ask če je Jože if Jože.NOM Maja.ACC invite AUX Majo povabil kam? where ‘Whom did he ask if Jože invited Maja somewhere?’ (42) Koga who.ACC je Miha spraševal, kdo je Majo AUX Miha ask Maja.ACC invite who AUX povabil kam? where ‘Whom did Miha ask who invited where?’ The interpretation of the bare wh-pronoun in Slovenian therefore hinges on the first potential licensor. Only when the licensor is the Q+wh-feature is the bare whpronoun interpreted as an interrogative pronoun. 111 In (42) then the wh-in-situ in the embedded clause can either take the embedded or the matrix scope and the answer can either be ‘Miha asked Črt who invited Maja where’ or ‘Miha asked Črt who invited Maja to the cinema’, respectively. 188 So far I have proposed an analysis of instances of wh-in-situ in multiple whquestions in Slovenian. I have argued that wh-phrases in situ are underspecified bare wh-pronouns which need to be licensed by some other element in the clause. In whquestions in Slovenian, the bare wh-pronoun is licensed by the Q+wh-feature on the Interrogative head and receives an interrogative meaning. On the other hand, there are also wh-pronouns specified with the wh-feature in Slovenian. These interrogative pronouns have to undergo wh-movement, as described in section 3.6. In the second part of this section I discuss some additional data that are of interest for the proposal. 5.2 Further issues In this section I look at data that might appear problematic given the proposed account. I show that the proposed analysis can be maintained despite apparent issues. I start the section with questions with a clause final subject wh-phrase, continue with dialectical variation in the use of different wh-pronouns and observe the behavior of koliko ‘how much/many’ and zakaj ‘why’. Finally, I also look at the data from Serbo-Croatian to see whether the proposal for instances of wh-in-situ in Slovenian can be extended to Serbo-Croatian. 5.2.1 Cases with a wh-subject in a clause final position There is another type of multiple wh-questions in Slovenian that needs to be considered. These are multiple wh-questions in which the wh-phrase referring to the subject appears in the sentence final position when another wh-phrase moves to the clause initial position.112 An example of such a question is shown below: 112 (i) Yes/no-questions with a subject in the clause final position are also possible: A pije kdo? QY/N drink who.NOM ‘Is anybody drinking?’ 189 (43) Kaj pije kdo? what.ACC drinks who.NOM ‘Who drinks what?’ These cases are potentially problematic because the wh-subject appears after the verb. If we assume that the nominative wh-phrase (the subject) is merged in the specifier of the vP and the wh-subject in these cases stays in situ, we would expect the reverse word order with the subject preceding the verb. While the expected word order is possible, as (44) shows, example (43) shows that the word order in which the verb precedes the wh-subject is also possible. The question is how to account for examples like these. (44) Kaj kdo pije? what.ACC who.NOM drinks ‘Who drinks what?’ The account of multiple wh-questions with a wh-subject in the clause final position is based on two separate analyses. First, in line with what was said in the previous section, the wh-phrase kdo ‘who’ in this case is a bare wh-pronoun which receives its interpretation in relation to the moved wh-phrase. The second part hinges on the behavior of verbs in Slovenian. Specifically, Ilc and Milojević Sheppard (2003) show that Slovenian has optional V-to-Aspect movement of the verb. They come to this conclusion based on examples with floating quantifiers – the fact that floating quantifiers can appear either before or after the verb indicates that the verb can either move or not move to Asp0 in the IP layer (see Ilc and Milojević Sheppard (2003) for the arguments behind movement to Asp0): (45) Ilc and Milojević Sheppard (2003: (30a, b)) a. Otroci vsi ljubijo čokolado. children all love chocolate.ACC ‘The children all love chocolate.’ b. Otroci ljubijo vsi čokolado. 190 Returning to the wh-question in (43), I propose that in these cases the whphrase kaj ‘what’ moves to the InterP, the bare wh-pronoun kdo ‘who’ stays in situ because it does not have a wh-feature that would motivate movement and finally, in these cases the verb undergoes movement to the Aspect head in the IP-layer. Since Asp0 is located above the Spec-vP projection in which the subject is merged, the word order in (43) is accounted for. As for example (44), there are two possible explanations – if we assume that the verb in these cases does not undergo optional movement to Asp0 and kaj ‘what’ moves to SpecInterP, then the wh-subject either stays in situ or moves to a Wh-Projection. Because multiple wh-questions with multiple movement and with wh-in-situ both receive the same interpretation, as I will show in section 7.1.1, we cannot distinguish between the two possibilities. 5.2.2 Two types of ‘what’ and other dialectical variation In some Slovenian dialects the standard Slovenian wh-pronoun kaj ‘what’ comes in two different varieties: kaj and kej. While we have seen that the standard Slovenian kaj ‘what’ has two meanings (i.e. interrogative and indefinite in polarity contexts), each variant in these dialects only gets one interpretation: kaj only gets an interrogative and kej only gets an indefinite reading. I will focus on data obtained from speakers from Ljubljana and Celje in this section and show that these data offer a confirmation for the proposal from the first part of this section (i.e. that the Q+wh feature in the Left Periphery is responsible for licensing of the interrogative reading of the wh-phrases in situ). Interestingly, this distinction only happens with kaj and with no other wh-pronouns and the distinction does not exist in all Slovenian dialects. Specifically, Western Slovenian dialects (for example, Slovenian as spoken in Nova Gorica) have a wide use of kej ‘what’ (or ki ‘kaj’ in some varieties of Western Slovenian dialects) as both an interrogative and indefinite pronoun. In these dialects the use of kej then parallels the use of kaj ‘what’ in standard Slovenian. I will leave Western Slovenian dialects aside. The described distinction between kaj ‘what’ and kej ‘something’ can be observed in single wh-questions, (46). In these questions the interrogative wh-phrase has to move to the clause initial position. In single wh-questions, speakers that use 191 the kaj/kej distinction only use the wh-phrase kaj ‘what’ and not kej ‘something’. This confirms our initial observation that only kaj has an interrogative reading. Also, neither can be used in declarative sentences, (47). (46) a. Kaj je Jože kupil? what AUX Jože buy ‘What did Jože buy?’ b. * Kej something (47) a. * Jože je Jože AUX b. ?* Jože je Jože AUX je Jože kupil? AUX Jože buy kupil kaj. buy what kupil kej. buy something Based on the proposal in the previous section, what we expect is that in polarity contexts one will find kej ‘something’ but in wh-questions, speakers of central Slovenian dialects will only use kaj ‘what’. This is exactly what we find. For example, in yes/no-questions (an instance of a polarity context) speakers only find the examples with kej ‘something’ grammatical, (48a). That is, in yes/no-questions, speakers of central Slovenian dialects cannot use kaj ‘what’, (48b). And as expected, in wh-questions, speakers can only use kaj ‘what’ in situ. Example (49) shows that wh-questions with kej ‘something’ in situ are judged as ungrammatical for the interrogative reading of the wh-in-situ. (48) a. A je Jože kupil kej? QY/N AUX Jože.NOM buy something.ACC ‘Did Jože buy anything?’ b. * A je Jože kupil kaj? QY/N AUX Jože buy what (49) a. Kdo je kupil kaj? who.NOM AUX buy what.ACC ‘Who bought what?’ 192 b. * Kdo je who AUX kupil kej? buy something This means that a different interpretation of the bare wh-pronoun results in a different phonological form for the bare wh-pronoun. When it is licensed by a yes/no-question marker and receives an indefinite reading it gets the kej form. When it is licensed by the Q+wh-feature and is interpreted as an interrogative pronoun, it receives the kaj form. A potentially problematic piece of data comes from multiple wh-fronting. In questions with multiple wh-fronting, speakers only get a true wh-question reading with kaj ‘what’, but questions with kej ‘something’ are also possible. (50) a. Kdo je kaj kupil? who.NOM AUX what.ACC buy ‘Who bought what?’ b. Kaj je kdo kupil? c. # Kdo je who AUX kej kupil? what buy That kaj ‘what’ can appear in multiple wh-fronting is not surprising. If we assume that it is specified with a wh-feature, then we also expect it to move. It is however surprising that kej can appear fronted in a wh-question. But (50c) does not get a multiple wh-reading nor does the kej in that example get an indefinite reading. These questions are different and kej seems to have a different meaning in such questions. Such meaning is also possible with the standard Slovenian kaj. The dictionary of standard Slovenian language (Slovar slovenskega knjižnega jezika, Bajec et al. 2000) recognizes that kaj ‘what’ in standard Slovenian has meanings other than the indefinite or the interrogative meaning. For example, it can be used for expressing small amounts (komaj kaj jem ‘I hardly eat’) but it also has a function of generalizing a question. For example, in (51), kaj is used as such in standard Slovenian – both (51a) and (51b) are interpreted as single wh-questions in which speakers only assign a value to the clause initial wh-phrase. 193 (51) (Bajec et al. 2000) a. Kako je kaj how what home AUX doma? ‘How is it at home?’ b. Kakšni ljudje what-kind people so kaj to? AUX what this ‘What kind of people are they?’ Speakers of central Slovenian dialects use kej in such cases to express a small amount, which helps us to account for the acceptability of (50c). In addition, speakers accept example (50c) but note that they cannot give a simple pair list answer to it (i.e. they cannot answer with ‘Homer beer, Bart a cake’). Rather, they can answer with ‘everybody/somebody did’ and then potentially offer a pair list answer (but they need to start with a general answer). This indicates that kej in these cases in fact has a different meaning from kej ‘something’ (or even ‘what’). This means that questions such as (50c) do not contradict our initial observations about the behavior of kej ‘something’, i.e. that it is a realization of bare wh-pronoun with an indefinite interpretation. To summarize, in some Slovenian dialects kaj only gets an interrogative reading and kej only gets an indefinite reading. In relation to this, kaj can be used in wh-questions and kej in polarity contexts. This means that the bare wh-pronouns that need to be licensed get a different phonological form in these dialects. In addition to the kaj/kej distinction, some variation can also be found with complex wh-phrases. I have already explored some standard Slovenian examples with D-linked phrases, e.g. (30) in 5.1.2, and have shown that in these the pronoun kateri ‘which’ is used. Other wh-words can appear in complex wh-phrases in addition to kateri ‘which’. These wh-pronouns, in standard Slovenian, have three distinct meanings, (52). In standard Slovenian, all of these can be used in yes/no-questions, as the examples (53)–(55) below show. All three can also appear in situ in whquestions, (56)–(58): (52) kakšen - what kind of kateri - which 194 čigav - whose (53) A je čigav ata že tukaj? Q is whose dad already here ‘Is anybody’s dad here?’ (54) A je kakšen Q is any zdravnik tukaj? doctor here ‘Is there a doctor here?’ (55) A Q je naredil izpit kateri od študentov? AUX pass exam which of students ‘Did any of the students pass the exam?’ (56) Kdo je who AUX opral čigav avto? wash whose car ‘Who washed whose car?’ (57) Kdo je who AUX opral kakšen avto? wash what-kind-of car ‘Who washed what kind of car?’ (58) Kdo je who AUX opral kateri avto? wash which car ‘Who washed which car?’ Variation between dialects and their complex wh-phrases can be observed with respect to which wh-word is used in multiple wh-questions with a wh-phrase in situ and in yes/no-questions. For example, speakers form Nova Gorica use kakšen/kəšn lit. ‘what kind of’ for both ‘what kind of’ and ‘which’, speakers form Laško use kateri/kir lit. ‘which’ for both ‘what kind of’ and ‘which’, and both use čigav ‘whose’, which is a separate pronoun in both. This simply means that speakers of the Laško dialect will find cases such as (58) more acceptable than (57), while speakers from Nova Gorica will find (57) more acceptable. While this might account for differences in acceptability between dialects, it does not give us any additional evidence about the behavior of wh-phrases in situ, since, contrary to the distinction with kaj/kej, variation in complex wh-phrases is not influenced by context in which a wh-phrase appears. 195 5.2.3 Koliko ‘how much/many’ and zakaj ‘why’ There are two exceptions with respect to which wh-pronouns can appear in both polarity context and in situ in multiple wh-questions. The first exception is koliko ‘how much/many’ and the second zakaj ‘why’. Koliko ‘how much/many’ is a wh-phrase that appears in both single and multiple wh-questions: (59) Koliko knjig si prebral? how-many books AUX.2SG read ‘How many books did you read?’ (60) a. Koliko knjig how-many books je kdo prebral? AUX who read ‘How many books did who read? b. Kdo je koliko knjig prebral? Wh-phrases which include koliko ‘how much/many’ behave differently from other wh-phrases in that they cannot appear in polarity contexts, as (61) shows for yes/no-questions and (62) for conditionals. Examples (61b) and (62b) show that we get the intended meaning ‘any’ with kaj lit. ‘what’. (61) a. b. *A si pojedel koliko hrušk? QY/N AUX eat how-many pears A kaj si pojedel QY/N AUX eat hrušk? what pears ‘Did you eat any pears?’ (62) a. * Miha ne b. ve, če bo bral koliko knjig. Miha not know if will read how-many books Miha ne ve, bo Miha not know if če bral kaj knjig. will read what books ‘Miha does not know if he will read any books.’ 196 Since koliko ‘how much/many’ cannot appear in polarity contexts, this implies that it lacks the ability to get an indefinite reading. This indicates that koliko ‘how much/many’ is not a bare wh-pronoun with featural deficiency. Based on this we might expect that it cannot appear in situ in wh-questions. As the examples below show, this is not the case: (63) Kdo je who AUX prebral koliko knjig? read how-many books ‘Who read how many books?’ That koliko ‘how much/many’ is indeed different from other wh-phrases is also evident from the fact that it is not compatible with the indefinite prefix ne- in the same way as other wh-pronouns (see Table 14). Specifically, while nekoliko exists, it is not simply an indefinite counterpart to koliko. If it were, we would find the pronoun nekoliko in declarative sentences where it would appear as an indefinite counterpart to the interrogative koliko ‘who much/many’, but this is not the case. Instead, nekaj ‘some’ is used. (64) a. * Maja je Maja AUX prebrala nekoliko knjig. read some books Intended: ‘Maja read a couple of books.’ b. Maja je prebrala nekaj knjig. Maja read some books AUX ‘Maja read a couple of books.’ Following The dictionary of standard Slovenian language (Bajec et al. 2000), nekoliko has a different reading. It can be used to express a small quantity, a small degree, and it can express an indefinite quantity – but it seems to typically do so for adjectives (nekoliko rdeč ‘somewhat red’). This means that the wh-pronoun koliko ‘how much/many’ does not have the indefinite counterpart with the prefix ne- or when it is licensed with a yes/no-question marker. Still, we can assume that the bare wh-pronoun koliko ‘how much/many’ exists and it gets an interrogative meaning when in the scope of the complex Q+wh-feature. However, when in the scope of a 197 yes/no-question operator QY/N- the sentence is not grammatical, because the assigned reading of the bare wh-pronoun is not available. Also incompatible with an indefinite reading is zakaj ‘why’. I have already shown that zakaj ‘why’ can appear in multiple wh-questions. This is again shown in (65). On the other hand, just like koliko ‘how much/many’, zakaj ‘why’ cannot appear in polarity contexts. (65) Zakaj je why AUX kdo opral avto? who.NOM wash car ‘Why did who wash the car?’ (66) * A si zakaj kupil avto? QY/N AUX why buy car (67) *Če je If AUX Tone zakaj opraskal avto, bom poklicala policijo. Tone why scratch car police will call But contrary to koliko ‘how much/many’, zakaj ‘why’ cannot appear in a clause final position. (68) ??Kdo je who AUX opraskal avto zakaj? scratched car why ‘Who scratched the car why?’ This is not surprising, since we have seen that zakaj ‘why’ behaves differently from the remaining wh-phrases in questions with multiple wh-fronting and needs to precede all wh-phrases in a multiple wh-questions, see section 3.4. One possible account of ungrammaticality of zakaj ‘why’ appearing in the clause final position is that zakaj ‘why’ is always specified with the wh-feature. Because of this, it also cannot appear in polarity contexts. In addition, it was suggested in the past that why is merged higher in the structure. Specifically, it was claimed that in Polish or Russian the reason why merges in CP and the purpose is why is merged in the vP and is then moved, see Stepanov and Tsai (2008). Assuming both types of why are merged higher in the structure in Slovenian too, is consistent with the fact that why does not appear in the clause final position in Slovenian questions – it cannot appear 198 in the clause final position, because it is merged high. I will not be considering zakaj ‘why’ any further, but see for example Stepanov and Tsai (2008) or Jędrejowski (2014). 5.2.4 Data from Serbo-Croatian Section 2.2.1 has shown that wh-in-situ can also be found in Serbo-Croatian, which I show again in (69).113 Crucailly, again, just as in Slovenian, the speakers need to give answers to both wh-phrases, e.g. the appropriate answer to (69) is ‘(He gave) a beer to Jim’. Aditionally, in Serbo-Croatian, just as in Slovenian, bare pronouns cannot be used in declarative sentences, (70). In these an indefinite pronoun needs to be used. 113 In fn. 16 I have shown that a wh-phrase can also stay in situ in wh-questions in Czech. However, despite the fact that there are some similarities in the behavior of wh-pronouns in Czech and Slovenian, the two languages differ in several respects. That is, in Czech (as in Slovenian) bare whpronouns cannot appear in declarative sentences and in both a wh-phrase has to appear in the sentence initial position for the question to be interpreted as a true wh-question. But in Czech bare wh-pronouns cannot appear in situ in yes/no-questions (or in conditionals). (i) a. b. (ii) a. (iii) b. a. b. Jan někoho pozval. Jan somebody.ACC invited ‘Jan invited somebody.’ * Jan koho pozval. Jan who.ACC invited * Jan komu koho představil? Jan who.DAT who.ACC introduce Intended: ‘Jan introduced who to whom?’ * Jan komu představil koho? Pozval Jan někoho? Invite Jan somebody.ACC ‘Did Jan invite somebody?’ * Pozval Tom koho? In addition to this, some note (see Šimík 2010) that in Czech the wh-phrase has to be stressed in clause final position. Also, in Czech and Slovenian questions with wh-in situ do not receive the same interpretations (see section 7). Given that bare wh-pronouns in these languages behave differently, the proposed analysis for Slovenian cannot simply be extended to Czech, as it seems that in Czech whpronouns are always specified with a wh-feature (and therefore cannot appear in yes/no-questions and conditionals). The difference between moved wh-phrases and wh-phrases in situ seems to originate from other sources, for example, as Šimík (2010) proposed for Czech, in the position of the Qparticle. 199 (69) a. Šta je kome Ivan dao? what.ACC AUX who.DAT Ivan give ‘What did Ivan give to whom?’ b. Kome je šta Ivan dao? c. Kome je Ivan dao šta? d. Šta je Ivan dao kome? (70) a. Maja je pozvala Maja AUX called nekoga. somebody ‘Maja called somebody. b. * Maja je pozvala Maja AUX called koga. who In Serbo-Croataian bare wh-pronouns can be used in yes/no-questions. There, the bare wh-pronoun does not have the interrogative reading (and there is no need to assign a value to them), but rather the same meaning as indefinite pronouns, as the example below shows.114 Bare wh-pronouns can also be used in conditionals in which they again receive the indefinite reading. (71) Da li That Q je Maja koga uvredila? AUX Maja.NOM who.ACC inslult ‘Did Maja insult somebody?’ (72) Ako Maja koga uvredi, if Maja who insults biće tužna. will-be sad ‘If Maja insults somebody, she will be sad.’ As in Slovenian, Serbo-Croatian speakers do not find examples in (73) acceptable as wh-questions. They do note that these examples are potentially better if the wh- 114 As Progovac (2005) shows, in Serbo-Croatian bare wh-pronouns that are interpreted as indefinites tend to raise, but also gives examples in which they stay in situ: (i) Serbo-Croatian (Progovac 2005: 28, (85), (ii)) a. Da li je on koga that Q AUX he who.ACC ‘Has he insulted someone?’ b. Da li je on uvredio koga? uvredio? insult 200 phrases are emphasized in which case the question is interpreted as an echo question – just as in Slovenian. (73) a. * Ivan je Ivan AUX šta kome dao? what.ACC who.DAT give b. * Ivan je kome dao šta? This means that Serbo-Croatian data are consistent with the data in Slovenian which indicates that the proposal for Slovenian can be extended to Serbo-Croatian. 5.3 Summary In this section I have argued that the difference between wh-phrases that undergo movement and wh-phrases that stay in situ in wh-questions in Slovenian is in the whphrases themselves. Specifically, I proposed that wh-phrases in situ in Slovenian are bare wh-pronouns that lack inherent quantificational force and are underspecified for the wh-feature. Because they are only interpreted as interrogative wh-phrases in multiple wh-questions in which one wh-phrase moves to the clause initial position, I proposed that it is the complex Q+wh-feature on the Interrogative head that is responsible for licensing interrogative wh-phrases in situ in Slovenian. So far I have only considered questions with multiple wh-fronting in Slovenian and cases in which a wh-phrase stays in situ in multiple wh-questions in Slovenian. In the following section I look at the third type of questions – wh-questions with short movement. 201 6 Multiple wh-questions with short movement One of the patterns of multiple wh-questions in Slovenian that was discussed in section 2.3 is the pattern in which one wh-phrase (or more) undergoes movement to the clause initial position, while another moves to some clause internal position, after the subject. I call these questions multiple wh-questions with short movement (cf. Dornisch 1998, Citko 2010). We observed the pattern for argument and adjunct whphrases in matrix and embedded questions and it can also be found with D-linked phrases. In addition, we saw that in this type of question the lower wh-phrase is located after the subject. I show this in (1) for two argument wh-phrases, but see 2.3 for more examples (including examples with more than two wh-phrases). Examples (1c) and (1d) show movement of all wh-phrases to the clause initial positions (all whphrases are located above the subject). (1) a. Kaj je Janez komu kupil? what.ACC AUX Janez who.DAT buy ‘What did Janez buy for whom?’ b. Komu je Janez kaj kupil? c. Kaj je komu Janez kupil? d. Komu je kaj Janez kupil? But before we try to account for wh-questions with short wh-movement we first need to show that this is in fact a question pattern which is distinct from multiple whfronting. The two are sometimes superficially very similar, especially given the fact that, as shown in chapter 3, wh-phrases that undergo multiple wh-fronting can be separated by several elements (cf. Rudin 1988). For example, in (2) the wh-phrases are separated by an auxiliary clitic and in (3) by the parenthetical po tvoje ‘in your opinion’ (both are repeated from 3.1.1). (2) a. Kdo je koga udaril? who.NOM AUX.3SG who.ACC hit ‘Who is hitting whom?’ b. Koga je kdo udaril? 202 (3) a. Kdo po tvoje kdaj kuha kosilo? who after you when cooks lunch ‘Who in your opinion cooks lunch when?’ b. Kdaj po tvoje kdo kuha kosilo? Further, fronted wh-phrases can be also separated by a topic phrase. This is shown in (4) in which, given the context, to prijateljico ‘this friend.ACC’ is the topic of (4a). (4) Maja je šla praznovat rojstni dan k svoji prijateljici. ‘Maja went to her friend’s birthday party.’ a. Kdaj je to prijateljico kje spoznala? when AUX this friend.ACC where meet ‘When and where did she meet this friend?’ The relevant question at this point is whether the subject that intervenes between the two wh-phrases in (1a, b) is a topic moved to the Topic position of the Left Periphery or something else. If it turns out to be a phrase moved to the Left Periphery, we could simply ignore wh-questions with short movement as a separate type of multiple wh-questions. But we can show that the nominative DPs are in fact subjects and not necessarily topics at the same time. This is evident by the fact that an indefinite phrase can act as a subject of a multiple wh-question. Since topics are typically already given in the discourse, they are also typically specific/distinct (as in example (4)). Because of this, I will assume that an indefinite DP, such as vsaj dva študenta ‘at least two students’ in (5), cannot be a topic. Still the subject can be a topic at the same time, for example ta študentka ‘this student’ is both a topic and the subject of (6).115 But it should also be noted that (6) would be an infelicitous question for the context in (5). 115 Rizzi (1997) writes that topics cannot be quantified in Italian, but this is not true for Slovenian. Given the appropriate context, such as the one in (i), one can quantify the topic, in this case vsem otrokom ‘all the children.DAT’ in (i.a): (i) Maja je v Celju spoznala otroke. ‘Maja met children in Celje.’ a. Kaj je vsem otrokom prinesla za božič? what AUX all children bring for Christmas ‘What did she bring all these children for Christmas?’ 203 (5) Maja vsak dan dobi kup daril. ‘Maja gets a lot of gifts every day.’ a. Kdaj sta vsaj dva študenta kaj Maji prinesla za darilo? when AUX at-least two student what Maja bring for gift ‘When did at least two students bring what as a gift to Maja?’ (6) Študentka vsak dan prinaša darila. ‘A student brings gifts every day.’ a. Kdaj je ta when AUX this študentka kaj student Maji prinesla what Maja bring za darilo? for gift ‘When did this student bring what as a gift to Maja?’ On the other hand, one could argue that the non-wh-subject that appears between the two wh-phrases is in the Focus Projection. However, focused phrases are pronounced with an emphasis in Slovenian, as in (7). Since subjects are not necessarily emphasized (there is no emphasis on ena študentka ‘some student’ in (5)), I will assume that they are not in the Focus Projection of the Left Periphery. Still, there are examples in which the subject is focused, as (7) shows. (7) Miha je predstavil Janeza Maji. ‘Miha introduced Janez to Maja.’ a. Ne, komu je TONE koga predstavil? no who.DAT AUX TONE who.ACC introduce ‘No, who did TONE introduce to whom?’ Given that even quantified subjects, as in (iii), can appear in the Topic position, we cannot take quantification as a test for determining if something is a Topic. (ii) (iii) Kaj je Maja komu prinesla za what.ACC AUX Maja who.DAT bring for ‘What did Maja bring whom for their birthday?’ Kaj je vsaka punca komu prinesla what.ACC AUX every girls who.DAT bring ‘What did every girl bring whom for their birthday?’ 204 rojstni dan? birthday za rojstni dan? for birthday Based on this we can conclude that the subject that is located between the two whphrases is not necessarily the topic or the focus – but that a subject can also be the topic or the focus of the sentence. Additional evidence for the behavior of multiple wh-questions with multiple wh-fronting being different from questions with short movement comes from the interpretation of adverbs that appear with multiple wh-questions. As we saw in section 3.2, Bošković (1997a et seq.) proposed an analysis of multiple wh-fronting in which he determined the position of wh-phrases, using as a test the interpretation of adverbs in Serbo-Croatian. I repeat the crucial examples below: (8) Serbo-Croatian (Bošković 1997a: (18), Stjepanović 2003: (19)) a. Ko mudro who wisely koga savjetuje? whom advises ??‘Who is it wise of to advise whom?’ ‘Who advises whom in a wise manner?’ b. Ko koga who whom mudro savjetuje? wisely advises *‘Who is it wise of to advise whom?’ ‘Who advises whom in a wise manner?’ c. Koga ko mudro whom who wisely savjetuje? advises ??‘Who is it wise of to advise whom?’ ‘Who advises whom in a wise manner? As Bošković notes (1997a: fn. 5) both (8a) and (8c) are degraded on the sentential reading, but they are better than (8b). He claims that the reason for this is a general incompatibility of sentential adverbs with questions (because of this, according to Bošković, who wisely advises whom is somewhat degraded when wisely has a sentential reading). Despite this general incompatibility Bošković claims that the adverb receives the sentential reading when it is located in the TP (or AgrsP) and that the adverb receives the manner reading when it is located in the VP (or AgroP). In (8c) the sentential reading is available, therefore we can assume that both whphrases (which precede the adverb) are located above the TP and that the adverb is in 205 the TP. The same holds for (8b), which is only different from (8c) in the word order of wh-phrases. Both readings are also available in (8a). Example (9) shows how Slovenian behaves with respect to adverbs in multiple wh-questions: (9) a. Kdo pametno komu pomaga? who wisely whom help ‘Who is it wise of to help whom?’ ‘Who helps whom in a wise manner?’ b. Kdo komu pametno pomaga? who whom wisely help ‘Who is it wise of to help whom?’ ‘Who helps whom in a wise manner?’ c. Komu kdo pametno pomaga? whom who wisely help ‘Who is it wise of to help whom?’ ‘Who helps whom in a wise manner?’ In (9), both the subject oriented and the manner readings are available when the adverb is located between the fronted wh-phrases or after them. The same also holds for examples in which fronted wh-phrases are divided by a clitic or a parenthetical, (10). (10) a. Komu po who.DAT after tvoje kdo pametno you who.NOM wise pomaga? help ‘Who is it wise of, in your opinion, to help whom?’ ‘Who, in your opinion, helps whom in a wise manner?’ b. Komu po tvoje pametno who.DAT after you wise kdo pomaga? who.NOM help ‘Who is it wise of, in your opinion, to help whom?’ ‘Who, in your opinion, helps whom in a wise manner?’ However, if we change the questions to multiple wh-questions in which the non-wh-subject intervenes between the wh-phrases, and compare them to the 206 examples in which the non-wh-subject follows fronted wh-phrases, there is a difference in the possible interpretations: (11) a. Kdaj je komu when AUX who.DAT pametno Miha pomagal? [both readings] wise Miha help ‘When was it wise of Miha to help whom?’ ‘When did Miha help whom in a wise manner? b. (12) Kdaj je pametno komu Miha pomagal? Kdaj je Miha komu when AUX Miha who.DAT [both readings] pametno pomagal? [manner reading] wise help ‘When did Miha help whom in a wise manner? When the subject follows fronted wh-phrases and the adverb in a multiple whquestion, the adverb gets a sentential or a manner reading, (11). This is true regardless of the position of the adverb (it can follow the wh-phrases or appear between them). When the subject intervenes between the fronted wh-phrases, (12), the adverb can receive only the manner reading. We can take this as an indicator that the lower wh-phrase does not appear in the Left Periphery of the clause (in which case both readings would be available) and as another argument for treating multiple wh-questions with short movement separately from multiple wh-questions with multiple wh-fronting to the clause initial position. This means that we need a separate account for short movement. But before proceeding, a side note is in order. I am here concentrating on examples in which the subject intervenes between two wh-phrases, but there are also cases in which the object does so. In (13) below, the accusative non-wh-object intervenes between two wh-phrases. Since it is an indefinite DP and it is not emphasized, it is not the topic or the focus of the sentence (I assume it is scrambled to its position). Such questions also behave exactly as do multiple wh-questions with a non-wh-subject with respect to the reading of the adverb. The adverb pametno ‘wisely’ in (14) can only receive the manner reading. (13) Kdo je who AUX neko knjigo kdaj kupil Tonetu za rojstni dan? some book when buy Tone for birthday ‘When did who buy a book for Tone for his birthday?’ 207 (14) a. Kdo je Mihatu pametno who AUX Miha.DAT wise kaj pojasnil? what explain ‘Who explained what to Miha in a wise manner?’ b. Kdo je Mihatu kaj pametno pojasnil? As mentioned in section 2.2, a similar phenomenon has been observed in Polish. Dornisch (1998) and Citko (2010) name this type of movement to the clause internal position short movement. Crucially, Citko notes that these questions are not echo-questions and that they are parallel to the ‘usual’ multiple wh-questions in Polish in that they allow a pair list reading. The same also holds for Slovenian. An example of short movement in Polish is given below in (15) where the dative whphrase komu ‘who.DAT’ is moved to a position beneath the subject Ewa.116 (15) Polish (Citko 2010: (12)) a. Co Ewa komu dała? what.ACC Ewa.NOM who.DAT gave ‘What did Ewa give to whom?’ b. Komu Ewa co dała? The two questions that need to be answered with respect to short movement are why this movement happens and what the landing site of the short-moved wh-phrase is. In section 3.6 I argue that wh-fronting proceeds to two separate positions in the Left Periphery, the Interrogative and the Wh-Projection, and that movement to both 116 Based on Dyakonova (2009), wh-questions with short movement are possible in Russian. Example (i) shows that short movement is possible in single wh-questions and, crucially, example (ii) shows short movement for multiple wh-questions. In these multiple wh-questions no superiority effects are exhibited. (i) (ii) Russian (Dyakonova 2009: 184, (6b)) Ty na seminare o čem budeš’ rasskazyvat’? you.NOM on seminar.LOC about what FUT.2SG tell.INF ‘What are you going to talk about at the seminar?’ (Dyakonova 2009: 202, (44)) a. Komu Ivan čto podaril? who.DAT Ivan.NOM what.ACC give.PST.MASC ‘What did Ivan give to whom?’ b. Čto Ivan komu podaril? 208 is motivated by the wh-feature on the wh-phrase (which agrees with the features of the heads in the Left Periphery) in addition to the EPP features on the Interrogative and the Wh-head. On the other hand, we have seen that wh-phrases in situ lack whfeatures and do not move, because there is no motivation for them to move. If we assume these two options for wh-phrases, there are also two possibilities for short movement. The first one is that wh-phrases are short wh-moved to some clause internal position, which means that short wh-movement is parallel to wh-movement to the clause initial position (i.e. it is motivated by wh-features). The second possibility is that wh-phrases are bare wh-pronouns which are scrambled to some clause internal position, which would be expected based on the fact that Slovenian exhibits scrambling, as shown in section 3.4.5 and as I will show in what follows. The issue, however, is that the two possibilities are hard to differentiate in multiple wh-fronting languages. 6.1 Short movement in Slovenian: Scrambling vs. wh-movement While the fact that Slovenian has wh-movement was established in chapter 3, more needs to be said about scrambling before proceeding. I will take scrambling to be optional semantically vacuous movement – that is, movement that results in different word orders, but does not change grammatical relations, truth conditions or morphology (as in Ross 1967, Saito 1989, see also Bailyn 2001).117 Crucially, Slovenian, just like other Slavic languages (see for example Bailyn 2001, Bošković 2004, 2009), allows scrambling of one or more constituents, as (16) shows. (16) a. Ivan je Ivan.NOM AUX napisal pismo Maji. wrote letter.ACC Maja.DAT ‘Ivan wrote a letter to Maja.’ b. Ivan je pismo napisal Maji. c. Ivan je Maji napisal pismo. 117 A different approach was proposed by Bošković and Takahashi (1998) who argue for Japanese style scrambling in which the scrambled element is based generated in its surface position and undergoes LF lowering. I will not be considering if Slovenian displays such scrambling, but see Bailyn (2001) and Bošković (2009) who consider Russian and Serbo-Croatian in this respect. 209 d. Maji je Ivan napisal pismo. As already indicated, fn. 100, and as I show below, there is an option to move bare wh-pronouns in yes/no-questions and conditionals. To recapitulate, I assume that the bare wh-pronoun in the examples below does not have a wh-feature and therefore cannot undergo wh-movement. But bare wh-pronouns can still move and since movement of bare wh-pronouns in the following examples does not change the meaning of the yes/no-question or the conditional sentence, I take the bare whpronouns to be scrambled to the clause internal position (I mark the position from which the pronoun was scrambled with an e). (17) A je Ivan kaj napisal e? QY/N AUX Ivan what.INDEF write ‘Did Ivan write anything?’ (18) Če je if AUX zdravnik koga izpostavil e virusu, bo dobil doctor who.ACC expose virus will get kazen. fine ‘If the doctor exposed anyone to a virus, he will get a fine.’ In what follows I will try to determine whether in wh-questions with short movement in Slovenian the wh-phrase is wh-moved or scrambled into a clause internal position. Short movement in Slovenian behaves similarly to short movement in Polish and in both languages this type of movement behaves similarly to ordinary multiple wh-fronting. For example, Citko (2010) shows that in questions with short movement Polish does not exhibit Superiority effects, (19), and the same holds for Slovenian, (20). (19) Polish (Citko 2010: (12)) a. Co Ewa komu dała? what.ACC Ewa.NOM who.DAT gave ‘What did Ewa give to whom?’ b. Komu Ewa co dała? 210 (20) a. Kaj je Črt komu kupil? what.ACC AUX Črt who.DAT buy ‘What did Janez buy for whom?’ b. Komu je Črt kaj kupil? However, the fact that wh-questions with short movement do not exhibit Superiority effects does not differentiate between wh-movement and scrambling since scrambling (by definition) results in free word order (see example (16) above). Second, in Polish, short moved wh-phrases can license parasitic gaps, (21), which means that short movement is an instance of A’-movement (see also Dornisch 1998). Similarly, parasitic gaps are licensed both in a wh-question with multiple whfronting, (22), and in multiple wh-questions with short movement in Slovenian, (23). (21) (Dornisch 1998: 160 (57)) Za co by Piotr kogo wyrzucił for what COND Piotr.NOM who.ACC nie wysłuchawszy throw-out not listen.PERF.PART przedtem? before ‘For what would Peter throw out whom without having listened to?’ (22) Slovenian (Golden 1997b: (2a)) Kogai je Peter povabil k sodelovanju na projektu, preden je srečal ti? who to join on project before AUX Peter invite AUX meet ‘Who did Peter invite to join the project before he met (them)?’ (23) Zakaj je Peter kogai povabil why AUX Peter who invite k sodelovanju na projektu, preden je to join on project before AUX srečal ti? meet ‘Why did Peter invite who to join the project before he met (them)?’ While both short movement and wh-movement to the Left Periphery license Parasitic Gaps and behave the same in this respect, this does not exclude scrambling as the reason for movement of the clause internal wh-phrase. In Slovenian, scrambling can license Parasitic Gaps. Witkoś (2008) shows the same holds for Polish. 211 (24) Majoi Maja.ACC je Peter povabil k sodelovanju na projektu, preden je srečal ei. AUX Peter invite to join on project before AUX meet ‘Peter invited Maja to join the project before he met.’ Dornisch (1998) also shows that short movement in Polish does not exhibit Weak Crossover Effects. Weak Crossover Effects also cannot be found with short movement in Slovenian, as (26) shows. Wh-movement to the clause initial position does not trigger Weak Crossover Effects in Polish or in Slovenian, as (27) and (28) show, respectively.118 (25) Polish (Dornisch 1998: 159 (56)) Kiedy Piotr przed kimi schował jegoi klucze ti? when Piotr from whomi hid hisi keys ‘When did Peter hide his keys from whom?’ (26) Kdaj je when is Peter pred komi skril njegovei ključe? Peter from whom hid his keys ‘When did Peter hide his keys from whom?’ (27) Polish (Szczegielniak 2001: (9e)) Kogoi jegoi przyjaciele podziwiali t? who his friends admire ‘Who do his friends admire?’ 118 In both languages WCO effects can be seen in long distance wh-questions and, as Marušič (2005) shows, Slovenian exhibits WCO in scrambling from finite complements – just as it does with whfronting from embedded clauses. (i) Polish (Szczegielniak 2001: (9f)) chcesz by jegoi przyjaciele * Kogoi ty who you want SUBJ his friends ‘Who do I want his friends to admire?’ (ii) Slovenian * Kogai hočeš da bi njegovii who want.2SG that SUBJ his ‘Who do you want his friends to admire?’ (iii) Slovenian (Marušič 2005: (22a)) Janezi je njegovj/*i oče reku, Janez.NOM AUX his father said ‘Johni, hisj/*i father said doesn’t play goalie.’ podziwiali ti? admire prijatelji friends občudovali? admire da __ ne that not igra play 212 golmana. goalie (28) Kogai who njegovii prijatelji občudujejo t? his friends admire ‘Who do his friends admire?’ In Slovenian, Weak Crossover Effects are also not triggered in cases with scrambling in matrix questions, as (29) shows. (29) Včeraj yesterday je Peter pred Tonetomi skril njegovei ključe. AUX Peter from Tone hid his keys ‘Yesterday Peter hid Tone’s keys from him.’ Again, questions with multiple wh-fronting and short movement and cases with scrambling behave similarly with respect to Weak Crossover Effects. Additionally, Dornisch shows that short movement does not fix scope. This means that the dative wh-phrase can take scope over the accusative wh-phrase in (30). According to Dornisch, this is shown by the fact that one can answer a question like (30) with a pair list answer in which the dative wh-phrase takes scope over the accusative wh-phrase, as in (31), in which the dative precedes the accusative. Again, the same holds for Slovenian as both questions in (32) can be answered with both (33a) and (33b). (30) Polish (Dornisch 1998: 148, (42a)) Co by Anna komu poleciła? what.ACC COND Anna who.DAT recommend ‘What did Anna recommend to whom?’ (31) (Dornisch 1998: 161, (60)) Ucznim podstawóki – students.DAT primary school.GEN matematyki – Tolstoja, Chandlera, profesorom Chandler.ACC professors.DAT chemikom – Plath. mathematics.GEN Tolstoy.ACC research-chemists.DAT Plath.ACC ‘To primary school students she recommended Chandler, to the mathematics professors Tolstoy, to the chemists – Plath.’ 213 (32) a. Kaj je what.ACC AUX Ana komu priporočila? Anna who.DAT recommend ‘What did Anna recommend to whom?’ b. (33) a. Kaj je komu Ana priporočila? Roku pravljice in ROK.DAT stories and me.DAT meni Hobita. Hobit. ‘(She recommended) stories to Rok and the Hobbit to me.’ b. Pravljice Roku in Hobita meni. Scope is also not fixed with scrambling in matrix clauses:119 (34) En fant je one boy AUX objel vsako punco. hug every girl ∀>∃/∃>∀ ‘A boy hugged every girl.’ (35) Vsako every punco je en girl one boy hug AUX fant objel. ∀>∃/∃>∀ ‘A boy hugged every girl.’ This means that with respect to Parasitic Gaps, Weak Crossover Effects, and Scope wh-fronting and scrambling behave similarly and we cannot determine if wh-phrases are wh-moved or scrambled to a clause internal position based on the data so far. But we can potentially use cases with long distance wh-movement and scrambling to differentiate between the two. That is, in Slovenian both long distance wh-movement and long distance scrambling are possible: (36) Kogai who.ACC je Janez rekel, da AUX Janez say Maja objema ti ? that Maja hugs ‘Who did Janez say that Maja hugs?’ (37) Anžetai Anže.ACC je Janez rekel, da AUX Janez say Maja objema ei. that Maja hugs ‘Janez said that Maja hugs Anže.’ 119 In cases with long distance scrambling out of embedded finite clauses scope is reconstructed to the base generated position, see Marušič (2005). 214 Wh-movement and scrambling, on the other hand, behave differently with respect to Relativized Minimality Islands, i.e. wh-islands.120 With scrambling, Relativized Minimality Effects are absent, as Bošković (2004, 2009) notes. This means that in Russian, but also Serbo-Croatian, a wh-phrase cannot move from a whisland, (38a), but a non-wh-phrase can be scrambled out of a wh-island, (38b): (38) Russian (Bošković 2004: (10a), (11a)) a. * Ktoi ty videl kogda ti who you saw when b. Ty doktori you doctor pod’ezžal? came videl kogda ei pod’ezžal? saw when came ‘Did you see when the doctor came?’ As we saw in sections 3.1.1 and 3.1.2, Slovenian does not allow wh-movement out of wh-islands (cf. Golden 1997a). The data is shown again below. Still, just as in Russian, scrambling out of wh-islands is acceptable: (39) a. ?? Kdo si who b. AUX videl, kdaj je prišel? saw when AUX came Zdravnik si videl, kdaj je prišel? doctor AUX saw when AUX came ‘Did you see when the doctor came?’ Based on this distinction we can make a prediction. If scrambling is responsible for short wh-movement, moving a wh-phrase out of a wh-island to a clause internal position, should be possible, but if short movement is the result of wh-movement, then such movement should not be possible. Crucially, speakers judge the example 120 (i) As already shown in section 3.4.5, Relativized Minimality is defined as: Relativized Minimality in the configuration: …X…Z…Y… X and Y cannot be connected by movement (or other local relations) if Z intervenes between them, and Z is of the same structural type as X. 215 below as ungrammatical (in the examples below, the matrix clause has a wh-phrase in a clause initial position, but this wh-phrase originates in the matrix clause; the nonwh-subject is followed by a wh-phrase that originates in the embedded wh-question). (40) *Kdaj when je Janez komui vprašal, za kaj Metka daje ti AUX Janez who.DAT ask for what Metka gives zdravilo? medicine ‘When did Janez ask to whom Metka gave the medicine for what?’ This can be contrasted with the example below in which komui ‘who.DAT’ is replaced with a non wh-phrase, which speakers find considerably more acceptable.121 (41) Kdaj when je Janez Rokui vprašal, za kaj Metka daje ei AUX Janez Rok.DAT ask for what Metka gives zdravilo? medicine ‘When did Janez ask for what Metka gives the medicine to Rok?’ While we have seen that we cannot distinguish between scrambling and whfronting in matrix questions, we can make a distinction between the two types of movements using extraction out of wh-islands, which indicates that short movement is in fact wh-movement. An additional argument for the fact that the movement mechanism responsible for placing wh-phrases in a clause internal position is whmovement comes from the behavior of topic and focus phrases in Slovenian. While 121 Additionally, we can test if movement of a wh-pronoun is acceptable when the wh-island is embedded in a yes/no-question. Speakers find these cases acceptable, but they note that in these cases the moved wh-phrase receive a contrastive reading and is emphasized: (i) A je Janez KOMU vprašal, kdaj je Metka dala zdravilo? QY/N AUX Janez who ask when AUX Metka give medicine ‘Did Janez ask, when Metka gave the medicine to WHO?’ Since we have seen that echo-wh-phrases in Slovenian do not have to move to the clause initial position, we can assume that komu ‘who’ does not undergo wh-fronting in (i). 216 we saw in section 3.4.4 that these phrases can appear in the Left Periphery, we can also find them following the subject in Slovenian. Example (42) shows this for a focus phrase and example (43) for the topic phrase. Example (44) shows that, just as in the Left Periphery, the word order of a wh-phrase and a focus phrase is free in the clause internal position (the Low Periphery, see the following section), and example (45) shows the same for a topic phrase and a wh-phrase. (42) Miha je Maji TONETA predstavil. Miha AUX Maja.DAT TONE.ACC introduce ‘Miha introduced TONE to Maja.’ (43) Maja je izpostavila Toneta virusu ebole ... ‘Maja exposed Tone to the Ebola virus...’ Janez pa Janez.NOM PA je AUX tega fanta sprejel v bolnico. this boy in hospital accepted ‘and Janez accepted this boy to the hospital.’ (44) a. Kdaj je Miha komu when AUX Miha.NOM who.DAT TONETA predstavil? TONE.ACC introduce ‘When did Miha introduce TONE to whom?’ b. Kdaj je Miha TONETA komu predstavil? (45) Maja je izpostavila Toneta virusu ebole. ‘Maja exposed Tone to the Ebola virus.’ a. Kdaj je Janez koga when AUX Janez.NOM who.ACC tega virusa ozdravil? this virus.GEN cure ‘When did Janez cure whom of the virus?’ b. Kdaj je Janez tega virusa koga ozdravil? Since wh-phrases behave the same with respect to topic and focus phrases when they follow the subject as they do with respect to topic and focus phrases in the Left Periphery, I will treat wh-movement to the Left Periphery and short movement in the same manner. That is, I will assume that wh-phrases are wh-moved to the Low Periphery. Note, however, that we cannot completely exclude scrambling as the reason why wh-phrases are moved to the clause internal position, since bare wh-pronouns 217 are always interpreted as interrogative wh-phrases when in the scope of the complex Q+wh-feature in the Interrogative Projection. Also, only one test shows different behavior in scrambling versus wh-fronting contexts in Slovenian. While this test has indicated that wh-phrases are wh-moved to the clause internal position, this has only been shown for movement out of embedded clauses. Unfortunately, at this point I do not know any other way of differentiating between wh-movement and scrambling, but I will treat short movement as wh-movement based on the parallels between movement to the Left Periphery and to the Low Periphery. As the next section will show, a similar proposal has also been made for Polish. 6.2 Short movement as wh-movement to the Low Periphery We saw in the previous section that Slovenian and Polish behave similarly with respect to short movement. Because of this I will take the proposal for short movement that was suggested for Polish in Citko (2010) as a starting point for my proposal for Slovenian. Citko (2010) notices that in Polish, the lower wh-phrase moves above negation and the adverb często ‘often’, (46) and (47), respectively. The situation is comparable to Slovenian. Assuming negation marks the edge of vP, as Citko (2010) does, this indicates that the lower wh-phrase ends up above vP, as shown in (48). Examples such as (48c) are discussed in chapter 5. (46) Polish (Citko 2010: (18)) a. Czego Ewa what.GEN Eve.NOM komu nie dała? who.DAT not gave ‘What didn’t Ewa give to whom?’ b. * Czego Ewa nie komu dała? (47) Polish (Citko 2010: (19)) a. Co Ewa what.ACC Eve.NOM komu często pokazuje? who.DAT often shows ‘What does Ewa often show to whom?’ b. ?? Co Ewa często komu pokazuje? 218 (48) a. Česa Maja komu what.GEN Maja.NOM who.DAT ni dala? not gave ‘What didn’t Maja give to whom?’ b. ?* Česa Maja ni komu dala? c. (49) a. Česa Maja ni dala komu? Kaj Maja komu pogosto kuha? what Maja who.DAT often cooks ‘What does Maja often cook for whom?’ b. Komu Maja kaj pogosto kuha? For Polish, both Dornisch (1998) and Citko (2010) propose that the lower whphrase undergoes movement to some clause internal position. Dornisch (1998) proposes that the clause initial wh-phrase moves to SpecCP and that this movement is driven by the strong Q-feature on C. She assumes that this Q-feature is an operator feature and that it can be checked by the wh-phrase. On the other hand, the lower whphrase moves to a preverbal position – the Transitivity Phrase (TrP). The Transitivity Phrase hosts a wh-feature which the wh-phrase checks. Under this approach, the whfeature on the wh-phrase is an interpretable feature (it needs to be visible at LF) and because of this it is not deleted after it is checked in the TrP, hence the wh-phrase is able to undergo further movement. In addition, she assumes that the wh-feature in Polish is a multiply checked feature – when a wh-phrase checks the wh-feature on TrP, it does not erase it. Because of this the wh-feature will be available for further operations. Dornisch therefore suggests that in a multiple wh-question with short movement the clause initial wh-phrase moves to SpecCP and the wh-phrases that undergo short movement move only to TrP and not any further. The structure, taken from Dornisch (1998: 147, (42b)), of the question (30) is shown below122: 122 According to Dornisch (1998) the auxiliary is generated in the higher VP. I follow her and assume the auxiliary is generated in VP and that it later moves out of VP. 219 (50) CP 2 C’ 2 Coi C g [Q] g by TP 2 Anna T’ 2 T TrP 2 ti Tr’ 2 tsu Tr’ 2 Tr VP g 2 [wh] V TrP 2 komuj Tr’ 2 tsu Tr’ 2 Tr VP g 2 [wh] ti V’ 2 V tj poleciła Dornisch proposes that the two Tr-heads carry a strong wh-feature and C has a strong Q-feature. Note that according to Dornisch (1998) each TrP is projected above a VP and that both the clitic, such as by (the conditional auxiliary), and the lexical verb, such as poleciła recommend, projects a VP. The lower Tr-head attracts the closest wh-phrase – co ‘what’. Dornisch assumes that the Polish wh-feature on Tr is a multiply checked feature – even if the wh-phrase checks the wh-feature in Tr it does not erase it. The wh-feature of co ‘what’ is an interpretable feature that must be visible at LF – it does not get deleted after being checked and co ‘what’ can again be available for movement. This is again repeated when the higher Tr head is merged (co again being the closest wh-phrase). When C is merged, co moves to check the strong Q-feature on C. Komu ‘who.DAT’ also has a strong feature that needs to be checked and this happens at the closest possible projection – the lower TrP in which the wh-feature was not erased after being checked. 220 Citko (2010) points out that the proposal by Dornisch (1998) has several problems123 and proposes a different derivation. She adopts the Minimalist approach as proposed by Chomsky (2001), i.e. she assumes that derivation proceeds in phases, and assumes a clause internal Focus Projection (FP) to which the lower wh-phrase moves. For Citko (2010) the derivation of (51) then proceeds as follows: Both whphrases first move to the phase edge position vP. If there is an FP projection in the clause, one of the two wh-phrases moves to this position. This is an instance of operator movement (cf. Dornisch 1998) so this wh-phrase is not available for further movement. The wh-phrase that is still located in SpecvP can then be moved to SpecCP. Crucially, the word order of the moved wh-phrases is free since a non-phase head can attract a lower wh-phrase because locality is only evaluated at phase heads (hence, komu ‘who.DAT’ can move to the clause initial position, as (53) shows). (51) Polish (Citko 2010: (12)) a. Co Ewa komu dała? what.ACC Ewa.NOM who.DAT gave ‘What did Ewa give to whom?’ b. (52) Komu Ewa co dała? [CP Coi [C’ [C] [TP Ewak [T’ [T] [FP komui [F’ [F] [vP ti [V’ tj [V’ tk [V’ [V] [vP dala ti tj (53) [CP Komui [C’ [C] [TP Ewak [T’ [T] [FP coj [F’ [F] [vP ti [V’ tj [V’ tk [V’ [V] [vP dala ti tj The idea is then that short movement proceeds to an IP internal position which is adjacent to the vP – according to Citko (2010) this is the clause internal Focus Projection. That wh-fronting can proceed to some IP-internal position had already been proposed in Jayaseelan (2001), who also showed that in Malayalam wh- 123 According to Citko (2010), the proposal by Dornisch (1998) would allow for the direct object to be reconstructed above the indirect object. If this was indeed the case, we would expect example (i) to be possible, but (i) is not grammatical. For further issues see Citko (2010): (i) * Ile zdjęć Janai Ewa how many pictures.GEN Jan.GEN Eve.NOM ‘How many pictures of Jan has Ewa shown to him?’ 221 mui pokazała? he.DAT show movement proceeds to a Focus Projection (FP) which is located directly above vP (based on the observation that a question word needs to directly precede the verb). In addition to the Focus Projection, Jayaseelan proposes that Topic Projections are also possible above the vP in Malayalam. Essentially, what we then get is a clause internal Left Periphery. A similar Clause Internal Periphery (referred to also as the low IP area/Low Periphery of the vP phase) is also proposed in Belletti (2004, but see also Belletti and Shlonsky 1995), who again argues for a Focus Projection that can be surrounded by clause internal Topic Projections. Together they are adjacent to the VP, e.g. are located in the low IP area.124 Poletto (2006) further showed that an array of elements, i.e. all arguments, adverbials and verbal modifiers (but not clitic or weak elements as these cannot be focalized) can be hosted in the low IP area/Low Periphery of the vP phase. This lack of specialization is, according to Poletto, typical of positions of the Left Periphery. Under this approach, a Left Periphery of each phase is set up, that is “by merging a ‘Topic-Focus’ field before the highest projection ‘closing up’ the phase (corresponding to Force in the CP system proposed by Rizzi (1997); see again Belletti (2004) for modern Italian among others” (Poletto 2006: 5).125 Based on the observations by Citko (2010) and previous proposals regarding the existence of the Low Periphery, I propose that a short moved wh-phase is located 124 Assuming the Clause Internal Periphery and the Clause Initial Periphery appear at the edges of different phases, this would mean the clause initial wh-phrase and clause internal wh-phrase are in different phrases. Based on this we can predict that we will not find Distinctness effects in cases with short movement. We can take example (i) and compare it to example in (ii), which is an example of short movement in a multiple wh-question. We can take grammaticality of (ii) as evidence that the two wh-phrases are in fact in two different phases. Also acceptable is (iii) in which the wh-phrase is located in situ (the wh-phrases are again in different phases). (i) ? Katero punco katero snov uči Božo? [which girl].ACC.F [which subject].ACC.F teaches Božo (ii) Katero punco Božo katero snov uči? [which girl].ACC.F Božo [which subject].ACC.F teaches (iii) Katero punco uči Božo katero snov? [which girl].ACC.F teaches Božo [which subject].ACC.F ‘Which girl is Božo teaching which subject?’ 125 The prediction is then that each phase will have its own left periphery. Based on Slovenia, this prediction is borne out since we can also find wh-items and a Focus Projection in Slovenian DPs, as (i) shows. For more on wh- and focus-fronting in DPs in Slovenian see Mišmaš (2014). (i) Kakšen NEMŠKI what-kind-of German Majin avto je popraskan? Maja’s car is scratched 222 (Rdeč.) (Red) in the Low Periphery of the vP phase. As already mentioned, this Low Periphery is (in addition to wh-phrases) able to host Focus Phrases, as example (54) below shows. An internal Focused Phrase can also be found in a wh-question with short whmovement, (55). In these cases the focused phrase can either precede of follow the clause internal wh-phrase. Because the wh-phrase can co-occur with a focused phrase in the Low Periphery, I propose they occupy different projections – WhP and FocP, respectively. Also possible in the Low Periphery is the clause internal topic, see (43) in the previous section. (54) Miha je Maji TONETA predstavil. Miha AUX Maja.DAT TONE.ACC introduce ‘Miha introduced TONE to Maja.’ (55) a. Kdaj je Miha komu when AUX Miha.NOM who.DAT TONETA predstavil? TONE.ACC introduce ‘When did Miha introduce TONE to whom?’ b. Kdaj je Miha TONETA komu predstavil? Just as in the Left Periphery, there is no strict word order between topic and focus phrases, (56). Example (57) shows that there is no strict word order between the topic, focus and wh-phrases in the Low Periphery: (56) Maja je izpostavila Toneta virusu ebole. ‘Maja exposed Tone to the Ebola virus.’ a. Kdaj je Janez BOŽOTA temu virusu izpostavil? when AUX Janez.NOM Božo.ACC this virus.DAT expose ‘When did Janez expose BOŽO to this virus?’ b. Kdaj je Janez temu virusu BOŽOTA izpostavil? (57) Maja je izpostavila Toneta virusu ebole. ‘Maja exposed Tone to the Ebola virus.’ a. Zakaj je Janez why Janez.NOM Božo.ACC this virus.DAT when expose AUX BOŽOTA temu virusu ‘When did Janez expose BOŽO to this virus?’ b. Zakaj je Janez temu virusu BOŽOTA kdaj izpostavil? 223 kdaj izpostavil? c. Zakaj je Janez kdaj temu virusu BOŽOTA izpostavil? Also, just as in the Left Periphery, more than one wh-phrase can be located in the Low Periphery (I only give three possible word orders below): (58) a. Kaj je what.ACC AUX Miha kdaj komu Miha.NOM when who.DAT dal? give ‘What did Miha give when to whom?’ b. Kaj je Miha komu kdaj dal? c. Komu je Miha kdaj kaj dal? Based on these examples I argue for the existence of a Low Periphery in Slovenian in which clause internal topic, focus and wh-phrases are located. Assuming such a vP periphery explains the position of the short moved wh-phrases (as more than one wh-phrase can appear clause internally). However, just as in questions with wh-in-situ, again at least one wh-phrase has to appear clause initially for the entire sentence to be interpreted as a true multiple wh-question. I return to the interpretation of wh-questions with short movement in section 7.1.2. 6.3 Clause initial wh-phrase and short movement The one limitation that applies in cases with short movement is that at least one whphrase has to appear in the clause initial position in order for the question to be interpreted as a true multiple wh-question and not, for example, an echo or a yes/noquestion. For instance, the question in (59e) would be interpreted as an echo question if the wh-phrases were pronounced with an emphasis (this is true regardless of the word order of wh-phrases) or a yes/no-question if the question was pronounced with a rising intonation, which is generally typical of Slovenian yes/no-questions. Examples (59a–d), on the other hand, receive a true question reading. In addition, (59a–d) show that the Wh-Projection in the Low Periphery can be repeated (that is, more than one wh-phrase can be located in the Low Periphery), just as it can be repeated in the Left Periphery of the clause. 224 (59) a. Kaj je Maja kdaj komu povedala? what.ACC AUX Maja when who.DAT tell ‘When did Maja tell what to whom?’ b. Kaj je Maja komu kdaj povedala? c. Komu je Maja kaj kdaj/kdaj kaj povedala? d. Kdaj je Maja kaj komu/komu kaj povedala? e. # Maja je kdaj kaj komu povedala? The obligatoriness of the clause initial wh-phrase can also be observed in questions in which all wh-phrases move to the Left Periphery, section 3.5, and in multiple wh-questions with a wh-phrase in situ, chapter 5. To recapitulate, following Soare (2007), I proposed in section 3.6 that in Slovenian the Interrogative Projection hosts an interpretable complex Q+wh-feature (which comes with an EPP subfeature). The Q+wh-feature with an EPP-subfeature are checked by a wh-phrase, which carries an uninterpretable valued wh-feature and that is moved to SpecInterP. This complex feature is also responsible for Clause Typing, defined as in Cheng (1991). In addition, I assumed that in questions with multiple wh-fronting, non-initial whphrases are moved to Wh-Projections in the Left Periphery where they check the interpretable wh-features with EPP-subfeatures. This proposal essentially also holds for questions with short movement, but with a slight modification: Wh-Projections are located in the Left Periphery (sometimes referred to as the High Periphery, cf. Poletto 2006) and in the Low Periphery. The Interrogative Projection, on the other hand, is only ever located in the Left Periphery. Assuming that the Interrogative Projection is always located in the Left Periphery of wh-questions, see section 3.5, accounts for the clause initial wh-phrase in wh-questions with short movement. Because the Interrogative Head is projected with its features, one wh-phrase has to move to check them. This also means that Clause Typing can only happen in the Left Periphery. In addition, the remaining whphrases then move to Wh-Projections – either in the High or the Low Periphery. I discuss what this means in relation to optionality in section 7.2.2. Assuming this, short movement (i.e. wh-movement to the Low Periphery) is essentially parallel to wh-movement to the Left Periphery – in both a wh-phrase 225 moves to a Wh-Projection to check an interpretable wh-feature with an EPP subfeature. This implies that movement of a wh-phrase with an uninterpretable whfeature is in fact obligatory and it is only the wh-pronouns without wh-features that stay in situ (however, as was observed in this section, scrambling of bare whpronouns cannot be excluded). I return to this and to the interpretation of whquestions in the next chapter. 226 7 How optional is multiple wh-movement in Slovenian I started this thesis with the observation that multiple wh-fronting in Slovenian is optional. This claim was made based on examples such (1) in chapter 1, repeated below, that show that in Slovenian multiple wh-questions only one wh-phrase needs to appear in the clause initial position, while the others can move to the beginning of the clause, some clause internal position, or stay in situ. (1) a. Koga je čemu znanstvenik izpostavil? who.ACC AUX what.DAT scientist expose ‘Who did the scientist expose to what?’ b. Koga je znanstvenik čemu izpostavil? c. Koga je znanstvenik izpostavil čemu? d. Čemu je koga znanstvenik izpostavil? e. Čemu je znanstvenik koga izpostavil? f. Čemu je znanstvenik izpostavil koga? In this chapter, I discuss the interpretation of multiple wh-questions with multiple wh-fronting to the clause initial position, multiple wh-questions with short movement and questions with wh-in-situ and show that they can be used in the same contexts. In addition, I also show that questions with no wh-fronting never receive the same interpretations as questions with wh-fronting. In the second part I revisit the issue of optionality and show that wh-movement in Slovenian is in fact not optional – interrogative wh-phrases in Slovenian have to move but they can move to different projections. Non-interrogative bare wh-pronouns, on the other hand, stay in situ, because their movement is not motivated. 7.1 Interpretation of multiple wh-questions in Slovenian In section 2.3, I claim that all multiple wh-questions in Slovenian receive the same interpretation. In this section, I will show that this is in fact the case and that 227 questions with multiple wh-fronting, multiple wh-questions with short movement and questions with wh-in-situ can all be used in the same contexts. 7.1.1 The interpretation of multiple wh-questions with wh-in-situ Šimík (2010) discusses the behavior of multiple wh-questions in Czech. I will use Šimík’s description and compare it to Slovenian data to show that multiple whquestions with a wh-phrase in situ get the same interpretation as multiple whquestions with multiple wh-fronting. Czech uses two strategies to form multiple wh-questions: multiple and single wh-fronting, (2). Questions such as (2) are comparable to multiple wh-questions in Slovenian (see for example (1) in the beginning of the chapter 5). In Czech, pitch accent is placed on the final constituent, regardless of whether this constituent is a wh-phrase or not. But while both strategies of making wh-questions are available in Czech, Šimík (2010) shows that multiple wh-questions with wh-in-situ appear only in certain contexts in Czech. In Slovenian, on the other hand, multiple wh-questions with wh-in-situ receive the same meanings as questions with multiple wh-fronting. (2) Czech (Šimík 2010 (7)) a. Kdo co KOUPIL? who what bought ‘Who bought what?’ b. Kdo koupil CO? According to Šimík (2010), in Czech the form of multiple wh-questions depends on the context. If the context entails givenness (that is, whether something is given in the context), a wh-phrase can stay in situ. For example, when a predicate is discourse given, (3), a wh-phrase will appear in situ (and the given predicate will appear in the de-accented position). If the predicate is not given in the discourse, the felicitous question is the one with multiple fronting, (4). In Czech, the examples in (3) and (4) get a pair list reading. 228 (3) Czech (Šimík 2010: (13)) Proč se neposadite? why not.sit.down.2PL REFL – Neřekli nam... not.told.3PL us ‘Why don’t you sit down? – They didn’t tell us...’ a. # kdo se ma kam who REFL have.3SG POSADIT. where sit.down ‘who is supposed to sit down where.’ b. (4) kdo se ma posadit KAM. Czech (Šimík 2010: (14)) Proč tady tak stojite? – Neřekli nam... why here so stand.2PL not.told us ‘Why are you standing here like that?’ – ‘They didn’t tell us...’ a. kdo se ma kam POSADIT. who have.3SG where sit.down REFL ‘…who is supposed to sit down where.’ b. # kdo se ma posadit KAM. In Slovenian a wh-phrase can either move or stay in situ regardless of the givenness of the predicate, as examples (5) and (6) below show. Note however, that while all the questions below are grammatical, they differ in the information the speaker is most interested in (cf. section 3.4.5). Because of this some might seem more appropriate that others, but this varies from context to context. This is not only the case for the following two examples, but holds in general. (5) Zakaj se ne why REFL NOT usedete? – sit.down.2PL Niso nam rekli ... NEG.AUX3PL us ‘Why don’t you sit down? – They didn't tell us...’ a. kam naj se kdo usede. where should SELF who sit ‘where who should sit.’ b. kam naj se usede kdo. c. kdo naj se usede kam. d. kdo naj se kam usede. 229 say (6) Zakaj tam stojite? – Niso nam rekli ... why there stand.2PL NEG.AUX3PL us say ‘Why are you standing there?’ – ‘They didn’t tell us...’ a. kam naj se kdo usede. where should SELF who sit ‘where who should sit.’ b. kam naj se usede kdo. c. kdo naj se kdo usede kam. d. kdo naj se kam usede. In Czech, a wh-phrase that refers to the topic (where topic is defined as given, as in Schwarzschild 1999) has to move. For example, the boldfaced co ‘what’ is given in the context below: (7) Czech (Šimík 2010: (11)) They brought the things, but I don’t know … a. co jsme měli předat what AUX have give KOMU. whom ‘what we were supposed to given to whom.’ b. # komu jsme měli předat CO. As (8) shows, a wh-phrase can move or stay in situ when referring to a constituent that is given in the context in Slovenian. This means that both word orders are possible in Slovenian, even if kaj ‘what’ refers to something salient in the context. Also, multiple wh-questions with both word orders can receive pair list answers. This is also possible when both wh-phrases are fronted. (8) They brought the things, but I don’t know … a. kaj naj what.ACC should bi dali komu. AUX give whom.DAT ‘what we were supposed to given to whom.’ b. komu naj bi dali kaj. c. komu naj bi kaj dali. 230 d. kaj naj bi komu dali. Šimík (2010) also observes contexts in which both wh-phrases in the multiple wh-question need to move in Czech and which get single pair readings (as opposed to the questions with multiple wh-fronting and pair list answers above, as in example (4)): Šimík first observes ‘default’ single pair readings that only happen in a context in which there are no requirements on the context. These questions are understood as conversation starters with an all new information structure (both wh-phrases in these cases refer to constituents that were not previously given in the context).126 In these both wh-phrases need to move to the clause initial position in Czech, which is not the case in Slovenian. (9) Czech (Šimík 2010: (15)) [Context: I meet a friend about who I know that he recently decided to be polite, do good deeds, and help people as much as possible. I can start the conversation by asking.] a. Tak co, so komu what whom jsi dnes s čim POMOHL? aux.2SG today with what help ‘So, who did you help with what today?’ b. # Tak co, komu jsi dnes pomohl s ČIM? For the same context in Slovenian a completely felicitous question can be formed both with multiple fronting or single fronting and in both the word order of wh-phrases is free. (10) [Context: I meet a friend about who I know that he recently decided to be polite, do good deeds, and help people as much as possible. I can start the conversation by asking.] 126 Zavitnevich-Beaulac claims that questions with multiple interrogation (i.e. multiple wh-questions) cannot be requests for new information in Russian, but can only be interpreted as clarifying questions, that is questions that only require additional identification. Given the examples from Šimík (2010) and from Slovenian in which multiple wh-questions are conversation starters, this is not always the case. 231 a. Torej, komu si s so AUX.2SG with what today help whom čim danes pomagal? ‘So, who did you help with what today?’ b. Torej, komu si danes pomagal s čim? c. Torej, s čim si komu danes pomagal? d. Torej, s čim si danes pomagal komu? While a single pair response is completely felicitous to all the questions in (10), it does not seem to be the case that in the given context the answer necessarily has to be a single pair answer. Pair list answers such as I baked a cake for my dad, I cleaned the car for my grandma, and I opened the door for a complete stranger are completely acceptable. Returning to Czech, according to Šimík (2010) reciprocal questions in Czech receive a single pair answer.127 In these questions we are “asking for the clarification of which of two contextually salient people was active in some salient event” (Šimík 2010: 498–499). In reciprocal multiple wh-questions a wh-phrase needs to stay in situ in Czech: (11) Czech (Šimík 2010: (17)) I heard that Karel and Marie broke up, but I don't know a. kdo se rozešel S who broke.up with whom REFL KYM. ‘who broke up with whom.’ b. # kdo se s kym ROZEŠEL. In the same contexts in Slovenian one can ask for clarification with a multiple wh-question in which one or both wh-phrases undergo wh-fronting: 127 In Pires and Taylor (2007) these questions are taken to be reference questions and are wh-in-situ questions in English and Brazilian Portuguese (which are typically wh-fronting languages). Also, Wachowitz (1974: fn. 3) observes that in these questions a wh-phrase can also stay in situ in Polish. However, as Pires and Taylor note, some researchers (i.e. Ginzberg and Sag 2001) take them as an instance of echo-questions. 232 (12) I heard that Karel and Marie broke up, but I don’t know… a. kdo je koga pustil. who AUX who.ACC leave ‘who left whom.’ b. koga je kdo pustil. c. kdo je pustil koga. d. koga je pustil kdo. Turning away from the comparison with Czech for a moment, Polish has single wh-fronting at least in clarification questions, as in the one in (13). Clarification questions are related to echo questions and are used when an interlocutor fails to assign a referent to a pronoun (Šimík 2010: fn. 11). According to Wachowicz (1974), the example cited in (13) can be used in a situation in which there were various tasks and several people were assigned to them and while proposals have been made on how to partner people, there is no fixed plan. Because the speaker is confused and wants a fixed plan, he asks the question below. (13) Polish (Wachowicz 1974: (15)) W końcu kto robi co? finally who does what ‘Finally, who’s doing what?’ A multiple wh-question, such as the one in (14), can be used as a clarifying question for the same context described above for Polish. As the example shows, both wh-phrases can be fronted or a wh-phrase can stay in situ in a clarifying question in Slovenian. (14) a. No, kdo bo kaj naredil? well who will what do ‘Well, who will do what?’ b. No, kaj bo kdo naredil? c. No, kdo bo naredil kaj? 233 In Czech, on the other hand, both wh-phrases need to be fronted in clarifying questions (Šimík 2010: fn.11). This means that wh-questions with a wh-phrase in situ have a wider distribution in Slovenian, compared to Czech (for an account of the interpretation of wh-in-situ in Czech, see Šimík 2010). In addition, the examples in this section show that multiple wh-questions with multiple wh-fronting and wh-in-situ can get the same interpretations in Slovenian. 7.1.2 Interpretation of wh-questions with short movement It will be shown in this section that multiple wh-questions with short movement can receive the same interpretations and appear in the same contexts as multiple whquestions in which all wh-phrases move to the clause initial positions or multiple whquestions with a wh-phrase in situ. This will be shown on the basis of the previous section. I use examples with a non-wh-subject and assume that the wh-phrase below the subject is short moved to this position. Starting with givenness, in Slovenian multiple wh-questions a wh-phrase can undergo short movement in cases with a given predicate or in cases in which a predicate is not given (that is, the same context in which one also finds multiple whquestions with multiple wh-fronting and wh-in-situ, see previous section). Given the context, the following examples get pair list answers. (15) Zakaj ne why NOT a. kaj kuhate? Niso nam povedali … cook NEG.AUX3PL us naj say posamezen kuhar what.ACC should komu individual cook.NOM who.DAT skuha. cook ‘what each individual cook should cook for whom.’ b. komu naj posamezen kuhar kaj skuha. (16) Zakaj gostje ne jedo? Natakarju niso povedali … why gustes NOT eat.3PL waiter tell a. kuhar kaj naj komu NEG.AUX3PL skuha. what AUX cook.NOM who.DAT cook 234 ‘what the cook should cook for whom.’ b. komu naj kuhar kaj skuha. Multiple wh-questions with short movement are behaving the same as multiple wh-questions with wh-in-situ in cases in which one wh-phrase refers to a topic (where topic is defined as given, as in Schwarzschild 1999). The topic is in boldface below. In this context the questions in (17) receive a pair list answer. (17) They brought the things, but I don’t know … a. kaj naj what.ACC should bi Tone komu dal. AUX Tone.NOM whom.DAT give ‘what Tone is supposed to have given to whom.’ b. komu naj bi Tone kaj dal. Multiple wh-questions with short movement can also be used in questions that are used as conversation starters with an all new information structure (both whphrases in these cases refer to constituents that were previously not given in the context). A response to these can be either a single pair or a pair list answer, again the same as with wh-in-situ above. (18) [Context: I talk to a friend about a friend about whom I know that he recently decided to be polite, do good deeds, and help people as much as possible. I can start the conversation by asking.] a. Torej, komu si danes s so AUX.2SG today with what help whom čim pomagal? ‘So, who did you help with what today?’ b. Torej, s čim si danes komu pomagal? Single pair answers are also available with multiple wh-questions with short movement. A reciprocal question, such as (19), receives a single pair answer, just like a multiple wh-question with a wh-in-situ. 235 (19) I heard that Karel introduced Tina and Miha but I don’t know… a. koga je Karel komu predstavil. who.ACC AUX Karel who.DAT introduce ‘who Karel introduced to whom?’ b. komu je Karel koga predstavil. Multiple wh-questions with short movement can also be used in clarifying questions. This means that multiple wh-questions with short movement behave like the multiple wh-questions with wh-in-situ in the previous section. (20) I heard that Tina bought a bunch of food for a lot of people, but I don’t know… a. kaj je Tina komu dala. what.ACC AUX Tina who.DAT give ‘what Tina gave to whom?’ b. komu je Tina kaj dala. Based on this we can conclude that multiple wh-questions with short movement receive the same interpretations and are used in the same contexts as multiple wh-questions in which all wh-phrases move to the Left Periphery and multiple wh-questions with a wh-phrase in situ. 7.1.3 Excluding cases with all wh-phrases in situ Questions in which all wh-phrases stay in situ behave differently from questions in 7.1.1 and 7.1.2. Specifically, questions in which all wh-phrases stay in situ are ungrammatical under a true question reading and are only acceptable as echo questions or yes/no-questions in which wh-phrases are interpreted as bare indefinite pronouns. The echo-interpretation is available if wh-phrases are pronounced with an emphasis, as (21) shows. The question in (21) can be interpreted either as a surprise 236 or a request-for-repetition question. Also, such questions cannot be used as conversation starters, (22).128 (21) Zdravnik doctor je AUX izpostavil KOGA ČEMU? expose who.ACC what.DAT ‘The doctor exposed WHO to WHAT?’ #‘Who did the doctor expose to what?’ (22) [Context: I talk to a friend about a friend about who I know that he recently decided to be polite, do good deeds, and help people as much as possible. I hear people are talking about it, but I can’t hear what they are asking.] # Torej, so Miha je danes pomagal KOMU S ČIM? Miha AUX.3SG today help whom with what ‘So, who did you help with what today?’ The unavailability of true wh-question reading in such cases is expected. We saw in section 3.5 that movement of one wh-phrase to the clause initial position is crucial for Clause Typing in a true wh-question. Since there is no wh-movement to the clause initial position in examples such as (21), a true wh-question reading is impossible. But, as I have mentioned above, these questions receive an echo-question interpretation, despite the lack of Clause Typing. This is not surprising, since, according to Bošković (2002), echo-questions do not need to be clause typed. When a question with wh-phrases in situ is pronounced with a yes/no-question (rising) intonation or has a yes/no-question marker in the clause initial position, the question will be interpreted as a yes/no-question. In these cases wh-phrases in the in situ position are interpreted as indefinite pronouns. (23) (A) je zdravnik QY/N AUX doctor izpostavil koga čemu? expose what.DAT who.ACC ‘Did the doctor expose somebody to something?’ 128 Such questions can be used as clarifying questions. For the example (21), we know that the doctor had several patients which he exposed to different viruses. However, we want to know precisely who was exposed to which virus. This is not surprising, since clarifying questions are sometimes taken to be similar to echo questions (see 7.1.1). 237 #‘Who did the doctor expose to what?’ Again the lack of true wh-question reading is not surprising. Because no wh-phrase moves to the clause initial position, the question is not typed as a wh-question. It is, on the other hand, typed as a yes/no-question with an overt or covert question marker. In this thesis I am only concerned with multiple wh-questions that receive a true wh-question reading. Because of this I will not be considering the structure and interpretation of yes/no- and echo-questions any further. But one crucial conclusion needs to be made based on the lack of true question reading in examples in which no wh-phrase moves to the clause initial position: Movement to the clause initial position is necessary for the question to be interpreted as a true wh-question. And importantly for optionality, this also means that movement to the clause initial position is obligatory in true multiple wh-questions in Slovenian. 7.2 Obligatory and non-obligatory wh-movement in Slovenian In section 2.1 I described the optionality of multiple wh-fronting in multiple whfronting languages as instances of multiple wh-questions in which not all wh-phrases have to move to the clause initial position but rather at least one wh-phrase has to be moved to the clause initial position while the rest can undergo short movement or stay in situ. In fact, section 7.1.3 shows that movement of one wh-phrase to the clause initial position, a projection I have argued to be to SpecInterP in 3.5, is obligatory as a wh-phrase needs to move to the clause initial position in order for the question to get a true wh-question reading. This means that we have obligatory movement of the clause initial wh-phrase on the one hand, and (apparent) optional movement of the non-initial wh-phrases. While these properties might seem to be contradictory, they can be accounted for if we assume the existence of two types of wh-pronouns (see also section 5.1.2): (i) Interrogative wh-pronouns which are specified with an uninterpretable valued wh-feature. These interrogative wh-pronouns move to check the wh-feature: 238 (ii) a. in the Left Periphery (the Interrogative or the Wh-Projection). b. in the Low Periphery (the Wh-Projection) Underspecified bare wh-pronouns which have no wh-feature. These stay in situ (and need to be licensed). Such a division means wh-movement is in fact obligatory, but it also does not exclude wh-in-situ in Slovenian. Among the wh-phrases that must move because, (i), are the clause initial wh-phrases, which can move to SpecInterP because the Interrogative Projection hosts an interpretable complex Q+wh-feature, responsible for Clause Typing, with an EPP-subfeature which needs to be checked, cf. Soare (2007), and it is checked by an interrogative wh-phrase, see section 3.5. In addition, wh-phrases in (i) that must move can appear in other positions (but not in situ) – in these cases the wh-phrases obligatorily move, but they are moved either to WhProjections in the Left Periphery or in the Low Periphery. This results in what seems to be optional movement. As for the wh-phrases that appear in situ, wh-phrases with no wh-feature, (ii), have no motivation to move and therefore stay in situ. In what follows, I will argue that it is in fact not optional whether the non initial wh-phrase moves or not. In fact, wh-movement is obligatory and the apparent optionality is a consequence of the existence of two separate lexical items, (i) and (ii). This is also wanted from the Minimalist perspective, which takes optional, semantically vacuous movement as an undesirable property of grammar (cf. 2.1). 7.2.1 Optionality and wh-in-situ In section 2.1 I started off with a description of what optionality is considered to be. To recapitulate, in Minimalism the edge of a phase is made available for a head H by an EPP feature and it is determined in the lexicon whether the H has such a feature or not (Chomsky 2004). In this approach, apparent optionality is related to the EPP feature and is thus determined in the lexicon. When the head has the EPP feature, movement is obligatory. The prediction is also that movement to the edge position will result in an interpretation different from the one that a structure with no movement receives. Biberauer and Richards (2006), on the other hand, claim that 239 semantically vacuous (true) optionality exists and that such optional movement does not result in a new interpretation. Crucially, in both of these two approaches movement has to be motivated. Typically, movement is considered to be a Last Resort operation where Last Resort in defined in Chomsky (1995) as a principle that moderates movement and forces movement to happen only when necessary: (24) Move F raises F to target K only if F enters a checking relation with a sublabel of K. This means that movement will only happen when there is a feature that motivates movement. What was proposed in this thesis, based on Soare (2007), is that in cases of multiple wh-fronting to the Left Periphery, the Inter0 comes with a complex Q+whfeature and a Wh-head comes with a wh-feature – both are interpretable, but unvalued. The wh-feature on the wh-phrase is uninterpretable, but valued. Features on the wh-phrases match the interpretable features on the heads of Wh-Projections. And finally, because the Interrogative Projection and the Wh-Projections in Slovenian come with an EPP-subfeature, movement occurs.129 This is schematized below. (25a) shows the agreement relation between the probes (Inter0 and Wh0) and the goals (wh-phrases). (25b) shows the structure of the multiple wh-question after wh-movement. Note that the reverse word order of wh-phrases is also possible. (25) a. b. [InterP [Inter0] …[Wh-P [Wh0-] … wh-phrase1 … wh-phrase2 ]] iQ+wh iwh EPP EPP uwh uwh [InterP [wh-phrase1] [Inter0] … [Wh-P [wh-phrase2] [Wh-0] … ]] In turn, if any of the wh-phrases in structure such as (25) do not move, the sentence will be ungrammatical because the EPP-feature on Inter0 or Wh0 will be left unchecked. The question would also be ungrammatical if only an Interrogative head 129 Given the Minimalist understanding of optionality, this means that movement to Wh-Projection is obligatory – if it were optional, the Wh-head would come with or without an EPP feature. 240 is projected in a case with two interrogative wh-phrases. In this case, only one whphrase gets its uninterpretable wh-feature checked, while one does not, and because all uninterpretable features need to be checked for a derivation to converge, this leads to ungrammaticality. But as we have seen, there are also multiple wh-questions with a wh-phrase in situ in Slovenian. In these, the wh-phrase that seems to be an interrogative phrase is in fact a bare wh-pronoun, which means that it comes without a wh-feature. For a questions such as (1b) in the beginning of chapter 5 this means that only a single Interrogative Projection, without any additional Wh-Projections, is projected in the Left Periphery. A wh-phrase with no wh-feature is merged in addition to a wh-phrase with a wh-feature. The features on the probe (Inter0) and the goal (the interrogative wh-phrase) will agree and the interrogative wh-phrase will move because of the EPPsubfeature – all in accordance with wh-movement being a Last Resort operation. And crucially, because the non-interrogative wh-phrase (wh-phrase2 in (26)) does not have a wh-feature, it requires no feature matching with a probe (a Wh-head that would potentially also carry an EPP-feature), hence no movement happens. The apparent wh-in-situ is then interpreted as an interrogative for other reasons, see chapter 5. (26) a. [InterP [Inter0]… … wh-phrase1 … wh-phrase2] iQ+wh uwh [ ] EPP b. EPP InterP [wh-phrase1] [Inter0] … wh-phrase2 This also means that in these cases a derivation will not converge if a Wh-head is projected in addition to an Interrogative head. Agreement between the Interrogative head and the interrogative wh-phrase is not problematic and happens just as described above, but if a Wh-head is projected, it requires agreement between the interpretable, unvalued wh-feature on the Wh-head and the uninterpretable whfeature on the only remaining candidate – the bare wh-indefinite (wh-phrase2 in the structure below). This cannot happen simply because the bare wh-indefinite has no uninterpretable wh-feature. Because the EPP subfeature on the Wh-head will be left unchecked and the interpretable wh-feature unvalued, such a derivation will crash: 241 (27) a. * [InterP [Inter0] …[Wh-P [Wh0-] … wh-phrase1 … wh-phrase2 ]] iQ+wh iwh uwh [ ] EPP EPP b. * [InterP [wh-phrase1] [Inter0] … [Wh-P [Wh-0] … wh-phrase2]] iwh EPP This means that wh-in-situ in Slovenian is not a result of optional movement. Wh-in-situ occurs because wh-movement is ‘well-behaved’ in Slovenian, i.e. it only happens when it is motivated and therefore necessary. Still, both multiple whquestions with a wh-in-situ and with multiple wh-movement receive the same interpretation since the wh-phrase in situ is interpreted as interrogative because of the reasons described in chapter 5. 7.2.2 Optionality and movement to the peripheries As I have already shown in the previous section, wh-movement is obligatory when a wh-phrase with a wh-feature enters the derivation. In a multiple wh-question, a whfeature of one wh-phrase needs to be checked in the Interrogative Projection which is located in the Left Periphery of the clause, but the wh-features of the remaining whphrases are checked in the Wh-Projections. The Wh-Projections can either be located in the Left Periphery of the clause (the clause external, High Periphery) or in the Low Periphery (the Clause Internal Periphery). Crucially, movement to the WhProjection is obligatory, which means that multiple wh-movement is not optional in Slovenian. What seems to be optional is the periphery in which the Wh-Projection, to which a wh-phrase moves, is located. But before we turn to optionality, we first need to ask, why several peripheries even exist. That there is more than just the Left Periphery was already observed in Belletti (2004)130, who explores the Low Periphery of the VP in Italian in addition to the clause external periphery in the sense of Rizzi (1997). Belletti also asks why a 130 Others have also argued for additional peripheries. See for example Jayaseelan (2001), Poletto (2006), etc. 242 parallel configuration introduces the IP (the configuration in question is then the clause external Left Periphery, the extended CP) and the VP (the Low Periphery). While Belletti does not provide an answer to this question, she notes that the described parallel is related to Chomsky’s (2000) observation that the CP and the VP are both (strong) phases (in Chomsky (2001), the two strong phases are vP and CP), which means that both are units of the derivation that are to some extent independent, as Belletti (2004) writes. In fact, Chomsky (2008) observes that C is a ‘shorthand’ for the Left Periphery, in the sense of Rizzi (1997). In addition to CP and V/vP, a Focus and a Topic Projection was also proposed for the Italian DP in Giusti (1996). This means that the DP is also assumed to have a periphery (cf. Mišmaš (2014) for Slovenian) which is in line with the observation in Chomsky (2008) that DP is a phase. It then seems that the existence of a periphery is closely related to the notion of a phase, that is, a phase is ‘closed off’ with a periphery. Another related question (already discussed in, for example, Bianchi (2014)), is if the hierarchy of functional projections is the same in all of the peripheries. Specifically, it has been noted that Force is only projected in (some) CPs, see Bianchi (2014). It seems that the hierarchies are not always the same. As we have seen, for example, in Slovenian, the Interrogative Projection is only found in the clause external periphery (and not in the Low Periphery). This is not surprising, given that the role of the wh-phrase in the Interrogative Projection is to type the entire clause as an interrogative, true wh-question. Wh-Projections, on the other hand, do not type the sentence but rather just host the interrogative wh-phrases. A similar claim was made in Cheng (1991) who observed that in multiple wh-fronting languages not all wh-phrases are fronted for Clause Typing, but that fronting is necessary to obtain an interrogative reading for wh-words. The fact that the noninitial wh-phrases do not necessarily need to appear in the clause external periphery is therefore not surprising.131 131 A Wh-Projection is not always available in a clause internal periphery. Shlonsky (2012), for example, observes that the Low Periphery of the vP in French includes Focus, but not a WhProjection. However, French is a language with (optional) single wh-movement. Assuming that whmovement in French is responsible for Clause Typing, it is not surprising that it proceeds to the High Periphery of the clause, since Clause Typing happens in the High Periphery. A clause internal WhProjection is not needed in French. 243 Returning to optionality and different peripheries, Slovenian is not the only language that has a clause internal and clause external periphery in which similar projections are available. For example, Belletti (2004) shows that focus movement in Italian can proceed to the low or high Focus Projection (that is, to the Focus Projection in the Low or in the High Periphery). According to Belletti (2004), however, in Italian, movement to the two different Focus Projections results in different interpretation: the Focused phrase is either interpreted as having a new (when in the Low Periphery) or a contrastive (when in the High Periphery) interpretation. Even more, the two focused phrases are pronounced differently (i.e. different stress, see Belletti (2004) for more). Focus movement in Italian is then well behaved if we assume the Minimalist understanding of optionality – movement to two different locations results in two different interpretations. Considering a different approach to optionality, developed in Biberauer and Richards (2006), this movement is not truly optional as movement is not semantically vacuous and the two different structures receive different interpretations. Slovenian, on the other hand, is different. As we have seen in section 7.1.2, questions in which one wh-phrase undergoes short movement in Slovenian receive the same interpretation as questions with multiple wh-fronting to the Left Periphery in Slovenian. Note also that a different proposal was made for the Italian data in Brunetti (2004), who claims that the interpretative distinction is not as sharp as described in Belletti (2001) – that is, Brunetti claims that a contrastive focus in Italian can either move to the High or stay in the Low Periphery. Brunetti proposes that movement of the contrastive focus to the High Periphery has a pragmatic effect. Because of the pragmatic effect, movement is again not semantically vacuous and therefore not problematic from the viewpoint of optionality. Based on the interpretation and context in which short wh-movement is available in Slovenian, see section 7.1.2, movement to different peripheries does not result in any special pragmatic effects. Comparing focus movement in Italian and wh-movement to a non-initial position in Slovenian, we can conclude that both Italian focused phrases and Slovenian non-initial wh-phrases can appear in either the High or Low Periphery. But the crucial difference is that different positions in Italian lead to different interpretation s (or at least some pragmatic effects) while in Slovenian being in the Low or in the High Periphery does not influence the interpretation of the non-initial 244 wh-phrase or the multiple wh-question as a whole. Still, the non-initial wh-phrase needs to move, when carrying a wh-feature. This means that while wh-movement by itself is not optional, the final position of the non-initial wh-phrase is truly optional (and therefore does not lead to different interpretations). 245 8 Conclusion Writing about multiple wh-fronting at this point in time might seem unnecessary since so many researchers have explored the phenomena for such a long time and from so many different perspectives. However, with the exception of Golden (1996a, 1997a), Slovenian was never really considered in the discussion. Even more, it seems that in light of the existence of multiple wh-fronting, wh-questions in which not all wh-phrases are clause initial were not considered among multiple wh-questions (again, with some exceptions; for example, Šimík’s (2010) exploration of wh-in-situ in Czech or restrictions on movement in Bošković (2002)). A possible explanation for this is that speakers of various languages with multiple wh-fronting disagree about the relevant data. Speakers of Slovenian, however, seem to agree that whphrases can in fact stay in situ in multiple wh-questions, and thus Slovenian can function as an interesting source of data. However, because Slovenian had been left out of the multiple wh-fronting discussion up to now, multiple wh-fronting in Slovenian needed to be considered in this thesis before questions with wh-in-situ and optional wh-movement, as these phenomena seem to be related to multiple whfronting. Because of this, the thesis was intended both as a systematic overview of multiple wh-questions in Slovenian and as an inquiry into why multiple whmovement in Slovenian seems to be optional. In relation to the first goal of the thesis, the analysis of multiple wh-fronting in Slovenian, I tried to establish how approaches proposed in Rudin (1988) and Bošković (1997a et seq.) cope with Slovenian data (and data on multiple wh-fronting in general), and proposed an analysis of multiple wh-questions in Slovenian assuming the Cartographic approach. I proposed that in multiple wh-questions with multiple wh-fronting all wh-phrases move to the Left Periphery of the sentence, specifically to the Interrogative Projection (the clause initial wh-phrase) and to WhProjections. This raised a further question of how the Left Periphery in Slovenian is structured and how movement to the Left Periphery proceeds. In relation to the second point, I proposed that movement of wh-phrases to the Left Periphery is in fact wh-movement and not focus movement – despite the fact that there are no Superiority effects in Slovenian. I suggested that the condition on chains that holds in other languages, for example in Bulgarian, as proposed in Krapova and Cinque 246 (2005), does not hold in Slovenian and correlated this with the availability of other free word order phenomena, e. g. scrambling. However, further exploration will be necessary to determine whether or not there is a general correlation between the ability to violate conditions on chains and free word order. Returning to the structure of the Left Periphery, some initial observations about the structure of the Left Periphery were made in the thesis, but again, the extended CP needs to be further investigated (examining, for example, the question of whether or not all Topic or Focus Projections in Slovenian are equal). In relation to the second goal, optionality, I have in fact argued against whmovement in Slovenian being optional. This, on the one hand, makes the title of my thesis terribly misleading, but is, from a wider perspective, wanted, since optional, semantically vacuous movement is taken to be an undesirable property of grammar (cf. 2.1). What I argued for is that when a wh-phrase comes with an uninterpretable valued wh-feature, it will also move to the Interrogative Projection or the WhProjections. This also means that wh-movement is obligatory, but it also does not exclude wh-in-situ in Slovenian. That is, while a wh-phrase with a wh-feature has to move, a wh-phrase does not necessarily come with a wh-feature (the bare whpronouns). It is also obligatory that a wh-phrase needs to appear in a clause initial position (the Interrogative Projection) for a question to receive a true wh-question reading, which is related to the fact that the Interrogative Projection hosts a interpretable complex Q+wh-feature, responsible for Clause Typing, with an EPPsubfeature which needs to be checked, cf. Soare (2007). However, it does seem to be optional whether wh-phrases move to WhProjections in the Left Periphery or in the Clause Internal (Low) Periphery. I argued for the existence of the Low Periphery based on the fact that Topic, Focus and whphrases can all move to a clause internal position in which they follow the subject. Postulating the Low Periphery accounts for multiple wh-questions with short movement, but it also opens a discussion on the difference between scrambling and wh-movement in languages with multiple wh-movement. That is, if a language has wh-in-situ with wh-phrases without wh-features and the language has scrambling – can these bare wh-pronouns be scrambled? An intuitive answer is yes, but unfortunately there is no clear test to determine whether a wh-phrase in matrix questions is scrambled or wh-moved, which is an issue that needs to be resolved. 247 Short movement also begs the question about the precise structure of the Low Periphery. While it again seems that in Slovenian the word order of Topic, Focus and wh-phrases is free, more attention needs to be given to this topic. As I have already mentioned, I proposed that wh-in-situ in Slovenian is a result of the existence of bare wh-pronouns in Slovenian. These do not move, because they do not have to. The proposal is based on Cheng’s (1991) account of wh-pronouns that can appear in polarity contexts in multiple wh-fronting languages and on the proposal for the interpretation of wh-phrases in wh-in situ languages. This account is simply extended to include the interrogative interpretation of wh-in-situ in multiple wh-languages. Crucially, this proposal for Slovenian relies on the fact that a whphrase needs to appear in a clause initial position in wh-questions with a wh-in-situ which is again linked to the Interrogative Projection with the complex Q+wh-feature which needs to be projected in the Left Periphery, and it is the Q+wh-feature that licenses the interrogative reading of the wh-phrase in situ. An advantage of the suggested account is that it simply complements the existing proposal based on the data from Slovenian. This means that what seems to be optional wh-movement in Slovenian is not a result of restrictions on movement (though, as we have seen, the Principle of Distinctness restricts multiple wh-movement in Slovenian to some extent), it is not movement that would result in different interpretations, and it is also not dependent on context, it is simply a result of the existence of two types of wh-pronouns. And again, what seems to be optional movement to a clause internal or clause initial position is in fact obligatory wh-movement that dislocates wh-phrases to WhProjections in the Low or the Left Periphery (and one wh-phrase obligatorily to the Interrogative Projection in the Left Periphery). What then needs to be further explored is the relation between the Low and the Left Periphery in Slovenian and in general and why movement to these different projections can result in the same interpretation. The primary focus of this thesis was Slovenian, which means that it is an open question to what extent the proposals made here can be applied to other languages. While one can take this is as a weakness, this allowed me to concentrate on the three types of questions within the language and consequently relate them to each other, which I think is an advantage. Also, several issues were only touched upon in the 248 thesis and will need to be addressed in future work, which I think is a good thing as it shows that more can be said, primarily about Slovenian, multiple wh-questions, and multiple wh-fronting but naturally also about many other related topics and issues. 249 9 References Aboh, Enoch O. & Pfau, Roland. 2011. What’s a wh-word got to do with it? In Beninca, Paola & Munaro, Nicola (eds.), Mapping the left periphery: The cartography of syntactic structures 5, 91–124. Oxford: Oxford University Press. Aoun, Joseph & Hornstein, Norbert & Lightfoot, David & Weinberg, Amy. 1987. Two types of locality. Linguistic Inquiry 18(4). 537–577. Bailyn, John Frederick. 2001. On scrambling: A reply to Bošković and Takahashi. Linguistic inquiry 32(4). 635–658. Bajec, Anton et al. 2000. Slovar slovenskega knjižnega jezika (SSKJ) [The dictionary of standard Slovenian language]. Ljubljana: Slovenska akademija znanosti in umetnosti. (Available at http://bos.zrc-sazu.si/sskj.html) Baker, Carl L. 1997. Notes on the description of English questions: The role of an abstract question morpheme. Foundations of language 6. 197–219. Beck, Sigrid. 2006. Intervention effects follow from focus interpretation. Natural Language Semantics 14(1). 1–56. Belletti, Adriana. 2001. ‘Inversion’ as focalization. In Hulk, Aafke C. J. & Pollock, Jean-Yves (eds.), Inversion in Romance and the theory of Universal Grammar, 60–90. Oxford: Oxford University Press. Belletti, Adriana. 2004. Aspects of the low IP area. In Rizzi, Luigi (ed.), The structure of CP and IP. The cartography of syntactic structures 2, 16–51. Oxford: Oxford University Press. Belletti, Adriana & Shlonsky, Ur. 1995. The order of verbal complements: A comparative study. Natural Language & Linguistic Theory 13(3). 489–526. Bianchi Valentina. 2014. Focus fronting and the syntax-semantics interface. To appear in Shlonsky, Ur (ed.), Beyond the functional sequence. Oxford: Oxford University Press. Biberauer, Theresa & Richards, Marc. 2006. True optionality: When the grammar doesn’t mind. In Boeckx, Cedric (ed.), Minimalist Essays, 35–67. Amsterdam: John Benjamins. Błaszczak, Joanna & Fischer, Susann 2001. Multiple Wh-Konstruktionen im Slavischen: State of the Art Report. Linguistics in Potsdam 14. 1–116. 250 Bošković, Željko. 1994. ECP, Spec-Head agreement, and multiple wh-movement in Overt Syntax. In Avrutin, Sergey & Franks, Steven & Progovac, Ljiljana (eds.), Formal Approaches to Slavic Linguistics: The MIT Meeting, 1993, 119−144. Ann Arbor: Michigan Slavic Publications. Bošković, Željko. 1997a. Fronting wh-phrases in Serbo-Croatian. In Lindseth, Martina & Franks, Steven (eds.), Formal Approaches to Slavic Linguistics: The Indiana Meeting, 1996, 86–107. Ann Arbor: Michigan Slavic Publications. Bošković, Željko. 1997b. On certain violations of the Superiority Condition, AgrO, and the Economy of Derivation. Journal of Linguistics 33. 227–254. Bošković. Željko. 1997c. Superiority effects with multiple wh-fronting in SerboCroatian. Lingua 102. 1−20. Bošković, Željko. 1998. Wh-phrases and wh-movement in Slavic. (Position paper for the Comparative Slavic Morphosyntax Conference in Bloomington, Indiana, June 1998). Bošković, Željko. 1999. On multiple feature-checking: Multiple wh-fronting and multiple head movement. In Epstein, Samuel & Hornstein, Norbert (eds.), Working Minimalism, 159−187. Cambridge: MIT Press. Bošković, Željko. 2000. Sometimes in [Spec,CP], sometimes in situ. In Martin, Roger & Michaels, David & Uriagereka, Juan (eds.), Step by step: Essays on minimalist syntax in honor of Howard Lasnik, 53–87. Cambridge: MIT Press. Bošković, Željko. 2001a. On the interpretation of multiple questions. In Pica, Pierre (ed.), Linguistic Variation Yearbook, 1−15. Amsterdam: John Benjamins. Bošković, Željko. 2001b. On the nature of the syntax-phonology interface. Clitization and related phenomena. Amsterdam: Elsevier. Bošković, Željko. 2002. On multiple wh-fronting. Linguistic Inquiry 33(3). 351−383. Bošković, Željko. 2003. On wh-islands and obligatory wh-movement contexts in South Slavic. In Boeckx, Cedric & Grohmann, Kleanthes (eds.), Multiple whfronting, 27–50. Amsterdam: John Benjamins. Bošković, Željko. 2004. Topicalization, focalization, lexical insertion, and scrambling. Linguistic Inquiry 35(4). 613–638. Bošković, Željko. 2007. A note on wh-typology. In Kosta, Peter & Schurcks, Lilia (eds.), Linguistic investigations into formal description of Slavic languages: 251 Contributions of the Sixth European Conference held at Potsdam University, 2005, 159–70. Frankfurt am Main: Peter Lang. Bošković, Željko. 2008. The NP/DP analysis and Slovenian. In Proceedings of the Novi Sad Generative Syntax Workshop 1, 53–73. Novi Sad: Filozofski fakultet u Novom Sadu. Bošković, Željko. 2009. Scrambling. In Kempgen, Sebastian & Kosta, Peter & Berger, Tilman & Gutschmidt, Karl (eds.), The Slavic Languages, 714–724. Berlin: Mouton de Gruyter. Bošković, Željko & Takahashi, Daiko. 1998. Scrambling and last resort. Linguistic Inquiry 29(3). 347–366. Browne, Wayles & Alt, Theresa. 2004. A Handbook of Bosnian, Serbian, and Croatian. Slavic and East European Language Research Center (SEELRC), Duke University. (Available at http://www.seelrc.org:8080/grammar/mainframe.jsp?nLanguageID=1) Bruening, Benjamin. 2007. Wh-in-situ does not correlate with wh-indefinites or question particles. Linguistic Inquiry 38(1). 139–166. Brunetti, Lisa. 2004. A unification of focus. Padova: Unipress. (Ph.D. dissertation) Cable, Seth. 2010. The grammar of Q: Q-particles, wh-movement, and pied-piping. Oxford: Oxford University Press. Cheng, Lisa Lai‐Shen. 1991. On the typology of wh-questions. Cambridge: MIT. (Ph.D. dissertation) Cheng, Lisa Lai‐Shen & Rooryck, Johan. 2000. Licensing wh‐in‐situ. Syntax 3(1). 1−19. Chomsky, Noam. 1973. Conditions on Transformations. In Anderson, Stephen R. & Kiparsky, Paul (eds.), A Festschrift for Morris Halle, 232–286. New York: Holt, Rinehart & Winston Chomsky, Noam. 1986. Barriers. Cambridge: MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge: MIT Press. Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Martin, Roger & Michaels, David & Uriagereka, Juan (eds.), Step by step: Essays on minimalist syntax in honor of Howard Lasnik, 89–155. Cambridge: MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Kenstowicz, Michael (ed.), Ken Hale: A life in language, 1–52. Cambridge: MIT Press. 252 Chomsky, Noam. 2004. Beyond Explanatory Adequacy. In Belletti, Adriana (ed.), Structures and beyond: The cartography of syntactic structures, vol. 3, 104–131. Oxford: Oxford University Press. Chomsky, Noam. 2008. On phases. In Freidin, Robert & Otero, Carlos & Zubizarreta, Maria Luisa (eds.), Foundational issues in linguistic theory, 133– 166. Cambridge: MIT Press. Cinque, Guglielmo. 1999. Adverbs and functional heads. Oxford: Oxford University Press. Cinque, Guglielmo & Rizzi, Luigi. 2008. The cartography of syntactic structures. Studies in linguistics 2. 42–58. Citko, Barbara. 1998. On multiple wh-movement in Slavic. In Bošković, Željko & Franks, Steven & Snyder, William (eds.), Formal Approaches to Slavic Linguistics: the Connecticut Meeting 1997, 97−113. Ann Arbor: Michigan Slavic Publications. Citko, Barbara. 2010. On the (a)symmetric nature of movement: A view from Polish wh- and passive movement. In Browne, Wayles & Cooper, Adam & Fisher, Alison & Kesici, Esra & Predolac, Nikola & Zec, Draga (eds.), Formal Approaches to Slavic Linguistics 18: The Cornell Meeting 2009, 38–57. Ann Arbor: Michigan Slavic Publications. Citko, Barbara & Gračanin-Yuksek, Martina. 2013. Towards a new typology of coordinated wh-questions. Journal of Linguistics 49(1). 1–32. Comorovski, Ileana. 1996. Interrogative phrases and the syntax-semantics interface. Dordrecht: Kluwer Academic Publishers. den Dikken, Marcel & Giannakidou, Anastasia. 2002. From hell to polarity: “Aggressively non-D-linked” wh-phrases as polarity items. Linguistic Inquiry 33(1). 31–61. Denham, Kristin. 1997. A minimalist account of optional wh-movement. Washington: University of Washington. (Ph.D. dissertation) Denham, Kristin. 2000. Optional wh-movement in Babine-Witsuwit’en. Natural Language and Linguistic Theory 18(2). 199–251. Dornisch, Ewa. 1998. Multiple-Wh-Questions in Polish: The Interactions between Wh-Phrases and Clitics. Ithaca: Cornell University. (Ph.D. dissertation) 253 Dyakonova, Marina. 2009. A phase-based approach to Russian free word order. Amsterdam: University of Amsterdam. (Ph.D. dissertation) Fanselow, Gisbert. 2006. Partial wh‐movement. In Everaert, Martin & van Riemsdijk, Henk (eds.), The Blackwell companion to syntax: 437–492. Malden: Blackwell Publishing. Fanselow, Gisbert. 2010. Semantic type effects on crossing movement in German. Language and Logos: Studies in Theoretical and Computational Linguistics 72. 48–63. Fanselow, Gisbert & Aäussler, Jana & Weskott, Thomas. 2013. Constituent Order in German Multiple Questions: Normal Order and (Apparent) Anti-Superiority Effects. The Mind Research Repository (beta). Vol. 1. Fox, Danny, 1995. Economy and scope. Natural language semantics 3(3). 283–341. Franks, Steven. 1998. Clitics in Slavic. (Paper presented at the Comparative Slavic Morphosyntax Workshop, Spencer Creek, June 1998) Gigafida. [A written corpus of modern Slovenian.] Available at: http://www.gigafida.net/. Giusti, Giuliana. 1996. Is there a FocusP and a TopicP in the noun phrase? University of Venice Woking Papers in Linguistics 6(2). 105–128. Golden, Marija, 1996a. Interrogative Wh-Movement in Slovene and English. Acta Analytica 14. 145–186. Golden, Marija. 1996b. K-premik in skladenjski otoki v slovenski skladnji [Whmovement and syntactic islands in Slovenian syntax]. Razprave XV, Razred za filološke in literarne vede. 237−253. Golden, Marija. 1997a. Multiple wh-questions in Slovene. In Browne, Wayles & Dornisch, Ewa & Kondrashova, Natasha & Zec, Draga (eds.), Formal Approaches to Slavic Linguistics: The Cornell Meeting, 1995, 240−266. Ann Arbor: Michigan Slavic Publications. Golden, Marija. 1997b. Parasitic Gaps in Slovene. In Junghanns, Uwe & Zybatow, Gerhid (eds.), Proceedings in the First European Conference on Formal Description of Slavic Languages, 387−398. Frankfurt am Main: Vervuert Verlag. Golden, Marija & Milojević Sheppard, Milena. 2000. In Beukema, Fritz & den Dikken, Marcel (eds.), Clitic phenomena in European languages, 191–207. Amsterdam/ Philadelphia: John Benjamins. 254 Grebenyova, Lydia 2006a. Sluicing Puzzles in Russian. In Filip, Hana & Franks, Steven & Lavine, James E. & Tasseva-Kurktchieva, Mila (eds.), Formal Approaches to Slavic Linguistics: The Princeton Meeting, 2005, 157−171. Ann Arbor: Michigan Slavic Publications. Grebenyova, Lydia 2006b. When multiple wh-fronting is not multiple. (Talk given at The 1st Annual Meeting of Slavic Linguistic Society, Bloomington, Indiana, 8–10 September 2006) Greenberg, Marc L. 2008. A short reference grammar of Slovene. München: Lincom. (Available at www.seelrc.org:8080/grammar/pdf/stand_alone_slovene.pdf) Grewendorf, Günther. 2001. Multiple Wh-fronting. Linguistic Inquiry 32(1). 87−122. Grohmann, Kleanthes K. 2003. German is a multiple wh-fronting language. In Boeckx, Cedric & Grohmann, Kleanthes (eds.), Multiple wh-fronting, 99–130. Amsterdam: John Benjamins. Hagstrom, Paul. 1998. Decomposing Questions. Cambridge: MIT. (Ph.D. dissertation) Halupka-Rešetar, Sabina. 2013. Left peripheral matters in Serbian: The role of discourse-pragmatics in word order variation and how to motivate it. (Talk given at Sinfonija 7, Niš, Serbia, 26–28 September 2013) Heim, Irene. 1982. The semantics of definite and indefinite noun phrases. Amherst: University of Massachusetts. (Ph.D. dissertation) Ilc, Gašper. 2008. O zanikanju in nikalnici v slovenščini [On negation and the negative word in Slovenian]. Jezik in slovstvo LIII(2). 65–81. Ilc, Gašper & Milojević Sheppard, Milena. 2003. Verb movement in Slovenian: a comparative perspective. STUF-Language Typology and Universals 56(3). 266– 286. Jayaseelan, Karattuparambil A. 2001. IP-internal Topic and Focus Phrases. Studia Linguistica 55(1). 39–75. Jędrejowski, Łukasz. 2014. Again on why. But why? In Chapman, Cassandra & Kit, Olena & Kučerová, Ivona (eds.), Formal Approaches to Slavic Linguistics. The McMaster Meeting 2013, 151–169. Ann Arbor: Michigan Slavic Publications. Kayne, Richard. 1994. The antisymmetry of syntax. Cambridge: MIT Press. Kiss, Katalin É. 1998. Identificational focus versus information focus. Language 74. 245–273. 255 Krapova, Iliyana. 2004. Word order in Topic-Focus structures in the Balkan languages. In Romània Orientale XVII. Atti del IX convegno internazionale di studi sul sud-est europeo (Tirana, 30.08–3.09. 2004): 139–162. Krapova, Iliyana & Cinque, Guglielmo. 2005. On the order of wh-phrases in Bulgarian multiple wh-fronting. University of Venice Working Papers in Linguistics 15. 171–197. Kuno, Susumu. 1993. Grammar and discourse principles: Functional syntax and GB theory. Chicago: University of Chicago Press. Kuno, Susumu & Robinson, Jane J. 1972. Multiple wh-questions. Linguistic Inquiry 3(4). 463–487. Laenzlinger, Christopher & Soare, Gabriela. 2005. On merging positions for arguments and adverbs in the Romance Mittelfeld. In Brugè, Laura & Giusti, Giuliana & Munaro, Nicola & Schweikert, Walter & Turano, Giuseppina (eds.), Contributions to the thirtieth Incontro di Grammatica Generativa 2005, 105–128. Venice: Università Ca’Foscari Venezia. Lambova, Mariana. 2003. On the interaction of multiple (non) wh-fronting and multiple Topicalization in Bulgarian. University of Pennsylvania Working Papers in Linguistics 8(1). 127–142. Liakin, Denis. 2005. On the Wh-Movement and Wh-in-situ in Russian. In Franks, Steven & Gladney, Frank & Tasseva-Kurktchieva, Mila (eds.), Formal Approaches to Slavic Linguistics 13: the Columbia Meeting, 218–230. Ann Arbor, Michigan Slavic Publications. Marušič, Franc. 2005. On non-simultaneous phases. Stony Brook: Stony Brook University. (Ph.D. dissertation) Marušič, Franc. 2006. Slovenian Clitics Have No Unique Syntactic Position. In Antonenko, Andrei & Bailyn, John & Bethin, Christina (eds.), Formal Approaches to Slavic Linguistics 16. The Stony Brook Meeting, 266–281. Ann Arbor: Michigan Slavic Publications. Marušič, Franc. 2008. CP under control. In Zybatow, Gerhild & Szucsich, Luka & Junghanns, Uwe & Meyer, Roland (eds.), Formal Description of Slavic Languages, 408–422. Frankfurt am Main: Peter Lang. Marušič, Franc & Mišmaš, Petra & Plesničar, Vesna & Razboršek, Tina & Šuligoj, Tina. 2014. On a potential counter-example to Merchant's Sluicing-COMP 256 generalization. (Paper presented at SinFonIJA 7, Graz, 25–27 September 2014, to appear in Grazer Linguistische Studien) Marvin, Tatjana & Stegovec, Adrian. 2012. On the syntax of ditransitive sentences in Slovenian. Acta Linguistica Hungarica 59(1–2). 177–203. Merchant, Jason. 2001. The syntax of silence: Sluicing, islands, and the theory of ellipsis. Oxford: Oxford University Press. Meyer, Roland. 2004. Superiority effects in Russian, Polish and Czech: Judgments and Grammar. University of Leipzig/ University of Regensburg. (Manuscript) Mihaliček, Vedrana. 2010. The ordering of wh words in multiple wh extraction in Serbo-Croatian. In Smirnova, Anastasia & Mihaliček, Vedrana & Ressue, Lauren (eds.), Formal Studies in Slavic Linguistics, 103–120. Newcastle: Cambridge Scholars Publishing. Mišmaš, Petra. 2013. The influence of grammatical features on linearization: evidence from Slovenian. In Franks, Steven (ed.), Formal Approaches to Slavic Linguistics. The Third Indiana Meeting, 218–231. Ann Arbor: Michigan Slavic Publications. Mišmaš, Petra. 2014. An Argument for Wh-fronting in the Slovenian DP. In Veselovská, Ludmila & Janebová, Markéta (eds.), Nominal Structures: All in Complex DPs, Olomouc modern language monographs; vol. 2, 175–186. Olomouc: Palacký University. Mišmaš, Petra. 2015. Wh-in Situ in a Multiple Wh-fronting Language. In Ziková, Markéta & Caha, Pavel & Dočekal, Mojmír (eds.), Slavic Languages in the Perspective of Formal Grammar Proceedings of FDSL 10.5, Brno 2014, 199–213. Frankfurt am Main: Peter Lang. Miyagawa, Shigeru. 2011. Optionality. In Boeckx, Cedric (ed.), The Oxford Handbook of Linguistic Minimalism, 354–376. Oxfrod: Oxford University Press. Müller, Gereon. 1999. Optionality in Optimality-Theoretic Syntax. GLOT International, vol. 4(5). 3–8. Müller, Gereon. 2002. Free word order, morphological case, and sympathy theory. In Fanselow, Gisbert & Féry, Caroline (eds.), Resolving Conflicts in Grammars: Optimality Theory in Syntax, Morphology, and Phonology (Linguistische Berichte Sonderheft 11), 9–49. Hamburg: Buske. 257 Nevins, Andrew. 2012. Haplological dissimilation at distinct stages of exponence. In Trommer, Jochen (ed.), The Morphology and Phonology of Exponence, 84–116. Oxford: Oxford University Press. Nishigauchi, Taisuke. 1990. Quantification in the Theory of Grammar. Dordrecht: Kluwer Academic Publishers. Pereltsvaig, Asya. 2007. On the Universality of DP: A View from Russian. Studia Linguistica 61(1). 59–94. Pesetsky, David. 1987. Wh-in-situ: Movement and unselective binding. In Reuland, Eric & ter Meulen, Alice G. B. (eds.), The representation of (in) definiteness, 98– 129. Cambridge: MIT Press. Pesetsky, David. 2000. Phrasal movement and its kin. Cambridge: MIT Press. Pires, Acrisio & Taylor, Heather Lee. 2007. The syntax of wh-in-situ and common ground. In Masullo, Pascual (ed.), Romance Languages: Structure, interfaces, and microparametric variation: Proceedings of the 37th LSRL/ Linguistic Symposium on Romance Languages. 201–215. Amsterdam: John Benjamins. Poletto, Cecilia. 2006. Parallel Phases: a study on the high and low left periphery of Old Italian. In Frascarelli, Mara (ed.), Phases of interpretation, 261–294. Berlin: De Gruyter. Potsdam, Eric. 2004. Wh-questions in Malagasy. In Law, Paul (ed.), Proceedings of AFLA 11 (ZAS Papers in Linguistics 34), 244–258, Berlin: Zentrum für Allgemeine Sprachwissenschaft. Potsdam, Eric. 2006. More concealed pseudoclefts in Malagasy and the clausal typing hypothesis. Lingua 116(12): 2154–2182. Progovac, Ljiljana. 2005. A syntax of Serbian: clausal architecture. Bloomington: Slavica Publishers. Richards, Norvin. 1997. What moves where when in which language? Cambridge: MIT. (Ph.D. dissertation) Richards, Norvin. 2001. Movement in language: Interactions and architectures. Oxford: Oxford University Press. Richards, Norvin. 2010. Uttering trees. Cambridge: MIT Press. Rizzi, Luigi. 1990. Relativized minimality. Cambridge: MIT Press. 258 Rizzi, Luigi. 1996. Residual verb second and the Wh-Criterion. In Belletti, Adriana & Rizzi, Luigi (eds.). Parameters and functional heads 2, 63–90. Oxford: Oxford University Press. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Haegeman, Liliane (ed.), Elements of Grammar. A Handbook in Generative Syntax, 281–337. Dordrecht: Kluwer Academic Publishers. Rizzi, Luigi. 2001a. On the position ―Int(errogative) in the left periphery of the clause. In Cinque, Gugliermo & Salvi, Giampaolo (eds.), Current studies in Italian syntax: Essays offered to L. Renzi, 287–296. New York: Elsevier. Rizzi, Luigi. 2001b. Relativized minimality effects. In Baltin, Mark & Collins, Chris (ed.), The Handbook of Contemporary Syntactic Theory, 89–110. Oxford: Blackwell. Rizzi, Luigi. 2004. Locality and Left Periphery. In Belletti, Adriana (ed.), The structure of CP and IP. The cartography of syntactic structures 3, 223–251. Oxford, New York: Oxford University Press. Rizzi, Luigi. 2011. Minimality. In Boeckx, Cedric (ed.), The Oxford handbook of linguistic minimalism, 220–238. New York: Oxford University Press. Rizzi, Luigi. 2013. Locality. Lingua 130. 169–186. Rojina, Nina. 2011. The syntactic structures of Russian wh-questions. Geneva: Université de Genève. (Ph.D. dissertation) Rudin, Catherine. 1988. On multiple questions and multiple fronting. Natural Language and Linguistic Theory 6. 445−501. Roussou, Anna & Vlachos, Christos. 2011. Introduction to Optionality in WhMovement. Linguistic Analysis 37.1/2. 3–10. Sabel, Joachim. 2003. Malagasy as an optional wh-fronting language. In Boeckx, Cedric & Grohmann, Kleanthes (eds.), Multiple wh-fronting, 229−254. Amsterdam: John Benjamins. Sabel, Joachim. 2006. Typologie des W-Fragesatzes [The typology of wh-questions]. Linguistische Berichte 206. 147–194. Sabel, Joachim & Zeller, Jochen. 2006. Wh-question formation in Nguni. In Mugane, John & Hutchison, John P. & Worman Dee A. (eds.), Selected proceedings of the 35th annual conference on African linguistics, 271–283. Somerville: Cascadilla Proceedings Project. 259 Shlonsky, Ur. 2010. The cartographic enterprise in syntax. Language and linguistics compass 4(6): 417–429. Shlonsky, Ur. 2012. Notes on wh in situ in French. In Brugè, Laura & Cardinaletti, Anna & Giusti, Giuliana & Munaro, Nicola & Poletto, Cecilia (eds.), Functional Heads. The Cartography of Syntactic Structures, vol. 7, 242–252. New York: Oxford University Press. Shlonsky, Ur & Soare, Gabriela. 2011. Where’s ‘Why’? Linguistic inquiry 42(4). 651–669. Soare, Gabriela. 2007. A cross-linguistic typology of question formation and the antisymmetry hypothesis. Generative grammar in Geneva 5. 107–133. Stepanov, Arthur. 1998. On Wh-fronting is Russian. In Tamanji, Pius N. & Kusumoto, Kiyomi (eds.), Proceedings of NELS 28, 453–467. Amherst: University of Massachusetts, GLSA. Stepanov, Arthur. 2000. Wh‐scope marking in Slavic. Studia Linguistica 54(1). 1–40. Stepanov, Arthur & Tsai, Wei-Tien Dylan. 2008. Cartography and licensing of whadjuncts: a cross-linguistic perspective. Natural Language & Linguistic Theory 26(3). 589–638. Stjepanović, Sandra. 1995. Short-distance movement of wh-phrases in SerboCroatian matrix clauses. Storrs: University of Connecticut. (Unpublished manuscript) Stjepanović Sandra. 1999. What do second position cliticization, scrambling, and multiple wh-fronting have in common? Storrs: University of Connecticut. (Ph.D. dissertation) Stjepanović, Sandra. 2003. Multiple wh-fronting in Serbo-Croatian matrix questions and the matrix sluicing construction. In Boeckx, Cedric & Grohmann, Kleanthes (eds.), Multiple wh-fronting, 255–284. Amsterdam: John Benjamins. Strahov, Natalya. 2001. A scrambling analysis of Russian wh-questions. In Franks, Steven & Holloway King, Tracy & Yadroff, Michael (eds.), Formal Approaches to Slavic Linguistics: The Bloomington Meeting 2000, 293–310. Ann Arbor: Michigan Slavic Publications. Szczegielniak, Adam. 2001. Polish Optional Movement. In Alexandrova, Galina M. & Arnaudova, Olga (eds.) The Minimalist Parameter: Selected papers from the 260 Open Linguistics Forum, Ottawa, 21–23 March 1997, 137−159. Amsterdam: John Benjamins. Šimík, Radek. 2010. Interpretation of multiple interrogatives: An information structure sensitive account. In Browne, Wayles & Cooper, Adam & Fisher, Alison & Kesici, Esra & Predolac, Nikola & Zec, Draga (eds.), Formal Approaches to Slavic Linguistics 18: The Cornell Meeting 2009, 486–501. Ann Arbor: Michigan Slavic Publications. Toman, Jindrich. 1981. Aspects of multiple wh-movement in Polish and Czech. In May, Robert & Koster, Jan (eds.), Levels of syntactic representation, 293–302. Dordrecht: Foris Publications. Toporišič, Jože. 2004. Slovenska slovnica [Grammar of Slovenian]. Maribor: Obzorja. Veselovská, Ludmila. 1993. Wh-Movement and multiple questions in Czech. Durham Durham University. (Masters dissertation) Wachowicz, Kristina. 1974. Against the universality of a single wh-question Movement. Foundations of Language. International Journal of Language and Philosophy 11: 155–166. Witkoś, Jacek. 2008. On the correlation between A-type scrambling and the lack of Weak Crossover effects. Studia Anglica Posnaniensia 44. 297–328. Yanovich, Igor. 2005. Choice-functional series of indefinite pronouns and Hamblin semantics. Proceedings of SALT, vol. 15. 309–326. Yeo, Weichiang Norman. 2010. Unifying Optional Wh-movement. York: University of York. (Ph.D. dissertation) Zavitnevich, Olga. 2001. On wh-questions and on nature of wh-words in Russian in comparison with English. Working Papers in English and Applied Linguistics 6. 109–134. (Cambridge: Re-search Centre for English and Applied Linguistics, University of Cambridge) Zavitnevich-Beaulac, Olga. 2005. On wh-movement and the nature of wh-phrases– case re-examined. Skase Journal of Theoretical Linguistics 2(3). 75–100. 261
© Copyright 2024