English/spanish discourse properties: the identification of A/A'-positions

English/Spanish discourse properties: The identification of A/A’-positions José Miguel Ruiz-Villaécija © José Miguel Ruiz-Villaécija, 2014 A mi director de tesis, Ángel Jiménez, por permanecer siempre a mi lado en este largo viaje. A Ramón Jiménez Pérez y Alfonso Jiménez Fernández, cuya pérdida ha acompañado a mi director en la última etapa de esta tesis. 4 The Beatles The Long and Winding Road 5 TABLE OF CONTENTS 0. INTRODUCTION…………………………………………………………….. 7 1. STATE OF THE ART………………………………………………………… 11 1.1. Introduction.................................................................................................... 12 1.2. Fixed word order and free word order languages………………………….. 12 1.3. Discourse functions………………………………………………………... 14 1.3.1. On the definition of topic and focus……………………….………. 14 1.3.1.1. Focus.…………………………………………………………. 15 1.3.1.2. Topic………………………………………………………….. 17 1.4. A typology of foci…………………………................................................. 20 1.5. A typology of topics...................................................................................... 26 1.6. On main differences between Topic and Focus…………………………… 38 1.7. On the A/A’-distinction…………………………………………………… 39 1.8. Phases……………………………………………………………….……... 41 1.9. Concluding remarks…...…………………………………………………... 49 2. THE SYNTAX OF DISCOURSE FUNCTIONS.……………..….…………. 52 2.1. Introduction………………………………………………………………... 53 2.2. The base-generation analysis…………...………………………....…...….... 53 2.3. Arguments supporting movement-analysis…..……………………………... 57 2.4. Syntactic approaches to discourse functions…..………………………….... 64 2.4.1. A unified movement-analysis.............................................................. 67 2.5. What motivates displacement?.............…………………...……………...... 75 2.5.1. Prosodic motivations.......................................................................... 75 2.5.2. A formal feature system..................................................................... 76 2.6. The cartography of syntactic structures………………..………………….. 79 2.7. CP/TP............................................................................................................ 86 2.8. Concluding remarks...……………………………………………………... 88 6 3. DISCOURSE MOVEMENT IN SPANISH IS ARGUMENTAL AS OPPOSED TO ENGLISH.................................................................................................... 90 3.1. Introduction……………………………………………………………….. 91 3.2. On the independence relation between syntax and discourse ……………. 91 3.3. The A-/A’-properties of movement……………………………………….. 94 3.3.1. Evidence for A-movement analysis………………………………... 94 3.3.2. Evidence for A’-movement analysis……………………………….. 99 3.4. Is Focus Aor A’-movement? A feature-based analysis………………….. 100 3.5. Concluding remarks...……………………………………………………... 109 4. MAIN CLAUSE PHENOMENA……………………………..……………… 111 4.1. Introduction……………………………………………………………….. 112 4.2. Factive clauses and discourse movement....………………………………. 113 4.3. Intervention effects……………………….……………………………….. 115 4.3.1. Embedded Topicalisation and Focalisation in English....................... 117 4.3.2. Embedded CLLD and Focalisation in Spanish.................................. 120 4.4. Empirical evidence…………...…………………………………................. 128 4.4.1. Objective............................................................................................ 128 4.4.2. Hypothesis......................................................................................... 128 4.4.3. Methodology...................................................................................... 128 4.4.4. Tests................................................................................................... 129 4.4.5. Discussion.......................................................................................... 139 4.4.6. Conclusion.......................................................................................... 145 4.5. Concluding remarks...................................................………….................... 147 5. CONCLUSIONS……………………………………………………………… 149 6. REFERENCES………………………………………………………………... 153 7. LIST OF ABBREVIATIONS AND SYMBOLS............................................ 170 7 0. INTRODUCTION 8 The present study focuses on the comparison of the nature of discourse constituents and their positions in the sentence in English and Spanish, within a generative approach to language. Specifically, this dissertation argues for a reconsideration of the A/A’-distinction. Evidence for the reformulation includes binding facts, reconstructions effects, weak-cross-over phenomena, parasitic gap constructions and the syntax of floating quantifiers. In this way, we demonstrate that discourse elements behave in a different way and have a different location in English and Spanish. The languages we concentrate on the dissertation are mainly English and Spanish but many results can be carried over to other languages. The main body of the study is organized as follows. In Chapter 1, we deal with an old claim: languages differ with respect to word order. In this case, we introduce the notions of focus and topic which are two key concepts in discourse analysis and information structure. In addition, we analyse the different types of focus and topics that have been traditionally assumed to exist in natural language. Finally, we introduce the concept of phases. Discourse constituents undergo movement cyclically and through different phases. Chapter 2 discusses the syntax of discourse functions. We state that topics and foci move to a high position in the sentence, i.e. they are not base-generated. More concretely, we give further evidence in favour of analysing clitic left dislocation (CLLD) and hanging topic (HT) as involving movement and base-generation respectively. We essentially entertain the same dual analysis for English, with topic fronting involving movement and left dislocation (LD) being a base-generation structure. We also assume a movement analysis for focus fronting in English and Spanish. As for the landing-site of these discourse elements that move in a clause, the cartography linguistics of Rizzi (1997, 2004) and others consider that topic and focus 9 are part of the articulated CP system. However, an unsplit CP analysis is adopted for English. At the same time, the Generalized TP analysis (Zubizarreta 1998), proposes that fronted topics and foci in Spanish have the specifier of Tense Phrase (TP) as their landing site. We will discuss these analyses in our next chapter. Chapter 3 concentrates on the difference between A-movement and A´- movement, argumental and non-argumental movement. In this case, we see how the properties of a movement determine whether it is argumental or non-argumental. So, focus movement in Spanish is of an A-nature. Dealing with English, the specifier of TP is not an available landing site for discourse constituents, that is, foci and topics have to move to the Complementiser Phrase (CP). In Chapter 4, we analyse the differences between English and Spanish with respect to the application of the operations of Topicalisation and Focalisation in main clauses and its possible extension to subordinate contexts. A movement analysis of factive clauses allows us to analyse this restriction in terms of an intervention effect. In this regard, factive clauses could be defined as finite object clauses. More concretely, as originally noted by Kiparsky & Kiparsky (1970), factive predicates are distinguished from non-factives in that the former presuppose the truth of their complement clause while the latter do not. Indeed, following Haegeman (2004, 2010, 2012), we outline an operator movement analysis of factive clauses, and show how this account derives the lack of discourse movement from an intervention effect between the moved operator and some left peripheral element. In other words, discourse movement in English is incompatible with factive clauses since topics and foci move to [Spec, C], which accounts for the incompatibility of English Focalisation/Topicalisation and factive clauses in terms of intervention effects. In contrast to English, Spanish discourse elements arguably stay within the TP domain. Therefore, focus and topic elements do 16 sentence which are not presupposed are the focus of a sentence. As we can see through the question and answer pairs in (6)-(7), any constituent, or even the entire sentence, may be the focus. The focus constituent in the (b) sentences occurs in square brackets. (6) a. ¿Qué comió Marcos? What eat-PAST-3SG Marcos ‘What did Marcos eat?’ b. Marcos comió [salchichas FOCUS]. Marcos eat-PAST-3SG sausages ‘Marcos ate [sausages].’ (7) a. ¿Qué hizo Marcos? What do-PAST-3SG Marcos ‘What did Marcos do?’ b. Marcos [comió salchichas FOCUS]. Marcos eat-PAST-3SG sausages ‘Marcos [ate sausages].’ (8) a. ¿Qué pasó? What happen-PAST-3SG ‘What happened?’ b. [Marcos comió salchichas FOCUS]. Marcos eat-PAST-3SG sausages ‘[Marcos ate sausages.]’ In (6) and (7), we can observe how the question in (a) gives us information about what is presupposed in the answer. In (8), the question provides no presupposition; likewise, in discourse-initial contexts, an entire sentence may be a focus. So far, we have dealt with information focus. However, Rizzi (1997) claims that a focalised element can be marked by special prosodic features and it can be preposed to the beginning of the sentence. In this case, the focus is contrastive 2 . Such an operation is known as Focalisation. Let us consider the example below: (9) YOUR PEN you should give to John (not mine). 2 It is also called identificational focus by some authors such as É. Kiss (1998). 17 In (9) the DP your pen provides new information and it has a kind of contrastive function in the context. The remaining part of the sentence transmits given information which is shared by the participants of the communicative situation. This second part is the presupposition of the sentence. Additionally, Pan’s (2007) study explored the influence of focus on syllable duration of lexical tones in Taiwenese. More concretely, his purpose was to examine the surface duration of Twainese lexical tones under different focus conditions. The four focus conditions included a broad focus condition with focus on the entire sentence, and three narrow focus conditions with narrow focus falling on the first, second, and third words. Results revealed that the duration of narrow focus syllables are longer than broad focus syllables, which in turn are longer than post-focus syllables. 1.3.1.2. Topic As in the case of focus, the identification of topic in a sentence depends on discourse context. In line with Jiménez-Fernández (2005), Jayaseelan (2001), Belletti (2004), Zubizarreta (1999), Zagona (2002), Göbbel (2005), Lambrecht (1994), Erteschik-Shir (2006) and Crystal (1991), the topic of a sentence can be described as the starting point for the rest of the sentence, that element which expresses what the sentence is about. For instance, given the context (10), (10) ¿Qué ocurrió con David? What happen-PAST-3SG with David ‘What happened with David?’ we can deduce that the answer is a sentence whose topic is David: (11) [David TOPIC] se emborrachó. David CL drink-PAST-3SG ‘David got drunk.’ 18 According to Biskup (2007), Frascarelli (2007) and Lahousse (2009), the concept of topic is generally taken to be determined by the notion of “aboutness”. This notion is based on the presuppositions of speaker and hearer. In the exchange starting in (10), it is presupposed that something happened to David; therefore the phrase [DP David] is presuppositional, and in the context of the question (10), is the natural discourse topic of answers in (11). From a syntactic point of view, topic constituents can be dislocated. For instance, (13) is an alternative to (12): (12) I don’t really like this book. (13) This book I don’t really like. In (13), the constituent this book is followed by a juncture and, optionally, by a pause. In this connection, topic constituents must receive a specific interpretation which is a consequence of their nature as presupposed material. Compare the (a) and (b) examples below: (14) a. Algunos compañeros leyeron ese libro, pero no sé cuáles. ‘Some partners read that book, but I don’t know which (ones).’ b. ?*Algunos compañeros, leyeron ese libro, pero no sé cuáles. ‘Some partners, (they) read that book, but I don’t know which ones.’ (15) a. Ningún compañero leyó ese libro. ‘No partner read that book.’ b. ?*Ningún compañero, leyó ese libro. ‘No partner, (s/he) read that book.’ The (a) examples indicate that non-specific preverbal subjects are grammatical. In the (b) examples, however, where the subject is set off by comma intonation and is obligatorily interpreted as topic, the non-specific topic is ungrammatical. Indeed, in accordance with Goodall (2001), in Spanish it is the EPP which triggers movement of the subject to the specifier of TP. At the same time, he points out that Spanish has very 19 robust Topicalisation and Focalisation processes, so it could be that a DP to the left of V is there by virtue of that type of movement, rather than by being attracted by an EPP feature. These processes would involve A’-movement to the CP layer of the clause. Consequently, Goodall states that subjects are fronted by a mechanism different from that which fronts topics. Although İşsever (2003: 1040) arrives at the same conclusion for Turkish, he states that non-specifics can also appear in topic position: 3 (16) A-Yalın üniversiteyi kazanırsa ne yapacak? ‘What will Yalın do if he is accepted to the university?’ a. Bir fakir-i yemeğ-e götürecekmiş. Herhangi bir fakir olurmuş. a poor-ACC dinner-DAT bring-FUT-PERF any one poor be-AOR-PERF ‘He will take a poor person to the dinner. Any poor person will do, he said.’ b. Bir çocuğ-u tepeden tırnağa giydirecekmiş. a child-ACC from.top.to.toe clothe-FUT-PERF ‘He will clothe a child from top to toe.’ c. Bir kiz-la hemen evlenecekmiş. Hangi kı olduğu a girl-comitative immediately marry-FUT-PERF which girl be-REL-ACC hiç farketmezmiş. never matter ‘He will marry a girl immediately. (He said that) it didn’t matter which one.’ d. Bir hayır kurumun-a 1000 dolar bağışlayacakmış. a charity-DAT dollar donate-FUT-PERF ‘He will donate 1000 dollars to a charity.’ In each sentence in (16), we can observe how non-specific DPs can appear in initial position of a sentence. According to Enç (1991) and Kornfilt (2003), Specificity and Case are closely related in Turkish. This fact suggests that the concept of topic is not determined by the notion of specificity. Therefore, we conclude that the relation between the linguistic terms of specificity and topicality seems to be language particular. 3 Such an account contradicts Breul’s (2004) intuition that referentiality entails specificity. 20 1.4. A typolology of foci Following linguists like É. Kiss (1998), Zubizarreta (1998), Donati & Nespor (2001), Benincà & Poletto (1999) or Belletti (2001), two different types of focus exist in natural languages in general. Consequently, as anticipated in Section 1.3.1.1, contrastive focus, which expresses exhaustive identification, must be distinguished from information focus, which merely expresses new information. To be more precise, information focus is characterized as purely new information (for the hearer), whereas contrastive focus involves new information which corrects or makes a contrast with a previous assertion: 4 (17) a. Where did you see my brother? b. We saw your brother in the restaurant. (18) a. I thought you met John. b. PETER we met (not John). In (17), in the restaurant stands for the information focus. It is identified with the interrogative operator in the question. By contrast, in (18), there is a contrast between the DPs Peter and John, indicating that the unit that receiving prominence is the correct one in comparison with other possibilities. Peter is then a contrastive focus. Even though the existence of these focus types has been recognized in the literature, the distinction between them has often been ignored. The former has occurred in Brunetti’s analysis of Italian (2004), for example. In her approach, focus elements never express exhaustive identification. There is only one type of focus which expresses new information 5 . 4 To differentiate between contrastive focus and information focus, we use capital letters for the former and underlining for the latter. 21 By contrast, Bianchi (2013) discusses optional movement of focus constituents to the left periphery of the clause in Italian. She shows that the fronted position and the ‘low’ position are not completely equivalent. In particular, she sketches out a characterization of two distinct interpretations for the focus structure: a truly corrective interpretation (where focus occurs in a reply that denies the preceding assertion), and a merely contrastive one: (19) a. Gianni ha invitato Lucia. John have-PERF-3SG invite Lucy ‘John invited Lucy.’ b. MARINA has invitato (non Lucia). Marina have-PERF-3SG invite (not Lucy) ‘Marina he invited (not Lucy).’ (Bianchi 2013: 1) (20) a. Vi saluto, devo tomare a casa. CL greet-PRES-1SG. have-to-PRES-1SG go-back-INF. to home ‘Good bye, I have to go back home.’ b. A quest’ora, ti conviene prendere IL TAXI, non la metro. At this hour, CL suit-PRES-2SG take-INF. the taxi, not the underground ‘At this time of day, you’d better take a taxi, not the underground.’ (Bianchi 2013: 12) The reply in (19b) conveys a correction of speaker A’s assertion. Meanwhile, (20b) is an instance of purely contrastive focus without any corrective import. According to Bianchi, focus fronting in (20b) would be infelicitous. In this way, the necessary condition for focus fronting relates not to the contrastive import per se but rather, to the corrective import which establishes a contrast across utterances. Bianchi holds that although fronted focus implements a corrective move, it can occur in certain embedded clauses. As we will see in Chapter 4, Hooper & Thompson (1973) points out that those 5 According to Brunetti (2004), the distinction between contrastive and information focus does not exist at any level of the grammar. From an interpretative point of view, there is only one focus expressing non-presupposed information. From a prosodic point of view, the rules of accent placement and the relation between accent and focus domain are always the same. From a syntactic point of view, a focus can always move to the left and it always possesses operatorlike properties. 22 embedded clauses that allow ‘root transformations’ are endowed with assertive force. Assertion is a property of declarative root clauses; in order to be compatible with Main Clause Phenomena (MCP), embedded clauses must be asserted, i.e., non-presupposed. Nevertheless, as we will see in Chapter 4, Section 4.2, such an account is problematic. On the contrary, the Spanish equivalent of (20) allows focus fronting: (21) a. Adiós, he de regresar a casa. good bye have to-PRES-1SG go-back-INF to home ‘Good bye, I have to go back home.’ b. A esta hora, UN TAXI deberías coger (no el metro). at this hour a taxi had better-PRES-2SG take-INF not the underground ‘At this time of day, a taxi you’d better take, not the underground.’ In (21), the constituent a taxi is an instance of purely contrastive focus without any corrective import. Such a constituent has undergone movement to the front of the clause. Therefore, it is obvious that in Spanish focus fronting does not imply a corrective interpretation. According to Gussenhoven (2007), languages that make a formal distinction between information focus and corrective focus include Efik, where a focused answer to a wh-question is not expressed in the same way as a focused correction, which requires a corrective focus particle (de Jong 1980; Gussenhoven 1983). Lekeito Basque, too expresses corrective focus and information focus differently (Elorditea 2007). In this respect, contrastive focus has syntactic and semantic properties that a mere information focus does not share. These properties are illustrated below: I) In languages like Hungarian the position of contrastive focus is inaccessible for some elements such as universal quantifiers and is ‘also’-phrases since they do not involve any exclusion (É. Kiss 1998: 251): 23 (22) a. Mari EGY KALAPOT nézett ki magának. Mary a hat.ACC pick-PAST-3SG out herself.DAT ‘It was a hat that Mary picked for herself.’ b. *Mari MINDED KALAPOT nézett ki magának. Mary every hat.ACC pick-PAST-3SG out herself.DAT ‘It was every hat that Mary picked for herself.’ c. *Mari EGY KALAPOT IS nézett ki magának. Mari a hat.ACC also pick-PAST-3SG out herself.DAT ‘It was also a hat that Mary picked for herself.’ The Hungarian examples (22b) and (22c), containing a universal quantifier and an is ‘also’-phrase in their immediately preverbal focus position respectively, are ungrammatical. In the grammatical versions of these sentences, the universal quantifier or ‘also’-phrase would occupy a quantifier position between the topic and the contrastive focus. The English equivalents of these sentences display a similar restriction, that is, universal quantifiers and also-phrases cannot occur as cleft constituents, the English realizations of contrastive focus (É. Kiss 1998: 251): (23) a. It was A HAT that Mary picked for herself. b. *It was EVERY HAT that Mary picked for herself. c. *It was also A HAT that Mary picked for herself. As for Spanish, these types of constituents can appear in the position of contrastive focus, as illustrated in (24): (24) a. UNA MANZANA compró María en el mercado. an apple bring-PAST-3SG Maria at the market ‘It was an apple that Maria bought at the market.’ b. TODAS LAS MANZANAS compró María en el mercado. every apple bring-PAST-3SG Maria at the market ‘It was every apple that Maria bought at the market.’ c. También UNA MANZANA compró María en el mercado. also an apple bring-PAST-3SG Maria at the market ‘It was also that Maria bought at the market.’ 24 The universal quantifier and the is ‘also’-phrase in (24b) and (24c) respectively are located in the preverbal section of such sentences. This fact suggests that the syntactic properties of contrastive focus in Hungarian, English and Spanish are not the same. Indeed, according to É. Kiss (1998), the contrastive focus itself is not uniform across languages 6 . Meanwhile, in Hungarian (25) and English (26) information focus does not share these distributional restrictions (É. Kiss 1998: 253): (25) a. Kiket hívtál meg a születésnapodra? who.PL.ACC invite-PAST-2SG.you PERF your birthday.to ‘Who did you invite to your birthday?’ b. Minden kollégámat meg hívtam. every colleague.my.ACC PERF invite-PAST-1SG.I ‘I invited every colleague of mine.’ c. Egy szomszédomat is át hívtam. a neighbour.my.ACC also over invite-PAST-1SG.I ‘I called over also a neighbour of mine.’ (26) a. Who did you invite to your birthday? b. I invited every colleague of mine. c. I called over also a neighbour of mine. In (25) and (26), we observe that universal quantifiers and ‘also’-phrases can have the properties of information focus: they can convey nonpressuposed information, and can bear a pitch accent. In other words, in Hungarian and English certain types of constituents cannot function as contrastive foci; but the type of constituents that can function as information focus is not restricted. In particular, it seems safe to conclude that contrastive focus, realised as an immediately preverbal constituent in Hungarian, and as a cleft constituent in English, is subject to distributional restrictions; it cannot be constituted by a universal quantifier or 6 In Chapter 3, Section 3.4, we will explain the reasons why contrastive focus behaves in a different way in English and Spanish. 25 an also-phrase (except for some languages like Spanish). Information focus, on the other hand, is exempt from most of these distributional restrictions. II) The contrastive focus occupies a scope position both in English and in Hungarian; it marks the sentence part following it and c-commanded by it as the scope of exhaustive identification. More specifically, exhaustive identification has narrow scope with respect to an operator c-commanding the contrastive focus, and has wide scope with respect to an operator c-commanded by the contrastive focus. In the example below, the contrastive focus Mary is in the scope of the universal quantifier always, and takes scope over the universal quantifier every boy. Hence, the sentence means that on every relevant occasion, of all the relevant girls it is Mary and no one else that all the boys want to dance with (É. Kiss 1998: 254): (27) It is always MARY that every boy wants to dance with. Regarding Spanish, the situation is the same, that is, the focalised constituent con María is in the scope of the universal quantifier siempre, and takes scope over the universal quantifier todos los chicos: (28) Siempre CON MARIA quieren bailar todos los chicos. According to Authier & Haegeman (2012), sentence-initial adjuncts do not block argument fronting in English. In particular, they state that a fronted argument is compatible with an adjunct to its right. Similarly, we claim that the left peripheral adjunct siempre does not block the fronting of the focused element con María. However, the adjunct is to the left of the focus. Information focus as such, however, cannot enter into a scope relation with a clause-mate operator. More concretely, the only possible interpretation of the Hungarian 32 (40) a. Mi hermanak, las empanadasj, los domingos, prok lasj hace de pollo. my sister the turnovers the Sundays CL make-PRES-3SG of chicken ‘My sister, the turnovers, on Sundays, she fills them with chicken.’ b. Las empanadas, mi hermana, los domingos, pro las hace de pollo. the turnovers my sister the Sundays CL make-PRES-3SG of chicken c. Los domingos, mi hermana, las empanadas, pro las hace de pollo. the Sundays my sister the turnovers CL make-PRES-3SG of chicken (41) a. Mi hermanak, las empanadasj, los domingos, prok lasj hace esa my sister the turnovers the Sundays CL make-PRES-3SG that tontitak de pollo. silly-goose of chicken ‘My sister, the turnovers, on Sundays, that silly-goose fills them with chicken.’ b. Las empanadas, mi hermana, los domingos, pro las hace esa the turnovers my sister the Sundays CL make-PRES-3SG that tontita de pollo. silly-goose of chicken c. Los domingos, mi hermana, las empanadas, pro las hace esa on Sundays my sister the turnovers CL make-PRES-3SG that tontita de pollo. silly-goose of chicken Besides, CLLDs do not license parasitic gaps (PGs) (42), or cause weak-crossover (WCO) effects (43) (Suñer 2006: 137): (42) a. A una candidate el jefe la descartó sin to a candidate.ACC the boss CL.ACC scrap-PAST-3SG without entrevistar*(la). interviewing*(her) ‘One candidate the boss scrapped her without interviewing her.’ b. A una candidata el jefe la descartó sin entrevistar*(la) a la pobrecita. ‘One candidate the boss scrapped her without interviewing*CL the poor woman’ (43) a. A Teresak suk madre lak quiere mucho. to Teresa.ACC her mother CL.ACC love-PRES-3SG a.lot ‘Teresa her mother loves her a lot.’ b. A Teresak suk madre lak quiere mucho a la muy loquita. ‘Teresa her mother CL loves the little dear a lot.’ Finally, Suñer points out that a CLLD phrase reconstructs to a site c-commanded by a preverbal subject but not by a postverbal one (Zubizarreta 1998, Ceccheto 2000). 33 Consider (44a), where the bound reading is impossible because of a Principle C violation, while (44b) with a postverbal subject in its VP internal merged position allows for the bound reading (Suñer 2006: 137): (44) a. Al primer hijo de un progenitork pro*k lo recibe siempre to.the first offspring of a father he CL receive-PRES-3SG always con los brazos abiertos. with the arms open ‘The first offspring of a progenitor he always receives him with open arms.’ b. Al primer hijo de un progenitork lo recibe siempre élk to.the first offspring of a father CL receive-PRES-3SG always he con los brazos abiertos. with the arms open And the same happens when the CLLD structure includes an epithet (Suñer 2006: 138): (45) a. Al primer hijo de un progenitork pro*k lo recibe al pequeñín con los brazos abiertos. ‘The first offspring of a progenitor he CL receives the tiny being with open arms.’ b. Al primer hijo de un progenitork lo recibe élk al pequeñín to.the first offspring of a father CL receive-PRES-3SG he the tiny being con los brazos abiertos.’ with the arms open So far, we have seen that Hanging Topic diverges from Clitic Left Dislocation since they have different properties 10 in Italian and Spanish. 10 These two types of topic can co-occur in a fixed order: HT-CLLD. This order is sketched in the following example (Rizzi 2004: 65): (i) a. Giorgio, ai nostri amici, non parlo mai di lui. Giorgio to the our friends not talk-PRES-1SG never of him ‘Giorgio, to our friends, I never talk of him.’ b. *Ai nostri amici, Giorgio, non parlo mai di lui. to our friends Giorgio not talk-PRES-1SG never of him (ia) is grammatical since it displays the order HT-CLLD. (ib), which represents the reverse order, is ungrammatical. 34 In addition, we have to make reference to Object shift which has been extensively studied in the German languages (Holmberg 1986; Diesing 1992; Vikner 1990, 1995; Bobalijk & Thráinsson 1998). In this respect, Suñer (2000) states that Spanish has two types of Object shift, each with its own properties. The first manifestation is semantically motivated and only occurs with presupposed/specific DPs. In this case, Spanish uses clitic-doubling (CL-D), an operation that is obligatory with strong pronouns but dialectally circumscribed with direct object non-pronominals. Consider sentences in (46) from Suñer (2000: 262): (46) a. ¿Cuando *(te) nombraron a ti? when you appoint-PAST-3PL to you ‘When did they appoint you?’ b. Y luego lo miré a él. and then him look-PAST-1SG to him ‘And then I looked at him.’ In (46), the CLs form a chain with the strong pronouns. As clitics are by nature affixal, which means that they require a host, they must attach to V which always raises to Tense in Spanish (Suñer 1994). In this sense, the obligatory CL-doubling of Spanish strong pronouns is due to Diesing’s (1992) Mapping Hypothesis. More specifically, Spanish strong pronouns are definite and referential. At the same time, strong pronouns cannot be interpreted existentially, because they would violate Heim’s (1982) Novelty Condition which requires that entities inside the VP to be new to the discourse; therefore, they must move out of the VP. In such a situation, CL-D achieves object shift without really moving the object itself, but by using the CL as a scope marker to signal where the object is to be interpreted. The second type of Object shift is p-movement (Zubizarreta 1998), a prosodically motivated rule which has the effect of changing the asymmetric ccommand hierarchy of metrical sisters with contradictory prosodic properties, so that 35 the most prominent element ends up being the lowest in the structure. In other words, prominence is assigned by two rules. The first is the Nuclear Stress Rule (NSR), which identifies the last constituent in the VP as prominent, under conditions of neutral stress and intonation. The second prominence rule, the Contrastive Stress Rule (CSR), assign contrastive stress freely. These two prominence rules may assign prominence to different constituents and give rise to a conflict. Then, p-movement becomes relevant (Suñer 2000: 282): (47) a. #Anoche se lastimó SARA la rodilla. CSR NSR b. Anoche se lastimó la rodilla Sara. last night CL hurt-PAST-3SG the knee Sara ‘Last night {SARA/Sara} hurt her knee.’ This problem is solved when p-movement moves the DO around the postverbal subject as in (47b), so that main stress can be achieved through the NSR. The distinction between CLLD and HT that we find in Romance seems to have correlates in other languages. In German and Dutch we have Contrastive Left Dislocation/Left Dislocation (see van Riemsdjik’s 1997 discussion, among many others). In the same vain, following Casielles-Suárez (2004: 5), in English we must distinguish between Topicalisation in (48a), and Left-dislocation in (48b): (48) a. Julia I couldn’t reach. b. John, I saw him yesterday. Casielles-Suárez suggests that the English structures in (48a) and (48b) have both been considered topicalising mechanisms in the sense that they seem to prepose an element which is taken to be the topic of the sentence. However, they are syntactically 36 different; while topicalisations have a gap, where the topicalised phrase would have appeared in the nonpreposed version, left-dislocations include a resumptive pronoun. At the same time, Casielles-Suárez (2004: 73) holds that, at first sight, the English construction in (48b), repeated below as (49), and the Spanish structure in (50) appear to be the same: (49) John, I saw him yesterday. (50) A Juan lo vi ayer. Syntactically, they both involve the preposing of the direct object and the co-occurrence of a resumptive pronoun. From a pragmatic point of view, both structures are considered to be topicalising mechanisms in the sense that they are used to introduce a discourse topic, John in (49) and Juan in (50). Nevertheless, syntactic evidence supports the heterogeneous nature of Spanish CLLD and English LD. First, Spanish CLLD allows not only for the dislocation of noun phrases (NPs), but also for an adjective, a quantifier or even a clause (Casielles-Suárez 2004: 73): (51) a. Listo no lo parece. clever not CL seem-PRES-3SG ‘Clever he doesn’t seem.’ b. A todos no los he visto todavía. to all not CL have-PERF-1SG seen yet ‘Everybody I’ve yet to see.’ c. Que fumas lo sabemos todos. that smoke-PRES-2SG CL know-PRES-1PL all ‘That you smoke we all know.’ On the contrary, English LD only allows for the dislocation of referential NPs. Hence, the English counterparts of (51a-c) would be ungrammatical (Casielles-Suárez 2004: 74): 37 (52) a. *Clever, he doesn’t seem it. b. *Everybody, I haven’t seen them yet. c. *That you smoke, we all know it. Second, there is no limit to the number of phrases that a Spanish CLLD can prepose (Casielles-Suárez 2004: 74): (53) a. A Pedro, los libros, ya se los compré. to Pedro the books already CL CL buy-PAST-1SG ‘The books for Pedro I already bought.’ b. Estos libros, yo, a Juan, nunca se los dejaría. these books I to Juan never CL CL lend-COND-1SG ?‘To Juan these books I would never lend.’ c. Un libro, a mí, Juan, en Navidad, nunca me lo ha regalado. a book to me Juan in Christmas never CL CL have-PERF-3SG given (as a gift) ‘A book for Christmas John has never given me.’ Meanwhile, English LD does not allow for multiple left-dislocations. Thus, the following multiple left-dislocations are ungrammatical (Casielles-Suárez 2004: 74): (54) a. *These books, to John, I would never lend them to him. b. *John, a gift, he has never bought it. Third, the Spanish construction can accomplish two tasks simultaneously: it can place a direct object in sentence-initial position and at the same time place a focal subject in sentence-final position (Casielles-Suárez 2004: 75): (55) La casa la limpié yo. the house CL clean-PAST-1SG I ‘The house I cleaned.’ The English counterpart of (55) would be ungrammatical since English LD does not allow for the order object-verb-subject. 38 With this background in mind, we can conclude that in order to understand the role of topical elements in a particular language we have to take into account the specific characteristics of the language, and examine the ways in which that language encodes topical phrases. This will be done with particular attention to Spanish and English in the following chapters. 1.6. On main differences between Topic and Focus Topic and focus structures are similar since they involve the left periphery of the clause. In fact, they seem to move phrases to the same position. So, how can we distinguish between topic and focus? These two constructions differ in a series of aspects, which emphasize a different nature. What we intend to do now is to make a clear distinction between them. Firstly, wh-phrases in main questions are compatible with topicalised expressions, while they are incompatible with a focus (Jiménez-Fernández 2005): (56) And the milk where did you put it? (57) *THE MILK where did you put (not the wine)? A second difference between focus and topic is that a focalised negative element triggers inversion of auxiliares, whereas Topicalisation does not (Haegeman & Guéron 1999): (58) UNDER NO CIRCUMSTANCES will I write a paper during the holidays. (59) *During the holidays will I write a paper under no circumstances. The third distinguishing property is the fact that a topic never displays any Weak-cross-over effect. These effects are detectable with focus (Rizzi 1997): 39 (60) Mark his mother always appreciated him. (61) ?? MARK his mother appreciated, not Robert. The last difference makes reference to the idea that the number of topics is subject to processing factors but unlimited. By contrast, there is a unique structural focus position: (62) The book, to John, tomorrow, I’ll give it to him for sure. (63) *TO JOHN THE BOOK I’ll give, not to Piero, the article. The next section is devoted to analysing the syntactic derivation of these two discourse notions in a detailed way. 1.7. On the A/A’-distinction Recent debates about clause structure and discourse movement raise significant problems regarding the traditional A/A’-distinction. Within the LGB framework (cf. Chomsky, 1981), an A-position was a position to which theta role could be assigned, i.e., VP internal argument positions and the specifier of TP position. Traditional theories within generative grammar that support the view that the subject of a clause is generated VP internally (Sportiche 1988; Larson 1988; Fukui & Speas 1986; Kitagawa 1986 among others) assume that all the theta roles of V are assigned inside the VP 11 . This implies that [Spec, T] is not even a ‘potential’ theta position. This raises an important question: What is the status of the specifier of TP position? Is it an A-position or an A’-position? The answer to this question is not straightforward. Note that if all arguments get their theta roles VP internally, then the 11 We take into consideration the VP-INTERNAL SUBJECT HYPOTHESIS (ISH) which implies that the lexical shell of the clause, that is, the V and its arguments, are generated inside V so that theta-assignment takes place uniformly under sisterhood. 40 combined assumptions of the classical LGB view and the VP internal subject hypothesis force us to classify the specifier of TP as an A’-position – since no theta role is assigned to an argument in that position. VP internal subjects will have to be Case marked in their base generated (VP internal) position. This would force us to conclude that VP internal subjects (as well as other arguments) can only move to A’-positions. Indeed, Uribe-Etxebarria (1992) considers that the preverbal slot occupied by preverbal subjects in Spanish will behave as a non-argument position. Similarly, Barbosa (2009) holds that argumental subjects (overt or null) are not subject to A-movement to preverbal position and remain in situ; apparent pre-verbal subjects would actually be leftdislocated or A-bar moved to preverbal position. Hence, under this view, in a language like English which has pre-T subjects in declarative sentences, this pre-T position would be an A’-position. Let us analyse the example below (Mahajan 1990: 19): (64) [TP Johni [VP seems [PP to himselfi/himj] [TP Johni to have [VP Johni shot Bill]]]] Since himself must be bound by John while him must be free from it, John must occupy a position that is in the domain of binding conditions A and B. that is, John must be in an A-position 12 . This argument in itself shows that there must be VP-external Apositions. We will put forward a theory of movement that treats this phenomenon as a systematic syntactic operation. In that respect, this study supports views such as Saito (1985), Hoji (1987) and Saito and Hoji (1983). However, we will depart from these studies by showing that movement is not a unitary phenomenon, i.e., it is not simply an instance of A’-movement. As a matter of fact, the discussion in this thesis will rise a number of significant issues especially concerning the typology of positions within a 12 Linguists like Lasnik (1999, 2003), Gutiérrez-Bravo (2006, 2007), Zubizarreta (1998, 1999) or Grewendorf & Sabel (1999) describe [Spec, T] as an A-position. 41 clause, the A/A’ distinction. In Chapter 3, we will argue that such a distinction needs to be revised. But first, we will analyse the syntactic behaviour of topic and focus in Chapter 2. 1.8. Phases As we have already seen, discourse factors trigger the movement of certain elements in a sentence. This movement takes place in a specific and determined way. Thus, lexical items undergo movement cyclically through different phases. But, what are phases? Phases are roughly cycles of syntactic computation that are sent to the semantic and phonological interfaces, where they receive a Logical Form (LF) and a Phonological Form (PF) interpretation, respectively. Due to their propositional nature, Chomsky (1999) suggests that phases are only CP and transitive vP. Keeping to these two phrases, Chomsky (2006, 2008) states that the objects derived have an underlying structure where C and v are the labels triggering internal operations, and are also the points of feature valuation and transfer. As explained below, phasal heads are responsible for the activation of the operation of agree, since they contain unvalued agreement features. In this sense, a pure cyclic computation is fundamental for the simplest account of uninterpretable features. They are unvalued in the lexicon, and when assigned a value, must be eliminated at the phase level to avoid crash. At the same time, the phase head can assign its uninterpretable features to the head it selects by a process of feature inheritance. So, grammatical features are transmitted from C to T and from v to V. In this connection, following Miyagawa (2010), Jiménez-Fernández (2008) and Jiménez-Fernández & Spyropoulos (2013) hold that discourse features can also be inherited by T and V. Moreover, phase heads may also have an edge feature called ‘EPP-feature’. This edge feature allows raising to the phase edge without feature 48 From the analysis of these sentences, the conclusion is that there is a close connection between the concept of phases and discourse movement. This relation implies that the processes of Topicalisation and Focalisation obey the Phase Impenetrability Condition in a strict way. 49 1.9. Concluding remarks From the preceding considerations we can draw the following specific conclusions: - Languages differ with respect to word order. In some of them, sentence constituents can appear in many different positions depending on discourse requirements. They are known as “free word order” languages. In other cases, the distinction between given and new information in a sentence does not cause any variation in the order of the elements. So, we are talking about “fixed word order” languages. - In the relevant literature many discussions can be found about the discourse function of different constituents in a sentence. For this reason, sentences are divided into given and new information which are represented by topic and focus constituents respectively. - Two different types of focus exist in natural languages in general. Hence, contrastive focus, which expresses exhaustive identification, must be distinguished from information focus, which merely expresses new information. As a matter of fact, contrastive focus has syntactic, semantic and phonological properties that a mere information focus does not share. - Regarding Romance languages, we must distinguish between two types of topic: clitic left dislocation and hanging topic. The main difference between them is that the hanging topic only occurs at the periphery of the whole message, whereas the clitic left dislocated topic appears at the periphery of the clause. In the same way, in English we must distinguish between Topicalisation and Left-dislocation. They are syntactically 50 different; while topicalisations have a gap, where the topicalised phrase would have appeared in the nonpreposed version, left-dislocations include a resumptive pronoun. In this sense, we can conclude that in order to understand the role of topical elements in a particular language we have to take into account the specific characteristics of the language, and examine the ways in which that language encodes topical phrases. - Topic and focus structures are similar since they involve the left periphery of the clause. In fact, they seem to move phrases to the same position. However, these two constructions differ in a series of aspects, which emphasize a different nature. - Recent debates about clause structure and discourse movement raise significant problems regarding the traditional A vs. A’-distinction. Traditional theories that support the view that the subject of a clause is generated VP internally assume that all the theta roles of V are assigned inside the VP. This implies that [Spec, T] is not even a ‘potential’ theta position. However, we will depart from these studies by showing that movement is not a unitary phenomenon, i.e., it is not simply an instance of A’- movement. - Due to the limitations of the Language Faculty, lexical items move cyclically through different phases (CP and vP). These phases imply an agree relation between a probe and a goal. Once a phase is formed, the complement of the phase head undergoes transfer to the phonological and semantic components so that neither the complement nor any element it contains can take part in further syntactic operations from that point on. In this sense, we must emphasize the relation between the concept of phases and discourse roles. Since CP is a phase and it can serve as a landing site for discourse elements we 51 conclude that topics and foci could move to the specifier of CP and obey the Phase Impenetrability Condition (PIC). 52 2. THE SYNTAX OF DISCOURSE FUNCTIONS 53 2.1. Introduction Some linguists like Chomsky (1977), Lasnik & Saito (1992) or Rooth (1992) state that discourse elements are generated in the left-periphery of the clause from the very beginning. Other authors such as Rizzi (1997) or Belletti (2004) claim that topic and focus elements move to that position during the derivation. In this chapter, we show evidence that topics and foci move to a higher position in the sentence, i.e. they are not base-generated. In fact, it is conventionally accepted that English Topicalisation and Focalisation involve movement rather than base-generation. However, the analysis of discourse phenomena in Spanish has been much more controversial. We try to clarify such a controversy in this second chapter. 2.2. The base-generation analysis The hypothesis that topics may be either base-generated or derived by movement seems to be an interesting starting point. Following Chomsky (1977), there are similarities in the syntax of Wh-movement and the operation of Topicalisation. He also notes the relation between Topicalisation and LD. Let us observe the following example: (1) a. *This book to whom should we give? b. *John who do you think saw? (2) a. This book, to whom should we give it? b. (As for) John, who do you think saw him? (Chomsky 1977: 94) According to Chomsky, one is derived by movement, whereas the other is basegenerated. In (1), we find an instance of Topicalisation. More specifically, (1a) is ungrammatical since movement of the topicalised expression results in a doubly-filled 54 COMP. In the same way, (1b) is ruled out because the topic is extracted from a Whisland 14 . Meanwhile, the analogous cases with LD (2a-b) are fine. In this connection, Chomsky claims that LD does not involve movement as a number of principles assumed to be diagnostics for movement are violated (Complex Noun Phrase Constraint, Specified Subject Condition, Wh-Islands, Subjacency etc.). 15 Lasnik & Saito (1992), henceforth L&S, approach the situation differently. They point out correctly that the framework following Chomsky & Lasnik (1977) and Chomsky (1981), in which the that-trace effect is explained by a constraint on traces, rules out the examples shown in (3a-b) by the Empty Category Principle (ECP), in contrast to (3c). (3) a. *Johni I think that ti won the race. b. *Who do you think that ti won the race. c. John, I think that he won the race. (L&S 1992: 76) However, they reconsider Chomsky’s (1977) analysis on the basis of a wrong prediction. One of Chomsky’s predictions is that under certain circumstances where LD is possible, Topicalisation should not be possible. On the other hand, wherever Topicalisation is available, LD should always be unavailable. This is so because the structures are identical except that Topicalisation involves movement. The former prediction is strongly confirmed, as Chomsky demonstrates. The latter prediction is not borne out, however, as shown in (4) and (5): 14 Chomsky (1977) argues that in a framework that rejects structure-building transformations, the variety of structures that precede NPs in TOP position eliminates a transformational solution in principle. 15 The insight that LD freely violates island constraints goes back to Ross (1967), as so many observations. 55 (4) a. I believe that this book you should read. b. ... that this solution I proposed last year is widely known. (5) a. *I believe that this book, you should read it. b. *... that this solution, I proposed it last year is widely known. (L&S 1992: 77) On the basis of such data, L&S propose that Topicalisation is adjunction to IP or TP (as similarly proposed by Baltin 1982). 16 By assuming that the position for topics (TP in L&S which I will call TopP, following Rizzi 1997) 17 in English is restricted to one base-generated TopP per sentence, L&S can account for the contrast in (6): (6) a. John, Mary, he likes. (L&S 1992: 78) b. *John, Mary, he likes her. (L&S 1992: 79) The upshot of L&S’s discussion with respect to the present issue is that LD may involve only one LD-ed element as there is only one TopP available, and LD involves base-generation, rather than movement. As commented in Section 2.1, some linguists also consider that foci are generated in the left-periphery of the clause from the very beginning. In this respect, the alternative semantics theory for focus proposed by Rooth (1992) has proven fruitful in 16 Baltin (1982) has pointed out that there are cases where embedded Topicalisation is possible but embedded LD is not. He cites contrasts such as the following: (i) (= Baltin’s (69)) the man to whom liberty we could never grant. (ii) (= Baltin’s (86)) *the man to whom liberty, we could never grant it. In order to account for this contrast, Baltin proposes that LD involves a base-generated topic, whereas Topicalisation involves adjunction to IP. According to him, it is the relativization in (ii) that is illicit. Such an analysis of the contrast between (i) and (ii) is impossible if it is assumed that embedded LD and embedded Topicalisation have identical structures, as proposed by Chomsky (1977). We will discuss the differences between root and embedded Topicalisation in Chapter 4. 17 Under the cartographic approach to the left periphery inaugurated by Rizzi (1997), the functional projection CP is broken down into a number of distinct functional projections. One of these projections is TopP. In Section 2.6, we will analyse such an approach in detail. 56 the analysis of focus in Germanic languages, from both a theoretical and empirical view. Indeed, it is at the base of many recent works which address issues related to the syntax-prosody interface in English (see, for instance, Truckenbrodt 1995; Selkirk 2000, 2007, and related work). Such a theory excludes the possibility that focus associates with movement, be it overt or covert movement. In this connection, Barbosa (1996) is one of the most outstanding linguists who insist on the idea that there is no movement in topic dislocation. Following Cinque (1990), she suggests that there is a series of differences between Focus and CLLD which lead us to think that Focus is similar to wh-movement whereas CLLD implies some different form of construal. In other words, Barbosa assumes that CLLD involves base-generation of the dislocated topic in a position of adjunction to the XP that is predicated of it; by contrast, it is suggested that non-referential quantified phrases, when fronted, must be associated with a gap, i.e. they move to an A-bar position. More specifically, they imply A-bar extraction to the specifier position of a functional projection to the left of T, as we can see in (7) from Barbosa (2009: 7): (7) Algoi lhe disseram algoi mas não sei o quê. something to.him say-PAST-3PL but not know-PRESENT-1SG DET what ‘They must have told him something, but I don’t know what.’ Meanwhile, unlike CLLD, Barbosa suggests that there are several pieces of evidence that English Topicalisation involves movement: it licenses parasitic gaps (PGs) and shows weak-cross-over (WCO) effects (see Duarte (1987) and Raposo (1997) for an analysis that is compatible with the one proposed here and captures these basic facts). However, Cecchetto (2000: 3) claims that CLLD shows island sensitivity, a fact that can be taken as an argument for a movement-based analysis: 57 (8) ??Gianni temo la possibilità che lo arrestito. Gianni, (I) fear-PRES-1SG the possibility that (they) CL arrest-SUBJ-3PL (9) *A Juan temo la posibilidad de que lo arresten. to Juan fear-PRESENT-1SG the possibility of that CL arrest-SUBJ-3PL ‘John, I fear the possibility that they punish him.’ (adapted from Cecchetto 2000: 3) The degradation of (8) is a clear example of illicit extraction from a complex NP. Nevertheless, (8) is not as ungrammatical as standard examples of strong island violations in Italian. In this connection, Cinque (1977) proposes that in Italian two constructions must be distinguished: CLLD and the hanging topic one. In a hanging topic sentence, island effects are not found. Since sentence (8) is ambiguous between a CLLD and a hanging topic reading, its marginal acceptability is due to the possibility of interpreting it as a hanging topic sentence. Interestingly, (9) suggests that in Spanish it is possible to distinguish hanging topic from CLLD sentences, since a personal direct object in a CLLD sentence is preceded by the preposition ‘a’ whereas a hanging topic DP is not. Therefore, strong island violations are only observed in the former case (cf. Escobar Alvarez 1995). This is not discussed in Barbosa’s article. So, we consider her argumentation insufficient to abandon the movement-analysis of CLLD. 2.3. Arguments supporting movement-analysis The question of movement vs. base-generation has been relevant throughout the history of generative discussions of CLLD. In this sense, we will argue for a movementbased analysis, suggesting that CLLD is derived by movement. Basically, it involves a left-dislocated phrase and a pronominal element resuming its reference somewhere lower in the structure. Such a pronoun is standardly taken to be a resumptive pronoun (RP). This is a property of Romance CLLD. 64 (24) a. THIS PROBLEM we resolved yesterday (not that one). b. ESTE PROBLEMA resolvimos ayer (no ese). In this way, we can conclude that English Topicalisation and Spanish CLLD involve movement, whereas English LD and Spanish HT do not. In the next section, we will study the nature of such a movement and the landing-site of these topic and focus elements that move in a clause. 2.4. Syntactic approaches to discourse functions So far, we have seen that some types of topics 19 and foci move, that is, they are not base-generated. Now, we provide a syntactic account of Focalisation and Topicalisation in minimalist terms. We have given special emphasis to the conditions which cause the movement of discourse elements. As we said at the beginning of this chapter, both operations involve movement to the initial part of the clause, but what is the specific landing-site of these constituents? Following ideas proposed by Grimshaw (1997), Topicalisation implies movement of a constituent to the left-most position within TP. Thus, the topicalised structure in (25) has the representation in (26): (25) This problem I can solve. (Jiménez-Fernández 2005: 277) 19 Since movement does not take place when dealing with hanging topics, we will restrict our syntactic analysis to dislocated topics. 65 (26) TP DP TP D T´ This problemi I T VP can V DP solve this problemi In (26), the DP this problem is moved from its initial object position to the left side of TP, leaving a copy behind. However, according to Jiménez-Fernández (2005), such an analysis presents some contradictions. Wh-movement involves moving the wh-phrase to the specifier of CP. Since topics adjoin to the left side of TP, wh-phrases should precede topicalised elements. Recall that CP dominates TP due to selectional constraints. This is true for relative clauses but not for wh-questions (Jiménez-Fernández 2005: 277): (27) a. John is the man to whom the prize they have given this year. b. *John is the man the prize to whom they have given this year. (28) a. The milk, where did you put it? b. *Where the milk did you put it? c. *Where did the milk you put it? The TP-adjunction analysis of Topicalisation establishes that TP should dominate CP, so that it does not account for the order ‘Topic+wh-phrase’. For this reason, Topicalisation cannot be viewed as adjunction to TP. 66 As for Focalisation, Chomsky (1977) holds that it implies movement of the focalised element to the specifier of CP. Hence, (29) should have the derivation in (30): (29) OUT OF THE ROOM I want you to go (not into the room). (30) CP PP C´ COMP TP OUT OF THE ROOMi D T´ T VP Ij D V´ V TP Ij want you to go out of the roomi Nevertheless, we have to face some problems that arise from this analysis. The main shortcoming is that in embedded that-clauses the focalised element should precede the complementiser that. This is not possible as shown by the example in (31): (31) a. I said that OUT OF THE ROOM I want you to go (not into the room). b. *I said OUT OF THE ROOM that I want you to go (not into the room). 67 In (31), the focus has to move to a position following the complementiser that. But, we have a problem since the only possible position is [Spec, T], which is already occupied by the subject I. Therefore, where does the focus constituent land? 2.4.1. A unified movement-analysis As far as CLLD is concerned, we have to talk about Boeckx’s (2003) theory of A’-dependencies. In this case, all chains involving a resumptive pronoun are taken to be the result of movement. Since in some languages resumptive chains can violate even strong islands, an immediate consequence of his approach is that movement can generally overcome islands, including strong ones, given the right conditions. More specifically, Boeckx suggests that there is no coherent notion of ‘island’, and movement is always unbounded. Apparent islands turn out to be constraints on Agree. However, if movement can be made dependent on Match but independent from Agree, then movement can escape an adjunct. Resumption is the tool that enables us to separate Move from Agree. Arguably, the seeming lack of coherence of the island notion has been replaced by an equally incoherent notion of ‘boundary for Agree’ plus an arbitrary distinction between Match and Agree. Boeckx’s proposal does not seem to be an option since HT is base-generated while CLLD implies movement – see the arguments presented in Chapter 2, Section 2.3. In this sense, Boeckx’s approach involves a theory of resumption and a theory of A’-movement. Let us discuss both of them: I) Resumption Following Boeckx, all forms of resumption imply a structure in which the resumptive takes the displaced element as a complement: 68 (32) DP D[def] NP ResPron Wh/Op Notice that this goes against a long tradition that takes wh-words to be determiners, not NPs. In fact, what the syntactic category of [wh] is left unanalyzed. The problem is that if [wh] is not D, a lot of interesting generalizations are lost, for example: determiners, pronouns, and wh-words in German show the same nominative, accusative, dative, and genitive case morphology while nouns only show genitive morphology. We can establish a generalization for German according to which D reflects case morphology in all four cases, but this generalization can be formulated only if [wh] is D. Assuming that the structure in (32) is the initial one, the wh/Op phrase raises, stranding the definite determiner as a resumptive pronoun. Movement must go through [Spec, D] as an escape hatch: (33) wh … [DP t(wh) [D´ D t(wh)]] Such an analysis raises two questions: why does not it violate the Left Branch Condition 20 ? Why is there no definiteness barrier 21 ? Boeckx’s answer for these two questions is as follows: whether one can or cannot extract from the specifier of a category depends on the feature composition of that category. More specifically, if D agrees with the extracted constituent, extraction is not possible. 20 Ross (1967) proposed the Left Branch Condition, which blocks movement of the leftmost constituent of an NP. So, in (33), if the wh/Op stops in [Spec, D], there should be a violation of such a condition. 21 Jiménez-Fernández (2009, 2012) analyses definiteness effects on the extractability of DPs. He claims that definite/specific DPs are phases and therefore show island effects. 69 Boeckx’s extension of his idea to clitic doubling is problematic. He cites two examples of clitic doubling with non-agreement between clitic and double: Sicilian (from Ledgeway 2000) and Spanish datives. Consider the Spanish dative, which he takes from Gutiérrez-Rexach (2000): (34) No le tiene miedo a las balas. NE CL.DAT.SG has fear DAT the bullets ‘He’s not afraid of bullets.’ In (34), it seems as if number agreement between clitic (le) and double (a las balas) were momentarily cancelled. In spite of this, there are no grounds for claiming there is no agreement between the dative clitic and the double. First, if the clitic is plural, the double must be plural: (35) *No les tiene miedo a la bala. 22 CL.DAT.PL DAT the bullet In (35), it seems that the form le has become unspecified for number. Second, there is [person] agreement: if we have a strong first or second person pronoun, the clitic must have the same [person] feature. Finally, in the first and second persons, number agreement is also obligatory: (36) a. No te tengo miedo a ti. neg CL-2SG have-1SG fear DAT you ‘I am not afraid of you.’ b. *No le tengo miedo a ti. CL-3SG DAT you c. *No te tengo miedo a vosotros. CL-2SG DAT you-PL 22 Villa-García (2009) claims that certain grammatical sentences in Spanish exhibit lack of agreement between the verb and the subject in terms of one of the three φ-features, namely person, number, and gender. So, he postulates a condition on the output of Agree which states that if optimal agreement fails to obtain, at most one feature can be left syntactically unvalued, and this remaining feature is handled by alternative mechanisms. 70 Besides, when we have an accusative clitic, we also have obligatory agreement between clitic and double. Therefore, Spanish is an argument against Boeckx’s proposals. II) Islands and the Principle of Unambiguous Chains Let us now move onto Boeckx’s theory of movement, islandhood, and the role of resumptives in overcoming islands. He takes the following Principle of Unambiguous Chains as a point of departure: (37) Principle of Unambiguous Chains (PUC): Chains cannot include more than one strong OCC 23 /EPP position. The PUC successfully rules out cases of Superraising: (38) *John seems t is intelligent. (38) is ungrammatical because the chain (John, t) includes two strong OCC/EPP positions, corresponding to the two instances of [Spec, T]. However, a subject wh-phrase should in principle create a problem, since there are two strong OCC positions: [Spec, T] and [Spec, C]. Boeckx suggests that there are two ways that a language can go to overcome this difficulty. The first way is to have T and C agree, so that they become a single unit. In this way, we do not have two strong OCC for one chain, but only one OCC for each chain. The other way is by means of resumption. In this case, the constituent is split in two blocks, one of which is involved in the A-dependency (the resumptive pronoun) so the other one is free to enter the A’-dependency (the wh-phrase or the dislocate). Hence, 23 OCC is short for occurrence, see Chomsky 2000. Its function is exactly the one that the EPP takes. 71 there are now two chains, not one, and the PUC is obeyed. Resumption is a consequence of PUC. Nevertheless, in languages such as Spanish chains can include more than one strong EPP position, as exemplified in (39): (39) Juan parece que es inteligentísimo. The second leg of his argument implies the trigger of movement. In this sense, Chomsky (2000) distinguishes between Match and Agree. If a probe finds features in a goal of the same type as those in the probe, then we can say that they Match. If, additionally, there is a transfer of features from goal to probe, then we have Agree (Chomsky 2000: 122, Boeckx 2003: 2). Chomsky proposes that Agree may trigger movement if the probe has an additional EPP feature. Instead, Boeckx claims that Match suffices to cause movement. Match occurs without limits. In particular, Match can target an adjunct, which has inert φ–features, and even probe inside it. Agree, however, is stricter, since it can only involve a goal with active φ–features (Boeckx 2003: 99-100). Now, we are ready to deal with the problem of islandhood and resumption. We first discuss Hebrew, a language in which resumptive chains are sensitive to no islands whatsoever (Boeckx 2003: 20): (40) Raʔiti ʔet ha-yeled ʔaser/se-ha-cayad harag ʔet ha-arie saw-PAST-1SG ACC the child C-the-hunter kill-PAST-3SG ACC the lion ʔaser/se-radaf ʔexarav. C-that after him ‘I saw the child that the hunter killed the lion that chased him.’ In accordance with Boeckx, a probe can enter an adjunct and Match its features against a goal. If the goal includes a resumptive pronoun, the Wh/Op does not need to satisfy 72 any Case/Agreement requirements, Agree between Wh/Op and probe is not necessary. Since Match is all that is required for movement, it can take place. What about Greek, a language in which resumptive chains are sensitive to strong islands only? (Boeckx 2003: 111): (41) Gnorisa mja gineka pu den ksero pjos tin pndreftike. meet-PAST-1SG a woman C NEG know-PRES-1SG who CL marry-PAST-3SG ‘I met a woman that I don’t know who married.’ (42) *Pira mia efimerida pu o Petros apokimithike eno tin got-PAST-1SG a paper.ACC C the Petros fall-asleep-PAST-3SG while CL diavaze. read-GEN-3SG ‘I got a paper that Petros fell asleep while reading (it).’ Boeckx’s solution is that Agree must be involved. His analysis of this type of example is detailed as following (2003: 112): (43) 1. Agr (in T or v) enters an A-dependency with the resumptive element. 2. The resumptive moves (cliticizes, or some such). 3. The probe establishes an Agree relationship with Wh/Op. 4. Wh/Op is extracted. Since extraction in this type of example involves Agree, it is subject to the type of restrictions that Agree is subject to: for concreteness, Agree cannot penetrate adjuncts, with the direct consequence that extraction out of adjuncts is not possible 24 . As evidence, Boeckx claims that the presence of a relative pronoun indicates the (abstract) presence of an agreeing complementizer. 24 In accordance with Stepanov (2007) and Chomsky (2008), adjuncts are strong islands for extraction. In this connection, Fábregas & Jiménez-Fernández (2012) analyse the conditions which determine extraction possibilities from fake adjuncts. More concretely, they state that the gerunds and depictive adjectives that permit extraction of one of their constituents are projected as PathP inside the first phase syntax that includes the main verb, with which they share the same syntactic space. 73 However, this solution seems to undermine Boeckx’s enterprise. His main idea is that resumption is a last resort operation, with the result that one chain that would have two strong OCC positions would become two chains with one strong OCC each. But in the Greek example we have a clitic that does not prevent the A’-dependency to require Agree – what is the role of the clitic then? Let us further analyze Boeckx’s discussion of Greek resumption. As we mentioned before, the PUC for subject extraction can be satisfied either by having T agree with C or by means of resumption. The probe in (42) must be the complementizer pu. According to Boeckx, pu has unvalued φ–features and can agree with Wh/Op in Greek even in the presence of a resumptive. The question is why can’t C agree with the T that it selects? If the φ–features of C are thus satisfied, it should be able to simply Match with the goal and overcome the island. There is a final group of languages (Vata, Serbo-Croatian) that are sensitive to all islands. Boeckx deals with this group by suggesting that complement clauses in these languages are actually relative clauses in disguise. This is Boeckx’s theory of islands and resumptives. Nevertheless, there is some evidence against it. Recall that Romance languages include HT as well as CLLD: (44) a. Pedro, hace tiempo que no lo invito a cenar. HT Pedro does time that NEG CL.ACC invite-PRES-1SG to dinner b. A Pedro hace tiempo que no lo invito CLLD to Pedro do-PRES-3SG time that NEG CL.ACC invite-PRES-1SG a cenar. to dinner ‘I have not invited Peter to dinner in a long time.’ Among the properties that tease them apart are, as listed previously, (i) connectivity effects, (ii) reconstruction effects, and (iii) islandhood. 80 The preposed element, bearing focal stress, introduces new information, while the rest of the sentence expresses given information which is shared by the speakers involved in the situation. Going back to Brody (1990, 1995), these articulations are realized through the usual process of syntactic representations: the X-bar schema. Let us observe the following structure: (58) TopP XP Top´ Top YP XP = topic YP = comment In (58) the Top head, which belongs to the COMP-system, projects its own X-bar schema. This Top head takes the topic as its specifier and the comment as its complement. In the same way, a Foc head projects a similar structure with a specific functional interpretation: its specifier is the focus, its complement is the presupposition: (59) FocP ZP Foc´ Foc WP ZP = Focus WP = Presupposition 81 Taking into account the Economy Principle, Focus and Topic movement should be considered as “last resort”. This type of movement must be triggered by the satisfaction of a specific criterion. In this sense, a constituent with a topic or focus feature must be in a Spec/Head configuration with Top or Foc, respectively. If there is no constituent with a topic or focus feature, the topic-focus system will not be present 29 . Consequently, if an element is topicalized it is moved to CP into the specifier of a functional projection known with the name of Topic Phrase. In this way, the moved element enters into a Spec-Head relation with the head of the Topic Phrase. The syntactic configuration is as follows: (60) a. This story I believe 30 . (Jiménez-Fernández 2005: 281) 29 In addition to Focus and Topic Phrases, CP is composed of two more projections, Force and Finiteness Phrase. A difference between the discursive categories of Topic and Focus and the functional categories of Force and Finiteness is that the former, but not the latter, expresses selectional restrictions between the COMP-system and the immediately higher and lower structural systems. For this reason, the order of these functional projections in the COMPsystem is something similar to (i), as proposed in Rizzi (1997: 288): (i) … Force … (Topic) … (Focus) … Fin TP 30 This is a partial derivation since we only consider the features [TOP] and [FOC]. The rest of morphosyntactic features are ignored. 82 b. TopP Spec Top´ this story [TOP] Top TP [TOP] Spec T´ I T VP D V´ I V DP believe this story Through a process of Topicalisation the DP this story is moved from its original object position to the specifier of a TopP, leaving a trace behind in VP. This movement is triggered by the Top head which attracts an element with the [TOP] feature. Since this CP is a phase, the c-command domain of its head C undergoes transfer to the phonological and semantic components. As a result, this domain is no longer accessible to further syntactic processes. Regarding Focalisation, we have a similar situation. The focalised element is moved into the left periphery of the clause, into the specifier of a functional projection which is called Focus Phrase. As in the former case, the moved element is in a local relation with the head of the Focus Phrase through the [FOC] feature. (61) a. NO OTHER TEACHER will I pay attention to. (Jiménez-Fernández, p.c.) 83 b. FocP DP Foc´ NO OTHER TEACHER [FOC] Foc TP will [FOC] D T´ I T VP will D V´ I V DP pay attention to no other teacher By a process of Focalisation, the focus head carries out the movement of that element in the sentence containing a [FOC] feature. Consequently, the DP object no other teacher is moved into the specifier of the Focus Phrase. Since the head Foc is strong, it attracts the auxiliary will 31 . Again, according to the Phase Impenetrability Condition (PIC), the domain of this head C (i.e. its TP complement) undergoes transfer to the phonological and semantic components to be assigned an appropriate phonetic and semantic interpretation respectively. As a result, that complement is no longer accessible to further syntactic processes from that point on. In this connection, Breul (2004: 32) holds that we may have movement of an XP into [Spec, Foc] not only in those cases where fronting is noticeable on the surface as in usual cases like 31 A strong head is defined as that head which attracts not only the grammatical information of a sentence constituent, but also the lexical material which shapes it (Radford 2004: 254). Instead, Breul (2004) illustrates the mechanism of deriving various constituent orders on the basis of α/γ–optionality of features. Nevertheless, such an optionality violates the Economy Principle. 84 (62) a. Beans he doesn’t like. b. BEANSi he doesn’t like beansi. but also in cases where fronting is not noticeable on the surface, as in (63) a. Peter doesn’t like beans. b. PETERi Peteri doesn’t like. Breul points out that (62) and (63) are associated with different types of focus structure. Breul argues that, in languages like English and German, every root clause is either categorical, identificational or thetic. The categorical/identificational/thetic distinction is conceived of as one in terms of focus structure. Each of the three types of focus structure is characterized by a specific property of its syntactic structure. The syntactic aspects of focus structure are summarised in the FocP-hypothesis. More exactly, he considers that for English the assumption that there are two distinct functional phrases TopP and FocP potentially dominating TP is problematic and not well motivated. He proposes alternatively that there is only one FocP dominating TP in categorial and identificational root clauses and that there is no CP. In whis way, a categorial sentence has a clause structure in which Foc contains [-foc]. In order for the derivation of the clause to converge, a correspondingly [-foc]-featured XP in the structure of TP has to move to [Spec, FocP] to check its [-foc] against its corresponding [-foc] in the Foc-head. This [-foc]-featured XP is the topic expression in terms of Rizzi (1997). In an identificational sentence a [+foc]-featured XP from the TP, the identificational focus expression, moves to [Spec, FocP] to check its [+foc] against the corresponding [+foc] in the Foc-head. In a thetic sentence there is no FocP above TP. At the same time, Breul analyses the relation between the syntactic manifestation of focus structure and the presence of an intonational focus constituent (i-focus). More precisely, 85 he highlights that at the point of the derivation where a phrase XP is assigned the [+/- foc]-feature, XP has to be licensed as i-focus by F-marking some item contained in it which is capable of F-projecting and thus F-marking XP. Breul suggests that topic and focus expressions do not target the same structural position universally. More precisely, TopP and FocP being potentially distinct in a language entails that there is a forth type of focus structure in that language, a type in which there is both a topic expression and an indentificational focus expression. However, we argued in Chapter 1, Section 1.6 that topic and focus differ in a series of aspects, which emphasize a different nature. Hence, Breul’s FocP-hypothesis is not valid. Additionally, we have to indicate that Emonds (2004, 2012) proposes the existence of a Discourse Shell, that is, a categorically unspecified projection that may immediately dominate only TPs specified as Discourse Projections. The specifier of this projection is proposed as the landing site for root movements like Focalisation and Topicalisation. More concretely, he claims that a construct of unlabeled Discourse Shells is the best way to analyse the left periphery of root clauses. He uses these Shells in place of certain clausal categories of other authors, i.e., the TopP and FocP phrases of Rizzi (1997) and papers using his framework. In this regard, Emonds’ Tense S Constraint holds that any trace of a fronted constituent in a Discourse Projection TP has its closer binder in the first Discouse Shell XP just above TP. As commented in Chapter 1, Section 1.8, discourse constituents undergo movement cyclically through different phases and the head C is a phase. Consequently, we will depart from Rizzi’s cartographic approach and adopt an unsplit CP analysis. In this way, in English discourse constituents would move to the specifier of CP. 86 2.8. CP/TP Crosslinguistic investigations have revealed that languages vary with respect to the systematic properties of syntactic reordering. In some cases, the word order of clauses is not determined by syntactic conditions such as Case or agreement, but rather by structural conditions that are closely related to the thematic role of the arguments of the verb. In fact, languages differ in a number of respects, which highlight a fundamentally different nature. So, taking into account the Generalized TP analysis of Zubizarreta (1998), we can say that fronted topics in Spanish have [Spec, T] as their landing site. More specifically, Gross & Bok-Bennema (1986), Gutiérrez-Bravo (2007) and JiménezFernández (2010), among others, state that [Spec, T] is a multifunctional position in Spanish, which can be occupied by preverbal subjects, but also by foci, interrogative operators and topics. In this sense, EPP is a purely structural condition that requires some specifier position to be filled, independently of the category or grammatical relation of the constituent that fills it (Branigan 1992, Jonas & Bobaljik 1993, Babyonyshev 1996, Grimshaw 1997, Fernández-Soriano 1999, Chomsky 2000, Holmberg 2005). Yiddish (as analysed in Diesing 1990 and Santorini 1992) is like Spanish in that subjects, topics, and wh-operators all have the specifier of TP as their landing site. Outside the Indo-European family, Finnish, as analysed in Holmberg & Nikanne (2002), is similarly a language where there is an active EPP requirement that is satisfied in the specifier position of TP, and which can be satisfied by both preverbal subjects and fronted topics. Similarly, Jiménez-Fernández & İşsever (2012) claim that subjects, topics and foci undergo movement to the specifier of TP in languages such as Turkish. Following Miyagawa (2005, 2010), in Japanese topic and focus constituents are also 87 preposed to the specifier of TP. In this connection, addressing the cross-linguistic applicability of the multiple-specifier analysis requires a detailed and careful investigation of each relevant case that will be undertaken in the next chapter. 88 2.9. Concluding remarks From the preceding considerations we can draw the following specific conclusions: - We argue for a movement-based analysis, suggesting that topic fronting and CLLD are derived by movement. In the same way, we assume a movement-based analysis for focus fronting in English and Spanish. Evidence for such an analysis comes from factors such as weak-cross-over or intonational break. - Following the cartographic approach, Topicalisation involves the articulation in topic and comment. Analogously, Focalisation leads us to a division between focus and presupposition. These articulations are realized through the usual process of syntactic representations: the X-bar schema. Taking into account the Economy Principle, Focus and Topic movement should be considered as “last resort”. This type of movement must be triggered by the satisfaction of a specific criterion. This topicalised/focalised element moves to CP in the left periphery of the clause. More specifically, it lands in the specifier of a functional projection called TopP/FocP. So, the moved element is in a Spec-Head relation with the head Top/Foc through the feature [TOP]/[FOC]. Nevertheless, we will depart from Rizzi’s cartographic approach and adopt an unsplit CP analysis. In this way, in English discourse constituents would move to the specifier of CP. - Languages vary with respect to the systematic properties of syntactic reordering. In some cases, the word order of clauses is not determined by syntactic conditions, but rather by structural conditions related to the thematic role of the arguments of the verb. 89 In this connection, [Spec, T] is a multifunctional position in Spanish, which can be occupied by preverbal subjects, but also by foci, interrogative operators and topics. Additionally, some languages such as Yiddish or Finnish are like Spanish in that subjects, topics, and wh-operators all have the specifier of TP as their landing site. 96 referential expression a cada niño is bound. This situation leads us to think that the anaphor has undergone A-movement. Observe that the binding relation between the anaphor and its antecedent has changed after the movement of su to the beginning of the sentence. This movement has made the sentence ungrammatical. Nevertheless, reconstruction effects are obtained in some cases, which is indicative of A’-movement. If the relationship between binder and bindee is reversed no asymmetry is established; i.e., binding is possible with both orderings, as can be seen in (7) from Jiménez-Fernández (p.c.) 32 : (7) a. Ángelai puso sui chaqueta en el armario. Angela put-PAST-3SG her jacket in the closet b. [Sui chaqueta]j la puso Ángelai en el armario [sui chaqueta]j. her jacket CL put-PAST-3SG Angela in the closet ‘Angela put her jacket in the closet.’ The solution proposed by Jiménez-Fernández & İşsever (2012) is based on the (LF-)adjunction of anaphoric/pronominal features to functional heads, namely v0 and T0. In line with Kural (1992), Zubizarreta (1998), Lahousse (2009) and Williams (2009), they state that the interaction of focus and binding is fundamental to predict the interpretive properties of displaced constituents. 32 Although the classical approach to this phenomenon suggests that topic movement can target both [Spec, T] and [Spec, C] in a language (Mahajan 1990; Miyagawa 2003; among others), Aand A’-positions respectively, there are also studies proposing a unified approach to A/A’ distinction. According to Chomsky’s (1993, and subsequent work) ‘copy theory of movement’, Saito (2003) suggests that movement can be accounted for as a uniform operation, which is based on the selectional interpretation of lexical features. Following such an approach, all cases of movement are to a position where A-binding is possible. He also claims that movement is subject to reconstruction. Reconstruction from an A-position, however, contradicts much of the proposals previously made. Saito nevertheless claims that this problem can be solved under the copy theory of movement. In this way, each copy in a copy chain is interpreted derivationally. More concretely, all features of an item undergoing movement are copied into the next target position in the derivation and later deletion applies on the relevant copy. Nevertheless, the fact that some types of A-movement undergo LF-reconstruction cannot be taken as evidence for a uniform movement-analysis. 97 As we will see in Section 3.4, in line with Chomsky (1998) and Miyagawa (2005, 2010), Jiménez-Fernández (2009, 2010) argues for a three-fold classification of languages based on feature inheritance. In this system, languages are parameterized as to which features of C, agreement-(φ) and/or discourse-features (δ), are inherited by T. In particular, if topics move to [Spec, T], the A-properties are fully expected. Observe the derivation of (5b), depicted in (8), where the topicalised antecedent object can bind the anaphoric feature in T0 associated with the in situ subject. Note that the first instance of object movement to the outer specifier of v0 is forced by p-movement (Zubizarreta 1998) so that focus is assigned to the subject (Jiménez-Fernández & İşsever’s (2012: 2)): (8) TP O-anti [+δ] T’ [+anp] + T0 φ/δ vP O-anti [+δ] v’ Si [+anp] [+foc] v’ v0 … At the same time, Jiménez-Fernández & İşsever (2012: 2) state that A’- properties are also possible when the antecedent is a defocused subject as shown in (7ab). In the first step in (9), the anaphoric feature of the object anaphor adjoins to v0 since it is the next higher functional head to which this feature can adjoin. This is followed by p-movement of the object to the outer specifier of v0 in Step 2. Hence, the binding relation between the antecedent subject and the anaphoric feature in v0 is established in 98 the vP domain. The subsequent movement of the object anaphor to [Spec, T] in Step 3 is drawn by its need to check its topic feature and does not alter the binding relations already established in vP. (9) [TP Oi [+top] [vP Oi [+top] [v’ S-anti [vº v0 + [+anaphor]] [VP … O[+anp] [+top] … ]]. 3 1 2 Summarizing, this type of movement allows an anaphor to be c-commanded by its antecedent in its local domain. Besides, the variation between examples without reconstruction and examples with reconstruction reduces to a single difference: the stage in the derivation at which binding conditions are satisfied. In other words, binding interpretations can be established either in the CP phase or in the vP phase depending on which category the anaphoric feature is attached to. Interestingly, Jiménez-Fernández & İşsever’s analysis is also applicable to English as well. However, discourse movement in English is best explained in terms of A’-movement. According to Gutiérrez-Bravo (2007), in English the priority is that the subject receives nominative case in [Spec, T]. Hence, the EPP is always satisfied by the subject. In other words, the specifier of TP is not an available landing site for discourse constituents, and therefore, topic constituents must move to CP. (10) [CP Oi [+top] [TP S-anti [vP Oi [+top] [v’ S-anti [vº v0 + [+anaphor]] [VP … O[+anp] [+top] … ]]. 3 1 2 99 3.3.2. Evidence for A´-movement analysis As we saw in section one, discourse movement can also be non-argumental. According to Zagona (2002), discourse elements can move to the left periphery of the clause through A´-movement. As stated in the former subsection, there are clear-cut characteristics which define this type of movement. First, when an element moves to CP, it can reconstruct for the purposes of binding. As a result, the moved element does not violate either the (A) or (C) conditions of Binding Theory. This is confirmed by (11): (11) Himi friends of Johni introduced himi to Mary. (Bailyn 2003: 162) In (11), binding is fully acceptable since the preposed item can reconstruct to its base position. In this way, him can be bound by its antecedent and at the same time the referential noun John is not bound. This situation provides evidence to suggest that the preposed element has carried out A´-movement. Besides, parasitic gaps is normally taken to be a diagnostic of A’-movement. Parasitic gaps involve an invisible formative which did nothing but license otherwise illicit movement steps or selection relations. In this sense, parasitic gaps are licensed only by A’-movement, if the trace (t) does not c-command the parasitic gap (e) and the A’-moved element c-commands the parasitic gap as well as its trace. Let us consider the following example (adapted from Alexopoulou & Kolliakou 2002: 205): (12) The paper we filed t before we could sign e. In (12), the gap is parasitic on the trace of the DP the paper and the sentence is grammatical because the DP-trace does not c-command the parasitic gap. In other 100 words, if the fronted element the paper can license a parasitic gap then it must have undergone A´-movement (movement to a “non-argumental” position). Basically, discourse movement can allow reconstruction in order not to violate the conditions (A) and (C) of Binding Theory. Moreover, parasitic gap licensing is possible in English. Hence, these arguments highlight the idea that discourse movement can also be non-argumental. As mentioned in the previous section, A’-properties in Spanish can be derived by assuming Jiménez-Fernández & İşsever’s (2012) analysis. In other words, if binding interpretations are established in the CP phase, A’-properties are fully expected. Nevertheless, in the following section we will see that discourse movement in Spanish is of an argumental nature. 3.4. Is Focus Aor A’-movement? A feature-based analysis As commented in Chapter 2, Section 2.7, crosslinguistic investigations have revealed that languages vary with respect to the systematic properties of syntactic reordering. This section analyses how focus-related/agreement features interact with the EPP feature in the process of Agree in the Minimalist Program. Chomsky (2001, 2006, 2008) holds that uninterpretable features enter the derivation in phasal heads and by a process of feature inheritance, they are lowered onto the next head. This lowering process only affects φ-features. However, Miyagawa (2005) assumes that the phasal head C has both agreement and focus-related features. In this connection, he states that Focus and agreement are commonly thought to be located on different heads: the focus element is on the FocP in the region of C and agreement on T. Nevertheless, in accordance with Chomsky (2001), agreement is associated with a higher head than T, that is, C. Let us observe the following example (from Miyagawa 2005: 5): 101 (13) a. [[ei geçeen yaz ada-da ben-i gör-en] kiş i-leri] last summer island-Loc I-ACC see-(y)An person-PL ‘the people who saw me on the island last summer.’ b. [[pro geçen yaz ada-da ei gör-düğ-üm] kiş i-leri] last summer island-Loc see-DIK-1SG person-PL ‘the people who(m) I saw on the island last summer.’ In (13), the agreement initially appears on C and gets copied onto T, as seen in (13b) above. If the subject appears in [Spec, C], however, the agreement on C takes this subject, and the agreement is not inherited by T. This is what we see in (13a). Taking into account that the EPP is on T, we have the syntactic representations below: (14) Focus CP C’ TP CAGREEMENT FOCUS FEATURE INHERITANCE TEPP (15) Agreement CP C’ TP CAGREEMENT FOCUS FEATURE INHERITANCE TEPP 102 As it turns out, the focus/agreement feature works in conjunction with EPP in order to value the relevant uninterpretable features and to move the category agreed with. In this connection, Miyagawa (2005, 2010) sees focus and agreement as constituting the two polarities of a parametric variation 33 . Consequently, a language can be classified according to whether it is focus or agreement prominent. For instance, he states that Japanese is a focus prominent language since it does not have any over agreement. However, drawing on Chomsky‘s (2001) Uniformity Principle, all languages instantiate both features in some way, although they differ in the specific type of feature that they highlight. In fact, Jiménez-Fernández (2008, 2010) holds that Spanish can be considered as agreement and focus prominent. Firstly, Spanish overtly marks subject/verb concord. As a result, the subject moves to the specifier of TP and the verb into T; these movements result in an SVO word order, as illustrated in (16): (16) Pedro lavó el coche. Pedro wash-PAST-3SG the car ‘Pedro washed the car.’ 33 Miyagawa’s analysis is an extension to Kuroda’s (1988) work. He holds that some languages such as Japanese differ from English due to the lack of ‘forced agreement’. 103 (17) CP C TP DP T’ Pedro φ T EPP, φ vP lavó DP v’ Pedro v+V VP lavó V DP lavó el coche As we have already seen, T has an EPP, which in combination with the φ-features inherited from C, attracts the subject. Thus, we can conclude that Spanish is an agreement prominent language. Secondly, as discussed in length by Jiménez-Fernández (2008, 2010), Spanish is a free word order language since sentence constituents can appear in many different positions. Such a situation implies that phrases other than the subject DP can satisfy the EPP. In this sense, GutiérrezBravo (2007), Holmberg (2005) and Grimshaw (1997) suggest that the EPP is a purely syntactic restriction that requires some specifier position to be filled, regardless of the grammatical relation of the element that fills it. Accordingly, if the subject agrees with the verb and does not undergo movement, something else moves into the specifier of TP. In fact, as commented in Section 3.2, all positions that allow a topic interpretation also allow an interpretation as focus. Such a situation involves a double dissociation between position and interpretation. Notwithstanding, as we mentioned in Chapter 1, Section 1.2, we have to analyze the context in which the varying orders are felicitous. Depending on the 104 context, only some orders are acceptable. Indeed, grammatical rules about discourse movement have been developed to explain this contextual dependency. To put it explicitly, the derivation of (18) is provided in (19): (18) EL COCHE lavó Pedro (no la bicicleta). (19) CP C TP DP T’ El cocheFoc T EPP, Foc, φ vP lavó DP vP el coche DP v´ Pedroφ v+V VP lavó V DP lavó el coche In (19), we have an identificationally focused sentence where the DP el coche has entered the Numeration with an interpretable [FOC] feature. The EPP of T in conjunction with the unspecified unvalued discourse feature inherited from C probes the suitable goal el coche. In this case, the [FOC] feature of T is valued through the operation of Agree. Such a feature is deleted in the process of Transfer since this is uninterpretable. Some properties corroborate this analysis. Let us begin with the binding theory. In this sense, note that if focus elements can be preposed to an A-position then we 105 expect them to be able to serve as antecedents to a reflexive in subject position. This prediction is borne out by (20) and (21) below: (20) *Sui hijo echó a Juani de la casa his son throw out-PAST-3SG to Juan of the house ‘His son threw Juan out of the house.’ (21) A JUANi echó sui hijo de la casa (no a Pedro). to Juan throw out-PAST-3SG his son of the house not to Pedro (=25) Reflexive binding in (21) can only be possible if the focus element a Juan is in an Aposition. Therefore, we can say that focus movement is argumental. A further diagnostic commonly used to determine whether movement is argumental or non-argumental relates to the phenomenon of weak-cross-over. According to Mahajan (1990), ‘to be construed as a bound variable, a pronoun must be c-commanded by a binder and its variable’. More concretely, a pronoun that is not ccommanded by a binder at s-structure cannot be construed as a bound variable. Consider the following example: (22) A PEDROi vio sui madre (no a Juan) 34 . to Pedro see-PAST-3SG his mother not to Juan ‘To Pedro, his mother saw (not to Juan).’ In (22), we do not get WCO effects since the focus element a Pedro c-commands the pronoun his at s-structure. Hence, it is obvious that focus movement is an A-movement. In addition, certain properties of parasitic gap constructions corroborate our claim that focus movement in Spanish is of an A-nature. As discussed in Section 3.3.2, parasitic gaps are licensed only by A’-movement, if the trace does not c-command the 34 This example was accepted by eight out of our ten informants with the intended focus interpretation. 112 4.1. Introduction Once we have defined the argumental/non-argumental nature of discourse movement, we are going to analyse the differences between English and Spanish with respect to the application of the operations of Topicalisation and Focalisation in main clauses and its possible extension to subordinate contexts. With this in mind, we will analyse pioneering works on factivity (Hooper & Thompson 1973), which claim that English Topicalisation and Focalisation are possible in the complement clauses of nonfactive predicates and of the semantic class of the main predicate: 35 (1) I exclaimed that NEVER IN MY LIFE had I seen such a crowd. (A) (H&T (43)) (2) I think that this book he read thoroughly. (B) (3) I found out that NEVER BEFORE had he had to borrow money. (E) (H&T (119)) (4) *It’s likely that SELDOM did he drive that car. (C) (H&T (96)) (5) *He was surprised that NEVER IN MY LIFE had I seen a hippopotamus. (D) (H&T (103)) These differences are studied by Haegeman & Ürögdi (2010). They propose that factive complements are derived via leftward movement of a TP-internal clause-typing operator to the left periphery of the clause. Under this analysis their incompatibility with Topicalisation and Focalisation will be due to intervention. At the same time, JiménezFernández & Miyagawa (forthcoming) suggest that such complement clauses are compatible with CLLD in Spanish –illustrated in (6b)– and Jiménez-Fernández & Camacho-Taboada (2014) argue that Spanish focus fronting is also compatible with all 35 In some cases, discourse movement is unacceptable in the complement clauses of non-factive predicates. As we will observe in Section 4.4.5, this is predicted if we assume Miyagawa’s idea that English is an agreement-prominent language, since in that case discourse is not so frequently reflected in the syntax of the language. On the contrary, Spanish is a discourseprominent language, and therefore, it is less restrictive than English. Additionally, in Section 4.4.5, we will suggest that the complement of non-factive predicates may be asserted or nonasserted. In this way, the absence of assertion would explain why non-factive verbs may not be compatible with focus/topic fronting. 113 types of subordinate clauses, regardless of the class of matrix predicate, as shown in (6a): (6) a. Es probable que SOLO ALGUNA VEZ haya conducido be-PRES-3SG probable that only sometimes have-PERF-3SG driven Juan ese coche. (Class C) Juan that car ‘It is probable that Juan has driven that car only sometimes.’ b. Ángela estaba sorprendida de que los regalos los hubieran Angela be-PAST-3SG surprised that the prizes CL have-PAST-PERF-3PL dejado los Reyes Magos debajo del árbol. (Class D) left the Reyes Magos under the tree ‘Angela was surprised that the Reyes Magos had left the prizes under the tree.’ The purpose of our study is to find out the reasons that explain this parametric variation in light of the different syntactic positions that the languages use for the topic and focus elements. 4.2. Factive clauses and discourse movement The notion of factivity is normally defined in terms of the truth value of the propositions expressed by sentential complements. Factive verbs are thus identified on the basis of the behaviour of their declarative that-clause complements. More concretely, as originally noted by Kiparsky & Kiparsky (1970), factive predicates are distinguished from non-factives in that the former presuppose the truth of their complement clause while the latter do not. Since Emond’s (1970) dissertation, it has been acknowledged that the so called Main Clause Phenomena (MCP) are restricted to occurring in root clauses and a limited subset of subordinate clauses. In this sense, following Hooper & Thompson (1973), we might think that factive clauses lack a left periphery ‘space’ entirely, but this option is problematic since factive clauses are compatible with adjuncts appearing to the left of 114 the subject. In (7a), last week precedes the canonical subject position. If, as proposed by, among others, Rizzi (1997), Aboh (2004), and Endo (2007), fronted adjuncts are topicalised, then (7a) implies that factive clauses are compatible with a Topicalisation, yielding the question why (7b) is not grammatical (Maki et al. 1999: 3): (7) a. John regrets that last week Mary did not turn up for the lecture. b. *John regrets that this book Mary read. Further support for the hypothesis that factive clauses do have a left peripheral space comes from the Romance languages, in which CLLD, which is also commonly considered to be associated with a left peripheral topic projection (Rizzi 1997), is available in factive clauses. This is traced back to Cinque (1990): (8) Es extraño que este problema no hayan sido capaces de be-PRES-3SG strange that this problem haven’t-PERF-3PL been able to resolverlo los estudiantes. resolve-INF-CL the students ‘It’s strange that the students haven’t been able to resolve this problem.’ In the same direction and within cartography, Haegeman (2003, 2006) suggests that subordinate clauses have a truncated structure since Force does not project and for this reason Topic and Focus do not project either. However, this account is not uncontroversial. In fact, imperatives, which are said to be associated with illocutionary force, do not allow Topicalisation in English (Haegeman 2010: 6): (9) Your essay leave *(it) in my pigeon hole this afternoon. As an alternative, in Chapter 1, Section 1.4, we anticipated that for Hooper & Thompson (1973) MCP depend on assertion. Assertion is a property of declarative root clauses; in order to be compatible with MCP, embedded clauses must be asserted, i.e., 115 non-presupposed. Nevertheless, it is assumed that complements of factive predicates resist MCP in English so we would have to assume that they are not assertive in the relevant sense. But Zubizarreta (2001: 201) comments ‘it is likely that factive predicates, which presuppose the truth of their propositional complement, contain an Ass(ertion) operator in its CP.’ Therefore, this account is not valid. Meanwhile, Breul (2004) emphasizes that the generation of MCP in subordinate contexts is syntactically free, but constrained by purely pragmatic principles acting as filters on the syntactic output. Thus, while it is syntactically possible to generate a FocP in any embedded clause, a FocP should be generated only if the embedded clause actually does manifest focus structure. Though pragmatic factors may influence on the availability of discourse categories, we are going to stick to a syntax-based explanation. As we will see in next section, intervention effects play a crucial role. 4.3. Intervention effects A movement analysis of factive clauses allows us to analyse the incompatibility of MCP with factivity in terms of an intervention effect. More specifically, Munsat (1986) argues that factive clauses contain an event operator that moves to C (see also Melvold 1986, 1991; Hegarty 1992; Watanabe 1993, 1996; Roussou 1994; Bianchi 2000; Zubizarreta 2001; Hiraiwa 2010; among many others). Using a previous proposal in Haegeman (2007), Haegeman & Ürögdi (2010) argue that this operator movement is what causes an intervention effect in factive clauses, leading to blocking of MCP. As we can see in (10) (Haegeman & Ürögdi 2010: 128): (10) [CP OPi … [FP ti [TP V… ]]] 116 This operator movement to [Spec, C] blocks anything else from moving to this position. As a matter of fact, we will support this general approach of using syntactic intervention to account for the absence of MCP in factive clauses. The question arises as to whether there is any independent evidence for the movement derivation of factive complements. Suggestive empirical support comes from languages in which factive complement clauses look superficially similar to relative clauses. So, for example, Collins (1994) and Aboh (2005) discuss the derivation of factive complements in Gungbe. As shown in (11) the internal syntax of complements of factive predicates is similar –though not identical– to that of relative clauses. (11a), from (Aboh 2005: 266, (4)), illustrates relative clause formation; (11b) from (Aboh 2005: 266, (2c)), illustrates a clausal complement of a factive predicate. Observe crucially that the same relative determiner dĕ is instantiated. (11c) (Aboh 2005: 279, (29a)) is a variant of (11b): rather tan instantiating relativization through the movement of a DP, relativization is achieved through V-fronting, coupled with doubling of the verb: (11) a. [Àgásá dàxó lò lε [dĕ mí wlé] ] vε ná Kòfi crab big Det Num that[REL] 1PL catch-PAST-1PL hurt-PAST-3SG for Kofi ‘The fact that we caught the aforementioned big crabs hurt Kofi.’ *‘The aforementioned big crabs that we caught hurt Kofi.’ b. Kòfi wε xò [àgásá dàxó [dĕ mí wlé] lò lε ] Kofi FOC buy-PAST-3SG crab big that[REL] 1PL catch-PAST-1PL Det Num ‘Kofi bought the (aforementioned) big crabs that we caught.’ c. [Wlé [dĕ mí wlé àgásá dàxó lò lε]] catch-PAST-1PL that[REL] 1PL catch-PAST-1PL crab big Det Num vε ná Kòfi hurt-PAST-3SG for Kofi ‘The fact that we CAUGHT the (aforementioned) big crubs for Kofi.’ In this respect, Edmonds (1976: 40) writes: ‘there is evidence that all the root transformations that front phrasal constituents without inducing comma intonation are 117 substitutions for the sentence-initial COMP node’. In this way, if factive clauses are derived by operator movement to COMP, then they are predicted to be incompatible with MCP that also imply movement to COMP since the fronted operator in COMP will preclude any additional movement. At the same time, in an articulated CP along the lines of Rizzi (1997) the impossibility of various fronting operations to the left periphery can no longer be derived by the fact that there is only one landing site available. Instead, it must be related to intervention effects related to multiple movements. 4.3.1. Embedded Topicalisation and Focalisation in English In the literature it has often been observed that English Topicalisation is not compatible with factivity. In other words, factive complements are expected to resist Topicalisation, while non-factive complements admit it. Let us observe the following examples: (12) a. The researcher explained that each part he had examined very carefully. (adapted from Hooper & Thompson 1973:474) b. *Peter resents that this book John read. (adapted from Maki et al 1999: 3) A movement analysis of factive clauses allows us to analyse this restriction in terms of an intervention effect. As already mentioned, in English the priority is that the subject receives nominative case in [Spec, T]. Thus, the specifier of TP is not an available landing site for discourse constituents. Instead, we suggest that topics and foci have to move to CP since focus features are not lowered from C to T. Recall that English is an agreement-prominent language. Furthermore, as we said in Chapter 1, CP is a phase and the domain of the head of a phase is spelled out at the end of a phase. Thus, the TP in the embedded clause, which is the complement of the embedded CP, is sent to the 118 phonological and semantic components at that point. Consequently, this language has a relatively strict word order. To be more precise, in (12a), there is no operator, and consequently, the topic element each part can move freely to [Spec, C]. On the contrary, in (12b), since the factive operator targets CP, it will block topic fronting, which also lands in CP. In this case, the topic constituent this book would have to move across the factive operator. The result would be ill formed: (13) a. [CP each parti [TP he [VP had examined each parti very carefully]]]]]. b. [CP OPi [FP OPi [TP John [VP read this book]]]]]. However, Bianchi & Frascarelli (2010) observe that not all factive clauses are incompatible with argument fronting, casting doubt on the validity of the canonical judgements shown in (12b). In this connection, Haegeman & Ürögdi (2010) argue that the received judgements in (12b) do arise in a neutral context, non-contrastive context, while potential counterexamples require specific contexts and the very presence of such contexts creates the necessary licensing conditions. As a matter of fact, examples like (14) pose a potential problem for the canonized pattern: (14) a. His parents resented that the maths exam he had not passed, and the biology exam he had not even taken. (Haegeman & Ürögdi 2010: 130) b. The entire office resented that Bill she had fired, and John she had decided to promote. (Haegeman & Ürögdi 2010: 130) Observe that both examples involve a contrast between two events. In (14a) the entire event of ‘not passing the maths exam’ must be contrasted with another event (in this case, ‘not even taking the biology exam’) and in (14b) ‘firing Tom’ is contrasted with 119 ‘promoting John’. We think that this is not a coincidence, as the contrast on the events is required for the examples to be felicitious. By virtue of being contrasted with another event, such an event is now part of a reference set and is thus D-linked in a way that it is not in the unmarked case. In other words, the contrast is actually not encoded on the fronted argument itself but rather on the event and – by virtue of the movement of the factive operator to CP – on the entire clause. While the relation between Topicalisation and factive clauses has received considerable attention in the generative literature, the compatibility of Focalisation with factivity has not been examined in such a great detail. Interestingly, we must note that in English focus movement is permitted with non-factive verbs (15a) but not with factives (15b). (15) a. We are sure that MEDICINE Peter will study next year (not psychology). b. *I am surprised that MATHS John has passed (not chemistry). In (15), the DPs Peter and John undergo movement to [Spec, T]. In this way, these DPs receive nominative case. Therefore, we claim that foci move to CP since focus features are not lowered from C to T in English. Later, following the Phase Impenetrability Condition (PIC), since this CP is a phase, the c-command domain of its head C undergoes transfer to the phonological and semantic components. As a result, this domain is no longer accessible to further syntactic processes. More explicitly, in (15a), there is no operator, and therefore, the focus element medicine can move to CP. In (15b), however, we can see how the focus element Maths gives rise to intervention effects. More concretely, this clause is derived via operator movement into [Spec, C], which accounts for the incompatibility of Focalisation and factive clauses in terms of intervention effects. In particular, since the factive operator is hosted by CP, it will block focus fronting, which also involves CP. In this case, the focused constituent 120 Maths would have to move across the factive operator. Such a result would be deviant. Therefore, English factive clauses are incompatible with Focalisation: (16) a. [CP MEDICINEi [TP Peter [VP will study medicinei next year]]]]]. b. [CP OPi [FP OPi [TP John [VP has passed MATHS]]]]]. 4.3.2. Embedded CLLD and Focalisation in Spanish There are crucial contrasts between Spanish and English, and therefore the mechanism we assume for English Topicalisation cannot simply be transposed to Spanish. Indeed, Spanish CLLD does not give rise to the same intervention effects as English Topicalisation and it is allowed in factive clauses. (17) Me alegra que esta canción la cantaras en la fiesta. CL glad-PRES-3SG that this song CL sing-PAST-2SG in the party ‘I am glad that you have sung this song in the party.’ According to Haegeman & Ürögdi (2010), the reasons for why Romance CLLD is allowed and English Topicalisation is disallowed in factive clauses are unclear. They suggest two possibilities: CLLD does not imply contrastivity or the DPs in the left periphery are actually base-generated there. However, they don’t elaborate any further and leave this issue aside. In this connection, we argued in Chapter 2 that CLLD is the result of movement. More precisely, we can use some of those arguments to justify that CLLD also undergo movement in factive clauses: 121 I) Island sensitivity As we have already said, a classic argument for movement is island sensitivity. Under traditional conception, it is unlikely that a dependency is the result of movement if it may span across an island; in turn, if it shows island sensitivity, it may have arisen from movement. Indeed, in factives CLLD shows island sensitivity, a fact that is standardly assumed to indicate the occurrence of a movement (adapted from Aboh 2010: 21): (18) *Lamento que a Pedro haya sólo un niño que regret-PRES-1SG that to Pedro there be-PRES-3SG only a child that lo invite a la fiesta 36 . CL invite-PRES-3SG to the party ‘I regret that there is only one child that invites Pedro to the party.’ If we take islandhood to be a property of movement, then the example in (18) would be evidence that CLLD is derived by movement in factive clauses. II) Condition A In this case, only when an anaphor is inside CLLD, it can be coreferential with a lower R-expression. Thus, the lack of Condition A effects supports the idea that CLLD involves movement in factives: (19) Me sorprende que a las amigas de sui madre no las CL surprise-PRES-3SG that to the friends of his mother, not CL salude Juani. say hello-PRES-3S Juan ‘I am surprised that Juan do not say hello to his mother’s friends.’ 36 This example was considered unacceptable by seven out of our ten informants. 128 4.4. Empirical evidence 37 4.4.1. Objective My goal in this section is to analyze the differences between English and Spanish with respect to the application of the operations of Topicalisation and Focalisation to main clauses and its possible extension to subordinate clauses, bearing in mind the distinction assumed here between factive and non-factive clauses (Hooper & Thompson 1973). This task will help us find out the reasons that explain the parametric variation detected in the two languages in light of the different syntactic positions that the languages use for the topic and focus elements. 4.4.2. Hypothesis We suggest that discourse movement implies different landing-sites in English and Spanish. Consequently, both languages interact with factivity in a different way. To be more precise, discourse movement is more constrained in English factive clauses than in Spanish ones. This restriction is due to intervention effects and the distinct syntactic positions used in each language. 4.4.3. Methodology To test the relation between Topicalisation/Focalisation and factivity in English, a series of written dialogues were administered to ten English native speakers. All of them were students at the University of Seville during the first semester of the academic year 2013/2014. Similarly, to test the relation between Topicalisation/Focalisation and factivity in Spanish, a series of written dialogues were presented to ten Spanish native 37 We wish to thank our informants for judgements and suggestions. 129 speakers. All these informants have studied English Philology at the University of Seville and hence a relatively good knowledge of language is presupposed. Additionally, to make discourse movement more natural, we included a context sentence before the critical sentence. The aim of this context sentence was to introduce a referent for the fronted constituent at the beginning of the embedded sentence. Informants were asked to rate the syntactic processes in the sentences, depending on how natural the sentences sounded to them in the relevant context. 4.4.4. Tests In this section, we present the two tests that we have used in order to study factivity in English and Spanish. In this sense, sentences are randomized so that informants are not influenced with regard to their judgements. Specifically, the first test is concerned with English Topicalisation/Focalisation in factive clauses. In this case, sentences 1, 3, 8, 9, 13, 18, 25, 27, 28, 31, 32, 34, 41, 44 and 46 combine a non-factive predicate with a topicalised constituent. Meanwhile, in sentences 5, 7, 14, 15, 19, 21, 24, 37, 38 and 48, it is a factive predicate that interacts with a topic. By contrast, in sentences 2, 6, 10, 11, 12, 16, 17, 20, 22, 23, 33, 35, 36, 42 and 43, we put together a non-factive predicate and a focused constituent. Finally, in sentences 4, 26, 29, 30, 39, 40, 45, 47, 49 and 50, we can find a factive-predicate with a focus. The second test deals with Spanish CLLD/Focalisation in factive clauses. More explicitly, in sentences 4, 6, 13, 17, 19, 24, 26, 29, 30, 32, 35, 36, 40, 45 and 48, we put together a non-factive predicate and a topicalised constituent. In the same way, sentences 10, 12, 14, 18, 21, 23, 25, 42, 43 and 50 combine a factive predicate with a topic. On the contrary, in sentences 2, 5, 7, 9, 11, 15, 16, 22, 27, 37, 39, 41, 44, 46 and 130 47, a non-factive predicate interacts with a focused constituent. Similarly, in sentences 1, 3, 8, 20, 28, 31, 33, 34, 38 and 49, we can observe a factive-predicate with a focus. 131 TEST on English Topicalisation/Focalisation in factive clauses Instructions for informants: Please, give your opinion on the following sentences. Sentence A gives the context and sentence B is the part to be considered. Your opinion can be expressed as follows: Acceptable (√) Marginal (??) Unacceptable (X) THANKS!!! (B) 1. A: He may have read this book. B: In fact, it appears that this book he read thoroughly. (adapted from Hooper & Thompson 1973: 478) (A) 2. A: You are supposed to have seen our sister this morning. B: Oh no. We said that YOUR MOTHER we saw this morning (not your sister). (C) 3. A: Supposedly he examined each part carefully. B: It was impossible that each part he had examined carefully. (adapted from Hooper & Thompson 1973:474) (E) 4. A: The suspect is accused of killing a boy. B: Police discovered that A GIRL the suspect had killed (not a boy). (D) 5. A: He went to see that film. B: Peter regretted that that film he went to see. (adapted from Haegeman 2010: 119) (C) 6. A: My father will buy me a car. B: It is probable that A MOTORBIKE your father will buy you (not a car). (D) 7. A: He didn’t pass the English exam. B: His parents resented that the English exam he had not passed. (C) 8. A: Regrettably Adam might lose the job. B: It is probable that the job Adam will lose. (A) 9. A: They won’t give the book to Peter. B: It is certain that the book they won’t ever give to Peter. (adapted from Cinque 1990: 63) 132 (C) 10. A: Maybe Mark knows the woman. B: It is possible that THE MAN Mark knows (not the woman). (B) 11. A: Government policies should help the poor. B: We think that THE RICH government policies will help (not the poor). (A) 12. A: She went to the supermarket and bought chocolate. B: However, she vowed that BREAD she would buy (not chocolate). (A) 13. A: Carol must have seen a huge crowd. B: Oh yes. Carol exclaimed that such a crowd she has never seen in her life. (adapted from Hooper & Thompson 1973: 474) (E) 14. A: Perhaps she didn’t tell the truth. B: I realized that the truth she hadn’t told. (adapted from Rizzi 2001: 288) (E) 15. A: She should examine each part carefully. B: I saw that each part she had examined carefully. (adapted from Hooper & Thompson 1973: 479) (B) 16. A: My sister is upset. Mary should call her as soon as possible. B: I’m sorry. I imagine that YOUR BROTHER Mary will call (not your sister). (B) 17. A: They decided to fix the motorbike last week. B: We suppose that THE CAR they fixed (not the motorbike). (adapted from Hegarty 1992: 1) (C) 18. A: John should visit Peter. B: It is likely that Peter John will visit. (E) 19. A: Mary had to read the novel. B: We found out that the novel Mary read from beginning to end. (B) 20. A: Did they run with Rubalcaba? B: It happened that WITH RAJOY they ran (not with Rubalcaba). (adapted from Rivero 1980: 372) (D) 21. A: She saw a monkey for the first time. B: I was surprised that a monkey she hadn’t seen before. (B) 22. A: Boys, the cup is broken! B: I believe that THE VASE the boys have broken (not the cup). (A) 23. A: Your neighbours are looking for a dog. B: It is impossible. We reported that A CAT we lost (not a dog). (D) 24. A: Unfortunately, Peter has forgotten the meeting. B: It bothers me that the meeting Peter has forgotten. 133 (A) 25. A: John has never seen this movie. B: It isn’t true that this movie John has never seen. (adapted from Haegeman 2012: 10) (D) 26. A: I adore flowers. I love their smell. B: Thus, it is odd that A CAKE he has sent you for your birthday (not flowers). (A) 27. A: He hasn’t read the report. B: However, he claims that the report he has read. (B) 28. A: Mark always makes the beds. B: As a matter of fact, Bill supposed that the beds Mark had made. (E) 29. A: He had to borrow a pencil. B: That’s not right. I found out that A PEN he had to borrow (not a pencil). (D) 30. A: Paul was obliged to submit a proposal. B: So, I’m surprised that AN ABSTRACT Paul submitted (not a proposal). (adapted from Haegeman 2010: 123) (B) 31. A: Anthony is unable to steal the cookies. B: Nevertheless, we think that the cookies Anthony stole. (A) 32. A: My father doesn’t like pizza. B: I don’t think so. My sister said that pizza your father ate. (C) 33. A: Nowadays, the republic is widely criticized by the voters. B: We doubt that THE MONARCHY the voters would prefer (not the republic). (B) 34. A: Henry expects to win the election. B: I believe that the election Henry will lose. (A) 35. A: We are happy because Robert’s uncle won the championship. B: Robert claimed that THE MATCH his uncle won (not the championship). (C) 36. A: Supposedly Wendy didn’t open the door. B: It is unlikely that THE WINDOW Wendy had opened (not the door). (adapted from Hooper & Thompson 1973: 479) (D) 37. A: The student is able to solve very difficult problems. B: Hence, it is strange that this problem the student hasn’t been able to solve. (adapted from Haegeman 2004: 16) (E) 38. A: During the Second World War, I led a rifle platoon. B: We know that a rifle platoon you led during the Second World War. (D) 39. A: You had to finish the thesis in time. B: My tutor resented that THE REVIEW I didn’t finish in time (not the thesis). 134 (E) 40. A: According to Philip, your father stole a tricycle. B: Philip knows that A BICYCLE my father stole (not a tricycle). (C) 41. A: The accident could have been avoided. B: We doubt that the accident she could have avoided. (A) 42. A: In my opinion, Peter was insulted by our partner. B: It is certain that JOHN our partner insulted (not Peter). (C) 43. A: They must have eaten the jam because I can’t find it. B: Indeed, they denied that THE CHEESE they had eaten (not the jam). (B) 44. A: He was the main candidate for the prize. B: In spite of this, he never imagined that the prize he would win. (E) 45. A: Your friends accused you of hiding the keys. B: That’s not true. My friends realized that THE TICKETS I hid (not the keys). (C) 46. A: He said those things. B: Actually, he never denied that those things he had said. (adapted from Rizzi 2004: 232) (D) 47. A: They didn’t attend the opera. B: Well, I regret that THE CONCERT they didn’t attend (not the opera). (adapted from Hooper & Thompson 1973: 479) (E) 48. A: He shouldn’t tell the secret. B: I’m sorry. I discovered that the secret he had told. (adapted from Rizzi 2001: 288) (E) 49. A: You should have given your sister the book. B: I recognize that THE PENDRIVE I didn’t give to my sister (not the book). (D) 50. A: Did you want to photograph the boy? B: I’m sorry that THE WOMAN I couldn’t photograph (not the boy). 135 TEST on Spanish CLLD/Focalisation in factive clauses Instrucciones para informantes: Por favor, da tu juicio sobre las siguientes oraciones. La oración A facilita el contexto y la oración B es la sección que debes valorar. Tu respuesta puede expresarse de la siguiente manera: Aceptable (√) Marginal (??) Inaceptable (X) ¡¡¡GRACIAS!!! (D) 1. A: Miguel es mi mejor amigo y cuando estoy mal lo suelo llamar. B: Por ese motivo, me sorprende que A ALBERTO hayas llamado (no a Miguel). (C) 2. A: Javier debe tomar zumo para desayunar. B: Javier negó que LECHE haya tomado para desayunar esta mañana (no zumo). (D) 3. A: Según Juan, Susana debería estudiar medicina. (Carlos, Campos) B: De hecho, Juan lamenta que HISTORIA quiera estudiar Susana (no medicina). (C) 4. A: Para el examen de lingüística debo leer el manual de Radford en un mes. B: Dudo que el manual de Radford lo puedas leer en un mes. (A) 5. A: El fontanero fue despedido hace poco tiempo. B: Hugo dijo que AL ALBAÑIL despidió hace poco tiempo (no al fontanero). (B) 6. A: Hace más de una hora que llamé a la pizzería y pedí algo de comida. B: Me imagino que el pedido no lo traerán hasta después de las doce. (C) 7. A: La selección española está obligada a ganar la Eurocopa. B: Dudo que EL MUNDIAL pueda ganar la selección española (no la Eurocopa). (D) 8. A: ¡Se comió la pera en menos de un minuto! B: Es interesante que EL PASTEL se comiera en tan solo un minuto (no la pera). (adapted from Hooper & Thompson 1973: 479) (C) 9. A: Me gustaría acabar la tesis antes de Mayo. B: Es probable que EL QUINTO CAPÍTULO acabes antes de Mayo (pero no la tesis entera). 136 (D) 10. A: Pedro lleva varios años sin trabajar. B: De hecho, es raro que este espléndido trabajo lo haya rechazado. (C) 11. A: Quizás seleccionen a Raquel para el puesto vacante. B: Es posible que A MARTA seleccionen para el puesto vacante (no a Raquel). (D) 12. A: Ángel y José no se ponen de acuerdo sobre qué referencias han de consultar. B: Precisamente, Ángel lamenta que este libro lo haya estado leyendo José durante todas las Navidades. (B) 13. A: Las elecciones generales se celebrarán en Noviembre de este año. B: Creo que las elecciones las ganará el Partido Socialista. (D) 14. A: Mi hijo tiene la costumbre de llevar siempre las llaves en el bolsillo. B: Por ello, es extraño que las llaves las dejara tu hijo encima del mostrador. (C) 15. A: Sueña con ganar al menos un diploma en las Olimpiadas. B: Es improbable que UNA MEDALLA gane en las Olimpiadas (no un diploma). (A) 16. A: Podría haber tenido una luxación. B: Por suerte, me informaron de que UN ESGUINCE tenía (no una luxación). (A) 17. A: En comisaría nos dijeron que el artefacto encontrado era bastante obsoleto. B: Sin embargo, el inspector dijo que el artefacto lo tuvo que examinar minuciosamente. (adapted from Hooper & Thompson 1973: 474) (D) 18. A: Pedro no puede asistir a su cita puesto que está enfermo. B: A pesar de ello, me fastidia que la cita la haya aplazado Pedro para después de vacaciones. (B) 19. A: ¡A ver cuando echan el tiempo! B: Pues supongo que el tiempo lo echarán al final del telediario. (D) 20. A: La asignatura preferida de Eva es biología. B: Por lo tanto, es raro que MATEMÁTICAS haya aprobado Eva (no biología). (E) 21. A: En una relación siempre se debe de decir la verdad. B: No obstante, me di cuenta de que la verdad la había ocultado mi marido durante todo nuestro matrimonio. (A) 22. A: Tienes una cita con Mario para el próximo miércoles. B: Juraría que A MANUEL cité para el próximo miércoles (no a Mario). (E) 23. A: En el supermercado procuran que las bebidas estén siempre frías. B: Es más, vimos que las bebidas las guardan en unas cámaras frigoríficas. (C) 24. A: los vuelos a Manchester van a ser cancelados. B: No, la compañía ha negado que los vuelos a Manchester los vaya a cancelar. (adapted from Emonds 2004: 277) 137 (E) 25. A: Hace tiempo que cedí la parcela en propiedad. B: No sabía que la parcela la habías cedido en propiedad. (A) 26. A: A Alicia le encanta la blusa que le han regalado. B: Todo lo contrario. Alicia exclamó que la blusa la va a devolver hoy mismo. (B) 27. A: Habíais pedido lasaña, ¿no? B: Resulta que MACARRONES habíamos pedido (no lasaña). (E) 28. A: Según el profesor, mi hijo amenazó a Pablo. B: Pues me he enterado de que A MARCOS amenazó tu hijo (no a Pablo). (A) 29. A: Necesitamos encontrar un merendero donde poder comer. B: El guarda forestal nos ha informado de que el merendero lo podemos encontrar tras ese valle. (A) 30. A: Nuestros políticos han de cumplir con sus obligaciones. B: Concretamente, nuestro presidente reivindicó que sus obligaciones las cumplirá durante todo su mandato. (E) 31. A: ¿Toca Claudia la guitarra? B: Sabemos que EL PIANO toca Claudia con gran soltura (no la guitarra). (B) 32. A: El hermano de Carolina teme por su despido. B: Carolina piensa que a su hermano lo despedirán a final de mes. (E) 33. A: ¡Un hombre ha intentado forzar la cerradura de mi casa! B: Vi que UNA MUJER intentó forzar la cerradura de tu casa (no un hombre). (E) 34. A: Los jóvenes escondieron mi monedero. B: Los jóvenes reconocieron que TUS LLAVES escondieron (no tu monedero). (B) 35. A: Luisa ha leído la novela por encima. B: Efectivamente, parece que la novela la ha leído Luisa por encima. (A) 36. A: Europa ha de solventar una gran crisis institucional. B: Es obvio que la crisis institucional la intentará solventar Europa en breve. (A) 37. A: ¡Has perdido mi camisa! B: Es verdad que TU ABRIGO he perdido (no tu camisa). (D) 38. A: Soy un desastre. Nunca llamo a Ramón. B: Me fastidia que A TUS PADRES no llames nunca (no a Ramón). (B) 39. A: Desgraciadamente la nueva multinacional no hace más que generar pérdidas. B: Esperaba que BENEFICIOS generara la nueva multinacional (no pérdidas). (C) 40. A: Estoy deseando recibir el paquete. B: Es imposible que el paquete lo recibas antes del lunes. 144 data above, that CLLD is at least more easily available in factive clauses in Spanish than Topicalisation would be in English, showing it is not subject to the same licensing requirements. Indeed, some authors have indeed signalled that CLLD has a wider distribution than English topics (Cinque 1990, Hirschbuhler 1997, Zubizarreta 1998). These findings confirm our previous claim that Spanish CLLD is of an argumental nature. To be more precise, in Chapter 1, Section 1.2, we pointed out that if a sentence constituent moves to an argumental position, it is a case of A-movement. In this way, since topics move [Spec, T] they do not have to move across the factive operator. Consequently, the fronted argument in the CLLD construction does not lead to intervention. As a matter of fact, in Chapter 2, Section 2.7, we highlighted that fronted topics in Spanish have [Spec, T] as their landing site. More specifically, [Spec, T] is a multifunctional position in Spanish, which can be occupied by preverbal subjects, but also by foci, interrogative operators and topics. In this case, EPP is a purely structural condition that requires some specifier position to be filled, independently of the category or grammatical relation of the constituent that fills it. Furthermore, as stated in Chapter 1, Section 1.7, the VP-internal subject hypothesis implies that the lexical shell of the clause, that is, the V and its arguments, are generated inside V so that, topic constituents would precede the subject. Regarding Spanish Focalisation, it must be noted that for most informants it is at least marginally compatible with factive clauses. The availability of such a phenomenon in Spanish factive clauses confirms our idea that focus movement in Spanish is also argumental. In this way, the displaced constituent does not interact with the null operator of factive clauses: 145 Focalisation Type of verb Acceptable Marginal Unacceptable A (non-factive) 13 (26%) 22 (44%) 15 (30%) B (non-factive) 14 (28%) 23 (46%) 13 (26%) C (non-factive) 8 (16%) 19 (38%) 23 (46%) D (factive) 8 (16%) 26 (52%) 16 (32%) E (factive) 15 (30%) 25 (50%) 10 (20%) The experimental results provide evidence that Spanish Focalisation is possible in factive clauses. More concretely, Spanish Focalisation is compatible with most nonfactive verbs (A, B and C). In the same way, factive verbs (D and E) show a high compatibility with the operation of focus fronting in Spanish. Thus, we demonstrate that Focalisation does not give rise to the typical intervention effects of other types of movement to the CP domain. Nevertheless, unlike it has been noted by JiménezFernández & Camacho-Taboada (2014), Focalisation is more restricted than CLLD in factive clauses. In this regard, factive and non-factive predicates may be asserted or non-asserted. The absence of assertion would explain why in some cases these verbs may not be compatible with Spanish Focalisation. 4.4.6. Conclusion To conclude, we have seen that factive clauses resist topic or focus fronting in English while remaining compatible with discourse movement in Spanish. This asymmetry follows from intervention. As stated in Chapter 3, Section 3.4, it implies that 146 discourse elements behave in a different way and have a different location in English and Spanish. In particular, in English the priority is that the subject receives nominative case in [Spec, T]. Thus, the specifier of TP is not an available landing site for discourse constituents, that is, topics and foci have to move to CP since focus features are not lowered from C to T. However, as we have just seen, on their way to CP, focused and topicalised constituents would have to move across a factive operator. Such a movement would cause intervention. Hence, discourse movement in English factive clauses is illicit. On the contrary, in Spanish since the subject agrees with the verb and does not undergo movement to TP, discourse elements can move freely into the specifier of such a projection. In this way, the factive operator is higher up and discourse constituents do not have to move across it. Therefore, we claim that CLLD and Focalisation are possible in Spanish factive clauses since these discourse constituents do not create intervention effects and the resulting sentence would be grammatical. 147 4.5. Concluding remarks From the preceding considerations we can draw the following specific conclusions: - It has been acknowledged that the so called Main Clause Phenomena (MCP) are restricted to occurring in root clauses and a limited subset of subordinate clauses. In this sense, we might think, for example, that factive clauses lack a left periphery ‘space’ entirely. Nevertheless, such an account is controversial. - In English factive complements resist Topicalisation, while non-factive complements admit it. A movement analysis of factive clauses allows us to analyse this restriction in terms of an intervention effect. Specifically, English factive clauses are derived by movement of an operator to their left periphery. The operator would move from within the TP domain to the CP area. Since the factive operator targets CP, it will block topic fronting, which also lands in CP. Hence, English Topicalisation is not compatible with factivity. - English focus movement is permitted with non-factive verbs but not with factives. As the factive operator is hosted by CP, it will block focus fronting, which also involves CP. In such a situation, the focused constituent would have to move across the factive operator. Such a result would be deviant. Therefore, English factive clauses are incompatible with Focalisation. - There are crucial contrasts between Spanish and English, and therefore the mechanism we propose for English Topicalisation cannot simply be transposed to Spanish. Indeed, 148 Spanish CLLD does not give rise to the same intervention effects as English Topicalisation and it is allowed in factive clauses. It suggests that English Topicalisation and Spanish CLLD implies different landing-sites. In contrast to English, Spanish focus arguably stays within the TP domain (rather than raising to the CP domain). Therefore, the factive operator is higher up and topic constituents do not have to move across it. For this reason, we claim that CLLD is possible in Spanish factive clauses since these discourse constituents do not create intervention effects and the resulting sentence would be grammatical. - As with CLLD, embedded focus is readily available in Spanish factive clauses without restriction. More precisely, in Spanish since the subject agrees with the verb and does not undergo movement, focus elements can move freely into the specifier of such a projection and satisfy the EPP. In this way, we have a straightforward explanation for the fact that this low focus does not result in an intervention effect the same way as English focusing does. - Ten native speakers of English and some others of Spanish determine the degree of acceptability of different word orders in both languages. They corroborate the idea that unlike English Focalisation/Topicalisation, Spanish focus and topic elements do not give rise to intervention effects. Consequently, discourse movement implies different landing-sites in English and Spanish factive clauses. 149 5. CONCLUSIONS 150 The general tendency in generative grammar is to bring languages together and to discover common ground existing among them, without denying their individual special features in the process. In this study, we have tried to show that recent developments in linguistic theory within generative grammar provide an excellent framework for the comparison of both very subtle and more general differences and similarities between English and Spanish. In Chapter 1, we showed that languages such as Spanish are known as “free word order” languages since sentence constituents can appear in many different positions. On the contrary, other languages such as English are generally regarded as having relatively “fixed word order”. In this case, the sentence constituents normally appear in a specific order. However, the canonical pattern in English may be manipulated in order to give special emphasis to a specific constituent. Furthermore, we stated that contrastive focus, which expresses exhaustive identification, must be distinguished from information focus, which merely expresses new information. In the same way, we must differentiate between two types of topic in Spanish: clitic left dislocation and hanging topic. More concretely, the hanging topic only occurs at the periphery of the whole message, whereas the clitic left dislocated topic appears at the periphery of the clause. Similarly, in English we must distinguish between Topicalisation and Left-dislocation. They are syntactically different; while topicalisations have a gap, where the topicalised phrase would have appeared in the nonpreposed version, left-dislocations include a resumptive pronoun. Then, we claimed that due to the limitations of the Language Faculty, lexical items move cyclically though different phases (CP and vP). Once a phase is formed, the complement of the phase head undergoes transfer to the phonological and semantic 151 components so that neither the complement nor any element it contains can take part in further syntactic operations from that point on. In Chapter 2, we argued for a movement-based analysis, suggesting that the processes of Topicalisation and Focalisation imply displacement. In this sense, languages vary with respect to the systematic properties of syntactic reordering. So, in languages like English, we depart from Rizzi’s cartographic approach and adopt an unsplit CP analysis. Hence, in English discourse constituents would move to the specifier of CP in the left periphery of the clause. By contrast, [Spec, T] is a multifunctional position in Spanish. Therefore, discourse elements have the specifier of TP as their landing site. In Chapter 3, we demonstrated that the A/A´-contrast observed in discourse movement can be accounted for in a precise way by taking into account certain facts. In this connection, we confirmed that discourse/agreement features interact with the EPP feature in the process of Agree in the Minimalist Program. Thus, because of the subjectverb concord, Spanish is an agreement-prominent language. In addition, due to informational movement, Spanish is also a discourse-prominent language. In this case, focus elements can satisfy the EPP in the specifier of TP in the same way that the subject does it. The reason is that the EPP is a purely syntactic restriction that requires some specifier position to be filled, regardless of the grammatical relation of the element that fills it. Meanwhile, English only gives prominence to agreement features, which spread into T. Consequently, it has a relatively strict word order. To be more precise, the priority is that the subject receives nominative case in [Spec, T]. So, the specifier of TP is not an available landing site for discourse constituents, that is, topics and foci have to move to CP since focus features are not lowered from C to T. 152 In Chapter 4, we claimed that there are crucial contrasts between Spanish and English, and therefore the mechanism we propose for English Topicalisation and Focalisation cannot simply be transposed to Spanish. To put it explicitly, factive clauses are derived by movement of an operator to their left periphery. The operator would move from within the TP domain to the CP area. In this way, we can provide an explanation for the incompatibility of English factive clauses with discourse movement. Nevertheless, unlike English Topicalisation and Focalisation, CLLD and Focalisation in Spanish are available in factive clauses since they do not give rise to the typical intervention effects of other types of movement to the CP domain. More specifically, in English factive constructions, topics and foci move to [Spec, C]. This supports the idea that MCP such as Topicalisation and Focalisation are incompatible with factive clauses. In Spanish, however, each topic and focus lands in a specifier of the head T so that there is no intervention effect between the operator and the topic and focus constituents. We are conciouss of the fact that many of the topics included in this study need further analysis and elaboration. Nevertheless, we hope that we have been able to open the door for a comparative analysis that may limit the relationships between languages, while at the same time enriching our view of their specific differences and enhancing our knowledge of language. 153 6. REFERENCES