論西班牙文第一連接詞之一致關係

全文

(1)國立臺灣師範大學英語學系碩. 士. 論. 文. Master’s Thesis Department of English National Taiwan Normal University. 論西班牙文第一連接詞之一致關係 On First Conjunct Agreement in Spanish. 指導教授：颯楊博士 Advisor: Dr. Gerardo Fernández-Salgueiro 研究生：薛. 博. 帆. Student: Bo-Fan, Andrew, Syue. 中華民國一百零六年六月 June, 2017.

(2) 摘要. 本文探討西班牙文中第一連接詞之一致關係 (first conjunct agreement) 的現象，此現象在許多語言中已經被大量地研究了。然而，幾乎沒有任何研究著重在西班牙文不同句構中的第一連接詞之一致關係。此外，在最簡方案 (Minimalist Program) 中並沒有任何關於西班牙文第一連接詞之一致關係的分析有包含階段理論 (Phase Theory) (Chomsky 1998) 或談論到連接詞組 (&P) 附加的時機。本研究採納了 Soltan (2007) 的晚附加 (late adjunction) 分析方式，也納入了階段 (phase) 的概念來檢視西班牙文中五種不同的句構，包括存在句、及物句、非限定句、被動句及反身被動句。另外，我提出以下的看法 — 和一致投射 (agreement projection) 有關連的動詞與第一連接詞之間的完整標記一致 (ϕ-complete Agree) 阻饒了第一連接詞之一致關係；然而，分詞和第一連接詞之間的缺陷標記一致 (ϕ-defective Agree) 顯示了第一連接詞之一致關係。透過了三個成分 (晚附加、一致和階段)，本文可以成功地解釋在本研究中五種句構的第一連接詞之一致關係的不同可能性；另外，連接詞組的附加時機也被展示出來。我預期本研究的分析可以被運用到西班牙文中的任何句構。. 關鍵詞: 第一連接詞之一致關係、西班牙文、一致、階段、晚附加. i.

(3) Abstract. This thesis investigates the first conjunct agreement (FCA) phenomenon in Spanish. This phenomenon has been widely studied in many languages. However, there are few studies, if any, that focus on FCA in different constructions in Spanish. Besides, there is no analysis on Spanish FCA in the Minimalist Program that incorporates the phase theory (Chomsky 1998) or deals with the timing of the adjunction. In this study, I adopt the late adjunction approach in Soltan (2007) and take into account the phase concept to look into five different constructions, including existential, transitive, nonfinite, passive, and reflexive passive, in Spanish. In addition, I propose that ϕ-complete Agree between the verb associated with the agreement projection and the first conjunct blocks FCA; ϕ-defective Agree between the participle and the first conjunct, however, displays FCA. Through the three ingredients, i.e. late adjunction, agreement, and phase, the different possibilities of FCA among the five constructions in this study can be successfully accounted for; besides, the timing of the adjunction of the &P is also demonstrated. It is expected that the analysis of this study can be applied to any constructions in Spanish.. Keywords: first conjunct agreement, Spanish, agree, phase, late adjunction. ii.

(4) Acknowledgements. It has not been an easy journey; however, with the help from a lot of people, I am able to stand where I am today. I would like to express my sincere appreciation to the following people. Ante todo, me gustaría expresar mi agradecimiento al asesor de mi tesis, Prof. Gerardo Fernández-Salgueiro por su paciente orientación. Hemos pasado mucho tiempo trabajando en este tema desde el segundo semestre. La discusión sobre numerosas citas fue eficaz y me inspiró a pensar más. Sus comentarios constructivos me ayudaron a mejorar mi tesis. Como un hablante no nativo de español, no fue fácil analizar muchos datos. Pero siempre que me encontré dificultades durante el proceso de escribir la tesis, Gerardo estuve allí para ayudarme y guiarme a través de las dificultades. La finalización de esta tesis no hubiera sido posible sin su apoyo. I would also like to show my appreciation to my committee members, Prof. HsiaoHung Iris Wu and Prof. Chao-Ting Tim Chou, for their insightful comments and suggestions, which helped me to improve the quality of my thesis. I thank them again for being willing to spend their valuable time reading and commenting my thesis and being an important part of my thesis journey. In addition, I would like to express my thankfulness to the professors who taught. iii.

(5) me during my three years in the linguistics program in National Taiwan Normal University: Prof. Miao-Hsia Tammy Chang, Prof. Chun-Yin Doris Chen, Prof. MiaoLing Hsieh, Prof. Hsi-Yao Hsu, Prof. Hui-Shan Nissa Lin, Prof. Jen-I Li, and Prof. Jen Ting, by alphabetical order. After taking the courses they offered, I gained a lot of knowledge from various subfields in linguistics. I really appreciate their detailed instruction and their patience in guiding me. I am very glad to have a cohesive group of classmates in the linguistics program: Amy Chen, Francis Chen, Kabby Chen, Sarah Chen, Aries Cheng, Eliza Cheng, Vivien Cheng, Lucy Chiang, Pris Ho, Lily Hsiao, Johnny Hu, Nick Hu, Lin Kong, Johnny Lee, Terry Li, Eileen Lin, Shawn Lin, Treak Lu, Amber Weng, Louisa Zheng, by alphabetical order. I enjoyed their company during the time in the linguistics program. I also appreciate their encouragement and help whenever I encountered problems. Besides, I would also like to thank my dear roommates and friends Anit and Monica. Thanks for always making delicious food. I enjoy your company a lot. Also, I would like to say thanks to the following friends who brought joy and happiness into my life: Anne, Artemis, Bryan, Caique, Chieh-En, Daniel, Doug, Diogo, Ender, Fred, Joanna, Kun-Hao, Marshell, Mary, Ming-Yang, Noah, Rui, Zach, Ying-Yu, Hsin, HsinYa (and many more). I love y’all. También me gustaría dar mi reconocimiento especial a Héctor por idear unas frases en español que necesité para mis argumentos en la tesis.. iv.

(6) ¡Muchas gracias! 最後，我要感謝我親愛的家人一直以來對我的支持和照顧。謝謝我的爸爸讓我在沒有後顧之憂下求學，我也很感謝我的奶奶、叔叔們、姑姑、哥哥和妹妹對我的關心。我也要謝謝我在天上的媽媽，謝謝她從小到大對我的教育的重視和無條件的愛，我想跟您說:「我愛您，我也很想您!」我要把我的論文獻給我所有的家人。. v.

(7) Table of Contents Chinese Abstract ............................................................................................................. i English Abstract .............................................................................................................ii Acknowledgements ...................................................................................................... iii Glossary .......................................................................................................................vii Chapter One – Introduction........................................................................................ 1 1.1 Overview .................................................................................................... 1 1.2 Motivation .................................................................................................. 8 1.3 Data Concerned in This Study ................................................................... 9 1.4 Purpose of the Study ................................................................................ 11 1.5. Organization of the Thesis ....................................................................... 12. Chapter Two – Previous Studies ............................................................................... 13 2.1 Munn (1999) ............................................................................................ 13 2.2 Camacho (2003) ....................................................................................... 16 2.3 2.4 2.5 2.6 2.7. Soltan (2007) ............................................................................................ 17 Bošković (2009) ....................................................................................... 20 Demonte and Pérez-Jiménez (2012) ........................................................ 21 Aljović and Begović (2016) ..................................................................... 24 Implications for FCA in Spanish.............................................................. 26. Chapter Three – FCA in Various Types of Constructions in Spanish ................... 28 3.1 Basic Structure in Spanish ....................................................................... 28 3.2 Existentials ............................................................................................... 30 3.3 Transitives ................................................................................................ 37 3.4 Non-finites ............................................................................................... 43 3.5 Passives .................................................................................................... 48 3.6 Reflexive Passives ................................................................................... 52 3.7 Summary of the Five Construction Types Concerning FCA ................... 56 Chapter Four – Analysis: Timing of Adjunction, Agreement, and Phases ........... 58 4.1 Completeness and Defectiveness of ϕ-Features....................................... 60 4.2 The Phase Theory – Chomsky (1998) ..................................................... 62 4.3. Deriving FCA with Late Adjunction, Agreement, and Phase .................. 63 4.3.1 One Phase – Existentials, Passives & Reflexive Passives ........... 63 4.3.2 Two Phases – Transitives & Non-finites ...................................... 68. Chapter Five – Conclusion ........................................................................................ 73 References ................................................................................................................... 75. vi.

(8) Glossary Person 1 : First person 3 : Third person Number SG : Singular PL : Plural Gender M : Masculine F : Feminine N : Neuter Tense PRES PAST. : Present : Past. Mood IND : Indicative Others Refl : Reflexive. vii.

(9) Chapter One Introduction. 1.1 Overview First conjunct agreement (henceforth FCA) phenomenon is quite prevalent among languages in the world. As much cross-linguistic empirical evidence shows (Aoun, Benmamoun and Sportiche 1994, 1999; Sobin 1994; Munn 1999; Doron 2000; Camacho 2003; Soltan 2007; Marušič, Nevins and Saksida 2007; Bošković 2009; Alijović and Begović 2016; among many others), FCA occurs when a verb, an adjective or a determiner agrees with the first conjunct of a coordinate DP. (1) illustrates the phenomenon in English, (2) in Serbo-Croatian, and (3) in Spanish. As shown in (1), the existential verb is agrees with the first conjunct a man in person and number. The participle uništena ‘destroyed’ in (2) agrees with the first conjunct sva sela ‘all villages’ in number and gender. As for (3), the singular and feminine determiner una ‘a’ searches for the first conjunct lluvia ‘rain’ to agree with in person and gender. Notice that agreement in these examples exclude the second conjunct.. (1) There is a man and a woman in the park.. 1.

(10) (2) Juče yesterday. su. uništena. are destroyed.PL.N. sve varošice.. sva sela. i. all. and all. villages.N. towns.F. ‘All villages and all towns were destroyed yesterday.’ (Bošković 2009). (3) Una a.SG.F. fuerte. lluvia. y. viento. azotaron la. ciudad.. heavy.SG rain.SG.F and wind.SG.M battered the city. ‘A heavy rain and wind battered the city.’ (Demonte and Pérez-Jiménez 2012). From the examples (1-3) above, we can notice that the verb, the adjective (participle), and the determiner are always to the left of the coordinate structure when FCA occurs, which means that they should always c-command the coordinate structure if we expect to observe FCA.1 This concept can be captured if we follow Munn (1993, 1999), Marušič, Nevins, and Saksida (2007), Soltan (2007), Bošković (2009), Benmamoun, Bhatia, and Polinsky (2010), Demonte, and Pérez-Jiménez (2012), Aljović and Begović (2016), among many others, in assuming that the coordinate structure is DP1 plus the adjunction of a conjunction phrase (henceforth &P) which contains a conjunction head and DP2. Notice that I label the highest DP as DP1, the first. 1. In this study, I assume the Linear Correspondence Axiom (or LCA) proposed in Kayne (1994) even though in more general terms this cannot be used to account for head-final languages. 2.

(11) conjunct as DP1 as well and second conjunct as DP2. 2 The tree diagram of the coordinate structure is shown in (4) below:. (4). DP1 DP1. &P &. DP2. Before moving on, I would like to mention the probe-goal theory introduced by Chomsky (2000), which will be a fundamental part of my analysis. Under the operation Agree presented by Chomsky (1999, 2000, 2001), the probe is an element that has an uninterpretable feature, i.e. the verb, the adjective, and the determiner in examples (1)(3) above; during the derivation, the probe searches downward to find a goal with a matching interpretable feature to agree with, as illustrated in (5).3. 2 3. In this study, the higher DP1 is a ‘coordinate structure’ and &P is a ‘conjunction phrase. I assume the Activity Condition (see Boeckx 2003). 3.

(12) (5) The illustration of operation Agree by Chomsky (2000) α αProbe. β. [-interpretable] F βGoal…… AGREE. [+interpretable] F. We understand that FCA occurs when the probe sitting higher in the structure targets the first conjunct instead of the coordinate structure as the goal to agree with, as shown in the first possible agreement in (6). Notice that the agreement of the second route, i.e. full agreement, is also possible, where the probe targets the coordinate structure to agree with.4. (6). 2 Full agreement ○. Probe DP1 1 ○. FCA. DP1. &P. Goal &. 4. DP2. Notice that there is no semantic difference for these two agreement patterns. 4.

(13) However, it is not entirely clear why the probe targets the first conjunct but not the coordinate structure given the fact that they are both equidistant to the probe under the assumption of May’s (1985) theory of adjunction. Here, I would say it is related to the timing of the adjunction of the &P. Before the adjunction of the &P, there is only one DP. Therefore, when the operation Agree applies, the probe has no other choice but to agree with the DP, as indicated in (7). Example (8) shows that the participle colgada ‘hung’ agrees with the only DP in the sentence una lámpara ‘a lamp’ in number (singular) and gender (feminine).5. (7). Probe DP Goal AGREE. (8) Tenemos colgada una have .1PL hung.SG.F a.SG.F. lámpara en. el. techo.. lamp.SG.F in. the ceiling. ‘We have a lamp hung in the ceiling.’. 5. The properties of the construction will be introduced in Chapter 3. 5.

(14) As shown in (9), after the operation Agree between the probe and the first conjunct takes place, &P then adjoins to the first conjunct, yielding two DP1 segments and the FCA phenomenon. In example (10), the participle colgada ‘hung’ agrees with the DP una lámpara ‘a lamp’ in number (singular) and gender (feminine) before the &P y un candelabro ‘and a chandelier’ late adjoins to it. The approach where the &P late adjoins to the DP1 is called “acyclic adjunction”. This approach is adopted from Soltan (2007) (See also Lebeaux 1988; Chomsky 1993; Fox and Nissenbaum 1999; and Uriagereka 2002). More details about the derivation will be discussed in the following chapters.. (9). Probe DP1 2 Adjunction ○ 1 AGREE ○. DP1. &P. Goal &. (10) Tenemos colgada una have.1PL hung.SG.F a.SG.F el. DP2. lámpara y. un. lamp.SG.F and a.SG.M. techo.. the ceiling 6. candelabro. en. chandelier.SG.M. in.

(15) ‘We have a lamp and a chandelier hung in the ceiling.’. There is an interesting interaction between the late adjunction of the &P and the movement of the phrase that is attached to – the adjunction of the &P cannot apply to an A-moved phrase but it can apply to an A’-moved phrase. If we late adjoin the &P to a DP after the DP moves to an argument position, it will lead to an ungrammatical sentence. In (11a), if the adjunct that John made late adjoins to the subject the claim after it moves to the specifier of TP (an A-position), it will result in an ungrammatical sentence, as indicated in (11a). However, if we late adjoin the &P to a phrase that moves to an A’-position, the outcome is acceptable. As the classic example from Lebeaux (1988) presented in (11b) shows, the adjunct that John made late adjoins to the whphrase which claim and it results in a grammatical sentence.. (11) a.. b.. *The claim [ADJUNCT that Johni made] seems to himi to be wrong.. Which claim [ADJUNCT that Johni made] was hei willing to discuss?. This asymmetry also suggests that after A-movement FCA is not possible. Remember that FCA occurs when an &P late adjoins to a DP that has already undergone. 7.

(16) Agree with the probe. In (12a), the existential verb be agrees with the subject a man in the specifier of vP before the subject A-moves to the specifier of TP. If we late adjoin an &P and a woman to the A-moved subject, it will lead to a mismatch in the ϕ-features between the “new” subject a man and a woman and the probe is and thus the sentence will be ungrammatical, as illustrated in (12b).. (12) a. b.. A man is in the park. *A man [ADJUNCT and a woman] is in the park.. 1.2 Motivation FCA has also been discussed in Spanish (Camacho 2003; Demonte & PérezJiménez 2012); however, to the best of my knowledge, there is no analysis on Spanish FCA in the Minimalist Program concerning the phase theory (Chomsky 1998) as well as the timing of adjunction. Besides, this phenomenon has not been worked on with respect to existentials in Spanish, in which presumably there is no Agr projection associated with the existential verb haber ‘to be’ and thus no overt subject agreement can be observed. To bridge the gaps, in this thesis I would like take into account the phase theory and the timing of adjunction, look into five constructions including existentials, transitives, in which the participle modifies an object, non-finites, passives. 8.

(17) and reflexive passives, and try to provide a uniform analysis to account for the FCA phenomenon in Spanish.6. 1.3 Data Concerned in This Study In this study, I investigate the FCA phenomena with data from Spanish, which show that existential, transitive and non-finite constructions allow FCA while passive and reflexive passive constructions don’t, as illustrated in (13)-(17) below.7 In each of the example, there is a post-verbal and post-adjectival coordinate phrase subject consisting of two singular nouns with different gender features, i.e. una lámpara (a lamp, feminine) and un candelabro (a chandelier, masculine).. (13) Existential: Hay. colgada una. have.PRES,IND hung.SG.F a.SG.F juntos. en. together in. el. lámpara y. un. lamp.SG.F and a.SG.M. candelabro chandelier.SG.M. techo. the ceiling. ‘There is a lamp and a chandelier hung together in the ceiling.’. 6. In this study, the transitive construction is a specific type in which the participle modifies an object (so it doesn’t agree with the verb). 7 For their judgement on the sentences used in this study, I would like to thank my advisor Prof. Gerardo Fernández-Salgueiro and the members in the Issues in Romance Grammar group on Facebook. 9.

(18) (14) Transitive: Tenemos. colgada una. lámpara y. have.PRES/PAST.1PL. hung.SG.F a.SG.F. lamp.SG.F and a.SG.M. juntos. techo.. en. Together in. el. un. candelabro chandelier.SG.M. the ceiling. ‘We have a lamp and a chandelier hung together in the ceiling.’. (15) Non-finite: Al. estar. because to.be juntos. en. together in. el. colgada una. lámpara y. hung.SG.F a.SG.F. lamp.SG.F and a.SG.M. techo,. the ceiling. un. candelabro chandelier.SG.M. nos marchamos. refl left. ‘Because a lamp and a chandelier are hung together in the ceiling, we left.’. (16) Passive: *Fue. colgada una. lámpara y. un. was.3SG. hung.SG.F a.SG.F. lamp.SG.F and a.SG.M. en. el. techo.. in. the ceiling. candelabro. juntos. chandelier.SG.M. together. Intended meaning: ‘A lamp and a chandelier were hung together in the ceiling.’ 10.

(19) (17) Reflexive Passive: *Se mantiene colgada una. lámpara y. refl keep.3SG hung.SG.F a.SG.F. lamp.SG.F and a.SG.M. juntos. en. together in. el. un. candelabro chandelier.SG.M. techo.. the ceiling. Intended meaning: ‘A lamp and a chandelier were hung together in the ceiling.’. 1.4 Purpose of the Study The gist of this study is to account for the impossibility of FCA in passive and reflexive passive constructions, as opposed to the other three construction types. By examining the data, it looks like ϕ-complete Agree between the verb in the agreement projection (Agr projection, henceforth) and the first conjunct blocks FCA; ϕ-defective Agree between the participle and the first conjunct, however, displays FCA.8 Aside from this observation, the analysis adopted in Soltan (2007), i.e. late-Merge of the adjunction phrase, and the phase concept from Chomsky (1998) both help to explain the FCA phenomenon in Spanish. I argue that the actual timing of the late adjunction of the second conjunct depends on whether there is an Agr projection and whether a phase has been completed.. 8. The definitions of ϕ-complete Agree and ϕ-defective Agree will be provided in Chapter 4. 11.

(20) 1.5 Organization of the Thesis This thesis is organized as follows. Chapter 2 reviews previous studies on FCA and the underlying issues. In Chapter 3, I introduce the basic syntactic structure in Spanish and the basic properties of the five constructions, i.e. existentials, transitives, non-finites, passives, and reflexive passives. Besides, the tree structures of these constructions are also provided. At the end of Chapter 3, I point out that Soltan’s (2007) analysis is not sufficient to account for the data in this study. The analysis of FCA of the five constructions, which involves the operation Agree, the details of the agreement of the ϕ-features between the probe and the goal, the acyclic merge of the &P, the application of the phase concept, is included in Chapter 4. Chapter 5 concludes the thesis.. 12.

(21) Chapter Two Previous Studies. In this chapter, I review some previous studies concerning the FCA phenomenon, including Munn (1999), Camacho (2003), Soltan (2007), Bošković (2009), Demonte and Pérez-Jiménez (2012) and Aljović and Begović (2016). In these studies, some issues related to FCA are also covered, e.g. the structure of the conjoined DPs, the clausal and the phrasal analyses (i.e. the elliptical and non-elliptical analyses) of FCA, and the distinction between syntactic plurality and semantic plurality. Through reviewing these works, we can see several different approaches the previous studies adopt to account for FCA and some implications can be drawn from these studies. In the following, the works will be presented in chronological order.. 2.1 Munn (1999) Munn (1999) is a reply to Aoun, Benmamoun, and Sportiche (1994) (henceforth ABS), where they claim that FCA results from ellipsis in the second clause of a coordinate structure. ABS’s main argument is that FCA cannot co-occur with elements that are sensitive to plurality requirement, e.g. the modifier sawa ‘together’ in Lebanese Arabic, reflexives and reciprocals, the intransitive meet in Arabic and others. Munn. 13.

(22) points out that the data provided in ABS’s study is not conclusive because they fail to distinguish between syntactic plurality and semantic plurality. After taking into account both syntactic plurality and semantic plurality, Munn re-examines ABS’s arguments that support a clausal analysis and provides data from other languages such as Irish, Moroccan Arabic and English to show that the data ABS use to support their clausal analysis does not work in some cases in these languages. Besides, the phrasal analysis is preferred to the clausal analysis for its simplicity and less stipulation. After re-examining ABS’s data and proving that their arguments are falsified, Munn proposes that FCA comes from government configurations and it is impossible in specifier and head relations. He supports this concept with examples from existential constructions in English, and VS agreement in Brazilian Portuguese and in Irish (all with a conjoined subject). Munn further provides more data (mixed agreement, FCA within the noun phrase, FCA in participial absolutes) to prove that the clausal analysis cannot account for those data because the clausal analysis does not predict group readings. With this fact, Munn argues that phrasal coordination as well as agreement under government are the two main ideas that are required to explain FCA. In the analysis section, Munn adopts the definitions of government from Aoun and Sportiche (1981) and Chomsky (1986) and assumes that there are two configurations of government. Munn calls the first configuration government where “the complement. 14.

(23) of a head is governed by the head” and the second configuration exceptional government where “the specifier of the complement is governed by the head provided the complement is made transparent in some way” (Munn 1999: 661). The structures of these two configurations are sketched in (18a-b), with the X head governing YP in both configurations.. (18) a. Government. b. Exceptional government. XP. X0. XP. X0. YP. ZP. governs governs YP. Z’. Munn argues that under government FCA occurs when a conjoined DP structure is in the position of YP in these two configurations. Under such analysis, Munn assumes that the conjunction is the head of its phrase with the first conjunct and the second conjunct as the specifier and the complement respectively.. 15.

(24) (19) NP1. NP1. BP. B. NP2. and. Later in the study, Munn mentions another structure for the coordination phrase, i.e. the first conjunct plus the adjunction of a Boolean phrase, which he has already argued for in Munn (1992, 1993). The structure is shown in (19). Munn suggests that the unavailability of “exceptional” exceptional government supports this kind of structure.9 In the current study, I follow this coordination structure with the adjunction of the &P. However, I will not appeal to the idea of government.. 2.2 Camacho (2003) Camacho points out that partial agreement happens in the non-canonical word order of the languages. According to Camacho, there are two types of conjoined structure with partial agreement, i.e. PF agreement (agreement patterns that do not have interpretive consequences) and LF agreement (agreement patterns that do have According to Munn (1999), “exceptional” exceptional government refers to government into the specifier of a specifier. 9. 16.

(25) interpretive consequences), which means that Camacho suggests that partial agreement is not handled in the narrow syntax.10 Following Aoun, Benmamoun and Sportiche (1994), Camacho suggests that partial agreement involves gapping with the examples from verb and subject agreement (non-canonical word order) as well as the coordination of preposition phrases and determiner phrases in interaction with adjectives. However, this analysis gains very few support in the literature of FCA studies. In my study, I will show that the analysis involves gapping or deletion is not correct. Besides, partial agreement phenomenon can be accounted for in syntax.. 2.3 Soltan (2007) In this paper, Soltan tries to show what built-in mechanism can best account for agreement based on the FCA phenomenon in Standard Arabic (SA). With the data presented in this paper, Soltan was able to provide some evidence that shows that a primitive built-in operation Agree in the grammar is preferred to the Spec-head (government) approach in terms of agreement phenomenon. Besides, Soltan proposes that FCA is derived from the interaction between Agree operation and the acyclic adjunction of the &P.. 10. Camacho (2003) eventually proposes that there are actually two types of PF agreement languages. 17.

(26) Soltan first presents the inadequacy of the Spec-head approach Aoun, Benmamoun, and Sportiche (1994) propose to deal with FCA. According to Aoun, Benmamoun, and Sportiche, FCA is derived through the application of coordination reduction to an underlying clausal coordination structure. They back up their argument with the fact that the coordinate structure fails the semantic plurality tests with the data from Lebanese Arabic and Moroccan Arabic. However, Soltan applies the semantic plurality tests with collective predicates and expressions such as maʕan ‘together’, baʕD-a-hum ʔal-baʕD ‘each other’, and ʔiltaqa ‘meet’ in Standard Arabic and shows that they are all compatible in FCA contexts. Besides the data in Standard Arabic, Harbert and Bahloul (2002) and Johannessen (1996) also provide data in Welsh and Czech respectively to show that the analysis Aoun, Benmamoun, and Sportiche offer cannot be applied to some other languages. Later in this paper, Soltan presents his analysis on deriving FCA. He follows the theoretical assumption of Agree operation proposed by Chomsky (1999, 2000, 2001); in addition, he adopts the following assumptions of the structural and morphological properties of conjoined DPs. First, Soltan chooses to follow the coordination structure presented by Munn (1993, 1999). Second, he assumes that adjuncts can be late merged into the structure noncyclically, which is first proposed by Lebeaux (1988) (c.f. Chomsky 1993, Fox and Nissenbaum 1999, and Uriagereka 2002). The Postcyclic. 18.

(27) Merge approach is mainly adopted to account for cases where some certain LF effects, e.g. binding, cannot be dealt with under a strictly cyclic derivation. An example of such cases is provided in (20):. (20) Which picture [COMPLEMENT of Billi] [ADJUNCT that Johnj liked] did he*i/j buy?. As we can see in (20), in the deep structure both Bill and John are bound by the pronoun he; therefore, in principle the coreference between either Bill or John and he should be ruled out due to the violation of Binding Condition C, i.e. an R-expression cannot have an antecedent that c-commands it. However, as the indexation suggests, the coreference between John and he is possible. The postcyclic merge approach can help to solve the problem by simply saying that the binding conditions are evaluated before the merge of the adjunct relative clause. The third and the fourth assumption that Soltan follows are that the ϕ-features of the whole coordinate structure are obtained through feature resolution rules, and that the T head has ϕ, CLASS and EPP features. With the assumptions mentioned in the previous paragraph, the derivation of FCA is easily understood as the probe agrees with the first conjunct before the late merge of the &P applies. Soltan mentions three possible overgeneralization of ungrammatical structures in natural languages with such analysis; however, he rules out all of them. 19.

(28) with some basic assumptions or constraints.. 2.4 Bošković (2009) Bošković (2009) aims to provide a unified analysis of FCA and LCA (which stands for last conjunct agreement) based on the operation Agree. Bošković proposes that FCA occurs when the probe searches for the conjunction phrase for the number feature and the first conjunct for the gender feature in Serbo-Croatian. As for LCA, due to the existence of two potential feature valuators (the conjunction phrase and the first conjunct), it causes a problem for the probe in cases where movement, i.e. pied-piping of a valuator, is involved since both of the valuators are in principle mobile. However, Bošković suggests that the probe would initiate a second probing targeting at the second conjunct because the second conjunct is not movable and will be the only candidate for being a pied-piping valuator. Since most of the study focuses on the analysis of LCA in Serbo-Croatian which involves detailed discussion of interpretable and uninterpretable features as well as valued and unvalued features during the derivation. The analysis of LCA is way more complicated than that of FCA, here I will not go into details because the main focus of the current study is on FCA. However, I would like to mention that in this study Bošković suggests that using the ellipsis approach to analyze conjunct sensitive FCA is. 20.

(29) not plausible. As (21) indicates, the presence of zajedno ‘together’ would prove that the ellipsis approach is not possible as plural predicates like zajedno require distribution over both of the conjuncts.. (21) Telad. i. krave. calves.N and cows.F. su. juče. pasle. are yesterday grazed.PL.F. zajedno. together. ‘Calves and cows grazed together yesterday.’. 2.5 Demonte and Pérez-Jiménez (2012) Demonte and Pérez-Jiménez (2012) analyze the closest conjunct agreement in Spanish DPs, which include FCA and LCA. They focus on the agreement between determiners, adjectives and nouns in Spanish with the following structure: [DPD (A) [N1 y N2] (A)]. As we can see in this structure, the adjective can either be prenominal or postnominal. Here I provide some examples with both of the word orders in their study. (22) shows that if the adjective is prenominal the determiner and the adjective should both agree with the first conjunct; plural agreement is not possible. However, as (23a) and (23b) suggest, postnominal adjectives alter between singular and plural agreement.. 21.

(30) (22) [La. (fascinante). /. the.SG.F fascinating.SG relieve. *Las. (fascinantes)] flora. y. the.PL.F. fascinating.PL flora.SG.F. and. que lo. rugged-landscape.SG.M that it. (23) a.. [El. trigo. y. the.SG.M wheat.SG.M presentaron. componen. constitute.PRS.3PL. sorgo. disponible]. and sorghum.SG.M available.SG. no not presentaron. variantes . . .. show.PST.3PL changes ‘The wheat and sorghum available did not show any change (in their levels).’. b.. [La. agricultura. y. ganadería. the.SG.F. farming.SG.F. and cattle.SG.F. ido. industrializando.. europeas]. se. han. European.PL.F refl have.PRS.3PL. become industrializing ‘The European farming and cattle sectors have become progressively industrialised.’. 22.

(31) In this study Demonte and Pérez-Jiménez provide arguments against an ellipsis analysis of closest conjunct agreement. They do not agree with the ellipsis analysis proposed in Camacho (2003). According to Demonte and Pérez-Jiménez, the problems with the ellipsis analysis are that it cannot explain the fact that a quantificational reading is attainable in a hypothetically elliptical sentence but not in its non-elliptical counterpart and that this analysis would require some stipulation to explain why the noun in DP2 should be a remnant from the ellipsis. Besides these empirical problems, one theoretical problem is left to be accounted for – is the ellipsis a phrasal ellipsis or a head ellipsis? Both of the options would remain problems to be solved. Demonte and Pérez-Jiménez’s analysis is based on Agree operation (Chomsky 2000, 2001) and they propose that there should be two sets of ϕ-features, i.e. concord ϕ-features and index ϕ-features. According to Demonte and Pérez-Jiménez (2012: 48), “Concord phi-features are formal features related to the morphosyntactic/declensional properties of lexical items and codify instructions to the PF interface. Index phi-features are formal features related to semantic properties of lexical items and codify instructions to the LF interface.” Concord features include gender, number and case, while index features contain gender, number and person. Based on Agree operation and the assumption that there are two sets of ϕ-features, Demonte and Pérez-Jiménez derive the closest conjunct agreement by showing how the nodes with unvalued features get. 23.

(32) the values they need.. 2.6 Aljović and Begović (2016) Aljović and Begović (2016) focuses on the morpho-syntactic properties of verbal and adjectival FCA in Bosnian, Croatian and Serbian. They propose that FCA is derived through the Agree operation (Chomsky 1999, 2000, 2001) and a new post-syntactic mechanism serving as a filter on Vocabulary Insertion within the framework of Distributed Morphology. They call this filter Vocabulary Item Feature Harmony, which is concerned with the ϕ-features between the Vocabulary Items. Aljović and Begović intend to utilize feature harmony to explain the distribution of FCA and another agreement pattern, i.e. full agreement (masculine plural agreement). Although they mainly discuss FCA phenomenon in Bosnian, Croatian and Serbian, they suggest that the findings of their study can be extended to other languages as well. In this paper, Aljović and Begović address the following issues, including the locality and scope of agreement operations, the syntax-morphology interface, and the roles syntax and morphology play in deriving agreement patterns, mainly FCA. Besides, they also try to account for the variation with respect to the acceptance of the data in Bosnian, Croatian and Serbian. Aljović and Begović also discuss about the issue whether FCA is the result of. 24.

(33) clausal ellipsis. In line with Munn (1999), Marušič, Nevins, and Saksida (2007), Demonte, and Pérez-Jiménez (2012) and many others, Aljović and Begović use some collective predicates and expressions such as zajedno ‘together’, which require plural controllers, to show that the verbal FCA is not the result of ellipsis. As we can see in (24), the non-elliptical version of the sentence (24b) is ungrammatical, which suggests that the FCA sentence in (24a) does not come from clausal ellipsis.. (24) a.. Negdje. u. somewhere in. šumi. zajedno živi. forest. together live.PRES.3SG fox.SG.F and lion.SG.M. lisica. i. lav.. ‘Somewhere in the forest, live together the/a fox and lion.’. b. *Negdje u šumi zajedno živi lisica i negdje u šumi zajedno živi lav.. In the analysis section, Aljović and Begović apply Vocabulary Item Feature Harmony to account for the FCA phenomenon because their one-Probe-two-Goals approach alone cannot capture two different agreement patterns, i.e. FCA and full agreement. Within the Vocabulary Item Feature Harmony approach, if the ϕ-features and the case of the probe are harmonious with those of the two conjuncts, FCA would be more acceptable; however, if there are more differences between the ϕ-features and. 25.

(34) the case of the probe and those of the two conjuncts, FCA will be gradiently less acceptable.. 2.7 Implications for FCA in Spanish The previous works have some implications for the FCA in Spanish. First, as many of these works mention, the ellipsis analysis of FCA is not desirable. In the current study, I incorporate a plural expression juntos ‘together’ in the data to make sure that the ellipsis analysis is ruled out (contra Aoun, Benmamoun, and Sportiche 1994 and Camacho 2003). More details about this will be covered in the next chapter. Second, the coordination structure which involves adjunction in Munn (1999) is adopted in this study. However, the ideas of FCA occurring under government and exceptional government in Munn’s study will be discarded as the current study will be conducted in Minimalist Program without appealing to those concepts in the generative grammar. Third, the current study adopts Soltan’s (2007) late merge approach to FCA and if we assume the coordinate structure that involves adjunction in Munn (1999) the FCA phenomenon can be easily accounted for. However, in the next chapter I will point out that the analysis in Soltan (2007) is not adequate to account for the data in the current study. Lastly, this study focuses on the participial (adjectival) FCA in Spanish. The. 26.

(35) discussion of verbal FCA and FCA involves determiners will not be addressed here.11 Besides the implications mentioned above, in this paragraph I would like to remark on some gaps in the previous studies. First, there are few studies, if any, that focus on FCA in different constructions in Spanish. In the current study, I investigate the FCA phenomenon in five different constructions including existentials, transitives, non-finites, passives, and reflexive passives, among which FCA has not been worked on in existential in Spanish. Through this study, I attempt to provide a uniform analysis for all these constructions. Second, as far as I know, there is no analysis on Spanish FCA in the Minimalist Program that incorporates the phase theory (Chomsky 1998) or deals with the timing of adjunction. To sum up this section, in my study I incorporate a plural expression juntos ‘together’ in the data of all five constructions to discard the ellipsis analysis. Besides, following Soltan (2007), I adopt the late adjunction approach to FCA. In addition to the late adjunction approach, the phase theory (Chomsky 1998) and the operation agree (Chomsky 1999, 2000, 2001) will help to show the timing of the adjunction. With all these “ingredients”, we will be able to show how FCA in Spanish is derived.. 11. Although there are two constructions (passives and reflexive passives) that involve both verbal FCA and adjectival FCA, the examples of FCA for these two constructions are not possible. 27.

(36) Chapter Three FCA in Various Types of Constructions in Spanish. In this chapter, I first present the basic structure in Spanish with a simple sentence, and then the basic properties of the constructions concerned in this study will be introduced. Besides, I will also provide the FCA and full agreement data of these constructions. The data show that in Spanish FCA is observable in existential, transitive and non-finite constructions while it is not possible in passive and reflexive passive constructions. Such differentiation comes from the interaction between the operation Agree and adjunction. More details will be discussed in the next chapter.. 3.1 Basic Structure in Spanish As a start of this chapter, I would like to show the basic structure in Spanish through a simple sentence, as illustrated in (25) below. As the syntactic tree (26) of this sentence demonstrates, there are two separate projections above vP, i.e. AgrP and TP.12 The reason for having an extra projection is due to the morphology of the verbs. From the verb conjugation, besides the root of the verb, we can also find a morpheme for the tense inflection and another morpheme for the agreement inflection (see Baker 1985;. 12. I assume that the ϕ features are in Agr and no specifier for T. 28.

(37) Legate 1999; Belletti 2001; Fernández-Salgueiro 2011a; among many others for the morpho-syntax representation). For example, the verb tenemos ‘we have’ can be decomposed into ten-e-mos, with the e being the tense inflection and the mos the agreement inflection.. (25) Nosotros tenemos we. have.PRES.1PL. un. libro.. a.SG.M. book.SG.M. ‘We have a book.’. (26) AgrP Agr’. Nosotros Agr T. Agr mos. v V. T e v. TP T. vP. [pres] v’. DP <Nosotros> v. VP. ten V <tener>. DP un libro. 29.

(38) Thus, in the structure (26) we can notice that the verb tener ‘to have’ goes through head movement from V to v, v to T, and then finally adjoins to Agr. In the trees in this study, the infinitive form of the verb would sit in V with angle brackets, while the verb with inflection would end up in Agr with separate morphemes representing the root, the tense morpheme, and the agreement morpheme. After having introduced the basic structure in Spanish, in what follows I will present the basic properties of the constructions in discussion, i.e. existential, transitive, non-finite, passive, and reflexive passive, as well as the FCA and the full agreement data of these various constructions.. 3.2 Existentials In English existential constructions, the verb be agrees with the following nominal in number, as can be seen in (27) and (28).13 However, in standard Spanish, the verb haber in existential constructions does not agree with the following nominals.14 As we can see in (29) and (30), the present tense verb hay does not change its form no matter what entity it precedes, e.g. a singular nominal in (29) and a plural nominal in (30).. 13. However, note that FCA is obligatory when the verb be precedes a conjoined NP. According to Rodríguez-Mondoñedo (2005), Spanish can be grouped into two dialects based on the agreement in existential constructions with haber ‘to be’. In the first type, the existential verb haber has a default agreement, i.e. third person singular. As for the second type, the verb haber agrees with the internal nominal. In this study I only investigate the first type since the no agreement dialect is the standard one. 14. 30.

(39) This phenomenon can also be observed in French. From (31) and (32), the present tense existential verb a does not show agreement with the following entities.. (27) There is a shining star in the sky. (singular, English). (28) There are ten koalas in the zoo. (plural, English). (29) Hay. un. have.PRES.IND a.SG.M. niño. en. boy.SG.M in. el. parque. (singular, Spanish). the park. ‘There is a boy in the park’. (30) Hay. diez árboles en. have.PRES.IND ten trees. in. el. parque. (plural, Spanish). the park. ‘There are ten trees in the park’. (31) Il y it. a. un. there have a.SG.M. garçon. dans la. boy.SG.M on. ‘There is a boy on the street.. 31. rue. (singular, French). the street.

(40) (32) Il y it. a. six. personnes. there have six persons. dans la in. maison. (plural, French). the house. ‘There are six persons in the house.’. Based on the facts provided above, we assume that there is no Agr projection associated with the existential verb haber, as opposed to regular sentences. As a result, probe-goal agree between haber and the following nominal does not apply, so no agreement can be observed. In this case, if we are to observe the FCA phenomena in such constructions, there should then exist other elements that exhibit partial agreement with the coordinate structure. In this study I place a participle between the existential verb and the coordinate structure; therefore, the structure of the sentence is as follows: [TP haber [AP participle] [DP1 DP1 and DP2]], where the participle is the probe, which agrees with the first conjunct in number and gender. (33a) is a sentence with such structure. In (33a) the participle colgada ‘hung’ agrees with the first conjunct una lámpara ‘a lamp’ in number (singular) and gender (feminine), showing the FCA phenomenon. In addition, the sentence with full agreement between the participle and coordinate structure is also presented here as (33b), where we can see the participle with the plural and default masculine markings.15 15. Notice that the optionality between the two patterns, i.e. FCA and full agreement, is not on FCA itself but on the timing of the adjunction of the &P. For the transitive and non-finites constructions that will be discussed in Chapter 3.3 and 3.4, this is the same case. 32.

(41) (33) a.. Hay. colgada. una. lámpara y. have.PRES.IND. hung.SG.F. a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. chandelier.SG.M. together in. en. el. techo.. un. (FCA). the ceiling. ‘There is a lamp and a chandelier hung in the ceiling.’. b.. Hay. colgados. una. lámpara y. have.PRES.IND hung.PL.M. a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. en. techo.. chandelier.SG.M. together in. el. un. (Full agreement). the ceiling. ‘There are a lamp and a chandelier hung in the ceiling.’. In (33a), one thing worth noting here is the use of juntos ‘together’. Aoun, Benmamoun and Sportiche (1994, 1999, henceforth ABS) and Camacho (2003) propose that partial agreement involves ellipsis as I mentioned in the previous chapter.16 According to ABS and Camacho, partial agreement constructions appear due to the ellipsis of the predicate of one of the conjuncts. Hence, in (33a) above the sentence would have been like (34) originally.17. 16 17. Partial agreement includes first conjunct agreement and last conjunct agreement. Notice that example (34) is not grammatical. 33.

(42) (34) *Hay. colgada una. lámpara y. hay colgado. have.PRES.IND hung.SG.F a.SG.F. lamp.SG.F and have hung.SG.M. candelabro. juntos. el. chandelier.SG.M. together in. en. un a.SG.M. techo.. the ceiling. Intended meaning: ‘There is a lamp and a chandelier hung together in the ceiling.’. Later the existential verb hay and the masculine participle colgado ‘hung’ in the second conjunct are elided. However, according to Bošković (2009), the elements being coordinated under this analysis are clauses instead of traditional conjuncts. Not convinced with the ellipsis analysis mentioned above, Marušič, Nevins, and Saksida (2007, MNS henceforth) incorporate some plural predicates such as collide and together in their Slovenian data (see (35) below) to argue that partial agreement in terms of ellipsis is not applicable to conjunct-sensitive agreement since plural predicates treat a coordinate structure as a single entity. 18 Therefore, if partial agreement is still possible in the presence of such predicates, it is not an outcome of ellipsis. With the help of juntos in (33a), we can rule out the ellipsis analysis.. Aljović and Begović (2016) also provide arguments against an ellipsis analysis with the data from Bosnian/ Croatian/ Serbian (see also Johannessen 1996, Munn 1999, Harbert and Bahloul 2002, Soltan 2007, and Demonte and Pérez-Jiménez 2012 among many others). 18. 34.

(43) (35) Krava. in. njena. cow.SG.F and her. teleta. so. se. pasla. calves.N are refl graze.PL.N. skupaj together. ‘A cow and her calves were grazing together.’ (MNS 2007). In order to know the internal structure of the sentences (33a) and (33b), I provide the tree diagrams in (36) and (37).19 In both of the tree structures, we can see that the existential verb haber raises from V to v and from v to T.20 The v in Spanish existential constructions is defective since it cannot host an external argument. As the sentence *La clase hay un alumno ‘Intended meaning: There is a student in the class.’ shows, the external argument la clase ‘the class’ results in the ungrammaticality of the sentence. However, if no external argument exists in the sentence, e.g. Hay un alumno en la clase ‘There is a student in the class’, the sentence would be grammatical. As a result, the ungrammaticality of the sentence with an external argument suggests that the v in Spanish existentials is defective. One main difference between the derivations of (33a) and (33b) is that for (33a) the operation Agree between the participle and the first conjunct applies before the adjunction of the &P, i.e. acyclic adjunction as mentioned in the first chapter, yielding FCA; whereas, for (33b) adjunction of the &P applies. Notice that the sequence juntos en el techo ‘together on the ceiling’ is not presented in the tree because the position of these phrases is irrelevant to the current study. 20 The morphology of the verb hay is also presented in the tree. 19. 35.

(44) before the operation Agree between the participle and the coordinate structure, showing full agreement. The derivations are shown in (36) and (37) respectively.21. (36). TP T vdef. vP T. AP. vP. [pres] colgada V ha. vdef. y. vdef. VP V. DP1. <haber> DP1 una lámpara. &P &. DP2. y un candelabro. Notice that in this study I treat the AP colgada ‘hung’ as an adjunct to vP. The position of the AP is not important to the current study. 21. 36.

(45) (37). TP T vdef. vP T. AP. vP. [pres] colgados V ha. vdef. y. vdef. VP V. DP1. <haber> DP1. &P. una lámpara. &. DP2. y un candelabro. 3.3 Transitives In the transitive construction discussed in this study, the participle modifies the object. As can be seen in (38a-b), the participle inclinada ‘leaned’ agrees the object la cabeza ‘the head’ in number (singular) and gender (feminine).. (38) a.. Pepito. tiene inclinada. la. cabeza. Pepito. has. the.SG.F. head.SG.F to. leaned.SG.F. ‘Pepito has his head leaned to the right.’. 37. a. la. derecha.. the right.

(46) b.. Pepito tiene la cabeza inclinada a la derecha.. In (39), the participle inclinados ‘leaned’ agrees with the object los pies ‘the feet’ in number (plural) and gender (masculine). There are two possible positions for the participle, i.e. either pre-object (38a) or post-object (38b). However, if we would like to observe the FCA phenomenon, we will only consider the pre-object option.22. (39) Pepito Pepito. tiene inclinados. los. pies. has. the.PL.M. feet.PL.M to. leaned.PL.M. a. la. izquierda.. the left. ‘Pepito has his feet leaned to the left.’. In addition, although Agr projection does exist in this construction, the verb raising to Agr does not agree with the object; instead, it agrees with the subject in the specifier of vP. From (39) and (40), we can notice that the verbs change their forms based on the subjects, with Pepito agreeing with tiene (3rd, singular) and nosotros agreeing with tenemos ‘we’ (1st, plural) and that the ϕ-features of the DP(s) in the object position do not affect the agreement between the verb and the subject.. 22. Remember that in the first chapter I mentioned that late adjunction cannot be applied to an A-moved phrase. If we late adjoin an &P to the object in the post-object configuration (i.e. having undergone Amovement), it will cause a mismatch in the ϕ-features between the object and the already agreeing participle and the ungrammaticality of the sentence. 38.

(47) (40) Nosotros tenemos inclinada we. have .1PL leaned.SG.F. la. cabeza. a. the.SG.F. head.SG.F to. la. derecha.. the right. ‘We have our head leaned to the right.’. The tree diagram in (41) presents the structure of (40). As can be seen in (41), the verb tener ‘to have’ raises from V to v, v to T, and T to Agr to agree with the closest DP, i.e. the silent subject nosotros ‘we’ in the specifier of vP before the subject further moves to Spec-AgrP. In addition, the participle inclinada ‘leaned’ agrees with the object la cabeza ‘the head’ in number (singular) and gender (feminine).23. Notice that the preposition phrase a la derecha “to the right” is not present in tree. Its position in the tree is not relevant to the current study. 23. 39.

(48) (41). AgrP Agr’. DP Nosotros Agr T v V te. Agr mos T e. v. TP T vP [pres] <Nosotros> v. v’ VP AP inclinada. VP V DP <tener> la cabeza. In order to observe FCA phenomena in Spanish transitive constructions, we need a coordinate structure, i.e. una lámpara y un candelabro ‘a lamp and a chandelier’, in the object position. To make the sentence similar to sentence (33a), the existential verb hay is replaced with a transitive verb tenemos ‘we have’, yielding (42a), whereby the participle colgada agrees with the first conjunct una lámpara in number (singular) and gender (feminine), showing that FCA is possible. Besides, the sentence with full agreement between the participle and the coordinate structure is presented in (42b). The. 40.

(49) difference in the agreement pattern lies in the relative order of the application of operation Agree and the adjunction of the &P. The tree structures of (42a-b) are provided in (43) and (44).24. (42) a.. Tenemos. colgada una. lámpara y. have.PRES/PAST.1PL. hung.SG.F a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. techo.. chandelier.SG.M. together in. en. el. un. (FCA). the ceiling. ‘We have a lamp and a chandelier hung in the ceiling.’. b.. Tenemos. colgados. una. lámpara y. have.PRES/PAST.1PL. hung.PL.M. a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. chandelier.SG.M. together in. en. el. techo.. un. (Full agreement). the ceiling. ‘We have a lamp and a chandelier hung in the ceiling.’. 24. In the tree structures, I put pro in both Spec-vP and Spec-AgrP. In this study I am abstracting away from the details of pro. See Holmberg (2005), Sheehan (2006), Fernández-Salgueiro (2011b) among many others for more details. 41.

(50) (43) AgrP Agr’. pro Agr T. Agr mos. v. T. TP T. vP. [pres/past] v’. pro. e V ten. v. v. VP AP. VP. colgada V <tener> DP1. DP1. 2 Adjunction ○. of the &P &P. una lámpara & DP2 y un candelabro. 42.

(51) (44) AgrP Agr’. pro Agr T v. Agr mos T. TP T [pres/past] pro. vP v’. e V. v. v. VP. ten AP. VP. colgados V <tener>. 2 Agree ○. DP1 ○ 1 Adjunction of the &P. DP1. &P. una lámpara &. DP2. y un candelabro. 3.4 Non-finites In Spanish non-finite clauses (45a-b), the PRO in (45a) is “singular” while the PRO in (45b) is “plural” based on the form of the adjectives feliz ‘happy (singular)’ and felices ‘happy (plural)’. Although the PROs are different, the ser “to be” verb in both cases is unaffected; it remains an infinitive. The T head in non-finite clauses is defective and there is no Agr projection. The structure of (45a-b) is shown in (46).. 43.

(52) (45) a.. [TP Ser to be. PRO. feliz]. es. PRO. happy.SG is. muy difícil. very difficult. ‘To be happy is very difficult.’. b.. [TP Ser to be. PRO. felices]. es. PRO. happy.PL is. muy difícil. very difficult. ‘To be happy is very difficult.’. (46). TP T’. TP Ser PRO feliz/felices. T v. V. vP T. v’. TP. v <Ser PRO feliz/felices> v. VP. es V <ser>. AP muy difícil. The non-finite clause showing FCA phenomenon is presented in (47a). Notice that there is a clause following the first one. By doing so, the proposition of this sentence 44.

(53) will be more natural. However, the detailed structure of that part of the clause will not be shown in the tree. As can be seen, the participle colgada agrees with the first conjunct una lámpara in number (singular) and gender (feminine). Once again, FCA can be observed. In addition, the sentence with full agreement between the participle and the coordinate structure is provided in (47b). The different agreement pattern once again arises from the relative order of the application of the operation Agree and the adjunction of the &P. The tree structure of (47a-b) is presented in (48) and (49) below.25. (47) a.. Al. estar. colgada. una. lámpara y. because to.be. hung.SG.F. a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. chandelier.SG.M. together in. en. el. techo,. the ceiling. un. nos marchamos. refl left. (FCA) ‘Because a lamp and a chandelier are hung together in the ceiling, we left.’. b.. 25. Al. estar. colgados. una. lámpara y. because to.be. hung.PL.M. a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. en. el. techo,. Here I am abstracting away from whether PRO moves or not. 45. un. nos marchamos..

(54) chandelier.SG.M. together in. the ceiling. refl left. (Full agreement) ‘Because a lamp and a chandelier are hung together in the ceiling, we left.’. (48) CP C Al. TP Tdef v V estar. Tdef v. vP v’. PRO v. VP AP colgada. VP. V <estar>. DP1. DP1. &P. una lámpara &. DP2. y un candelabro. 46.

(55) (49) CP C Al. TP Tdef v V estar. Tdef v. vP v’. PRO v. VP AP. VP. colgada 2 Agree ○. V <estar> DP1 una lámpara & y. DP1 &P DP2 un candelabro. 47.

(56) 3.5 Passives The basic pattern of Spanish passive constructions is illustrated in (50) below:. (50) AgrP Agr’. DP Subj Agr T ser. Agr. TP T <ser>. vP vdef. VP V. DP. Verb. Subj. In Spanish passive constructions, the subject can either be preverbal (through AMovement) or post-verbal (base position). Due to the fact that FCA appears only when the subject is post-verbal, the pattern with a preverbal subject will not be considered in this study.26 The agreement relation in passive constructions is much more complicated, with the auxiliary verb ser agreeing with the subject in person and number, as well as the participle agreeing with the subject in number and gender. Therefore, there are two. 26. Please refer to footnote (22). 48.

(57) probes in passive constructions. As shown in (51) and (52), the auxiliary verb ser agrees with un coche ‘a car’ in (51) and Juan y María ‘Juan and Maria’ in (52), giving fue (3rd, singular) and fueron (3rd, plural) respectively. Regarding the participles, robado ‘stolen’ in (51) agrees with un coche in number (singular) and gender (masculine), while entrevistados ‘interviewed’ in (52) agrees with the coordinate DP Juan y María ‘Juan and Maria’ in number (plural) and gender (masculine).2728. (51) Fue was.3SG. robado. un. coche. ayer.. stolen.SG.M. a.SG.M. car.SG.M. yesterday. ‘A car was stolen yesterday.’. (52) Fueron entrevistados were .3PL interviewed.PL.M. Juan. y. María. John.SG.M and Maria.SG.F. esta mañana. this morning. ‘Juan and Maria were interviewed this morning.’. The passive example in this study is provided in (53a). As suggested by the asterisk, FCA is prohibited in Spanish passive constructions. Neither the auxiliary verb fue nor. 27. In Spanish, when two DPs with different genders are coordinated, the whole DP then has a default masculine gender. 28 Here, I suggest that the two probes also have to Agree. However, this assumption will lead to more discussion that is beyond the scope of the current study. I will leave this issue to future studies. 49.

(58) the participle colgada can only agree with the first conjunct una lámpara of the postverbal subject. Here, only the pattern with full agreement, i.e. fueron colgados ‘were hung’, is possible, as shown in (53b). In (53b), both the auxiliary verb fueron and the participle colgados agree with the coordinate structure una lámpara y un candelabro. Therefore, only the situation that the adjunction of the &P happens before the operation Agree applies is possible. The tree structure of (53a-b) is shown in (54).. (53) a.. *Fue. colgada. una. lámpara y. to-be.PAST.3SG. hung.SG.F. a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. chandelier.SG.M. together in. en. el. un. techo. (FCA). the ceiling. Intended meaning: ‘A lamp and a chandelier were hung together in the ceiling.’. b.. Fueron. colgados. una. lámpara y. to-be.PAST.3PL. hung.PL.M. a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. chandelier.SG.M. together in. en. el. un. techo. (Full agreement). the ceiling. ‘A lamp and a chandelier were hung together in the ceiling.’. 50.

(59) (54) AgrP Agr’. pro Agr T. Agr. TP T. vP. Ø *fue [past]. <ser> [past] vdef. VP. fueron V. DP1. *colgada colgados. DP1. una lámpara & y. &P DP2 un candelabro. 51.

(60) 3.6 Reflexive Passives The basic structure of reflexive passives in Spanish is shown in (55) below:. (55) AgrP Agr’. DP Subj Agr T vdef. Agr T. TP T [pres]. vP vdef. VP. <se> V. vdef. V. DP. Verb. se. <Verb>. Subj. According to Kelling (2006), in Spanish reflexive passive construction, the verb has to be a transitive one and it is only possible with the third person. Besides, the theme in reflexive passives is realized as the subject. When the agent of the action is present in the sentence, it will lead to an ungrammatical sentence as in (56a). In line with passives constructions, the subject of the reflexive passives in Spanish can also either be preverbal (56b) or post-verbal (56c). Notice that the post-verbal one is more prevalent, i.e. less marked. Due to the same reason as in the passive constructions, in this study we will only focus on the post-verbal position of the subject.. 52.

(61) (56) a.. *Los. contratos se. firmaron por el. futbolista.. the. contracts refl. sign.PAST by the soccer player. ‘The contracts were signed by the soccer player.’. b.. Los. contratos se. firmaron.. the. contracts refl sign.PL. ‘The contracts were signed.’. c.. Se. firmaron los. contratos.. refl. sign.PL. contracts. the.PL. ‘The contracts were signed.’. (Kelling, 2006). In this study, the examples of reflexive passive is shown in (57). As the asterisk suggests, FCA is not possible in reflexive passives just like in passives. In this sentence we can see that neither the verb mantenerse ‘to keep’ nor the participle colgada ‘hung’ can only agree with the first conjunct.29 Again, only the pattern with full agreement, i.e. se mantienen colgados ‘were kept’, is available. In (57b), both the reflexive verb se. 29. Notice that just like the passive construction, in reflexive passive construction, there are also two probes, i.e. the reflexive verb and the participle. 53.

(62) mantienen and the participle colgados agree with the coordinate structure, meaning that the adjunction of the &P is applied before the operation Agree. (58) demonstrates the tree structure of (57).30. (57) a.. *Se mantiene colgada. una. lámpara y. un. refl keep.3SG hung.SG.F. a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. en el. techo. (FCA). chandelier.SG.M. together in the. ceiling. Intended meaning: ‘A lamp and a chandelier are kept together in the ceiling.’. b.. Se. mantienen. colgados. una. lámpara y. refl keep.3SG. hung.PL.M. a.SG.F. lamp.SG.F and a.SG.M. candelabro. juntos. chandelier.SG.M. together in the. en el. un. techo. (Full agreement) ceiling. ‘A lamp and a chandelier are kept together in the ceiling.’. 30. In the tree, I put the reflexive se in vdef so that vdef assigns an accusative case to the theme (i.e. the subject) and the subject gets a person ϕ-feature. See Richards (2008) and Cuza, Czerwionka and Olson (2016). 54.

(63) (58) AgrP Agr’. pro Agr T. Agr. vdef T vdef. V *se mantiene se mantienen. TP T [pres]. vP AP *colgada. vP. colgados vdef <se>. VP V <mantener>. DP1. DP1 una lámpara & y. &P DP2 un candelabro. In both the passive construction and the reflexive passive construction, FCA is not observable; instead, the sentences with full agreement are the legitimate ones. The ungrammaticality of (53a) and (57a) cannot be explained within the analysis in Soltan (2007). If we follow Soltan’s analysis, we can say that the two probes, i.e. fue and colgada as well as mantiene and colgada, both agree with the first conjunct una lámpara, and then the late adjunction of the &P, y un candelabro, happens afterwards.. 55.

(64) However, these operations result in an ungrammatical sentence. Therefore, it seems that Soltan’s analysis is not quite adequate to account for the data in Spanish.31 I propose that the observed pattern results from the interaction among the operation Agree, the timing of adjunction and phases. Besides, ϕ-complete Agree between the verb associated with the Agr projection and the first conjunct blocks FCA; ϕ-defective Agree between the participle and the first conjunct, however, displays FCA. More details will be mentioned in Chapter 4.. 3.7 Summary of the Five Construction Types Concerning FCA To summarize this section, from sentences (33a), (42a), (47a), (53a) and (57a) restated below as (59-63), we can capture a pattern with the existential, transitive and non-finite constructions allowing FCA; whereas, for passive and reflexive passive constructions FCA is not observable.. (59) Existentials32 Hay colgada una lámpara y un candelabro juntos en el techo. ‘There is a lamp and a chandelier hung together in the ceiling.’. Soltan’s analysis works well for the existential sentence (33a), the transitive sentence (42a) and the non-finite sentence (47a) though. 32 For the glosses, please refer to the content above. 31. 56.

(65) (60) Transitives Tenemos colgada una lámpara y un candelabro juntos en el techo. ‘We have a lamp and a chandelier hung together in the ceiling.’. (61) Non-finites Al estar colgada una lámpara y un candelabro juntos en el techo, nos marchamos. ‘Because a lamp and a chandelier are hung together in the ceiling, we left.’. (62) Passives *Fue colgada una lámpara y un candelabro juntos en el techo. Intended meaning: ‘A lamp and a chandelier were hung together in the ceiling.’. (63) Reflexive Passives *Se mantiene colgada una lámpara y un candelabro juntos en el techo. Intended meaning: ‘A lamp and a chandelier were hung together in the ceiling.’. 57.

(66) Chapter Four Analysis: Timing of Adjunction, Agreement, and Phases. In this chapter, I attempt to answer the question arises in the previous chapter: why do existentials, transitives, and non-finites allow FCA but passives and reflexive passives don’t? At first glance, it may seem to be a construction specific phenomenon. However, I argue that it is actually related to the agreement of the ϕ-features between the probe and the goal. Remember that we have already looked at acyclic adjunction (Soltan 2007) and how it works in interaction with the operation Agree to derive FCA in the previous chapters. I will now include agreement and phases in my analysis and discuss them in detail. By observing and examining the tree structures of the five construction types, I propose that ϕ-complete Agree between the verb associated with the Agr projection and the first conjunct blocks FCA; ϕ-defective Agree between the participle and the first conjunct, however, displays FCA. Besides the details of the agreement between the probe and the goal, the phase concept from Chomsky (1998), i.e. the interaction between the adjunction of the &P and phases, also help to explain FCA phenomenon in Spanish and to show the timing of the acyclic adjunction of the &P. Based on the data in this study, for those constructions with one phase, i.e. existentials, passives, and. 58.

(67) reflexive passives, I argue that FCA occurs when the operation Agree applies prior to the late adjunction of the &P to the first conjunct; while for those constructions with two phases, i.e. transitives and non-finites, FCA happens when the ϕ-defective Agree between the participle and the first conjunct applies before the &P late adjoins to the first conjunct in phase one right before the complement of the phase head is sent to spellout. If there is an Agr projection, the ϕ-complete Agree between the verb moving to the Agr projection and the unpronounced subject doesn’t apply until phase two. Notice that in both cases the proposal on agreement mentioned earlier in the paragraph still applies. Therefore, the ungrammaticality of the FCA examples of passives and reflexive passives can be accounted for. In the following, I would like to use the three “ingredients” (late adjunction, agreement, and phases) and go through each of the five constructions to show how FCA is possible or impossible as well as the timing of the adjunction of the &P. Before proceeding to do this task, first I will talk about the completeness and defectiveness of ϕ-features as I mentioned the ideas of ϕ-complete and ϕ-defective in the proposal on agreement.. 59.

(68) 4.1 Completeness and Defectiveness of ϕ-Features In order to understand what ϕ-complete Agree and ϕ-defective Agree are, we need to know about the details of the ϕ-features and case assignment during the operation Agree33. I will use examples in English to illustrate the distinction between ϕ-complete Agree and ϕ-defective Agree.. (64) John will sing opera.. (65) John seems [. to sing opera]. In the example (64), when the subject John is in the specifier of vP, it has the interpretable person and number features but lacks the case feature. When it agrees with the T head will, which has an uninterpretable nominative case feature but lacks the person and number values, the subject John gives the T head will the person and number values and then the T head will assigns the subject John a nominative case. After the operation Agree takes place, the subject John A-moves from the specifier of vP to the specifier of TP. While in the example (65)3435, the subject, i.e. the trace of the subject 33. In the current study, only the nominative case is related to the analysis; therefore, the accusative case will not be discussed. 34 Notice that we only focus on the discussion of the sentence in the square brackets. 35 I am aware of the fact that the ϕ features in T is inherited from C according to Chomsky (2008). However, since the example (65) is a raising sentence, there is no CP layer. Here, I would assume that the ϕ features are in T. 60.

(69) John, has the person and the number features. After undergoing Agree with the defective T head to, John moves from the specifier of vP to the specifier of TP. During the operation Agree, the defective T head to only gets the person value but not the number value; hence, it cannot assign a case to John. From the discussion above, a conclusion can be drawn. When an operation Agree leads to case assignment, there should be ϕ-complete Agree, involving both the person and the number feature. However, when an operation Agree doesn’t lead to case assignment, it is ϕ-defective Agree, meaning that either the person or the number feature is missing. This is supported by the description in Radford (2009), where he briefly summarizes the completeness condition in Chomsky (1999) as the following (α is the probe and β is the goal):. (66) Completeness Condition (Radford 2009) An uninterpretable case/agreement feature on a constituent α is deleted when α agrees (in respect of one or more ϕ-features) with a ϕ-complete constituent β (i.e. with a nondefective β which carries a complete set of ϕ-features – both person and number).. 61.